Chapter  39:  Management of Newly Diagnosed Patients with Epilepsy: A Systematic Review of the Literature: Evidence Report/Technology Assessment Number 39

Exclusion Criteria

Exclusion criteria comprised the following:

• Reviews and meta-analyses, letters, and case reports.

• Non-English language studies.

• Animal studies.

• Pharmacodynamic/pharmacokinetic studies.

Studies were rejected for the following reasons: (a) not newly diagnosed patients; (b) outcomes not extractable for newly diagnosed patients; (c) reviews; (d) case studies; (e) other languages; (f) studies of patients with secondary seizures, i.e., seizures not idiopathic or cryptogenic. These included studies selecting for patients with seizures secondary to tumors, trauma, infection, stroke, chronic medical or neurologic disease, psychogenic or pseudoseizures, and febrile convulsions. It should be noted that in some studies, information about secondary causes of seizures was not supplied or was incomplete, so we could not be sure that all included studies consisted completely of patients with primary epilepsy, but the intent was to develop a database as consistent as possible in this regard, given the vagaries of reporting in the literature.

All studies passing Level I screening were retrieved for a second screening (Level II) applying the following inclusion criteria.

Inclusion Criteria

Level II inclusion criteria were as follows:

• Study designs: observational (prospective, retrospective, and cross sectional), or interventional (randomized controlled trials [RCTs], nonrandomized controlled trials [nRCTs], and uncontrolled case series [UCS]).

• At least 10 patients, adults or children, with a first seizure, a first presentation with epilepsy or seizures, or a first diagnosis of epilepsy.

• Studies addressing any of the following diagnostic interventions: history; neuropsychological assessment; physical examination; imaging with CT, MRI, or PET scans; EEG (standard, video, invasive, ambulatory); lumbar puncture; or blood test.

• Studies addressing any of the following monitoring interventions: EEG (standard, video, invasive, or ambulatory), laboratory tests (hematology/biochemistry) and drug assays (standard or sophisticated), and clinical/pharmacologic expertise.

• Studies addressing any treatment intervention: clinical/pharmacologic expertise, counseling/psychosocial, surgery (resective or vagus), physical therapy/occupational therapy (PT/OT), speech language, investigational study, and education.

• Studies reported in English only.

The most difficult aspect of screening papers at both levels was establishing the presence of the population of interest, i.e., newly diagnosed patients. This term was defined to include patients presenting with a first seizure, patients presenting for a first diagnosis, and patients with a new diagnosis just prior to study entry. It did not include studies in patients with a long-time, established diagnosis of epilepsy. Studies of patients with a misdiagnosis, or uncertain diagnosis, were eligible as long as the patients were diagnosed with recent onset of seizures. Making this distinction in retrospective observational studies was the most difficult. Such studies could be included as long as information about patients at the time they were newly diagnosed (i.e., time zero) was included in the observation window and extractable for that group.

The second most difficult challenge in screening studies for inclusion was establishing whether data were extractable for the population of interest. In many cases where a series of patients attending a clinic, for instance, were studied, those with a new diagnosis were not distinguished from those with established epilepsy in the reporting of interventions and/or outcomes. This difficulty was most pronounced in study designs that were observational as opposed to interventional.

Since study selection is a critical component of any systematic review, the Level II screen required the agreement of two reviewers for each rejected paper. In cases of disagreement, a third reviewer adjudicated. The list of eligible studies was also subsequently shared with the project TEP for review and comment prior to locking the database.

Diagnosis

There were five observational studies, all in Europe, reporting incidence of new diagnoses (Braathen and Theorell, 1990; Forsgren, 1990; Forsgren, Bucht, Eriksson, et al. 1996; Jallon, Goumaz, Haenggeli, et al., 1997, Luhdorf, Jensen, and Plesner, 1986). The incidence results ranged from 0.03 percent to 0.08 percent in patients with all seizure types and all ages represented. These figures are similar to the incidence estimates for established market economies noted above (see Background section in Chapter 1).

Evidence Tables 4, 4a, 5, and 5a summarize selected results from studies with diagnostic information. Several studies of diagnostic interventions described what could be loosely construed as a "gold standard" for the diagnosis of epilepsy. These gold standards tended to include both a clinical component and an EEG component. The clinical requirements for diagnosis were highly variable, however, and included such signs and symptoms as tonic/clonic movements, with or without post-ictal confusion, tongue biting, sphincter disturbance, aura, and loss of consciousness. Some studies required the events to be unprovoked; others did not. Some studies required the events be witnessed; others did not. The seizure type was usually diagnosed by clinical features and the epilepsy syndrome, by seizure type and EEG findings. Only a minority referred to established classification schemas, e.g., ILAE or WHO.

There were 12 prospective interventional trials reporting results of a diagnostic intervention: 6 were in patients at the time of first diagnosis, 4 at the time of first presentation, and 2 at the time of first seizure. There were 26 observational studies, which included "time zero" for diagnosis in the observation window. Most of the studies reporting the use of diagnostic interventions did not provide the numbers of patients with a new or changed diagnosis as a result of the intervention. Rather, these studies typically reported the use of the intervention as discretionary by the study investigators or as occurring in some patients at the time of study entry, but no further details are provided.

The role of history and physical examination was touched on in several studies. Berg, Shinnar, Levy, et al., (1999a, 1999b) reported that 609 of 613 children were assigned a syndromic diagnosis on the basis of clinical features. Arts, Geerts, Brouwer, et al., (1999) reporting on 466 children suggested the history alone yielded a 29 percent sensitivity and 89 percent specificity. The study by Hoefnagels, Padblerg, Overweg, et al., (1991) noted that it is impossible to find a gold standard for the diagnosis of epilepsy and therefore developed its own to distinguish epilepsy from syncope. Sensitivity and specificity of several components of a history were computed, e.g., particular symptoms before, during, and after the paroxysmal event. Those before the event tended to the highest sensitivity (88 percent to 98 percent), and those during the event, the highest specificity (64 percent to 94 percent). Camfield, Camfield, Dooley, et al. (1985a) reported that in a retrospective analysis of 168 children seen after their first seizure, an abnormal neurologic examination (in 30 patients) was predictive of recurrence, as was seizure type (partial seizure associated with increased risk). Neither the sleep-wake status at the first seizure nor a history of febrile seizures predicted recurrence. In three additional retrospective studies, the utility of various interventions in diagnosis and/or prediction of recurrence was reported. Ambrosetto, Giovanardi, and Tassinari (1987) reported on history (and EEG findings) in 72 patients and concluded that only generalized seizures as the sole ictal phenomenon, and a long interval between the first and second seizures, were predictive of seizure frequency subsequently.

King, Newton, Jackson, et al. (1998) reported that early EEGs, within 24 hours of seizure, yielded a diagnosis in 77 percent (232/300) patients. Martinovic and Jovic (1997) related the risk of recurrence in children after a first seizure to the finding of epileptiform patterns on standard EEG. Carpay, de Weerd, Schimsheimer, et al. (1997) reported the diagnostic yield of standard EEGs and sleep-deprived EEGs after a normal standard EEG. Of 552 standard EEGs in children with first seizures, 243 were normal and 309 were epileptiform. Sleep-deprived EEGs in the normal EEG patients added 11 percent more diagnoses to the 56 percent of patients with diagnosis after abnormal standard EEGs. Hughes and Gruener (1985) reported the use of standard EEGs in confirming seizures in patients with a new diagnosis of epilepsy. The history was noncontributory, but the standard EEG was abnormal in 300 of 358 patients, supporting its use in first diagnosis. Sperling and Engel (1985) studied the utility of additional electrodes in the EEG diagnosis of 98 patients who had normal standard EEGs but who were suspected of having temporal lobe epilepsy. Ear, anterior temporal, or nasopharyngeal electrodes did not increase the diagnostic yield. Chu (1991) studied the utility of sphenoidal recording during both standard and ambulatory EEGs. The increased yield was marginal.

There were several studies of video EEG in diagnosis. Oldani, Zucconi, Smirne, et al. (1998) reported that daytime video EEG was not diagnostic in any of 23 patients suspected of nocturnal frontal lobe epilepsy. After sleep deprivation, however, video EEG was diagnostic in 12 of 20, compared with nocturnal video polysomnography in 20 of 23 patients. Shihabuddin, Abou-Khalil, and Fakhoury (1999) reported on 123 patients with attacks of unknown nature, in whom standard EEG, CT, and MRI were nondiagnostic. Ambulatory EEG with video monitoring yielded a diagnosis in 81. The number of patients judged to have epilepsy was decreased as a result of this intervention, from 31 to 20 patients. Bye, Lamont, and Healy (1990) reported that 66 of 82 children had paroxysmal events documented on video EEG, 23 of which were ictal, leading to diagnosis. Foley, Legido, Miles, et al. (1995) reported that outpatient video EEG was diagnostic in 32 of 36 children with suspected epilepsy. Cull (1985) assessed the utility of 24-hour ambulatory EEG versus standard EEG. The latter was abnormal in 16 of 62 patients and the former, abnormal in 21 patients. Ambulatory EEG was not superior to standard EEG if attacks occurred at frequencies of fewer than one per week. Seven of 46 patients with normal standard EEGs were abnormal on ambulatory recordings. Conversely, 2 of 16 abnormal standard EEGs were normal during ambulatory EEGs.

