The prevalence and demographic distribution of treated epilepsy: a communitybased study in tasmania, australia

Acta Neurol Scand 2012: 125: 96–104 DOI: 10.1111/j.1600-0404.2011.01499.x The prevalence and demographicdistribution of treated epilepsy:a community-based study in Tasmania,Australia DÕSouza WJ, Quinn SJ, Fryer JL, Taylor BV, Ficker DM, OÕBrien TJ, Pearce N, Cook MJ. The prevalence and demographic distribution of treated epilepsy: a community-based study in Tasmania, Australia.
Acta Neurol Scand: 2012: 125: 96–104.
Ó 2011 John Wiley & Sons A ⁄ S.
1The Menzies Research Institute, The University ofTasmania, Hobart, Tasmania; 2The Department of Objectives – To estimate the prevalence and demographic distribution Medicine, St Vincents Hospital, The University of of treated epilepsy in a community-based population. Materials & Melbourne, Melbourne, Vic., Australia; 3Flinders Clinical methods – We surveyed all residents in Tasmania, Australia, who were Effectiveness, Flinders University Adelaide, SA, supplied at least one antiepileptic drug prescription between July 1, Australia; 4The University of Cincinnati Academic Health 2001 and June 30, 2002, recorded on the national prescription Centre, Department of Neurology, Cincinnati, OH, USA; database. We adjusted for the effect of disease-related non-response 5The Department of Medicine, The Royal Melbourne bias by imputation methods. Results – After three mail contacts, Hospital, The University of Melbourne, Melbourne, Vic., 54.0% (4072 ⁄ 7541) responded, with 1774 (43.6%) indicating treatment Australia; 6Centre for Public Health Research, MasseyUniversity Wellington Campus, Wellington, New for epilepsy, representing 86.0% of the estimated total possible cases in Zealand; 7Department of Medical Statistics, London Tasmania. The adjusted treated epilepsy prevalence was 4.36 per 1000 School of Hygiene and Tropical Medicine, London, UK (95% CI 4.34, 4.39); lower in women (prevalence ratio 0.92 (95% CI0.84, 1.00)); greater with increasing age (P < 0.001); similar in the Key words: antiepileptic drugs; epidemiology; three main geographic regions; and similar with socioeconomic status epidemiological methods; epilepsy; prevalence of postcode of residence. Conclusions – Although our estimates are W. DÕSouza, The Department of Neurology & Neuro- likely to be affected by access to health services, overall treated epilepsy logical Research, 5th Floor, The Daly Wing, St VincentÕs prevalence of 4.4 per 1000 is similar to previous studies. Our finding of Hospital, The University of Melbourne, Melbourne, PO high elderly prevalence has been reported in a few recent studies in Box 2900, Fitzroy 3065, Melbourne, Vic., Australia developed countries and has important clinical and public health implications in populations with similar aging demographics.
Accepted for publication January 20, 2011 extends from India, in the west, to the Pacific Islands in the east. Despite its great geographic and Studying the distribution and determinants of cultural diversity, the lifetime prevalence in this epilepsy in population-based samples has the region is between 3.4 and 7.5 per 1000 (2, 3), a potential to improve our understanding of this similar range to that seen in the developed world, common disorder (1). Such comparative studies with treated epilepsy having a prevalence of about are important as they provide information on the 4–5 per 1000 (2, 4). However, with few exceptions, burden of the condition, both at the population most prevalence studies have had insufficient power level and in specific subpopulations, and may to include analyses by age, gender, region (5, 6), or also indicate etiological factors acting at these In Australia, the national prescription database The International League Against Epilepsy appears representative of community-treated epi- divides the world into six regions comprising: lepsy and can provide an effective and efficient method for large-scale patient recruitment for European, Latin America, and North America.
epidemiological research (9). Utilizing this data- The Asia–Oceania region, which includes Australia, base, we estimated the prevalence of treated Prevalence & distribution of treated epilepsy in Tasmania epilepsy in Tasmania and its distribution by age- netÕ arrangements limit the total annual contribu- group, gender, region, and socioeconomic status.
tion that a family can make toward prescriptioncosts for each of these categories of patient. Oncethese limits are reached, any PBS ⁄ RPBS prescrip- tions dispensed are either free or with a much The design and methods of patient recruitment reduced co-payment for the remainder of the safety have been described previously in detail (9) and net period (11). These arrangements have implica- will only be outlined briefly here.
