Chapter  121:  Use of Spirometry for Case Finding, Diagnosis, and Management of Chronic Obstructive Pulmonary Disease (COPD)

Question 1 What is the prevalence of chronic obstructive pulmonary disease (COPD) and airflow obstructions in various adult populations as defined by: 1) spirometry and 2) clinical examination?

Diagnosis and case-finding recommendations for spirometric testing include all adults with a history of exposure to risk factors including current and former smokers and any adult with persistent respiratory symptoms of cough, phlegm, wheeze, or dyspnea. Because smoking is the main risk factor in causing COPD, the analytic framework begins with adults presenting to a primary care clinic where an assessment of COPD risk factors (smoking and symptom status) is performed. Decision nodes are based on smoking and respiratory status. Spirometry characterizes an individual as having airflow obstruction (and the stage of severity) while history and physical examination assess the presence or absence of signs or symptoms. Among former and current smokers, spirometry would be utilized regardless of symptom status (case-finding in asymptomatic individuals or those with nonspecific symptoms). Thus, the prevalence of abnormal spirometry in these two groups regardless of symptom status is assessed and subsequently the prevalence of individuals within each spirometric category that have respiratory symptoms. In adults that have never smoked, proposed spirometric recommendations are limited to those with respiratory symptoms. An unknown percentage of individuals might not be diagnosed or would be misdiagnosed in the absence of spirometry. Spirometry may detect a large reservoir of asymptomatic individuals, those with mild airflow limitation, or individuals with minimal symptoms that might not benefit from detection and treatment. Adverse effects would include increased health care costs, distraction from other interventions of proven effectiveness, or labeling individuals with disease unnecessarily or incorrectly. Spirometry could create unnecessary patient worry, increase health care expense and use of ineffective therapies with adverse effects, provide false reassurance, or lead to lower utilization of treatments of known effectiveness for other conditions.¹²

The analytic framework takes the perspective that abnormal airflow (as detected by spirometry) is a likely surrogate or risk factor for COPD but is not the sole criterion for defining clinically important disease or adults requiring treatment. Compared to clinical evaluation, spirometry would be useful if it improved diagnostic accuracy of individuals with airflow obstruction who would benefit from disease-specific interventions and ruled out individuals who are otherwise being misdiagnosed and/or receiving ineffective/unnecessary treatment. Improvement in process measures include increased smoking cessation rates and more appropriate utilization of effective interventions. Clinical outcomes include improved respiratory symptoms, health status, morbidity, and mortality in the spirometrically tested group. Determining the prevalence and severity of airflow obstruction in primary care adults according to symptom and smoking status and prior clinical diagnosis is necessary to assess the number of individuals that may benefit (or be harmed) by spirometric case-finding and diagnosis compared to clinical examination.

Definitions of “airflow obstruction” and “lower limits of normal” vary and typically have become more expansive over time. Normal lung function (and thus criteria for airflow obstruction) has been statistically derived from population-based surveys rather than directly based on pathological/clinical criteria of disease.¹³ Spirometrically-detected airflow impairment has been defined using equations according to subjects having an FEV₁/FVC ratio below the lowest 5 percent of the reference population (controlled for gender, height, age, and race) rather than documenting a disease state or symptom status. Most population based surveys have not conducted bronchodilator reversibility testing and thus estimates of a patient's best lung function or the presence of asthma or partial reversibility in airflow obstruction may not be accurately known. Additionally, airflow obstruction as measured by spirometry does not fully describe the disability in COPD that is manifested by dyspnea, exercise intolerance, and exacerbations. Some individuals with airflow obstruction are asymptomatic. Others with respiratory symptoms compatible with COPD may have normal spirometry. This may be due to the fact that other physiologic abnormalities (dynamic hyperinflation of the lungs and peripheral muscle abnormalities) as well as psychologic variables (coexisting anxiety) affect these clinical outcomes. Even among symptomatic individuals with airflow obstruction other conditions may be the cause of the respiratory symptoms (e.g., heart failure).

GOLD has developed recommendations for the diagnosis, management, and prevention of COPD. Their recommendations rely on results of spirometry in addition to clinical evaluation (e.g., physical examination, chest x-ray, eliciting symptoms based on clinical history).⁸ Diagnosis and treatment include individuals without respiratory symptoms but who have airflow obstruction. Changing definitions of disease can profoundly alter disease prevalence.¹⁴ In the case of COPD, this could occur by classifying individuals with disease based solely on spirometric findings rather than a combination of symptoms and physiologic measures or changing the level of spirometry that constitutes the presence or severity of disease. Table 1 on page 7 reflects the effects of using varying spirometric definitions of airflow obstruction. The effect of new definitions on disease prevalence/incidence, symptom severity, treatment, and outcomes is not known.

Clinically significant COPD includes individuals with dyspnea or other respiratory symptoms that reduce quality of life. Spirometry may be useful to assess the presence and severity of airflow obstruction, determine if symptoms are likely due to COPD (both in confirming a diagnosis and establishing spirometric severity or in excluding airflow obstruction as a cause), and institute appropriate disease-specific intervention. In the absence of airflow obstruction, a clinical diagnosis of and treatment for COPD is inappropriate (though individuals with asthma or a large bronchodilator response may have normal spirometry during symptom free periods). Assessing airflow in the absence of disabling symptoms or effective preventive interventions is limited to prognostic information or improving smoking cessation rates.

Question 2 Can use of spirometry lead to increased smoking cessation rates?

Smoking cessation is the most effective way to reduce the risk of developing COPD and prevent or improve respiratory symptoms. While smokers with symptoms have the greatest improvement, reduction in future respiratory symptoms is seen even among asymptomatic individuals with airflow obstruction.¹⁵ It is the only intervention demonstrated to prevent or delay the development of airflow limitation and reduce its progression. In patients with mild to moderate airflow obstruction, abstinence from smoking results in a sustained 50 percent reduction in the rate of lung-function decline over time.¹⁶

Clinical Practice Guidelines issued by the U.S. Department of Health and Human Services¹⁷ recommend that health care providers identify all smokers and advise them to quit regardless of spirometric or symptom status. Individuals attempting to quit smoking should be offered pharmacological interventions, unless there are medical reasons to withhold this form of treatment. Interventions that improve smoking cessation rates and maintain abstinence would be very valuable. However, reducing the prevalence of smoking has proven to be a formidable task.¹⁸ Approximately 35 percent of smokers with mild to moderate airflow obstruction enrolled in the Lung Health Study achieved abstinence at 1 year, but only 22 percent reported continued abstinence at 5 years. The 16 percent absolute reduction compared to enrollees assigned to receive “usual care” occurred with an intensive intervention that consisted of nicotine replacement (chewing gum, inhaler, spray, and a transcutaneous patch that was provided free of charge), cessation behavioral counseling, which consisted of 12 group sessions in the first 10 weeks, and a maintenance program for people who quit smoking.¹⁹ Cost effectiveness analyses have shown that smoking cessation interventions with incremental quit rates of 3 percent to 6 percent are economically acceptable because of the large health benefits (many beyond airflow obstruction) due to smoking cessation.⁴

A key question in case-finding is to determine if obtaining spirometry and providing individuals with measures of their lung function improves smoking cessation rates among current smokers and maintains abstinence among former smokers or never smokers. Benefits could occur regardless of symptom status or spirometric value. The potential roles of spirometry in improving smoking cessation rates include its use as a: 1) “biomarker assessment of lung health” to provide feedback and encouragement for smoking cessation and continued abstinence (regardless of symptom status); 2) risk stratification or prognostic tool for identification of an individual's (or group's) likelihood of smoking cessation, and 3) guide for targeting types of smoking cessation programs. Smoking cessation counseling could be enhanced by incorporating results from spirometric testing into routine clinic visits. Health care providers may be more likely to counsel patients or recommend additional smoking cessation therapies based on spirometric findings. Smokers may be more likely to quit if presented with information about their “lung health.” Adverse effects include added costs and resource use associated with initial and confirmatory spirometric testing and decreased smoking cessation rates due to false reassurance or nihilism. The potential role for, and outcome from, spirometry used as a motivational tool for smoking cessation are shown in Figure 2 on page 15.

Question 3 Does the effectiveness of specific therapies to improve clinically relevant outcomes in COPD vary based on baseline or followup spirometry, short-term spirometric response due to initial therapy, or spirometric progression over time?

Treatment goals are to reduce spirometric decline in lung function, relieve disabling respiratory symptoms (particularly dyspnea), improve exercise tolerance and health status, prevent and treat complications and exacerbations, and reduce mortality. Recommendations encourage use of spirometry to assess baseline severity of airflow obstruction and acute treatment response. Clinicians are encouraged to periodically assess symptoms and monitor objective measures of airflow limitation for development of complications and to determine when to adjust therapy. The effectiveness of this strategy is not known.

If treatments are effective in adults with mild to moderate airflow obstruction or those with absent or relatively mild respiratory symptoms, then one potential benefit of case-finding with spirometry could be identification and treatment of a large number of individuals not readily detected by clinical examination. However, if effectiveness is limited to the much smaller cohort of subjects with severe airflow obstruction and activity limiting respiratory symptoms, then population-based spirometric case-finding is less likely to be beneficial compared to spirometric identification and treatment targeted at individuals with bothersome respiratory symptoms.

Spirometry may be useful as a guide for initial and followup management among individuals with established airflow obstruction/COPD. Among asymptomatic individuals, spirometry could be effective if it resulted in initiation of interventions for airflow obstruction that prevented the development of symptoms or reduced the decline in lung function. In symptomatic individuals, spirometry could improve diagnostic accuracy and determination of whether or not spirometric thresholds of airflow obstruction exist prior to appropriate initiation of COPD specific therapy. Monitoring patients with periodic spirometry would be useful if modification of therapeutic interventions according to spirometric response to therapy, spirometric change over time, or achieving a certain spirometric threshold reduced respiratory symptoms including exacerbations and hospitalizations and improved quality of life. Adverse effects would include the costs of using spirometry to monitor treatment or disease progression, harms related to medication use, and unnecessary or improper initiation/modification of treatments based on spirometry compared to clinical evaluation. To assess the effectiveness of interventions for COPD beyond smoking cessation we will focus on whether effectiveness varies according to symptom status (presence or absence, type, severity, or frequency of symptoms), previous clinical diagnosis of COPD, baseline or followup spirometry, acute spirometric response to treatment, spirometric slope over time, and intervention type or dose.

Question 4 Is prediction of prognosis based on spirometry, with or without clinical indicators, more accurate than prognosis based on clinical indicators alone?

Spirometry could provide independent prognosis related to quality of life, progression to more severe and symptomatic COPD, and mortality (both overall and COPD specific). Spirometry may help identify individuals at increased risk for future health problems who are in need of effective COPD-specific interventions. Spirometry may provide more accurate risk stratification and appropriate utilization of interventions for other chronic medical conditions.

The analytic pathway includes the ability of clinical examination and history to determine respiratory symptom status and etiology, spirometry to assess presence and severity of airflow obstruction, spirometry to alter smoking cessation and abstinence rates in current and former smokers, spirometry to guide initiation and modification of pharmacologic or rehabilitation therapy for individuals with established COPD, and finally spirometry as a prognostic tool for future COPD-related outcomes (especially worsening symptom status).

Final synthesis of this information will result in a pathway that evaluates the number of adults needed, according to smoking and symptom status, to receive office-based spirometry in order to identify candidates for treatment. We will estimate the number of individuals likely to have improvement in specific outcomes, the type and relative effectiveness of interventions, whether monitoring of spirometry improves clinical management and outcomes, and prognosis based on spirometric findings.

Question 1

Data sources. Articles published in the English language from 1966 to January 2005 were identified by searching MEDLINE accessed through PubMed and Cochrane Database using the following terms: diagnosis, epidemiology, bronchospirometry, COPD, emphysema, bronchitis, respiratory function tests, airway obstruction (or airflow limitation), cohort studies, case reports, case-control studies. Because our goal was to estimate the prevalence of COPD and airflow obstruction likely to be encountered by casefinding in primary care settings, we examined population based or primary care cohort or case-control studies.

Study selection. Studies were eligible if they reported the results of spirometry testing of community-based adult populations or primary care settings and were published in English. Studies limited to patients with known COPD or symptoms such as cough, sputum production, dyspnea, or wheeze were excluded unless results were reported separately for asymptomatic individuals. Emphasis was placed on community-based studies conducted in the U.S.

Outcomes. The primary outcome was the prevalence of airflow obstruction according to GOLD stage (or other consensus criteria such as ATS) according to: spirometry, race, gender, age, symptom, and smoking status (current, past, or never), and presence of a clinical diagnosis of COPD.

Quality assessment. Quality and strength of evidence was determined by whether the included studies adequately addressed our key outcome by providing information related to spirometrically-detected COPD in general adult populations or primary care settings according to GOLD stage or other consensus criteria, race, gender, age, smoking, and symptom status. Because this report was intended to guide clinical decisions in the United States, we placed greatest emphasis on studies conducted in the U.S.

Question 2

Objective. Our primary goal was to determine if providing smokers with results from spirometric testing improves smoking cessation rates.