Four retrospective studies also assessed video EEGs. Duchowny, Resnick, Deray, et al. (1988) studied 60 young children; a first diagnosis was made possible in 9 or the 60 and a change in diagnosis was made in 24. Mohan, Markand, and Salanova (1996) reviewed results in 444 patients with video EEG. This intervention led to a diagnosis of psychogenic seizures in 141 patients and an epilepsy diagnosis in 180 patients. In 123 patients, video EEG was not diagnostic. The authors noted that if the frequency of episodes is greater than one per week, the video EEG is more likely to yield a diagnosis. Chen, Mitchell, Horton, et al. (1995) retrospectively studied video EEG in diagnosis and treatment decisions in 138 children. A diagnosis was made possible in 96 (70 percent). Roberts and Fitch (1985) reported the diagnostic utility of video and ambulatory EEGs in 102 patients. In 57 patients, the video EEG led to a diagnosis, which was epilepsy in 19. In a further 20, the ambulatory EEG led to a diagnosis, which was epilepsy in 3.

Blood tests in diagnosis were evaluated in very few studies. Anzola (1993) found normal serum prolactin levels in patients with syncope, but elevated levels in patients with epilepsy. Positive predictive value (PPV) was 89 percent and negative predictive value (NPV), 61 percent. Neufeld, Treves, Chistik, et al. (1997) studied the use of sequential creatine kinase (CK) measurements to differentiate syncope from generalized tonic clonic seizures in adults at the time of first seizure. Admission CK was increased in 25 percent of seizure patients versus 6 percent of syncope patients. At 24 hours, CK was increased in 56 percent of seizure patients and 12 percent of syncope patients. An absolute increase in CK of 15 milliunits/ml gave a diagnostic sensitivity of 69 percent and specificity of 94 percent.

The evidence is clear that diagnostic interventions will vary in their utility according to the patient population studied. For example, there are several studies restricted to elderly patients with new epilepsy, most of which suggest an increased likelihood of abnormal EEGs and CT scans and the relative lack of utility of the history alone in diagnosis. Holt-Seitz, Wirrell, and Sundaram (1999) assessed retrospectively the diagnostic yield of various interventions in 84 elderly patients with new-onset seizures. Standard EEG was abnormal in 73 percent and CT, abnormal in 68 percent. MRI was performed in only 11 patients and was abnormal in 7. Dam, Fuglsang-Frederiksen, Svarre-Olsen, et al. (1985) studied different etiologies in 84 patients with new late-onset epilepsy. They reported that EEG was abnormal in 56 percent and CT was abnormal in 60 percent. However, only 7 of the 84 patients had unilateral signs on physical examination, but of these, 6 had focal findings on CT scan. Kilpatrick, Tress, O'Donnell, et al. (1991) reported on 50 patients with late-onset epilepsy with normal or nondiagnostic CT scans. MRIs within 3 months revealed diagnostic abnormalities in 10 of these patients. Luhdorf, Jensen, and Plesner (1986) studied prognostic factors in the elderly for 12 months after a first seizure. History did not correlate either to the type of seizure or to the presumed cause. A paroxysmal EEG, however, did. Lee (1985) studied ictal confusion as a first presentation of epilepsy in later life in 11 patients. Again, the history was not useful in diagnosis, but a standard EEG did result in a diagnosis in 5 of 11 patients.

One retrospective study assessed features of typical absence seizures. Panayiotopoulos, Chroni, Daskalopoulos (1992) noted that in all of 23 adults studied, events were documented in standard ± video EEGs (performed in 12). All 23 patients had normal neurologic and mental status.

Diagnosis of juvenile myoclonic epilepsy (JME) was the focus in four studies. Atakli, Sozuer, Atay, et al. (1998) reported that in a retrospective series of 76 patients with JME, a complete history led to a correct diagnosis in all 76, although 40 had presented with a misdiagnosis. In all patients, the neurologic examination was normal; and in 69 of 76 standard EEGs, background activity was normal. CT (in 32) and MRI (in 3) did not contribute to the diagnosis. Grunewald, Chroni, and Panayiotopoulos (1992) reported an average delay in the diagnosis of JME of 14.5 years; yet in all 15 children studied retrospectively, the diagnosis was made by history and standard EEG alone. In 14 of 15 patients, a misdiagnosis had been made originally. Panayiotopoulos, Tahan, and Obeid (1991) reported that the history and EEG alone were diagnostic in a retrospective series of 70 patients with JME. In 22 of these patients, a misdiagnosis had been originally made. Sharpe and Buchanan (1995) also reported a retrospective series of JME patients in whom diagnosis had been delayed for an average of 9.7 years. The history and CT were normal in all, and the value of the standard EEG in diagnosis had been reduced by the use of AEDs.

The new diagnosis of temporal lobe epilepsy was reported by Harvey, Berkovic, Wrennall, et al. (1997) in a retrospective series of 63 children. The neurologic examination was abnormal in 9, and the neuropsychiatric examination was abnormal in 24. Fewer than one-fourth of CTs (performed in 48) were abnormal, but nearly one-half of MRIs (performed in 58) were abnormal. Video EEG had a high yield in selected patients, with 18 of 20 revealing ictal activity.

Henneman, DeRoos, and Lewis (1994) reported an emergency room (ER) study of adults presenting with new-onset seizures. Of 333 seizure patients assessed in the ER, the clinical examination was clearly abnormal in 75 and routine blood tests were abnormal in 48. In 227 patients undergoing a lumbar puncture (LP), it was abnormal in <10 percent. Of 325 patients having a CT scan, a clinically significant abnormality was seen in about 40 percent. The standard ER workup consisted of a clinical examination, routine blood tests, and LP and CT in selected patients. The authors concluded that only 136 of the 254 patients admitted to hospital actually needed admission. Of these 136, the vast majority (129) were identified by a standard ER workup.

ILAE criteria for diagnosis were studied in two retrospective series. In Jallon, Goumaz, Haenggeli, et al. (1997), despite ILAE-determined epilepsy diagnoses in 273 patients, the standard EEG was normal in 67. In 190 of these patients, a CT was performed and was normal in 61. In 56 patients, a MRI was performed and was normal in 17. Reutens, and Berkovic (1995) assessed agreement of ILAE criteria with clinical features in 101 patients. In particular, not all patients fit ILAE syndromic diagnoses, and they reported phenotypic overlap between syndromes in adolescent generalized epilepsy.

In summary, the literature supports the diagnostic role of a complete history, especially in diagnosing JME, to elucidate an adequate description of the seizures to permit categorizing by seizure type, since a history suggestive of a focal seizure predicts recurrence. A clinical examination that includes a careful neurologic examination is essential, since an abnormal examination after a first seizure also predicts recurrence. The chance of recurrence is critical to decisions about initiating AEDs. Data are insufficient to conclude whether any blood tests at the time of a first seizure are useful except to rule out the presence of any secondary causes of seizures. As for standard EEGs, the literature suggests that it is an absolute requirement and should be performed as early as possible after the first seizure. The evidence reviewed does not permit conclusions about sleep EEGs or sleep-deprived EEGs. The studies in this review reported that only about 50 to 60 percent of standard EEGs are abnormal, but if the standard EEGs are performed within 24 hours of the first seizure, it is suggested that the diagnostic yield increases. Another reason to perform the standard EEG as soon as possible is the confounding influence of AEDs if started prior to diagnosis, which occurred in this literature in from 2 to 20 percent of patients. The role of MRI in first diagnosis is best established in late-onset seizure patients with nondiagnostic CTs, although other roles for MRI, such as in localization of seizure foci in operative candidates, have not been addressed in this review. The literature suggests that ambulatory and video EEGs are useful in a first diagnosis if standard EEG, CT, and MRI are nondiagnostic. Video EEGs are also useful in diagnosis in very young children, in patients with poorly characterized seizure types, and in those with suspected psychogenic seizures, especially if episodes are frequent.

Monitoring

Evidences Tables 6, 6a, 7, and 7a display selected results of studies with monitoring information. None of these studies had as a primary objective the testing of monitoring interventions necessary for optimal patient care. There were 24 prospective interventional studies (8 RCTs, 7 nRCTs, 5 UCS, and 4 XS) that had a monitoring component, but in nearly all, this was a monitoring intervention dictated by a research study protocol and not optimal patient care. None of these studies enrolled patients after a first seizure but rather, on first presentation or first diagnosis.

The interventions covered included periodic drug levels and hematology and chemistry panels for the most part. Other interventions used in monitoring new patients after initiation of treatment with AEDs included measures of cognitive function and behavior (Camfield, Camfield, Dooley, et al., 1985; Frost, Hrachovy, Glaze, et al., 1995; Isojarvi, Parkarinen, and Myllyla, 1993; Kalviainen, Aikia, Helkala, et al., 1992; Mandelbaum and Burack, 1997; O'Dougherty, Wright, Cox, et al., 1987; Pulliainen and Jokelainen, 1994; Williams, Bates, Griebel, et al., 1998); EEG changes (Camfield, Camfield, Smith et al., 1985b; Drake, Weate, Newell, et al., 1994; Frost, Hrachovy, Glaze, et al., 1995; Konishi, Naganuma, Hongou, et al., 1995; Marciani, Maschio, Spanedda, et al., 1995; Sannita, Gervasio, and Zagnoni, 1989), median nerve evoked potentials (Carenini, Bottacchi, Camerlingo, et al., 1988), peripheral benzodiazepine receptor density (Ferrarese, Tortorella, Bogliun, et al., 1997), and QoL (Steiner, Dellaportas, Findley, et al., 1999).