tions on the PBS ⁄ RPBS data set. When a patientpays the entire cost of the medication, there is noHIC record of the prescription. Prescription records for drugs costing less than the general Tasmania is an island state off the southeastern patient co-payment will not be complete (only coast of Australia comprising an area of 68,000 recorded for concessional beneficiaries and those square km, with a population of 472,672 in 2002 who have reached safety net entitlements). There (10). The state is divided into three main geo- will be complete capture for more expensive drugs, graphic, administrative, and postcode regions: as the government would have made a contribution southern (231,662), northern (134,701), and north- in every case. Only five of these anticonvulsants (phenobarbitone, phenytoin suspension, carba-mazepine liquid, carbamazepine 50 mg, and lamo-trigine 5 mg) cost <$22.40, and so had limited Pharmaceutical benefits scheme and the Australian national capture only for ÔGeneralÕ beneficiaries. However, barbiturates are infrequently prescribed in Tasma- In Australia, consultative and investigational nia (2.3% of all antiepileptic drugsÕ (AED) pre- health services are significantly funded by the scriptions – data not shown), liquid preparations federal government, allowing universal mean-inde- are prescribed for patients with swallowing diffi- pendent access to health services. National pre- culties and children typically <8 years of age, and scription data records are generated when the low-dose tablet preparations are usually used in government contributes to the cost of a product dose titration (rather than dose end-point). With dispensed under the Australian Pharmaceutical less ÔgeneralÕ patients in Tasmania than other Benefits Scheme (PBS) and Repatriation Pharma- Australian states, the HIC sample frame is ceutical Benefits Scheme (RPBS). The PBS ⁄ RPBS expected to be a comprehensive AED treatment is a subsidization program monitored by the database for all ÔconcessionÕ card holders and, with Health Insurance Commission (HIC). Patients are these few exceptions, the majority of ÔgeneralÕ classified into one of two categories, determining patients (Table 1). Benzodiazepines and acetazol- the amount the patient contributes and the amount amide prescriptions were excluded from selection of subsidy paid by the government. General because they were more likely to be prescribed as beneficiaries make a maximum patient contribu- single agents for non-epilepsy conditions.
tion (A $22.40 in 2002) per prescription item;concessional beneficiaries (primarily social security recipients) or veteran affairs (returned servicemenand women) categories purchase drugs at a con- An invitation letter asked potential participants to cession rate (A $3.60 in 2002). Additional Ôsafety agree to participate in a longitudinal disease cohort Table 1 Response to the health insurance commission mail invitations Do not have blank spells, seizures, epilepsy Do not know whether have blank spells, seizures, or epilepsy Did not specify whether have blank spells, seizures, or epilepsy – The Tasmanian Epilepsy Register (TER) and mate, and Vigabatrin. During the study period, asked ÔAre you taking antiepileptic medications for blank spells, seizures or epilepsy?Õ(9). Mailout one Rufinamide, Locosamide, and Zonisamide were occurred on November 20, 2002. A second mail not available in Australia. Patients had to be alive invitation was sent on February 19, 2003 to those and have a postcode listed in Tasmania at some- invited patients who had failed to respond, except time during this time frame and to indicate that those where a Ôreturn to senderÕ notification had AEDs were prescribed for epilepsy. Respondents been received or we had received information that were classified as having epilepsy when they self- they were deceased. A third anonymous HIC mail affirmed to Ôhave blank spells seizures or epilepsyÕ contact was sent to AED prescribers of non- or when their treating doctor affirmed they Ôhad responders who did not respond to the second been prescribed anticonvulsant medications for mailout, in February 2004. This letter requested epilepsyÕ. Mail respondents who answered ÔdonÕt the treating doctor to provide the indication for knowÕ or ÔdidnÕt specifyÕ were classified as disease AED treatment and to advocate participation onto non-responders for imputation correction.
To correct observed prevalence estimates for disease non-response bias, the method proposed byDrane was used to determine imputed prevalence (13). By assuming an exponential decay in disease As an indicator of socioeconomic status, each response rate with each subsequent contact, the patientÕs postcode of residence was scored into one true prevalence can be estimated by fitting a of five ordered categories according to the Socio- regression model to the observed disease response economic Index for Area (SEIFA) constructed by rate at each mail contact and summing over all the Australian Bureau of Statistics (using the extrapolated mail contacts (Fig. 1). Confidence subcategory for Relative Advantage / Disadvan- intervals for each imputed Tasmanian demo- graphic prevalence estimate (age-group, gender,region, and socioeconomic status) were obtainedby converting the corresponding confidence inter- val endpoints of each imputed sample estimate into Approval was obtained from the southern Tasma- nia Health and Medical Human Research EthicsCommittee, the HIC, and the Department of Vet- Validation study – Validation of the question Ôhave eransÕ Affairs Human Research Ethics Committee.
blank spells seizures or epilepsyÕ to confirm theoverall (but not strata specific) false positive rate of Representativeness of sample – To assess the repre- sentativeness of our final participant cohort, wecompared this cohortÕs demographic features for differences with that of the Tasmanian population,mail responders, and non-responders with a chi- square test for differences between proportions. A test for trend in the proportion responding across levels of the demographic characteristic [age, socio-economic status (SES)] was also conducted by fitting a univariable log binomial model and fittingthe characteristic as a linear predictor (9).
