Abstract
SUMMARY
Purpose:
To validate a brief screening instrument for identifying people with epilepsy in epidemiologic or genetic studies.
Methods:
We designed a nine-question screening instrument for epilepsy and administered it by telephone to individuals with medical record–documented epilepsy (lifetime history of ≥2 unprovoked seizures, n = 168) or isolated unprovoked seizure (n = 54), and individuals who were seizure-free on medical record review (n = 120), from a population-based study using Rochester Epidemiology Project resources. Interviewers were blinded to record-review findings.
Results:
Sensitivity (the proportion of individuals who screened positive among affected individuals) was 96% for epilepsy and 87% for isolated unprovoked seizure. The false positive rate (FPR, the proportion who screened positive among seizure-free individuals) was 7%. The estimated positive predictive value (PPV) for epilepsy was 23%, assuming a lifetime prevalence of 2% in the population. Use of only a single question asking whether the subject had ever had epilepsy or a seizure disorder resulted in sensitivity 76%, FPR 0.8%, and estimated PPV 66%. Subjects with epilepsy were more likely to screen positive with this question if they were diagnosed after 1964 or continued to have seizures for at least 5 years after diagnosis.
Discussion:
Given its high sensitivity, our instrument may be useful for the first stage of screening for epilepsy; however, the PPV of 23% suggests that only about one in four screen-positive individuals will be truly affected. Screening with a single question asking about epilepsy yields a higher PPV but lower sensitivity, and screen-positive subjects may be biased toward more severe epilepsy.
Keywords: Epilepsy, Epidemiology, Sensitivity, Specificity, Questionnaire, Validity
Valid screening questions to identify people with a history of seizures or epilepsy are crucial for epidemiologic studies. They are also essential for identifying affected family members in genetic studies. Screening relies on careful history-taking, but it is not clear what questions should be asked, or how accurately they identify affected individuals. Several previous epidemiologic studies have assessed epilepsy prevalence using a two-stage screening strategy in which an initial broad screen identified possible cases and subsequent, more detailed assessment separated true from false positives (Osuntokun et al., 1982; Haerer et al., 1987; Schoenberg, 1987; Meneghini et al., 1992; Placencia et al., 1992; Nicoletti et al., 1999; Borges et al., 2004; Melcon et al., 2007; Noronha et al., 2007). The initial screens included both items that targeted recognized seizures and items that targeted symptoms possibly reflecting unrecognized seizures [e.g., “Have you ever had attacks in which you lose contact with the surroundings?”(Placencia et al., 1992)]. The use of symptom-based questions is essential for maximizing sensitivity (i.e., the proportion of true cases in the population who screen positive), especially in settings with limited access to medical care, and many of these questionnaires were found to have sensitivity ≥95% (Osuntokun et al., 1982; Meneghini et al., 1992; Placencia et al., 1992; Borges et al., 2004; Melcon et al., 2007); however, these high sensitivities came with a “cost” of false positives, and estimates of specificity (i.e., the proportion of unaffected individuals who did not screen positive) were usually lower (Osuntokun et al., 1982; Meneghini et al., 1992; Placencia et al., 1992; Borges et al., 2004; Melcon et al., 2007).
False-positive rates have major implications for the effort and cost involved in a study because they affect positive predictive value (PPV): the proportion of screen-positive individuals subsequently confirmed to be affected. Because epilepsy is relatively uncommon (life-time prevalence 0.5–3% in many studies), even a small false-positive rate will be applied to more than 95% of the individuals sampled, so that the proportion of screen-positive individuals who are truly affected could be quite low. For many of the screening questionnaires used previously, PPV was as low as 20% (reviewed in Placencia et al., 1992).
We designed a nine-question screening instrument to identify subjects with epilepsy in a telephone interview. We assessed sensitivity and false-positive rates for the instrument by administering it to individuals with medical record–documented epilepsy or isolated unprovoked seizure and individuals who were seizure-free on medical record review, from a population-based study. We also explored the effect of restricting the screen to different subsets of questions. Finally, we used our estimates of sensitivity and false-positive rates to extrapolate to the anticipated PPV that would be obtained in a population survey with our instrument.