Data sources and study selection. A detailed search strategy was used to identify potentially relevant articles and is provided in Appendix B ^*. Studies were eligible if they were randomized controlled trials (RCTs), published in English, had a minimum of 25 subjects per treatment arm, involved subjects that smoked (regardless of respiratory symptoms or spirometry status), had a followup time of 6 months or longer, and provided outcomes smoking cessation rates (as measured by self-report or biochemical validation such as carbon monoxide level). The intervention had to include spirometry alone or in conjunction with other treatments as a motivational tool for smoking cessation. Studies were excluded if the control group also received notification of spirometric results. Non-controlled reports that merely reported smoking cessation rates according to spirometric value or respiratory status were excluded. However, these studies were reviewed and findings described in order to estimate whether spirometric values or respiratory status could predict smoking cessation rates. Of the 212 references identified, seven met eligibility criteria (Figure 3 on page 16). Additionally, in order to provide a context for potential magnitude and biologic plausibility of various smoking cessation strategies, we included information related to the effectiveness of established strategies for smoking cessation and rationale for use of biomarkers as a tool for enhancing smoking cessation counseling.

Literature search strategy. The literature search used Ovid MEDLINE until May 2005. To supplement this search, we examined the Cochrane Database of Systematic Reviews of Effectiveness as well as bibliographies of published articles and contacted experts in the field. Listserv members of the World Health Organization's Society for Research on Nicotine and Tobacco were contacted and invited to identify additional published, unpublished, or ongoing relevant studies. Search terms included: spirometry; smoking therapy; smoking psychology; COPD; airflow limitation; randomized controlled trials; controlled clinical trials; and case-control studies. Identified articles were reviewed along with their references to identify other key articles and to refine our search strategy. Our search strategy included articles identified to evaluate the effectiveness of interventions for patients with COPD (Question 3). Titles and abstracts of identified references were reviewed using standardized data abstraction sheets (Appendix C ^*). All references received an identification number.

Interventions. We considered the process of obtaining and providing the results of spirometry to smokers in combination with focused smoking cessation counseling as a single intervention consistent with a pragmatic approach likely to be employed in health care settings. Other differences in interventions between treatment and control groups such as the incorporation of results from biomarker testing (carbon monoxide or cotinine levels, chest x-rays, etc.), varying frequency, intensity, methods of counseling, or pharmacologic treatments were considered concomitant interventions that might differentially effect cessation rates.

Outcomes. Smoking cessation outcomes in clinical trials are measured in a variety of ways including short- and long-term abstinence and point-prevalent or sustained abstinence. In general, short-term abstinence refers to outcomes at less than 3 months following initiation of treatment and may include in-treatment results, depending on the duration of interventions. Long-term abstinence refers to outcomes measured at 6 to 12 months after initiation of treatment (or later). In addition, at the measurement point, abstinence can be described as point-prevalent (usually 7–30 days prior to the measure) or sustained (ranges from 6 months to continuous from point of intervention). Finally, abstinence can be self-reported, or validated by biomarkers of exposure such as carbon monoxide (CO) or cotinine. Quit attempts are generally regarded as a less robust, secondary process outcome. Our primary outcome was long-term sustained abstinence that was validated by biomarkers. Subgroups of interest included spirometric categories, (e.g., GOLD or ATS), symptom status, race, and gender.

Quality assessment and quantitative synthesis. Quality and strength of evidence was based on the method of Schulz et al.²⁰ We also assessed loss to followup and whether studies provided information that would allow for determination of the independent effect of conducting spirometry and providing their results on smoking cessation rates. Because of the clinical heterogeneity of study interventions, pooled analyses were not conducted.

Question 3

Literature search strategy. Search terms were identical to those published by Sin⁹ (adults >19 years of age, COPD, RCTs) to identify RCT/controlled clinical trials (CCT), meta-analyses, or reviews published since the completion of their search (i.e., between 2002 and January 2005; for inhaled therapy between 2002 and January 2005). For each of these therapies we conducted a literature search using Ovid MEDLINE. To supplement this search, we examined the Cochrane Database of Systematic Reviews of Effectiveness as well as bibliographies of published articles and contacted experts in the field. We limited our search to English-language articles. These were categorized according to type of intervention: 1) inhaled medications including: β2 agonists, long-acting anticholingerics (tiotropium), combination β agonists and anticholinergics, inhaled corticosteroids, combination inhaled corticosteroids and long-acting β 2 agonists, pulmonary rehabilitation, 2) disease management programs (which include any combination of patient education, enhanced followup, and/or self-management session); 3) long-term administration of non-invasive mechanical ventilation (NIMV); and 4) oxygen therapy.²¹ We obtained additional information from the data coordinating center for one large trial (LH-1) that evaluated pharmacologic interventions in subjects with mild to moderate airflow obstruction.

Titles and abstracts of identified references in addition to those included in the report by Sin were reviewed using standardized and piloted data abstraction sheets. All references received an identification number. The number of excluded studies and reasons for exclusion are described in Figure 4 on page 16. Studies meeting preliminary eligibility criteria were retrieved in full for further assessment and data extraction.

Eligibility criteria. For intervention studies we restricted our analysis to trials that were randomized, defined by clinical diagnosis or spirometry, and provided clinically relevant outcomes. Trials of inhaled therapies were required to enroll at least 50 subjects per treatment arm. A followup time of 3 months was used as the threshold for inclusion (with the exception of pulmonary rehabilitation programs, for which 6 weeks was used as the threshold).

Studies were excluded if they only reported physiologic variables such as changes in FEV₁, because the correlation between spirometric changes and long-term clinical outcomes in COPD has been shown to be weak.²² We examined bibliographies of these reviews and meta-analyses.²³–³⁷ The studies that contained the different domain of comparison between the baseline and the ending point³⁸ or no baseline data of spirometry as FEV₁ ³⁹ or no comparison groups at same design,⁴⁰ or cost-effectiveness analysis^{41, 42} were excluded. Information from the original publication was used unless additional relevant data were available in subsequent reports.

Quality of studies and strength of evidence. Two researchers independently extracted study and patient characteristics onto data sheets.⁴³ Disagreements were resolved by discussion or cross checking of other co-workers through project meetings.⁴⁴ The methods of Schulz et al.²⁰ were used to assess the quality of RCT. We evaluated whether studies were blinded, used intention-to-treat analysis, and reported attrition. The magnitude of effect across different outcomes and pharmacologic interventions (e.g., exacerbations, mortality, dyspnea, etc.) was assessed based on absolute and relative reductions as well as in comparison to previously determined minimally important clinical differences in respiratory health status measures. Subgroup analysis was attempted to determine if results varied according to disease severity based on baseline symptom and/or spirometric status, acute change in spirometry, or spirometric change in time. We attempted to focus on individuals most likely to be identified through spirometric casefinding (i.e., individuals with mild to moderate airflow obstruction and respiratory symptoms who were not diagnosed by clinical examination). We evaluated whether any trial utilized spirometry as a guide for monitoring subjects' clinical status or to modify therapy. Of the 53 studies that were eligible, 20 were new references not included in the report by Sin.

Quantitative synthesis of study outcomes. All analyses were conducted using Review Manager Version 4.2 (Revman; The Cochrane Collaboration, Oxford, England). For each end point we combined the results from individual studies to produce pooled effect estimates (relative risk ratios and absolute risk ratios). Heterogeneity of results across individual studies was checked using the Cochrane Q test. If heterogeneity was observed (p<.10), we used the Dersimonian and Laird random-effects model to synthesize the results; otherwise, a fixed-effects model was used.⁴⁵ As part of a sensitivity analysis for the latter situation, we used a random-effects model to determine the robustness of the data. In all cases, the results obtained from the random-effects and fixed-effects models were similar. Continuous variables were pooled using weighted mean difference technique.

Outcomes. Our primary outcome was the number of individuals with at least one exacerbation as defined by authors. Secondary outcomes included changes in St. George's Respiratory Questionnaire (SGRQ) scale scores; number of subjects with respiratory symptoms including dyspnea, cough, or sputum production; mortality; and overall and respiratory-specific hospitalizations and changes in health status between intervention and control. We attempted to evaluate results according to the following subgroups: spirometrically-determined severity of disease (GOLD or ATS stages and mean baseline FEV₁), symptom status, smoking status, gender, age (≥65 vs. <65), and race. We restricted analysis of health status and dyspnea to two well-standardized and validated instruments in COPD, SGRQ Chronic Respiratory Disease Questionnaire (CRQ).⁴⁶ These instruments quantify the extent of physical and psychological impairments related to COPD and allow investigators to determine the (beneficial) effects of specific interventions on the functional status of patients with COPD.⁴⁷ Dyspnea and exacerbations are the two most bothersome symptoms and the aspects of COPD that most influence health status. The CRQ is a 20-item COPD specific questionnaire that measures: dyspnea (five items), fatigue (four items), emotion (seven items), and mastery (four items). A 0.5 unit change per question (on a seven-point scale) is considered the minimally important clinical change. Composite scores range from 20–140 with higher scores indicating improved health status. The SGRQ is a respiratory-specific 50 item questionnaire with domains of symptoms, activity, and impacts plus a summary total score. Lower scores indicate improved health status, and a change of four units (out of 100) is considered clinically significant. While validated these questionnaires have been found to have weak correlations with physiologic variables including FEV₁ and mild to moderate correlation with exercise capacity and assessment of dyspnea, anxiety, and depression.

Question 4

Data sources and study selection. Articles published in English from 1966-January 2005 were identified using a search strategy similar to Question 1, which also included the key word “prognosis.” Eligible studies included cohort or case-control studies that assessed the prognostic effect of spirometry on COPD progression and outcomes. Additional studies were evaluated to determine the independent effect on overall mortality, though this was not the primary focus as directed by our TEP. We obtained additional results from one of the identified studies through personal communication with the author.

Outcomes. The primary outcome was progression to more severe airflow obstruction (GOLD or other stage criteria) and development of respiratory symptoms.

Data synthesis. Data were described for each study and not pooled.

Does Clinical Examination Predict Airflow Limitation?

There are no data that directly describe the sensitivity and specificity of spirometry (i.e., the probability of developing clinical obstructive airways disease given a particular FEV₁ or FEV₁/FVC). Instead, patients with an abnormally low FEV₁ and FEV₁/FVC are said to have “airflow limitation.” An FEV₁/FVC lower than the fifth percentile for age, height, and gender has been described as abnormal.⁴⁸

Despite the lack of a “gold standard” for defining the clinical presence of COPD, a systematic review by Holleman and Simel evaluated 19 articles that assessed the clinical examination for detecting airflow limitation according to spirometry.⁴⁸ Spirometric reference standards used in studies yielding operating characteristics for individual clinical examination items varied across the studies. None used the GOLD 2003 classification. Only two studies incorporated both FEV₁ and the FEV₁/FVC as the reference standard. Factors considered in the clinical examination included history (background information such as cigarette smoking and occupational or environmental pollutants and symptoms of wheezing, dyspnea, coughing, and sputum production) and physical examination (inspection, vital signs, palpation, percussion, auscultation, and clinical measures of airflow [match test, forced expiratory time test]).

Smoking status (ever vs. never) is only a moderately good predictor of airflow limitation. Compared to “never smokers” patients who have “ever smoked” are only slightly more likely to have airflow limitation (+LR [likelihood ratio] = 1.8). Never having smoked is moderately associated with decreased likelihood of disease. The most powerful predictor was at least a 70 pack-year history of smoking (+LR = 8.0) though the sensitivity was only 40 percent. Symptoms of sputum production or wheezing are associated with a moderate increase in the likelihood of airflow limitation. However, symptoms of cough or exertional dyspnea are associated with only a slight increase in the likelihood. Additionally, the absence of dyspnea or exertional dyspnea is only moderately useful in ruling out disease (any dyspnea: +LR = 1.2; -LR = 0.55; sensitivity = 82 percent; specificity = 0.33 percent).

Physical examination findings to predict airflow limitation all had a specificity of >90 percent but were limited by poor sensitivity. Patients who have wheezing on unforced expiration almost certainly have airflow obstruction, and this increases with the severity of airflow limitation and the prior probability of disease (Positive Likelihood Ratio = 36). The presence or absence of wheezing on forced expiration is of no value in diagnosis or ruling out airflow limitation.⁴⁹,⁵⁰ Absent wheezing, normal breath sound intensity, or absent rhonchi are associated with only a moderate decrease in the likelihood of disease. (Negative Likelihood Ratio 0.85, 0.70 and 0.95 respectively). Neither the presence nor absence of rales was useful in diagnosing airflow limitation.⁵¹–⁵³

Can the clinical examination predict severity of airflow limitation? Two studies reported on whether the presence of positive clinical findings could predict the severity of airflow limitation. The number of positive findings predicted the severity of airflow limitation in patients with known disease. The findings were present only if the FEV₁ was less than 50 percent predicted. Similarly, the number of positive findings predicted the severity of airflow limitation (r = 0.6).

Accuracy of the overall clinical impression for predicting airflow limitation. Three studies evaluated the accuracy of the overall clinical impression, or a clinician's ability to integrate all aspects of the clinical examination in forming an impression about the likelihood of airflow limitation. Clinicians' overall impressions predicted any airflow limitation only moderately well. The ability to diagnose airflow limitation clinically is variable but seems to improve as the severity of the disorder increases.

Combinations of individual findings. Six studies assessed the utility of combining clinical examination items to predict airflow limitation. Combinations of findings do not effectively rule out airflow limitation. The best combination is never having smoked, no reported wheezing, and no wheezing on examination (LR = 0.18). A patient with any combination of two findings (≥70 pack-year history of smoking, history of COPD, or decreased breath sounds) can be considered to have airflow limitation.