Period prevalence of treated epilepsy – Prevalent cases of epilepsy were identified from individualssupplied at least one prescription for an AED in Figure 1. Captured and imputed cases used in estimating epi-lepsy prevalence in Tasmania. *For the purposes of disease Tasmania above the ÔreportableÕ retail price thresh- response, mail respondents were classified as epilepsy cases old during the twelve-month period from July 1, when they affirmed to Ôhave blank spells seizures or epilepsyÕ 2001 to June 30, 2002: Carbamazepine, Ethosuxi- and classified as not being epilepsy case when they affirmed to mide, Gabapentin, Lamotrigine, Methylphenobar- ÔdonÕt have blank spells, seizures or epilepsyÕ. Epilepsy preva-lence was imputed from total disease non-respondents, which bitone, Phenobarbitone, Phenytoin, Primidone, comprise mail respondents who ÔdidnÕt knowÕ, ÔdidnÕt specify,Õ Sodium Valproate, Sulthiame, Tiagibine, Topira- Prevalence & distribution of treated epilepsy in Tasmania this question against a neurologist diagnosis of with increasing age (trend P < 0.001), or when epilepsy was undertaken in a random sample of from a higher socioeconomic quintile (linear trend 293 mail respondents using a validated modified P < 0.001) with over-representation if female diagnostic questionnaire administered by a final- almost reaching significance (P = 0.053). In addi- year trainee neurologist with the responses inter- tion, patients taking anticonvulsants were more preted by an epilepsy specialist using standardized likely to respond to our invitation if their prescrip- diagnostic guidelines (14). For these 293 respon- tion was obtained from a neurologist and less likely dents: none reported they were taking AEDs for to respond if their prescription was obtained from indications other than epilepsy; epilepsy status remained uncertain after interview in 4 (1.4%); a Table 1 shows the breakdown of response to the false positive diagnosis of epilepsy was reached in three HIC mail invitations. There was evidence of a 23 (7.8%) (one single seizure, three migraine, nine diminishing response rate with response time. The psychogenic, and 10 syncope); and a true positive 1774 indicating treatment for epilepsy (=captured diagnosis was confirmed in 266 (90.8%). Including cases) and 289 (=imputed cases) giving an overall the ÔuncertainÕ category as false positives gave a estimated prevalence of 2063 cases in Tasmania positive predictive value (PPV, the number of cases found correctly to have the disease by the gold Table 2 shows the estimated prevalence of standard) of 0.91, and PPV was 0.92 when these treated epilepsy in Tasmania by twenty-year age- four uncertain cases were excluded. It was not group, gender, region, and SEIFA. The crude possible to estimate the false negative and true prevalence of epilepsy was estimated to be 4.36 per negative rate of self-identification of epilepsy for 1000 (2063 ⁄ 472,672 · 1000) persons in the Tasma- this question (i.e., respondents who answered ÔnoÕ, nian population with 95% confidence interval (CI) ÔdonÕt know,Õ or ÔdidnÕt specifyÕ) for privacy 4.34–4.39. The Tasmanian figures are shown, as age–sex standardization to the World StandardPopulation or Australian Standard Population did Associations between demographic characteristics and not alter our results appreciably. Adjusted treated prevalence of treated epilepsy – Assuming an expo- epilepsy prevalence was 8% lower (95% CI 6– nential decay in response rate with response time 38%) in women than men, similar in the three within each demographic subgroup, the distribu- geographic regions and not associated with SEIFA tion of treated epilepsy by 5-year age-group, (P = 0.50). We examined the association between gender, region, and socioeconomic status by age and prevalence of epilepsy using fractional SEIFA quintiles were estimated. We calculated polynomials and found that the data were best univariate prevalence ratios to compare differences fitted to a log binomial model by transforming age within variable categories by fitting a log binomial to 1 ⁄ Öage. Fig. 2 displays the nonlinear relation- model to test for trend in prevalence for age and ship between prevalence of epilepsy and increasing SIEFA, and whether prevalence is associated with region and gender, after adjusting for the increased Finally, Table 3 shows the observed percentage confidence intervals obtained using DraneÕs impu- and imputed prevalence of patients prescribed tation method. Only a univariate analysis was concurrent AED among those responding to our conducted as our final estimates for total preva- mail invitation indicating treatment for Ôblank lence and its distribution by age, gender, region, spells, seizures, or epilepsyÕ. More than two con- and SES yielded imputed rather than actual cases current AED for epilepsy were taken by 142 (8.0%) not allowing a multivariate approach. All analyses were performed using Stata Version 9.
