Chapter  8:  Diagnosis and Treatment of Swallowing Disorders (Dysphagia) in Acute-Care Stroke Patients: Evidence Report/Technology Assessment Number 8

Objectives

This report was requested by the Health Care Financing Administration, which sought an evidence-based assessment of methods for diagnosing and treating swallowing disorders (dysphagia) in elderly individuals with neurologic diseases, and specifically those methods associated with the services provided by speech-language pathologists. About 6,228,000 Americans over age 60 have dysphagia, and about 300,000 to 600,000 people are affected by dysphagia resulting from neurologic disorders each year.

This report addresses four questions: (1) How does diagnosis of dysphagia or aspiration affect treatment courses and patient outcomes? (2) What are the indications for diagnosing patients using a full bedside exam (BSE), modified barium swallow, fiberoptic endoscopy, or other instrumented exams? (3) Is one diagnostic technology more effective than any other? and (4) When is noninvasive therapy appropriate, and when should feeding tubes be used in certain patient populations?

Search Strategy

We used broad-based literature searching to ensure that we found all relevant information. Thus, we searched 23 electronic databases, including MEDLINE and Embase. Search terms included those relevant to the disorder, diagnostics, epidemiology, etiology, and treatment of dysphagia. Search dates depended on the database, but ranged from 1945 to July 1998. We also searched 23 World Wide Web sites, conducted hand searches of article bibliographies, and sought published and unpublished information from experts in the field.

Selection Criteria

We adopted broad criteria to ensure retrieval of all relevant information. Articles were retrieved if they were clinical studies of any design containing information derived from 10 or more patients. Two researchers independently requested articles. Disputes were resolved in favor of obtaining the article.

Data Collection and Analysis

Data from clinical trials on diagnostic test performance and/or patient outcomes were entered directly into evidence tables. This information included study size, design, diagnostic method(s), patient selection criteria and characteristics, and patient outcomes.

Studies were separated by their design, but their quality was not necessarily ranked according to it. We did not rigidly adhere to a formal quality ranking scheme because some randomized controlled trials (RCTs) had flaws serious enough to render their results unreliable. Similarly, we did not always consider case-controlled studies to provide more reliable information than case series because of certain flaws in the former.

To reach our conclusions, we performed two cost-effective analyses, a meta-analysis of the efficacy of the 3-ounce water test, a similar meta-analysis of the BSE, an exploratory meta-analysis (so called because the control data were derived from historical sources) on the effects of dysphagia diagnosis and treatment programs on patient outcomes, an illustrative meta-analysis that addresses the technical issue of the low statistical power of vote-counting procedures, and numerous other quantitative analyses. The need to perform these original calculations is a reflection on the relatively poor reporting in this literature.

Main Results

We focused on dysphagia diagnosis and treatment programs because assessment of dysphagia diagnosis and treatment independent of one another provides only limited information. The emphasis of our analysis was on stroke victims, which reflects the preponderance of the literature. Stroke victims also comprise the largest group of patients with neurologic disease who have dysphagia. Outcomes of particular interest were aspiration pneumonia, malnutrition, dehydration, and quality of life (QOL), but the most data were available for aspiration pneumonia.

The evidence regarding the first question, which is the most significant of our key questions, suggests that dramatic reductions in the occurrence of pneumonia are observed when a systematic program of diagnosis and treatment of dysphagia in an acute stroke management plan is implemented. Because these data are from historically controlled studies rather than RCTs (which may be unethical in this context), the exact magnitude of this reduction in pneumonia rates is difficult to determine. Also for this reason, it is equally difficult to determine the contribution of the dysphagia-specific aspects of the management programs to these rate reductions (as opposed to those aspects of the stroke management program not related to diagnosis and treatment of dysphagia). However, because the effects observed in these studies are substantial, it would be imprudent to ignore them. Therefore, these results must be taken as evidence of efficacy of these programs. Malnutrition, dehydration, and QOL were not addressed in available studies.

Regarding the second question, the risk for developing aspiration pneumonia cannot be accurately predicted from any single clinical sign or symptom. There is a clear-cut need to optimize a brief initial exam that employs several key signs and symptoms to accurately detect patients with possible unsafe swallows and who therefore need more extensive testing.

The results of the studies of diagnostic test performance that address the third question are erroneous because treatment necessarily follows diagnosis of a swallowing disorder. This treatment makes it impossible to distinguish between the true- and false-positive results of these tests. Because of this, the purely diagnostic abilities of the various tests cannot be determined; thus, the tests can only be compared by assessing how they direct treatment and influence adverse patient outcomes, as addressed in our first question above.

We constructed two cost-effectiveness models to address this issue, and their results suggest that use of the full BSE in dysphagia programs reduces costs. These models, which are based on published protocols and findings, assume that patients will receive a preliminary bedside assessment, and that one result of this assessment is that no more than about 39 percent of patients will be referred for further evaluation, as occurred in the best study we evaluated. Under these conditions, our cost-effectiveness analysis suggests that dysphagia diagnosis and treatment programs that employ the full BSE would either save money or have very little net cost if they reduced pneumonia rates by amounts similar to those obtained in certain published studies. In addition, our results indicate that the slightly higher costs of videofluoroscopy would be offset if it provided slightly less than an additional 10 percent reduction in pneumonia rates, again assuming that no more than about 39 percent of patients are referred for videofluoroscopy or the full BSE.

Evidence related to the fourth question consists of limited data from a single RCT concluding that a soft mechanical diet including thickened liquids results in lower morbidity than does a pureed diet. Although numerous other studies have been conducted on dysphagia treatments, their designs make it impossible to assess the efficacy of individual treatments. A single RCT with low statistical power reported inconclusive results about the effect of treatment intensity level on patient outcomes.

A well-designed trial that compares dysphagia management programs using different diagnostic modalities is needed. This trial should take into account the current lack of a demonstrated gold standard diagnostic test, have acceptable statistical power, and consider the fact that treatment of patients distorts measures of diagnostic test performance. Because of this urgent need, we provide a detailed description of the design and analysis of a trial that would answer several major unanswered questions.

Conclusions

Available evidence on the diagnosis and treatment of dysphagia is extremely limited. Nevertheless, current data suggest that implementation of dysphagia management programs for stroke patients in the acute-care setting is accompanied by a reduction in pneumonia rates. Use of the full BSE in these programs appears to be cost-effective. The limitations of available evidence do not allow us to determine the extent to which videofluoroscopy or fiberoptic endoscopy reduce pneumonia rates compared with the full BSE despite the fact that it is reasonable to believe the additional information they provide would lead to improved patient outcomes. Although the added benefits provided by instrumented exams are likely to be small, videofluoroscopy may be cost-effective if used in programs in which no more than about 39 percent of all patients are referred to this test, and if its use leads to an additional 10 percent reduction in pneumonia rates, compared with the full BSE. There is a clear-cut need to optimize a brief initial exam that accurately detects patients with possible unsafe swallows who may therefore need more extensive testing. Study designs used in the evaluation of noninvasive therapy have made it impossible to assess the efficacy of individual treatments. However, there is evidence supporting the use of a soft mechanical diet over a pureed diet for preventing aspiration pneumonia in stroke patients with dysphagia who have a history of aspiration pneumonia. The evidence on whether treatment that is more intensive yields better patient outcomes than less intensive treatments is inconclusive. Well-designed studies, particularly those that compare diagnostic modalities, are needed.

This document is in the public domain and may be used and reprinted without permission, except for those copyrighted materials noted for which further reproduction is prohibited without specific permission of copyright holders.

Suggested Citation

Diagnosis and Treatment of Swallowing Disorders (Dysphagia) in Acute-Care Stroke Patients. Evidence Report/Technology Assessment No. 8. (Prepared by ECRI Evidence-based Practice Center under Contract No. 290-97-0020.) AHCPR Publication No. 99-E024. Rockville, MD: Agency for Health Care Policy and Research. July 1999.

1. How does diagnosis of dysphagia or aspiration affect the subsequent course of treatment and outcomes?

In answering this question, we focus on dysphagia diagnosis and treatment programs in general. We address the specific diagnostic tools and the specific treatments that might be used to optimize these programs in Questions 3 and 4, respectively.

Current evidence suggests that implementation of a systematic program of diagnosis and treatment of dysphagia in an acute stroke management plan yields dramatic reductions in pneumonia rates. Because these data are derived from historically controlled studies rather than RCTs, the exact magnitude of this reduction in pneumonia is difficult to determine. Also for this reason, it is equally difficult to determine the contribution of the dysphagia-specific aspects of the management programs to these rate reductions (as opposed to those aspects of the stroke management program not related to diagnosis and treatment of dysphagia). However, because the effects observed in these studies are substantial, it would be imprudent to ignore them. Therefore, these results must be taken as evidence of efficacy of these programs.

Partly for the reasons noted in the preceding paragraph, it seems prudent to include dysphagia-specific management with formal diagnosis and treatment as part of the standard protocol of stroke management in acute-care settings, despite the sparse available data. Also, these programs appear to have a minimal potential to harm patients. Finally, dysphagia diagnosis and treatment programs may be cost-effective. Thus, withholding such programs would deny potential benefits to patients, perhaps at some additional costs.

2. What are the appropriate indications for having patients diagnosed using a full BSE, MBS, fiberoptic endoscopy, or another instrumented exam?

The risk for developing aspiration pneumonia cannot be accurately predicted from any single clinical sign or symptom. There is a clear-cut need to optimize a brief initial exam that accurately detects patients with possible unsafe swallows and who therefore need more extensive testing. An optimum combination of signs and symptoms for such an initial test has not been determined, and further appropriate research is needed. This research should seek to devise an initial test that has high sensitivity and moderate specificity. This would ensure that the great majority of patients with unsafe swallows are appropriately referred for further testing and that only a minimum number of patients with safe swallows are erroneously referred for more extensive testing.

3. Is there any evidence that one diagnostic technology provides more useful information than any other diagnostic?

Neither videofluoroscopy nor fiberoptic endoscopy can serve as a perfect gold standard for detection of aspiration, because each yields false-negative and false-positive results. Without a third, better reference standard, the ability of these two methods to detect aspiration cannot be compared with each other.

Full BSEs can have sensitivities for aspiration near 80 percent, with specificities near 70 percent. Epidemiological evidence indicates that about half of the patients with dysphagia who aspirate do so silently (without a cough). These two points, taken with the very low pneumonia rates observed in dysphagia management programs that used full BSEs, indicate that these exams are capable of detecting most aspiration, even silent aspiration (or that any undetected aspiration does not contribute greatly to the pneumonia rate).

The ability of the full BSE to detect silent aspiration should be optimized in future research. Whether it can be conclusively stated that an optimized full BSE can entirely replace imaging exams such as videofluoroscopy and fiberoptic endoscopy depends partly on the degree to which the BSE is optimized, and partly on the additional benefit that results from the direct internal visual information provided by the imaging exams.

In all studies that attempted to predict pneumonia from the results of a diagnostic test, measurements of the ability of a diagnostic test to predict pneumonia, when expressed in terms of test sensitivity and specificity, were distorted by intervening treatment that prevented some pneumonia from occurring. This distortion cannot be circumvented by withholding treatment, because this would be unethical. Therefore, it does not appear possible to ethically measure either the absolute or relative sensitivities or specificities of dysphagia diagnostic modalities for predicting pneumonia. This also appears to be true for other important patient outcomes such as malnutrition, dehydration, and QOL. This means that the only ethical method of comparing various diagnostic modalities is to conduct controlled trials that measure the combined effects of diagnosis and treatment on the rate of pneumonia and/or other patient outcomes. This limitation on study design provides one reason why the superiority of any given diagnostic method is not conclusively shown by currently available data.

Another reason that current studies do not conclusively show the superiority of any diagnostic test relates to their small size. These studies do not have enough statistical power to allow one to detect the effects of interest, which may be small.

The differences in the ability of various diagnostic tests to predict pneumonia are likely to be small because the two studies using BSEs were so successful (nearly all pneumonia was prevented). It is difficult to obtain a statistically significant improvement upon their results.

Although currently available data do not allow one to determine the degree to which, or even whether, use of videofluoroscopy or other instrumented exams lead to lower pneumonia rates than the full BSE, it is entirely reasonable to expect that their use might lead to lower pneumonia rates. This is because instrumented exams provide more information than the full BSE. Partly for this reason, and even though no data currently demonstrate a difference in effectiveness among diagnostic modalities, one should not conclude that our results prove that the BSE and instrumented exams are equivalent. In terms of patient outcomes, we have found no evidence of a difference between these technologies, not evidence of no difference. Research that appropriately addresses this issue is needed.

Because no data satisfactorily compare the abilities of the full BSE and videofluoroscopy to prevent pneumonia, we constructed two cost-effectiveness models to address this issue. These models, which are based on published protocols and findings, assume that patients will receive a preliminary bedside assessment, and that one result of this assessment is that no more than 39 percent of patients will be referred for further evaluation. Under these conditions, our cost-effectiveness analysis, although based on less than perfect data, suggests that dysphagia diagnosis and treatment programs that employ the BSE would either save money or have very little net cost if they reduced pneumonia rates by amounts similar to those obtained in certain published studies. In addition, our results indicate that the slightly higher costs of videofluoroscopy would be offset if it provided slightly less than an additional 10 percent reduction in pneumonia rates, again assuming that no more than 39 percent of patients are referred for videofluoroscopy or the full BSE.

Some epidemiological evidence suggests that dysphagia patients who aspirate have about a 50 percent greater risk of developing aspiration pneumonia than dysphagia patients who do not aspirate during videofluoroscopy exams. However, the relationship between aspiration and pneumonia is not perfect. Other patient characteristics such as oral hygiene and immune system strength may play equal or even greater roles in causing pneumonia. This loose relationship between aspiration and pneumonia means that aspiration should not be considered a definitive marker for the patient outcome of pneumonia.

Large RCTs measuring patient outcomes such as aspiration pneumonia, malnutrition, dehydration, and QOL are needed to determine the comparative effectiveness of the various dysphagia diagnostic and treatment methods. These trials should enroll patients in reasonably homogeneous disease groups or stratify patients into appropriate subsets. We have included detailed suggestions for such studies along with some estimates of the number of patients needed.

4. When is noninvasive swallow therapy appropriate? Does it work particularly well or particularly poorly in any particular patient population? What can the evidence tell us about this therapy? Are feeding tubes useful or a last resort that might be avoided for some patients by dysphagia diagnosis and therapy?

Most study designs used in the evaluation of noninvasive therapy have made it impossible to assess the effectiveness of individual treatments.

The results of a single RCT supported the use of a soft mechanical diet over a pureed diet for preventing aspiration pneumonia in stroke patients with dysphagia who had a history of aspiration pneumonia.

A single RCT reported inconclusive results about the effect of treatment intensity level on patient outcomes. The statistical power of this trial may have been too low to detect the appropriate differences.

Elderly Defined

Elderly is usually not well defined. The ages encompassed by this classification vary from researcher to researcher. While Medicare may consider the elderly to be those over 65 years, many others have included those as young as 55. The oldest old are those 85 years old and over. For the purposes of this report, we attempt to compare results of different studies in such a way that the ages included are comparable. Results for different subgroups of the elderly are presented when available; the results of the oldest old, who are likely to be the most disabled, are presented separately.

Dysphagia Defined

Table 1. Examples of Definitions of Dysphagia

Table 1. Examples of Definitions of Dysphagia

Author	Year	Definition
Kahrilas	1989	Implies difficulty in swallowing
Dorland's Illustrated Medical Dictionary	1994	Difficulty in swallowing
American Academy of Otolaryngology-Head and Neck Surgery	1995	Feeling of difficulty passing food or liquid from the mouth to the stomach
Buchholz	1996	Deficiency in the achievement of the purposes of eating: (1) pleasure, both personal and social, and (2) nutrition and hydration
Boyce	1997	Sensation of delay in passage of a food bolus within 10 seconds of initiation of a swallow

These definitions vary along a continuum from the specific and quantitative (e.g., Boyce, 1997) to those with subjective and qualitative elements (e.g., Buchholz, 1996). Differences among them are not insignificant, and have implications for the scope of the diagnosis and treatment of dysphagia. For example, Boyce (1997) confined the disorder to sensations of delay in food passage, thus presumably excluding silent aspiration (aspiration resulting in no symptoms and thus not detectable to the patient or on BSE) as an outcome of dysphagia. The definition by Buchholz (1996) would include not only silent aspiration, if classified as a deficiency in nutrition and hydration, but also deficiencies in eating pleasure that arise from oral problems such as poorly fitting dentures.

For the purposes of this assessment, we will define dysphagia here very generally, following Kahrilas (1989), as disordered swallowing. This definition does not require the patient to consciously perceive the disorder and, therefore, includes silent aspiration. However, the focus from all disorders that may arise during feeding is narrowed to only those that originate after the bolus has passed through the initial oral stages of mastication (oral dysphagia) and is at least at the point when transit from the mouth to the pharynx is initiated reflexively (oropharyngeal dysphagia). We will not consider esophageal dysphagia in this report, which falls more into the realm of gastroenterology than within the realm of disorders dealt with by speech-language pathologists (SLPs) and otolaryngologists who sometimes identify these problems and bring them to the attention of gastroenterologists. In our discussion and analyses, we generally accept each study's definition of dysphagia but discuss any outcome differences that may have resulted from these different definitions and diagnostic technologies used. In particular, the differences in diagnosed occurrence among dysphagia diagnosed at BSE versus VF or fiberoptic methods [videofluoroscopic swallowing studies (VFSS), modified barium swallow (MBS), fiberoptic endoscopic examination of swallowing (FEES), and fiberoptic endoscopic examination of swallowing and sensory test (FEESST)] are noted, because BSE only detects dysphagia with overt symptomatology, and instrumented methods may detect cases that are asymptomatic but will not necessarily cause a problem to the patient.

Physiology and Symptomatology of Dysphagia

This section will focus on the physiological dysfunctions and the particular symptoms demonstrated by patients with dysphagia due to neurologic disease. While the Epidemiology section of this report focuses on the overall occurrence of dysphagia in each of several neurologic disorders, this section will center specifically on what complaints these patients have, as well as what particular abnormalities in physiological processes during oropharyngeal dysphagia are present during instrumented exam (usually VFSS or MBS).

Physiology: Background

Table 2. The Oral and Pharyngeal Stages of a Normal (more...)

Table 2. The Oral and Pharyngeal Stages of a Normal Swallow Table 2a. Oral Stage

Function	Results
Tongue progressively elevates anteriorly to posteriorly.	Carries material through the oral cavity.
Groove is created on the midline of the tongue.	Guides material down the tongue.
Material reaches the anterior faucial arch region.	The pharyngeal swallow is triggered, which ends the oral stage of the swallow.

Table 2b. Pharyngeal Stage

Table 2b. Pharyngeal Stage

Function	Results
Velopharyngeal musculature elevates, closes nasal cavity, and approximates the pharyngeal wall.	Prevents material from entering the nasal cavity.
Tongue base retracts toward the posterior pharyngeal wall.	Creates pressure to drive the bolus through the pharynx.
Posterior pharyngeal wall contracts toward the tongue base.	Creates pressure to drive the bolus through the pharynx.
Hyoid elevation occurs.	Lowers the epiglottis and lifts the larynx.
Larynx closes at three levels: True vocal folds False vocal folds Base of epiglottis to aryepiglottic folds.	Protects airway.
Larynx elevates and moves forward.	Protects airway; pulls open cricopharyngeal and upper sphincter.
Cricopharyngeal sphincter relaxes and is pulled open.	Allows material to pass into esophagus; pharyngeal stage of swallowing ends.

Although a discussion of the physiology of a normal swallow is beyond the scope of this report, a certain degree of understanding is necessary to discuss abnormalities demonstrated by patients with neurogenic dysphagia. Swallowing is generally discussed as occurring in three stages: oral, pharyngeal, and esophageal (some authors include oral preparatory as a fourth stage) (Hardy and Robinson, 1993). Tables 2a and 2b describe the oral and pharyngeal stages of normal swallowing.

Swallowing is a complex function that involves several nerve and muscle groups. Six of the 12 cranial nerves and 4 cervical nerves play intrinsic roles in the swallowing process; they mediate communication between the brain stem and the muscles of the oral and pharyngeal regions to perform a complex set of processes necessary for a safe swallow. The muscles of the oral and pharyngeal region send messages to the brain stem reticular formation, medullary integrative center, and frontal cortex through the 7th (facial), 9th (glossopharyngeal), and 10th (vagus) cranial nerves; messages are received from the brain through the 9th, 10th, and 12th (hypoglossal) cranial nerves (Bass and Morrell, 1992; Lugger, 1994); these nerves also innervate muscles involved in swallowing. In addition to these four cranial nerves, two other cranial nerves (trigeminal and spinal accessory) also innervate muscles used in chewing and swallowing, while four cervical nerves are integral to the pharyngeal phase of swallowing. Thus, damage to any of these nerves, muscles, or lesions in the brain stem, medulla, or cortex can lead to disorders of swallowing.

Clinical Symptoms and Reported Complaints

Table 3. Detectable Symptoms of Oral-Pharyngeal Dysphagia (more...)

Table 3. Detectable Symptoms of Oral-Pharyngeal Dysphagia During a Physical Exam or BSE

Neurogenic symptoms: oral	Neurogenic symptoms: pharyngeal
Drooling Difficulty chewing Difficulty initiating swallow Dry mouth Compensatory strategies Retention of food in lateral sulcus, anterior sulcus, beneath tongue, or hard palate Spitting food out Slowed oral transit Excessive lingual movement	Throat clearing Wet voice Hoarseness Nasal regurgitation Coughing Choking Laryngospasm Pneumonia Stridor Feeling food catch in throat Excessive saliva or mucus Delayed elevation of hyoid bone

A variety of symptoms manifest themselves as a result of the above-mentioned physiologic abnormalities. Table 3 lists some of the general clinical signs and symptoms of swallowing disorders readily detectable during a physical exam or a BSE. It is also necessary for the clinician to talk to the patient and ask questions, because the patient is often aware of problems he or she is having. The patient may report an inability to chew, a feeling of food catching in his or her throat, problems in holding food in the mouth, collection of food low in the throat, or coughing or choking before or during the swallow (Logemann, 1983a).

Some of these observable symptoms may be indicative of a more serious problem, such as aspiration. The Results section discusses the predictive value of some of these signs and symptoms in predicting aspiration and pneumonia.

Diagnosis of Oropharyngeal Dysphagia

Oropharyngeal swallowing disorders are diagnosed with three levels of diagnostic methodologies. First, there is a preliminary clinical examination of patients with suspected dysphagia or with a condition considered to put the patient at risk for dysphagia. Second, in some centers, patients with clinically suspected dysphagia are then given a formal BSE by an SLP or other specialist with training in dysphagia diagnosis and treatment. Third, based on the clinical or BSE and input from support staff involved in feeding activities, the patient may be referred for an imaging study. We describe these diagnostic steps below. Comparison and evaluation of the diagnostic efficacy of these exams is deferred to the Results section of this report; however, we discuss here some peripheral issues such as the physical limitations and convenience factors involved in these exams. For these latter issues, literature was not sufficient for a formal evaluation.

Instrumented and Imaging Exams

Videofluoroscopic swallowing studies

The VFSS is recognized as the reference standard for assessment of oropharyngeal swallowing, although it has not been shown to be a perfect gold standard. This technique was originally developed for observation of the esophageal region. It was later modified for assessment of the oropharyngeal component of swallowing (Ekberg, 1982; Logemann, 1983b), and is thus called the MBS. The MBS is often carried out in a radiological suite, although portable units are available. Recently, the MBS has begun to be conducted in mobile clinics set up for fluoroscopic exams, or with transportable units that can be taken into facilities lacking an inhouse fluoroscopic unit. The MBS is administered by a team consisting of a radiologist to perform the fluoroscopy, an SLP to assess swallowing function and administer therapy, and often a nurse or paramedical attendant to assist the patient. Administration of radiation is legally limited to medical doctors; however, a radiologist per se is not a legal requirement, and mobile units may use other medical doctors.

The MBS is often administered to the patient in an upright seated position, although the patient's typical eating position can also be used if it is different. The patient attempts to swallow barium-impregnated boluses of different consistencies, progressing from solids to pudding to thick liquids and ending with thin liquid (although some practitioners reverse this order). If aspiration is observed, patient maneuvers or various diet consistencies are tried to see if aspiration can be eliminated or minimized. The exam is ended if unpreventable or dangerous aspiration is observed. The patient is observed from the front and from the side, with the side view being the most useful. Because of limitations on radiation exposure, the entire test usually is conducted within approximately 5 minutes. This exam allows direct observation of not only aspiration and timing, but also of other structural and functional anomalies of the swallow. The dynamic fluoroscopic images are captured on videotape and can be viewed repeatedly and in slow motion following the exam.