Prevalence and Severity of Airflow Obstruction

Population-based studies from seven different countries were identified that assessed the prevalence and severity of airflow obstruction and respiratory symptoms (Table 2 on page 19, Figures 5 and 6 on page 19, Summary Tables 1–4 on pages 55–60, and Evidence Tables 1–4 in Appendix D ^*). The prevalence and severity of airflow obstruction and COPD in general populations varied widely according to definitions utilized and country studied. Postbronchodilator testing or response to bronchodilators was rarely performed in these large population surveys. Respiratory symptoms were usually assessed according to responses to single item questions rather than detailed clinical probing. Additionally, subjects were generally categorized as having COPD based on a patient reported diagnosis of emphysema or chronic bronchitis. Some reports provided outcomes according to age, race, gender, smoking, respiratory symptom, and spirometric status (Summary Tables 1 and 2 on pages 55–58 and Evidence Table 1 in Appendix D ^*). However, none provided additional subgroup data required for assessment of a specific respiratory symptom (presence or absence or type) according to postbronchodilator spirometric value (e.g., GOLD stage) in a particular demographic category (e.g., age, race, gender, smoking). Thus, estimates of these are extrapolated from prebronchodilator results provided for larger aggregate groups. Data from one study of Spanish adults ages 40–65 indicated that the 26 percent of adults with airflow obstruction had a positive bronchodilator response of at least 200 mL and a relative increase of at least 12 percent. However, only 5 percent had normal airflow after bronchodilator inhalation (i.e., asthma). The methodology reported in these surveys is likely to introduce only a small misclassification. Findings may more accurately reflect the results obtained using primary-care spirometry and brief respiratory symptom assessment than those obtained in pulmonary specialty practice. Determining normal airflow “at-risk” populations (approximately GOLD 0) across studies was difficult because some provided the prevalence of “any respiratory symptom” (wheeze, cough, sputum, or dyspnea) but not specifically chronic cough and sputum in subjects with GOLD stage normal airflow. Data from the NHANES I and III (Summary Tables 1–3 on pages 55–59 and Evidence Tables 1–3 in Appendix D ^*) survey provide the most comprehensive and relevant assessment of obstructive lung disease, low lung function, and respiratory symptoms in adults in the United States. NHANES III assessed adults ages 17 years and older from 1988-1994 who classified themselves as whites or blacks and had pulmonary function testing performed without bronchodilators based on 1987 American Thoracic Society recommendations and had complete information on race, smoking status, height, and presence of respiratory symptoms. Subjects were asked if they had ever been told by a doctor that they had Obstructive Lung Disease (OLD) of asthma, chronic bronchitis, or emphysema (and if yes whether they still had that condition). Individuals who reported ever being told they had a diagnosis of emphysema or currently reported a diagnosis of chronic bronchitis were categorized by the authors as having current COPD. There was no information regarding whether any of these individuals had previously undergone spirometry or whether spirometry led to the clinical diagnosis of these conditions. However, we considered that individuals with a “current” or “previous” diagnosis of emphysema or chronic bronchitis were not detected by “primary care case finding” and that they had “clinically detected COPD.”

Subjects in NHANES were classified as reporting respiratory symptoms if they gave a positive response to respiratory specific questions related to cough, phlegm, wheeze, and dyspnea. The question for dyspnea read: “Are you troubled by shortness of breath when hurrying on level ground or walking up a slight hill?” For cough and sputum a positive response was considered if subjects affirmatively answered the question: “Do you usually cough (bring up phlegm) on most days for 3 consecutive months or more during the year?” Data were stratified according to national population estimates by race, sex, and smoking status. Overall estimates were then age-adjusted to all study participants. Subjects were defined as having “low lung function” based on the 1987 ATS recommendations (i.e., subjects with an FEV₁/FVC <0.70 and an FEV₁ <80 percent predicted. This group was further divided according to ATS-1995 criteria (Stage 1 vs. Stage 2 or 3) into subjects with an FEV₁ of ≥50 percent predicted and those with <50 percent predicted). Because activity-limiting or “troubling” shortness of breath (as reported in the NHANES questionnaire) is the most clinically bothersome and relevant outcome, we judged this to be the most “clinically significant” symptom of COPD if accompanied by spirometric evidence of airflow obstruction performed in the absence of bronchodilators.

Spirometry performed in the absence of inhaled bronchodilators identified a relatively large proportion of individuals with airflow obstruction who did not report respiratory symptoms and conversely was also normal in a large percentage of adults who report respiratory symptoms. An estimated 8.5 percent of the population reported current OLD (asthma, emphysema, or chronic bronchitis) and another 4.3 percent reported OLD in the past but not currently. Approximately 28 percent of adults reporting a current diagnosis of OLD had asthma as their only type of OLD. The proportion of the population with past or current OLD and COPD varied by sex, race, and smoking status, with women reporting more disease than men, whites reporting more disease than blacks, and current or former smokers reporting more disease than never smokers. OLD was reported among 12.5 percent of current smokers, 9.4 percent of former smokers, and 5.8 percent of never smokers. Former smokers were on average older than current smokers and never smokers. Mean level of lung function was lower among smokers than never smokers and increased with age. Results from the NHANES III survey indicated that the prevalence of 1987-ATS defined “low-lung function” increased from 6.0 percent in adults ages 25–44 to 40.7 percent in those ages ≥75 years (Summary Table 3 on page 59). NHANES I results indicate that ATS 2 or 3 (approximately GOLD postbronchodilator Stage 3,4) airflow obstruction was present in 2.6 percent of adults 50–59 years old and 4.2 percent of adults ages 70–74. The prevalence of mild versus moderate to severe airflow obstruction in the Po Delta Survey in adults ages >45 years increased from 8 percent versus <3 percent to 35 percent versus 5 percent respectively when using ATS rather than European Respiratory Society (ERS) criteria (Summary Table 3 on page 59). The prevalence of low lung function was similar in whites and blacks (13.8 percent vs. 11.6 percent in NHANES III) (Evidence Table 2 in Appendix D ^*). Results from the NHANES I survey indicated that the percentage of whites and non-whites having normal spirometry and no respiratory symptoms was 67.6 percent and 65.3 percent respectively.

The prevalence of adults having normal spirometry and not reporting respiratory symptoms (normal/individuals not reporting respiratory symptoms) varied by country. However, the criteria used to define symptoms and airflow obstruction was the greatest factor contributing to varying prevalence estimates. It ranged from 52 percent (U.S.: NHANES III) to 89 percent (Italy: ERS Criteria) (Summary Table 2 on pages 56–58). The prevalence of normal spirometry and no respiratory symptoms in the Italian Po Delta Survey decreased from 89 percent to 60 percent when subjects were classified by ATS criteria instead of ERS criteria.

An estimated 6.8 percent of the U.S. population had ATS-1987 criteria for low lung function and 7.2 percent had an FEV₁/FVC <0.7 but an FEV₁ >80 percent predicted. When recategorizing NHANES subjects according to a 2003 GOLD staging system that uses postbronchodilator spirometric values to define airflow obstruction, the percentage of individuals labeled as having “airflow obstruction” or “at-risk” increased by more than three-fold. Only 56.4 percent of the population had both normal spirometry and reported no chronic respiratory symptoms. Greater than 20 percent had airflow obstruction or were considered “at risk.” Specifically, 7.2 percent of subjects had GOLD Stage 0 (chronic sputum and phlegm but normal spirometry), and an additional 13.9 percent of adults had airflow obstruction (approximate GOLD Stage 1, 2, 3,4 = 7.2 percent, 5.4 percent, and 1.5 percent respectively). Prevalence was higher in current smokers and with increasing age (Evidence Table 4 in Appendix D ^* and Figure 5 on page 19). The percentage of individuals reporting respiratory symptoms increased with worsening airflow obstruction (Summary Table 4 on page 60). However, 23 percent of individuals with normal spirometry reported respiratory symptoms and 21 percent of individuals with severe to very severe airflow obstruction (ATS 2–3; FEV₁ less than 50 percent predicted, approximate GOLD Stage 3,4) had no symptoms. Furthermore, 35 percent of individuals with an FEV₁ less than 50 percent predicted did not report being troubled by shortness of breath, the symptom felt to be most clinically bothersome and warranting intervention. (Figure 6 on page 19) Therefore, the overall prevalence of adults having both low lung function and “any respiratory symptom” is GOLD 1 = 3.6 percent, GOLD 2 = 3.2 percent, GOLD 3,4 = 1.2 percent. Findings from other population-based studies are consistent with these when attempting to account for differences in definitions of airflow obstruction and symptom status as well as use of prebronchodilator spirometric values. Between 40 and 80 percent of individuals with spirometrically determined “low lung function” had no prior diagnosis of OLD.

To assess diagnostic accuracy of spirometry the additional number of adults with clinically significant disease that would be detected by case-finding is required. These would be defined as an adult with spirometrically determined airflow obstruction who reports bothersome respiratory symptoms but not a diagnosis of COPD. However, there were no data according to previous reported diagnosis of COPD, stage severity of airflow obstruction, and symptom status (particularly dyspnea). In adults who reported a clinical diagnosis of COPD (emphysema or chronic bronchitis), (approximately 3 percent of the total NHANES respondents) only 17.4 percent had 1987-ATS defined low lung function suggesting that the vast majority of these individuals do not have COPD. Among individuals reporting a clinical diagnosis of COPD 25.6 percent reported chronic phlegm, and 48 percent reported shortness of breath.

Summary of Interventions Used to Enhance Smoking Cessation

Two major categories of effective and potentially effective strategies, pharmacologic therapy and counseling/behavioral treatments, are shown in Figure 7 on page 23. Interventions can be used either alone or in combination. A summary of effectiveness is provided below.

Pharmacologic therapy. Nicotine replacement therapies (NRTs) and Buproprion SR are considered “first line” medications. The overall odds of smoking cessation in those who used NRTs (except lozenges) were 1.7 fold greater than those who did not use NRT (95 percent CI: 1.6, 1.8).⁵⁴ Separate meta-analyses of each of the first four NRTs shown in Figure 7 on page 23 were conducted and the results were statistically significant ranging from an odds ratio of 1.5 for gum to 2.7 for nasal spray. The absolute differences ranged from 6.6 percent for gum to 16.6 percent for nasal spray.¹⁷ The results of a meta-analysis of two studies indicate that Buproprion SR improved smoking cessation rates twofold with an absolute rate difference of 13.2 percent.¹⁷

Counseling/behavioral therapy. Several counseling and behavioral therapy strategies have been shown to be effective with absolute differences in smoking cessation rates between control and intervention ranging from 2.3 percent to 8.0 percent. These include advice to quit by a physician,¹⁷ nurse, or other health professional; intensive counseling, either at the group or individual level;¹⁷ general problem-solving,¹⁷ such as providing general information about smoking cessation and relapse, identifying potential stumbling blocks, and creating solutions to overcome them; self-help materials, including written and computer-based materials and audio-and videotapes;⁵⁴ and the technique of rapid smoking.¹⁷ A systematic review found inconclusive evidence of effectiveness from motivational counseling,⁵⁵ including a discussion of the benefits of quitting, the risks of continued smoking, or a discussion of personal risk based on biological markers (including spirometry). The results of a meta-analysis of smoking cessation trials by Kottke et al., suggest that a smoking cessation message reinforced consistently and repeatedly over time is the best predictor of success.⁵⁶

Use of Biological Markers in Smoking Cessation

Biological markers may have a unique role as motivational aids in smoking cessation programs.⁵⁷,⁵⁸ Three categories of biomarkers include markers of: 1) tobacco exposure (e.g., carbon monoxide, cotinine, thiocyanate); 2) physiologic effects (e.g., pulmonary function tests—including spirometry, histopathological changes, x-rays, plethysmography, electron beam tomography, and other diagnostic tests); and 3) genetic susceptibility (e.g., CYP2D6).⁵⁹ Several observational studies have assessed biomarkers as motivational tools for smoking cessation. However, the lack of controls makes assessment problematic. (Studies without controls: CO,⁶⁰–⁶² CT scans,⁶³ airflow limitation/spirometry,⁶⁴–⁶⁶ plus others.⁶⁷–⁶⁹

Rationale for the Use of Spirometry in Smoking Cessation

Determination of smoking and respiratory symptom status as well as advice and interventions to aid cessation or maintenance of abstinence should be provided to all smokers, regardless of pulmonary function or the presence or absence of symptoms. Results from the Lung Health Study (LH-1), a 5-year multicenter randomized control trial in the United States and Canada, indicated that smoking cessation is beneficial in slowing both the clinical and spirometric progression of patients with mild/moderate airflow obstruction. If smokers quit prior to the development of symptoms, the rate of lung function decline approaches that of nonsmokers.⁷⁰,⁷¹ In LH-1, smokers who quit experienced an initial increase in lung function in the first year after quitting (mean increase of 47mL/year), followed by an annual decline comparable to declines observed in nonsmokers attributed to age (mean annual decline of 31mL/year). Subjects who continued to smoke had an annual decrease in lung function that was twice the rate seen in those who quit (mean annual decline of 62mL/year). Similar findings have been observed in other studies.⁷² The standard deviation of the annual rates of decline in FEV₁ in the LH-1 (48mL/year in quitters and 55mL/year in continuing smokers) indicates that, even over 5 years of followup, confidence in a value for the annual decline in an individual is low.