Methods
The cases for this study comprised all residents of the city of Rochester, MN, U.S.A., who were born in 1920 or later and had incidence of either epilepsy (two or more unprovoked seizures) or an isolated unprovoked seizure between 1935 and 1994. Patients meeting these criteria had been identified in previous studies using the resources of the Rochester Epidemiology Project (REP) (Melton, 1996) as described previously (Hauser et al., 1993, 1996). The restriction to cases born in 1920 or later was introduced to maximize the number of cases that could be interviewed, since those born earlier were likely to be deceased. For each case, we selected as a control a patient who had not had an unprovoked seizure before the case's diagnosis date and who matched the case by sex, birth year (±5 years), and length of contact with the medical records linkage system (first contact with an REP provider within one year of that of the case, and medical visit to an REP provider within one year of the case's diagnosis date). Potential controls were not excluded if they had new-onset unprovoked seizures after the case's diagnosis date or if they had febrile or other acute symptomatic seizures. No other exclusions were made in the selection of either cases or controls.
Between 2003 and 2008, we carried out a comprehensive review of the medical records of each case or control at the Mayo Clinic and all other local providers. Abstraction involved initial review by trained nurse abstractors followed by expert review by the study epileptologists (JRB and WAH) and provided detailed information for the duration of each subject's residence in the Rochester area, including all outpatient examinations, home and emergency room visits, hospitalization records, laboratory tests, and neurologic and other special examinations. Results were recorded on standardized, Web-based forms.
Independently of the medical record abstraction, we attempted to interview each surviving case and control. Interviews were administered through a computer-assisted telephone interview implemented in Ci3 software (Sawtooth Technologies, Northbrook IL, U.S.A.). The interview included a screening instrument to screen for lifetime history of seizures (Table 1), followed by a diagnostic interview to obtain further clinical details in subjects who screened positive. The diagnostic interview was a modified version of the form we described previously (Ottman et al., 1990, 1993). Interviewers were blinded to record-review findings.
Table 1.
Questions from Screening Instrumenta
|
Each question could be answered no, yes, possible, or don't know.
Phrase “Other than the seizure[s] you had because of a high fever” added only if subject responded “yes” or “possible” to question 1.
For each question in the screening instrument, we classified the response as positive if the subject answered “yes” or “possible.” We considered four definitions of a positive screen, consisting of a positive response to: (1) any question in the screening instrument (any positive), (2) any of Q2 through Q3G (any positive excluding febrile seizure), (3) Q2 or Q3A (epilepsy or any seizure), and (4) Q2 only (epilepsy only). We also examined the effect of combining these definitions with a positive response to the diagnostic interview question, “Have you ever taken medications for seizures?” Subjects who responded positively to only the febrile seizure question (Q1) were not given a diagnostic interview; therefore, we were unable to examine this question in conjunction with the medication question.
For each screen definition, sensitivity was defined as the proportion of subjects with medical record–documented unprovoked seizures who screened positive, and the false-positive rate (1-specificity) was defined as the proportion of subjects who screened positive among subjects found to be seizure-free on record review. Some of the controls had been interviewed after moving away from the Rochester area, and thus could have had seizures that were not captured by the REP medical records linkage system. To address this problem, we restricted analyses of the false-positive rate to seizure-free controls who had one or more visits to an REP provider in the year of interview or later.
We also estimated the PPV of each screen definition. Because PPV is highly dependent on prevalence, it would have been incorrect to calculate PPV directly from our case–control data, in which interviewed cases comprised approximately half of the subjects. Instead, we estimated PPV according to the formula
where p is the assumed prevalence of epilepsy in the population (1%, 2%, or 3%).
This study was approved by the institutional review boards of the Mayo Clinic Rochester, Olmsted Medical Center, and Columbia University Medical Center. All interviewed subjects gave written informed consent.
