SEARCH

SEARCH BY CITATION

Keywords:

  • Epilepsy;
  • Treatment gap;
  • Anti-epileptic drugs;
  • Adherence;
  • Interventions;
  • Developing countries

Summary

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

In many developing countries, people with epilepsy do not receive appropriate treatment for their condition, a phenomenon called the treatment gap (TG). We carried out a systematic review to investigate the magnitude, causes, and intervention strategies to improve outcomes in developing countries. We systematically searched MEDLINE, EMBASE, and PsycINFO databases, supplemented by a hand search of references in the key papers. The degree of heterogeneity and a pooled TG estimate were determined using metaanalysis techniques. The estimates were further stratified by continent and location of study (urban, rural). Twenty-seven studies met the inclusion criteria: twelve from Africa, nine from Asia and six from Latin America. We observed a high degree of heterogeneity and inconsistency between studies. The overall estimate of the TG was 56/100 [95% confidence interval (CI) 31.1–100.0]. The variation in estimates could possibly be explained by nonuniform TG estimation methods and the diverse study populations, among other factors. The TG was mainly attributed to inadequate skilled manpower, cost of treatment, cultural beliefs, and unavailability of antiepileptic drugs (AEDs). These factors have been addressed using different intervention strategies, such as education and supply of AEDs. Future research should estimate the TG coherently and develop sustainable interventions that will address the causes.

Epilepsy is the most common chronic neurological disorder, affecting approximately 50 million people worldwide, of whom 40 million are estimated to live in developing countries1 (WHO, 2004). Several studies have reported that over 90% of people with epilepsy (PWE) in developing countries do not receive appropriate treatment for their condition, a phenomenon known as the treatment gap (TG) (Shorvon & Farmer, 1988; Scott et al., 2001; Diop et al., 2003, 2005). However, none of the studies provide a confidence interval (CI), suggesting that the estimate is not data driven. The TG is defined as the number of people with active epilepsy not on treatment (diagnostic and therapeutic) or on inadequate treatment, expressed as a percentage of the total number with active epilepsy (Meinardi et al., 2001; Kale, 2002). The epilepsy treatment gap (ETG) has a broader definition to include the influence of epilepsy on mental and social well-being (Meinardi et al., 2001). For the purpose of this review, active epilepsy is defined as having at least one unprovoked seizure in the last 5 years. The TG has been estimated by the direct method during prevalence studies and indirectly from the amount of antiepileptic drugs (AEDs) consumed in the country and the number of people with active epilepsy (Kale, 2002). The gap is reported to be influenced by various factors, including lack of access to or knowledge of AEDs, poverty, cultural beliefs, stigma, poor health delivery infrastructure, and shortage of trained professionals (Meinardi et al., 2001; Scott et al., 2001).

The Global Campaign Against Epilepsy, a partnership between the World Health Organization (WHO), International League Against Epilepsy (ILAE), and the International Bureau for Epilepsy (IBE) was launched in 1997 to bring epilepsy “out of the shadows” by addressing some of the factors outlined above (Sander, 2002; Diop et al., 2003). In 2002, the Global Campaign entered its second phase, and several demonstration projects were set up in various countries, to reduce the TG and the physical, economic, and social burden of epilepsy through community-level interventions (Sander, 2002; Li et al., 2007). Despite this concerted effort by different stakeholders, there are no systematic reviews on the magnitude, causes, or intervention strategies to improve the outcomes in developing countries. Therefore we conducted a systematic review of the literature on the TG to answer the following questions: (1) What is the magnitude of the TG in developing countries? (2) What are the factors responsible for the TG in developing countries? (3) What interventions have been implemented to address causes of the TG?

Methods

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

Data sources and search strategy

A literature search was carried out using three databases: MEDLINE (1966 through June 2007), EMBASE (1980 through June 2007), and PsycINFO (1887 through June 2007) using a combined text word and medical subject headings (MeSH) to identify relevant papers (Appendix 1). Additional articles were identified by searching references cited in the key papers. The strategy was developed by breaking the review question into its elemental facets: exposure, outcome, population, publication language, and keywords according to the recommendations of the National Health Service Centre for Reviews and Dissemination (Khan et al., 2001) (Appendix 1). Publication language was left open to ascertain how many studies were available in languages other than English.