Despite the evidence that smoking cessation improves clinical outcomes and measures of airflow obstruction, the concept of performing spirometry and using these test results to provide personalized encouragement for smoking cessation is controversial. ⁷³–⁷⁵ Spirometry might be useful in motivating individuals to quit smoking or maintain abstinence. It may identify individuals likely to benefit from more intensive smoking cessation counseling as well as those unlikely to quit smoking. It may motivate physicians to more carefully assess symptom and clinical status and/or provide smoking cessation counseling. Spirometric results can be provided to those at-risk in several formats, including percent FEV₁, FEV₁/FVC, FEV₁/FEV₆, or “lung age” estimations, wherein a patient's chronological age is contrasted with the physiologic age of his/her lung tissue.⁷⁶,⁷⁷ However, spirometry entails time and costs, and may result in false labeling or reassurance and it is not yet known whether it improves smoking cessation rates. The available evidence suggests that cessation rates are relatively low and require fairly intensive counseling and pharmacologic intervention. Therefore, spirometry may not offer additional motivational benefit nor serve as a reliable predictor for an individual's likelihood of quitting. A conceptual model for the role of spirometry in smoking cessation is provided in Figure 2 on page 15.

Smoking Cessation Strategies in People with COPD

Most smoking cessation studies did not specifically recruit subjects with airflow obstruction or clinically diagnosed COPD, nor do they report outcomes according to spirometry or symptom status. A systematic review published in 2004 evaluated the effects of interventions for smoking cessation in people with established COPD.⁷⁸ The authors identified five randomized trials comprising 6,491 patients with COPD conducted in the U.S., Canada, and Denmark between 1991 and 2001. None of these studies used spirometry as a motivational tool for smoking cessation. However, three studies, including the largest study, used spirometry as the method to identify subjects eligible for participation. Lung Health Study One (LH-1) enrolled 5,887 current smokers who had spirometric evidence of mild to moderate airflow obstruction. Nearly 30 percent had a previous clinical diagnosis of bronchitis but only 3.2 percent had a diagnosis of emphysema. Because studies were clinically heterogeneous regarding study population (severity of obstruction and symptoms) and types of interventions, abstinence rates were not pooled.

Three trials involved 179 subjects and evaluated four different behavioral intervention strategies. Two studies involved smokers who were admitted to the hospital and may not be representative of large population-based strategies. Behavioral interventions included: 1) use of the term “smokers' lung” rather than “chronic bronchitis” when talking to patients plus an informational brochure; 2) individual counseling responsive to patients needs and questions combined with a self-help manual; and 3) behavioral reinforcement schedules that provided lottery tickets for reduced breath carbon monoxide, self-reported smoking cessation, or attendance at clinic visits. Absolute differences in self-reported and biochemically validated point prevalence or continuous abstinence at 6–12 months ranged from 10–16 percent. However, the confidence intervals were wide and there were no statistically significant differences in any of the studies.

The Lung Health Study evaluated the effect of an intensive smoking cessation intervention (combined with either the inhaled bronchodilator ipratropium bromide or placebo) on the rate of decline in FEV₁. The comparison (usual care) group received no study prescribed smoking intervention. The smoking intervention group received intensive cessation counseling (advice to quit by physician at one session plus group counseling—12 sessions in 10 weeks), nicotine gum provided at no cost, and a maintenance program for those who quit smoking. After 12 months, the smoking intervention program was significantly more effective in helping smokers to quit (RD at 5 years = 0.26, 95 percent CI: 0.23, 0.28). The differences declined but persisted throughout the 5 years of study followup (RD = 0.16, 96 percent CI: 0.14, 0.18).

Can symptom status and/or baseline spirometric values be used as risk stratification tools to assess the likelihood of smoking cessation? Observational studies reported in the 1970s provide conflicting information regarding the motivational effects of spirometric test results on smoking rates or the ability of spirometric values or symptom status to predict smoking cessation rates. Loss et al. examined the prevalence of pulmonary abnormalities at baseline and the subsequent 6-month abstinence rates in a group of 73 smokers.⁶⁷ Subjects completed pulmonary function testing and received these results along with brief counseling 1 week later via telephone. Twenty-nine percent of subjects had abnormal spirometric results and 89 percent had cough, excess sputum production, shortness of breath, or wheezing. At 6 months followup, 7 percent of those with abnormal Pulmonary Function Test (PFT) results were abstinent as compared with 11 percent of those with normal results. The authors concluded that pulmonary testing did not provide sufficient motivation to induce smoking cessation and that the costs of the testing outweighed the benefits.

Petty et al. examined smoking cessation rates among 101 smokers that were followed for up to 7 years.⁶⁸ All subjects were notified of their spirometric results via mail. The abstinence rate at the end of followup was 18 percent in those with abnormal lung function at baseline (FEV₁/FVC <60 percent) versus 11 percent in those with values above this threshold. Cessation rates were similar in subjects with chronic bronchitis (18 percent) and those without bronchitis (19 percent).

Hepper et al. conducted a series of community screening programs for COPD.⁶⁹ Participants with abnormal lung function received their test results within 2 weeks after screening. They were encouraged to follow up with their primary care physician, who was also provided with the results of the tests, and could give them further information. Subjects from randomly selected communities (n = 553) were contacted 2 to 3 years after baseline testing to assess smoking status. The quit rate among smokers with abnormal results and no prior COPD diagnosis was 21.4 percent, compared with 11.7 percent among those with normal results and 11.9 percent among those with abnormal results and prior COPD diagnosis. These authors concluded that providing the spirometric results was enough motivation to compel smokers to quit if they had no prior clinical diagnosis of COPD and were not already aware that they had reduced pulmonary function.

Gorecka et al. reported results of a case-series of adult smokers who received smoking cessation advice along with baseline spirometric screening and 1 year followup.⁶⁴ The authors attempted to assess factors associated with smoking cessation in adult smokers (n = 558) categorized as either having “airflow limitation” (defined as FEV₁/FVC ratio <85 percent or normal lung function. Subjects with airflow limitation were further categorized as having mild (FEV₁ <70 percent of normal), moderate (FEV₁, 50–69 percent of normal) and severe (FEV₁ <50 percent of normal) airflow limitation. There was no difference in 1 year sustained smoking cessation among individuals with normal lung function compared to those with spirometrically determined airflow limitation. However, in post hoc multivariate analyses (and in contrast to findings from the LH-1 study) FEV₁ was independently inversely associated with likelihood of abstinence at 1 year. Individuals with poorer lung function as defined as an FEV₁ <88 percent had greater odds of having sustained smoking cessation than individuals with an FEV₁ >88 percent. However, the confidence intervals were wide and included one (OR = 1.61; 95 percent CI: 0.91, 2.87). The authors provide no explanation for the selection of the FEV₁ comparison values used in post hoc analyses.

Results from the LH-1 study suggest that the use of symptom status and baseline spirometric values including percent FEV₁, percent FEV₁/FVC and bronchodilator response reported as a percent of baseline are of limited clinical value in determining the likelihood of future smoking cessation. Spirometric values are strongly inversely associated with intensity of smoking, degree of addictiveness, and thus smoking cessation rates. In LH-1, differences in spirometric values between symptomatic individuals and individuals not reporting respiratory symptoms and across the type of symptoms were typically small (5–10 percent). At 5 years of followup, smoking cessation rates among enrollees with baseline respiratory symptoms (i.e., cough for ≥3 months/year, phlegm for ≥3 months/year, wheezing, dyspnea), were less than those without symptoms (14.7–15.8 percent vs.16.9–17.4 percent).¹⁵ However, the absolute differences in quit rates according to presence or absence of baseline symptoms or type of symptoms were small (1.2–2.8 percent). Thus the presence or type of symptoms is not a reliable clinical indicator for assessing future quit rates. Conversely, regardless of the presence or type of symptoms at baseline, there were significant differences in the point prevalence of symptoms according to the three smoking categories. All four respiratory symptoms were most common in those who continued to smoke, least common in sustained quitters, and intermediate in subjects who abstained intermittently. Symptoms were more prevalent at followup in all smoking groups among those who reported the symptom at baseline.

Additional analyses assessed the association between symptoms and changes in FEV₁ (percent predicted) during the 5-year study period. Regardless of treatment assignment or symptom status individuals with a greater loss in FEV₁ had a greater occurrence of symptoms at 5 years. The quintiles of change in spirometry ranged from a loss of ≥11 percent to a gain of ≥2 percent. The 5-year occurrence of symptoms in the intervention group from highest to lowest quintile of change ranged from 33 percent to 10 percent among individuals without baseline symptoms and 68 percent to 29 percent if a baseline symptom was present. Similar findings were observed in the usual care group. Thus there appears to be an association between change in spirometry and occurrence of symptoms. Because of the known intra-individual variability in spirometric values, it is not clear how useful these findings would be for individual patient counseling and therapeutic decisions. Both smoking cessation and the reduction in the number of cigarettes smoked per day were associated with less severe airflow obstruction at baseline. However, the magnitude of these differences was small and unlikely to be useful in clinical decisionmaking.

One study compared bupropion sustained release to placebo in 404 patients with a FEV₁/FVC ≤70 percent and clinically defined COPD.⁷⁸ At 12 months there was no statistically significant difference between subjects randomized to buproprion or placebo (10 percent vs. 7 percent abstinence; RD = 0.02; 95 percent CI = -0.04, 0.07). Prolonged abstinence rates after 26 weeks were lower in patients with more severe COPD (FEV₁ between 35 percent and 50 percent predicted) than those with more mild airflow obstruction. However, the difference was not statistically significant.

Summary of Included Study Interventions

A summary of the study characteristics, including duration, sample sizes, descriptions of control and intervention, and a brief description of participants and inclusion criteria, is provided in Summary Table 5 on pages 61–62. Only one study⁷⁹ evaluated the independent effect of obtaining and providing results of spirometry combined with targeted counseling on smoking cessation rates. In six studies⁸⁰–⁸⁵ individual smoking cessation counseling was provided to intervention and control participants, although the duration, format, and intensity of this counseling varied widely across the studies. In the remaining study⁸³ no intervention was provided to the control group. Six study designs⁸⁰–⁸⁵ involved more than one intervention being evaluated in the experimental group as compared to the control group. CO levels were incorporated into the intervention arms of two studies;⁸⁰,⁸¹ written smoking cessation materials were provided to participants assigned to the experimental group in three studies;⁸²–⁸⁴ and to both treatment arms in three studies.⁸⁰,⁸¹,⁸⁵ Blood tests,⁸² chest x-rays,⁸⁵ and symptom questionnaires with feedback⁸⁰ were each included in one study.⁸⁰ Spirometric results were provided in-person to participants in six studies⁷⁹–⁸²,⁸⁴,⁸⁵ and via mail in one study.⁸³

Methodological Quality and Characteristics of Included Studies

Study strengths and limitations are shown in Summary Table 6 on pages 63–64. Randomization to a treatment arm was clearly adequate in one study,⁷⁹ unclear in four,⁸⁰,⁸³–⁸⁵ and inadequate in two of the studies.⁸¹,⁸² Six studies⁷⁹–⁸⁴ provided data such that intention to treat results could be calculated. In the study by Li et al. the analysis completed was not intended-to-treat, as the randomization was not maintained in the analysis due to poor physician compliance in delivering the appropriate intervention.⁸⁵

Length of followup, sample size, and loss to followup. Each of the seven studies provides followup data at 9 months or longer, and two studies provide followup data at 36 months.⁸²,⁸⁴ Approximately one-fifth of the 6,052 participants randomized to a treatment group (six trials reporting) were lost to followup (n = 1,137, 19 percent). The number lost to followup in each study is shown in Evidence Table 5 in Appendix D ^*. The range of attrition rates in the seven studies was between 7 percent and 36 percent of participants per treatment arm. The rates were greater in the intervention than in the control group in three of the five studies that reported attrition rates by treatment arm. The study by Risser et al. is small with large, uneven attrition; at 12 months followup 13 of 45 participants were lost to followup in the intervention group versus 6 of 45 in the control group, although their reasons for not participating were similar across the two groups. Additionally, the results of this study may not be generalizable to other populations since the participants in this study had an average of five active medical conditions, one-quarter were enrolled in psychiatric programs, and 21 percent consumed four or more alcoholic beverages daily, all characteristics that make smoking cessation more difficult.⁸⁰

Compliance. In the Segnan study, physician compliance to the randomized treatment groups was low and study subject compliance to complete the followup visits and spirometry, if applicable, was also low.⁷⁹ Among participants, there was less than 40 percent attendance at the return visits and among those randomized to receive spirometric testing, only 50.2 percent of subjects attended this appointment. One factor contributing to the low compliance to spirometry was that subjects were asked to make a separate appointment for spirometry at another facility. In the study by Li et al., two of the four participating physicians carried out the study protocol as expected and the remaining two did not.⁸⁵ In the study by Risser et al., 7 percent of controls and 5 percent of those in the treatment group did not complete the initial 1-hour intervention.⁸⁰ Only 37 percent completed all six visits in the Richmond study.⁸²

Baseline Characteristics

Summary baseline characteristics are shown in Table 3 on page 29. Six of the seven studies provided subject age at baseline; the mean age of subjects was 42.1 years (range 16–75, n = 5,962). Gender was reported in all seven studies and the vast majority of subjects were male (90.1 percent, n = 5,453). A large proportion of the participants in the two studies that supplied race data were white (84.5 percent, n = 662); no further information regarding race or ethnicity was provided.

The average intensity of smoking among participants in six of the seven studies was 18.3 cigarettes per day (n = 5,129). The average amount of smoking was 38.5 pack-years (two studies, n = 295). Just over 70 percent of subjects had previously made at least one quit attempt (71 percent; four studies n = 1,275) and one study indicated that 36 percent of participants were in the “prepared” motivational state at baseline (n = 74).