Results
The study included 910 cases with incident unprovoked seizures and 941 controls (including some controls identified for cases later found to be ineligible). Among these 1851 subjects, 529 could not be interviewed because they were deceased (n = 295), could not be located (n = 130), or were unable to answer the interview questions because of illness or disability (n = 104). Thirty-four percent (449 of 1,322) of the remaining subjects agreed to be interviewed. Participation rates were higher in cases than in controls [37% (221 of 594) vs. 31% (228 of 728), p = 0.025], in women versus men (38% vs. 30%, p = 0.001), and in older versus younger subjects (<35 years 23%, 35–49 years 30%, 50–64 years 39%, ≥65 years 47%, p < 0.001). Among the cases, participation rates were higher in subjects aged ≥20 versus <20 years at the first unprovoked seizure (48% vs. 33%, p < 0.001), but did not differ significantly between subjects with epilepsy versus isolated unprovoked seizure or between subjects with focal versus generalized epilepsy.
Three interviewed cases were excluded because the information in the medical records was too sparse for classification (n = 2) or all of the interview screening questions were answered “don't know” (n = 1). Among the remaining 218 interviewed cases, 166 were classified on medical record review as having epilepsy and the remaining 52 isolated unprovoked seizure. Review of the medical records of the 228 interviewed controls revealed that two had had onset of epilepsy and two others an isolated unprovoked seizure, in each instance after the case's diagnosis (therefore, not disqualifying them as controls). For analysis of the sensitivity of the interview questions, we combined the information from these affected controls with that from the cases, so that 168 subjects with epilepsy and 54 subjects with isolated unprovoked seizure were included. Five other interviewed controls had had seizures on medical record review (one febrile seizure, four other acute symptomatic seizures), and were excluded from our analyses, as were 99 controls whose last medical visit to an REP provider was a year or more before interview. Analyses of false-positive rates were restricted to the 120 remaining controls whose medical records contained no indication of seizures and who were presumed to be seizure-free.
Among all included subjects, 61% were women, 44% were college graduates, and 97% were white non-Hispanic. These factors did not differ among subjects who were seizure-free and those who had epilepsy or isolated unprovoked seizure. Age at interview averaged 54 years ± 0.9 (standard error [SE]) overall; seizure-free subjects were older at interview (58 years) than those with epilepsy (52) or an isolated unprovoked seizure (50) (p = 0.002).
Among subjects with epilepsy on record review, 76% responded positively to the “epilepsy” screening question (Q2), whereas only 46% of those with isolated unprovoked seizure on record review did so (Table 2). Among subjects who did not respond positively to Q2, 15% of those with epilepsy and 35% of those with isolated unprovoked seizure responded positively to the “any seizure” question (Q3A). Few subjects responded positively to the questions relating to seizure symptoms and, with the possible exception of the question asking about “a change in mental state or level of awareness” (Q3C), subjects with and without unprovoked seizures were equally likely to respond positively to these questions (Table 2).
Table 2.
Number (percentage) of subjects who responded “yes” or “possible” to specific questions
| Medical record–based diagnosis |
|||
|---|---|---|---|
| Questions from Screening Instrument |
Epilepsy (n = 168) |
Isolated unprovoked seizure (n = 54) |
Seizure-free controla (n = 120) |
| Q1. Febrile seizureb | 4 (2.4) | 1 (1.9) | 1 (0.8) |
| Q2. Epilepsy/seizure disorder | 128 (76.2) | 25 (46.3) | 1 (0.8) |
| Q3A. Any seizure (if “No” to Q1) | 25 (14.9) | 19 (35.2) | 3 (2.5) |
| Q3B–Q3G.≥1 seizure symptomc | 4 (2.4) | 2 (3.7) | 3 (2.5) |
| Q3B. Uncontrolled body movements | 1 | 2 | 1 |
| Q3C. Change in mental state | 3 | 0 | 0 |
| Q3D. Daydreaming/staring | 2 | 0 | 2 |
| Q3E. Reactions to light | 1 | 0 | 0 |
| Q3F. Jerking/clumsiness on awakening | 1 | 0 | 0 |
| Q3G. Other repeated unusual spells | 1 | 0 | 1 |
Restricted to controls with a medical visit to a local provider (at the Mayo Clinic or another Olmsted County site) in the year of interview or later.
Number of subjects who responded positively to Q1 but not to any of the other questions.