Only the study by Camfield, Camfield, Smith et al., (1985b) could be considered as showing monitoring interventions that impact optimal patient care. In this study of standard EEG and neuropsychological changes with long-term use of AEDs in 82 children, behaviorial side effects or allergies led to treatment change in 11 patients and hematologic abnormalities led to change in 2. Another study (Sannita, Gervasio, Zagnoni, 1989) of 10 children, showed that EEG results did not relate to AED levels or to behavioral side effects, suggesting no role for EEG in routine monitoring of patients. The lack of relationship of EEG monitoring results to clinical seizure frequency in treated patients was also suggested by the results of the Drake study of spectral analysis of EEGs in 30 adults with new epilepsy after AED monotherapy. Although 25 patients demonstrated nocturnal sleep EEG abnormalities, none had seizures on treatment. There are several other studies in this review wherein standard EEGs were applied, all at highly variable intervals, ranging from every month (for the first 3 months) in one, to every 3, or 6, or 8 months in others, to once before AED withdrawal in one. None of these studies reported any effect of such monitoring on seizure frequency, dropout rates from AED side effects, or noncompliance. Furthermore, there were no studies reporting any change in quality of life or change in diagnosis or treatment as a result of standard EEGs performed as part of patient monitoring.

Monitoring cognitive effects of AEDs were the subject of several studies, with inconsistent results. Williams, Bates, Griebel, et al. (1998) showed no difference in cognitive or behavioral effects before and after treatment in 35 of 37 children initiated on AED monotherapy. On the other hand, Gunduz, Demirbilek, and Korkmaz (1999) reported that, in 20 newly treated children after 1 year, neurological dysfunction was detected on a battery of tests. In addition, Frost, Hrachovy, Glaze (1995) reported that language and Wechsler Intelligence Scale for Children (WISC)-derived measures declined at 1 year of AED treatment in 16 children. These studies could be interpreted as suggesting that monitoring of cognitive and neuropsychological measures may detect changes with long-term AEDs, especially in children. O'Dougherty, Wright, Cox, et al. (1987), however, reported that neuropsychological test results in 11 children did not necessarily correlate with AED drug levels.

Standard drug levels were measured "regularly" or "periodically" in followup in numerous studies, but the test frequency either was typically not specified or variable. There were no studies directly linking monitoring of standard drug levels to seizure frequency, dropout rates from AED side effects, or noncompliance. There were no studies reporting of any change in quality of life or change in diagnosis as a result of monitoring using standard drug levels. One interventional study, however, reported a change in treatment as a result of drug levels in 19 percent of monitored patients (Callaghan, Kenny, O'Neill, et al., 1985). Luhdorf, Jensen, and Plesner (1986) reported that subtherapeutic drug levels in a population of elderly patients with new epilepsy predicted recurrence.

As for periodic chemistries and/or hematology measures, one interventional study (Stefan, Plouin, Fichsel, et al., 1988) reported an increase in dropouts with adverse effects in 2 of 10 patients as a result of blood testing. The paper by Steiner, Dellaportas, Findley, et al. (1999) further supported the notion that understanding the distinctive side effects of different AEDs is important to optimal patient monitoring. In 181 patients with newly diagnosed epilepsy enrolled on a RCT comparing efficacy and tolerability of phenytoin and lamotrigine, abnormalities of liver enzymes occurred more often in the phenytoin patients, whereas abnormal elevations in serum creatinine were seen more often in the lamotrigine patients.

Chen, Mitchell, Horton, et al. (1995) reported on the use of video EEG in diagnosis and subsequent treatment decisions. It was performed once at entry and once in followup. One-fourth of all children had a diagnosis changed as a result of video EEG, and 88 percent of a subgroup referred especially for seizure classification were successfully diagnosed. The procedure was also useful to assist in preoperative assessment of surgical candidates. These results suggest a place for video EEG in new patients not readily classified or those who may be judged to be candidates for surgery.

In summary, the evidence, although scant, suggests there is no role for standard EEG in routine monitoring of patients after a new diagnosis of epilepsy. Video EEG has a role subsequent to a new diagnosis if the diagnosis is or becomes uncertain or if surgery is considered. Neuropsychological monitoring for behavioral and cognitive changes may have a role in identifying subtle AED side effects, especially in children. Blood testing (hematology and chemistries) is important to monitor for infrequent but potentially serious side effects of AEDs. Periodic drug levels may be important in assessing AED compliance, although this remains controversial. Since there is evidence that different AEDs impact QoL differently, there may also be a role for QoL monitoring.

Treatment

Evidence Tables 8, 8a, 9, and 9a summarize selected results of studies with treatment information. Monotherapy AED as a treatment intervention was reported in over one-half the studies in the review for durations ranging from 2 weeks to 6 years. Monotherapy was a treatment intervention in 5 studies of patients at time of first presentation, 44 studies at first diagnosis, and 6 studies at first seizure. Monotherapy was associated with a report of remission of seizures in 56 to 75 percent of these patients, with lower rates in observational studies of real-world patients, mostly with a mixture of seizure types or epilepsy diagnoses. Rates of remission (seizure freedom) in general in newly diagnosed patients begun on AED monotherapy decrease over time, with 60 to 90 percent of patients in remission at 1 year, 30 to 60 percent at 2 years, and approximately 30 percent at 3 years. Patients are not "cured," but these studies suggest that seizure freedom for prolonged periods is possible with AED monotherapy. In most of these studies, there was no apparent difference in efficacy of AEDs. Only the side effect profiles of AEDs differed.

Polytherapy AED as a treatment intervention was unusual in newly diagnosed patients. In the eight studies that included any newly diagnosed patients with AED polytherapy, the proportions of such patients were low, ranging from 14 percent and 20 percent of all. One very small study (Penry, Dean, and Riela, 1989) reported a remission rate in six children on polytherapy (50 percent at 1 year). One interventional study of 622 adults at a Veterans Administration hospital (Smith, Mattson, Cramer, et al., 1987) reported a remission rate of 45 percent in newly diagnosed patients (n) receiving AED polytherapy (n=122) after first failing monotherapy. No studies reported dropout rates from side effects or from noncompliance or alterations in QoL associated with AED polytherapy.

In 15 studies, patients received no treatment: 8 after a first seizure, 2 at the time of a new diagnosis of epilepsy, 1 at first presentation, and 4 observational studies with "time zero" of diagnosis in the observation window. The durations within which patients remained with no treatment ranged from 2 weeks to 5.8 years. Remission rates with no treatment were highly variable; and given the many differences in patients across these studies, no conclusions can be drawn. These studies do underscore the fact, however, that some patients will do well without treatment. The difficulty is in identifying them.

Last, several studies were categorized as clinical/pharmacologic expertise studies. Clinical/pharmacologic expertise was defined as that expertise necessary for: (a) selecting optimal AEDs based on epilepsy diagnosis and patient characteristics, (b) adjusting drugs and dosages to reduce seizures, (c) monitoring and limiting adverse drug reactions and interactions, (d) monitoring patients' tolerance and compliance with particular drug regimens, (e) recognizing changes in seizure characteristics, and (f) ordering and interpreting appropriate laboratory tests based on knowledge of specific adverse events associated with different drugs. This was associated with epilepsy centers in 27 studies. In Atakli, Sozuer, Atay, et al., (1998), 35 of 40 misdiagnosed patients had, after correction of the diagnosis in an epilepsy center, their medications changed and became seizure free. In Grunewald, Chroni, and Panayiotopoulos (1992), inappropriate AEDs had been prescribed in 7 of 15 children prior to presentation to an epilepsy center. Furthermore, in Oller-Daurella and Oller (1991), a hospital-based study of adults, a shorter first seizure to treatment interval was associated with better remission durations, thus demonstrating that the timing of a correct diagnosis and treatment is also important to optimal patient management. There is considerable evidence for certain conditions, e.g., JME, that suggests that diagnosis and AED choice requires expertise. Seizure remission rates in studies categorized as clinical/pharmacologic expertise studies were high: in three observational studies, 79 percent of patients achieved remission, as did 84 percent of patients in the three interventional trials in this category.

There were no interventional or observational studies of any of the following treatment interventions: counseling, speech therapy, occupational therapy, physical therapy, education, or surgery in newly diagnosed patients.

In summary, the evidence supports the notion that optimal patient care would include access to health care professionals who are experts in the choice and timing of initial AED monotherapy, but it does not describe the training or experience of such caregivers. The shorter the interval from first seizure to diagnosis and initiation of treatment, the better. Although most studies reporting remission rates reported results ranging from 35 to 60 percent, the highest remission rates (79 to 84 percent) were reported in studies categorized with clinical/pharmacologic expertise. There is some evidence suggesting that clinical/pharmacologic expertise minimizes misdiagnoses, with resulting mistakes in choosing AEDs.

Optimal outcomes (Evidence Tables 10 and 10a) were explicitly or implicitly defined in terms of seizure frequency in 15 observational and 26 interventional studies. Interestingly, an optimal outcome was always defined as no seizures, but the requisite time period to maintain that goal varied widely (<3 months up to 2 years), suggesting lack of consensus as to what is an optimal outcome for newly diagnosed patients. Far less often (Aikia, Kalviainen, Sivenius, et al., 1992; Camfield, Camfield, Dooley, et al., 1985; Kalviainen, Aikia, Saukkonen, et al., 1995), the occurrence of a stipulated number of side effects was also reported as part of a definition of an optimal outcome. In all three studies that included side effects in the definition of "optimal," the optimal number of side effects was zero. Compliance, QoL, and costs were never noted as components of a definition of optimal patient outcomes.

Background

In February 1999, the Agency for Health Care Policy and Research (AHCPR) awarded a task order contract to MetaWorks to develop an evidence report on "Criteria for Referral of Patients with Epilepsy." The topic was nominated by the Centers for Disease Control and Prevention (CDC) for the purpose of developing a framework for organizing and optimizing the care of persons with epilepsy. The CDC's request for an evidence report on this topic followed from their September 1997 conference entitled "Living Well With Epilepsy." The participants divided into work groups which generated several recommendations. Among these was the development of a testable framework for effective care of persons with epilepsy. The CDC determined that this framework was to be the result of a preliminary literature search examining models of care in epilepsy and other chronic conditions and a stakeholders meeting with representatives from various constituencies involved in improving epilepsy care. A systematic literature review based on the framework would then be undertaken to draft an evidence report. This evidence report could then be used by providers, professional societies and payors as a basis for decision-making about effective care for patients with epilepsy.