Few studies reported respiratory symptom status or spirometry at baseline (51.6 percent of participants had any symptoms, n = 476). Of those evaluated, most subjects had relatively minimal symptoms and/or mild airflow obstruction. Three studies provided specific respiratory symptom data. In three studies, 35.6 percent of participants indicated that they had excess phlegm (n = 1,608), while 29.2 percent of subjects in two studies had cough (n = 908) and 2.8 percent of those in one study reported dyspnea (n = 16). Baseline spirometry results were presented in two of the studies. The mean FEV₁ of the 1,445 participants in one study was 2.64. The other study presented FEV₁ as a percentage (87 percent) as well as the ratio of FEV₁/FVC (76 percent) among those assigned to the intervention group (n = 103). None of the studies provided outcome data according to symptom status.

None of the participants were selected based on their motivation to quit smoking. Subjects were selected as part of a cohort of outpatients in three studies,⁷⁹,⁸¹,⁸² as part of a cohort of workers in two studies,⁸⁴,⁸⁵ and as volunteers in two studies, including volunteers of a health promotion clinic⁸⁰ and a community health survey.⁸³ Participants were selected for the study by Rose et al.⁸⁴ because they were at high cardiorespiratory risk. Likewise, smokers included in the study by Humerfelt et al.⁸³ were at high risk due to previous occupational asbestos exposure and/or adjusted FEV₁ in the lowest quartile and subjects of the study by Li et al. may have additional motivation to quit due to long-term occupational exposure to asbestos.⁸⁵

Results

Smoking cessation outcomes data for each of the seven included trials are summarized in Summary Table 7 on pages 65–66. Smoking cessation outcomes in clinical trials are measured in a variety of ways including short- and long-term point-prevalence or sustained abstinence. In general, short-term abstinence refers to outcomes less than 3 months following treatment, and may include in-treatment results depending on the duration of treatment. Long-term abstinence refers to outcomes generally measured at 6 to 12 months. In addition, at the measurement point, abstinence can be described as point prevalent (usually 7–30 days) or sustained (ranges generally from 6 months to continually from point of intervention). Finally, abstinence can be self-reported or validated by biomarkers of exposure such as carbon monoxide or cotinine. Quit attempts are regarded as a less robust, secondary process outcome.

Due to the heterogeneity of the interventions and the diverse manner in which results were reported, the calculation of a pooled estimate of cessation rates was considered inappropriate. A summary of the individual study results is provided and displayed in Figure 8 on page 30, Figure 9 on page 31, and Summary Table 7 on pages 65–66. The study by Rose et al.⁸⁴ also provided data on change in participant pulmonary function over the course of followup.

Abstinence rates. Six studies reported greater smoking cessation rates among those in the experimental groups compared to those in the control groups after 6 to 12 months of followup. The results were statistically significant in two studies.⁸²,⁸⁴ The largest study involving 2,610 subjects compared multiple interventions using a letter + informational pamphlet + questionnaire + spirometry to a group that received no intervention. The absolute difference in sustained abstinence at 12 months between groups was 1.5 percent and of borderline statistical significance.⁸³ One study showed a lower rate of abstinence among the intervention group as compared to the control group; this difference was not statistically significant.⁸¹ The range of abstinence rates for the control groups was 2.8 percent to 14 percent and among intervention groups was 6.5 percent to 39.3 percent. The range of absolute rate differences (abstinence rate _intervention - abstinence rate _control) was 1.0 percent to 33.0 percent (Figure 8 on page 30). Results for two studies are biologically verified.⁷⁹,⁸⁵ Results for the remaining five studies are based on participant self-report.⁸⁰–⁸⁴

Caution must be taken in attributing differences in cessation rates to the independent contribution of spirometry (Summary Tables 6 and 7 on pages 63–66) because most studies used interventions in addition to spirometric testing that have been proven to independently improve smoking cessation. Only one study⁷⁹ assessed the independent contribution from the process of obtaining and providing the results of spirometry to smokers in combination with focused smoking cessation counseling. Two others approximated this process.⁸⁰,⁸¹ The results of these three studies⁷⁹–⁸¹ are mixed and none were statistically significant. The study most closely adhering to this principal demonstrated a nonsignificant 1 percent greater point prevalent quit rate at 12 months in the group assigned to receive spirometry plus repeat counseling compared to repeat counseling alone (6.5 percent vs. 5.5 percent).⁷⁹ Quit rates were lower in this group than in the group that received repeat counseling plus nicotine replacement therapy (7.5 percent). The self-reported 6 month point prevalent abstinence rates for the intervention group assigned to receive spirometry in combination with advice plus carbon monoxide values was lower than the group that received advice alone (9 percent vs. 14 percent).⁸¹ The one study that showed a beneficial effect compared a 50-minute educational intervention in the control group with a similar intervention plus spirometry, carbon monoxide values, and a questionnaire and discussion of symptom status. At 12 months the biologically verified point prevalent quit rates were 20 percent in the intervention group and 6.7 percent in the control group.⁸⁰ The effect on cessation rates that occurred in the intervention group due to carbon monoxide testing and symptom assessment/discussion is not known. A summary of the study strengths and limitations are included in Summary Table 6 on pages 63–64.

Self-reported abstinence rates. Results were similar across studies when various measures of abstinence, including self-reported point prevalence abstinence at 6 to 12 months followup and sustained abstinence over the course of the study, were examined. In the study by Richmond et al., the 6-month point prevalent self-reported abstinence rate among controls was 3.0 percent compared with 35.0 percent among those in the intervention group (p <0.0001).⁸² Sippel et al. reported a higher self-reported abstinence rate among controls than among those in the intervention group at 9 months followup (14 percent vs. 9 percent); however, this result was not statistically significant (p = 0.10).⁸¹ The 12-month self-reported abstinence rates were 11.1 percent among controls versus 24.4 percent among the intervention group in the study by Risser et al. (p = 0.08),⁸⁰ 9.1 percent in controls and 11.4 percent in the intervention group in the study by Humerfelt et al. (p = 0.05),⁸³ and 8.9 percent and 39.3 percent, respectively, in the study by Rose et al. (p < 0.0001).⁸⁴ Rose et al. also provided self-reported abstinence rates at 36 months of followup of 14.5 percent among controls and 35.5 percent among those in the experimental group (p <0.0001).⁸⁴

Biologically verified abstinence rates. Biological validation of cessation, using varying definitions of abstinence, was performed in four studies.⁷⁹,⁸⁰,⁸²,⁸⁵ Studies with self-reported abstinence rates showed higher rates of abstinence than studies with biologic confirmation of abstinence. In each of the biologically validated studies, the abstinence rate among those in the intervention group was greater than the rate among the controls; these results were statistically significant in two of the four studies.⁸²,⁸⁵

Li et al. reported biologically verified abstinence rates at 11 months of followup of 2.8 percent among controls and 6.5 percent among those in the intervention group (p <0.0001).⁸⁵ In the study by Richmond et al.,⁸² the biologically validated abstinence rates at 36 months of followup were 8.0 percent among controls and 35.7 percent among those in the experimental group (p <0.001).

Sustained abstinence over the course of the study. Among the three studies that reported sustained abstinence,⁸²,⁸³,⁸⁵ higher abstinence rates were reported in the intervention groups compared to the control groups in all three studies. The results were statistically significant in two of the three.⁸²,⁸⁵ In the study by Li et al.,⁸⁵ the self-reported sustained abstinence rates over the course of the study at 11 months of followup were 3.6 percent among controls and 8.4 percent among those in the experimental group (p = 0.01) and the self-reported 12-month rates were 3.2 percent among controls receiving no intervention versus 4.7 percent among the intervention group that receives a letter providing spirometry test results, advice to quit, and a pamphlet emphasizing behavior modification (p = 0.05).⁸³ Richmond et al. reported biologically validated sustained abstinence rates at 36 months of 2.0 percent among controls and 23.5 percent among those in the intervention group (p <0.001).⁸² As previously noted the intervention group had six visits to a primary care provider for counseling and smoking cessation support versus only two visits in the control group.

Quit attempts. Risser et al. reported that 15.6 percent of participants in the control group versus 35.6 percent of participants in the intervention group had made one or more quit attempts over the 12 months of followup (p = 0.03)⁸⁰ and Sippel et al. reported that 36 percent and 48 percent, respectively, had attempted to quit over 9 months of followup (p = 0.09).⁸¹

Change in pulmonary function. Participant changes in pulmonary function were reported at 1 and 3 years post-intervention by Rose et al. At 1 year, those in the intervention group had a mean decline in FEV₁ of -0.075 compared with -0.115 in the control group. Likewise, the intervention group experienced a mean reduction in FVC of -0.132 versus -0.153 in the control group. At 3 years, the mean change in FEV₁ was -0.056 in the intervention group and -0.037 in the control group, while the mean change in FVC was -0.001 versus -0.002, respectively. This corresponded to an overall rate of change in lung function (FEV₁ and FVC) that was 14 percent less in the intervention group compared with the control group over 3 years, a difference that was highly significant statistically.⁸⁴

Demographic and Baseline Characteristic of Studies

1) Pharmacological therapies. Among the 53 studies, there is some overlap across each intervention. Most intervention trials 1) were of short duration (i.e., 6 months or less), 2) enrolled subjects with a previous clinical diagnosis of COPD, 3) had subjects with severe to very severe airflow obstruction, and 4) enrolled subjects who had symptomatic, stable COPD but relatively frequent episodes of exacerbations. Almost all studies used spirometric criteria for inclusion criteria. However few studies used spirometry for casefinding or population-based recruiting and many did not assess postbronchodilator spirometric values or categorize enrollees according to spirometric response to bronchodilators.

All studies compared a fixed dose of medications though some studies evaluated different doses of a given pharmacologic agent. Five studies were multiarm trials that compared combination therapy with inhaled long-acting beta agonists (LABA) plus corticosteroids to placebo or monotherapy with either LABA or inhaled corticosteroids. Three studies compared ipratropium monotherapy with combination therapy consisting of ipratropium with either a short or long-acting beta agonist. None titrated interventions according to short-term spirometric response to therapy, change in spirometry over time, or based on an enrollee crossing a threshold value of spirometry. None of the studies used spirometry to begin, discontinue, adjust, or monitor treatment effectiveness.

The range of mean study baseline spirometry of enrolled subjects was typically quite narrow. Only four studies evaluating inhaled corticosteroids and one study of short-acting inhaled anticholingerics had mean baseline FEV₁ percent predicted values that were greater than GOLD stage 3,4 airflow obstruction (i.e., mild-moderate severe airflow obstruction). None of the studies published subgroup outcomes according to smoking status, previous clinical diagnosis of COPD, age, race, or gender. Only two studies reported outcomes according to spirometric stage of disease⁸⁶,⁸⁷and these involved inhaled corticosteroids. Additional outcome data according to baseline symptom and spirometric status were obtained from one study of short acting anticholinergics through personal communication with the Data Coordinating Center for LH-1 (John Connett, personal communication, 2004). The few studies that followed groups of patients for longer than 1 year did not report outcomes separately according to baseline symptom status (i.e., wheezing, dyspnea, sputum production, cough, or respiratory symptoms).

The definition of our primary outcome (COPD exacerbation) varied across studies. Most studies defined exacerbations as a subjective worsening of cough, sputum, or dyspnea that required treatment with antibiotics and/or oral/intravenous corticosteroids. Other studies defined exacerbations based only on acute changes in respiratory symptoms and did not specifically require the use of additional medications. For our analyses, an exacerbation event was defined as a subject having at least one exacerbation during the treatment period. If this outcome was not available, an exacerbation was denoted by the alternative events: 1) subject having a COPD adverse event; 2) subject requiring additional treatment for COPD; 3) exacerbation/deterioration of COPD leading to study withdrawal.