Number of subjects who responded positively to any of Q3B through Q3G but not to either Q2 or Q3A. The numbers shown for individual items in Q3B through Q3G are not mutually exclusive.
Sensitivity was highest for the broadest screen definition (positive response to any screen question) among subjects with either epilepsy (96%) or isolated unprovoked seizure (87%) on record review; however, 7% of subjects who were seizure-free on record review also screened positive according to this definition (Table 3). When the febrile seizure question was excluded, sensitivity declined slightly for both subjects with epilepsy (94%) and those with an isolated unprovoked seizure (85%). Among four subjects with epilepsy who responded positively to the febrile seizure question but no other screen question (Table 2), three had had febrile seizures before epilepsy diagnosis.
Table 3.
Sensitivity and false-positive rates for different definitions of a positive screen
| N (%) Who screened positive, by medical record diagnosis |
|||
|---|---|---|---|
| Definition of positive Screen |
Epilepsy (n = 168) |
Isolated unprovoked seizure (n = 54) |
Seizure-free (n = 120) |
| Screening Instrument questions only | |||
| Any positive (any of Q1 through Q3G) | 161 (95.8) | 47 (87.0) | 8 (6.7) |
| Any positive excluding febrile seizure (any of Q2 through Q3G) | 157 (93.5) | 46 (85.2) | 7 (5.8) |
| Epilepsy or any seizure (Q2 or Q3A) | 153 (91.1) | 44 (81.5) | 4 (3.3) |
| Epilepsy only (Q2 only) | 128 (76.2) | 25 (46.3) | 1 (0.8) |
| Screening Instrument questions and yes to medications for seizures in Diagnostic Interview | |||
| Any positive excluding febrile seizure (any of Q2 through Q3G) | 140 (83.3) | 25 (46.3) | 3 (2.5) |
| Epilepsy or any seizure (Q2 or Q3A) | 136 (81.0) | 24 (44.4) | 2 (1.7) |
| Epilepsy only (Q2 only) | 117 (69.6) | 15 (27.8) | 1 (0.8) |
When the screen definition was restricted to “epilepsy or any seizure” (Q2 or Q3A), sensitivity was only slightly lower, and the false-positive rate declined to 3%. With the “epilepsy only” screen definition (Q2), sensitivity declined to 76% among epilepsy patients, and the false-positive rate declined to 0.8%. Sensitivity for all definitions was reduced when the screen definition was amended to include a positive response to the medication question, but the false-positive rates were also lower (Table 3).
Figure 1 illustrates the effect of different screen definitions on sensitivity and PPV in a screen for epilepsy with assumed lifetime prevalence in the population of 1%, 2%, or 3%. Although sensitivity was highest for the broadest screen definition, PPV was lowest. For example, if lifetime prevalence is assumed to be 2% in the population, the expected PPV is approximately 23%. As sensitivity declined, in general, so did the false positive rate; for example, in a screen using just the epilepsy question (Q2), the expected PPV was 66% with a lifetime prevalence of 2%. Use of the epilepsy question with the additional requirement of a positive response to the medication question from the diagnostic interview lowered sensitivity without changing the false-positive rate, and thus the PPV for this screen definition was slightly lower than with the epilepsy question alone (64%, Fig. 1).
Figure 1.
Sensitivity and expected positive predictive value (PPV) in a screen for epilepsy with different screening definitions. Sensitivity: proportion of medical record–documented epilepsy patients who screened positive. PPV: proportion of screen-positive subjects predicted to have epilepsy = (p*sensitivity)/[(p*sensitivity) + (1 − p)*(false positive rate)], where p = assumed lifetime prevalence of epilepsy (1%, 2%, or 3%). Screen definitions are as shown in Table 3.
Epilepsia © ILAE
To explore the reasons for the false-positive rate, we examined the histories of the eight subjects who screened positive but were found to be seizure-free on medical record review. The single seizure-free subject who responded positively to the epilepsy question had psychiatric problems noted in the medical record. Although she reported having taken antiepileptic medications, the medical record did not confirm this. She had multiple electroencephalography (EEG) studies for dizziness and falls, none of which showed epileptiform abnormalities. Three of the other screen positive subjects had benign positional vertigo, and one of them also had syncope. The remaining four subjects had no relevant diagnoses in the medical record.