While such VFSS are generally considered the most direct and comprehensive method for assessing swallowing function, these exams are not without limitations. In a study involving 11 clinicians and 182 patients with dysphagia in 10 nursing homes on the U.S. East Coast and Midwest, Hageman (Hageman, unpub.) recently found that 55 percent of these patients with dysphagia of typical etiologies were not able to undergo VFSS. The top nine reasons in order were: status too mild, poor patient cooperation, transportation difficulty, status too severe, insufficient oral motor ability, patient was combative, physician denial, payer denial, and patient refusal. Some of these reasons are the results of the healthcare system, and are not inherent limitations of the exam itself. Other authors (Bastian, 1993; Kaye, Zorowitz, and Baredes, 1997; Kim, Goodhart, Aviv et al., 1998; Langmore, Schatz, and Olson, 1991; Spiegel, Creed, and Selber, unpub(a)) have also criticized the VFSS method for the above limitations, as well as for its inability to image soft tissue, observe dry swallows, observe a normal meal throughout its full time course, and its limitation to passive observation of bolus passage without the ability for direct sensory testing in the absence of a bolus.

It is an important aspect of VFSS and other imaging methods that they are not limited to passive observation of signs and symptoms, but instead provide immediate feedback to the patient and clinician in terms of maneuvers, therapies, and diet changes that may prevent or improve aspiration and other abnormal aspects of swallowing. Thus, they can be a major part of the planning, initiation, and monitoring of treatment.

In terms of safety, the radiation dosage involved in modern fluoroscopy is not considered a major concern (Beck and Gayler, 1990; Martin and Hunter, 1994; Potvin, Kaur, and Williams, 1991; Wright, Boyd, and Workman, 1998), particularly for the elderly patients discussed in this assessment. We found no evidence-based reports in the literature of morbidity or deaths caused by aspiration of barium compounds during oral-pharyngeal swallowing studies. There were case reports of two deaths (Gray, Sivaloganathan, and Simpkins, 1989) and one serious morbidity report (Penington, 1993) from barium aspiration; however, all of these appeared to be cases in which lack of attention to the possibility of aspiration allowed aspiration of large quantities of barium typically used in gastrointestinal (GI) tract barium swallows. This seems unlikely in exams intended to assess oral-pharyngeal dysphagia, because they begin with the specific purpose of observing whether there is any aspiration of small initial quantities of barium. Allergic reactions to barium sulfate appear to be quite rare and have been estimated to occur at a rate less than two per million (Muroi, Nishibori, Fujii et al., 1997).

Fiberoptic endoscopy

A more recent development is the FEES (Bastian, 1991; Bastian, 1993; Kidder, Langmore, and Martin, 1994; Langmore, Schatz, and Olsen, 1988; Selkin, 1984). This exam is administered by an otolaryngologist and an SLP; in some states, an SLP alone may administer the exam, while in other states, this exam is beyond the scope of an SLP's practice. It is performed with portable equipment at the patient's bedside. A fiberoptic device attached to a video camera is inserted nasally so that the pharynx and larynx can be observed from above. The image is observed on a small video monitor and is recorded on videotape. Patients are first observed swallowing their own secretions. Next, food and liquid of decreasing viscosity are swallowed. Various postural positions can be tried. Barium is not necessary, but dye is often added to the food to increase visibility (the image is in color). FEES cannot directly observe the oral region during swallowing, and in the pharyngeal and laryngeal regions, it is limited to observations immediately before and after a swallow, because pharyngeal closure obscures the exact moment of swallowing. However, aspiration can be clearly detected before the swallow, and aspiration during the swallow is usually detectable by post-swallow residue in the trachea or by the act of coughing up aspirated material. Because a fiberoptic exam can be carried out over the full course of a meal, fatigue factors can be noted that might be missed on the necessarily briefer VFSS evaluation. Also, the portability and repeatability of FEES allow followup assessment-therapy sessions as needed throughout the course of dysphagia therapy. This not only can provide feedback necessary for changes or reinforcement of therapy but may also allow the discontinuation of ineffective or unnecessary therapy.

The fiberoptic exam is not limited to passive observation. Sensory and reflex information can be obtained by touching the tip of the device to different pharyngeal structures or administering regulated air puffs (Aviv, 1997; Aviv, Martin, Keen et al., 1993; Kim, Goodhart, Aviv et al., 1998). If a pneumatic aspect is involved, it is called a FEES and sensory test. Advocates of FEES and FEESST believe the sensory measures, particularly bilateral deficits, are prognostic for intermittent aspiration that may not always be observable on short VFSS imaging tests (Aviv, unpub.). Because of the pneumatic equipment involved, FEESST is not as easily portable as FEES. In the present report, we will sometimes use FE as a generic term to refer to all of the above forms of fiberoptic endoscopic exams.

The MBS and FE methods can both provide patient feedback on swallowing performance. Both methods provide both overlapping and unique information; thus, it is not necessarily a matter of replacing one test with another. The issue is the proper use of both tests. However, some practitioners prefer to be able to directly view the swallow and possible aspiration with a VFSS, if this type of exam is available and the patient's condition is suitable.

Repetitive oral suction swallow (the ROSS test)

The ROSS test has only been used in Sweden (Nilsson, Ekberg, and Hindfelt, 1995; Nilsson, Ekberg, Olsson et al., 1998); however, it illustrates the use of simple noninvasive instruments to evaluate swallowing, and addresses oral function, a stage that cannot be well evaluated by either VFSS or FE methods. In the ROSS test, the patient repetitively sucks water through a straw from a glass that sits on a scale. The scale is connected to a strip-chart recorder that plots the removal of the water from the glass. A pressure detector in the straw records the magnitude of the suction through the straw. A Doppler probe is attached to the patient's neck just to the right of the midline at the level of the cricothyroid membrane, and it records Doppler shift at the mid-pharyngeal region. At the same level on the left side of the neck, a piezoelectric movement sensor records movement of the larynx. Finally, a temperature sensor placed in one of the patient's nostrils records airflow. These devices provide a simultaneous plot of the magnitude and timing of all of the above functions as the patient attempts to drink 200 ml of water.

Piezoelectric computerized laryngeal analysis (CLA)

In the CLA exam, a miniature piezoelectric strip is taped on the exterior of the neck at the laryngeal prominence. As the patient is asked to dry swallow or to swallow various consistencies and sizes of bolus, the speed, extent, and timing of laryngeal elevation are measured and recorded as a strip-chart readout on a laptop computer. Normal swallow references for all of these measurements are a part of the computer's program, so that instantaneous comparison to these standards are made, and the chart readout indicates with colors and markers where deviations from normal occur. While the information provided appears to be mostly limited to laryngeal movement, as discussed in the Physiology and Symptomatology of Dysphagia section above, this movement is an integral part of laryngeal protection. Delay, insufficiency, or aberrations of this movement are strong indicators of aspiration and other swallowing problems. Thus, while this exam is not as comprehensive as the MBS or a FE exam, it may provide sufficient information to detect and treat aspiration in some cases.

As with FEES, the CLA exam can be performed at the bedside, be used for dry swallows, assess swallowing during the entire course of normal meals, and be carried out repeatedly throughout the course of treatment to evaluate results. There is a less expensive and simpler version of the CLA that substitutes a sensing and signaling device for the computer readout; a patient's acceptable swallowing parameters can be programmed into the device so that it signals when the patient is not achieving adequate parameters. It can thus be used to monitor swallowing on a long-term, ongoing basis by the patients themselves or by minimally trained support staff and family members in nursing facilities and at home.

CLA is a relatively new technology, but a small, unpublished study compares it with VFSS for prediction of the important patient outcome of aspiration pneumonia. Because we did not comprehensively solicit unpublished studies from all dysphagia device manufacturers, we did not consider it fair to other manufacturers to present this CLA study in the Results section of this assessment. However, we describe this study here. In spite of this study's small size (16 patients), it was a prospective trial in which patients were each given both tests, and in which the readers of one test were blinded to the results of the other test. The endpoint in this trial was pneumonia, and patients were followed for 6 months. This would be a strong study design if patients positive for aspiration on either test were treated identically. Unfortunately, such was not the case for this study, because treatment was based only on the VFSS results. The results of CLA were not used in patient management. The fact that some patients were treated, which likely prevented some cases of pneumonia, means that sensitivity and positive predictive value (PPV) may have been underestimated for both tests. Furthermore, because only the patients who received a positive VFSS test result were treated to prevent pneumonia, the VFS sensitivity and PPV were likely underestimated to a greater degree. The results of this study cannot be reliably interpreted because of this design flaw; thus, the effectiveness of CLA for predicting patients at risk for pneumonia compared with VFSS is not currently known. However, further research on this new technology, as well as others described below, should be encouraged.

Other dysphagia analysis techniques in development

All of the above methods had at least one study that attempted to assess the impact of the diagnostic method on aspiration pneumonia frequency. Although the lack of comparative or controlled studies does not allow us to formally assess other instrumented methods in this evidence report, brief descriptions of some of these other methods are worthwhile. Among them are manometry, manofluorography, electromyography (EMG), electrical impedance tomography, scintigraphy, pulse oximetry, respiratory pattern analysis, echoplanar magnetic resonance imaging (MRI), cervical auscultation, and ultrasound.

EMG measures electrical activity of muscles. This activity increases during muscular contraction. EMG can be used to monitor the activity of the muscles involved in swallowing, particularly the middle pharyngeal constrictor and the cricopharyngeus (Hanson, Lawson, and Remacle, 1995). Lack of coordination of contraction of these two muscles is taken to be indicative of a swallowing disorder. Other abnormalities include increased resting contraction or decreased activity of one or the other of these muscles.

Manometry is used to measure pharyngeal or esophageal motility. A pressure-sensing device, or manometer, is inserted into the lumen, and the pressure exerted by the walls on various parts of the manometer during swallowing is measured. The timing of muscular contraction is noted. The presence or absence of peristalsis can be measured, as well as the timing of peristalsis (Johnston, Collins, McFarland et al., 1993). These measurements are influenced by the consistency of the bolus, body position, manometer placement, and gastric fullness, as well as the presence of any swallowing disorder. The utility of this method may be limited by difficulties in ascertaining the exact position of the manometer unless performed simultaneously with VF (Ergun, Kahrilas, and Logemann, 1993). Combining the two techniques (manofluorography) allows coordinated analysis of bolus transport and pharyngeal or esophageal pressure (Olsson, Nilsson, and Ekberg, 1994).

The citric acid cough test can be used to measure the cough reflex. The patient irritates his or her throat by inhaling nebulized citric acid dissolved in saline (Sekizawa, Ujiie, Itabashi et al., 1990). Greater concentrations of citric acid produce greater irritation. The concentration of citric acid required to produce cough is determined. Higher concentrations indicate a diminished cough reflex.

Oximetry has been used to indicate the occurrence of aspiration. Aspiration of food or liquid is thought to cause reflex bronchoconstriction and can cause a rapid alteration in the degree of oxygen saturation of arterial blood, which can be measured using a finger probe attached to a pulse oximeter (Zaidi, Smith, King et al., 1995). Simultaneous measurement of arterial oxygen saturation and assessment of dysphagia using VF suggest that pulse oximetry may detect the presence or absence of silent aspiration (Collins and Bakheit, 1997). Although there are several studies on pulse oximetry, none have addressed whether use of this technique has an impact on patient pneumonia rates and, for this reason, we do not formally assess this technology.

Respiratory inductance, respirometry, plethysmography, and nasal thermistor airflow recording are other methods of detecting cardiopulmonary adaptation during and after swallowing (Nilsson, Ekberg, Bulow et al., 1997; Rogers, Msall, and Shucard, 1993).

Scintigraphy during the swallowing of a radionuclide may detect aspiration (Muz, Mathog, Miller et al., 1987). This method allows determination of flow dynamics and quantification of the amount of liquid aspirated (Hamlet, Muz, Farris et al., 1992).

Critically ill patients are often fed a high-glucose formula through NG tubes. If this formula is aspirated, glucose will be found in the trachea. When these patients have a tracheostomy or translaryngeal intubation, the presence of glucose in the tracheal secretions can easily be assayed (Winterbauer, Durning, Barron et al., 1981).

Echoplanar magnetic resonance imaging, a form of MRI, allows superior temporal resolution compared with conventional MRI (Gilbert, Daftary, Woo et al., 1996). This rapid imaging may enable the clinician to distinguish the rapid events of swallowing better than conventional MRI.

Respiratory patterns during swallowing can be measured electronically (Selley, Flack, Ellis et al., 1989a). Sounds produced during swallowing and breathing can be recorded and timed. In neurologically impaired patients, these patterns differ from those of healthy controls (Selley, Flack, Ellis et al., 1989b). Use of this technique, the Exeter dysphagia assessment technique (EDAT), to assess oropharyngeal motor and sensory function may aid in diagnosis of dysphagia (Selley, Flack, Ellis et al., 1990). Simultaneous recording of swallowing behavior using VF and EDAT allow events recorded by VF to be related to events in the respiratory pattern as recorded by EDAT (Selley, Ellis, Flack et al., 1994).

Listening to the patient's breathing patterns through direct, stethoscopic auscultation may enable the physician to detect unusual sounds such as bubbling, which may indicate aspiration (Zenner, Losinski, and Mills, 1995).

The thickness of the esophageal wall and the width of the lumen can be measured using ultrasonography (Sobin, Nathanson, and Engstrom, 1996). Patients with esophageal motility disorders have been reported to show a prolonged swallowing time and widened liquid- and air-filled lumina. This method of measurement may prove useful in the diagnosis of dysphagia. Ultrasonography can also be used to detect abnormal tongue movements during swallowing that might be associated with dysphagia (Wein, Bockler, and Klajman, 1991). Abnormal movements of the hyoid bone and associated muscles can be detected using ultrasound duplex-Doppler imaging (Sonies, Wang, and Sapper, 1996).

Electrical impedance tomography measures conductivity of tissues. Pharyngeal conductivity changes during swallowing. Measuring the timing of this conductivity change may allow measurement of the speed and timing of a swallow (Hughes, Liu, Griffiths et al., 1996).

Summary of Instrumented Diagnostic Tests

Table 4. Reported Advantages and Disadvantages of Each (more...)

Table 4. Reported Advantages and Disadvantages of Each Instrumented Diagnostic Test

Test	Reported advantages	Reported disadvantages
Technologies supported by published clinical trials measuring pneumonia as the outcome
VFSS	1 Visualizes frequency and amount of aspiration. 2 Visualizes physiologic variables such as timing of swallow and paralysis. 3 Can test several amounts and viscosities of boluses. 4 Can test therapies during the diagnostic exam.	1 Not portable. 2 Requires patient to be able to sit upright and understand instructions. 3 Time limitation: cannot observe entire meal. 4 Cannot image soft tissue. 5 Cannot examine dry swallows. 6 Cannot measure accumulated secretions in laryngopharynx. 7 Only analyzes motor component of swallow. 8 Small radiation risk.
FEES	1 Portable to bedside of patient. 2 Can document entire course of meal to record fatigue effects. 3 No radiation risks. 4 Can assess sensory and reflex status of pharynx through endoscopic touch.	1 Cannot visualize exact moment of swallow or aspiration due to white out. 2 Cannot visualize oral region.
FEESST	1 Tests sensory aspects of swallow through quantitative pneumatic measures: identifies patients who may experience aspiration due to lack of laryngeal sensation. 2 Can document entire course of meal to record fatigue effects. 3 No radiation risks.	1 Not as easily portable as FEES. 2 Cannot visualize exact moment of swallow or aspiration due to white out. 3 Cannot visualize oral region.
Technologies supported by other clinical trials
Manometry	1 Allows quantitative analysis of pressure dynamics within the oropharynx. 2 Detects elevated resting pressure indicative of gastroesophageal reflux. 3 No radiation risks.	1 Variability precludes intercenter comparison of absolute measurements. 2 Does not measure the axial component of pharyngeal contraction. 3 Perfusion manometers may not register rapid pressure changes. 4 Does not provide information on the degree of esophageal sphincter opening.
Manofluorography	1 Allows quantitative analysis of pressure dynamics within the oropharynx. 2 Detects elevated resting pressure indicative of gastroesophageal reflux. 3 Fluorography allows more precise control of manometer placement, reducing variability of measurements. 4 Allows coordinated analysis of bolus transport and intraluminal pressure changes.	Perfusion manometers may not register rapid pressure changes.
EMG	1 No radiation risks. 2 Surface EMG is noninvasive.	1 There is often no correlation between electrical events on the pharyngeal muscles and intraluminal pressure changes. 2 Needle electrodes may cause patient discomfort. 3 Surface EMG cannot distinguish the contribution of individual muscles to the signal.
CLA and other piezoelectric movement detectors	1 Portable to bedside of patient. 2 No radiation risks. 3 Noninvasive. 4 Requires little patient cooperation. 5 Can document entire meal.	1 Aspiration cannot be viewed directly, but rather is inferred by the degree of oral/laryngeal dysfunction. 2 Cannot measure accumulated secretions in laryngopharynx.
Pulse Oximetry	1 Portable to bedside of patient. 2 No radiation risks. 3 Noninvasive. 4 May detect silent aspiration.	1 Older patients, smokers, and patients with chronic lung disease may confound test results.
Scintigraphy	1 Enables detection and quantification of small amounts of aspirated material. 2 Allows determination of flow dynamics	Cannot examine dry swallow. Radiation risk.
Technologies supported by published initial demonstrations
Citric Acid Cough Test	1 Measures cough reflex.	1 Only measures cough reflex.
ROSS Test	1 Noninvasive. 2 Provides objective information describing the oral phase of swallow. 3 Portable to bedside of patient. 4 No radiation risks.	1 Piezoelectric transducer is subject to distortion from patient movement. 2 Piezoelectric transducer measures only extent, but not direction of movement.
Tracheal Glucose Assay	1 Detects aspirated feeding formula.	1 Only useful in tube-fed patients.
Respirometry	1 Respirometry has been reported only as a supplement to VFSS. It therefore shares the advantages and disadvantages of this technique.	1 Nasal obstruction may hamper recording. 2 Respirometry has been reported only as a supplement to VFSS. It therefore shares the advantages and disadvantages of this technique.
Echoplanar MRI	1 Provides soft tissue imaging. 2 Enables visualization of the entire swallow.	1 The adduction-abduction procedure cannot be visualized entirely within a single axial slice. 2 Multislice acquisitions place restrictions on temporal resolution. 3 Limited spatial resolution allows visualization of gross surfaces only.
Stethoscopic Auscultation	1 Noninvasive. 2 Portable to bedside of patient. 3 No radiation risks.	1 No nomenclature exists to describe swallowing sounds. 2 No data are available ascribing specific sounds to specific swallowing events.
EDAT	1 Noninvasive. 2 Portable to bedside of patient. 3 No radiation risks.	1 Provides only limited data and is therefore a supplemental, rather than a replacement technology. 2 May be confounded by patient deficits unrelated to swallow, such as visual impairment.
Ultrasonagraphy	1 May detect abnormal tongue and hyoid bone movements. 2 Noninvasive. 3 No radiation risks.	1 Can be confounded by motion of the patient or the equipment. 2 Doppler imaging requires careful selection of the appropriate wavelengths.
Electrical Impedance Tomography	1 No radiation risks. 2 Noninvasive. 3 Portable to bedside of patient.	1 Poor spatial resolution. 2 Incorrect electrode placement may lead to misinterpretation of the temporal aspects of the signal. 3 May be subject to limitations based on the size and shape of the patient's laryngeal framework. 4 Can be confounded by patient movements unrelated to swallow. 5 Results become less reproducible with decreasing bolus size.

Table 4 summarizes the suggested advantages and disadvantages of each of the three major instrumented diagnostic tests. While the information provided by these tests overlaps, because of their differences they may each detect patients at risk that the other tests do not detect. It may ultimately not be a matter of deciding which test is the best one to use, but which patients should undergo which test based on symptoms. However, evidence is not yet available to selectively guide patients to different diagnostic methods. Similarly, evidence is not yet available to determine whether any single difference in the physiological parameters measured by any of these tests is important. For this reason, we do not attempt to assess whether any particular difference leads to improved patient outcomes.

Treatment of Oropharyngeal Dysphagia

Noninvasive Therapies

Numerous noninvasive approaches exist for treating oropharyngeal dysphagia, and the method used depends upon the type of underlying disorder and its manifestations. Logemann (1983, 1991, 1994; Logemann and Kahrilas, 1990) classifies these approaches into three categories: compensatory techniques, indirect therapy, and direct therapy.

Compensatory techniques

Compensatory techniques attempt to eliminate the symptoms of dysphagia, but not to change the actual swallow physiology. This is usually accomplished by teaching patients to position their heads and bodies to control the flow of food or liquid, by modifying the consistency and volume of food, and by modifying the rate at which the food is given. Thermal stimulation with special appliances or food, as well as prosthetics, are also sometimes used as compensatory measures (Logemann, 1991, 1994).

Table 5. Postural Techniques for Aspiration Elimination (more...)

Table 5. Postural Techniques for Aspiration Elimination in Dysphagia

Observed disorder (VFSS)	Posture applied	Rationale
Inefficient oral transit	Head back	Utilizes gravity to clear oral cavity
Delayed swallow reflex	Chin down	Widens valleculae to prevent bolus from entering airway
Reduced posterior motion of tongue	Chin down	Pushes tongue base backward toward pharyngeal wall
Unilateral laryngeal dysfunction (aspiration during swallow)	Head rotated to damaged side	Places extrinsic pressure on thyroid cartilage, increasing adduction
Reduced laryngeal closure (aspiration during swallow)	Chin down Head rotated to damaged side	Puts epiglottis in more protective position; narrows laryngeal entrance; increases vocal fold closure by applying extrinsic pressure
Reduced pharyngeal contraction	Lying down on one side	Eliminates gravitational effect on pharyngeal residue
Unilateral pharyngeal paresis	Head rotated to damaged side	Eliminates damaged side from bolus path
Unilateral oral and pharyngeal weakness on the same side	Head tilt to stronger side	Directs bolus down stronger side
Cricopharyngeal dysfunction	Head rotated	Pulls cricoid cartilage away from posterior pharyngeal wall, reducing resting pressure in cricopharyngeal sphincter

Postural techniques are usually used temporarily until the patient's swallow function recovers, or until direct therapy begins to have an effect. Occasionally, patients with extreme neurologic or structural damage must use these techniques permanently to eliminate aspiration. Postural techniques used to alter the flow of food are shown in Table 5.

Table 6. Diet Modification Techniques

Table 6. Diet Modification Techniques

Swallowing disorder	Easiest food consistencies	Food consistencies to avoid
Reduced range of tongue motion	Liquid	Thick foods
Reduced tongue coordination	Liquid	Thick foods
Reduced tongue strength	Liquid	Thick, heavy foods
Delayed pharyngeal swallow	Thick liquids and thicker foods	Thin liquids
Reduced airway closure	Pudding and thick foods	Thin liquids
Reduced laryngeal movement; cricopharyngeal dysfunction	Liquid	Thicker, higher viscosity foods
Reduced pharyngeal wall contraction	Liquid	Thick, higher viscosity foods
Reduced tongue base posterior movement	Liquid	Higher viscosity foods

Food consistency alterations commonly used for different dysfunctions are shown in Table 6. (Logemann, 1994).

Indirect therapy

In indirect swallow therapy, the patient is given exercises to improve neuromuscular control over chewing and swallowing (Logemann, 1991). These are especially useful in patients who lack tongue control in any of the following ways (Logemann, 1983b):

• Lateralization of tongue during chewing.

• Elevation of tongue to hard palate.

• Cupping of tongue around bolus.

• Elevation of tongue against palate to hold bolus.

• Range of anterior to posterior movement.

• Coordination of anterior to posterior movement.

Exercises include (Logemann, 1991):

• Range of motion or resistance exercises for tongue and jaw.

• Tongue coordination and chewing exercises utilizing a material (such as gauze controlled by the clinician) with which the patient practices movements.

• Laryngeal adduction exercises.

• Bolus control exercises: the patient manipulates food or liquid in the mouth without actually swallowing it.