Long-acting β agonists. The baseline demographic and pulmonary characteristics of the 18 studies⁴⁶,⁸⁷–¹⁰³ evaluating long-acting β2 agonists are summarized in Evidence Table 6 in Appendix D ^*. One published report was a pooled analyses of a published RCT⁹⁵ and an unpublished RCT.⁸⁹ The quality of the randomization allocation concealment method was adequate in only three trials ⁸⁷,⁹³,⁹⁴ and unclear in the remaining studies. Intention-to-treat analysis was reportedly used in 14 trials.⁴⁶,⁸⁸,⁹⁰–⁹⁴,⁹⁶–¹⁰¹,¹⁰³ All studies were double-blinded. Enrolled subjects had symptomatic COPD and severe to very severe airflow obstruction. A total of 12,390 patients, with a mean FEV₁ of 1.24L (range 0.96–1.51L) and pretreatment FEV₁ range of 33–55 percent predicted at baseline spirometry, were evaluated during 3 months to 1 year in studies assessing long-acting β agonist alone or in combination with other therapies. Long-acting β2 agonists (salmeterol or formeterol) alone were compared to placebo in 14 trials that provided exacerbation outcomes (n=6,544). Subjects randomized to active controls received tiotropium (2 trials), ipratropium (4 trials) sibenadet (1 trial), inhaled long acting corticorticosteroids fluticasone or budesonide (5 trials), or combination therapies (five with inhaled corticosteroids and one with ipratropium). Three studies compared different doses of formoterol⁹³,⁹⁷,⁹⁹ and two studies assessed different doses of salmeterol.⁴⁶,¹⁰³ The mean age was 63.5 years (n=8,029) and males were 74 percent of the subjects. Five trials provided ethnicity information.⁹⁰,⁹¹,⁹⁶,¹⁰⁰,¹⁰² Nearly 95 percent of subjects were white. Smoking history was recorded in ten trials⁸⁷–⁹²,⁹⁵,⁹⁶,⁹⁹,¹⁰² (mean of 46 pack-years) and the duration of COPD diagnosis was about 8 years (nine trials reporting).⁸⁹,⁹¹,⁹³,⁹⁵–⁹⁹,¹⁰²

Long-acting anticholinergics. In Evidence Table 7 in Appendix D ^* the general characteristics of the five clinical trials,⁸⁹,⁹⁵,¹⁰⁴–¹⁰⁶ with tiotropium (18 ug/day) are summarized. Two of the published reports⁸⁹,¹⁰⁵ were pooled analyses of two published RCTs⁹⁵,¹⁰⁶ and two unpublished RCTs. The quality of the randomization allocation concealment method was unclear in all studies. Intention-to-treat analysis was reportedly used in three trials.¹⁰⁴–¹⁰⁶ All studies were double-blinded. A total of 2,663 patients were enrolled. All had severe to very-severe airflow obstruction, (a mean FEV₁ of 1.11L with range 1.04L to 1.25L, and pretreatment FEV₁ of 38–42 percent predicted), respiratory symptoms and a previous diagnosis of COPD. Treatment duration ranged from 3.3 months to 1 year. Of the 2,663 patients, 1,308 (49.1 percent) received monotherapy with tiotropium. Twenty-nine percent received placebo (two reports, n=771 subjects) or ipratropium bromide (one report, n=179 subjects). The others (15 percent) were only treated by long-acting β2 agonists (salmeterol; one study). The mean age was 64.5 years and 70 percent of subjects were male. No further information regarding race or ethnicity was provided. Subjects had a mean of 48 pack-years smoking and had COPD for approximately 9 years.

Short-acting anticholinergics. There were eight published reports, five multi-armed, involving the short-acting anticholinergic ipratroprium (typically 40 ug three to four times/day).¹⁹,⁹⁸,⁹⁹,¹⁰⁰–¹⁰²,¹⁰⁵,¹⁰⁶ Seven involved ipratropium monotherapy, (including five versus placebo, four versus long-acting β agonists and one versus tiotropium) and one combined ipratropium with salmeterol in comparison to salmeterol alone and placebo. One published report was a pooled analyses of a published RCT and an unpublished RCT¹⁰⁵,¹⁰⁶. The quality of the randomization allocation concealment method was unclear in all studies, except the LH1.¹⁰⁷ Intention-to-treat analysis was reportedly used in five trials.⁹⁸–¹⁰⁰,¹⁰⁵,¹⁰⁶ All studies were double-blinded. Two studies compared outcomes to the long-acting anticholinergic tiotroprium. Studies providing data on exacerbations were 3 months in duration. General characteristics are summarized in Evidence Table 8 in Appendix D ^*. A total of 8,489 patients with moderate to severe COPD (a mean FEV₁ of 2.21L with range 1.18L to 2.64L and pretreatment FEV₁ of 33–46 percent predicted) were evaluated during 3 months to 1 year. Of the 8,345 patients, 2,667 (31 percent) were randomized to ipratropium monotherapy and 5,466 patients (66.7 percent) to placebo, tiotropium, or usual care. Combination therapy with salmeterol was evaluated in 47 subjects. In comparison to other interventions, studies evaluating ipratroprium enrolled a higher percentage of relatively young individuals, women, subjects with mild to moderate airflow obstruction, and those without symptoms or previous clinical diagnosis of COPD. Therefore, subjects enrolled in these studies more closely represent a spectrum of the population likely to be detected by case finding in primary care settings. The mean age was 52.9 years (n=5,523). Sixty-five percent were men. Nearly 95 percent were white (n=two studies).¹⁰⁰,¹⁰² The mean smoking history was 41 pack-years (n=five studies)¹⁹,⁹⁹,¹⁰²,¹⁰⁵,¹⁰⁶ and the mean duration of COPD was 9.8 years.⁹⁸,⁹⁹,¹⁰²,¹⁰⁵,¹⁰⁶

Combination therapy with inhaled short-acting anticholinergics and β2 agonists. Combination therapy (inhaled short acting anticholinergics bronchodilators [ipratropium bromide] + short (or long)-acting β2 agonists (albuterol or salmeterol) studies are summarized in Evidence Table 8 in Appendix D ^*.¹⁰⁸–¹¹⁰ A total of 1,186 patients with severe to very severe airflow obstruction and symptomatic COPD (a mean FEV₁ of 0.95L with range 0.91L to 1.00L and pretreatment FEV₁ of 34 percent to 37 percent predicted) were evaluated during 85 days. Of all patients, 404 (34.1 percent) had been treated by combination therapy. Ipratropium and albuterol, were given to 393 (33.1 percent) and 389 (32.8 percent), respectively. The mean age was 64.4 years (n=1,186) and approximately 65 percent were men. Over 93 percent of enrollees were white and the mean duration of COPD was 8.9 years.

Inhaled corticosteroids. Thirteen studies evaluated inhaled corticosteroids and are summarized in Evidence Table 9 in Appendix D ^*. The quality of the randomization allocation concealment method was adequate in five studies,⁸⁶,⁸⁷,¹¹¹–¹¹³ and unclear in the other studies. Intention-to-treat analysis was reportedly used in 11 trials.⁸⁶,⁸⁸,⁹¹,⁹²,⁹⁶,¹¹¹–¹¹⁶ All studies were double-blinded. A total of 8,849 patients were enrolled. Studies evaluating inhaled corticosteroids enrolled subjects with a relatively wide spectrum of clinical and airflow severity. Studies also assessed treatment over several years. Enrolled subjects (a mean FEV₁ of 2.0L with range 0.91L to 2.53L and pretreatment FEV₁ of 36–77 percent predicted) were evaluated from 6 months to 4.5 years. Of all patients, 3,247 (36.7 percent) had been treated by inhaled corticosteroids (fluticasone or triamcinolone; or budesonide or beclomethasone) and 3,257 patients had only taken a placebo (36.8 percent). The mean age overall was 60 years (n=6,504). The percent of males overall was about 71 percent. According to three trials⁹¹,⁹⁶,¹¹⁴ with ethnicity information, the proportion of white subjects was 94 percent. All subjects in ten trials⁸⁶–⁸⁸,⁹¹,⁹²,⁹⁶,¹¹¹,¹¹⁵–¹¹⁷ had a smoking history with a mean of 44 pack-years.

Combination corticosteroids and long-acting β agonists. Five studies evaluated inhaled corticosteroid and long-acting β agonist combination therapy (Evidence Table 9 in Appendix D ^*).⁸⁷,⁸⁸,⁹¹,⁹²,⁹⁶ All studies were parallel-grouped, placebo-controlled, and double-blinded. The quality of the randomization allocation concealment method was adequate in one trial⁸⁷ and intention-to-treat analysis was reportedly used in four studies.⁸⁸,⁹¹,⁹²,⁹⁶ Study duration ranged from 6 to 12 months. A total of 4,713 subjects were enrolled, approximately 25 percent randomized to combination, placebo, and monotherapy arms each. The subjects were, on average, 64 years old, had a baseline FEV₁ of 1.2L (range 0.96 to 1.3, FEV₁ of 36–45 percent predicted), had a smoking history with a mean of 46 pack-years, and a median duration of COPD of 6 years (two studies reporting.⁹¹,⁹⁶ Two trials reported ethnicity and nearly all subjects were white (94 percent).⁹¹,⁹⁶

D2 antagonist. Sibenadet was evaluated in three studies involving four study protocols⁹⁰,¹¹⁸,¹¹⁹ and summarized in Evidence Table 10 in Appendix D ^*. One study included two trials according to study duration (3 or 6.5 months).¹¹⁹ A total of 4,077 patients with a mean FEV₁ of 1.29L (range 1.2–1.4L) and pretreatment FEV₁ of 39–42 percent predicted at baseline were evaluated during 3 to 13 months. Of all patients, 1,977 (48.5 percent) had only been treated by sibenadet. A placebo was given as treatment to 1,547 (37.9 percent). The others (13.6 percent) were treated with salmeterol. The mean age was 63.9 years (n=3,523). The percent of males was 71 percent. According to one trial⁹⁰ with ethnicity information, the proportion of whites was 97.2 percent. In four trials subjects had a mean of 47 pack-years.

Oral Purified Bacterial Extracts. A systematic review and meta-analysis identified 13 trials (1,971 patients) of oral purified bacterial (active) extracts in patients with chronic bronchitis and COPD.¹²⁰ Ten studies tested OM-85BV, two trials tested LW-50020, and one study tested SL-04.¹²⁰ Study duration ranged from 3–12 months. In trials that reported demographic information, 60 percent were male. Inclusion criteria were COPD in six trials, chronic bronchitis in ten trials, and more than three episodes of exacerbation within the previous year in eight trials. In the seven studies that reported smoking habits, almost half of analyzed patients were smokers or ex-smokers. Lung function was reported in five trials (mild to moderate COPD, four trials; severe COPD, one trial).¹²⁰

2) Nonpharmacological therapies. The general characteristics of the seven studies¹²¹–¹²⁷ of pulmonary rehabilitation intervention are summarized in Evidence Table 11 in Appendix D ^*. One study included two trials according to disease severity (moderate or severe).¹²⁷ Subjects enrolled in nonpharmacological therapy trials typically had severe to very severe airflow obstruction and respiratory symptoms, already received a clinical diagnosis of COPD, and were already receiving a wide assortment of pharmacologic agents. Thus they are unlikely to be representative of the vast majority of subjects detected by casefinding with spirometry in primary care settings, whom this report is targeted to address. We provide this information for the sake of completeness.

A total of 693 patients with a FEV₁ range (0.71–1.07L) and pretreatment FEV₁ of 31–50 percent predicted at baseline were evaluated from 8 weeks to 2 years. The general characteristics of the nine studies using interventions of disease management, education, and followup are summarized in Evidence Table 12 in Appendix D ^*. A total of 1,997 patients with a FEV₁ range (0.78-post 1.71L) and pretreatment FEV₁ of 37–59 percent predicted at baseline were evaluated from 3 months to 1 year. Also, the general characteristics of the two studies intervened by NIMV are summarized in Evidence Table 13 in Appendix D ^*. A total of 220 patients with a FEV₁ range of 0.73L and pretreatment FEV₁ of 30 percent predicted at baseline were evaluated from 6 months to 2 years.

Outcomes by Intervention

1) Pharmacological therapies.

Long-acting β2 agonists as monotherapy (LABA). (Figure 10 on page 67, Evidence Figures 1 and 2 in Appendix D ^*, and Evidence Tables 14 and 15 in Appendix D ^*.) Thirteen placebo-controlled trials (6,544 patients, baseline) followed patients from 3 to 12 months. Compared to placebo, both formoterol and salmeterol reduced exacerbations as well as improved St George's Respiratory Questionnaire scores. There was a pooled 18 percent relative risk reduction (95 percent CI, 10–24 percent) and 4 percent pooled absolute risk reduction [95 percent CI, -6 to -2] in the percentage of individuals having one or more COPD exacerbation events during study followup. Reductions were consistently seen across studies with each agent and were similar in studies utilizing formeterol and those using salmeterol. Only three studies reported rates of hospitalization. They were reduced by about 5 percent compared to placebo in one study and not different in two others. The few dose comparison studies of the long acting β2 antagonists salmeterol or fometerol found similar improvements in all clinical outcomes at differing doses of medication. This suggests that increasing the dose of β2 antagonist beyond salmeterol 50 ug/bid or fometerol 12 ug/bid is not beneficial.

There was no significant difference in all-cause mortality between placebo and LABA (12 studies, 5,700 patients) (RR=1.07; 95 percent CI: 0.65–1.78). Absolute risk reduction was 0 percent [95 percent CI, -1 to 1]. Improvement was demonstrated in health-related quality of life scores as measured by the SGRQ in seven studies of subjects with GOLD Stage 3,4 disease (1.98-unit improvement; 95 percent CI: 0.81–3.15 vs. placebo). However, the pooled weighted mean difference in SGRQ compared to placebo failed to achieve a previously specified level of clinical significance (i.e., ≥4). Additionally, the only two studies that individually reported a clinically important difference in SGRQ enrolled subjects with severe to very severe airflow obstruction (mean FEV₁ predicted <50 percent).