Finally, in subjects with epilepsy on record review, we examined demographic and clinical epilepsy features associated with a positive screen using the “epilepsy only” and “any positive” screen definitions (Table 4). Sensitivity was not significantly associated with sex, education, age at diagnosis, or major epilepsy syndrome for either screen definition. For both screen definitions, sensitivity was significantly lower in subjects diagnosed before 1965 than in those diagnosed more recently. Sensitivity also appeared to be lower in subjects who were older at the time of interview than in those who were younger, especially for the “epilepsy only” screen definition, but this was partly due to confounding with time-period of diagnosis, since subjects diagnosed before 1965 were older at the time of interview than those diagnosed more recently (average 61 vs. 49 years, p < 0.001).
Table 4.
Sensitivity of “Epilepsy only” (Q2 only) and “Any Positive” (any of Q1 through Q3G) screen definitions in subjects with epilepsy, by demographic and clinical features
| Epilepsy only |
Any positive |
||||
|---|---|---|---|---|---|
| Demographic and clinical features |
Total subjects |
Screen positive n (%) |
p-valuea | Screen Positive n (%) |
p-valuea |
| Sex | |||||
| Male | 69 | 55 (79.7) | 0.371 | 68 (98.6) | 0.141 |
| Female | 99 | 73 (73.7) | 93 (93.9) | ||
| Educationb | |||||
| High school graduate or less | 42 | 30 (71.4) | 0.719 | 40 (95.2) | 0.967 |
| Some college | 54 | 42 (77.8) | 52 (96.3) | ||
| College graduate | 71 | 55 (77.5) | 68 (95.8) | ||
| Age at interview | |||||
| <45 years | 49 | 41 (83.7) | 0.040 | 48 (98.0) | 0.095 |
| 45-59 years | 58 | 46 (79.3) | 57 (98.3) | ||
| ≥60 years | 61 | 41 (67.2) | 56 (91.8) | ||
| Time-period of diagnosis | |||||
| 1935–1964 | 49 | 31 (63.3) | 0.006 | 43 (87.8) | 0.003 |
| 1965–1984 | 68 | 60 (88.2) | 68 (100.0) | ||
| 1985 or later | 51 | 37 (72.5) | 50 (98.0) | ||
| Age at diagnosis | |||||
| <10 years | 56 | 40 (71.4) | 0.221 | 52 (92.9) | 0.263 |
| 10–19 years | 32 | 28 (87.5) | 32 (100.0) | ||
| ≥20 years | 80 | 60 (75.0) | 77 (96.2) | ||
| Major epilepsy syndrome | |||||
| Idiopathic generalized | 31 | 25 (80.6) | 0.196 | 31 (100.0) | 0.094 |
| Cryptogenic focal | 71 | 57 (80.3) | 67 (94.4) | ||
| Symptomatic | 38 | 29 (76.3) | 38 (100.0) | ||
| Unclassifiable | 28 | 17 (60.7) | 25 (89.3) | ||
| History of convulsive seizures | |||||
| Positive | 115 | 90 (78.3) | 0.353 | 114 (99.1) | 0.002 |
| Negative | 53 | 38 (71.7) | 47 (88.7) | ||
| Seizure-free in fifth year following diagnosisc | |||||
| No | 30 | 28 (93.3) | 0.016 | 29 (96.7) | 0.903 |
| Yes | 105 | 76 (72.4) | 101 (96.2) | ||
p-value from χ2 test (for age at interview, from χ2 test for trend).
Information on education was missing in one subject.
Restricted to 135 subjects followed for more than 5 years after diagnosis.
With regard to two measures of epilepsy severity, the findings differed between the two screen definitions. For the “epilepsy only” screen definition, sensitivity was similar in subjects with and without convulsive seizures, but was significantly lower in subjects who were seizure-free versus those who continued to have seizures in the fifth year after diagnosis. This pattern was reversed for the “any positive” screen definition; that is, sensitivity was significantly lower in subjects who had not had convulsive seizures than in those who had, but did not differ between subjects who were seizure-free versus those who continued to have seizures in the fifth year after diagnosis.
