SEARCH

SEARCH BY CITATION

Keywords:

  • epilepsy;
  • epilepsy surgery;
  • referral;
  • underutilization

Abstract

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

Background:  Epilepsy surgery is a treatment that can cure patients with intractable epilepsy. This study investigates whether referrals for epilepsy surgery evaluation are underutilized.

Methods:  Patients with epilepsy aged 18–60 years were identified in a computerized registry held by public health care providers in a Swedish county using ICD codes. Clinical data and data on referral status for epilepsy surgery were obtained from the patients’ medical records. Potential candidates for epilepsy surgery evaluation were identified using pre-specified criteria. Obstacles for referral were analysed by comparing clinical data in patients who were considered for referral and those who were not. Appropriateness of non-referral was evaluated against recommendations from the Swedish Council on Technology in Health Care (SBU).

Results:  Of 378 patients with epilepsy in the registry, 251 agreed to participate. Of 251, 40 were already referred patients and 48 patients were identified as potential candidates for epilepsy surgery evaluation by study criteria. Referral had been considered but not performed in 15 of the potential candidates. Potential candidates not considered for referral were less likely to have seen a neurologist, to have had an EEG, CT and MRI, and more likely to have cognitive disturbances. Following the recommendations by the SBU, 28 of 48 potential candidates were identified as inappropriately not referred patients.

Conclusion:  The number of missed referrals for epilepsy surgery evaluation was estimated to be 60 per 100 000 inhabitants. Several important obstacles were found for not referring patients for epilepsy surgery evaluation.


Background

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

Epilepsy is a disease that affects nearly one per cent of the population. It is characterized by recurrent unprovoked seizures with complex effects on a person’s social, vocational and psychological functioning [1]. From 20% to 30% of persons with epilepsy are pharmaco-resistant and continue to have seizures despite treatment with anti-epileptic drugs (AEDs) [2].

Epilepsy surgery is a safe and effective treatment for medically refractory partial epilepsy [2–5]. However, the use of surgery procedures seems to be low [6] and referral of patients to epilepsy centres occurs late in the course of the disorder [7,8]. Data from the Swedish National Epilepsy Surgery Register indicate that the yearly number of epilepsy surgeries has declined during the past 10 years and is now <50 in a population of 9 million inhabitants.

There are three official Swedish recommendations on the treatment of epilepsy [9–11]. These recommendations have been written by Swedish experts in these areas and edited by the authorities. The most extensive recommendations were published in 1991 by the Swedish Council on Technology Assessment in Health Care (SBU) [10]. The council recommended that patients with severe partial epilepsy not responding adequately to medical treatment should be evaluated for epilepsy surgery. Less stringent criteria for evaluation have been published by the Medical Products Agency (...‘patients with pharmaco-resistant epilepsy should be referred to a tertiary epilepsy centre’...) [9] and by the National Board of Health and Welfare [11] (…‘epilepsy monitoring is indicated in patients having seizure despite treatment with drugs’…).

The research question in this study is whether referrals for epilepsy surgery evaluation is lower than expected amongst neurologists practising outside a tertiary epilepsy centre. We set out to assess the prevalence of referred and non-referred patients with epilepsy during 1998–2002 in a county of Sweden. We also assessed to what extent non-referred patients should have been referred if national recommendations for evaluation for epilepsy surgery had been used by the treating physicians.

Methods

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

Analysis was made in the following five steps. First, we identified the study population and its demographic and clinical characteristics (Table 1). Secondly, potential candidates for epilepsy surgery evaluation and already operated patients were identified. Thirdly, surgery candidates were described according to referral status and as to whether the treating neurologists considered referring the patients (Table 2). Fourthly, we compared the clinical characteristics according to the referral status of the patients (Table 3). Fifthly, the group of non-referred surgical candidates was divided into those who were referred in accordance with the national recommendations and those who were not.

Table 1.   Characteristics of study population (= 251)
  1. a= 238 due to missing values.