Direct therapy

Direct therapy attempts to change swallow physiology through special swallowing techniques, or medical or surgical management. Three swallowing maneuvers are commonly used to alter physiology during the swallow:

• Mendelsohn Maneuver: Patients are instructed to feel their larynx elevate during the swallow and attempt to prolong the period of maximal elevation. The rationale for this is that maximal cricopharyngeal opening occurs during maximal elevation of the larynx and hyoid (Logemann and Kahrilas, 1990). This technique is useful for patients with reduced hyolaryngeal movement (Logemann, 1991).

• Supraglottic Swallow: This technique is used to minimize aspiration. Patients voluntarily hold their breath before and during the swallow, thereby closing the true vocal folds. The patient then coughs when the swallow is completed, to clear any residual material from the pharynx.

• Super-Supraglottic Swallow: In a variation on the supraglottic swallow, patients apply increased effort to holding their breath before the swallow.

No medical (i.e., pharmaceutical) treatments are known to specifically improve oropharyngeal swallowing in neurologic patients. Drugs given to treat the underlying disease, however, may result in some improved swallow function.

Natural History of Dysphagia

Epidemiology

Dysphagia is not a condition that occurs in isolation; rather, it can be caused by many different disorders, including stroke, degenerative neurologic conditions, head and neck cancer, and head injury. It can also occur as a natural part of the aging process. Ideally, patients should be selected for testing on the basis of clinical signs and symptoms detected during physical exam or noninstrumented swallow exam. However, as will be discussed in the Results section of this report, few clinical signs and symptoms have been found that predict dysphagia-related morbidity such as aspiration or pneumonia. This means that the only currently available means of determining which patients are most likely to benefit from a dysphagia diagnosis and treatment program involves determining whether patients with any given disease or condition are particularly likely to become dysphagic. If so, it may then be possible to selectively choose patients for extended diagnostic testing based on the disease from which they suffer. Additionally, if most cases of dysphagia can be shown to result from a particular disease or condition, then treating patients suffering from this condition will alleviate most of the morbidity and mortality due to dysphagia. For these reasons, we discuss in this section not only the incidence of dysphagia overall, but also the incidence of dysphagia associated with a variety of neurologic conditions.

Arriving at the exact incidence and prevalence of dysphagia is, however, not possible. This is because dysphagia is not a single disease, but a cluster of symptoms, not all of which may be detected with current diagnostic technology (see the section entitled Diagnosis of Oropharyneal Dysphagia), and because of variations in definitions and interpretation of what constitutes dysphagia. It is therefore not possible to discuss the incidence or prevalence of dysphagia; rather, we can only discuss diagnosed occurrence because no current diagnostic test is 100 percent accurate. Also, precise data on the number of patients affected by dysphagia-related morbidity and mortality are not available. Because of this, the present section and the section on Burden of Illness contain numerous original calculations. Appendix H contains further details about these calculations.

Table 7. Range of Reported Epidemiological Findings (more...)

Table 7. Range of Reported Epidemiological Findings for Each Disease: Prevalence/Incidence of Disease Overall and Occurrence of Dysphagia Within Each Disease

	Overall prevalence/incidence (per 100,000)		Dysphagia (percent)
Condition	Whole population	Elderly population	Whole population	Elderly population
Overall	NA	NA	BSE: 12-17 VFSS: NR	BSE: 14 -34 VFSS: NR
Stroke	Annual incidence:145-290	Annual incidence: 65-74: 142-235 85+: 1,390 -2,574	BSE: 19-59 VFSS: 50-90	47
PD	Prevalence: 107 1 Annual incidence: 13-65	Prevalence: 790 to 2,250	BSE: 23-77 VFSS: 63-81	NR
AD	Prevalence:260 1	Prevalence: 65+: 2,634 -6,240 85+: 18,018-28,850	VFSS: 84 1	NR
MS	70-171 Prevalence:	265 1	Questionnaire: 13-33 1 BSE: NR VFSS: NR	NR
MND ALS	Prevalence: NR Annual incidence: 6 Prevalence: 3.12 1 Annual Incidence: 1.14 - 1.8	Annual incidence: 18-25	Questionnaire: 71 1 BSE/VFSS: 51.2 BSE/VFSS: 29	NR
PSP	Prevalence: 1.39 1 Incidence: 1.1 1	Prevalence: NR Incidence: 7 1	BSE/VFSS: 56 1	NR
HD	Prevalence: 1.9 1 Incidence: 0.2 1	NR	BSE: NR VFSS: 100 2	NR

Table 8. Epidemiological Data From the Published Literature: (more...)

Table 8. Epidemiological Data From the Published Literature: Neurologic Diseases and the Rate of Dysphagia Within Each

Disease	Prevalence (per 100,000)	Incidence (per 100,000)	Study	Reason	Diagnosed occurrence of dysphagia (%)	Study	Reason
Stroke	NA	145 289	Brown, Whisnant, Sicks et al., 1996; Modan and Wagener, 1992	Mayo Clinic Mayo Clinic seemed low: this provides an upper estimate	VFSS: 74.6 BSE: 41.7	Daniels, McAdam, Brailey et al., 1997; DePippo, Holas, and Reding, 1992	Median of VFSS studies Median of BSE studies
Parkinson's disease	106.9	13	Mayeux, Marder, Cote et al., 1995	Only number on general population that included elderly	VFSS: 69.1	Bushmann, Dobmeyer, Leeker et al., 1989; Fuh, Lee, Wang et al., 1997	Mean of 2 studies in which L-dopa was withheld
Alzheimer's disease	259.8	NR	Beard, Kokmen, Offord et al., 1991	Only published number	VFSS: 84	Horner, Alberts, Dawson et al., 1994	Only published number
Multiple sclerosis	170.8	NR	Wynn, Rodriguez, O'Fallon et al., 1990	Only number; Mayo Clinic	NR	NA	NA
Motor neuron disease	170.8	6.2	Lilienfeld, Sprafka, Pham et al., 1991	Only published number	51.2 (method not reported)	Leighton, Burton, Lund et al., 1994	Exam, not survey
Amyotropic lateral sclerosis	NR	1.8	McGuire, Longstreth, Koepsell et al., 1996	Exam, not survey	29 (method not reported)	Leighton, Burton, Lund et al., 1994	Only published number
Progressive supranuclear palsy	1.39	1.1	Golbe, Davis, Schoenberg et al., 1988;Bower, Maraganore, McDonnell et al., 1997	Only published number	VFSS: 55.6	Litvan, Sastry, and Sonies, 1997	Only published number
Huntington's disease	1.9	0.2	Kokmen, Ozekmekci, Beard et al., 1994	Only published number	VFSS: 100	Kagel and Leopold, 1992	Only published number

Table 7 shows a summary of all studies located on the epidemiology of major neurologic diseases and the rate of dysphagia within each of these diseases. These studies are further described in Evidence Table 1 through Evidence Table 9. Table 8 displays the best studies out of all those reviewed, according to our evaluation. When choosing the best studies, we took into consideration the research methods, populations included, and generalizability to the U.S. population as a whole. These methods are described in Appendix A. Unfortunately, many of these studies were chosen because they were the only published study that reported a particular statistic. A critical discussion of this literature is also found in Appendix A. The rates shown in Table 8 were used in the Burden of Illness section to determine the number of individuals in the United States affected by these diseases.

It is currently difficult to determine whether patients with any specific disease should automatically undergo extensive diagnostic testing for dysphagia. It appears, as shown in Table 8, that stroke, Alzheimer's disease, and Huntington's disease all have high enough rates of dysphagia to warrant diagnostic testing for all patients with these conditions. However, in the case of Huntington's and Alzheimer's diseases, only a single study on dysphagia was available for each; therefore, current data cannot be considered reliable. Further, these rates were determined by VFSS, which does not necessarily detect only cases of dysphagia that actually cause problems, but also potentially many cases that never will.

With stroke, there is a large body of literature (found in Evidence Table 3) on the diagnosed occurrence of dysphagia, and the reported rates of dysphagia range from 20 to 90 percent. Table 8 shows the median as determined by each diagnostic method. It is not possible to determine if these rates are accurate and what proportion of these cases are actually a risk for serious morbidity.

Because of these limitations in available evidence, we cannot make a recommendation about diagnosing patients for dysphagia solely on the basis of the neurologic disorder from which they suffer.

Burden of Illness

To determine the burden of illness on society resulting from dysphagia in these disorders, we need to determine the total number of people affected. An important question is how many of these patients would suffer serious morbidity or death if preventive treatment for dysphagia were not available? Dysphagia may lead to malnutrition because the patient is unable to take in adequate amounts of food. However, it is often not the dysphagia, per se, that leads to serious morbidity (pneumonia) and death, but aspiration. It is therefore necessary to determine what proportion of patients with dysphagia aspirate, what proportion of aspirators are expected to contract pneumonia if not treated, and what proportion of elderly patients with pneumonia die of the illness. Through these steps, we can determine the number of patients with dysphagia who will subsequently die of resulting morbidity without directed dysphagia management (if data directly making this link are not available). Knowing the raw number of individuals affected yearly by morbidity may help determine priorities in clinical management and insurance coverage. It would be interesting also to link the occurrence of dysphagia to the incidence of malnutrition, but the data do not exist in the literature to support this calculation.

It is important to point out that the figures we discuss in this section are not rates of morbidity or mortality that are found in the absence of treatment; rather, we have limited the discussion to pneumonia and to those studies in which there was no apparent dysphagia-oriented treatment. For this purpose, we have not limited the publication date of the studies we examined; historical data are particularly important in addressing these questions to find morbidity and mortality rates before current dysphagia treatment was implemented.

The focus of this discussion will be on stroke patients, as they make up the greatest proportion of neurologic dysphagia patients, and because most of the published literature focuses on stroke patients. We are currently unable to document the rate of aspiration or pneumonia in patients with neurologic diseases other than stroke, because the data do not exist. We can, however, calculate the number of patients whose dysphagia was caused by a certain disease and/or the rate of dysphagia caused by certain diseases.

This discussion will first review the literature on the rates of illness in stroke or patients with dysphagia; we then translate the rates discussed here and in the Epidemiology section above into total numbers of patients affected in the United States.

Malnutrition

There is little information on the incidence of malnutrition in elderly patients with dysphagia. Two groups of researchers have looked at malnutrition in the nursing home elderly and its possible link to swallowing and chewing problems (Keller, 1993; Keller, 1995; Thomas, Verdery, Gardner et al., 1991). One group found a significant association between dysphagia and malnutrition, while the other did not. Neither of these researchers looked specifically at any neurologic disorder.

Only one study has been published examining the rate of malnutrition in acute-care stroke patients (see Evidence Table 10). Davalos et al. (1996) followed 104 stroke patients who were admitted within 24 hours of stroke. At admission, 16.3 percent were malnourished. At one week, this increased to 26.4 percent of 91 survivors, and at two weeks, it increased to 35 percent in 43 patients remaining in the hospital (Davalos, Ricart, Gonzalez-Huix et al., 1996). This last figure is likely skewed because only the most debilitated patients would stay in the acute-care hospital for that long.

Malnutrition in dysphagia

This same study reported that 43 (41.3 percent) of these patients had swallowing problems at admission. While there was no difference in the rate of malnutrition between patients with dysphagia and those with a normal swallow at admission, at 1 week, patients with dysphagia were significantly more likely to be malnourished than patients with normal swallows (48.3 percent versus 13.6 percent) (Davalos, Ricart, Gonzalez-Huix et al., 1996). Standard treatment for patients with dysphagia in this acute-care facility was enteral feeding. While tube feeding is generally a last resort measure that is supposed to prevent malnutrition, this study suggests that it does not always achieve its goal.

Dehydration in dysphagia

One would expect patients with dysphagia to have an increased risk of dehydration because of their inability to swallow thin liquids safely. Dehydration is generally measured through blood tests such as measurement of hematocrit or blood urea nitrogen/creatine ratio. Very few studies have reported these test results for patients with dysphagia. Smithard et al. (1996) reported no changes in hydration status regardless of the ability to swallow in acute-care stroke patients (Smithard, O'Neill, Parks et al., 1996). However, these patients were only followed for 1 week post-stroke, and therefore these findings do not address long-term dehydration risks in this population.

One study (Barer, 1989) examined the occurrence of hydration-related blood measurements comparing three groups of patients with different levels of swallow disability. Blood urea nitrogen measurements were significantly higher at day 8 post-stroke for patients on NG tubes compared with those with more minor swallowing problems. However, blood urea nitrogen changes examined in isolation do not provide definitive information about hydration status; they must be examined in relation to creatine levels.

Two studies have examined the occurrence of dehydration in aspirators versus nonaspirators (Holas, DePippo, and Reding, 1994; Schmidt, Holas, Halvorson et al., 1994). Neither found any significant difference in the occurrence of dehydration between these two groups of patients.

In conclusion, the current data examining the relationship between dysphagia and dehydration are sparse, and do not suggest a significant causal relationship. However, more research in this area is needed before a firm conclusion can be drawn.

Aspiration in dysphagia

Not all patients with dysphagia suffer from aspiration, but many do (see Evidence Table 11). Three studies specifically examined the proportion of stroke patients with dysphagia who present with aspiration. These three studies are generally in agreement that stroke dysphagics aspirate; 46.3 percent (Kidd, Lawson, Nesbitt et al., 1993); 53.5 percent (Holas, DePippo, and Reding, 1994); and 43.2 percent (Daniels, McAdam, Brailey et al., 1997)

Aspiration in stroke

Five studies (Chen, Ott, Peele et al., 1990; Daniels, Brailey, Priestly et al., 1998; Daniels, McAdam, Brailey et al., 1997; Kidd, Lawson, Nesbitt et al., 1993; Smithard, O'Neill, Parks et al., 1996) directly examined the occurrence of aspiration in a stroke population within 7 days of the event (without consideration of general dysphagia) (see Evidence Table 12). Their estimates ranged from 21.3 percent (Smithard, O'Neill, Parks et al., 1996) to 38.2 percent (Daniels, Brailey, Priestly et al., 1998). Diagnosis of aspiration will depend upon bolus size and consistency tested during VF. The study reporting the highest rate was the only one to recruit patients (Daniels, Brailey, Priestly et al., 1998); patients sensing a swallowing problem may have been more likely to volunteer for a study. Smithard et al. (1996), reporting the lowest rate with 21.3 percent, did not confirm stroke diagnoses with computed tomography (CT) or MRI, only through clinical assessment (Smithard, O'Neill, Parks et al., 1996); it is possible that some cases are not truly acute stroke, thus potentially lowering any estimation of morbidity. As a result, the best estimate is probably a median of all hospitalized strokes, 33.5 percent, which is the average of the two median numbers from Daniels et al. (1997) and Kidd et al. (1993).

If we assume, as discussed in the previous section, that approximately half of stroke dysphagics experience aspiration, then, from the numbers calculated above, approximately 42 to 76 percent of the stroke patients in these three studies would be expected to be dysphagic. These numbers are in rough accordance with rates discussed in Appendix A, in which a range of 46 to 90 percent of hospitalized CVA patients were diagnosed with dysphagia on VF.

In conclusion, studies indicate that approximately 43 to 55 percent of stroke patients with dysphagia experience aspiration; 21 to 38 percent of all stroke patients experience aspiration.

Pneumonia

Pneumonia is a fairly common occurrence after a stroke, especially if there is no preventative treatment. Four studies have examined the rate of pneumonia after a stroke (Dobkin, 1987; Haerer and Smith, 1974; Odderson and McKenna, 1993; Young and Durant-Jones, 1990) (see Evidence Table 13); rates range from 1.5 to 13.0 percent. Because we have calculated (in Appendix D) that most pneumonia resulting from a stroke occurs within the first week after the event, even though the followup for each of these studies was different, the rates are comparable. The lowest figure comes from a stroke rehabilitation unit (Dobkin, 1987), the highest from an acute-care hospital that included only aspiration pneumonia (Young and Durant-Jones, 1990). Both were retrospective case series that only followed patients for their length of stay. Two other studies also followed stroke patients in an acute-care hospital and found pneumonia rates of 6.7 percent (Odderson and McKenna, 1993) and 8.8 percent (Haerer and Smith, 1974). There is no obvious reason for the differences found in the three acute-care studies, although Odderson et al. (1993) only included nonhemorrhagic strokes (and found the lower of the two rates), while Young et al. (1990) did not have such an inclusion criterion. It may be the case that patients who suffer nonhemorrhagic strokes are less disabled and, therefore, less susceptible to pneumonia. Haerer and Smith (1974) included only cases of acute pneumonia, thus possibly lowering their reported rates slightly.

The fact that the highest rate from these four studies included only aspiration-specific pneumonia suggests that most pneumonia after stroke is aspiration related. This supposition may be confounded by the fact that these patients were followed longer than those in other studies. This may not be specifically aspiration resulting from oropharyngeal dysphagia but may also include many patients suffering from esophageal reflux. However, this does appear to be the most reliable figure (13.0 percent) for acute hospital stroke patients (Young and Durant-Jones, 1990).

Aspiration resulting in pneumonia: Relative risk compared with dysphagics and nonaspirators

Aspiration, a common side effect of stroke, can put individuals at risk for pneumonia. Five studies examined the occurrence of all types of pneumonia in patients experiencing aspiration (Croghan, Burke, Caplan et al., 1994; Holas, DePippo, and Reding, 1994; Schmidt, Holas, Halvorson et al., 1994; Smithard, O'Neill, Parks et al., 1996; Teasell, McRae, Marchuk et al., 1996) (see Evidence Table 14). The rates of pneumonia ranged from 11.9 percent (Teasell, McRae, Marchuk et al., 1996) to 50 percent (Croghan, Burke, Caplan et al., 1994). The highest rate came from a nursing home, the lowest from a stroke rehab unit, reflecting the relative debilitation of patients in these different care settings.

However, almost everyone experiences aspiration at one time or another, and, therefore, it is not a given that pneumonia will result. It is therefore important to determine whether aspiration is significantly associated with pneumonia, through a comparison of those who experience aspiration to dysphagics without aspiration and other nonaspirators. Evidence Table 15 displays those studies providing such comparisons. Because these are case series data, they cannot be subject to a meta-analysis; we can, however, pool the data in such a way that we can approximate the trend in pneumonia incidence in those with and without dysphagia.

Evidence Table 15 displays data from patients in all care settings, including acute-care, rehabilitative, and long-term care. Patients in these different care settings will have different risks of contracting pneumonia because of general differences in health status. We therefore pool the data separately for these three settings.

Table 9. Summary Comparison Among Different Patient (more...)

Table 9. Summary Comparison Among Different Patient Groups for Risk of Pneumonia

Patient population	Acute-care		Stroke rehabilitation		Long-term care
Condition	# with pneumonia	Total with condition	# with pneumonia	Total with condition	# with pneumonia	Total with condition
Aspirators	35	96	23	171	24	42
Proportion		0.365		0.135		0.571
Nonaspirators	17	142	13	517	6	18
Proportion		0.120		0.025		0.333
Dysphagic nonaspirators	32	135	3	75	8	16
Proportion		0.237		0.040		0.500
All dysphagics	56	196	21	220	32	58
Proportion		0.286		0.095		0.552

Table 9 provides a summary comparison from several studies, shown in Evidence Tables 14 through 18, among aspirators, nonaspirators, nonaspirating dysphagics, and all patients with dysphagia.

The most informative comparison in the table above is between those who aspirate and those who have dysphagia but do not aspirate; any difference between these two groups will be the net additional risk of contracting pneumonia due to the presence of aspiration. There appears to be a trend for acute-care patients that aspirators have a greater risk of contracting pneumonia than the other three groups of patients. However, because these are case series data, we cannot determine whether there is a significant difference between aspirators (36.5 percent) and nonaspirating dysphagics (23.7 percent). This difference is very strong in the rehabilitation group (13.5 versus 4.0 percent).

This difference is not apparent in the nursing home group. While 57.1 percent of aspirators contract pneumonia, 50.0 percent of nonaspirating dysphagics do. It is unlikely that this is a significant difference. It may be the case that the elderly in nursing homes are so debilitated that any such disability can lead to greater risk of illness.

Several of the studies shown in Evidence Tables 14 through 18 found a significant association between the presence of aspiration and the development of subsequent pneumonia (Holas, DePippo, and Reding, 1994; Kidd, Lawson, Nesbitt et al., 1993; Langmore, Terpenning, Schork et al., 1998; Reynolds, Gilbert, Good et al., 1998; Schmidt, Holas, Halvorson et al., 1994). One study found that silent aspiration was significantly more predictive of subsequent pneumonia than symptomatic aspiration (Holas, DePippo, and Reding, 1994). However, many of these studies also found a significant relationship between dysphagia and subsequent pneumonia (DePippo, Holas, and Reding, 1994; Kidd, Lawson, Nesbitt et al., 1993; Langmore, Terpenning, Schork et al., 1998; Reynolds, Gilbert, Good et al., 1998; Smithard, O'Neill, Parks et al., 1996). In fact, one study (DePippo, Holas, and Reding, 1994) found that the presence of cough alone detected on a bedside screening test (Burke Dysphagia Screening Test) predicted 100 percent of pneumonia cases (however, with a high false positive rate). It is unclear how much of the relationship between dysphagia and pneumonia is caused by dysphagic aspirators, because aspiration was not controlled for in the researchers' statistical analyses. Interestingly, two other studies found no relationship between aspiration and pneumonia/chest infection (Croghan, Burke, Caplan et al., 1994; Smithard, O'Neill, Parks et al., 1996) (although some of this may have been due to lack of statistical power).

In conclusion, while there appears to be a trend for aspirators, especially stroke patients in acute-care, to be more likely to contract pneumonia, it is currently not possible to determine if this is a statistically significant increase in risk. Statistical analyses in the published literature are currently equivocal about these possible differences.

Pneumonia mortality

Pneumonia can be successfully treated with antibiotics. However, in weakened elderly people, pneumonia is a leading cause of death; the Centers for Disease Control and Prevention (CDC) reported that in 1995 alone, 74,297 elderly people died of pneumonia or influenza, the fifth leading cause of death in the elderly (Department of Health & Human Services, 1997).

Pneumonia-specific mortality in elderly pneumonia patients

Three studies have examined the pneumonia-specific mortality rate among the elderly with pneumonia (Riquelme, Torres, El-Ebiary et al., 1996; Thompson, Hall, Szpiech et al., 1997; Venkatesan, Gladman, Macfarlane et al., 1990) (see Evidence Table 20). These estimates range from 13.7 percent (Venkatesan, Gladman, Macfarlane et al., 1990) to 39.5 percent (Thompson, Hall, Szpiech et al., 1997). Both the lowest and highest numbers were reported from a hospital on patients with similar mean age; however, the higher number included only nursing home ward patients. Both included all causes of pneumonia. A median number reported by Riquelme et al. (1996) provides probably the best estimate, a 19.8 percent death rate after 30 or more days of hospitalization.

Table 12. Dysphagia-Related Morbidity and Mortality (more...)