Given these associations of sensitivity with specific diagnostic features and the relatively low participation rate in the study, we examined the possibility of selection bias in our overall estimates of sensitivity. If the factors associated with a positive response to our screen questions differed between interviewed and noninterviewed subjects with epilepsy, we could have under- or overestimated sensitivity in our analyses. Among living subjects with epilepsy, we found no differences between those who were interviewed and those who were not (including subjects who refused, were not located, or were too ill or disabled to be interviewed) in the proportions either diagnosed before 1965 (30% vs. 27%, p = 0.65), with a history of convulsive seizures (32% vs. 28%, p = 0.30), or seizure-free in the fifth year after diagnosis (78% vs. 71%, p = 0.14). Interviewed subjects were older than those not interviewed (average 52 vs. 47, p < 0.001), possibly leading to some underestimation of sensitivity in the “epilepsy only” screen definition, since older subjects were less likely to screen positive.
Discussion
When all nine questions in our screening instrument were included, sensitivity was 96% and specificity (1-false positive rate) was 93%. Although this combination of sensitivity and specificity may appear to be very good, if lifetime prevalence of epilepsy is 2% we project a PPV of only about 23% in a population survey, suggesting that only about one of four screen-positive individuals would be confirmed to have epilepsy in the second stage of screening.
Inclusion of the question targeting febrile seizures in our screen added little to sensitivity except in patients with epilepsy who had prior febrile seizures. Inclusion of this question also raised the false-positive rate, and so did not improve the screening performance of our instrument overall. Similarly, the symptom-based questions improved sensitivity only slightly (from 91% to 94% when the febrile seizure question was excluded), at a cost of an increase in the false-positive rate. When only the epilepsy question was used for screening, however, sensitivity declined substantially (76% compared with 91% with “epilepsy or any seizure”), suggesting that the question about seizures under any circumstances was important for eliminating false negatives. On the other hand, use of the epilepsy question alone produced the highest PPV: We estimate that about two-thirds of screen positive individuals will have epilepsy with this question. These results are similar to those in a recent prevalence study in Washington Heights/Inwood and Central Harlem, New York, which used a screen modeled after the screen used here (Kelvin et al., 2007). PPV was much higher for the question asking directly about epilepsy or a seizure disorder (82%) than for the full set of questions (28%). Similarly, in another study assessing the accuracy of information from a questionnaire mailed to twins, seizure histories were verified in 82% of individuals who self-reported epilepsy but in many fewer of those who reported “other seizures” (35%) or “staring spells” (16%) (Corey et al., 2009).
The optimal choice for screening depends on the resources available and the objectives of a study. For estimation of prevalence, use of a screen with maximum sensitivity is extremely important to avoid underestimation. Many false positives will need to be evaluated in a second stage of screening, but this is unavoidable to ensure that case identification is as complete as possible. If the goal is to identify people likely to have epilepsy for further analysis [e.g., comorbidity studies (Kobau et al., 2006)], use of the epilepsy question alone may serve the purpose at minimal cost. However, our findings suggest that this approach introduces bias—people with epilepsy who screen positive are more likely than those who are missed to have recently diagnosed epilepsy or seizures that continued for five or more years after diagnosis (or perhaps more intractable epilepsy). With the broad screen definition (positive response to any screen question), those who screen positive are also more likely to have been diagnosed within 40 years before interview or to have had convulsive seizures, but this is not as problematic because sensitivity still approaches 90% in the subgroups with a lower likelihood of a positive screen.
Our study affords several advantages for the investigation of validity. First, the study is population-based. Cases ascertained from the population are more likely to have well-controlled or remitted epilepsy than are those ascertained in a clinic setting; thus we would expect that previous validation studies based in clinic populations would show inflated estimates of sensitivity compared with that expected in a population survey. The study in Ecuador confirms this expectation: Sensitivity of the same survey questions was 98% when evaluated in a clinic setting but only 79% when evaluated in a field setting (Placencia et al., 1992). In addition, the number of subjects included in our study is larger than in most previous validation studies, and extensive clinical detail was available through medical record review.