The development of this evidence report is being carried out in two phases. The first phase was preparatory to a meeting of representatives from the AHCPR, the CDC, and stakeholders from various constituencies involved in improving the care of patients with epilepsy. The purpose of this meeting was to define questions to guide a more comprehensive literature review to generate an evidence report on effective referral criteria for persons with epilepsy. The subsequent literature review and synthesis will constitute the second phase of the project.

This report summarizes phase I tasks and findings: the preliminary literature review and the stakeholders' meeting as well as the recommendation regarding direction for future work.

Literature Review

A preliminary literature review was conducted for models of care in epilepsy and other chronic diseases. We were particularly interested in care management issues that are associated with disease conditions of low prevalence and high chronicity. We were interested in population- and policy-level information as opposed to clinical patient-level information. The sources listed in Appendix A were searched for information relative to referral criteria, guidelines for referring categories of patients needing more specialized care, and studies assessing the impact of specialty care on patient outcomes.

The information sources searched and the yields for each are summarized in Appendix A. The results of the preliminary literature search are described below.

Summary of Stakeholders' Meeting

A meeting of stakeholders representing various constituencies in the field of epilepsy was held on March 25, 1999, in Philadelphia, Pennsylvania. The purpose of the meeting was to develop a testable framework that could be supported with evidence from the literature for organizing the care of persons with epilepsy. Present at the meeting were representatives from AHCPR, CDC, MetaWorks, and stakeholders from the following constituencies: epileptology, neurology, primary care, nursing, managed care, healthcare purchasing, as well as patient representatives. A list of attendees is attached in Appendix B. The agenda for the meeting (Appendix C) comprised a presentation of the project background, the results of the literature search (described above), and an overview of the Lovelace Health System's approach to provider education and outcomes research in epilepsy. After the presentations, attendees were divided into constituency groups to generate a summary of their experience with referral patterns for epilepsy in the real world, and how referrals might occur in an ideal world. Enclosure B, "Real" and "Ideal" Experiences With Referrals for Epilepsy Management (Appendix D) was employed as a starting point for organizing these discussions. A spokesperson from each constituency then presented a summary of their discussion to the group. Participants discussed several different referral strategies that emerged as a result of the literature review and discussion of stakeholders' experience and merged these strategies into a framework to guide future study.

A summary of key points emerging from the meeting is described below. Detailed meeting minutes are contained in Appendix E.

Direction For Further Study

A framework grid was developed outlining specific resources required at each point of care (diagnosis, treatment and monitoring) for each of the five highest-priority patient populations highlighted by the stakeholders (see Appendix F). This framework was created to aid in the retrieval, organization and synthesis of best available evidence for the management and referral of patients with epilepsy. The five priority populations are listed on the Y-axis. The X-axis lists healthcare services and all other resources that might be considered necessary for optimum care during the diagnosis, treatment and monitoring of epilepsy patients. Each white cell represents a particular area of inquiry that may be further researchable using the published literature. The darkened cells represent healthcare services or resources deemed by MetaWorks and University of Pennsylvania co-investigators as not applicable for the particular population on the Y-axis.

Using the matrix framework, researchers can assess the quantity and quality of available evidence examining the clinical outcomes resulting from the application of a particular service to a particular patient population at certain time points of care. Providers, professional societies and payors can use the evidence to guide decisions about referrals and resource utilization. The purpose of this exercise is to improve outcomes for these populations through the application of best evidence to clinical practice and management of epilepsy.

This targeted literature review and synthesis of evidence will constitute phase II of this project.

References

1.

Diagnosis and Management of Asthma in Adults. Institute for Clinical Systems Integration (ICSI). May 1998.

2.

Specialty Referral Guidelines for People With Diabetes. Diabetes Treatment Centers of America (DTCA). 1998. [Free Full text in PMC]

3.

Roman S. H., and Harris M.I. Management of Diabetes Mellitus from a Public Health Perspective. Current Therapies for Diabetes. 1997; 26: 443– 474.

4.

Position Statement: Implications of the Diabetes Control and Complications Trial. Diabetes. 1993; 42: –.

5.

Yang Y., Shah A. K., Watson M., and Mankad, V.N. Comparison of Costs to the Health Sector of Comprehensive and Episodic Health Care for Sickle Cell Disease Patients. Public Health Reports. 1995; 110: 80– 85. [PubMed]

6.

Pritchard J., Stiller C. A., and Lennox E.L. Overtreatment of Children With Wilms' Tumor Outside Pediatric Oncology Centers British Medical Journal 1989. 299:(6703): 835– 836.

7.

Piro L., and Doctor J. Managed Oncology Care: The Disease Management Model. Cancer (Supplement). 1998; 82: 2068– 2075.

8.

Bernard D. B., Townsend R.R., and Sylvestri M.F. Health and Disease Management: What is it and Where is it Going? What is the Role of Health and Disease Management in Hypertension? American Journal of Hypertension. 1998; 11: 103S– 108S.

9.

De Boer H. M, Aldenkamp A. P., and Overweg J. Epilepsy Care in the Netherlands. Acta Neurologica Scandinavia (Supplement). 1992; 140: 101– 105.

10.

Osservatorio Regionale per l'Epilessia (OREp), Lombardy. Epilepsia. 1997; 38: 1338– 43. [PubMed]

11.

Brown S., Betts T., Chadwick D., Hall B., Shorvon S., and Wallace S. An Epilepsy Needs Document. Seizure. 1993; 2: 91– 103. [PubMed]

12.

Taylor M.P., Readman S., Hague B., Boulter V., Hughes L., and Howell S. A District Epilepsy Service, With Community-Based Specialist Liaison Nurses and Guidelines for Shared Care. Seizure. 1994; 3: 121– 127. [PubMed]

13.

Scottish Intercollegiate Guidelines Network (SIGN). 1997

14.

Epilepsia (Supplement). 1990; 31: S1– S2.

Attachment A: Information Sources, Search Terms, Yield, and Papers Retrieved for Review

Attachment B: Stakeholders

List of Attendees

CDC/AHCPR Stakeholder Committee Meeting

Criteria for Referral of Patients with Epilepsy

March 25, 1999 -- Philadelphia, PA

• I. Epilepsy Specialists

• Frank Gilliam, M.D. (UNABLE TO ATTEND)

• University of Alabama Epilepsy Center

• 1719 6th Avenue South, Suite 312

• Birmingham, AL 35294

• Greg Barkley, M.D.

• Comprehensive Epilepsy Program

• Henry Ford Hospital

• 2729 West Grand Blvd.

• Detroit, MI 48202

• II. Primary Care

• Patricia Dean, MSN, MS

• Program Manager

• Comprehensive Epilepsy Center

• Miami Children's Hospital

• 3100 S.W. 62nd Ave.

• Miami, FL 33155

• Louis Trost, M.D.

• Lovelace Family Practice - Westside

• 2929 Coors, N.W.

• Albuquerque, N.M. 87120

• III. Health Services

• Margaret Gunter, PhD

• Lovelace Clinic Foundation

• Education Bldg., 2nd floor

• 5400 Gibson Blvd., S.E.

• Albuquerque, N.M. 87108

• Suzanne Mercure

• Director of Consumer Programs

• Institute for Health Policy Solutions

• Principal, Barrington and Chappell

• 6023 Madison Overlook Court

• Falls Church, VA 22041

• IV. Neurology

• David G. Cook, M.D.

• Pennsylvania Hospital

• 8th and Spruce Street

• Philadelphia, PA 19107

• David A. Stumpf, M.D.

• Northwestern University Medical School

• Department of Neurology

• 645 N. Michigan Ave.

• #1058

• Chicago, IL 60611-2814

• V. Consumers/Advocates

• Linda Warner

• 9156 Jewel Lane North

• Forest Lake, MN 55105

• Joyce Bender

• Chief Executive Officer

• Bender Consulting Services, Inc

• Penn Center West III, Suite 223

• Pittsburgh, PA 15276

• AHRQ

• Harry Handelsman, D.O.

• Task Order Officer

• Agency for Health Care Policy and Research

• 6010 Executive Blvd., Suite 310

• Rockville, MD 20852

• CDC

• Patricia Price, D.O., FACPM

• Epilepsy Program Coordinator

• Centers for Disease Control and Prevention

• Division of Adult and Community Health

• 4770 Buford Hwy, NEMailstop K-45

• Atlanta, GA 30341

• Suzanne M. Smith, M.D., M.P.H

• Chief, Health Care and Aging Studies Branch

• Centers for Disease Control and Prevention

• Mailstop K-45

• Atlanta, GA 30341

• CDC Ex Officio

• David Thurman, M.D.

• Centers for Disease Control and Prevention

• National Center for Injury Prevention and Control

• 4770 Buford Highway, N.E.

• Mailstop F-41

• Atlanta, Ga. 30341

• (770) 488-4715+

• Leonard Davis Institute/University of Pennsylvania

• Jacqueline French, M.D.

• (Project Co-Investigator)

• University of Pennsylvania Medical Center

• Department of Neurology, 3 West Gates

• 3400 Spruce Street

• Philadelphia, PA 19104-4283

• Joanne Levy, MBA, MCP

• (Project Manager)

• Associate Director

• Leonard Davis Institute of Health Economics

• University of Pennsylvania

• 3541 Locust Walk

• Philadelphia, PA 19104-6218

• MetaWorks

• Sameer Chopra, MA

• (Project Manager)

• MetaWorks Inc.