The narrow range of mean baseline spirometric values (FEV₁ range=1.1–1.5; percent predicted=33–54 percent; GOLD 3,4) precludes assessment of whether treatment effectiveness varied according to baseline spirometry. The high percentage of subjects in the placebo arm that had at least one COPD exacerbation suggests that in all but one study⁹³ subjects had severe symptoms and frequent exacerbations. In two trials lasting 6 months (n=1,212) salmeterol provided similar reductions in exacerbations compared to tiotropium (RR=1.07; 95 percent CI 0.92 to 1.25).⁸⁹,⁹⁵

Long-acting anticholinergics: tiotropium. (Figure 11 on page 68, Evidence Figure 3 in Appendix D ^*, and Summary Tables 8 and 9 on pages 69–71.) Five clinical trials of the long-acting, anti-cholinergic tiotropium (n=2,956) in patients with severe to very severe airflow obstruction (mean FEV₁ percent predicted = 39–41 percent) and respiratory symptoms demonstrated a reduction in exacerbations compared with either placebo (RR = 0.84; 95 percent CI, 0.74–0.95) or with the short-acting anti-cholingeric ipratropium bromide (RR = 0.77; 95 percent CI, 0.62–0.95). Pooled absolute risk reductions were 6 percent (95 percent CI, -11 to -2) compared to placebo and 11 percent (95 percent CI, -20 to -2) versus ipratropium. Subjects enrolled in these studies had frequent episodes of exacerbations. The weighted mean percentage of subjects with at least one exacerbation during the 3–12 month study period in the control group was approximately 40 percent. Hospitalization rates were reported in three studies and found to be lower by about 7 percent compared to placebo and 4 percent versus ipratropium. All-cause mortality was decreased versus placebo (RR = 0.50; 95 percent CI 0.17 to 1.24) in two trials (n=1,723),⁸⁹,¹⁰⁴ and the absolute risk reduction was one percent (95 percent CI, -2 to 0). The risk of all-cause mortality was increased compared with ipratropium in one trial, although not significantly (RR = 1.51; 95 percent CI 0.41 to 5.50).¹⁰⁵ Tiotropium also improved scores on the SGRQ health-related quality of life scale relative to placebo (2.7–3.7 unit improvement) and ipratropium (3.3 unit improvement) though the reduction failed to achieve a previously determined level of clinical significance. Exacerbations were similar compared to long-acting β2 agonists (RR for exacerbations vs. long-acting β2 agonists, 0.92; 95 percent CI, 0.75–1.11) in pooled results from two studies.⁸⁹

Short-acting anticholinergics alone or in combination with β2 agonists. (Figure 12 on page 72, Evidence Figure 4 in Appendix D ^*, and Summary Tables 10 and 11 on pages 73–75.) Ipratropium was no more effective than placebo and less effective than tiotroprium in reducing exacerbations and hospitalizations. In four trials of patients with GOLD Stage 3,4 COPD, (range of FEV₁ percent predicted = 33–45 percent) with followup of 3 months, ipratropium monotherapy did not significantly reduce the percentage of subjects having one or more COPD exacerbations compared with a placebo (ARR = 1.3 percent; RR = 0.95 [0.78 to 1.16]. Between 18 percent and 38 percent of subjects experienced a study-defined exacerbation suggesting that, on average, individuals had symptomatically severe COPD with frequent exacerbations. Three studies reported results from validated respiratory functional status questionnaires.⁹⁸,⁹⁹,¹⁰⁵ The change from baseline between control and ipratroprium was small and less than considered clinically significant (four point difference) in two studies⁹⁸,⁹⁹ and favored tiotropium by three points in the other trial.¹⁰⁵ Exacerbation rates for LH-1 (n=3,923; mean FEV₁ = 2.6; FEV₁ percent predicted = 75 percent) have not been published. Additional outcomes from LH-1, which is the only trial that enrolled subjects with mild to moderate airflow obstruction regardless of symptoms, are described below. Combination therapy with either short or long acting β agonist (four trials) in addition to ipratropium did not reduce exacerbations compared to ipratropium alone (RR=1.03 95 percent CI=0.64 to 1.67) but did versus β agonists (RR=0.68 95 percent CI=0-.51 to 0.91).¹⁰⁶,¹⁰⁸–¹¹⁰

In the one study comparing ipratropium to tiotropium (n=535)¹⁰⁵ the percentage of subjects having at least one exacerbation was higher in the ipratroprium group (46 percent) than subjects randomized to receive tiotropium (35 percent). There was no significant difference in mortality rates between ipratropium and control groups. However, only one study (LH-1) followed patients for more than 1 year. The overall mortality rate in the placebo arm of 2.2 percent was less than the ipratropium group (2.8 percent).

Inhaled corticosteroids. (Figure 13 on page 76, Evidence Figure 5 in Appendix D ^*, and Summary Tables 12 and 13 on pages 77–80.) In ten placebo-controlled trials (3,734 patients)⁸⁶–⁸⁸,⁹²,⁹⁶,¹¹¹–¹¹³,¹¹⁶,¹¹⁷ with at least a 6-month followup period, inhaled corticosteroids led to a 22 percent relative reduction in the percentage of subjects having a COPD exacerbation event (RR 0.78; 95 percent CI, 0.70–0.88). The pooled absolute risk reduction was 5 percent (95 percent CI, -8 to -3). Six of the studies were 1 year in duration or longer and the percentage of subjects in the placebo arm that experienced at least one exacerbation was 22.4 percent.⁸⁷,⁸⁸,⁹²,¹¹¹,¹¹²,¹¹⁷ An additional study¹¹³ enrolled 1,116 smokers with moderate airflow obstruction (FEV₁ percent predicted = 64 percent) and is the only trial of inhaled corticosteroids to report on rates of hospitalizations. Members of the triamcinolone group had fewer overall respiratory symptoms during the course of the study (21.1 per 100 person-years vs. 28.2 per 100 person years, p = 0.005), had fewer visits to a physician because of a respiratory illness (1.2 per 100 person-years vs. 2.1 per 100 person-years), and fewer hospitalizations for respiratory conditions (0.99 per 100 person years vs. 2.1 per 100 person years). There was no significant association between triamcinolone use and specific respiratory symptoms including chronic cough, production of phlegm, or wheezing.

As shown in the meta-analysis by Sin et al.⁹ and subgroup data from the study by van der Valk⁸⁶ the beneficial effect of inhaled corticosteroids was associated with the severity of airflow obstruction as measured by FEV₁. Whereas the study that had the highest mean FEV₁ value failed to demonstrate a beneficial effect of inhaled corticosteroids, trials that had a mean FEV₁ of less than 1.7L or lower (mean baseline FEV₁ percent predicted 36–57 percent) demonstrated a positive effect of inhaled corticosteroids on exacerbations, regardless of the duration of the study or the specific formulation used. Analysis of the subgroup of patients with a FEV₁ value less than 50 percent predicted (low FEV₁ group) suggests that the improvement time to first exacerbation due to fluticasone observed in the COPE trial is driven by this group. The hazard ratio was 2.1 (95 percent CI 1.1–36) and 1.2 (95 percent 0.8–2.0) in the low and high-FEV₁ groups respectively. Inhaled corticosteroids resulted in a 19 percent reduction in all-cause mortality though the confidence intervals were wide and not statistically significant (RR 0.81 95 percent CI = 0.60 to 1.10) and the absolute reduction was only 0.6 percent. Changes in the SGRQ reported in two trials involving 995 subjects followed from 6 months to 3 years were less than considered clinically significant.⁸⁶,¹¹¹

Combination corticosteroids and long-acting inhaled β agonists. (Figure 14 on page 81, Evidence Figures 6–12 in Appendix D ^*, and Summary Tables 14 and 15 on pages 82–84.) Five multi-arm trials (n=1,982 patients)⁸⁷,⁸⁸,⁹¹,⁹²,⁹⁶ evaluated monotherapy with either LABA or inhaled corticosteroids compared to combination therapy with these agents and to placebo. Thus they provide direct comparative evidence regarding the relative effectiveness of either agent or combination therapy to placebo as well as to their respective monotherapies. The ARR in exacerbations compared to placebo seen with both monotherapies and combination therapy were all statistically significant and of similar magnitude (ARR compared to placebo for LABA = 3.7 percent; corticosteroids = 5.2 percent and combination therapy = 5.9 percent). The addition of inhaled corticosteroids to LABA resulted in a borderline significant reduction in exacerbations compared to LABA alone (RR = 0.82 [0.65, 1.04]; ARR = 1.3 percent). When compared to monotherapy with corticosteroids, there was approximately one-half the reduction reported for comparison to LABA monotherapy (RR = 0.92) and the absolute risk reduction was less than 1 percent. The mean baseline FEV₁ ranged from 36–45 percent predicted indicating subjects had very severe airflow obstruction (GOLD Stage ≥3). A subgroup analysis reported by Calverley indicated that therapeutic effectiveness varied by severity of baseline spirometry. While the relative risk reduction for combination therapy compared to placebo was 39 percent for all enrollees, individuals with FEV₁ >50 percent predicted had only a 10 percent relative risk reduction. Improvements in respiratory symptoms compared to placebo as measured at 1 year by the SGRQ were less than considered clinically significant in one trial) (WMD = -2.2; 95 percent CI = -3.3 to -1.1)⁸⁷ and one study demonstrated a large and clinically relevant improvement of 7.5 units.⁸⁸ Compared to placebo, combination therapy reduced all-cause mortality by 44 percent, but the confidence intervals were wide and not statistically significant and the absolute reduction was 0.7 percent (RR vs. placebo, 0.66; 95 percent CI, 0.32–1.38). (Summary Table 14 on pages 82–83 and Evidence Figure 6 in Appendix D ^*.) The addition of LABA to inhaled corticosteroids did not reduce mortality compared to corticosteroids alone (RR = 0.98). However, when compared to LABA, combination therapy with corticosteroids resulted in nearly a 54 percent reduction in mortality though there were relatively few deaths. Thus, monotherapy with corticosteroids may be slightly more effective in reducing exacerbations than LABA. The addition of LABA to corticosteroids does not reduce exacerbations or improve mortality or respiratory status compared to monotherapy.

Sibanet (D2 receptor/β agonist). (Evidence Figures 13 and 14 and Evidence Tables 16 and 17 in Appendix D ^*.) Four trials evaluated the D2 receptor/β agonist, sibanet (n=3,524 patients)¹¹⁸,¹¹⁹,¹²⁸ in patients with GOLD Stage 3,4 airflow obstruction. Compared to placebo there was no significant difference in exacerbations (RR, 0.91; 95 percent CI, 0.81 to 1.03; ARR = -0.7). Changes in the SGRQ were small and less than considered clinically important. Only one study was at least 1 year in duration and all-cause mortality was not different between sibanet and placebo (RR vs. placebo, 1.14; 95 percent CI, 0.60–12.15).

Oral purified bacterial extracts. There was a large variety in the reported end points. No more than five trials reported on the same efficacy end point. Three trials reported on the prevalence of exacerbation with OM-85BV or SL-04 over a 6-month period. Two trials (731 patients) were judged to be of high methodologic quality. The combined RR for prevention of exacerbations was 0.83 (95 percent CI, 0.55 to 1.25). Itching or cutaneous eruptions were reported in 3.3 percent of subjects who received active extracts compared with 1.0 percent of control subjects. Data on hospital admission for respiratory problems was reported in 31 of 191 patients (16.2 percent) receiving OM-85BV and in 44 of 190 patients (23.2 percent). Urologic problems (primarily urinary tract infections) were reported in 8 percent of patients who received active extracts compared with 3.0 percent of control subjects.¹²⁰

Overall withdrawals from treatment, noncompliance, and adverse events were examined for trials 1 year or longer in duration. Subjects treated with a β agonist, tiotropium, or a corticosteroid were less likely to withdraw from treatment for any reason compared to placebo or control. The percent of β agonist subjects withdrawing was 30.8 percent compared with 37.9 percent of the placebo subjects in four trials >1 year (ARR = 7.1; RR = 0.86: 95 percent CI 0.77 to 0.96).⁸⁷,⁸⁸,⁹²,⁹⁷ The overall withdrawal rate for subjects treated with tiotropium was 17.3 percent compared with 27.8 percent and 21.2 percent of placebo and ipratropium subjects, respectively (ARR = 8.3; RR = 0.69: 95 percent CI 0.56 to 0.84).¹⁰⁴,¹⁰⁵ Subjects on corticosteroids had a withdrawal rate of 26.5 percent versus 31.9 percent of placebo subjects in seven trials reported withdrawal data (ARR = 5.45; RR = 0.83: 95 percent CI = 0.76 to 0.90).⁸⁷,⁸⁸,⁹²,¹¹¹,¹¹²,¹¹⁴ The one trial of ipratropium reporting withdrawal data favored the control, tiotropium, 15.2 percent to 21.2 percent (RR = 1.40: 95 percent CI 0.96 to 2.03).¹⁰⁵ In trials of combination corticosteroids and long-acting β agonist, withdrawals were lower for combination therapy compared with placebo but were similar compared with either monotherapy.