On the other hand, the unique features of the population we studied may limit the generalizability of our findings. Subjects included here were overwhelmingly white non-Hispanic, and unusually well educated (44% college graduates) compared with the general U.S. population (although we did not find that sensitivity was associated with education). Because of proximity to the Mayo Clinic and other healthcare providers included in the REP, the population has unusually good access to medical care and may be better informed about seizure disorders (and thus more likely to respond positively to questions in the screen) than would be expected in other settings. Moreover, the participation rate among eligible subjects was only 34%, and was greater in cases than in controls, raising the possibility that our estimates of sensitivity were inflated by participation of subjects who were more likely to recall their diagnoses and hence to respond positively to questions in the screen. On the other hand, clinical features associated with a positive response to our screening questions (Table 4) did not differ between living subjects with epilepsy who were and were not interviewed, providing some reassurance that the low response rate did not introduce substantial bias.
Incorporating a question in the screen about use of medications for seizures reduced false positive rates, but also reduced sensitivity so much that it was not very useful. However, we found that three of the eight seizure-free subjects who screened positive had benign positional vertigo, suggesting that adding a screening question designed to ascertain this differential diagnosis might improve screening performance.
Finally, we note that although one of our goals was to validate the screening instrument for use in genetic studies, all of the information assessed here was self-reported; that is, subjects answered the screening questions about their own medical histories rather than those of their relatives. The sensitivity of the screen for identifying affected family members in indirect, “family history” interviews in which subjects report of their relatives' seizure histories is likely to be lower than that reported here.
Acknowledgments
The study was supported by NIH grant R01 NS043472 (to RO). We thank Jane Emerson, RN, Melissa Petersen, RN, Diane Carlson, RN, Ann Van Oosten, and Thomas Bitz for assistance with data collection, and Keith Onken for programming of the CATI instrument. We are also grateful to the study participants, whose generous contribution of their time made this research possible.
Footnotes
We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Disclosure: Dr. Ottman was a consultant to Ortho-McNeil Janssen Scientific Affairs, LLC., and held stock options in and received consultant fees from Trigeminal Solutions, Inc. for serving on its scientific advisory board. Dr. Hauser was a consultant to Pfizer, Valiant Pharmaceuticals, Schwartz Biomedical, and Ovation Pharmaceuticals. Dr. Buchhalter served on the speaker's bureau for UCB, and is a site investigator for an anticonvulsant trial sponsored by Ovation Pharmaceuticals. None of the other authors has any conflict of interest to disclose.
References
- Borges MA, Min LL, Guerreiro CA, Yacubian EM, Cordeiro JA, Tognola WA, Borges AP, Zanetta DM. Urban prevalence of epilepsy: populational study in Sao Jose do Rio Preto, a mediumsized city in Brazil. Arq Neuropsiquiatr. 2004;62:199–204. doi: 10.1590/s0004-282x2004000200002. [DOI] [PubMed] [Google Scholar]
- Corey LA, Kjeldsen MJ, Solaas MH, Nakken KO, Friis ML, Pellock JM. The accuracy of self-reported history of seizures in Danish, Norwegian and U.S. twins. Epilepsy Res. 2009;84:1–5. doi: 10.1016/j.eplepsyres.2008.11.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Haerer AF, Anderson DW, Schoenberg BS. Survey of major neurologic disorders in a biracial United States population: the Copiah County Study. South Med J. 1987;80:339–343. doi: 10.1097/00007611-198703000-00016. [DOI] [PubMed] [Google Scholar]