Age, years, mean (SD)40.2 (11.7)
Age at onset, yearsa, mean (SD)19.1 (14.2)
Duration of epilepsya, years, mean (SD)20.8 (13.5)
Male sex, n (%)141 (56.2)
Number of seizures in the past 2 years, n (%)
 Seizure freedom113 (45.0)
 1–325 (10.0)
 4–619 (7.6)
 7–126 (2.4)
 13–2419 (7.6)
 >2456 (22.3)
 No information available13 (5.2)
Seizure type, n (%)
 Pure partial seizures20 (7.8)
 Partial seizures occasionally generalizing116 (46.2)
 Primary generalized seizures43 (17.1)
 Not determinable64 (25.5)
 No information available8 (3.2)
EEG findings, n (%)
 Normal31 (12.4)
 Generalized epileptiform discharges62 (24.7)
 Focal epileptiform discharges77 (30.7)
 Other pathologies55 (21.9)
 No information available26 (10.4)
Findings on computed tomography of the head, n (%)
 Normal98 (39.0)
 Focal temporal pathology7 (2.8)
 Focal extra-temporal pathology28 (11.2)
 Multifocal pathology12 (4.8)
 Other pathology16 (6.4)
 No information available90 (35.9)
Findings on MRI of the head, n (%)
 Normal49 (19.5)
 Focal temporal pathology 11 (4.4)
 Focal extra-temporal pathology19 (7.6)
 Multifocal pathology10 (4.0)
 Other6 (2.4)
 No information available156 (62.2)
Distribution of numbers of tested anti-epileptic drugs, n (%)
 None1 (0.4)
 175 (29.9)
 259 (23.5)
 324 (9.6)
 >381 (32.3)
 No information available11 (4.4)
Ever consulted a neurologist, n (%)
 Yes216 (86.1)
 No31 (12.3)
 No information available4 (1.6)
Table 2.   Distribution of study population according to candidacy for epilepsy surgery evaluation, surgery status, referral status and as to whether the treating physician had considered referral for epilepsy surgery evaluation
StatusNumber (%)
  1. aNot operated after evaluation in an epilepsy surgery centre.

Not candidate for epilepsy surgery evaluation163 (64.9)
Patients referred previously
 Referred and operated24 (9.6)
 Referred, but not operateda16 (6.4)
Potential candidates for epilepsy surgery evaluation
 Not referred, referral considered15 (6.0)
 Not referred, referral not considered33 (13.1)
Total251 (100)
Table 3.   Patients’ characteristics according to referral status
 Referred or referral considered (= 55) Referral not considered (= 33)P-value
  1. a= 82 due to missing data. bFinding present versus all other findings and no data present.

Age, years, mean (SD)42.3 (10.2)42.5 (11.4)0.93
Age at onset, yearsa, mean (SD)15.8 (15.3)17.7 (17.2)0.60
Duration of epilepsya, years, mean (SD)26.3 (14.9)24.6 (15.2)0.62
Male sex, n (%)29 (52.7)20 (60.6)0.47
No data on clinical assessment by local neurologist, n (%)2 (3.6)8 (24.4)0.003
No data on EEG available, n (%)4 (7.3)15 (45.4)<0.0001
Unifocal epileptiform activity, n (%)b37 (67.3)9 (27.3)0.0002
No data on neuroradiological examination, n (%)4 (7.3)11 (33.3)0.002
Unifocal abnormality on neuroradiology, n (%)b22 (40.0)11 (33.3)0.53
Tested more than three anti-epileptic drugs, n (%)41 (74.6)14 (42.4)0.003
Cognitive disturbances, n (%)b15 (27.3)19 (57.6)0.005

Study population

Patients were identified in a patient registry held by public health care providers in the county of Jämtland, which is located in Northern Sweden. Public health care is given by general practitioners in 29 health centres and in one hospital. To a minor extent, health care is also provided by private practitioners none of whom is a neurologist. At the time of the present study, two full-time neurologists served the Jämtland population with neurological care. They were the only physicians who were permitted to refer patients for epilepsy surgery evaluation. Patients in Jämtland are referred to tertiary epilepsy centre at the University Hospital of Umeå.

In December 2002, Jämtland had 70 197 inhabitants aged 18–60 years (Statistics Sweden). A search in this population between 1998 and 2002 using ICD 10 codes G40 (epilepsy) and R56 (convulsions, not elsewhere classified) yielded 501 unique patients. These registry diagnoses were validated by means of an interview with the treating physician. Patients with unprovoked seizures of epileptic nature according to the treating physician were considered to have epilepsy (= 378). Of these, 23 (6.1%) were deceased, 97 (25.7%) did not respond to any of three invitation letters to participate in our study, seven (1.9%) declined to participate and 251 (66.4%) agreed to participate in our study.