Table 12. Dysphagia-Related Morbidity and Mortality in Stroke Patients

Disease/disorder	Number or rate	Source	Comments
Stroke, annual incidence
1. All ages
Total population	265,284,000	U.S. Department of Commerce, 1997
Rate of stroke, low estimate	145/100,000	Brown, Whisnant, Sicks et al., 1996	Mayo Clinic
Rate of stroke, high estimate	290/100,000	Modan and Wagener, 1992	National Hospital Discharge Survey
Total affected	384,662
	768,528	High estimate
2. 65+
Men, total population:	13,880,892	Centers for Disease Control and Prevention, 1998
Rate of stroke	953/100,000	Barker and Mullooly, 1997
Total affected, male	132,285
Women, total population	19,979,990	Centers for Disease Control and Prevention, 1998
Rate of stroke	736/100,000	Barker and Mullooly, 1997
Total affected, female	147,053
Total affected, 65+	279,338	From above
All-case dysphagia in elderly population
Total population, age 60+	43,859,973	Centers for Disease Control and Prevention, 1998a
Rate of dysphagia	14.20%	Baum and Bodner, 1983	Only study
Total affected	6,228,116
Dysphagia in stroke
1. Total stroke population, low estimate	384,662	Calculated above
Total stroke population, high estimate	768,528	Calculated above
Rate of dysphagia, VFSS	74.60%	Daniels, McAdam, Brailey et al., 1997	Median of all studies, hospitalized
Total affected	286,958	Low
	573,322	High
Rate of dysphagia, BSE	41.70%	Kidd, Lawson, Nesbitt et al., 1993	Median of all hospitalized BSE numbers taken within 5 days of stroke
Total affected	160,404	Low
	320,476	High
2. Total elderly stroke population, age 65+	279,338	From above
Rate of dysphagia, BSE, age 70+	35.50%	Barer, 1989	1 day post-stroke
Total affected	99,165
Malnutrition in stroke
Total stroke population	384,662	Brown, Whisnant, Sicks et al., 1996	Mayo Clinic + CDC wonder
	768,528	Modan and Wagener, 1992	Modan, national, hospital discharge survey + CDC wonder
Rate of malnutrition	26.4%	Davalos, Ricart, Gonzalez-Huix et al., 1996	At 1 week
Total affected	101,551
	202,891
Malnutrition in dysphagia
Total dysphagic stroke	160,404	Low
	320,476	High
Rate of malnutrition	48.4%	Davalos, Ricart, Gonzalez-Huix et al., 1996
Total affected	77,636
	155,110
Aspiration in stroke
Total stroke	384,662	Low estimate, from above
	768,528	High estimate, from above
Rate of aspiration	33.50%	Daniels, McAdam, Brailey et al., 1997; Kidd, Lawson, Nesbitt et al., 1993	Average of Daniels,1997 and Kidd, 1993, median number for acute hospital strokes <5 days poststroke assessment
Total affected	128,862
	257,457
Pneumonia, annual incidence
Total population	265,284,000	U.S. Department of Commerce, 1997
Rate of pneumonia	1,600/100,000	Adams and Marano, 1995
Total affected	4,244,544
Total population 65+	33,860,882	Centers for Disease Control and Prevention, 1998c
Rate of pneumonia	3,032/100,000	Houston, Silverstein, and Suman, 1995	Mayo Clinic
Total affected	1,026,662
Pneumonia in stroke
Total stroke	384,662	Low, from above
	768,528	High, from above
Rate of pneumonia	13.00%	Young and Durant-Jones, 1990	Acute-care hospital, aspiration pneumonia only, LOS, no dysphagia program
Total affected	50,006
	99,909
Dysphagia to pneumonia in stroke
1. Total dysphagic stroke, BSE	160,404	Low, from above
	320,476	High, from above
Rate of pneumonia	14.30%	Nilsson, Ekberg, Olsson et al., 1998	No specific swallow treatment, acute hospital, LOS, BSE
Total affected	22,938	Low
	45,828	High
2. Total dysphagic stroke, VFSS	286,958	Low
	573,322	High
Rate of pneumonia	14.30%	Nilsson, Ekberg, Olsson et al., 1998	No specific swallow treatment, acute hospital, LOS, BSE
Total affected	41,035	Low
	81,985	High
3. Total dysphagic stroke, elderly	99,165
Rate of pneumonia	14.30%	Nilsson, Ekberg, Olsson et al., 1998	No specific swallow treatment, acute hospital, LOS
Total affected	14,181
Nondysphagia to pneumonia
Total nondysphagic in stroke, BSE	224,258	Low, from above
	448,052	High, from above
Nondysphagia to pneumonia	0.198	Reynolds, Gilbert, Good et al., 1998	BSE
Total affected	44,403	Low
	88.714	High
Aspiration to pneumonia in stroke
Total aspirating stroke	128,862	Low, from above
	257,457	High, from above
Aspiration to pneumonia	19.20%	Schmidt, Holas, Halvorson et al., 1994	Acute-care hospital, no specified dysphagia treatment
Total affected	24,741	Low
	49,432	High
Proportion of all CVA pneumonias occurring in dysphagic CVAs
Total pneumonia in CVA	50,006
	99,909
Dysphagia to pneumonia	22,938
	45,828
Proportion of pneumonias that occur after dysphagia	0.459
Dysphagia to pneumonia, BSE	22,938	Low
	45,828	High
Nondysphagia to pneumonia, BSE	44,403	Low
	88,714	High
Proportion of CVA pneumonias that result after dysphagia	0.341
Proportion of all CVA pneumonias occurring in aspirating CVA
Aspiration to pneumonia	24,741
	49,432
Total CVA pneumonia	50,006
	99,909
Proportion of CVA pneumonias that result after aspiration	0.495
Pneumonia mortality
1. Total pneumonia incidence, all ages	4,244,544	Adams and Marano, 1995; U.S. Department of Commerce, 1997
Total pneumonia deaths, all ages	81,972	Centers for Disease Control and Prevention (CDC), 1998a
Rate of death	0.019
2. Total pneumoniaincidence, 65+	1,026,662		Houston 1995 + CDC
Total deaths	74,297	Department of Health & Human Services, 1997	Health, U.S., 1996-7, for 1995, age 65+, Table 34
Rate of death	0.072		Calculated from above
Clinical literature on neurogenic pneumonia indicates 19.8 percent
3. Pneumonia hospitaladmissions, age 65+	690,000	U.S. Department of Health & Human Services, 1997	Health, U.S., 1996-7, Table 89
Total deaths	71,000	Centers for Disease Control and Prevention, 1998b	National Hospital Discharge Survey, 1995
Rate of death, hospitalized	0.103		Calculated from above
Dysphagia to pneumonia death
Total dysphagic pneumonia	22,938
	45,828
Rate of pneumonia death	19.8%	Riquelme, Torres, El-Ebiary et al., 1996
Total number of deaths	4,542
	9,074
Total dysphagic pneumonia, age 65+	14,181
Rate of pneumonia death	19.8%	Riquelme, Torres, El-Ebiary et al., 1996
Total number of deaths	2,808
Total dysphagic stroke, BSE	160,404	Low
	320,476	High
Proportion of dysphagic strokes leading to pneumonia deaths	0.028		Only diagnosed on BSE, so doesn't include silent aspirators who are at particular risk for pneumonia
Aspiration to pneumonia death
Total aspirating pneumonia	24,741
	49,432
Rate of pneumonia death	19.8%	Riquelme, Torres, El-Ebiary et al., 1996
Total number of deaths	4,899
	9,787
	128,862
	257,457
Proportion of aspiration resulting in pneumonia death	0.038
Nondysphagia to pneumonia death
Total nondysphagic pneumonia	44,403
	88,714
Rate of pneumonia death	19.8 percent	Riquelme, Torres, El-Ebiary et al., 1996
Total number of deaths	8,792
	17,565
Nondysphagic strokes, BSE	224,258	Low, from above
	448,052	High, from above
Proportion of nondysphagic strokes leading to pneumonia death	0.039

This is at odds with an estimate determined from vital statistics data (shown in Table 12), in which the estimate was 7.2 percent. The vital statistics data rely on reporting hospitals giving data on all pneumonia cases and related deaths; a single clinical study relies on the quality of care at a single institution. However, the national data on mortality include deaths occurring in all pneumonia cases, not just those resulting from neurologic disease. Therefore, it is prudent to use the Riquelme data for the purposes of this report.

Pneumonia-specific mortality in patients with aspiration

Three studies have reported on the pneumonia-specific mortality rate among patients with aspiration (Croghan, Burke, Caplan et al., 1994; Feinberg, Knebl, and Tully, 1996; Pick, McDonald, Bennett et al., 1996) (see Evidence Table 19). The rates range from 18.8 percent (Pick, McDonald, Bennett et al., 1996) to 31.8 percent (Croghan, Burke, Caplan et al., 1994). All three studies examined nursing home patients in particular. Both the lowest and highest rates came from studies with a 1-year followup. However, the lower rate determined aspiration on the basis of nurse observation rather than instrumented exam and, therefore, misses cases of silent aspiration. Both of the other two numbers were determined from VF on patients of mixed etiology. The most reliable rate comes from Feinberg et al. (1996), because they evaluated 152 patients versus Croghan's 42, with an estimated pneumonia-specific mortality rate for patients with aspiration of 20.9 percent after a mean followup of 29 months (Feinberg, Knebl, and Tully, 1996). There is no figure for the elderly population as a whole, and, therefore, an estimate cannot be made for stroke patients in acute-care.

All-cause mortality in pneumonia patients

Pneumonia may contribute to mortality resulting from other causes through its weakening of the immune system. Nine studies have reported on the overall death rates of patients with pneumonia (see Evidence Table 21). The overall mortality rates ranged from 5.8 percent (Beck-Sague, Villarino, Giuliano et al., 1994) to 53.5 percent (Thompson, Hall, Szpiech et al., 1997). These mortality rates are not specific to aspiration pneumonia; only one study (Jones, 1993) reported mortality rates specific to aspiration pneumonia, a rate of 47.5 percent, the second highest rate reported by all studies in Evidence Table 28. It has not been reported in the literature what proportion of elderly pneumonia patients suffer from aspiration pneumonia specifically. These studies are also largely not specific to stroke patients or even general neurologic patients.

Both the lowest and highest rates were found in nursing home patients, normally the frailest population represented. Beck-Sague et al. (1994) examined all nursing home residents with pneumonia, both those treated at the nursing home and those sent to acute-care (Beck-Sague, Villarino, Giuliano et al., 1994). Thompson et al. (1997) examined only those admitted to an acute-care hospital, which may account for some of the difference (Thompson, Hall, Szpiech et al., 1997). On the other hand, Houston, Silverstein, and Suman (1995) compared the mortality rates of nursing home residents treated in the home versus those treated in the hospital and found a higher mortality rate in those treated at the home (46 percent versus 35.5 percent), which would suggest that pneumonia in these patients should be taken more seriously, and more patients should be hospitalized.

Hospitalized elderly pneumonia patients demonstrate an overall mortality rate of 11 to 48 percent in a 30-day period (Garb, Brown, Garb et al., 1978; Houston, Silverstein, and Suman, 1995; Jones, 1993; Marrie, Durant, and Kwan, 1986; Marston, Plouffe, File et al., 1997; Reynolds, Gilbert, Good et al., 1998; Riquelme, Torres, El-Ebiary et al., 1996); length of followup time does not seem to positively correlate to mortality rate in these studies, nor does age. Among nursing home patients specifically, hospitalized patients demonstrated a mortality rate of 36 to 54 percent (Houston, Silverstein, and Suman, 1995; Marrie, Durant, and Kwan, 1986; Thompson, Hall, Szpiech et al., 1997), substantially higher than the general pneumonia population. Because many nursing home patients with pneumonia are treated in the nursing home, the nursing home elderly who are admitted to the hospital may represent the sickest of the sick.

The most reliable figure for the overall elderly population specific to those with neurologic disease is the only one specific to stroke; with only 21 patients, there is likely to be error, but Reynolds, Gilbert, Good et al. (1998) reported an all-cause mortality rate of 23.8 percent. This figure seems to correspond well with the pneumonia-specific mortality rate reported above of 19.8 percent (Riquelme, Torres, El-Ebiary et al., 1996).

Calculation of Burden of Illness

Table 10. Dysphagia-Related Morbidity and Mortality (more...)

Table 10. Dysphagia-Related Morbidity and Mortality in Stroke

Disease/disorder	Number affected/probability of occurrence
Stroke: individuals affected annually	384,662-768,528
Total affected, age 65 +	279,338
Pneumonia	4,244,544
Pneumonia, 65	1,026,662
Pneumonia in stroke	50,006-99,909
Pneumonia mortality	0.019
Total pneumonia rate of death, age 65+	0.198
Rate of death, hospitalized	0.103
BSE findings
All-cause dysphagia in elderly population	6,228,116
Dysphagia in stroke	160,404-320,476
Dysphagia in the elderly stroke patients	99,165
Malnutrition in stroke	101,551-202,891
Malnutrition in dysphagic stroke	77,636-155,110
Dysphagia resulting in pneumonia in stroke	22,938-45,828
Pneumonia in neurogenic dysphagic elderly	14,181
Nondysphagia resulting in pneumonia in stroke	44,403-88,714
Proportion of all CVA pneumonias occurring in dysphagic CVAs	0.34-0.459
Dysphagia resulting in pneumonia death	4,542-9,074
Dysphagia resulting in pneumonia death, age 65+	2,808
Nondysphagia resulting in pneumonia death	8,792-17,565
Proportion of dysphagic strokes leading to pneumonia death	0.028
Proportion of nondysphagic strokes leading to pneumonia Deaths	0.039
VFSS findings
Dysphagia in stroke	286,958-573,322
Dysphagia resulting in pneumonia in stroke	41,035-81,985
Aspiration in stroke	128,862-257,457
Aspiration resulting in pneumonia in stroke	27,834-55,611
Aspiration resulting in pneumonia death	5,511-11,011
Proportion of all CVA pneumonias occuring in aspirating CVAs	0.557
Proportion of aspiration resulting in pneumonia death	0.043

Table 11. Number of Patients Affected by Neurologic (more...)

Table 11. Number of Patients Affected by Neurologic Disease Other Than Stroke

Disease/disorder (annual incidence)	Number affected
Parkinson's disease	34,487
Dysphagia in Parkinson's	23,830
Motor neuron disease	15,917
Dysphagia in MND	8,150
Amyotropic lateral sclerosis	4,775
Dysphagia in ALS	1,385
Progressive supranuclear palsy	2,918
Dysphagia in PSP	1,622
Huntington's disease	531
Dysphagia in HD	531
Total burden of illness of dysphagia in neurologic diseases other than stroke	51,435
Total burden of illness, all neurologic diseases including stroke, per year	338,393-624,757

The total number of patients afflicted with each of the conditions, morbidities, and mortalities we have discussed in the Epidemiology and Burden of Illness sections of this report are shown in Table 10 and Table 11. Figures in these tables are based on our calculations using data from several studies, including government census figures. For population statistics, the CDC website (http://www.cdc.gov/), on which the most up-to-date population figures are stored, was accessed. Population statistics discussed here are from 1996, unless otherwise noted. The rates taken from the epidemiological literature are shown in Table 8. A discussion of why these rates were chosen is found in Appendices A and B. Table 12 shows burden-of-illness calculations specifically for a stroke population; addresses other neurologic disorders. These tables contain data from only those neurologic disorders for which there were reliable dysphagia statistics. Therefore, multiple sclerosis and Alzheimer's disease are not included.

A full discussion of these calculations is found in Appendix B. A summary of findings is found in Table 10 and Table 11.

Electronic Database Searches

Several electronic databases were searched with the intention of retrieving both clinical trials for analysis, and review articles for background knowledge. Specific search strategies are shown in Appendix C. The databases searched and general keywords used are listed below:

American Speech-Language-Hearing Association (ASHA) Database (1945 to 1996)

CANCERLIT (through September 4, 1997)

CATLINE (through August 25, 1997)

The Cochrane Database of Systemic Reviews (through 1998, Issue 3)

The Cochrane Registry of Clinical Trials (through 1998, Issue 3)

The Cochrane Review Methodology Database (through 1998, Issue 3)

Combined Health Information Database (CHID) (through July 29, 1998)

Current Contents (through July 1998)

The Database of Reviews of Effectiveness (Cochrane Library) (through 1998, Issue 3)

DIRLINE (through November 1997)

ECRI Health Devices Alerts (1977 through July 1998)

ECRI Health Devices Sourcebase (through July 1998)

ECRI Healthcare Standards Database (1975 through July 1998)

EMBASE (Excerpta Medica) (1974 through February 6, 1998)

Health Care Financing Administration Coverage (HCFA) Manuals CD-ROM (through July 1998)

HealthSTAR (Health Services, Technology, Administration, and Research) (1990 through May 20, 1998)

Incidence and Prevalence Database (1988 through August 25, 1997)

ECRI International Health Technology Assessment (IHTA) Database (1990 through July 1998)

MEDLINE (1964 through July 24, 1998)

NIH Grants Database (through December 12, 1997)

Nursing and Allied Health (NAHL) (1988 through April 30, 1998)

PsycINFO (1967 through September 10, 1997)

Sociological Abstracts (1963 through November 1997)

The search strategies employed a number of free-text keywords, as well as controlled vocabulary terms, including but not limited to:

Diagnostic modalities: barium sulfate; barium swallow; barium; fluoroscopy; cineradiology; videofluoroscopy; FEES (fiberoptic endoscopic evaluation of swallowing); ESE (endoscopic swallowing evaluation); FEED (fiberoptic endoscopic evaluation of dysphagia); FEESST (fiberoptic endoscopic evaluation of swallowing with sensory testing); VEED (videoendoscopic evaluation of dysphagia)

Disorder: deglutition disorders (exploded); deglutition (exploded); dysphagia; swallowing

Epidemiology: epidemiology; research design; epidemiologic study characteristics; epidemiologic methods; epidemiologic studies; evaluation studies; incidence; prevalence; statistics and numbers; aspiration pneumonia; neurodegenerative diseases (exploded); Parkinson disease; silent aspiration; stroke

Etiology: aging; Alzheimer's disease; dementia; multiple sclerosis; Parkinson disease; stroke

Miscellaneous: cachexia; wasting; weight loss; quality of life; QOL; life satisfaction; satisfaction

Treatment: speech therapy; speech-language pathology; electrical stimulation; enteral nutrition; intubation, gastrointestinal; nasogastric; nasointestinal; NG; percutaneous endoscopic gastrostomy; PEG tube feeding; rehabilitation, geriatric; rehabilitation; speech and language; rehabilitation, patients; elder care; mobile health units

In general, the searches were restricted to human. Case reports were excluded.

World Wide Web Searches

Searches of the World Wide Web were also conducted using various search engines including, but not limited to, AltaVista, Hotbot, and Yahoo. Pertinent websites include:

Hand Searches of Journal and Nonjournal Literature

In addition to searching Current Contents - Clinical Medicine on a weekly basis - more than 1,600 journals and supplements maintained in ECRI's collections were routinely reviewed. A hand search of the Cumulated Index Medicus (1960-1964) was conducted for the terms deglutition: deglutition disorders, speech therapy, and speech disorders. Nonjournal publications and conference proceedings from professional organizations, private agencies, and government agencies were also screened.

Other Mechanisms

Other mechanisms were used to retrieve additional relevant information, including review of bibliographies/reference lists from peer-reviewed and gray literature. (Gray literature includes reports, studies, etc., produced by local government agencies, private organizations, educational facilities, corporations, etc., that do not commonly appear in the published peer-reviewed journal literature.) Published and unpublished information was also solicited from a panel of experts in the field.

The Evidence Model as a Clinical Pathway

The evidence model approximates the clinical pathway followed by patients. Understanding of this pathway is critical to understanding not only the relationship(s) between the kinds of available evidence, but also for understanding the kinds of future research that might be conducted. Thus, the purpose of this section is to elaborate on the use of the evidence model as an illustration of the clinical pathway.

Use of the Evidence Model to Answer the Four Main Questions

Table 14. Levels of the Key Questions of this Report (more...)

Table 14. Levels of the Key Questions of this Report

Level of evidence	Type of question(s) posed by level	Examples of experimental end points
Level 6	Is a dysphagia management program cost-effective?	Cost-effectiveness
Level 5	Is a dysphagia management program efficacious?	Comparison of morbidity and mortality combined with demonstration of diagnostic efficacy
Level 4, Type 1 Level 4, Type 2	Are dysphagia diagnostic methods efficacious? Are dysphagia treatment methods efficacious?	Percentage of patients given specific treatments (without regard to treatment efficacy) Morbidity and mortality (without regard to diagnostic efficacy)
Level 3	Do dysphagia diagnostic methods change patient management?	Percentage of patients whose management was changed
Level 2	What is the performance of the various bedside exams, MBS, FEES, and FEESST?	Sensitivity, specificity, positive and negative predictive values
Level 1	What is the resolution of the MBS, FEES, and FEESST?	Not relevant to the present report

This hierarchy contains six levels. The lowest level (Level 1) addresses the technical efficacy of a technology. This level of evidence is illustrated by the situation often found in radiology in which it is determined that an imaging device can resolve a certain number of lines per centimeter.

Level 2 evidence provides information related to a more general question: how well a given diagnostic can determine whether a patient has (or does not have) a certain disease or condition. Evidence at this level, unlike that at the first level, provides the sensitivity and specificity of the diagnostic.

Level 3 evidence has even greater generality: it addresses how physicians use the results of a diagnostic test. For example, a study providing evidence at this level could measure the percentage of patients whose management was changed as a result of the diagnostic test. The precedence this evidence takes over Level 2 evidence becomes apparent when we consider that even a perfectly sensitive and specific diagnostic test would not benefit patients if its results were not used in patient management. This level of evidence, however, does not specify the treatments given to patients.

At Level 4, there are two types of evidence. The first type is evidence that examines the performance of a diagnostic without regard to treatment efficacy. The second type is evidence that examines the efficacy of treatments without regard to the performance of the diagnostic test. Neither type of evidence is as general as evidence about the efficacy of a program of diagnosis and treatment, and both are more meaningful than the previous levels, which do not provide much information about patient outcomes. An example of the first type of Level 4 evidence is a hypothetical study that specifies that a certain percentage of patients given a MBS exam were instructed to tuck their chins when they swallowed and that another percentage of patients were instructed to turn their heads to one side when they swallowed. Even though the treatments are specified by this type of Level 4 evidence, the efficacy of these treatments (e.g., how many patients died of aspiration pneumonia) is not considered. Rather, this level of evidence specifies the percentages of patients given treatments thought to be effective. Therefore, this type of evidence primarily gives information about the performance of the diagnostic test. An example of the second type of Level 4 evidence would be a study that compared the effectiveness of two treatments given to patients who all had the same results on a diagnostic test. This type of evidence primarily provides information about treatment efficacy and does not consider whether patients are receiving the best possible diagnostic test. As such, information about the outcomes of patients given diagnostic tests thought to be effective is provided.

Level 5 evidence addresses both the efficacy of the diagnostic test and the efficacy of treatment. Evidence at this level addresses whether patients are receiving both the best available diagnostic and the best available treatment. In other words, the efficacy of dysphagia programs is addressed. As such, information that is of greater interest to patients and society than that provided by the evidence in Levels 1 through 4 is provided.

The highest level of evidence is provided by Level 6, which addresses the cost-effectiveness of any given program. At one extreme are programs that might be highly effective but so expensive that society's finite amount of dollars would save more lives were they spent elsewhere. At the other extreme are inexpensive programs that are not particularly efficacious. Such programs give rise to debates about whether they should be implemented because of the remote possibility they might benefit some patients.

How this evidence hierarchy is used to rank the major questions of this evidence report is explained below, in the discussion of the evidence model links of interest.

Outcomes: Dysphagia Diagnosis Treatment Program for Prevention of Serious Morbidities

The primary reason to use any diagnostic test is to improve patient management and, thus, patient outcomes. Several dysphagia-related outcomes are important, including subjective patient quality of life (QOL) improvement, prevention of malnutrition, prevention of dehydration, prevention of pneumonia, and prevention of death.

We could not find sufficient suitable data for formal evaluation of QOL improvement due to dysphagia diagnosis, so this outcome is not evaluated in relation to our first question.

One of the most meaningful endpoints to patients is prevention of death. Again, there was little evidence on this in the literature, particularly on death due to malnutrition (if, despite the availability of feeding tubes, it occurs). We were, however, able to estimate potential death rates due to dysphagia, and these are addressed in the Epidemiology section of this report.

Pneumonia

For the purposes of this question, then, prevention of pneumonia is the primary endpoint, an endpoint that is reported relatively often in the literature. The advantage of this endpoint is that it allows us to ask whether use of any particular diagnostic leads to lower pneumonia rates. In addressing this question, we focused on stroke patients, either in acute care, rehabilitation units, or nursing homes, inasmuch as this reflects the emphasis of the literature.

Aspiration pneumonia, in particular, is the specific diagnosis of interest in a population with dysphagia, as these patients are at risk of aspirating food or liquid into their lungs. The definition of aspiration pneumonia is not clear in the medical literature, and it is often difficult to diagnose without the use of invasive procedures. It is therefore questionable in many studies whether patients diagnosed as having aspiration pneumonia were actually definitively diagnosed as such. Adding to this confusion is the fact that the causative relationship between aspiration and pneumonia has been demonstrated in the literature as somewhat tenuous (see the section entitled Burden of Illness). In addition to aspiration, the general debilitation, malnutrition, and dehydration that may accompany stroke and degenerative neurologic conditions could increase risk of pneumonia by weakening immune defenses. Also, poor oral hygiene might accompany these conditions and thus increase the presence of oral-pharyngeal bacteria or other microorganisms that cause increased risk of pneumonia. Nevertheless, in the presence of markedly increased dysphagia and aspiration in these neurologically impaired patients, aspiration remains the most plausible route into the lungs for microorganisms that might overwhelm a weakened immune system. Therefore, even though aspiration alone may not be sufficient to cause pneumonia in otherwise uncompromised people, aspiration does appear to participate in the etiology as the most plausible route of harmful microorganisms into the lungs. Thus, the term aspiration pneumonia is probably warranted in patients who are at such increased risk of both aspiration and pneumonia.