- Hauser WA, Annegers JF, Kurland LT. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935-1984. Epilepsia. 1993;34:453–468. doi: 10.1111/j.1528-1157.1993.tb02586.x. [DOI] [PubMed] [Google Scholar]
- Hauser WA, Annegers JF, Rocca WA. Descriptive epidemiology of epilepsy: contributions of population-based studies from Rochester, Minnesota. Mayo Clin Proc. 1996;71:576–586. doi: 10.4065/71.6.576. [DOI] [PubMed] [Google Scholar]
- Kelvin EA, Hesdorffer DC, Bagiella E, Andrews H, Pedley TA, Shih TT, Leary L, Thurman DJ, Hauser WA. Prevalence of self-reported epilepsy in a multiracial and multiethnic community in New York City. Epilepsy Res. 2007;77:141–150. doi: 10.1016/j.eplepsyres.2007.09.012. [DOI] [PubMed] [Google Scholar]
- Kobau R, Gilliam F, Thurman DJ. Prevalence of self-reported epilepsy or seizure disorder and its associations with self-reported depression and anxiety: results from the 2004 HealthStyles Survey. Epilepsia. 2006;47:1915–1921. doi: 10.1111/j.1528-1167.2006.00612.x. [DOI] [PubMed] [Google Scholar]
- Melcon MO, Kochen S, Vergara RH. Prevalence and clinical features of epilepsy in Argentina. A community-based study. Neuroepidemiology. 2007;28:8–15. doi: 10.1159/000097850. [DOI] [PubMed] [Google Scholar]
- Melton LJ., III History of the Rochester Epidemiology Project. Mayo Clin Proc. 1996;71:266–274. doi: 10.4065/71.3.266. [DOI] [PubMed] [Google Scholar]
- Meneghini F, Rocca WA, Anderson DW, Grigoletto F, Morgante L, Reggio A, Savettieri G, Di Perri R. Validating screening instruments for neuroepidemiologic surveys: experience in Sicily. Sicilian Neuro-Epidemiologic Study (SNES) Group. J Clin Epidemiol. 1992;45:319–331. doi: 10.1016/0895-4356(92)90033-j. [DOI] [PubMed] [Google Scholar]
- Nicoletti A, Reggio A, Bartoloni A, Failla G, Sofia V, Bartalesi F, Roselli M, Gamboa H, Salazar E, Osinaga R, Paradisi F, Tempera G, Dumas M, Hall AJ. Prevalence of epilepsy in rural Bolivia: a door-to-door survey. Neurology. 1999;53:2064–2069. doi: 10.1212/wnl.53.9.2064. [DOI] [PubMed] [Google Scholar]
- Noronha AL, Borges MA, Marques LH, Zanetta DM, Fernandes PT, de Boer H, Espindola J, Miranda CT, Prilipko L, Bell GS, Sander JW, Li LM. Prevalence and pattern of epilepsy treatment in different socioeconomic classes in Brazil. Epilepsia. 2007;48:880–885. doi: 10.1111/j.1528-1167.2006.00974.x. [DOI] [PubMed] [Google Scholar]
- Osuntokun BO, Schoenberg BS, Nottidge VA, Adeuja A, Kale O, Adeyefa A, Bademosi O, Olumide A, Oyediran ABO, Pearson CA, Bolis CL. Research protocol for measuring the prevalence of neurologic disorders in developing countries. Neuroepidemiology. 1982;1:143–153. [Google Scholar]
- Ottman R, Hauser WA, Stallone L. Semistructured interview for seizure classification: agreement with physicians' diagnoses. Epilepsia. 1990;31:110–115. doi: 10.1111/j.1528-1157.1990.tb05368.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ottman R, Lee JH, Hauser WA, Hong S, Hesdorffer D, Schupf N, Pedley TA, Scheuer ML. Reliability of seizure classification using a semistructured interview. Neurology. 1993;43:2526–2530. doi: 10.1212/wnl.43.12.2526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Placencia M, Sander JW, Shorvon SD, Ellison RH, Cascante SM. Validation of a screening questionnaire for the detection of epileptic seizures in epidemiological studies. Brain. 1992;115(Pt 3):783–794. doi: 10.1093/brain/115.3.783. [DOI] [PubMed] [Google Scholar]
- Schoenberg BS. Recent studies of the epidemiology of epilepsy in developing countries: a coordinated program for prevention and control. Epilepsia. 1987;28:721–722. doi: 10.1111/j.1528-1157.1987.tb03707.x. [DOI] [PubMed] [Google Scholar]