Four trials reported withdrawal from treatment due to noncompliance.⁸⁷,⁹⁷,¹⁰⁵,¹¹¹ The rates of withdrawal ranged from <1 to 16 percent for treatment and <1 to 16 percent for placebo or control. The LHS2 trial reported adherence to treatment based both on patient report and canister weight. Approximately 70 percent of triamcinolone and placebo subjects had satisfactory adherence to the treatment protocol based on self-report. However, these rates decreased to 53.7 percent and 58.5 percent respectively based on canister weights for triamcinolone and placebo.¹¹⁴

Treatments were generally well tolerated. Adverse events during the study followup period were usually minor and seldom more than placebo. Compared to placebo, an increased frequency of oropharyngeal candidiasis (5.1 percent vs. 2.1 percent), throat irritation (7.6 percent vs. 4.5 percent), and bruising (8.4 percent vs. 3.7 percent) was seen with corticosteroid use. Dry mouth was reported in 12 percent of subjects using anticholinergics.¹⁰⁴,¹⁰⁵ A separate meta-analysis of 20 RCTs involving 6,623 subjects by Salpeter and colleagues assessed the cardiovascular effects of β agonists (primarily the long-acting β agonists salmeterol and fometerol) in patients with asthma or COPD. Treatment with β agonists was associated with a significantly increased risk for adverse cardiovascular events (RR 2.54; 95 percent CI 1.59 to 4.05; 2.7 percent vs. 0.7 percent). The vast majority of the adverse events in the β agonist group were due to sinus tachycardia (87 percent) and thus of uncertain clinical significance. However, major cardiovascular events were also higher in this group compared to placebo though not statistically different (RR = 1.66; 95 percent CI 0.76 to 3.60).¹²⁹

2) Nonpharmacological therapies

1. Pulmonary rehabilitation program. (Evidence Table 18 in Appendix D ^*.) Patients with advanced COPD experience marked dyspnea and exercise intolerance related in part to generalized muscle weakness, cardiac impairments, and nutritional deficiencies. Pulmonary rehabilitation programs were developed to address some of these adverse physiological changes. The contents of pulmonary rehabilitation vary from center to center. However, most contain four major components: exercise training, education, behavioral modification, and outcome assessment. The intensity of the exercise training is heterogeneous. Most aerobic training is targeted at 60–90 percent of the predicted maximal heart rate for about 30 minutes. Most programs emphasize endurance training. The eight clinical trials (693 patients) indicate that pulmonary rehabilitation may improve the health status of patients with severe to very severe COPD (mean FEV₁, 0.71 to 1.07L; FEV₁ percent predicted = 31–50 percent) as assessed by SGRQ and increases exercise tolerance beyond that achieved by standard care alone (including inhaled bronchodilators), at least during the time patients are in the rehabilitation program. Three of the eight trials reported an improvement in the SGRQ between control and intervention greater than the four point minimally important difference.¹²³,¹²⁴,¹²⁷ As noted by Sin, pulmonary rehabilitation did not have any significant effect on mortality.⁹

2. Disease management, education, and followup studies. (Evidence Table 19 in Appendix D ^*.) Disease management is an approach to coordinate resources across the health care system with the aim of fostering continuity of care and increasing patients' knowledge and control over their chronic diseases.¹³⁰ Because the care of patients with COPD frequently requires multiple caregivers, including physicians, nurses, physiotherapists, pharmacists, and nutritionists, a process to promote integration and seamless care may improve clinical outcomes in COPD. Sin et al. noted that because of marked heterogeneity in the content of the programs and their effects, these data need to be interpreted cautiously and further study is required.⁹ Patients enrolled in these programs had moderate to very severe airflow obstruction (FEV₁ percent predicted = 37–59 percent), had been previously diagnosed clinically with COPD, and were taking inhaled bronchodilators. It is likely that individuals involved in these programs represent an extremely small fraction of adults likely to be detected by case finding with spirometry. Only one trial reported exacerbation rates and rates of hospitalization were not consistently different between intervention and controls.¹¹⁶ On average, these programs did not achieve a clinically meaningful improvement in health status of patients or a statistically significant impact in hospitalization rates.

3. Non-invasive mechanical ventilation (NIMV). Respiratory muscle fatigue and dynamic hyperinflation commonly are observed in patients with severe COPD.¹³¹–¹³³ Patients with severe COPD work harder than patients without COPD because they have to overcome dynamic lung hyperinflation and airflow obstruction.¹³²,¹³³ Long-term NIMV therapy theoretically unloads the inspiratory muscles of respiration and helps restore depleted energy stores, as well as partially reversing respiratory muscle fatigue.¹²⁸ Sin and colleagues concluded that the long-term use of NIMV cannot be recommended at this time because there is insufficient clinical trial evidence for its efficacy.²¹,¹³⁴–¹³⁶

4. Influenza and pneumococcal vaccinations. Elderly persons and persons with certain underlying medical conditions experience more than 80 percent of the serious complications of influenza, such as hospitalization and death.¹³⁷,¹³⁸ Among elderly persons, those with a clinical diagnosis of chronic lung disease are an especially high-risk group. Their hospitalization rates for pneumonia are two to seven times those of individuals without underlying pulmonary conditions. Influenza vaccination is recommended for all adults, especially those with chronic medical conditions. High-risk elderly persons, such as those with chronic lung disease, have inadequate vaccination rates.¹³⁹ Methods to improve influenza vaccinations in these individuals would be beneficial.

A retrospective, multiseason cohort study in a large managed care organization assessed the effects of influenza and the benefits of influenza vaccination in elderly persons with a diagnosis of chronic lung disease during the previous 12 months. Influenza vaccination was associated with fewer hospitalizations for pneumonia and influenza (adjusted risk ratio, 0.48 [95 percent CI, 0.28 to 0.82]) and with lower risk for death (adjusted odds ratio, 0.30 [CI, 0.21 to 0.43]) during the influenza seasons. It was also associated with fewer outpatient visits for pneumonia and for all respiratory conditions.¹⁴⁰ However, there is no information assessing whether vaccination rates are improved by case finding with spirometry. Additionally, there is no information to determine whether spirometry should be used to identify asymptomatic individuals with airflow obstruction who should be considered at increased priority for influenza vaccination, especially if vaccination supplies are limited.

Streptococcus pneumoniae is a major cause of morbidity and mortality. Pneumococcal vaccination is often recommended for preventing invasive disease for the elderly and others who are at increased risk for serious pneumococcal infections and their complications. This includes individuals with chronic lung disease. A recent meta-analysis of randomized or quasi-randomized controlled trials assessed the effectiveness of pneumococcal vaccination¹⁴¹ in preventing pneumonia, bronchitis, and mortality. Patient populations and type of vaccine varied in these trials. Two trials limited enrollment to subjects with COPD (n=292). The authors of the meta-analysis concluded that despite encouraging data from some very early trials, pooling trial results published from 1977 on suggest there is no significant effect on pneumonia (14 trials, n=75,008 subjects; OR = 0.77, 95 percent CI = 0.58, 1.02) or death (OR = 0.90, 95 percent CI = 0.90, 1.07). In the two small trials involving subjects with COPD, the odds of definitive pneumococcal pneumonia were actually higher in the groups receiving vaccine than control, though “all-cause pneumonia” was less common in vaccine recipients in one of these studies. The pooled results from case-control studies did demonstrate a significant efficacy in preventing invasive pneumoccocal disease (OR = 0.47 [CI = 0.37, 0.59]) as have other cohort studies among elderly persons with chronic lung disease. However, these pooled data from RCTs suggest that pneumococcal vaccination may not reduce morbidity and/or mortality, especially in individuals with COPD. Furthermore, even if a decision is made, despite the findings from this meta-analysis of RCT, to routinely vaccinate elderly individuals or those with chronic lung disease, there is no evidence that spirometry leads to improved vaccination rates or that outcomes are improved in individuals not reporting respiratory symptoms who have airflow obstruction.

Does Treatment Effectiveness Vary According to Baseline Spirometry, Spirometric Response to Treatment, and/or Change in Spirometry Over Time?

Periodic monitoring with spirometry has been recommended as a guide to treatment response and/or patient health status.¹⁴² However, correlation between spirometric changes and long-term clinical outcomes in COPD has been shown to be weak.²² As noted above, most treatment trials enrolled subjects who had both moderate to severe respiratory symptoms and severe to very severe airflow obstruction. All studies used spirometry to confirm and quantify the presence and severity of airflow obstruction and used spirometric values as entry criteria and most ruled out a clinically significant bronchodilator response. None of the trial protocols involved modification of treatment according to spirometry. Results from large long-term RCTs of inhaled corticosteroids and anti-cholinergics demonstrate that these interventions do not alter the course of spirometric decline. Two studies assessed clinical response according to short-term change in spirometry.

A conceptual model and flow diagram Figure 15 on page 85 illustrates how periodic monitoring with spirometry (e.g., annually or every 3–5 years) may be used to identify symptomatic individuals with normal airflow to mild to moderate airflow obstruction who may subsequently develop severe to very severe airflow obstruction and thus be candidates for treatment. The starting point is based on the pooled summary of treatment effectiveness indicating that interventions with the exception of smoking cessation and influenza vaccinations were only effective in symptomatic individuals (regardless of smoking status) who had severe to very severe (approximately GOLD Stage 3,4) airflow obstruction. Periodic monitoring with spirometry in patients reporting respiratory symptoms would assist the health care provider in initiating or modifying therapy if data demonstrated that outcomes are improved if treatment initiation or modification is based on 1) acute spirometric response to therapy, 2) change in spirometry over time (slope of FEV₁ decline), or 3) crossing a given followup spirometric threshold (e.g., transition from mild-moderate to severe or very severe; approximately GOLD 1 or 2 to GOLD 3 or 4).

Effectiveness of Treatment According to Baseline Spirometry

All studies of long- and short-acting inhaled anti-cholinergics and long-acting β agonist, except for LH-1 evaluating ipratropium, assessed individuals with severe to very severe airflow obstruction and respiratory symptoms (mean FEV₁ ranged from 0.96–1.51 and FEV₁ percent predicted from 33–55 percent; approximately equivalent to GOLD Stage 3). Therefore it is not possible to determine the effectiveness of long-acting anti-cholinergics or β agonist in subjects with spirometry demonstrating mild to moderate airflow obstruction. However, information is available from other inhaled agents suggesting that a spirometric threshold for treatment effectiveness exists and that treatments do not prevent the development of symptoms among individuals not reporting respiratory symptoms.

As shown by Sin and colleagues and confirmed in our results, the effectiveness of inhaled corticosteroids is associated with baseline spirometry as measured by FEV₁. Three trials enrolling approximately 2,500 subjects with a mean FEV₁ >2L (GOLD Stage 0–2) and followed for 3 or more years failed to demonstrate a benefit in clinical outcomes, although there was a trend towards a reduction of mortality. Analysis of the subgroup of patients with a FEV₁ value less than 50 percent predicted (low FEV₁ group) suggests that the improvement in time to first exacerbation due to fluticasone observed in the COPE trial is driven by this group. The hazard ratio was 2.1 (95 percent CI 1.1–36) and 1.2 (95 percent 0.8–2.0) in the low- and high-FEV₁ groups respectively. The study by Calverley and colleagues evaluated inhaled corticosteroids alone or in combination with long acting β agonists in subjects with a mean baseline FEV₁ percent predicted of 45 percent (GOLD Stage 3).⁸⁷ They observed that treatment effectiveness was associated with disease severity as measured by baseline spirometry. Compared to placebo, combination therapy resulted in a relative risk reduction in exacerbations of 39 percent. However, in the subgroup with baseline FEV₁ >50 percent (moderate airflow obstruction; GOLD Stage 2) the relative risk reduction was only 10 percent (P value and confidence intervals not provided).

The largest and longest study assessing inhaled bronchodilators (LH-1) compared outcomes of 3,923 adult smokers who were at risk for or had mild to moderate airflow obstruction and treated them with ipratropium vs. placebo over an average of 5 years. Prevalence of baseline symptoms (LH-1) according to spirometric category are shown in Table 4 on page 45. Only a small percentage of subjects had a previous diagnosis of COPD, less than one-half reported dyspnea, about 5 percent had normal spirometry and sputum production (GOLD 0), and almost 20 percent reported no respiratory symptoms. Therefore subjects enrolled in LH-1 are representative of adults likely to be detected by spirometric case finding. In unpublished data obtained from the Data Coordinating Center Director (John Connett, personal communication, 2004) there was no reduction at 3 years in respiratory hospitalizations for subjects with baseline post-bronchodilator assessed GOLD spirometric stages “Normal,” 0, 1, or 2 (Evidence Table 20 in Appendix D ^*) in subjects randomized to ipratropium compared with placebo. Ipratroprium did not improve outcomes of dyspnea (31.0 percent vs. 31.2 percent), cough and sputum (14.9 vs. 15.0 percent), or respiratory hospitalizations in the overall cohort. Results were not different when assessed according to baseline spirometric stage or symptom status (Table 5 on page 46). The presence of symptoms at baseline, rather than spirometry or treatment, was the best predictor of symptoms at the 3-year followup. Additional analysis demonstrated that ipratropium did not improve the percentage of subjects having dyspnea and cough and sputum at 3 years regardless of presence or absence of these symptoms at baseline (Evidence Figures 15–18 inAppendix D ^*). These results along with the primary study findings from LH-1 indicate that in smokers with normal airflow to moderate airflow obstruction ipratropium was not effective in altering spirometric decline or the development of respiratory symptoms or respiratory hospitalization.

The Lung Health Study-2 recruited 1,116 participants who had previously participated in or had been screened for the LH-1 study and randomized them to the inhaled corticosteroid triamcinolone or placebo. Almost 90 percent of subjects were current smokers, but fewer than 20 percent of subjects had a previous physician diagnosis of emphysema or chronic bronchitis. Approximately one-third had daily cough and phlegm and 40 percent had some level of dyspnea. The mean FEV₁ after bronchodilator was 2.3L (68 percent predicted: GOLD Stage 2). Thus, these individuals are representative of subjects who might be detected by case finding with spirometry. After a mean duration of followup of 40 months the rate of decline in the FEV₁ after corticosteroid use was similar in the 559 participants in the triamcinolone group and the 557 participants in the placebo group (mean approximately = 44mL/year). Members of the triamcinolone group had fewer overall respiratory symptoms during the course of the study (21.1 per 100 person-years vs. 28.2 per 100 person-years, p = 0.005) and had fewer visits to a physician because of a respiratory illness (1.2 per 100 person-years vs. 2.1 per 100 person-years). There was no significant association between triamcinolone use and the development of specific respiratory symptoms of chronic cough, production of phlegm, or wheezing.