Clinical data were collected in the medical records of the 251 patients (study population, Table 1). Information on the presence of cognitive disturbances was collected only from the charts of patients who had undergone surgery or were considered potential epilepsy surgery candidates (= 88, Table 2).

Potential candidates of referral for surgical evaluation

For this study, patients who previously had neurosurgery, or were seizure free for at least 2 years at the last counselling, and/or had tried <2 AEDs or whose EEG indicated primary generalized epilepsy, or any combination of these, were considered not be potential candidates for referral for epilepsy surgery (= 203), i.e. patients for whom it is not reasonable to consider referral. All other patients were considered potential candidates for evaluation (i.e. ‘patients not referred, referral considered’ and ‘patients not referred, referral not considered’, = 48, Table 2).

Criteria for appropriateness of non-referrals in potential candidates for evaluation

We evaluated whether it was appropriate not to refer one potential candidate for surgical evaluation by using the recommendations issued by SBU [10] which we considered more explicit than those of other authorities: ‘patients with severe partial epilepsy not responding adequately to medical treatment’ [9,11]. For this study, we interpreted the recommendation as follows: referral was indicated if (i) there was a description of partial seizures with or without secondary generalization in the medical records of the patient, (ii) the patient had tried at least two AEDs [2], (iii) the patient had one or more seizures in 2 years prior to the last counselling and d) the patient accepted to be referred (for obvious reasons).

Statistical analysis

Differences in proportions were analysed using the chi-square method using the statistical package JMP 5.1. Analysis of variance was used to test for differences in continuous variables between groups. In the analyses of differences in characteristics by referral patterns (Table 3), we considered lack of information on EEG and neuroradiological findings equal to the absence of a typical finding. For example, we tested whether there were differences in the proportions of patients with unifocal epileptiform discharges versus patients with unknown, other abnormal and normal EEG findings. This was performed to reduce the number of strata in the analysis and because lack of information can be considered equivalent to the absence of a typical finding in a retrospective study.

We decided to obtain information on seizure frequency in 2001–2002 mainly for descriptive purposes (Table 1). Some patients had medical records that extended over several years. Seizure frequency may vary considerable with time. For this reason, seizure frequency was not statistically compared between groups.

StatsDirect 2.6.4 was used to estimate the chance of having epilepsy surgery in Jämtland compared to the rest of Sweden. This programme used conditional maximum likelihoods to estimate the odds ratio, and 95% confidence interval was estimated using the exact method by Fischer.

The study was approved by the Regional Ethical Review Board in Uppsala. Written informed consent was given by the patients/next-of-kin.

Results

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

The prevalence of epilepsy (378/70,197) was 0.54% (95% CI 0.48–0.59). Participation rate (251/378) was 66.4% (95% CI 61.6–71.2). Age (= 0.73) and gender (= 0.82) did not differ statistically significant between responders and non-responders. The study population (= 251) is described in Table 1.

Of 251, 163 (64.9%, 95% CI 59.0–70.8) were not candidates for epilepsy surgery evaluation according to study criteria and 88 (35.1%, 95% CI 29.2–41.0) patients had already had been referred (= 40) or were potential candidates (= 48). The distribution of patients according to their candidacy status, epilepsy surgery status, referral status and as to whether the local neurologist had considered to referring them for evaluation is shown in Table 2.

There were some statistically significant differences between patients for whom referral had been considered including already operated patients and patients who were not considered for referral (Table 3). Patients who were not considered for epilepsy surgery work-up were less likely to have seen a neurologist, less likely to have had an EEG and a neuroradiological examination, less likely to have tested more than 3 AEDs, less likely to have an EEG finding of focal epileptiform discharges and more likely to show signs of cognitive deficits. The groups did not differ with regard to age, epilepsy duration, age at onset or in the proportions of neuroradiological findings indicative of a unifocal lesion.

Using recommendations by the SBU, it would have been appropriate to refer 28 of 48 non-referred patients. The 28 patients were distributed similarly amongst patients who were considered for referral (eight of 15) and not considered (20 of 33). Of 28, four patients were not seen by the local neurologist. Amongst the 28 patients, the number of seizures in 2001–2002 was 1–3 in four patients (14.3%), 4–6 in six patients (21.4%), 13–24 in five patients (17.9%) and more than 24 in 13 patients (46.4%).