responding to our survey and indicating treatment There are limited studies describing epilepsy prev- for epilepsy, the percentage and estimated preva- alence patterns by gender, age-group, region, or lence of patients prescribed concurrent AED was socioeconomic status. Although our estimates are influenced by factors that affect a person acquiringa diagnostic label for treatment purposes, caseascertainment by AED prescription provides an efficient recruitment method for epidemiological A total of 4072 persons responded to the survey, studies in communities characterized by high giving an overall response rate of 54.0%. Patients access to health services (2). The estimate of 4.4 taking anticonvulsant were more likely to respond per 1000 for overall treated prevalence is consistent Table 2 Estimated prevalence* of treated epilepsy in Tasmania by age-group, gender, region, and SEIFA  *Imputed prevalence reported. Age–sex standardization to the World Standard or Australian Standard Population did not alter my results appreciably and so the Tasmanianfigures are represented. For the purposes of disease response, mail respondents were classified as having epilepsy when they affirmed to Ôhave blank spells seizures orepilepsyÕ and classified as not having epilepsy when they affirmed to ÔdonÕt have blank spells, seizures or epilepsyÕ. Disease prevalence was imputed from total disease non-respondents, which comprise mail respondents who ÔdidnÕt knowÕ, ÔdidnÕt specify,Õ and all mail non-respondents.  As an indicator of socioeconomic status, each patientÕspostcode of residence was scored into one of five ordered categories according to the Socioeconomic Index of Relative Advantage ⁄ Disadvantage (SEIFA) constructed by theAustralian Bureau of Statistics. à472, 672 (38). §Per 1000. –Test for linear trend using a log binomial model.
Table 3 Observed percentage and estimated prevalence of epilepsy treated withconcurrent antiepileptic drug medications in Tasmania between July 1, 2001 andJune 30, 2002 *Imputed prevalence reported;  472, 672 (38); àPer 1000. AED, antiepileptic drugs.
Figure 2. Prevalence* of treated epilepsy in Tasmania by 5-year age-groups. Age-group 0–4 years is excluded because of public health implications in countries with similar small numbers. *Imputed prevalence reported. For the pur- poses of disease response, mail respondents were classified as Lack of community-based sampling has been a having epilepsy when they affirmed to Ôhave blank spells sei- major criticism of most case ascertainment meth- zures or epilepsyÕ and classified as not having epilepsy whenthey affirmed to ÔdonÕt have blank spells, seizures or epilepsyÕ.
ods in previous epidemiological research into Disease prevalence was imputed from total disease non- epilepsy (15). Previous studies have suggested respondents, which comprise mail respondents who ÔdidnÕt that the medical practitioner writing AED pre- knowÕ, ÔdidnÕt specify,Õ and all mail non-respondents. Imputed scriptions is also most likely to be responsible for prevalence reported with smoother and 95% confidence inter-vals was also shown. The smoother and 95% confidence disease supervision and follow-up (4, 16). If this is interval was obtained from a log binomial regression model true, with 70.9% of patients with epilepsy receiving their AED prescriptions exclusively from theirgeneral practitioner and 19.1% receiving them in with similar studies, while our observation of high part from a medical specialist in the preceding elderly prevalence has been suggested by a few twelve months, it suggests that our cohort repre- recent reports, and has important clinical and sents community-treated disease (9).
Prevalence & distribution of treated epilepsy in Tasmania However, to be an ideal recruitment approach further act to lower our prevalence estimates (25).
for prevalence estimation, AED prescription pen- Although it is uncertain whether this form of etration for epilepsy treatment should have uni- measurement bias would be overcome by complete versality and validity (17). We acknowledge that participation, a higher response rate would have allowed us more confidence in the precision and lower prevalence estimates. Our inclusion of accuracy of our prevalence estimates (26).