Because of the lack of uniformity in the criteria for distinguishing aspiration pneumonia from other pneumonia in dysphagia in published studies, we were forced to accept the terminology used by the authors of each study. This is unlikely to have had a major impact on the results of this report because there were no marked differences in the frequency of pneumonia in acute stroke patients, regardless of whether the authors stated they were studying aspiration pneumonia or just pneumonia. However, the term chest infection, which is common in the medical literature of the U.K., did appear to be a broader definition that was associated with markedly higher frequencies. For this reason, we generally did not use this type of data. Otherwise, we included and analyzed any dysphagia study that examined pneumonia, regardless of whether the authors stipulated that it was aspiration pneumonia. Because we must frequently refer to results from more than one study, we often use the generic term pneumonia, even though some of the studies being referred to may have used the term aspiration pneumonia. Under the circumstances, it does not appear practical to put too fine a point on the possible distinctions between these terms.

The links of the evidence model that are relevant to answering this first question are the links between each of the diagnostic techniques and aspiration pneumonia. Referring to the evidence model shown in Figure 1, these are the eight imaginary lines between each of the boxes labeled D1 through D8, and box M8. These links each ask whether there is a relationship between use of any of the diagnostic tests and pneumonia or, more specifically, does use of any of these tests prevent aspiration pneumonia? It is also possible to compare the results of one link with another to determine whether the use of one diagnostic leads to lower pneumonia rates than another. As such, these links presume that treatment intervened between diagnosis and the outcome and, therefore, these links address the efficacy of dysphagia diagnosis and treatment programs. To ensure that the meaning of each link is clear, we preface our discussion of these links by presenting in each heading the question posed by the link and parenthetically noting the codes for the specific boxes of the evidence model that comprise the link.

Conclusions

The conclusions in this subsection concern comparisons of pneumonia rates obtained with and without dysphagia diagnostic and treatment programs. The above-evaluated studies of Odderson, Keaton, and McKenna (1995), Daniels, Brailey, Priestly et al. (1998), and Spiegel, Creed, Selber et al. [unpub.(b)] have provided data (albeit data that are limited, and potentially confounded) that are consistent with, but do not prove, the idea that use of a BSE, VFSS, or FEES in a dysphagia diagnosis and treatment program may substantially reduce pneumonia rates in stroke patients with dysphagia. These data do not allow us to specify exactly how successful these programs are (i.e., how many cases of pneumonia they would prevent in a larger context), nor do they allow us to conclude definitively that these programs caused the reduction in pneumonia frequency (as noted above, that could have been caused by changes in hospital policy, physician practice, or some other unspecifiable factor). Nevertheless, it would be imprudent to dismiss the substantial effects observed in these studies, and thus they provide evidence about a question that is at the fifth level, near the top, in the hierarchy of the societal interest of the questions that we describe in the Evidence Model section of this evidence report.

The suggestion of apparently substantial efficacy of these dysphagia programs, while derived from historical prospective case series and case series compared with historical controls (which are typically not as reliable as RCTs), indicates that conducting a randomized controlled trial in which some patients were not allowed to participate in a dysphagia program would be unethical. Additional data from dysphagia programs and from historical controls or contemporary controls in institutions or geographic areas without dysphagia programs would be helpful in assessing the magnitude of the effect of dysphagia programs in preventing pneumonia.

Because of the lack of even a historically controlled study of a VF dysphagia program, because of a lack of studies of the long-term patient outcome of interest (prevention of aspiration pneumonia), and because of the small size of the one reported uncontrolled case series, there is no statistically significant efficacy demonstrated for VF studies in the prevention of aspiration pneumonia (although there is an apparent trend in the one study reported). Further, it has not been demonstrated that predictions of aspiration pneumonia derived from videofluoroscopy are superior to those derived from the BSE or FEES, although there are some studies demonstrating that VFSS can detect substantially more patients with the surrogate endpoint of aspiration than the BSE. This lack of statistically significant evidence does not mean VF swallowing diagnosis and treatment programs have no effect. Rather it reflects the lack of hard experimental evidence.

This section has only dealt with comparisons of diagnostic methods used in dysphagia programs compared with the no-program situation. Direct comparisons among BSEs and instrumented exams are discussed below in Questions 2 and 3.

Diagnostic Test Guidance of Treatments

The results of a diagnostic test are typically used to determine the course of subsequent treatment. The diagnosis of aspiration or dysphagia per se will simply determine that treatment is necessary; more specific symptoms (such as the severity of aspiration) will determine the specific treatment applied. Clinicians may do this to different degrees- while some clinicians may simply determine treatment on the basis of gross manifestations (such as the presence or absence of aspiration), other clinicians may use a more specific technique, such as assessment of particular physiologic functions, as predictors of which treatment will be most effective.

One implication is that a description of the diagnosis that guided treatment is relatively uninformative unless the outcomes after the treatment are described as well. A study may report, for example, that detection of aspiration of more than 10 percent of bolus volume led to the placement of a feeding tube. If the study does not report the outcomes from these patients placed on feeding tubes, and compare them with other patients placed on feeding tubes using different diagnostic criteria, there is no way to determine if the influence of the diagnostic test on treatment helped the patient, had no beneficial effect, or harmed the patient.

Outcomes reported after the treatment should be clinically meaningful. It is not sufficient to report that the pharyngeal transit time as measured by videofluoroscopy was significantly reduced after tactile-thermal stimulation (TTA). Rather, the outcome measure must in some way directly measure the patient's well-being. Clinically meaningful outcomes are occurrence of pneumonia, dehydration, malnutrition, QOL, and mortality. Intermediate or surrogate measures that would fit these criteria would be occurrence of aspiration, coughing, or choking; these are all measurable outcomes that directly affect the patient.

Diagnostic technologies can be used to guide treatment in two ways: the diagnostic can detect signs that predict the success of future treatments, or the diagnostic can be used while treatments are being attempted. Videofluoroscopy has often been used in the latter way. However, while videofluoroscopy may indicate that a particular treatment is effective in the first 5 minutes, it cannot determine whether the patient is able to continue receiving benefit long term.

Many studies, most using videofluoroscopy, have used a diagnostic technology to guide or assess treatment. This has been done to assess the appropriateness of specific dietary characteristics (Bisch, Logemann, Rademaker et al., 1994; Logemann, Pauloski, Colangelo et al., 1995), postural changes (Ekberg, 1986; Logemann, Kahrilas, Kobara et al., 1989; Rasley, Logemann, Kahrilas et al., 1993; Shanahan, Logemann, Rademaker et al., 1993), and other applied treatments (de Lama Lazzara, Lazarus, and Logemann, 1986; Rosenbek, Robbins, Willford et al., 1998). Only one study used a diagnostic method other than videofluoroscopy to make treatment recommendations (Eibling, 1994). No studies have reported the use of symptoms detected using any other diagnostic method to guide the course of treatment.

Conclusions Concerning Diagnostic Guidance of Treatment Management

Eight of nine studies discussed here used videofluoroscopy to assess treatment options. The most meaningful outcome, reported by three of these studies, was the occurrence of aspiration. All other measures reported were specific physiological swallow measures. These are considered a way of measuring the likelihood of having a safe swallow; however, they only imply changes in swallow safety. No studies have compared different diagnostic criteria in the determination of treatment to compare the relative predictive value of specific symptoms shown by the diagnostic tool. No diagnostic methodologies have been compared with one another as to their ability to guide treatment. In particular, no studies have assessed the ability to use specific symptomatology during a BSE to guide treatment.

What Clinical Signs and Symptoms Predict Pneumonia? (Link D1 to M1)

Recent studies have not been able to determine the absolute test characteristics [sensitivity, specificity, positive predictive value (PPV), Negative predictive value (NPV); for definitions see Appendix E] for signs, symptoms or BSEs. This is because treatment intervenes between the test and the outcome of pneumonia (for further discussion of this issue, see the section entitled The Relationship between Diagnosis and Treatment of Swallowing Programs above). In this situation, the test characteristics indicate the ability of the test to detect or predict only the unpreventable cases of aspiration pneumonia; we have no idea how many cases would have occurred without treatment. To the extent that some cases were prevented and false positives are unknown, sensitivity, specificity, and PPV would be underestimated, while NPV would remain unaffected. Nevertheless, within the patient population of a single study, it is possible to determine the relative measures of diagnostic examination performance carried out on the same subjects, provided that treatment following each exam is chosen in exactly the same way, and caregivers are preferably blinded as to which test was used to prescribe treatment.

We are interested in detecting as many of the patients at high risk for pneumonia as possible; thus, sensitivity needs to be high, over 80 or 90 percent (fewer than 10 to 20 percent of the high-risk patients would be missed). At the same time we want to minimize the number of false positives referred for the more expensive instrumented exam; thus, specificity needs to be at least moderate, say over 70 or 80 percent (fewer than 20 percent or 30 percent of the low risk patients would be wrongly referred for the instrument exam). Because, as discussed above, intervening treatment will result in underestimates of sensitivity and underestimates of specificity, we should possibly require 90 percent sensitivity and accept 70 percent specificity.

The signs and symptoms data reported by Harkness, Bentley, and Roghmann (1990) in a matched-control study show higher sensitivity and specificity than those examined by Langmore et al. (1998), but the methods are problematic. The biases in matched-control diagnostic studies are such that they typically overestimate measures of test performance. For a given diagnostic test, PPV and NPV are not set, but PPV increases as prevalence increases and NPV decreases (Fletcher, Fletcher, and Wagner, 1988). Yet in matched-control studies, prevalence is usually set artificially high, which overestimates PPV and underestimates NPV (see Appendix E). For example, in this study, each case in the disease arm was matched with two nondiseased controls; thus, pneumonia prevalence was artificially set at 33 percent, artificially inflating PPV and NPV for any population with a lower prevalence.

Sensitivity for a given test is also not fixed, but depends on the spectrum of disease severity (Brenner and Gefeller, 1997; Fletcher, Fletcher, and Wagner, 1988; Gann, 1996; Ransohoff and Feinstein, 1978). Most tests are more sensitive for severe disease (or risk). Case-control and matched-control studies usually preselect disease cases so that the spectrum of disease severity is not typical, which overestimates sensitivity. In the Harkness et al. study, only radiographically confirmed cases with one or more signs or symptoms possibly associated with pneumonia that occurred within an 18-week period were chosen for cases. Atypical cases of pneumonia that were not radiographically confirmed or that were slow to develop and had not yet occurred in the 18-week period may have been more difficult to predict but were not included in the sensitivity count, thus possibly inflating sensitivity. Specificity is also not fixed but is dependent on the presence of comorbidities with symptoms that overlap those of the disease of interest. In case-control studies, controls are usually selected to have no confounding comorbidities, which overestimates specificity. Such was the case in this study, in which controls were selected who had no obvious signs or symptoms of respiratory infection. To the extent that signs and symptoms of respiratory infection and dysphagia overlap, some patients with these overlapping symptoms were excluded from the controls even though they may have been falsely predicted to get pneumonia by these very signs and symptoms, thus artificially inflating specificity. It is not surprising then that the test characteristics given by the Harkness, Bentley, and Roghmann (1990) data are generally higher values than Langmore, Terpenning, Schork et al. (1998) reported, given the potential biases described above, and therefore are probably unrealistically inflated.

Conclusions

What Clinical Signs and Symptoms are Indicative of Aspiration? (Link D1 to S1)

Because aspiration does not inevitably cause pneumonia, it must be considered a surrogate endpoint for pneumonia, and thus of secondary importance. Nevertheless, because of our inability to determine true test characteristics for prediction of pneumonia (as discussed in the previous subsection), we were forced to consider an intermediate portion of the overall link between test effectiveness and the outcome of pneumonia. Therefore we now consider the link between clinical signs and symptoms on a BSE (D1 through D4) and detection of aspiration (S1).

For detection of aspiration, we would want to have signs and symptoms that are almost as good as aspiration detected at a VFSS above 80 percent sensitivity (less than 20 percent of patients with aspiration detected by VFSS would be missed by the sign or symptom), and we would want specificity to be moderately high, above 70 percent (no more than 30 percent of patients referred for VFSS would be needlessly tested by VFSS for aspiration). In general it can be seen in Evidence Tables 34 and 35 that items with higher sensitivities have low specificities. This is expected because of the inevitable tradeoff between sensitivity and specificity as the diagnostic threshold for a positive test is changed, either intentionally or unintentionally. Because of this tradeoff, sensitivity and specificity values are best considered as sensitivity-specificity pairs, and the relationship of the pair values to the threshold chosen by the examiner, or enforced by the symptom, is best appreciated on a plot of sensitivity versus specificity. Such a plot is known as a receiver operating characteristic (ROC) plot, and the regression curve for a set of sensitivity-specificity points is the ROC curve (see Appendix E for more details of this statistical method).

We examined the available data for each of the other individual signs and symptoms of the BSE, to determine whether they could be combined meta-analytically. Combining all the points from the different subtests into a summary ROC curve would not be an appropriate way to determine the effectiveness of the BSE. Such a curve would underestimate the sensitivity and specificity of the BSE because: (1) it would not consider the effects of employing multiple tests (results of other tests could cancel out a false-positive or false-negative result of any given test, making the whole more effective than the parts), and (2) some of the less-effective subtests are unlikely to be used in an optimized BSE, and therefore should not be allowed to affect the results of the meta-analysis. Also, there are only a small number of trials for some of the individual signs and symptoms (e.g., only one trial reported "multiple swallows"). For these reasons, we chose not to perform a meta-analysis on the data for each individual subtest and simply present these data in ROC space.

Figure 2. Sensitivity and Specificity of Oral-Pharyngeal (more...)

   Figure 2. Sensitivity and Specificity of Oral-Pharyngeal Symptoms for Detection of Aspiration Plotted in ROC Space

Figure 2

Thus, some of these signs and symptoms look more promising than others, but there needs to be more research on this topic before a conclusion can be reached. As noted above, more useful than the test characteristics for these items alone are the characteristics of various combinations of signs and symptoms, and some of these authors reported results of combinations that we present in the next two tables.

Meta-analysis of the 3-ounce Water Test: Sensitivity and Specificity for Aspiration

Figure 3. Sensitivity and Specificity of the 3-oz Water (more...)

   Figure 3. Sensitivity and Specificity of the 3-oz Water Test for Detection of Aspiration Plotted in ROC Space

Various versions of the BSE include some of the signs and symptoms discussed above along with other items. Evidence Table 36 shows the test characteristics obtained in several studies of BSEs for detecting aspiration confirmed by VFSS for patients with mixed etiologies. Evidence Table 37 shows BSE test characteristics for aspiration in stroke patients referred to VFSS with dysphagia symptoms. Evidence Table 38 is similar to Evidence Table 37, except that the former table presents results obtained from examining all stroke patients, not just those referred for VFSS. Sensitivities range from 0 to 91 percent. However, in general, specificity is high when sensitivity is low, and low when sensitivity is high. Four of the studies tabled used an abbreviated version of a full BSE, the 3-ounce (or 50-ml) water test. In Figure 3, these sensitivity-specificity points have been used to construct a summary ROC curve (Littenberg and Moses, 1993) (see Appendix E). The points of the individual signs and symptoms discussed above are also plotted. It can be seen that the 3-ounce water test summary ROC curve falls below our hypothetical curve passing through 80 percent sensitivity and 70 percent specificity. This indicates that, even if the threshold is somehow manipulated (which causes one to move along the curve as opposed to shifting the curve), this test is unlikely to achieve the necessary sensitivity without sacrificing too much specificity, and vice versa. Also, it can be seen that the 3-ounce water test curve is similar to the average ROC curve we would expect for the whole group of signs and symptoms (11 points are above and 12 points are below the 3-ounce water test curve). This indicates that the 3-ounce water test may not contain the best individual signs and symptoms available.

Meta-analysis of the Characteristics of the BSE: Sensitivity and Specificity for Aspiration

Figure 4. Sensitivity and Specificity of BSEs for Detection (more...)

   Figure 4. Sensitivity and Specificity of BSEs for Detection of Aspiration Plotted in ROC Space

In Figure 4, we plotted the summary ROC curve for the five studies that reported both sensitivity and specificity for a complete, formal BSE. These comprehensive BSEs produce a curve that is above both the 3-ounce water test curve and the average curve for all the individual signs and symptoms. Interestingly, the BSE curve falls very close to our hypothetical 80 percent sensitivity and 70 percent specificity curve. This reasonably high sensitivity of the BSE is one explanation for why the BSE dysphagia program reported by Odderson, Keaton, and McKenna (1995) (discussed under Question 1 above) had such a substantial apparent effect in reducing aspiration pneumonia spite of the likely existence of substantial silent aspiration (as defined by absence of cough). Full BSEs score several items that may increase their sensitivity for aspiration, and so are not limited to the single symptom of failure to cough.

The data in the above-mentioned tables and figures indicate that only a BSE using more and/or better signs and symptoms than the 3-ounce water test would be appropriate for detection of aspiration and/or referral of patients to an instrumented exam. In fact, these data, along with the effectiveness of the BSE program reported by Odderson, Keaton, and McKenna (1995), call into question the need for instrumented exams to detect aspiration. The 80 percent sensitivity of the BSE for aspiration using VFSS as a gold standard indicates that VFSS would only detect an additional approximately 20 percent of patients who aspirate. Instrumented exams, particularly those providing imaging, seem likely to be more useful than a BSE for guiding treatment and therapy, because of the additional physiological and functional information they provide. However, the effect of this additional information on patient outcomes such as pneumonia, malnutrition, dehydration, and QOL has been poorly demonstrated. Also, as discussed above under Question 1, the approximately equal effectiveness of the Odderson BSE program (Odderson, Keaton, and McKenna, 1995) and the Daniels VFSS program (Daniels, Brailey, Priestly et al., 1998) in preventing pneumonia indicates that the added benefit of VFSS over BSE is likely to be small, and, in fact, too small to be detected by anything other than a study much larger than either of these.

The potentially small benefit that accrues to patients as a result of using VFSS, and its somewhat greater cost than the BSE, mean that a comparison of the BSE with VFSS is primarily a cost-effectiveness issue. We address cost-effectiveness in our supplemental analysis.

Conclusions

Single symptoms detectable on clinical exam do not appear to reliably predict aspiration. It seems likely that there should be some combination of clinical oral-pharyngeal signs and symptoms that can predict aspiration with adequate sensitivity and at least moderate specificity. The approximately 80 percent sensitivity and 70 percent specificity for aspiration of full BSEs including several signs and symptoms bears this out, and could possibly be improved upon. Further research on this topic is of great interest, but it is important that researchers understand the tradeoff involved in the sensitivity-specificity relationship. They may be able to reach a consensus on some combination of these values that achieves both adequate sensitivity and specificity for prediction of aspiration or aspiration pneumonia. The reasonably high sensitivity and specificity of BSEs raises the issue of whether a BSE could entirely replace instrumented exams. The possibly small improvement of instrumented exams over BSEs and the increased cost of instrumented exams present a cost-effectiveness question that is discussed in our supplemental analysis. This question also raises the issue of the effect of diagnostic exam findings on subsequent treatment.

What Clinical Signs and Symptoms Predict Malnutrition? (Link D1 to M2)

Malnutrition is a significant predictor of mortality in the elderly (Elmstahl, Persson, Andren et al., 1997; Keller, 1995; Murden and Ainslie, 1994; Ryan, Bryant, Eleazer et al., 1995; Wallace, Schwartz, LaCroix et al., 1995). It can also lead to other morbidities through a weakening of the immune system (Chandra, 1990). It is therefore important to enact measures to prevent malnutrition if it is determined to be prevalent in a population of patients with dysphagia. To do this, it would be helpful to identify any signs or symptoms that predict eventual malnutrition in these patients.

The relationship between malnutrition and diagnosis of dysphagia is difficult to determine from available evidence. This is partly because the administration of parenteral or enteral nutrition is often introduced if there is even a threat of malnutrition, so the number of patients whose dysphagia might lead to malnutrition may be at least partially obscured. In addition, diagnosis of dehydration and malnutrition does not require extensive diagnostic methods for detection. For these reasons, in acute care, malnutrition may be largely dealt with outside of the context of formal diagnosis and treatment of dysphagia. This could explain why few studies have attempted to determine the extent to which (or even whether) dysphagia causes malnutrition. There is, however, the possibility that for some patients diagnosis of dysphagia may lead to therapy that prevents malnutrition and therefore obviates the need for feeding tubes.

There is limited evidence about the prevalence of malnutrition in patients with dysphagia, and whether patients with dysphagia are at increased risk for malnutrition compared with individuals with normal swallows. Current evidence on this is equivocal, as two studies have reported a significant association (Davalos, Ricart, Gonzalez-Huix et al., 1996; Keller, 1995) and another found no association (Thomas, Verdery, Gardner et al., 1991). Only one of these studies was specifically about acute-care stroke patients, and it found that 1 week after stroke, patients with dysphagia were significantly more likely to be malnourished than those with a normal swallow (Davalos, Ricart, Gonzalez-Huix et al., 1996).

However, no published studies examine signs or symptoms that may predict malnutrition in patients with dysphagia. Given that an association between dysphagia and malnutrition has not been definitively made, there may be no reason for such studies to be conducted. Therefore researchers must definitively document a link (if one does exist) between dysphagia and malnutrition in specific patient populations to determine the burden of illness in this population; then clinicians can work on ways to detect potential malnutrition risk factors before malnutrition occurs.

What Clinical Signs and Symptoms Predict Dehydration? (Link D1 to M2)

Dehydration may be a particularly important outcome in regard to neurologically disabled patients with dysphagia because they most often suffer from an inability to swallow thin liquids safely; less common is the inability to swallow thick liquids and solids. However, this is a logical inference, rather than one based on clinical studies. While clinicians recognize the importance of measuring dehydration (Langmore, 1998 ; Logemann, 1998b; Yorkston, 1998), few studies have yet been conducted on this outcome. Currently, no studies examine the predictive value of clinical signs and symptoms for development or presence of dehydration.

How Reliably do Videofluoroscopic Studies Detect Aspiration? (Link D5 to S1)

Conclusions

These studies together indicate that there is fairly good interobserver agreement on aspiration of thin liquids, but that thicker liquids and solids present a problem of interobserver agreement. Unfortunately, the inter- and intraobserver variability of a diagnostic technology are used as the reference standard for judging all newer technologies, yet they have not been thoroughly studied. Potentially, the absence of interobserver agreement on aspiration of solids and thicker consistencies could present a risk to some patients, because in the standard procedure (Logemann, 1983b), these consistencies are tried first as indications of whether it is safe to proceed on to thin liquids. We were unable to find any evidence that addressed this issue, and there is likely to be little risk for stroke patients, who usually have more difficulties when swallowing thin liquids.

Which is a Better Predictor of Pneumonia: BSE or Videofluoroscopy? (Links D4 and D5 to M1)

There is a general assumption in the field of dysphagia research that VFSS (or any instrumented exam for that matter) is better than any noninstrumented exam. This hypothesis needs to be tested in clinical trials. The sensitivity and specificity of the BSE for detection of aspiration using VFSS as a gold standard was discussed above in Question 2. However, the use of VFSS as a gold standard requires the assumption that VFSS is 100 percent accurate, which may not be the case. Thus, in order to compare the BSE with VFSS, a clinical outcome must be used as the gold standard. However, this is problematic because measuring effectiveness only in terms of the ability of the test to predict pneumonia is unreliable, because treatment intervenes and will bias the results. Examples of this will be presented below.

Case Series Comparisons of VFSS and BSEs

Because of the scarcity of trials comparing BSE with VFSS within the same patient population, we turned to the case series literature. Obvious differences in study design and patient characteristics among these studies make comparisons among them of questionable reliability. For example, treatments may have differed in significant ways, although none of these studies provided in-depth description of the therapies. Study design issues, such as retrospective versus prospective case identification, may have resulted in different findings so that retrospective case review could not identify all diagnoses or occurrence of illness, therefore skewing results. We seek, however, only to roughly approximate a comparison of test characteristics for these two tests. Given the purported advantages of VFSS in detecting cases of silent aspiration missed by BSE that may ultimately lead to pneumonia, even with study differences, VFSS should appear to have increased sensitivity.