• 470 Atlantic Ave, 10th floor

• Boston, MA 02210

• Rhonda Estok, RN, BSN, CNOR

• (Project Administrator/Manager)

• Clinical Information Specialist

• MetaWorks Inc.

• 470 Atlantic Ave, 10th floor

• Boston, MA 02210

• Susan Ross, M.D., FRCPC

• (Project Co-Director)

• Chief Scientific Officer

• MetaWorks Inc.

• 470 Atlantic Ave, 10th floor

• Boston, MA 02210

• Isabella Sledge, M.D.

• (Project Co-Investigator)

• Assistant Medical Director

• MetaWorks Inc.

• 470 Atlantic Ave, 10th floor

• Boston, MA 02210

Attachment C: Stakeholders Meeting Agenda

AGENDA

Attachment D: Enclosure B ("Real" and "Ideal" Experiences with Referrals for Epilepsy Management)

"Real" and "Ideal" Experiences With Referrals for Epilepsy Management Circle All Answers That Apply

REAL WORLD (YOUR EXPERIENCE)	IDEAL WORLD (IF DIFFERENT FROM YOUR EXPERIENCE)
I.    CURRENT CARE Referred to MORE SPECIALIZED CARE PATIENTS ARE REFERRED FOR SPECIALIZED CARE FOR THE FOLLOWING REASONS: Initial Diagnosis Unclear Diagnosis Treatment Initiation Treatment Failure: Efficacy Treatment Failure: Side Effects Long-term Management Need for Other Services Insurance Coverage/ Reimbursement Patient Preference   other:_______________________________________ WHAT CRITERIA ARE USED TO DETERMINE, THE TYPE OF SPECIALIZED CARE (GENERAL NEUROLOGIST, EPILEPTOLOGIST, EPILEPSY CENTER) TO WHICH THE PATIENT IS REFERRED? Accessibility Expertise/ Services Offered Insurance Coverage/ Reimbursement Patient Preference Collegial Relationship other:_______________________________________	Initial Diagnosis Unclear Diagnosis Treatment Initiation Treatment Failure: Efficacy Treatment Failure: Side Effects   Long-term Management Need for Other Services Insurance Coverage/ Reimbursement Patient Preference   other:_______________________________________   Accessibility Expertise/ Services Offered Insurance Coverage/ Reimbursement Patient Preference   Collegial Relationship other:_______________________________________
II.    SPECIALIZED CARE Returned to MORE GENERALIZED CARE PATIENTS ARE REFERRED FOR SPECIALIZED CARE FOR THE FOLLOWING REASONS: Long-term Management Need for Other Services Insurance Coverage/ Reimbursement Patient Preference Accessibility Collegial Relationship other:___________________________ ________________________________	Long-term Management Need for Other Services Insurance Coverage/ Reimbursement Patient Preference   Accessibility Collegial Relationship other:___________________________ ________________________________

Attachment E: Stakeholders' Meeting Transcript

AHRQ/CDC/MetaWorks Epilepsy Stakeholders' Meeting

Penn Towers Hotel, Philadelphia

March 25, 1999

1. Introductions/Participants
   Suzanne Smith, MD, MPH, Chief, Health Care and Aging Studies, CDC.
   Patricia Price, DO, Epilepsy Program Director, CDC.
   Frank Gilliam, MD, Epileptologist, University of Alabama Epilepsy Center (did not attend)
   David Cook, MD, Neurologist, University of Pennsylvania Medical Center.
   Jacqueline French, MD, Epileptologist, University of Pennsylvania Medical Center; Co-Principal Investigator.
   Margaret Gunter, PhD, Health Services Researcher, Lovelace Clinic Foundation.
   Greg Barkley, MD, Medical Director, Comprehensive Epilepsy Center, Henry Ford Health System.
   Suzanne Mercure, Institute Health Policy Solutions; (works with employers in purchasing Healthcare, developing standards and requirements for health plans, and assisting consumers with problems and issues; consultant).
   Harry Handelsman, DO, Task Order Officer, AHCPR.
   Louis Trost, MD, Family Practitioner at Lovelace; Co-Chairman of project for integrated epilepsy care.
   David Thurman, MD, Neurologist & Epidemiologist at CDC Injury Center.
   David Stumpf, MD, Pediatrician and Neurologist; participated in Academy of Neurologists Practice Guidelines Committee.
   Joyce Bender, President and CEO of Pittsburgh information technology consulting firm; Member Of President's Committee On Employment Of People With Disabilities ; epileptic patient.
   Linda Warner, Economist; Board member of Epilepsy Foundation; son has epilepsy.
   Patricia Dean, MSN, Nurse Practitioner at a comprehensive epilepsy center at Miami Children's Hospital; board member of Epilepsy Foundation.
   Joanne Levy, MBA, MCP, Project Manger, Leonard Davis Institute, University of Pennsylvania.
   Susan Ross, MD, Chief Scientific Officer, MetaWorks.
   Isabella Sledge, MD, Co-Principal Investigator, MetaWorks.
   Rhonda Estok, RN, Clinical Information Specialist, MetaWorks.
   Sameer Chopra, MA, Project Manager, MetaWorks.

2. Project Background
   (P. Price; S. Smith; S. Ross)
   P. Price (CDC)
   

   • 1997 Living Well With Epilepsy Conference -- 3 work groups (clincal, public health, advocacy)

   • Goal of clinical work group: no seizures + no side effects = control.

   • Barriers to achieving goal:

     1. Physicians and patients don't realize it's an achievable goal (need for public health messages);

     2. Physicians not familiar with important diagnostic/treatment issues;

     3. Patients do not receive integrated care (e.g., education and counseling);

     4. Restrictions on referral to specialists (in managed care capitated networks);

     5. Insurers make decisions without evidence;

     6. Outcomes' evidence not readily available;

   • Identifiable Needs:

     1. Testable management framework to guide clinicians, consumers, and advocates (supported by theory, experience, and expert opinion);

     2. Population-based strategy for optimal care of people with seizures.
        S. Smith (CDC)

   • Evidence will not provide all that is necessary. There are gaps and biases in published scientific literature, and it will be our task to also identify questions to be investigated by outcomes and health services researchers.
     S. Ross (MetaWorks Inc.)

   • Our goal is to develop a framework for organizing the "clinical processes" (including patient support and education) necessary to provide optimal care to patients with epilepsy.

   • This project will be carried out in two phases. Phase I, which includes a preliminary review of available evidence of models of care in epilepsy and other disease settings, this Stakeholder's Meeting, and a report presenting evidence and meeting summaries in addition to a framework and action plan for Phase II, will conclude with a formal meeting between MetaWorks, CDC and AHCPR. The project sponsors can use the report at the end of Phase I to decide whether to proceed with Phase II.

3. Explanation of Preliminary Literature Searches and Findings
   (I. Sledge)
   Literature Review

   • See attached presentation overheads; (Available upon request)
     Models in Alternative Disease Settings

   • Asthma

     1. Institute for Clinical Systems Integration practice guidelines recommend "step approach." Patient can manage disease;

     2. Does not mention, however, what type of specialist should be consulted or when;

   • Diabetes

     1. Existing guidelines from Diabetes Treatment Centers of America mention triggers for referral to different specialists (neurology/ophthalmology), but not evidence-based - based upon expert opinion;

     2. Diabetes Control and Complications Trial (DCCT) -

        1. Tight management of type 1 diabetics (10% of all diabetics) keeps blood sugar levels low over time.

        2. 4 models of care presented:

           • Intensive Care (endocrinologist and specialist team);

           • Co-management (primary care focused with referral specifications for specialty care);

           • Expert System (primary care with treatment algorithms for diabetic care);

           • Mini-Clinic (blocks of practice time carved out to bring in diabetics at one time - consulting specialists present).

        3. Important Message: stratification of patients for different levels of care.

   • Childhood Cancer:

     1. Important Observation: care in specialty center may decrease overall system costs (due to long-term reduction in morbidity and mortality).

     Disease Management Approach

   • Particular arrangements utilize a capitated approach for paying for the disease as a whole, not for particular services;

   • Someone manages care for patient and makes decision for type of care needed;

   • Models exist in cancer: chronic disease of high cost, high volume, and with complex management issues;

   • Referrals today are often tied to reimbursement;

   • Discussion Comment: (J. Bender): medical groups (not just MCOs) are sometimes offering care in this capitated method;

   • Discussion Comment: (D. Stumpf): Intervention in a small group of patients may make large changes (e.g. 10% of patients use 90% of resources);

   • Discussion Comment: Reimbursement decisions often made in isolation of total "systems" perspective - pharmacy benefits managers, those who collect utilization data, etc., perhaps make these decisions independently. Hence, the long-term cost-savings perspective (which includes lost work time and disability) is not always taken when decisions on reimbursement made.
     Disease Management and Epilepsy

   • Epilepsy: Scarce literature on how changing management/referral patterns can improve outcomes or decrease particular types of care utilization (long-term);

   • Several European models of integrated care are available in the literature:

     1. "Epilepsy Care in the Netherlands" (Netherlands, 1997)

        1. Survery of referral patterns/outcomes;

        2. All patients with seizures see neurologists;

        3. Only three specialized centers available;

        4. Mobile outpatient clinics with neurologist decreased admissions to specialized epilepsy centers;

     2. Survey of 14 Epilepsy Centers (Italy, 1997)

        1. Over 3000 patients completed >2 years of follow-up;

        2. Percent of patients achieving remission increased from 13% (prior to evaluation in center) to 28%.

     3. "Epilepsy Needs Document" (England, 1993)

        1. Patient always goes back to GP;

        2. Looked at incidence of epilepsy and how many patients are at each point in pathway;

        3. District epilepsy services - liaison nurses coordinate care between GP and specialist. Neurologist does not have sole care of patient, EVEN if they only have seizures/epilepsy;

        4. Regional guidelines take into account the availability of specialists;

        5. Patient keeps a card to assist communication between primary care physician and neurologist;

        6. Discussion Comment (G. Barkley): So few neurologists in Great Britain - perhaps a driving force for this kind of arrangement;

        7. Discussion Comment (I. Sledge): Region-specific guidelines - not published, but distributed to physicians in various regions.