ÔdonÕt knowÕ or ÔdidnÕt specifyÕ responders as Nevertheless, in a number of communities, disease non-responders for imputation is an AEDs have been demonstrated to have widespread attempt to at least correct for this possibility use and penetration in treating epilepsy, making among survey non-responders. We are also aware them potentially a good target for identifying cases that AED initiation and withdrawal practices for clinical epidemiological research (27). Every among local physicians and patients may affect person identified as having epilepsy in a door-to- prevalence estimates from prescription data (18).
door community survey in Australia had been In Australia, the typical practice is to commence prescribed AEDs at some time in their lifetime, AED treatment after the second unprovoked with virtually all of the one-third Ôoff medication seizure, and this is reflected in the low number on survey dayÕ seizure free in the preceding year of patients (0.34%) found to have been treated (2). Although this Australian study suggests no after a single seizure in our validation study, a lifetime AED treatment gap for epilepsy, it was similar figure found in a Swedish study (19).
based on Ôdoctor-diagnosed epilepsyÕ making it Therefore, patient recruitment by AED prescrip- likely to also be an underestimate. It is difficult to tion appears to have yielded few false positive determine the exact magnitude of our underesti- cases because of ÔearlyÕ treatment.
mate. We suspect that the overall situation in A potential limitation of the estimation of Australia is more likely to resemble countries with prevalence from studies using AEDs prescriptions similar universal primary and tertiary health care is that it may overestimate the number of epilepsy access such as Sweden or the UK, where AED cases because these medications are sometimes underascertainment of 8–25% has been observed prescribed for misdiagnosed epilepsy (9, 19) or for other diseases such as a migraine, depression, Non-response is the rule rather than the excep- mood stabilization, and chronic pain conditions tion in epidemiological surveys that generally (20). Therefore, we used the response to the worldwide has been increasing (29). Imputation is question ÔAre you taking antiepileptic medications one mechanism to correct for unmeasured disease for blank spells, seizures or epilepsy?Õ to confirm in survey non-responders. The theoretical basis for epilepsy status among mail respondents. Although this methodology does not require independent this is unlikely to be a sensitive screening tool for population sampling and comes from Little and epilepsy in household surveys, our aim was to Rubin (26) and Hansen et al. (30), which allows obtain a high PPV (0.92) for capturing patients imputation correction from two or three mailouts with epilepsy. We did not adjust our overall from a diminishing cohort of non-responders from prevalence estimate using this measured false the same initial population sample, i.e., each positive rate (7.8%) as it was similar to other subsequent strata is dependent on the previous studies presenting unadjusted estimates (21). We (31). The benefits of non-response imputation are, therefore expect our AED estimate to be a useful first it improves the accuracy and precision of measure of clinically active epilepsy prevalence prevalence estimates (26). Secondly, it takes advan- tage of typical human behavior previously shown Underestimating prevalence is the more likely to be present in conditions other than epilepsy, e.g., consequence of prevalence measures based solely asthma (32) where disease respondents tend to on AED utilization, with the lowest reported rates reply earlier allowing for a functional connection usually arising from developing countries (23) or between mail contact and changing disease preva- indigenous groups (24) where access to health lence over time. Thirdly, it is relatively easy to services result in few patients on regular treatment.
perform simply by designing surveys with either a Therefore, although demonstrated to be an effi- two or three mail contact at regular intervals.
cient mechanism for generating community-treated Finally, although derived from published theory cases, our ascertainment method utilizing the HIC (13, 33), it is also consistent with others and our prescription database almost certainly underesti- current study observations confirming an exponen- mates the true prevalence of epilepsy as it fails to tial fall in epilepsy prevalence among non-respond- sample undiagnosed and untreated disease. We ers with resistance to mail response primarily from also acknowledge that disease concealment will patients receiving AEDs for other conditions. Its main criticism is that it is methodologically vul- Our final cohort demonstrated similar trends nerable where there is evidence of Ôhard coreÕ with older age, female, and higher socioeconomic disease non-responders (34) as it is based on the groups that have been previously observed from assumption that disease responders are more mail surveys in other disease groups (41). Although sensitive to replying earlier rather than resisting these responder differences have the potential to reply because of disease concealment. In the case of reduce our prevalence estimates in the lower people with epilepsy, although we know disease socioeconomic groups and inflate our prevalence concealment occurs (25, 35), it is largely unknown estimates in the higher age-groups, they are whether this is a temporary or permanent state, unlikely to be the major explanation for the i.e., affected by time to response. Reassuringly, our differences observed in this study. Our estimated observations appear not to support this.