Several case series studies have been published that examined either the ability of BSE to diagnose an unsafe swallow (Evidence Table 41) or MBS detection of aspiration to predict pneumonia (Evidence Table 42). We have assumed for this discussion that care setting does not affect test characteristics in any significant way (though the differences in prevalence of disease in different care settings will obviously have some effect); therefore, acute-care and rehabilitation centers are analyzed together. We have limited the pool of studies to those reporting on 90 percent or more stroke patients. Because of a paucity of data, the formal BSE and water swallow test have been collapsed into a single category of noninstrumented exam. We are, therefore, disregarding many factors that may potentially confound results.

Evidence Table 41 includes trials that related pneumonia rates to results of the BSE or 3-ounce/50-ml water test. Key points regarding these trials and their suitability for between-studies comparison are presented below.

Gordon, Hewer, and Wade (1987), examining the 50-ml water test, excluded those patients known to have choked on fluids the day of the test. Since these patients likely would have choked during the 50-ml water test, this selection bias is likely to have excluded cases that would have been true positives, thus underestimating sensitivity and PPV of the test.

DePippo had very sensitive criteria for interpreting the Burke Dysphagia Screening Test (BDST). Failure on any of the seven elements was interpreted as failure of the test. Thus, sensitivity was high (92 percent) and specificity was low (45 percent). Patients who failed the BDST were referred for barium swallow. Patients failing the barium swallow were enrolled in a dysphagia program including diet modification and compensatory swallowing techniques. As a result, treatment probably intervened for most patients who had dysphagia.

Table 15. One Study's Results Illustrative of the Effects (more...)

Table 15. One Study's Results Illustrative of the Effects of Treatment Bias (Gottlieb, Kipnis, Sister et al., 1996)

		50-ml water test results in first group
		Pneumonia outcome
		Positive	Negative
		50-ml water test results in the third group
		Pneumonia outcome
50-ml water test result	Positive	6 TP	7 FP	46 percent PPV
	Negative	3 FN	44 TN	94 percent NPV
Pneumonia rate: 15 percent		67 percent sensitivity	86 percent specificity
		Positive	Negative
50-ml water test result	Positive	0 TP	13 FP	0 percent PPV
	Negative	2 FN	45 TN	96 percent NPV
Pneumonia rate: 3 percent		0 percent sensitivity	76 percent specificity

Gottlieb, Kipnis, Sister et al. (1996) called their trial a validation of the 50-ml water test, but because it does not compare the 50-ml water test with a reference test for diagnosis of dysphagia, it is not truly a validation. It is, however, a trial measuring the ability of the 50-ml water test to predict pneumonia. More than that, this is a trial measuring the effect of a comprehensive swallowing program on pneumonia rates. Three consecutive groups of 60 patients were enrolled in the swallow program, using the 50-ml water test as the criterion for selecting patients for swallow therapy. Patients who failed the 50-ml water test were given a modified diet, along with three sessions per week of swallow therapy by a speech therapist. Pneumonia rates in patients with and without the modified diet were compared. The change in results is graphic evidence of the effect of treatment bias (Table 15).

As the dysphagia program became established, true-positive results decreased while false-positive results increased. Sensitivity, specificity, and PPV all appeared to decrease as treatment became more successful at preventing pneumonia. (See Appendix G for further discussion of the effects of treatment bias.)

All of these trials found that the BSE had high NPV. Reported PPVs were very low; this could be the result of treatment bias as well as selection of a sensitive test threshold to minimize false-negative results. Smithard's trial found NPV to be 84 percent; all the others had NPV results over 90 percent. This demonstrates that BSE results can be used to identify patients who are unlikely to need swallow therapy.

Although Evidence Table 42 shows DePippo, Holas, and Reding (1994) twice, only the results for the BSE were graphed to avoid double counting of patients. Smithard et al. (1996) is included in both tables but not included in our graph, due to only a 1-week followup period that may affect its results relative to the results from other studies with longer followup times (it has, however, already been discussed above). Holas et al. (1994) is tabled but excluded from analysis because results were only reported for those patients who tested positive on VFSS, thus resulting in patient selection bias.

Johnson's group tested the ability of VF examination to predict pneumonia. Kinematic pharyngeal transit time (Em) was the primary measure of swallow function; pneumonia rates were stratified by group, with threshold Em times ranging from 2 to 5 seconds. This allows consideration of a range of thresholds and illustration of the tradeoff between sensitivity and specificity. At the lowest threshold (2 seconds), sensitivity was 100 percent and specificity was 42 percent; at 3 seconds, sensitivity was 86 percent and specificity was 68 percent; at 4 seconds, sensitivity was 76 percent and specificity was 81 percent; and at 5 seconds, sensitivity was 66 percent and specificity was 94 percent. Treatment was not reported, so it cannot be determined if treatment bias affected the results. The trial does not appear to have spectrum bias, because it was a retrospective review of consecutive cases referred for videofluoroscopy. The only exclusion criteria were dysphagia caused by disease other than stroke (two cases) and patient noncompliance causing an examination of poor quality (two cases, 3.2 percent). The latter exclusion causes a small bias against VFSS, but is appropriately conservative in design.

The Holas, DePippo, and Reding (1994) report on the correlation between MBS results and pneumonia showed that all patients selected for this trial had videofluoroscopy results that were positive for dysphagia. Patients also were given unspecified dysphagia treatment, so treatment bias probably affected the results.

Aviv, Sacco, Mohr et al. (1997) reported MBS results in 20 patients with known dysphagia. Although all patients got some treatment, treatment differed by group. The 10 patients whose barium swallow demonstrated aspiration were given a percutaneous endoscopic gastrostomy (PEG) tube, and the 10 that did not have aspiration were given swallow therapy and no PEG tube. Pneumonia risk was twice as great for those patients who had negative barium swallow results than for those who had positive results, a counterintuitive result. The reported barium swallow sensitivity of 33 percent and specificity of 43 percent are worse than would be obtained by random chance. This suggests that treatment bias affected the results.

Aviv et al. also tested laryngopharyngeal sensory discrimination testing (LPSDT) in a matched-control trial. Twenty patients with complaints of dysphagia and who were not taking food by mouth were paired with healthy age-matched controls. The controls had no dysphagia and no history of stroke. None of the controls contracted pneumonia or had abnormal LPSDT results. Six of the dysphagia patients (30 percent) contracted pneumonia, all of whom exhibited a bilateral sensory deficit on the LPSDT. Treatment bias affects these results too, though; since there were no false negatives, sensitivity remained at 100 percent. Specificity of the LPSDT was 57 percent while PPV was 43 percent.

Daniels et al. (1998) and Aviv et al. (1997) are interesting studies to discuss because of outlier VFSS test characteristic findings. Aviv reports a very low sensitivity of 33 percent from a very small pool of patients (20). Because all patients who tested positive for aspiration were put on a feeding tube (as opposed to noninvasive therapies reported in most other studies), we excluded these results from further analysis in an attempt to control for treatment effects. On the other hand, Daniels et al. (1998) reported a sensitivity rate of 100 percent; this was because only one case of pneumonia occurred in their patient population of 55; these results are thus rendered unreliable because of the low number of patients who developed pneumonia, resulting in low statistical power.

NPV is the parameter least affected by the known biases. For the barium swallow trials listed in Evidence Table 42, there was a wide range of NPV values, from 55 to 100 percent. The wide range of values could be explained by the possibility that some investigators used more sensitive thresholds than others. Summary ROC meta-analysis would be able to test this hypothesis, but meta-analysis of this evidence base is not appropriate because of treatment bias and selection bias previously described.

Because so many of the trials are confounded by treatment bias and/or spectrum bias, we could not use their results in a between-studies comparison or draw conclusions on the relative effectiveness of the 3-ounce water test, the barium swallow, and other tests based on these results. Trials measuring the effect of these diagnostic procedures on outcomes such as pneumonia rate are necessary to determine the value of these diagnostic tests.

Which is the Better Predictor of Pneumonia: Videofluoroscopy or Fiberoptic Endoscopy? (Links D5 through D7 to M1)

We located two studies that addressed this question, a pseudo-randomized controlled trial and a nonrandomized controlled trial. We address these two studies in separate subsections below.

Which Better Detects Aspiration, VFSS or FEES? (Links D5 through D7 to S1)

Although some studies used one diagnostic as a gold standard against which to compare the other diagnostic, this is not appropriate, because FEES may detect some cases of aspiration that VFSS does not, and VFSS may provide some false positives that FEES correctly identifies as negative, and vice versa. The Leder et al. (Leder, Sasaki, and Burrell, 1998) study only reported diagnostic accuracy, which is the combination of positive and negative agreement. Unfortunately, diagnostic accuracy does not tell one whether the disagreements were in the positive or negative direction, which is particularly important when the consequences of a false positive are different from those of a false negative. In the present case, the consequences of a false negative are pneumonia and, perhaps, death. On the other hand, the most serious possible consequences of a false positive are the complications caused by feeding tubes.

The remaining three studies all reported finding some apparent cases of aspiration detected by VFSS but not FEES, but they also reported that some cases detected by FEES were not detected by VFSS. Kaye, Zorowitz, and Baredes (1997) reported substantially more cases of aspiration detected by VFSS than by FEES. This was the largest study, but there was no indication that the readers of one test were blinded to the results of the other test. Also, the bolus volumes, consistencies, and number of boluses were not equivalent between the two exams, and it is difficult to detect aspiration of water by FE (milk or colored liquids are typically used). In the Wu, Hsiao, Chen et al. (1997) study, the readers were blinded, and more positives were reported for FEES than for VFSS. The remaining study by Langmore, Schatz, and Olson (1991) reported the same number of positives for each test that were missed by the other test; however, this was only 1 (out of 21 tests) in each case.

The important finding is that in all of these studies, each test detected some apparent positives that the other test did not. This means that neither test can be the reference standard for the other. The amount of disagreement over these positives is not reliable, because there was no outcomes-based gold standard for aspiration. Investigators on future studies should recognize these limitations of study design and they should also recognize that aspiration is a surrogate marker for the more useful endpoint of long-term pneumonia frequency.

Conclusions Concerning the Comparability of Videofluoroscopy and Fiberoptic Endoscopy for Prevention of Pneumonia and Detection of Aspiration

A single pseudo-randomized controlled trial with low statistical power and moderately strong internal validity reported a trend that incidence of aspiration pneumonia in dysphagia patients with mixed etiologies was lower after FEESST than VFSS but failed to show that either test was statistically significantly superior. This study did not address whether either of these tests is superior to the BSE, and the reader is referred to Question 2 for a discussion of this latter issue.

We found no reliably interpretable evidence comparing the abilities of VFSS and FE to detect aspiration. This difficulty was primarily due to the fact that the trials that examined aspiration failed to incorporate an appropriate gold standard external to VFSS and FE. However, even were such an aspiration gold standard possible, it is important to remember that aspiration is a surrogate endpoint for the long-term followup of pneumonia frequency. Thus, at the present, aspiration pneumonia frequency is both the most desirable and the only available gold standard for any instrumented exam. However, intervening treatment biases diagnostic test detection and prediction results. Thus, without withholding treatment, the absolute test detection and prediction results cannot be known. If treatment is not identical after each test, the relative detection and prediction abilities also cannot be known.

Summary

Measuring the performance of diagnostic tests for dysphagia is difficult because of the lack of a gold standard. While VFSS is the most widely used diagnostic imaging method, and therefore considered the gold standard, it does not demonstrate perfect sensitivity and specificity. As a result, it becomes difficult to interpret the results of studies that compare VFSS with other newer diagnostics, such as FEES and FEESST. This is because when one test yields a positive diagnosis and the other test yields a negative one, it is impossible to determine which test is correct. Similarly, if the two tests agree, they may both be wrong or they may both be right. The only way to assess test performance is to use clinical followup. In the studies we examined, however, followup was confounded by treatment intervention.

Because of these limitations, research thus far has been unable to demonstrate any significant advantage of one test over another. Comparisons have been made of the abilities of VFSS and FEESST for prevention or prediction of aspiration pneumonia and of the abilities of VFSS and FEES to detect aspiration. There was a statistically nonsignificant trend in one pseudo-randomized trial comparing VFSS and FEESST suggesting that incorporation of FEESST into a dysphagia diagnosis and treatment program results in lower aspiration pneumonia rates than in programs that employ VFSS. Further data are needed to determine whether this trend is reliable. Another within-subjects trial comparing the predictive value of these tests for aspiration pneumonia was flawed in that treatment decisions were made on the basis only of VFSS, and not on the basis of the FEESST results, thus leaving the results of each program noncomparable.

A comparison of VFSS and FEES for detection of aspiration in four studies showed that VFSS detects some aspiration that FEES does not, and that FEES detects some aspiration that VFSS does not. The ultimate gold standard, clinical outcome, was not used in any of these trials. Because of the above-mentioned limitations in studies of dysphagia diagnostic technologies, and because of the lack of a reliable gold standard, these trials provide no reliable information about relative efficacy of these two diagnostic tests.

A few studies have evaluated the interobserver reliability of videofluoroscopy. Two studies reported very high reliability overall when testing several different viscosities. These studies used retrospective viewing of tapes or film. As such, these results may not be generalizable to the more common procedure of viewing the test and making clinical decisions in real time. One study that analyzed each viscosity independently reported a high reliability only for thin liquids, and very low or nonexistent reliability for thick liquids and solids. While this low reliability is problematic because solids and thick liquids are the first substances tested on VFSS, the high reliability of thin liquids is reassuring because this is the viscosity most often problematic for patients with neurogenic dysphagia.

No studies have compared the test characteristics of FEES and FEESST. Such a comparison is warranted to determine if the quantitated sensory testing included as part of FEESST actually provides any clinically relevant information. Further within-subjects studies comparing VFSS and FE should make treatment determinations for patients tested positive by either test so that results, although confounded by the treatment, will be comparable between the two tests. Alternatively, a randomized controlled trial should be carried out in which the patients in each arm are diagnosed by only one method, and are treated solely based on that method. This would demonstrate the relative effect of each test on actual patient outcomes; however, the detection and prediction abilities of the tests in the absence of treatment would not be known.

In the absence of such a trial, and because current literature does not appropriately address the issue of differences between diagnostic tests, we further explored the answer to this third question in our supplemental analysis. This latter analysis, in turn, relies upon the data we discussed in our answer to Question 1.

Outcomes Considered

Figure 1

Many studies have looked at the short-term effects that noninvasive swallow therapies (such as chin tuck, head turning, and palatal training appliances) and diet modification (such as changes in bolus volume, flavor, or temperature) have on the physiological process of swallowing, such as total swallow duration and pharyngeal transit time (Bisch, Logemann, Rademaker et al., 1994; Lazarus, Logemann, Rademaker et al., 1993). However, little evidence in the literature directly links changes in these physiological processes to changes in the overall swallowing-related health of the patient; for example, there is no evidence linking improved (decreased) pharyngeal transit time to reduced incidence of aspiration pneumonia. Because of this, we will not discuss these findings, and we limit our discussion to the clinically meaningful outcomes outlined above.

Table 16. Treatments for which Outcomes of Interest (more...)

Table 16. Treatments for which Outcomes of Interest Were Not Reported

Swallow therapy		Diet modification		Combined swallow therapy and diet modification		Feeding tube
Outcome	Link	Outcome	Link	Outcome	Link	Outcome	Link
Change in food intake	T1 to O2	Change in food intake	T2 to O2	Malnutrition	T3 to M2	Malnutrition	T4 to M2
Weight change	T1 to O4	Prevention of feeding tube	T2 to O3	Dehydration	T3 to M2	Dehydration	T4 to M2
Malnutrition	T1 to M2	Weight change	T2 to O4	QOL	T3 to O7	QOL	T4 to O7
Dehydration	T1 to M2	Dehydration	T2 to O7
QOL	T1 to O7	QOL	T2 to M2

Table 17. Outcomes of Interest for which No Comparison (more...)

Table 17. Outcomes of Interest for which No Comparison Groups Exist in the Published Literature

Swallow therapy		Diet modification		Combined swallow therapy and diet modification
Outcome	Link	Outcome	Link	Outcome	Link
Aspiration	T1 to O5	Aspiration	T2 to O5	Aspiration	T3 to O5
Prevention of feeding tube	T1 to O3	Malnutrition	T2 to M2
Mortality	T1 to O8, O9	Mortality	T2 to O8, O9

The exception to this exclusion of case series is the literature on feeding tube usage, which may provide useful information from case series because this treatment is more applicable to chronic stroke patients who may not spontaneously recuperate.

Table 18. Outcomes Available for Analysis by Treatment (more...)

Table 18. Outcomes Available for Analysis by Treatment Category

Swallow therapy		Diet modification		Combined swallow therapy and diet modification		Feeding tube
Outcome	Link	Outcome	Link	Outcome	Link	Outcome	Link
Pneumonia	T1 to M1	Pneumonia	T2 to M1	Pneumonia	T3 to M1	Pneumonia	T4 to M1
				Mortality	T3 to O8, O9	Mortality	T4 to O8, O9, O10
				Change in food intake	T3 to O2	Aspiration	T4 to O5
				Prevention of feeding tube	T3 to O3	Weight change	T4 to O4
				Weight change	T3 to O4	Minor tube complications	T4 to O6
						Major tube complications	T4 to M4

We discuss each of these outcomes for each of these treatments (and, hence, each appropriate link in the evidence model) in the following sections and subsections. We will first discuss pneumonia, the most commonly reported outcome, followed by any other morbidities for which there were data; short-term outcomes (weight change, feeding method, energy intake, aspiration, minor tube complications) are then discussed, followed by mortality.

Noninvasive Therapies

Because of the inconvenience, discomfort, and cost of enteral feeding, it is important to patients, clinicians, caregivers, and payers that the patient be able to eat orally and independently whenever it is deemed safe to do so. This means that the patient must be able to ingest adequate amounts of food to maintain a healthy weight and not aspirate significant amounts of harmful substances. It must also be comfortable for the patient to eat; if they experience pain or the disquiet of coughing on every swallow, they will be reluctant to eat, and will ultimately lose weight and become susceptible to further illnesses.

Table 19. Characteristics of Controlled Studies on (more...)

Table 19. Characteristics of Controlled Studies on Noninvasive Therapies for Dysphagia

Study	N	Study design	Care setting	Mean age	Primary disease(s)	Treatment	Time frame
Groher 1987	23 23	RCT	Nursing home	71.8 74.2	CVA with history of aspiration pneumonia	Pureed diet Mechanically altered diet	6 mos
Kasprisin 1989	48 13 8	Retro CT	Hospital	NR	Various	Swallow therapy	NR
Martens 1990	16 15	HPCS	Acute hospital unit	49.3 46.1	Brain injury, tumor, CVA	Various (diet, exercise, counseling) No specific dysphagia treatment	NR
DePippo 1994	38 38 38	RCT	Rehab unit	76 74.5 73	CVA	Diet and swallow technique recommendations Therapist-prescribed diet and swallow techniques Therapist prescribed diet and techniques reinforced	1 year

All of these studies suffer from the same limitation that makes it impossible to make conclusions based on their results, no matter how well designed they are. The problem is an issue of statistical power-if the outcome of interest (in this case, pneumonia or mortality) is one that has a low incidence rate, to obtain statistically significant group differences, the total number of patients in the study must be high enough that a type II error can be avoided. An example about how to calculate the necessary number of subjects needed is included in the Future Research section. Studies on patients with dysphagia interested in reporting pneumonia or death as an outcome would require several hundred patients to obtain reliable findings.

PEG Tubes

In many cases, as discussed previously in this report, stroke patients spontaneously regain their ability to swallow. These are recuperative stroke patients with mild to moderate dysfunction. If these patients have severe difficulty swallowing immediately after the event, they may be given an NG tube temporarily to help them ingest adequate amounts of food. They may also receive noninvasive swallow therapy and/or diet modification until the dysphagia resolves. However, there are also chronic stroke patients who recover more slowly, or never recover full swallow ability. These patients would most benefit from effective noninvasive or invasive treatments because of the long-term risks of unsafe swallowing.

While NG tubes are used short term, PEG tubes, because they are invasive, are generally reserved for those patients requiring long-term enteral feeding (Lazarus, 1990). However, acute-care stroke patients are sometimes given a PEG tube for sustenance while undergoing noninvasive swallow therapy or while receiving recreational oral feeding. Those patients with chronic stroke accompanied by severe dysphagia may receive a PEG tube if noninvasive therapies prove ineffective in eliminating aspiration or sustaining adequate energy intake. With the advent of more specialized noninvasive therapies, PEG tube placement may become a last resort. The patients placed on PEG tubes are thus likely to be sicker than those who continue with noninvasive therapy. Consequently, it is not appropriate to compare the outcomes of PEG tube patients with those undergoing noninvasive therapy. Tube patients are not only at increased risk of morbidity and mortality because of their debilitated status, but the insertion of the feeding tube also can cause many different minor and major complications. These additional factors make it all the more important to identify efficacious noninvasive measures whenever possible and reserve PEG for those patients with unresolvable dysphagia or other feeding-related functional problems. Most studies conducted thus far on PEG tubes have included a mixture of patients with widely varying etiology (from paralysis to Parkinson's) and varying levels of cognition. Many studies included patients in a comatose state, and these studies will not be discussed here except in reference to tube-related complications and tube-related mortality, for which we assume the condition of the patient plays no significant role. For our discussion of morbidities and nontube related mortalities, we limit our discussion to studies including 90 percent or more cognizant patients, with a largely neurologic etiology (>75 percent), because those are patients most relevant to this report and most likely to be suffering from dysphagia, and therefore potentially eligible for noninvasive therapy. This reduces our pool of eligible studies from 20 to 7 (all studies are recorded in the appropriate evidence table, however). Only two studies were located that reported patients with dysphagia as a major part of their patient pool (Britton, Lipscomb, Mohr et al., 1997; Norton, Homer-Ward, Donnelly et al., 1996): We will pay particular attention to these two studies in this discussion.

Synthesis of Results Common to the Four Questions

Although this evidence report has concentrated on four main questions, these questions are not entirely independent of one another. Therefore, at this point, we present related information in one section. Thus, we present here the results of all of the controlled studies addressing our main question: Does a dysphagia diagnosis and treatment program prevent aspiration pneumonia? This, as noted in the Methodology section of this report, may be the question of most interest to society.

Thus, for the eight controlled studies above, we found four with highly significant results, two with near significant results and low power, and two with nonsignificant results and low power. Because of the low power, none of the nonsignificant results convincingly demonstrated no effect. This puts the emphasis on the four trials that found statistical significance. The most unbiased appraisal of these results is that there is an apparent effect (regardless of the cause of that effect) demonstrated by the studies that is not caused by random error. We again caution that a statistical analysis such as ours cannot eliminate confounding differences between control and experimental groups resulting from differences in treatment, patient characteristics, or other factors that contribute partially or wholly to the apparent effect.

Patient Selection

As mentioned above, most patients included in the studies thus far have been stroke patients; some researchers have examined exclusively stroke patients, while others have included a patient mix of many neurologic etiologies. Because stroke is a recuperative neurologic disorder, while many others that affect the elderly are degenerative, it creates problems to pool the outcomes data obtained from these patients. We suggest that in the future, researchers make sure to include patients with dysphagia of only one etiology in each study.

The lack of research on dysphagia in patients other than stroke victims ignores the considerable burden of illness resulting from these other diseases. More research on these latter kinds of patients is needed. It must be remembered that treatment recommendations made for a patient with Parkinson's disease, for example, on the basis of how a stroke patient with similar symptoms responded to that treatment is not necessarily appropriate.

Patients should all be at the same stage of disease, or plans should be made before a study is begun, to stratify patients by disease stage. The ambiguity caused by failure to use a homogeneous patient group or to stratify is illustrated by studies reporting that they examined stroke patients a mean of 15 days post-stroke. This could mean that some patients were seen immediately after the stroke while others were seen 2 months post-stroke. Symptoms may be different at these different time periods, making results difficult to interpret. The same holds true for other patients with neurologic disease: in neurodegenerative diseases, swallowing difficulties may be more pronounced as the disease progresses. Therefore, the efficacy of diagnosis and treatment should be examined by severity of disease.