Of the remaining 20 appropriately non-referred patients, there was no description of partial seizures in 16 (three in the ‘not referred -referral considered’ group and 13 in the ‘not referred – referral not considered group’) and four did not accept to be referred (four in the ‘not referred – referral considered’ group).

In December 2002, 70 197 people aged 18–60 years lived in the county of Jämtland and 5 027 497 of similar age lived outside the county of Jämtland (Statistics Sweden). Data from the Swedish Epilepsy Surgery Register showed that 194 patients had epilepsy surgery during 1998–2002, eight from Jämtland and 186 from the rest of Sweden [12]. The mean number of operated patients per 100 000 inhabitants was 11.4 in Jämtland and 3.7 in the rest of the country. Odds ratio of having epilepsy surgery in Jämtland when compared to the rest of Sweden was 3.1 (95% CI 1.3–6.2, = 0.008, Fischer’s exact test).

Discussion

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

To our knowledge, this is the first empirical evidence of underutilization of referrals for epilepsy surgery evaluation derived from the general population. Fifty-eight per cent of non-referred potential candidates were candidates for epilepsy surgery evaluation according to recommendations by Swedish medical authorities. These findings are important because epilepsy surgery is a safe [3], evidence-based therapeutic option [5] with good long-term prognosis in selected patients [13,14]. Underutilization of referrals for evaluation will evidently lead to an underutilization of epilepsy surgery.

Reasons for non-referral may differ between neurologists and primary care physicians, but we were unable to analyse this in detail in the present study. However, our data indicate that non-referral by the neurologists (24 of 28) is much more important than non-referral by primary care physicians (four of 28).

We were able to find results from EEG examinations in the medical records of 90% of patients (Table 1), a proportion that is somewhat higher than that of a previous Swedish incidence study in 1996, 80% [15]. Information on CT and MRI of the brain was available in 64% and 38% of patients, respectively, substantially lower than by Forsgren et al. [15], 80% and 58%, respectively. This may be explained by differences in attitudes to indications for neuroradiological examinations in a prevalence and incidence cohort, i.e. for patients who have had epilepsy for a long time and those who have not.

In patients who were not considered for referral for pre-surgical work-up by the treating physicians, information on EEG and neuroradiological findings was lacking in a substantial proportion, 45% and 33%, respectively. Swedish guidelines [9] recommend clinicians to obtain an aetiologic investigation which in most cases should include neuroimaging. In this study, the clinical work-up in some patients was performed in the pre-CT era, and neuroimaging examinations were later not ordered. This explains the lack of neuroimaging in many patients. Patients who had not been considered candidates for epilepsy surgery evaluation were less likely to have seen a neurologist, to have had an EEG, CT and MRI. Presence of a mono-focal lesion did not necessarily lead to a referral, probably because some lesions were considered not to be surgically remediable. It is reasonable to assume that all these factors represent obstacles in the process leading to a referral. Because of the retrospective nature of our study, we were unable to assess whether factors like status epilepticus, driving, employment and quality of life played a role in obtaining an epilepsy surgery referral.

We know from a recent study on the attitudes of Swedish referring neurologists that more than half of these neurologists consider lack of mental retardation to be a requirement for surgery (submitted for publication). Patients in this study who had not been considered candidates for epilepsy surgery evaluation were more likely to have cognitive defects.

This probably illustrates the attitude of neurologists to referrals of patients of with low chance of becoming seizure free after epileptic surgery in comparison to neurologists’ attitudes of trying an additional drug. For example, the chance of seizure freedom 2-years post surgery was 20% in patients with low IQ [16], whereas 12–17% of patients with intractable epilepsy became seizure free for 0.5–1 years in close temporal relation to the application of a systematic drug treatment protocol [17,18]. The fact that long transportation and hospitalization far away from home is necessary because of geographic reasons might be a contributing obstacle in this particular area.

Lhatoo et al. [6] estimated that yearly number of newly diagnosed patients who may eventually require surgery is similar to the number of patients who were operated yearly. The data of the present prevalence population indicate that the pool of surgery candidates may be large. Using recommendations by the SBU as a reference, 11.2% of the study population (28/251) were inappropriately non-referred candidates for epilepsy surgery evaluation. Give that the prevalence of epilepsy of 0.54% in adults (378/70,197) aged 18–60 years is true, the estimated number of missed referrals would be 60 per 100 000 inhabitants (28/251 × 378/70,197). If 70% of patients referred for epilepsy surgery evaluation were operated [19], our estimate would reduce to 42 in 100 000 inhabitants.