treated epilepsy prevalence of 4.36 per 1000 is We acknowledge our third mail invitation to consistent with a number of previous studies that AED prescribers of all non-responders to our have derived prevalence from AED prescription second mail invitation yielded a relatively poor (22), primary care (28), or household survey 18.0% (697 ⁄ 3871) response; however, it allowed us sources (2). Prevalence was 8% lower in women to confirm a number of important assumptions.
than men. While a few studies have found higher First, the poor physician response although disap- epilepsy prevalence in women than men (42), and pointing was not unexpected and justified our some have found no gender differences (6), the design to initially bypass physicians for recruitment majority are consistent with our findings, with and confirmed the universal difficulties in engaging greater prevalence in men compared to women busy primary physicians in non-core research (43), at all age-groups (24). The explanations for activities (36). Secondly, it also confirmed resis- tance to mail response was occurring dispropor- tionately among those taking AED for other At present, our knowledge of socioeconomic indications (48.5%, i.e., 338 ⁄ 697) rather than differences in epilepsy prevalence is limited and epilepsy (38.3%, i.e., 267 ⁄ 698). Finally, most conflicting (7, 8). The SEIFA index utilized in this importantly, it confirmed a further observed study is comprised of variables relating to income, decline in disease prevalence with time, allowing education, occupation, wealth, and living condi- prediction of the unobserved disease in non- tions with the highest weighting being given to the responders using the methods of Drane (13, 33).
first three variables (12). As SEIFA reflects the Given the high PPV from our validity study, socioeconomic well-being of an area, rather than of an individual, it is possible for a relatively advan- between self-reported and doctor-reported epilepsy taged person to be resident in an area of low index status was assumed (37), allowing us to combine score. SEIFA also has a varying impact on health classification sources in a three contact exponential service uptake, and potentially AED prescription decay rather than two contact linear prevalence treatment, with SEIFA 1 having higher general practice uptake in metropolitan and lower in rural Useful insights into epilepsy treatment in Tasma- Australian settings (44). Therefore, one explana- nia are obtained through the HIC AED prescription tion of our finding of no association in treated data. At 8.0%, the percentage of patients receiving epilepsy prevalence and SEIFA is that it is real and more than two concurrent AEDs for epilepsy does not reflect access to general practice services.
responding to our mail survey is at the high end of Alternatively, it may be as a result of misclassifi- polytherapy seen in most other developed countries cation of individual socioeconomic status or con- (19, 38). Although not always possible to avoid, AED polytherapy is currently only recommended In our study, no differences in treated epilepsy for inadequately controlled epilepsy when patients prevalence were found between the three main fail serial monotherapy and dual therapy (39).
geographic regions of Tasmania. Higher urban Previous studies have suggested that significantly prevalence has been noted in two previous studies more patients prescribed AEDs, or, when diagnosed (27, 45), but more commonly, studies have found and managed by doctors other than neurologists or higher prevalence in rural compared to urban pediatricians, are on multiple AEDs compared with settings (5, 46, 47). Although it has been speculated those treated by private specialists or hospital that these contrary findings may be attributed to doctors (4, 40). Therefore, the relatively high greater access to health services in the urban setting percentage of patients on multiple AED may either at the treatment (27) or primary prevention level reflect increased prevalence of severe epilepsy (27) or (47), no clear etiological reasons for these differ- inadequate epilepsy management in Tasmania.
Prevalence & distribution of treated epilepsy in Tasmania Epilepsy is often considered a disease of younger 4. Lambie DG, Johnson RH, Stanaway L. Prescribing patterns age-groups, and the highest prevalence rates occur for epilepsy. NZ Med J 1981;94:15–9.
in children in a number of developing countries 5. Placencia M, Shorvon SD, Paredes V et al. Epileptic sei- zures in an Andean Region of Ecuador. Brain 1992; (46). In earlier studies, this higher childhood pattern was also seen in developed countries (48).
6. CDC. Leads from the morbidity and mortality weekly re- However, recent studies have reported a reversed port: prevalence of self-reported epilepsy – United States pattern with prevalence rates being lowest in children <14 years (22, 43) and highest in elderly 7. Pond DA, Bidwell BH, Stein L. A survey of epilepsy in fourteen general practices. Psychiatr Neurol Neurochir people over 65 years of age (22, 28) in developed countries. Consistent with these latter studies, we 8. Cornaggia CM, Canevini MP, Christe W et al. Epidemi- also found the lowest prevalence rates in children ologic survey of epilepsy among army draftees in Lomb- and highest in the elderly. Although lower sam- ardy, Italy. Epilepsia 1990;31:27–32.