Control Groups

Issues of study design are especially important with stroke patients, because they often regain their swallowing ability spontaneously as they recover after the acute stroke episode. Thus, in case series following the treatment of a group of stroke patients after the event, there is a high potential for time confounds; this makes it extraordinarily difficult to interpret results of such studies because they do not conclusively show what caused the recovery of these patients. Similarly, in degenerative neurological disease, swallowing function could deteriorate spontaneously, thus masking the effectiveness of treatment. Only controlled studies can resolve these problems.

Also, only controlled studies can determine whether a particular treatment is best for any given patient population. Although we have argued in the Methods section of this report that comparing the efficacy of different treatments is not as interesting as comparing the efficacy of different diagnosis and treatment programs, we also recognize that refinements in treatment strategies will most likely come about as a result of controlled trials comparing treatment efficacy. In some situations (e.g., patients with advanced Alzheimer's disease), it may even be desirable to determine whether treatment is more effective than doing nothing at all.

In other situations, it would not be ethical to include a control group that receives no treatment if it has already been found that patient condition improves with treatment, and there is some limited evidence in the literature that this is the case. In the case of diet modification, there is enough evidence of treatment effect in some patients that employing an untreated control group would not be ethical. In cases where it is unethical to include an untreated control group, the control group would have to be either a group receiving a different level of treatment or a historical control. Historical controls have been used in some of the treatment studies evaluating the implementation of a dysphagia treatment program. There are, however, limitations to this type of control group, as there is no way to ensure patient or management equivalency, and patients are not chosen in the same way (consecutive versus record review). However, a historical control within the same institution is preferable to controls from a different institution. One study has been conducted comparing different levels of intervention (DePippo, Holas, and Reding, 1994 ); in such a study design with randomization, it could be ensured that patient characteristics were similar in each trial arm, and therefore not affecting results. We therefore suggest that all future research evaluating treatment efficacy include at least two randomized treatment arms in each study so that even if absolute efficacy cannot be determined, relative efficacy can.

We also recognize that refinements in diagnosis will come about primarily as a result of trials that compare the efficacy of different diagnostic modalities. Designing such studies at the current time is problematic. Determining the performance of a diagnostic method (sensitivity and specificity) requires comparison to a gold standard diagnostic that identifies every positive and negative case of disease correctly. It has not been satisfactorily demonstrated that any of the available tools are, in fact, gold standards. Further, determining diagnostic test performance requires that false-positive rates be measured. This, in turn, means that some patients who are diagnosed as having a swallowing disorder should not receive treatment. Therefore, the only ethical kind of studies comparing diagnostic test performance would have to be conducted on a patient population known not to benefit from treatment. Because disease in such patients may be more severe, it is not clear that sensitivities and specificities from such studies would be generalizable to other patient populations.

Partly for these reasons, determining the percent agreement between a diagnostic test and the modified barium swallow (MBS)is not an appropriate measure by which to gauge the efficacy of the first test. It is, after all, possible for the results of two relatively poor tests to perfectly agree with each other, and high correlations between the results of two tests do not mean that the sensitivities and specificities of the two tests are similar. This latter point can be illustrated by a hypothetical experiment in which two tests are given to ten patients. Both tests yield one false positive and four true negatives, and therefore have a specificity of 80 percent. However, one of these tests yields four true positives and one false negative, while the other yields three true positives and two false negatives. The correlation between these tests (0.9) is relatively high, but the sensitivity of the former test in 80 percent and that of the latter 60 percent, a difference that could be clinically important in spite of the apparently "good" correlation between the two tests.

Outcomes

Meaningful outcomes to consider in any study are those outcomes that measure aspects of health that are important to the patient. However, it is also important to clearly report these outcomes. It has been reported that aspiration leads to increased risk of aspiration pneumonia, and incidence of pneumonia has been the most commonly reported long-term outcome in any of the studies reviewed in this report. However, it was uncommon to find aspiration pneumonia distinguished from general pneumonia when reported as an outcome in the dysphagia literature. It would be interesting to see studies explore the changing rate of pneumonia over time after an acute stroke. With this information, studies following patients for different lengths of followup can be compared using meta-analysis, decision analysis, or other mathematical calculations. Malnutrition, quality of life (QOL), and disease-specific mortality are other potentially important outcomes of interest that have been seldom reported.

The relationship of malnutrition to dysphagia has not been definitively determined, and this relationship needs to be explored. If it does turn out that malnutrition is a serious health concern resulting from dysphagia, then this endpoint should be reported as an outcomes measure in evaluation of dysphagia diagnostic and treatment technologies. Thus far, only two studies have been conducted, both on nursing home patients (Keller, 1993; Keller, 1995; Thomas, Verdery, Gardner et al., 1991); this research should be expanded to include patients in other care settings with different severity of disease.

QOL is an important, often neglected, measure that should be more seriously considered by researchers. No diagnostic or treatment program is worth doing if it results in a patient who is just as physically, socially, or psychologically impaired as before the program was undertaken. QOL is a subjective measurement that can be made in numerous different ways, and no one measurement method has been judged to be superior, thus making this endpoint difficult to address. However, it should not be ignored.

Many studies have been conducted with such a short followup period that mortality is not addressed. In those cases where mortality has been addressed, overall mortality has generally been the only measure. It would be interesting to know the causes of mortality in dysphagia patients; obviously, death from pneumonia is a serious consideration. Another question of interest may be whether a general weakening of the patient's system increases the risk of dying from other causes. If the link between dysphagia and malnutrition is substantiated, then this is an important concern as malnutrition in the elderly has been found to weaken the immune system (Chandra, 1989, 1990). These possible links need to be explored and reported.

Other short-term outcomes of interest have seldom been reported specific to particular treatments. In the field of noninvasive swallow therapy using different postural techniques and exercises, it would be interesting to see more on changes in nutritional intake (as measured by kilocalories) or weight changes. In the field of diet modification, short-term outcomes seldom reported include the ability to maintain the recommended diet safely, nutritional intake, and weight changes. Elimination of aspiration is a specific measure of interest, which should be evaluated over an extended duration of treatment; many studies have measured it during a barium swallow as the treatment is initially being tested, but few have followed this outcome for these patients after they have attempted to use the treatment independently.

Investigators should attempt to identify particular symptoms that could be used to determine an appropriate treatment plan. It is not clear whether one set of signs and symptoms would serve all patients, or whether different signs and symptoms should be used for patients with different primary diagnoses. It is also not clear that the same set of signs and symptoms should be acted upon in the same way for patients at different stages in their disease. For example, a given set of signs and symptoms in patients with early-stage Alzheimer's disease might lead to one course of action whereas the same set of signs and symptoms in a patient with advanced Alzheimer's disease may lead to another course of action, if any action at all is taken.

With such measures, studies could be conducted over a short period of time and still be able to report meaningful treatment response measures.

Followup

To compare the outcomes of any two groups of patients, they need to have been followed for the same length of time after the onset of disease. This is because the risks of morbidity and mortality resulting directly from that disease will change as time progresses and the total incidence of these endpoints will change. Similarly, the number of patients experiencing morbidity and mortality will accumulate over time. Therefore, the outcome of a patient followed for 4 weeks after stroke cannot be compared with a patient followed for 8 weeks.

Unfortunately, most studies on dysphagia management published to date have followed patients for a mean length of followup; in acute care, this is often simply the length of stay in the hospital, which is different for each patient. However, even in other care settings, followup time has not been standardized.

Any meaningful analysis comparing study results without such standardization is impossible. It clearly is very difficult for a researcher to follow patients once they have left the inpatient care setting, and obvious confounds arise when a care setting changes. Analysis would then need to take into account care setting, and perhaps compare the outcomes of patients released to the community versus those sent to continuing care (either in rehabilitation or a nursing home).

Clinical Signs and Symptoms

Individual signs and symptoms for prediction of pneumonia during noninstrumented exam have been found to be unreliable; all tested symptoms have been found to have either a very low sensitivity or a low specificity. However, researchers have not examined the co-occurrence of multiple symptoms to find an algorithm that may successfully predict pneumonia. Such research is suggested for the future in order to isolate those patients who would be best served by undergoing more extensive, instrumented diagnosis and treatment.

Some individual symptoms, in particular dysphonia, have been found predictive of aspiration. As with pneumonia, however, co-occurrences of several symptoms have not been fully explored, and such research may produce a method of noninstrumented diagnosis that is accurate enough to selectively choose the appropriate patients at risk to undergo further testing.

Diagnosis

There is currently no conclusive evidence about the superiority of instrumented diagnostic tools over noninstrumented ones. It has been assumed by clinicians that imaging technology such as videofluoroscopy or fiberoptic endoscopy must be superior to a noninvasive BSE because of additional information provided. However, published research has yet to prove this superiority. Researchers must document what information provided by instrumentation but not by BSE results in better outcomes for patients.

It would also be interesting to see if there are any symptoms on BSE that are reliably predictive of particular physiological dysfunction. As there is currently some limited information on the relationship of particular physiological abnormalities and appropriate treatment, such BSE predictive value would make it a better diagnostic tool in isolation.

Treatments

Most treatment trials on noninvasive swallow therapy have thus far examined results of the program overall, rather than broken out by specific techniques. For example, a study may report that patients underwent strengthening exercises or postural techniques, and then report outcomes for all these patients together. It would be interesting to see long-term results after individual techniques, such as the chin tuck or Mendelsohn Maneuver. (Such results have been reported for the palatal training appliance and tactile-thermal stimulation.) Such studies, as discussed above, would have to be constructed with a control or comparison group of some sort, either a historical control or comparisons among different intensities of treatment; comparison of different therapies is inappropriate unless patients with identical indications could appropriately be referred for completely different treatments.

Clinical studies examining outcomes with the use of feeding tubes have demonstrated a lot of problems. The most obvious problem has been that the overwhelming majority of these studies did not isolate patients with dysphagia from patients put on a tube for other causes. Thus, the results of those with dysphagia are mixed with those suffering from dementia or paralysis that makes self-feeding difficult. It is then impossible to determine what the effects of tube feeding are for dysphagia patients specifically. Only two studies have been identified that examined dysphagia patients specifically [Norton, Homer-Ward, Donnelly et al., 1996; Spiegel, Creed, Selber et al., unpub.(a)]; one reported mortality and the other weight change. More of this type of research is recommended.

As with the other treatment literature, the feeding tube literature suffers from inconsistent followup times. The literature would benefit if results were reported using survival curves (e.g., Kaplan-Meier curves) so that the percentage of mortality, morbidities, and complications were standardized on the basis of the number of patients actually followed for the specific time intervals. We also suggest that the rates of all possible feeding tube complications be reported, even if none occurred, because currently these complications are inconsistently reported and it is unclear when they are not reported whether none occurred or whether the researchers did not consider them important.

Trial Design

The trial we suggest is a multicenter, multiarm trial, with patients randomly assigned to each arm. Randomization should be accomplished by accepted means (e.g., by using a table of random numbers), and not according to patient hospital number, the day of the week on which patients arrive at the hospital, or other means not accepted by methodologists.

This trial can consist of two to four groups, depending upon the number of questions one wishes to answer. For the purposes of this description, we will describe the trial as if it contained four groups, with the understanding that the number of groups can be reduced.

Patient Population

The patient population should be as homogeneous as possible. Thus, only patients in a certain age range should be enrolled. This is required inasmuch as very old patients may have had more co-morbidities (or have been otherwise more debilitated) than relatively younger ones. Additional patient inclusion criteria may also be desirable to ensure that the study is conducted on a homogeneous group. Regardless of the specific criteria, however, the characteristics of the enrolled patients should be recorded to determine, after the conclusion of the trial, whether the patients in any of the four groups were different from those in other groups.

Patients should be consecutively enrolled. We recommend that the trial be limited to patients with a single disease, preferably stroke. This ensures that the trial will study the population most likely to benefit from dysphagia programs. It also makes accruing patients to the trial easier than it would be if other patients were studied. This means that the trial can be concluded more rapidly. Further, we are unable to estimate how effective dysphagia diagnosis and treatment programs are in other patient populations. Enrolling patients from these other populations could, therefore, have an impact on our statistical power calculations (see below), making it possible that many more patients than we have specified might need to be enrolled. In practice, this could yield a worst-case scenario in which the trial found no effect because it was underpowered.

Diagnosis and treatment should begin at a uniform time, and as soon as is practical. Delaying diagnosis and treatment too long would mean they would be offered to patients increasingly less likely to benefit from a dysphagia program. Because the trial should be restricted to stroke victims immediately after the stroke event, only acute-care centers should participate.

Diagnostic Methods

All patients in each arm undergo the dysphagia diagnostic method to which they are randomized, regardless of whether they are exhibiting clinical symptoms of dysphagia. In the first arm (the control group), all patients receive diagnosis by a noninstrumented test and then treatment. Although it is commonly believed that noninstrumented tests are inferior to instrumented diagnostics, no definitive evidence of this exists and, in fact, there is some limited evidence that dysphagia programs using the BSE are effective in preventing aspiration pneumonia [Odderson, Keaton, and McKenna, 1995; Spiegel, Creed, Selber et al., unpub.(a)]. Therefore, there is no compelling evidence to suggest that inclusion of such a group would be unethical. If suspicions about such a group remain, it is important to understand that the trial we describe would be subject to the normal ethical stopping rules. Thus, accrual to this group could be immediately terminated should it be found during the trial that patients in this group fared significantly worse than did patients in another group. At the same time, we strongly recommend that this group be included, no matter how many groups the trial ultimately contains. Inclusion of this group answers a most fundamental question: Are instrumented exams more efficacious than structured formal BSEs?

Patients in the other three arms would also receive the noninstrumented test. The results of this exam, however, would not be used in patient management. In fact, physicians and other caregivers would be blinded to the results of this test (the reason for these BSEs are described below). We recommend blinding of providers to the results of these BSEs to ensure that interpretation of the results of the subsequent tests is not biased by the results of the BSE.

These patients would then randomly receive diagnosis with a single instrumented test (to be chosen by the researchers). Numerous instrumented tests are currently used in dysphagia management, and the clinical literature is currently equivocal about the superiority of any one over any other. We do not specifically recommend the inclusion of any particular instrumented tests, because to include only the widely used tests might be construed as exclusionary of new, emerging (and potentially superior) technologies; however, to recommend the newer technologies would discount the refinement and extensive development of the established technologies. We therefore defer to the clinical establishment to choose which instrumented exams to include. Treatment would be determined based upon the results of these diagnostic tests.

Because all patients in this trial receive noninstrumented diagnosis, it is important that this exam be standardized across centers. The specific components of this exam could be determined by a panel of experts (which, if desired, could include patient representatives). We recommend, however, that this exam include (but not be limited to) assessment of patients' oral hygiene habits, dysphonia, the ability to voluntarily cough (rated on a four-point scale), and the 3-ounce water test. As discussed previously in this report, limited data suggest these tests are effective at predicting aspiration. For reasons that will become apparent below, a thorough and standardized patient history should be taken as part of this test. Again, the expert panel could determine specific elements of this history.

Treatments

As with the diagnostics discussed above, treatments in different centers should also be standardized to the greatest extent possible; standardization should be accomplished by the same expert panel that standardizes the diagnostic methods. This is not meant to imply that all patients receive the same treatment; rather, patients are assigned to treatments based on symptomatology. However, the methods of determining the appropriate treatment (i.e., the symptom-based treatment choices) should be the same across the different diagnostic methods. We recognize that this is likely to be problematic. However, it is important to accomplish this standardization to reduce the variability of the trial's results (thus increasing its statistical power) and to ensure that apparent effects of diagnostics are not confounded by treatment differences.

Because this standardization should ensure that centers or individuals providing therapies are equally well trained, it is important that none of the sites (or providers) are in a startup phase of speech-language therapy, when results might be poorer than those obtained after more experience is gained.

Outcome Measures

The primary outcome measure of the trial should be pneumonia, as it is the most serious morbidity that may result from dysphagia. We recommend this as the primary outcome for several purely pragmatic reasons. First, the cost of treating pneumonia is among the greatest of the costs of treating morbidities. Second, although death rates due to dysphagia are of extreme importance, one could obtain low death rates by curing pneumonia, not by preventing it. Thus, any effect of dysphagia diagnosis and treatment programs observed in such a trial would be contaminated by the effectiveness of pneumonia treatment. Finally, there is so little information on other morbidities resulting from dysphagia, one cannot, on the basis of current data, even be assured that these other morbidities pose a major health problem. In fact, one of the purposes of this trial is to provide information about the health problem caused by these other morbidities. Thus, secondary outcomes should include, but not be limited to, measures of: (a) the number of patients placed on feeding tubes, (b) weight change, (c) body mass index, (d) serum albumin (e) dehydration (measured by the blood urea nitrogen:creatinine ratio), (f), morbidities resulting from feeding tubes, (g) dysphagia-related mortality, and (h) all-cause mortality. Data should be analyzed on an intent-to-treat basis.

The results of the BSEs can also be considered among the outcomes of this trial. These results will be used to provide two kinds of information. First, they will allow one to determine whether any signs and symptoms, and particularly any combination of signs and symptoms, predict morbidity (and thus whether the BSE can be used as a stand-alone diagnostic). Second, these results will allow one to determine, if warranted, which BSE results can be used to selectively refer appropriate patients for subsequent instrumented diagnostic tests. It may be possible to pool the BSE results from the three groups that receive subsequent diagnostics to enhance the statistical power of the analyses that will be required; such analyses will involve multivariate statistics, which are lower-power tests than univariate analyses. The fact that analysis of these data will be conducted using a form of multivariate statistics (and, more specifically, a form of multiple regression analysis) means that the results of any one trial, including the one we describe here, will not provide results completely generalizable to all settings. It is well-known that multiple regression equations conducted in one setting are less predictive when used in another (this reduction in predictive ability is termed shrinkage). It is even possible that some variables that appear to be predictive in this clinical trial will not be found predictive in later work. In other words, it is possible that the multivariate results obtained in a well-controlled clinical trial that employs a strict protocol may not be entirely generalizable to actual clinical practice, where protocols are often less strict. This does not imply that the results of these analyses are not worthwhile but it does mean that they will have to be checked, and possibly further refined based on how predictive these signs and symptoms are in actual clinical practice.

Followup

Patients should be followed for a uniform period of time, at least 1 year. This requires active monitoring of patients after they leave acute care. Following patients for an average (or mean) length of time is inappropriate. Means are distorted by outliers, and would be misleading if a few patients or any group(s) were followed for abnormally short or long periods. We recognize the difficulty in performing this type of followup, so it will most likely be prudent to offer patients remuneration to enhance compliance.

Even patients who receive negative diagnoses for swallowing disorders should be followed. This is required to ensure that results are analyzed on an intent-to-treat basis.

Statistical Power Issues

Table 20. Number of Patients Required in Proposed Trial (more...)

Table 20. Number of Patients Required in Proposed Trial

Baseline pneumonia rate	Program rate	Percent decrease	N per group	Total N 2 groups	Total N 3 groups	Total N 4 groups
2%	1.5%	25%	10,795	21,590	32,385	43,180
2%	1.0%	50%	2,318	4,636	6,954	9,272

The lefthand column displays the rate of aspiration pneumonia that, for the purposes of our statistical power calculations, we assume will occur in patients whose treatment is based on only the results of the BSE (see our discussion of Question 1 in the Evidence Report and our first Supplemental Analysis for further information about this rate). In the next column from the left, we portray two hypothetical pneumonia rates that might occur in a dysphagia diagnosis and treatment program that employs an instrumented diagnostic. These hypothetical rates translate to 25 percent and 50 percent reductions in the rate of aspiration pneumonia, as shown in the third column from the left. The next column shows the number of subjects needed in each group of the trial for each of the two hypothetical reductions in pneumonia rates. The final three columns show the total number of patients the trial must contain for each of these two rate reductions for a two-, three-, and four-group trial, respectively. Thus, if a program results in a 25 percent decrease in the rate of aspiration pneumonia, 10,795 patients will be required for each group, meaning that the trial must contain 21,590 to 43,180 patients, depending on the number of groups. Similarly, for a program resulting in a 50 percent reduction in the pneumonia rate, 2,318 patients per group will be required, meaning that the trial must contain 4,636 to 9,272 patients.

For these preceding calculations, we sought a trial that would give 80 percent statistical power at an alpha level of 0.05, and assumed that ² would be the test statistic (the reasons for this assumption are discussed below).

For this area of research, this is a relatively large trial. However, it is not large when compared with the size of other trials reported in the medical literature. The size of this trial, however, is one reason for our recommendation that this trial be a multicenter trial. Using a number of centers will make it easier to enroll this number of patients and will decrease the length of time required to complete the trial.

In constructing this table, we have chosen a 25 percent decrease in pneumonia as the lowest effect size of interest on the assumption that a smaller decrease in the aspiration pneumonia rate would be cost-prohibitive. This in fact may not be the case, so we believe it prudent to describe the calculations we used to reach this conclusion. In this way, others can both evaluate our results and use our calculations to employ their own judgment about what is and what is not cost-prohibitive. These calculations are based upon this trial's costs for preventing one case of aspiration pneumonia.

In a trial containing two groups, each patient would receive a BSE at a cost of $141. (The sources of this cost information are described in our first Supplemental Analysis.) Half of the patients in the trial would receive an instrumented exam at a cost of $218, so the average cost of instrumented exams would be $109 per trial enrollee. We also assumed that about 38 percent of all patients would receive diagnosis-directed treatment (derived from the estimate of aspiration in this population by Daniels, Brailey, Priestly et al., 1998), at a cost of $242 per treated patient, which yields an average cost per enrollee of $92. Summing the costs yields an approximate cost of $251 per enrollee. If a dysphagia program employing an instrumented diagnostic decreases the rate of aspiration pneumonia by 25 percent (i.e., from 2 to 1.5 percent), then the program will prevent one case of pneumonia in every 200 enrollees in a two-group trial. This means that the total cost of preventing one case of pneumonia in such a trial is $68,400. Similarly, in a trial with three groups, the cost to prevent one case of pneumonia is $75,600, and in a trial with four groups, the cost is $79,400. Although these numbers are admittedly crude, they far exceed the highest estimate of the cost of treating pneumonia, $19,000, in the literature (Aviv, unpub.). Hence, our assumption that a 25 percent program-related decrease in pneumonia is the smallest effect of interest is probably a liberal one.

Our estimates of the number of patients required for this trial are highly sensitive to the pneumonia rate predicted to occur on those patients whose treatment is determined solely by the BSE and to the anticipated magnitude of the effect of a dysphagia program. For example, the above table assumes that an in instrumented diagnostic will decrease pneumonia rates to 1 percent (instead of 1.5 percent), meaning that a four-group trial will require 33,908 fewer patients. Similarly, if we assume a 4 percent base pneumonia rate and, as above, a 25 percent decrease as a result of employing instrumented diagnostics, then only 21,100 patients are needed for the four-group trial, which is 22,080 fewer than if the base rate is 2 percent. This sensitivity arises because the power function is relatively steep in the area in which we are working, and not as a result of any inaccuracies in our calculations. Because of this sensitivity, it is wise to consider a pilot study to better estimate the base pneumonia rates and the anticipated difference in pneumonia rates between patients who receive only the BSE and those who receive an instrumented exam.

Our calculations of the number of patients needed for this trial assume that the statistical analysis of the data will be conducted to maximize power. This reduces the number of patients required for the study, and without analyses that maximize power, this number could become prohibitively large. The implication is that focused contrasts should be used. Taking the four-group trial as an example, a relatively low-power analysis could result if the analysis were conceived as a ² analysis of a 2 X 4 design in which the four treatment groups are the column headings and the presence or absence of pneumonia are the row headings. On the other hand, it is possible to analyze these data using three orthogonal contrasts. These contrasts might, for example, ask whether the results of the BSE are different from those of all other tests, whether the results of the MBS exam are different from those of the two fiberoptic exams, and whether the results of the two fiberoptic exams are different from each other. By employing these contrasts, one reduces the degrees of freedom from three (for the omnibus ² test) to one, and there is a resulting increase in statistical power. Obviously, the focused contrasts can also be used in the three-group trial and are not relevant to the one group trial, which is already analyzed on 1 degree of freedom.

Some Final Remarks

Although the trial need not contain all four groups we have described, using fewer groups should be approached cautiously. As a rule, data obtained from a single trial that directly compares the diagnostic strategies of interest are stronger that those of several trials, each of which only makes some of the comparisons of interest. This remains true even if the aggregate of all trials makes all of the desired comparisons. However, this would be feasible if the same centers took part in several, smaller studies, using the same documented diagnostic and treatment criteria.