     4. Scottish Intercollegiate Guidelines Network (SIGN) Recommendations

        1. Based on panel discussion, not a systematic review of available evidence;

        2. Assigned tasks to different types of providers;

        3. Suggests that roles of providers within health system should be clearly defined;

        4. National Health Service in Scotland provides universal coverage;

        5. Discussion Comment: If you deviate from referral paths/criteria, do you not get reimbursed?

        6. Discussion Comment: In a rationed system, you are limited on the number of particular tests, etc., you can use. After a certain amount, services are provided without further reimbursement.

     5. (USA) National Association of Epilepsy Centers - Referral Recommendations

        1. Provided a flow diagram and criteria for referral;

        2. PCP should refer for more specialized care if seizures are not controlled in 3 months;

        3. Neurologist should refer if no seizure control in 9 months;

        4. Discussion Comment (D. Cook): Does not address fact that there are easy and difficult patients. The goal is to find out which patients can and should be referred early. Patients often stratify very early as "easy" or "difficult" patients.
           Discussion Comments Following MetaWorks Evidence Review

   • Patients need early contact with someone who can say, "this is what you have, this is what you need."

   • People who need referral may not be referred perhaps because they are viewed as "not sick enough." Is it possible that only the very sick and the "squeaky wheels" get the specialized care they need in our health systems?

   • (D. Stumpf): looked at Medicare data in Chicago. Suggests that 10% of epileptic patients are consuming 90% of ER visits. That's one criteria for MCO's to monitor care, or for identifying patients for capitated care. We need to focus on the 10% of patients causing all of the resource utilization;

   • Some patients have infrequent seizures but frequently visit the ER. Other patients have frequent seizures but do not seek care. Such situations may be driven by the threshold of the patient and family to accept or react to seizures, their knowledge on what to do and where to go, and the local availability of care;

   • (G. Barkley): some of the "expensive" patients are not difficult to treat, but have been mistreated. Referred to Morrell papers about ER visits and the severity of epilepsy.

   • In managed care, an ER visit is often viewed as a care failure.

     1. Particular tests performed in the ER (CT, MRI, etc.) are expensive and may be completely unnecessary. Because a care-managing physician is not seen or present at the ER, crucial information (e.g., that particular tests may have already been recently performed) is absent.

     2. (J. French): A PCP or neurologist inexperienced with epilepsy may "panic" about the onset of seizures as much as a family learning to cope with the illness. That might explain why there are many instances where such physicians do not oppose their patients' visits to the ER.

   • Seizures are a symptom of multiple conditions. There is a great need to identify and understand the underlying illness that produces the seizures. Only certain providers have the ability and experience to accomplish this task.

   • (D. Stumpf): There is a tremendous need for data. Medicare/Medicaid data is perhaps the best outcomes data available. What kind of data do we need to get answers to our questions? (We can prepare hypotheses for outcomes research in addition to looking to the literature to identify patient groups who need referral.)

   • (D. Cook): There is data about prognosis that could perhaps allow one to predict who will have more problems in the course of their illness and who will not, and when such problems will occur;

   • (J. French): Cited the VA Coop study, which found that if a diagnosis of epilepsy is made, the choice of drug may surprisingly not be a significant factor in the patient's outcome (whether he/she is seizure free or not; the adverse effects of drugs will vary).

   • (D. Stumpf): What percentage of all patients with epilepsy in a population are being seen by neurologists/epileptologists? The literature published on patients reaching that level of care (perhaps the vast majority of available epilepsy literature) may be extremely skewed or biased. Only 20% of all patients reach that level of care...

   • (G. Barkley): Significant number of patients seen by specialists turn out not to have epilepsy (perhaps 25%). Data is skewed if we don't know who at the top of the funnel really has epilepsy. Proper diagnosis is essential. We have no idea what's out there at a population-level.

   • The Easy Part: identifying trigger groups who require specialized care.

   • The Harder Part: identifying a physician's comfort level with providing particular types of care at different levels.

4. Lovelace Seizure Disorder Project Overview
   (M. Gunter)

   • (S. Smith): Introduction of M. Gunter. 1n 1993, Lovelace embarked on a disease management program in epilepsy to improve the quality of care (assuming that cost savings would follow);

   • (M. Gunter): Lovelace Health System

     1. An integrated MCO in New Mexico;

     2. Spent time identifying the costliest diseases to manage. Epilepsy was not one of them, but one particularly influential individual was interested at looking at this condition. Most significantly, Novartis was willing to provide funding;

     3. Explored disease management philosophy, or the idea that improving quality of care will improve patient outcomes (and cost-savings will follow);

     4. 300 physicians on staff (multi-specialty practice);

     5. 1,600 network physicians;

     6. 229,000 member HMO;

     7. 1 (235-bed) acute care hospital;

     8. 9 primary care centers.

   • VISION of project team: to provide high quality, efficient, timely healthcare for patients with seizure disorder with emphasis on empowerment of patients and their primary care providers;

   • MISSION is to excel at seizure management. And, most importantly, improve quality of life of suffering patients;

   • Disease Management Teams were cross-functional/multidisciplinary (primary care, specialists, pharmacologists, nurses, and administrators); they have research fellows and data technicians to collect data, in addition to outcomes support from researchers at Novartis;

   • Major Question: how do we eventually facilitate the use of guidelines after they are developed?

   • Development Phase: development of guidelines that combined evidence-based Identifying and principles and expert consensus. Pilot tested when completed.

   • Tracking potentially epileptic patients:

     1. Requires a good data system to enable the electronic identification of patients;

     2. Identification markers included ICD-9 codes (780.3, 345 series: diagnosis of epilepsy), CPT-4 codes (particular diagnostic tests such as EEGs), and prescriptions for antiepileptic drugs;

     3. All three flags gave an 89% yield in identification of epileptic patients;

     4. Medical record review.

   • Data collection:

     1. Patient specific:

        1. Humanistic component: QOLIE-31 questionnaire (patient satisfaction assessment) sent after patient confirmed as having epilepsy. Responses revealed that patients are not receiving satisfactory education materials on epilepsy;

        2. Clinical component: medical record review;

        3. Economic component: cost of epilepsy patients determined from utilization data;

     2. Provider specific: mail and telephone surveys for providers -- how comfortable are you with your knowledge about epilepsy? The surveys asked questions on provider knowledge across a wide range of seizure disorder work-up and management decisions and also requested advice on the tools that could be developed to help physicians do a better job.

   • Practice Guidelines Products:

     1. Adult new-onset algorithm with chronic management;

     2. Pediatric new onset algorithm with chronic management;

     3. ER new onset algorithm;

     4. Status epilepticus algorithm.

   • Guidelines: include referral criteria. Not worded as "this is what you must follow" but rather as "consider this approach and these suggestions..." which have been developed by a team of your colleagues in the system.

   • Some neurologists are very comfortable with epilepsy and some not. The same is true for PCPs.

   • Simply mailing guidelines to physicians will not improve care or change provider behavior;

     1. How do we help physicians provide better care? (and, at the same time, not be burdensome);

     2. Physicians appear to need trainingin addition to information or algorithms. Training components included:

        1. "Pearls" page;

        2. Laminated pearls card;

        3. Appendices (drug information, seizure characteristics, women of child-bearing age, blood level monitoring);

        4. History forms (initial history, follow-up history) so nothing is forgotten;

        5. QOLIE-10 (quality of life forms);

        6. Health risk assessment;

        7. Case studies.

   • Implementation:

     1. Pilot guidelines tested at 3 sites;

     2. Revisions were made based on feedback;

     3. Practice guideline presentation at organization-wide training sessions;

     4. Development team went to the physicians, rather than vice versa. The result: 80% provider participation in pilot program;

     5. Need local "champs" at each satellite/pilot clinic to carry torch and assist implementation efforts.

   • Key indicators of improvement:

     1. Quality of life;

     2. Improvements in processes:

        1. better outcomes documentation;

        2. work-ups follow guidelines recommendations;

     3. Patient/provider satisfaction;

     4. Improvements in provider comfort level.

   • Accuracy of electronic identification efforts:

     1. Total patients: 2756 (AED, ICD-9, CPT-4s);

     2. Good positive predictive value of electronic data. (CPT and AED best);

     3. Sensitivity: If only looked at three measures, would get 60% of true epileptics.

   • Prevalence: 4.7 per 1000 patients (across all ages); high in 75+ males and children;

   • QOL (Patient) Survey:

     1. 779 patients provided surveys (450 replied);

     2. Frequency of seizures largely important (more significant than intensity);

     3. Socioeconomic status (medium importance);

     4. Co-morbidities (small to medium importance);

     5. For patients taking antiepileptic drugs, the frequency of seizures and adverse effects had large importance.

   • Provider Survey (PCPs):

     1. Assessed knowledge and comfort level with treatment and diagnostic work-up;

     2. Assessed when and why they refer patients for more specialized care.

     3. Those who did not attend presentation/training felt less comfortable, less knowledgeable (mailing guidelines clearly does not by itself change physician behavior).