The Swedish recommendations issued by authorities are rather imprecise. ‘Pharmaco-resistance’, ‘insufficient seizure control’ and ‘intractability’ can be defined in several ways [20]. The use of liberal criteria to enter a pre-surgical evaluation has been proposed by Ryvlin and Rheims [21]. ‘Severity’ and ‘intractability’ of epilepsy are not defined by the Swedish recommendations. For the purpose of this study, we interpreted ‘not responding adequately to medical treatment’ [10] as non-seizure freedom in 2 years prior to the last counselling, because the SBU recommendations state that a severe seizure situation does not necessarily mean a high frequency of seizures [10]. There is significant difference of opinion on whether low seizure frequencies should trigger a label of intractability; in the present study, nine of 28 inappropriately non-referred patients experienced six or less seizures in 2 years. This may partly explain failure to refer. Further, statements on contraindications are given only in the SBU recommendations. They state that ‘psychiatric difficulties, personality disorders, mental retardation and psychosis are not considered reasons for avoiding an operation’ [10] and that surgery is not performed in some eloquent brain areas. In everyday clinical practice, decisions to refer and – subsequently – operate must be individualized, taking contraindications into account. The clinical picture in five amongst 28 inappropriately non-referred patients may represent relative or absolute contraindications for referral; one patient had severe cardiac disease, one who did not accept physical examination and three had post-stroke epilepsy. All-in-all, in half (9 + 5) of the 28 patients who fulfilled referral criteria by the SBU, there were further potential obstacles for referral. This would indicate a possible reduction in the number of missed referrals from 60 to 30 patients per 100 000 inhabitants because of factors that are not strictly defined the official recommendations.

The present study suffers from limitations inherent to retrospective medical record reviews, for example the lack of information in some patients. To partially account for these limitations, we validated the registry-based diagnosis of epilepsy by interviewing the treating physicians. The use of physician-based diagnoses was justified by the fact that we aimed at describing referral patterns of patients who were considered to have epilepsy by their doctors. The prevalence of epilepsy was in agreement with the results of previous European studies [22], indicating that our physician-based diagnoses were fair. Further, physicians may have considered referral for epilepsy surgery without documenting this in the medical records. This failure to document may lead to an overestimation of the size of the not considered-not referred group. Another shortcoming is the small number of patients.

The strengths of the study are the extensive method of case finding and the well-defined population and catchment area. Non-responders were similar to responders with respect to age and gender.

Data from Swedish Epilepsy Surgery Register show that the chance of having epilepsy surgery was three times higher in Jämtland than in the rest of the country during 1998–2002, possibly indicating that the prevalence of non-referrals in this period of time was larger in other areas of Sweden than in Jämtland. The frequency of epilepsy surgeries differs substantially by geographical region and university hospital. Neurologists are unevenly distributed within Sweden with clusters in large cities where university hospitals are located. We believe that our findings are mainly generalizable to Swedish areas where neurologists are responsible for referring patients to tertiary centres. The problems observed in this study might be different in health care systems with routines and criteria for referral that are dissimilar from those in Sweden.

We need to know more about the reasons for referrals and non-referrals for epilepsy surgery, how patients are informed and how they understand and react when they are faced with the option to be referred.

Akbari et al. [23] recently showed that dissemination of referral guidelines is effective when they are combined with other measures such as the use of structured referral sheets and the collaboration between care providers of different levels. To avoid inappropriate non-referrals, it is essential to develop and implement guidelines on epilepsy surgery in close collaboration with referring neurologists and subspecialists in tertiary epilepsy centres.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

We thank Mrs Marie Hermansson and Mrs Eva-Britt Ogren for invaluable help with administrative matters.

Disclosure of sources of funding

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References

This study was sponsored by Selanders foundation, the Developmental Foundation of the University Hospital of Uppsala and by an educational grant from the Research and Development Unit, Jämtland County.

References

  1. Top of page
  2. Abstract
  3. Background
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Disclosure of sources of funding
  9. References