It is also possible to add additional groups to this trial, groups that incorporate combinations of diagnostics. Similarly, some diagnostic combinations may be substituted for the groups we have outlined here.

The results of this trial can be extended to answer other questions of interest, even though it (or any other feasible clinical trial) will likely contain too few patients to yield appropriate statistical power. This can be accomplished by using the data from this trial to construct a decision analysis. Such an analysis is further discussed in Appendix F.

Introduction

In this supplemental analysis, we conduct an analysis to determine whether dysphagia diagnosis and treatment programs are a cost-effective means to prevent aspiration pneumonia. As in our main analysis, we focused on stroke patients, because they are the largest group of elderly adults experiencing dysphagia. The clinical measure of effectiveness for this supplemental analysis is prevention of aspiration pneumonia, because it is the major cause of serious morbidity, mortality, and expense for patients with dysphagia. Also, as discussed in the main assessment, we were unable to find adequate literature to estimate the effect of dysphagia diagnosis and treatment on the other important patient outcomes of malnutrition, dehydration, and quality of life (QOL). Finally, we focused on the acute-care setting because patients are at greatest risk for aspiration pneumonia during this time, because the greatest efforts of dysphagia programs are at these times, and because most of the data required for cost-effectiveness analysis was for the acute-care setting. However, we also discuss the implications of our results for nursing home patients.

We performed this analysis by constructing a decision tree that specifically addresses the question of whether a typical program to diagnose and treat dysphagia is effective enough in preventing pneumonia to justify the cost of diagnosis and treatment. We first addressed the question for a dysphagia program that uses a preliminary bedside exam (BSE) followed by a full BSE for patients judged to have dysphagia by the preliminary exam. We next examined the cost-effectiveness of a program that uses a preliminary exam to refer dysphagia patients to a videofluoroscopic swallowing study (VFSS). To answer each question, we set up the decision tree that mimics a hypothetical randomized controlled trial of a cohort of acute stroke patients.

There are two primary arms of this decision tree. One arm models the historical situation of a typical acute-care hospital with no special dysphagia program. In this arm, the swallowing competence and needs of stroke patients are assessed informally by attending physicians, nurses, and dieticians without special training in diagnosis and treatment of dysphagia, and without the services of speech-language pathologists (SLPs) who are specially trained to diagnose and treat dysphagia. In this situation, a major involvement of physicians is in placing enteral tubes in patients judged to have a swallowing problem that could make it difficult for them to maintain hydration and nutrition or that could lead to aspiration of food and liquids. Patients judged less severely impaired may be provided with pureed food and possibly a soft mechanical diet including thickened liquids. The historical proportion of stroke patients who acquire aspiration pneumonia is used as the effectiveness measure of this arm. There is a branch for the proportion of patients who get aspiration pneumonia during a typical acute-care stay of 2 weeks. The tree counts the number of patients who get pneumonia and the number who remain free of pneumonia by setting the effectiveness payoff for the former patients at 0 and the effectiveness payoff for the latter patients at 1. Thus, the measure of effectiveness in this analysis is number of pneumonia cases prevented. The cost of pneumonia is calculated as the Medicare reimbursement rate for pneumonia. The expected value for this entire arm is calculated as the pneumonia treatment costs to Medicare per stroke patient.

Using Medicare reimbursements instead of true provider costs allows us to estimate Medicare's net costs for dysphagia diagnosis-treatment minus pneumonia costs. We could find no recent literature sources for comprehensive provider costs of dysphagia diagnosis-treatment and pneumonia, whereas Medicare reimbursement rates are known. If Medicare reimbursements underestimate the provider costs of treating pneumonia to a greater extent than dysphagia diagnosis-treatment costs, then any cost savings found by our analysis will be even greater for providers than for Medicare. Thus, our estimates will be conservative and for the minimum savings.

The second primary arm of the tree models a typical dysphagia program in the same setting as above. For the BSE program, clinicians, typically nurses, carry out a preliminary swallowing evaluation as described by Odderson et al. (Odderson, Keaton, and McKenna, 1995) (described in the main evidence report in the Results section under Question 1) and refer stroke patients suspected of dysphagia to SLPs specially trained to diagnose and treat dysphagia. These dysphagia specialists carry out a formal BSE. Treatment of these patients may include (alone or in combination) enteral intubation, positional maneuvers, strengthening exercises, and diet modifications such as the restriction of certain consistencies and the addition of thickening agents to certain consistencies. Patients also receive 15 minutes a day of therapy for 6 days to reinforce and maintain these measures. We took the incidence of aspiration pneumonia observed in recent studies of BSE dysphagia programs as the effectiveness of this arm. For the cost of this arm, we used the Medicare reimbursement rate for diagnosis and treatment of dysphagia and the reimbursement rate of treating pneumonia. Therefore, the costs associated with this arm are the net costs for dysphagia diagnosis-treatment and pneumonia per stroke patient. The decrease in the number of PEG tubes is also considered in the costs. However, rather than adding PEG costs to both arms, we merely subtracted the decrease from the dysphagia program arm.

The simple cost-effectiveness of each arm is calculated as the costs divided by the proportion of pneumonia-free patients. However, this is not a particularly useful measure, because we are interested in the costs for all patients. Of primary interest is the overall difference in cost-effectiveness between the two arms, called the marginal or incremental cost-effectiveness. This tells the cost-effectiveness of adding the dysphagia program to the routine care of stroke patients. The marginal cost-effectiveness between the two arms is calculated as the difference in total costs for each arm divided by the difference in effectiveness for each arm. This gives the cost (or savings) per pneumonia case prevented. These are the net direct medical costs (or savings) of preventing pneumonia.

It is important to place these medical costs in perspective, which can only be accomplished by adjusting the cost of pneumonia by some measure of QOL and the threat of death from pneumonia. Meaningful pneumonia-related QOL data are not currently available, so it is necessary for individual users of this analysis to employ their own data and judgment for this evaluation. Also, it is necessary to realize that because we were limited by published data to calculating only the pneumonia cost-effectiveness, one is forced to take pneumonia as representative of the other important patient outcomes such as dysphagia-related malnutrition, dehydration, and QOL. Dysphagia diagnosis and treatment costs relating to these have unavoidably been included here in our pneumonia analysis, while the costs of these additional patient outcomes and the savings for their prevention have not. Thus, the additional savings (monetary or QOL) from preventing or ameliorating these must be considered, although we were unable to explicitly calculate these.

Methods

We used the results and evidence tables of the main analysis as the probabilities and costs in the decision tree. The methods for searching the literature and for extracting and analyzing the data are described in detail in the Methods section of the main analysis. Briefly, the peer-reviewed literature was searched for publications containing information on the epidemiology, burden of disease, diagnosis, treatment and costs of stroke, aspiration pneumonia, and dysphagia in the elderly population. The titles and abstracts in the search results were examined independently by two analysts, and publications that appeared to have relevant data were ordered. The set of ordered publications was manually examined independently by two analysts to extract data, and bibliographies were examined for additional publications that were not in the searches. The only exclusion criteria were that the studies contain 10 or more patients and be conducted in the United States. In some cases where no appropriate data was available meeting those requirements, smaller studies and studies outside the United States were included, with appropriate indication of these exceptions.

The variables used in the decision trees are shown in S-1 and S-2. These same data, along with additional data provided for comparison purposes, are presented in the evidence tables at the end of the supplemental analysis.

The decision tree(s) were constructed, calculated, and illustrated using DATA^TM decision analysis software (version 3.0.18: TreeÂge Software, Cambridge, MA).

Cost-Effectiveness of a BSE Dysphagia Program Compared with No Program

Cost-Effectiveness of a VFSS Dysphagia Program Compared with No Program

Figure S-1. Cost-effectiveness Decision Tree for BSE

Figure S-2. Cost-effectiveness Decision Tree for BSE Program Showing Variable Base-case Values and Terminal Node Payoffs

Figure S-3. Cost-effectiveness Decision Tree for BSE Program Showing Expected Costs

Figure S-4. Cost-effectiveness Decision Tree for BSE Program Showing Expected Pneumonia Cases

Figure S-5. Tornado Diagram for BSE Dysphagia Program

No bars are shown for effect of swallow program on PEG rate or for cost of PEG placement because the dysphagia program is dominant and no cost-effectiveness ratio can be calculated for these variables.

Figure S-6. Two-way Sensitivity Analysis for Base Pneumonia Rate and Risk Reduction with BSE Dysphagia Program

Figure S-7. Two-way Sensitivity Analysis for Cost of Diagnostic Test and Cost of Swallow Therapy with BSE Dysphagia Program

Figure S-8. Two-way Sensitivity Analysis for Cost of Diagnostic Test and Risk Reduction with BSE Dysphagia Program

Figure S-9. Tornado Diagram for Dysphagia Program

Figure S-10. Two-way Sensitivity Analysis for Base Pneumonia Rate and Risk Reduction with VFSS Dysphagia Program

Figure S-11. Two-way Sensitivity Analysis for Cost of Diagnostic Test and Risk Reduction with VFSS Dysphagia Program

Figure S-12. Two-way Sensitivity Analysis for Cost of Diagnostic Test and Risk Reduction with VFSS Dysphagia Program

Epidemiology of Dysphagia Originating from Neurologic Conditions

Table 21. Neurologic Causes of Oropharyngeal Dysphagia (more...)

Table 21. Neurologic Causes of Oropharyngeal Dysphagia in the Elderly

Central nervous system cerebrovascular accidents/strokes Anterior cortical stroke Brainstem stroke Wallenberg's syndrome Pseudobulbar palsy Parkinson's disease Multiple sclerosis Alzheimer's disease Other neuromuscular disorders Motor neuron disease (MND) Polymyositis and dermatomyositis Muscular dystrophies Myasthenia gravis Hypothyroidism or hyperthyroidism Bulbar poliomyelitis Peripheral neuropathy (secondary to diabetes mellitus)

No information was found in a Medline search on the occurrence of dysphagia in bulbar poliomyelitis, muscular dystrophy, hyperthyroidism or hypothyroidism, peripheral neuropathy, or on the overall prevalence of these disorders in the U.S. population. Therefore, we will focus on those relatively common disorders in the elderly for which there are data on both the overall incidence or prevalence of the disorder and on the diagnosed occurrence of dysphagia in individuals with that disorder, including stroke (cerebrovascular accident, or CVA); Parkinson's disease, Alzheimer's disease, amyotrophic lateral sclerosis (ALS), and a few other less common degenerative neurologic diseases. Neoplasms are also not considered in this report.

Epidemiology of Stroke in the United States

Six published studies were located that reported the annual incidence rate¹ of CVA in the United States using various measurement techniques (Barker and Mullooly, 1997; Brown, Whisnant, Sicks et al., 1996; Kittner, White, Losonczy et al., 1990; McGovern, Burke, Sprafka et al., 1996; Modan and Wagener, 1992; Sacco, Boden-Albala, Gan et al., 1998). Prevalence rates cannot be examined for CVA, because it is an event rather than an ongoing disorder. The prevalence of cerebrovascular disease, on the other hand, can be examined; however, we are primarily interested in dysphagia that occurs shortly after an acute accident. Cerebrovascular disease overall was reported in the 1995 National Nursing Home Survey as the primary admitting diagnosis in 2.4 percent of elderly individuals (Dey, 1997). It is the third leading cause of death among individuals 65 and older (Department of Health & Human Services, 1997 ).

A sex discrepancy, in which women suffered fewer strokes than men, was noted in all studies reporting these statistics. Within each of these studies, there is a consistent trend that the stroke risk ratio for women to men is approximately 1:1.5; this seems to indicate that men have a 50 percent higher risk of having a stroke than do women.

Table 22. Annual Incidence of Stroke per 100,000 Individuals (more...)

Table 22. Annual Incidence of Stroke per 100,000 Individuals Age 65 to 74 Years

Study	Women	Men
McGovern, Burke, Sprafka et al., 1992	514.9 (age 70-74)	818.2 (age 70-74)
Brown, Whisnant, Sicks et al., 1996	524	885
Gillum and Ingram, 1996	1,172 to 1,805 (depending on geographic region) per 100,000 person-year at risk	1,676 to 2,315 (depending on geographic region) per 100,000 person-year at risk
Barker and Mullooly, 1997	736	953

Table 23. Annual Incidence of Stroke per 100,000 Individuals (more...)

Table 23. Annual Incidence of Stroke per 100,000 Individuals Age 85 Years and Over

Study	Women	Men
Brown, Whisnant, Sicks et al., 1996	2,574	1,563
Barker and Mullooly, 1997	1,522	1,816

To summarize, the annual incidence of stroke in the United States has been estimated to be between 145 and 290 per 100,000 individuals. Caucasians have a lower incidence than African-Americans and Hispanics, and women have lower incidence than men, except among the oldest elderly. Incidence increases with age: populations age 65 to 74 show incidence of 515 to 2,315 per 100,000; those age 85 and over range from 1,522 to 2,574 per 100,000. Rates vary depending on the study methodology (hospital survey versus medical records/physician survey) as well as population differences (age, sex, race, and place of residence). It therefore appears that both environmental and genetic influences affect risk for a CVA.

Epidemiology of Parkinson's Disease in the United States

The methods of these three studies are fairly distinct from one another. Lilienfeld et al. (1990), who reported the lowest rates, reviewed acute-care hospital discharges over a five year period (excluding individuals over 75 years of age) (Lilienfeld, Sprafka, Pham et al., 1991). Mayeux et al. (1995) established a community registry by polling health maintenance organizations, hospitals, and nursing homes; they interviewed everyone (except those over 75 years old) suspected of having Parkinson's disease for confirmation (Mayeux, Marder, Cote et al., 1995). Mitchell et al. (1996), who reported the highest rates, focused solely on nursing home residents, examining ICD-9 diagnoses (Mitchell, Kiely, Kiel et al., 1996). Their rates ranged from 45 per 100,000 (Lilienfeld, Sprafka, Pham et al., 1991) to 2,250 per 100,000 (Mitchell, Kiely, Kiel et al., 1996) when only primary Parkinson's disease is considered (Parksinson's disease not secondary to any other disorder). It is clear from these studies that the elderly experience more Parkinson's disease, and those in nursing homes have an even higher rate. One study (Lilienfeld, Sprafka, Pham et al., 1991) reported that women have a lower rate than men (45 versus 65 per 100,000).

Epidemiology of Alzheimer's Disease

Patients with Alzheimer's disease and other degenerative brain disorders made up 2.7 percent of all nursing home admissions in the 1985 National Nursing Home Survey (Hing, Sekscenski, and Strahan, 1989) (This was classified separately from senile dementia.) Two other epidemiological studies were found for the United States, one which looked at the population in general through review of clinical records, and one which focused on the African-American community in a cross-sectional interview study (Beard et al. 1991), using records from the Mayo Clinic record linkage system, identified all medical cases of Alzheimer's disease on prevalence date January 1, 1980. They reported age- and sex-adjusted prevalence rates of 259.8 per 100,000 overall and 2,634.1 per 100,000 people age 65 and older. Age-specific information indicated that this rate rose from 236/100,000 in those age 65-69, to 18,018/100,000 in those aged 85 and older (Beard, Kokmen, Offord et al., 1991). The Mayo Clinic serves a 96 percent Caucasian community; however, this was the only study on the overall population, and therefore is the study we will use for calculation of burden of illness.

Epidemiology of Dysphagia in Other Degenerative Neurologic Diseases

Dysphagia in Stroke

The annual incident stroke population, as calculated above, constitutes 385,000 to 769,000 people each year, depending on which study estimate is used. The rate of dysphagia within stroke has been calculated in various ways; for the purposes of this discussion, we will examine the occurrence of dysphagia as measured by videofluoroscopic swallowing study (VFSS) and bedside examination (BSE) separately. In general, more patients are diagnosed dysphagic when a VFSS is used than when a BSE is used.

Several studies have also examined the rate of dysphagia in hospitalized stroke patients as measured by BSE, water swallow test, or structured observation. Again, these studies report widely varying rates but do not appear to have study or patient characteristics that account for the differences. The median rate reported by those studies examining hospitalized patients within 5 days after stroke was 41.7 percent (Kidd, Lawson, Nesbitt et al., 1993). This calculates to an overall burden of illness of 160,000 to 320,000 individuals per year.

For elderly stroke patients specifically, the occurrence of dysphagia appears to be higher than the population as a whole, when measured by BSE. More than 279,000 people over age 65 suffer from stroke each year [Barker and Mullooly, 1997; Centers for Disease Control and Prevention (CDC), 1998]. One study examined specifically the rate of occurrence of dysphagia immediately after the stroke event in patients age 70 and older; using BSE, the rate was 35.5 percent (Barer, 1989). This calculates to a burden of illness of about 99,165 elderly stroke patients with dysphagia each year.

Malnutrition in Stroke

The results from a single, European study suggest that within 1 week of an acute stroke, 26.4 percent of patients are malnourished (Davalos, Ricart, Gonzalez-Huix et al., 1996). This calculates to 101,551 to 202,891 possible cases of malnutrition occurring within 1 week after an acute stroke.

The same European study as discussed above reported that 48.4 percent of all stroke patients with dysphagia became malnourished within 1 week of the event (Davalos, Ricart, Gonzalez-Huix et al., 1996). If we assume that 160 to 320 stroke patients have dysphagia, then the burden of illness resulting from malnutrition in dysphagia is 77,636 to 155,110 individuals.

Aspiration in Stroke

Aspiration, in particular, is a condition that can lead to serious morbidity, especially in stroke patients. As stated above, stroke affects 385,000 to 769,000 people each year. Several studies have examined the occurrence of aspiration shortly after the stroke incident. Of studies reporting this rate in acute hospital strokes within 5 days after the event as measured by VFSS or modified barium swallow (MBS), the median reported rate is 33.5 percent (an average from two median studies reporting 32.2 percent and 34.7 percent) (Daniels, McAdam, Brailey et al., 1997; Kidd, Lawson, Nesbitt et al., 1993). This calculates to an annual burden of illness of 128,862 to 257,457 stroke patients with documented aspiration on VFSS.

Pneumonia After Stroke

Pneumonia Mortality

Parkinson's Disease

Only one study looked at the incidence rate of Parkinson's disease for the entire population (all ages and both sexes): Mayeux et al. (1995) reported an incidence rate of 13 per 100,000 per year (Mayeux, Marder, Cote et al., 1995) from a Mayo Clinic study. As there are more than 265 million people in the United States (Department of Commerce, 1997), the number of people affected by Parkinson's disease each year is 34,487.

The diagnosed occurrence of dysphagia in Parkinson's disease has been reported by several studies, using both videofluoroscopy and noninstrumental methods. Of two studies using videofluoroscopy that withheld L-dopa on the day of testing (which can affect dysphagia), the mean rate, 69.1 percent, was calculated (Bushmann, Dobmeyer, Leeker et al., 1989; Fuh, Lee, Wang et al., 1997). This indicates that of the 34,487 individuals affected by Parkinson's disease each year, 31,394 of them demonstrate dysphagia on videofluoroscopy. Data are not available to determine burden of illness resulting from malnutrition, pneumonia, or death.

Motor Neuron Disease

The annual incidence rate of MND, reported by Lilienfeld, Sprafka, Pham et al., (1991), was 6 per 100,000 in 1984, the most recent statistics available. In a population of more than 265 million, this means that 15,917 individuals are affected by MND each year.

There was only one potentially reliable number on the rate of dysphagia in MND, provided by Leighton, Burton, Lund et al. (1994), who reported that 51.2 percent of MND patients demonstrated swallowing graded as moderate or poor on videofluoroscopy. The total annual burden of illness is therefore 8,150 individuals affected by dysphagia in MND.

Amyotrophic Lateral Sclerosis

The most common form of MND is ALS. The annual incidence of ALS has been reported at 1.8 per 100,000 (Annegers, Appel, Lee et al., 1991). Thus, the total number of people affected by ALS in the United States each year is 4,775.

Only one study examined the rate of dysphagia in ALS. Leighton, Burton, Lund et al. (1994) reported a rate of 29 percent of poor or moderate swallowing as demonstrated on VFSS. Thus, 1,385 people are affected by dysphagia resulting from ALS.

Progressive Supranuclear Palsy

The annual incidence rate of PSP has been reported at 1.10 per 100,000 (Bower, Maraganore, McDonnell et al., 1997) (the prevalence has not been reported). Thus, approximately 2,918 people are affected by PSP each year.

Only one study reported the rate of dysphagia in PSP; Litvan, Sastry and Sonies (1997) reported that 55.6 percent of PSP patients complained of swallowing problems. This calculates to 1,622 people each year experiencing swallow problems as a result of PSP.

Huntington's Disease

The annual rate of new cases of Huntington's disease is 0.2 per 100,000 (Kokmen, Ozekmekci, Beard et al., 1994). The burden of illness is therefore approximately 531 each year in the U.S. population. The diagnosed occurrence of dysphagia was reported by a single study, at 100 percent of all Huntington's disease patients (Kagel and Leopold, 1992). Thus, the total number of new dysphagic Huntington's disease patients each year is 531.

Summary

Approximately 51,435 people this year will be affected by dysphagia resulting from these neurologic disorders. The large majority of these cases are the result of Parkinson's disease (more than 34,000). Combined with stroke, approximately 340,000 to 625,000 people will be affected by dysphagia resulting from a variety of neurologic disorders this year. Many of these patients may subsequently be affected by aspiration or pneumonia as the result of a disordered swallow although, presently, lack of data in the published literature makes the calculation of these rates impossible. If, however, we were to assume that the rates of aspiration and pneumonia were the same for all neurologic patients as for stroke, then we would estimate that approximately 14.3 percent of these patients with dysphagia would contract pneumonia (48,390 to 89,340 cases of pneumonia each year). If pneumonia death occurs at a rate of 19.8 percent, then approximately 9,581 to 17,689 patients will possibly die as a result of their untreated dysphagia.

To identify swallowing disorders not associated with cancer

• 1

  explode deglutition disorders or deglutition

• 2

  (tw) all swallow: or dysphagia

• 3

  neoplasms(px) [(px) indicates a pre-explosion of a MeSH term]

• 4

  1 or 2

• 5

  4 and not 3

• 6

  5 and aged(px)

• 7

  6 and human

• 8

  7 and not case report

To identify documents related to barium swallow in stroke patients

• 1

  barium sulfate or barium adj sulfate

• 2

  explode deglutition disorders

• 3

  explode cerebrovascular disorders or stroke(tw)

• 4

  1 and 2 and 3

To identify documents related to barium swallow in patients over 65

• 1

  barium sulfate or barium adj sulfate

• 2

  explode deglutition disorders

• 3

  aged(px)

• 4

  1 and 2 and 3

To identify therapies

• 1

  explode deglutition disorders or explode deglutition

• 2

  (tw) all swallow: or dysphagia

• 3

  1 and di&(px) [this is a pre-explosion of a subheading category]

• 4

  1 and th&(px)

• 5

  3 or 4

• 6

  1 or 2

• 7

  6 and <speech therapy or speech-language pathology>

• 8

  6 and <barium sulfate or barium(tw) or fluoroscopy or cineradiography or all videofluoroscopy:(tw)>

• 9

  5 or 7 or 8

• 10

  9 and aged(px)

• 11

  10 and human

To identify complications associated with and costs of PEG tubes

• 1

  gastrostomy

• 2

  percutaneous(tw) and endoscopic(tw) and gastrostomy(tw) or <PEG(tw) and all tube:(tw)>

• 3

  1 and instrumentation(sh)

• 4

  2 and equipment a#d supplies(px)

• 5

  3 or 4

• 6

  1 or 2

• 7

  6 and <fatal outcome or treatment outcome or mortality or survival or explode morbidity>

• 8

  6 and explode economics

• 9

  7 and aged(px)

• 10

  5 or 7 or 8 or 9

• 11

  10 and human

• 12

  neoplasms(px)

• 13

  11 and not 12

• 14

  13 and not <infant(px) or child(px)>

To assess issues related to enteral nutrition

• 1

  enteral nutrition or intubation, gastrointestinal

• 2

  1 and <nasogastric(tw) or nasointestinal(tw)>

• 3

  1 contains NG

• 4

  2 or 3

• 5

  4 and <adverse effects(sh) or complications(sh) or mortality(sh) or hazard:(tw) or safe(tw) or safety(tw)>

• 6

  5 and not child(px)

• 7

  gastrostomy and <percutaneous(tw) or PEG(tw)>

• 8

  6 and not 7

• 9

  8 and human

To assess epidemiology and economics of various disorders and their relation to dysphagia