     4. If a panel of physicians sees more patients, they report feeling more knowledgeable/comfortable.

   • Next steps:

     1. Impact analysis (measurement of outcomes, post-implementation):

        1. Provider survey;

        2. Case finding;

        3. Provider confirmation (of patients);

        4. QOLIE-31;

        5. Medical record review;

        6. Utilization analysis;

        7. Cost analysis;

     2. Reinforcement by local champions;

     3. What's best for patient/providers to help them provide more efficient, cost-effective care?

     4. What can be accomplished without overwhelming physicians?

     
     Discussion of Presentation

   • (H. Handelsman): What would you like to learn from the project at Lovelace? How satisfied are you with the outcomes of your program?

   • (M. Gunter/L. Trost): We would like to learn when patients should be referred and to whom they should be referred. The impact of the project is still to be measured. But doctors in the system are familiar with each others' credentials and experience and can trust skills of the guidelines committee. In the system, trust has replaced "rules" on how to refer or when to refer.

   • (M. Gunter/L. Trost): We don't know if the results will be applicable to other health care organizations. Lovelace is an isolated, staff-model system (owned by Cigna), and our experience might not be generalizable.

   • Guidelines should be established to ensure that everything that is needed for patient gets done (history, diagnostics), regardless of who does it.

   • Poor documentation was an impediment to effectively managing care and measuring outcomes. Hard to see if important issues were not properly being dealt with!

   • After training, PCP has better idea of what he/she should be comfortable doing and what he/she shouldn't.

   • (J. French): Is it more cost-effective to teach 100 PCPs to manage particular components of care for 1 - 2 epileptics, or more cost-effective to just refer all of the patients to 1 specialist? Because epilepsy is a low prevalence condition, maybe PCPs don't have to know as much about epilepsy as they know about asthma, diabetes, etc.

   • (S. Smith): Maybe specialized care is not the most cost effective way to diagnose and manage epileptics. We need data to support our hypotheses.

   • (M. Gunter): For treating asthma within the Lovelace system, we have achieved great success in sending patients to specialized clinics.

   • To measure improvement, we need patients to be meticulous about recording when they have seizures and and the ways in which their condition improves. This is a very difficult task for obvious reasons!
     BREAK

5. Discussion of Stakeholders' Experiences: the "real" and the "ideal"
   Stakeholders broke into groups to discuss their perspective on real and ideal world referral experiences, using a framework provided to each before the meeting (attached to this transcript). The groups reconvened and reported their views:

Primary Care (L. Trost, I. Sledge)

• All of the issues listed on the survey sheets are currently taking place and reflect why referrals are made;

• In the ideal world, all of the referral reasons are situation-dependent and heavily influenced by provider knowledge and comfort level;

• Curbside consults = same level of care;

• Most patients in managed care networks are referred back to generalized care;

• Pediatricians hold on to patients longer than do providers of care to adults;

• Most effective efficacy path may actually raise costs. Let's consider efficacy, not cost, first;

• Is it worth educating PCPs or having them refer patients to specialists in every instance?

• Ideally, education for primary care providers is important even if only to reduce panic and improve triage;

• In the ideal world, reimbursement or insurance coverage should not be an issue;

• Patient preference should win in the ideal world;

• (L. Trost): Interesting concept from another meeting: Patient-centered practice policies should be based on what the patients need rather the role of particular providers. Guidelines daring to address "who" should emphasize competencies (Association of American Family Practitioners.)

• There is tremendous variation in practice style. It would make sense to study these differences and compare outcomes. High patient volume may lead to better outcomes; (there's a lot of literature in surgical specialties demonstrating that the more patients with a particular condition a provider sees, the better the outcomes).

Neurology (D. Cook, D. Stumpf)

• Diagnosis and treatment can be conceptualized as a timeline. Where are the weak-points where expert intervention would be necessary? Six break-points can be identified:

Researchable Question: Which patients are at risk for poor prognosis/problem courses? Can we classify patients by our ability to control their seizures?

Difficulty Experienced in Controlling Seizures of Particular Types of Patients

Issues to Consider When Discussing Referral at This Level of Care

• Thought process: there are some "low-hanging fruits" (patients whose seizures are easy to classify, control, and treat) who can be brought quickly to remission;

• Easy patients may not need to be referred (low hanging fruit), but they still need to be counseled;

• Are there variations in style of practice between neurologists? What are the differences?

• What dictates the choice between monotherapy vs. polytherapy?

• Organization of care... are "curbside" consults possible or not?

• Where is the waste in the system (use of ER, bad diagnosis, misuse of technology, duplicity of CT and MRI testing, overuse of EEG);

• Any breakdown at any point in the treatment timeline above triggers referral. (David Cook added the following comments in correspondence after the meeting: "I think that many GPs are not aware of when a breakdown occurs. Some of these 'breakdowns' may not cause seizures but may cause long-lasting problems in terms of difficulty with management and/or appropriate monitoring. Therefore, awaiting a breakdown in the system before the referral may be problematic.")

• Counseling takes lots and lots of time that a neurologist may not have (who will do it?);

• Our discussion is an issue regarding classification of patients: easy things that doctors can do vs. what needs to be done by other, more epilepsy-focused individuals;

• PRINCIPAL CARE ISSUE: Often patients don't go to PCP, but directly to neurologist who manages all care (because epilepsy is the primary concern). Neurologists have varying levels of comfort and expertise providing primary care (e.g., cholesterol management, pap smears, etc.) For a seizure-free patient, a neurologist can provide primary care and often does;

• (J. Stumpf, G. Barkley): Counseling concept -- sequential presentations (counseling). People can't assimilate important information all at once.

Epileptologists (J. French, G. Barkley)

• Variation in care is an important issue. Epileptologists receive patient referrals from all different points in the course of their epileptic condition. No matter which point of care a patient arrives, 90% of patients claim that they have little information or understanding about their condition and the necessary care they should seek;

• In the real world, reimbursement is a huge reason for why some people are referred to epileptologists and why others are not;

• No matter how good the care of the referring neurologist, there are always little things about management of patient that could be tweaked: epileptologists certainly provide value-added services;

• How to choose patients for referral that will most benefit from the epileptologist and his or her resources?

• There is a common misperception that the epileptologist charges more. Cost may hence be a confounding issue when making decisions about when to refer to such specialized care. The epileptologist charges about the same as a neurologist but provides more time and more specialized care to patients. Epileptologists don't make money based on number of patients they see in an hour. Most of the revenue at epilepsy centers comes from drug studies, surgery referrals, academic sources, etc. This arrangement is a luxury that people in office practice may not have;

• Why do epileptologists refer patients back to more generalized care? To maintain collegial relations. Difficult to get patients to go back! The patient realizes that so many important resources are available (social worker, psychologist, epilepsy nurse, etc.) at epilepsy centers. When they call in middle of night in case of an emergency, everyone who answers has some specialized epilepsy training;

• Other circumstances would warrant referral to more general care: remission of seizures without side-effects; the patient is seen, evaluated, and a care plan for a neurologist can be formulated. If any problems arise in long-term management, the patient can always be sent back to the epilepsy center;

• How do we get patients to drive the process of moving between different types of providers? Is this even a good idea? (Many patients are perhaps not as articulate and forthright as others);

• Empowering patients is very important in modifying patterns of care usage; patients need to know more about their condition;

• The mantra "seizure free with no side effects" often drives referrals. But there's a point where you have to consider the patient's quality of life and draw the line as to how intense or invasive the medical interventions will be;

• Do we control seizures to the patient's or physician's satisfaction? Patients should be educated enough to know what's best for them... and what to expect from their care providers;

• (S. Smith): "No seizures, no side effects" a physician-centered perspective when it emerged at the Orlando conference;

• (J. French): Physicians need to have realistic expectations. Probably 30% of patients will not achieve goal of "no seizures, no side effects."

Open Discussion

• Patients seem to have epiphanies about their condition at epilepsy centers. A multidisciplinary approach to patient care is clearly needed. People lose control in their lives because they are randomly attacked by a seizures - it's very different than, for instance, having predictable arthritis every morning, or having chest pain when exercising too hard (e.g., heart patients). Patients need knowledge to develop understanding about their lives and condition - this will lead to improved quality of life;

• Problem: sometimes doctors think they can be everything to their patients and don't refer;

• In Orlando, it came out in discussions that doctors were never referring well-informed patients to the epilepsy foundation for resources and information. Doctors can't assume that even these patients will not need support from other individuals. Doctor shouldn't feel that they have supplied all the information necessary for an epileptic to get by, because that is rarely the case;

• Phenytoin may not be the best drug of choice for all patients (adverse drug reactions), but it is the most commonly prescribed. Physicians are stubborn and need evidence-based information;

• Internet: source of both information and misinformation. Evidence-based information needs to become freely available via this medium. For instance, CDC needs to lead the way so people know where to turn when they have information needs;

• Many good practice guidelines (evidence-based) are available in other disease settings. Problem is how to implement and change behavior of everyone in a system who provides care. Need complex issues distilled to algorithms that can be easily followed in busy practice.

Health Services/Payors (S. Mercure, M. Gunter)

• Health plan and provider selection by employers is typically made in a complete vacuum of information. Employers simply assume that the health plan has important evidence-based information and is providing the right array of services and providers for patients within the plan. Employers do not feel the need for details;

• Cost-effectiveness and total health care costs are critical. The only way any referral patterns will be adopted is if they save total costs to the system and improve the quality of healthcare. Cost includes lost work time as well as direct health care expenditures;

• Employers will pay only if they know something is worth it. Is $5 in extra premium going to lead to better outcomes for employees and hence less days off from work?

• Health professions need to remove their own job title barrier problems. Rely on the mantra: "Getting people treated in the right way, at the right time, with the right skills, in the right setting, with the right outcome, at the right price;"

• Consumers buy into evidence-based practice ("I have to see a specialist"). Self-referral occurs when people make a fuss within their system;

• Patients may select plans based on relationships with particular providers, generalists or specialists;

• Expectations from providers need to be defined by consumer groups;

• Specialized services (e.g., education) do not necessarily have to be performed by specialists;

• Telephone conversations work well in organized systems (without a real referral). Cost-saving curbside consultations.