pling of liquid AED preparations in children 9. DÕSouza WJ, Fryer J, Quinn S et al. The Tasmanian epi- lepsy register – a community-based cohort: background <8 years probably partly explains the lower prev- and methodology for patient recruitment from the Aus- alence rates seen in this age, they persist beyond the tralian national prescription database. Neuroepidemiology age that these issues would continue to have an impact. Our results suggest that incidence may be 10. Australian Bureau of Statistics. Regional population growth, Australia and New Zealand, 2002–03. Canberra: lower in the younger age-group, or that the illness Australian Bureau of Statistics, 2003. Report No.: cat. no.
treatment duration is more short-lived, compared to adults and the elderly (49). If true, with 11. Edmonds DJ, Dumbrell DM, Primrose JG, Mcmanus P, demographers predicting a dramatically greater Birkett DJ, Demirian V. Development of an Australian elderly population in the future (50), these findings Drug Utilisation Database. PharmacoEconomics. 1993;3: have important implications for health service 12. Trewin D. Information paper: census of population and housing socio-economic indexes for areas (SEIFA). 2001.
ABS Catalogue No. 2039.0 ISBN 0 642 47936 4.
13. Drane JW, Richter D, Stoskopf C. Improved imputation of non-responses to mailback questionnaires. Stat Med Wendyl DÕSouza was supported from a Pfizer Alfred & St VincentÕs Hospital Electrophysiology Fellowship, St VincentÕs 14. DÕSouza WJ, Stankovich J, Bower S et al.
Hospital Neuro-epidemiology Fellowship and FRACP GSK computer-assisted-telephone interviewing to diagnose sei- Fellowship in Neurology. This project was generously sup- zures, epilepsy and idiopathic generalised epilepsy. Epi- ported by grants from The Booth Estate Launceston, Royal Hobart Hospital Research Foundation, GSK Neurology, 15. Sander JWAS, Shorvon SD. Epidemiology of the epilep- Clifford Craig Medical Research Trust – North West Tasmania sies. J Neurol Neurosurg Psychiatry 1996;61:433–43.
and Menzies Research Institute NHMRC Capacity Building 16. Rutgers MJ. Epilepsy in general practice: the Dutch situ- Grant. The Centre for Public Health Research is supported by a programme Grant from the Health Research Council of New 17. Kelsey JL, Thompson WD, Evans AS. Methods in obser- Zealand. We thank the Tasmanian Regional Divisions of vation epidemiology. New York: Oxford University Press, General Practice, Specialist Physicians of Tasmania, The Epilepsy Association of Tasmania, the Pharmaceutical Guild 18. Oun A, Haldre S, Magi M. Prevalence of adult epilepsy in of Tasmania, the Tasmanian Branch of the Pharmacist Society Estonia. Epilepsy Res 2003;52:233–42.
of Australia, and the Society of Hospital Pharmacists. We also 19. Forsgren L. Prevalence of epilepsy in adults in Northern thank our research assistants Nicola Mulcahy, Leanne Barnes, and Charlotte McKercher from the Menzies Research Institute 20. Lammers MW, Hekster YA, Keyser A, Meinardi H, Renier for their considerable efforts in participant liaison, data WO, Herings RMC. Use of antiepileptic drugs in a com- munity-dwelling Dutch population. Neurology 1996;46:62–7.
21. Maremmani C, Rossi G, Bonuccelli U, Murri L. Descriptive epidemiologic study of epilepsy syndromes in a district ofNorthwest Tuscany, Italy. Epilepsia 1991;32:294–8.
The authors report no conflicts of interest.
22. Wallace H, Shorvon SD, Tallis R. Age-specific incidence and prevalence rates of treated epilepsy in an unselected population of 2,052,922 and age-specific fertility rates ofwomen with epilepsy. Lancet 1998;352:1970–3.
1. Rose G. The strategy of preventive medicine. New York: 23. Coleman R, Loppy L, Walraven G. The treatment gap and primary health care for people with epilepsy in rural 2. Beran RB, Hall L, Pesch A et al. Population prevalence of Gambia. Bull World Health Organ 2002;80:378–83.
epilepsy in Sydney, Australia. Neuroepidemiology 1982;1: 24. Haerer AF, Anderson DW, Schoenberg BS. Prevalence and clinical features of epilepsy in a biracial United States 3. Stanhope JM, Brody JA, Brink E. Convulsions among the population. Epilepsia 1986;27:66–75.
Chamorro people of Guam, Mariana Islands: I seizure 25. Beran RB, Michelazzi J, Hall L, Tsimnadis P, Loh S. False- disorders. Am J Epidemiol 1972;95:292–8.
negative response rate in epidemiologic studies to define prevalence ratios of epilepsy. Neuroepidemiology 1985;4: by a survey for prescriptions of antiepileptic drugs: epi- demiology and patterns of care. Acta Neurol Scand 26. Little RJA, Rubin DB. Statistical analysis with missing data. New York: John Wiley & Sons, Inc. 1987.
39. French J. The long-term therapeutic management of epi- 27. Olafsson E, Hauser WA. Prevalence of epilepsy in rural lepsy [review]. Ann Intern Med 1994;120:411–22.
Iceland: a population-based study. Epilepsia 1999;40: 40. Giuliani G, Senigaglia AR, Scatanglini F, De Rosa M.
Drugs as epidemiological ‘‘tracers’’ for epilepsy: I First 28. Cockerell OC, Eckle I, Goodridge DMG, Sander JWAS, estimates of the prevalence of the disease in the Italian Shorvon SD. Epilepsy in a population of 6000 re-examined: NHS model of drug-use register. Boll Lega Ital Epil secular trends in first attendance rates, prevalence, and prognosis. J Neurol Neurosurg Psychiatry 1995;58:570–6.
41. Groves RM, Dillman DA, Eltinge JL, Little RJA. Survey 29. Locker D, Wiggins R, Sittampalam Y, Patrick DL. Esti- mating the prevalence of disability in the community: the 42. Nicoletti A, Reggio A, Bartoloni A et al. Prevalence of influence of sample design and response bias. J Epidemiol epilepsy in rural Bolivia. Epilepsia 1999;53:2064–9.
43. Hauser WA, Annegers JF, Kurland LT. Prevalence of 30. Hansen MH, Madow WG, Tepping BJ. An evaluation of epilepsy in Rochester Minnesota: 1940–1980. Epilepsia model-dependent and probability-sampling inferences in sample surveys. J Am Stat Assoc. 1983;78:776–93.
44. Turrell G, Oldenburg BF, Harris E, Jolley DJ, Kimman 31. Hook EB, Regal RR. The value of capture-recapture ML. Socioeconomic disadvantage and use of general methods even for apparent exhaustive surveys. The need practitioners in rural and remote Australia. Med J Aust for adjustment for sources of ascertainment intersection in attempted complete prevalence studies. Am J Epidemiol.
45. Gudmundsson G. Epilepsy in Iceland. Acta Neurol Scand 32. Marco EC, Verlato G, Zanolin E, Bugiani M, Drane JW.
46. Bondestam S, Garssen J, Abdulwakil AI. Prevalence and Nonresponse bias in EC Respiratory Health Survey in treatment of mental disorders and epilepsy in Zanzibar.
Italy. Eur Respir J. 1994;7:2139–45.
Acta Psychiatr Scand 1990;81:327–31.
33. Drane JW. Imputing nonresponses to Mail-back Ques- 47. Rwiza HT, Kilonzo GP, Haule J et al.
tionnaires. Am J Epidemiol. 1991;134:908–12.
incidence of epilepsy in Ulanga, a rural Tanzanian District: 34. Scott C. Research on mail surveys. J R Stat Soc.
a community-based study. Epilepsia 1992;33:1051–6.
48. Kurland LT. The incidence and prevalence of convulsive 35. Zielinski JJ. People with epilepsy who do not consult disorders in small urban community. Epilepsia 1959;1:143– physicians. Janz, eds. Stuggart: Thieme; 1976.
36. Macdonald BK, Cockerell OC, Sander JWAS, Shorvon 49. Olafsson E, Ludvigsson P, Gudmundsson G, Hesdorffer D, SD. The incidence and lifetime prevalence of neurological Kjartansson O, Hauser WA. Incidence of unprovoked disorders in a prospective community-based study in the seizures and epilepsy in Iceland and assessment of the epilepsy syndrome classification: a prospective study.
37. Sackett DL. Bias in analytical research. J Chron Dis 50. Rowland D. An ageing population: emergence of a new 38. Giuliani G, Terziani S, Senigaglia AR, Luccioni G, Foschi stage of life?Õ, The transformation of AustraliaÕs popula- N, Maffer C. Epilepsy in an Italian community as assessed tion: 1970–2030. Sydney: UNSW Press, 2003.

Source: http://publichealth.massey.ac.nz/assets/Uploads/Wendyl-2011-The-prevalence-and-demographic-distribution-of-treated-epilepsy-a-community-based-study-in-Tasmania-Australia.pdf