Consumers (L. Warner, P. Dean, J. Bender)

• Patient referrals are fueled by dissatisfaction with outcome of current care;

• Patient preference is the main driver behind referrals (and should be);

• Patient education fuels demand for appropriate care - often one call or the sound of the word "appeal" to an MCO gets them to pay;

• ER is a dangerous, undesirable place;

• Different patients have different thresholds and levels of education and preparation which also guide referrals;

• Common reason for referral: access to advanced treatments;

• Many people with epilepsy do not have PCPs but neurologists as their primary caregivers. Many advocacy groups want the specialist to be the gatekeeper. A dangerous trend for people with multiple conditions? Does this promote gaps in preventive care?

• Some doctors are satisfied with their control of seizures when in actuality, the seizures can be better controlled. It comes down to seizures vs. drug side effects... Satisfaction of patient will stem from them knowing this trade-off and accepting the right balance of intervention;

• There is still a lot of stigma against epilepsy - which prevents people from going to a doctor in the first place;

• Doctors and other health care workers often overreact to epilepsy/seizures - how are patients supposed to act and feel? Education of the primary care givers is extremely important;

• What is patient's goal (for treatment)? This has to be identified by providers AFTER the patient is educated. Quality of life vs. remission of seizures. What risk are patients willing to take to control their seizures?

• GOAL: "no seizures, no side effects" - a goal for most folks. A good piece of public education for patients and providers (that it is possible at all). Important for this word to be out there.

____________________

POST LUNCH

____________________

V. Approaches to Framework/Model for Referral Criteria (using best available evidence)

Referral Strategies That Could Be Utilized in Framework (J. French, I. Sledge)

1. Point of care - referral at important points in care that may require expert opinion and resources: first diagnosis, choosing medications, drug delivery, changing medications;

2. Time-based - no referral while patient is symptom free and without side effects Referral if remission/control of seizures not achieved in a particular amount of time;

3. Population-based - particular at-risk populations need referral (e.g., epilepsy in childbearing age, epilepsy with co-morbid condition, epileptic on three or more drugs, etc.);

4. Service-oriented approach - what can a PCP do? What facilities/resources/services are needed to be able to manage epilepsy? And where are they located? Patient referrals driven by availability of services.

Discussion

• How do we find evidence for particular referral paths? ("back door" approach);

• Strategies above are not necessarily mutually exclusive (particularly 1 and 3);

• Where do we think we can find evidence?

• What evidence will be useful to people whom we want to use the evidence?

• If no evidence available, then develop research questions...

• Who accredits epilepsy centers? What defines the resources available at the comprehensive epilepsy centers? National Association of Epilepsy Centers established expectations for the centers 10 years ago. Currently, there is no national standard established by a funder, government agency, etc;

• Goal for group: not to get locked into particular types of providers, centers, etc., since what they can deliver is varied across the United States;

• Focus on critical elements of care. Could centers help define what needs to be available for epilepsy patients?

• Where do patients get stuck when they really should be referred for further care? At a specialist or a generalist? Patients may get stuck when they don't realize that there are better services available. Consumer needs to be educated as to what the services are. Their own advocacy may drive treatment;

• The group representing consumers warned, however, that consumer driven health care is a blight on the system. Many people cannot speak for themselves to get what they need;

• (S. Ross) Strategies 3 and 4 seem complementary and researchable. Develop a menu of potentially important services, consider specific populations, and then bring a time model to the issue. These are researchable strategies... and it's easy to identify where the gaps are. This approach gets away from talking about specific types of providers.

• Need to survey evidence to back up pieces of the framework. Outcomes data will be a big missing piece of any puzzle, but outcomes studies are a second step needed to answer the question: "Will the evidence assembled have merit in real life?"

• Need to help: CONSUMERS, PRIMARY CARE PROVIDERS, and INSURERS make decisions about referral. We have to get them information in a concise, logical way...

• Stumpf warns that the existing literature is published mainly by neurologists and subspecialty neurologists whose patient populations are small subsets of the real population. How to develop population-based guidelines from skewed evidence?

Populations Who May Need Referral for More Specialized Care

(Numeric ranking shown below is derived from a consensus vote and indicate research priority based on potential impact on healthcare systems, patients, and providers)

Note: Support system should be independent variable for every population category.

(1) All those undergoing 1^st diagnosis (children, adults, elderly) : Patients presenting with seizures could have any one or more of a number of different conditions. Someone needs to interpret different types of diagnostic information to classify the seizures. Are there predictors of changing diagnosis? Features to search in the literature: epidemiology (including misdiagnosis), history, work-up results, differential diagnosis, initiation of treatment, and venue of first diagnosis (ER or office, for instance).

Women with epilepsy (peri-childbearing, peri-menopausal)

Elderly patients

Patients presenting with febrile seizures

(3) Patients with comorbidities (brain injured stroke, motor syndromes, mental retardation, etc.): More likely to need specialized care to manage seizures or both conditions simultaneously.

(4) Patients demonstrating intractability

Patients with drug interactions

(4) Patients with side effects from Antiepileptic Drugs

(5) Withdrawals of Antiepileptic Drugs (Adult, Child)

Discussion

• (D. Stumpf) was concerned that there is not even sufficient data to show that EEGs are important to improving patient outcomes. Stumpf was very concerned that there is not enough data out there to create a menu of services required for the diagnosis of epilepsy;

• (J. French) agreed in part, but suggested that part of our responsibility to find out where further studies are needed;

• (D. Stumpf) thought of this project as a hypothesis-testing exercise.

• (S. Smith) reminded the group that this is neither a hypothesis-testing or hypothesis-generating exercise, but a policy planning exercise;

• (G. Barkley) refers to an organization of care in epilepsy article by Gruman in the Journal of Ambulatory Care Managment (S. Smith has the paper);

Menu of Services

First Diagnosis

General categories of things to look at:

• Technology

• Therapy

• Differential Diagnosis

• History

• Epidemiology

• Work-up

Things that need to be done:

• Take history

• Full physical

• Work-up (lumbar taps, blood work)

• Counsel the patient

• OT, PT, speech and language assessment

Things that might need to be available:

• EEG

• Scans (CT, MRI)

• Someone with knowledge of AEDs

• Someone with knowledge of classification

• Special EEGs with invasive monitoring, video monitoring, etc.

Discussion

• There's a lot that can be done at the time of first diagnosis that can save $$ and improve outcomes down the line.

• Issue: do we need evidence for what's needed to make a diagnosis of the first seizure in a population? The literature on comparative diagnostics is enormous.

• Is there evidence to show what services/resources should be on the menu? Is that what we need to look at?

• Problem: If roles are not established alongside what needs to be done, then people who are unqualified to perform certain roles may end up performing them but incorrectly... but the literature is not likely to be there to support how comfortable or knowledgeable physicians feel about making diagnosis. This will be a major gap.

• Consumers (or payors) can be alerted to know that there are questions that should or can be asked to a provider.

• Issue: incorrect diagnosis is a huge issue. Where do we draw the line when talking about diagnosis? Should we look at related disease states to dig up people who are epileptics but misdiagnosed? (e.g., look up syncope and see how many people are epileptic.)

• We need to alert people that misdiagnosis is extremely common and that there are consequences.

VI. Next Steps
(S. Ross)

• Meeting report will go to CDC/AHCPR for review;

• Final meeting in Phase I will occur in April/May;

• This will be followed by a go-no-go Phase II decision by the CDC;

• Suggest a framework for organizing researchable issues by doing a cross tab of item 3 (populations) and 4 (services). Each selected cell in the matrix could be researched for quantity and quality of evidence available in literature.

VIII. Epilogue (CDC/AHCPR/MW/LDI organizers)

By next week :

• MW will develop the cross tab framework described above;

• MW will produce and distribute meeting transcript;

• We will all continue weekly meetings but with potential change of day and time;

• CDC will send MW a copy of Gruman paper referenced earlier: (Gruman J., VonKorff M., Reynolds J., Wagner E.H., and Barkley G.L. Organizing health care for people with seizures and epilepsy. J Ambulatory Care Manage 1998; 21/2: 1 - 17.

According to Task Order timeline:

• MW will develop and submit first draft of Final Report on Phase I of Task Order;

• MW will coordinate all schedules for Phase 1 close-out meeting in Washington, DC to review Report & Recommendations for Phase II.

N. IX. Draft Sample Matrix for Evidence Review

Attachment F: Matrix - Healthcare Services and Resources

Critical Questions for Systematic Review AHCPR/CDC Epilepsy October 20, 1999

LEGENDDomains for Prioritizing Questions for Systematic Review

Purpose: To refine our literature review, members of the project technical expert panel (TEP) were asked to rate each population (and critical questions associated with that population) against the prioritization criteria listed on the y-axis (left-hand side) of the following spreadsheet. The cells of the spreadsheet contain aggregate TEP ratings. The tallies at the bottom of the spreadsheet facilitated the decision to pursue a systematic review of the literature concerning patients undergoing the first diagnosis of epilepsy ("Population One"). The critical questions relevant to each population can be found in Appendix C.

Population One: Patients undergoing their first diagnosis.

Population Two: Patients demonstrating intractability.

Population Three: Patients presenting with comorbidities.

Population Four: Patients with side effects from anti-epileptic drug therapy (AEDs).

Population Five: Patients withdrawing from AED therapy.

Criteria for Referral of Patients with EpilepsyPhase II Work Plan: Part BNovember 23, 1999December 14, 1999 (Revision)November 23, 1999

Bibliographic Record