Study selection

The online abstracts of studies identified from the database search were reviewed, and reprints of potential eligible studies obtained. Studies meeting one of the following criteria were chosen for more detailed review: (1) cross-sectional studies that provide the prevalence of the TG, or studies using the indirect method to estimate the gap; (2) qualitative or observational studies that identify potential causes of the TG; and (3) interventions to address some of the attributed causes of the TG: prospective cohort or those that have used randomized control or prepost designs and focused on either education or supply of drugs.

Studies excluded

Studies with the following characteristics were excluded from review: (1) epilepsy prevalence studies that did not estimate the TG; (2) studies reported in languages other than English; (3) reviews; (4) editorials, letters, or reports; (5) studies conducted in developed countries; and (6) studies that reported same results in different papers.

Data extraction

Data were extracted using a proforma designed for the review. The first reviewer extracted all the data, and the fourth reviewer reextracted data from a sample of half of the studies. The studies were organized into three broad categories: those on magnitude, attributed causes, and interventions. Information from each study was then obtained on author, year of publication, country, study design, sample size, ascertainment method, length of study, age of participants, and outcome. We extracted data on magnitude of the TG and calculated a 95% CI around the estimates. We further stratified the TG estimates by continent and location of study (urban, rural). All data were entered in an Excel spreadsheet and transferred to STATA version 9.2 (College Station, TX, U.S.A.) for analysis.

Analysis

The 95% CI for the TG estimates were calculated, and heterogeneity was investigated. The TG prevalence estimates were transformed to logits [log (p/1−p)] to improve their statistical properties and later back-transformed to prevalences and expressed as percentages. For further analysis, the data was stratified by continent and location of the study. Attributed causes of the TG were listed and compared across continents using descriptive statistics, but such comparison was not performed for the interventions, because they measured different outcomes and few provided numerical estimates.

Description of heterogeneity and summary estimates

We used a forest plot (Lewis & Clarke, 2001) to visualize the extent of heterogeneity among the studies that investigated the magnitude of the TG. The standard test for heterogeneity, the Cochrane chi-squared (χ2) test, was used to examine the null hypothesis of homogeneity. We used a method that quantifies inconsistency across studies, thus assessing its impact on the metaanalysis. This statistic is I2=[(Q – df)/Q]× 100%, where Q is the χ2 statistic, and df is its degrees of freedom (Higgins & Thompson, 2002; Higgins, 2003). I2 describes the percentage of the variability in estimates that is due to true heterogeneity (true differences in TG prevalence), and a value greater than 50% was considered substantial heterogeneity.

The mean TG prevalence and its CI were derived from random effects metaanalysis, an analytical approach used when heterogeneity cannot be readily explained. This assumes that the outcomes being estimated in the different studies are not identical, but follow a normal distribution, allowing for variation between studies. However, the usual CI of mean in the random effects model does not take into account the between-study variance and so can be narrow where there is substantial heterogeneity. We therefore calculated the 95% CI for the true TG prevalence as the mean of logits ± 1.96ℓ, where ℓ is the among-study standard deviation (Goodman, 1989).

Assessment of methodological quality

The studies were appraised by two independent reviewers based on the criteria outlined in a critical appraisal guide (Crombie, 1996) and guidelines on how to appraise a paper (Greenhalgh, 2001). The relevant methodological aspects were identified and assessed individually for each study. The studies were then rated as good, average, or poor as outlined in Appendix 2. We did not use a composite numerical score to reflect overall methodological quality, because there is no gold standard for the “true” methodological quality, and such scores are probably neither valid nor reliable in practice (Greenhalgh, 2001). Empirical evidence and theoretical considerations suggest that although summary quality scores may in some circumstances provide a useful overall assessment, they should not generally be used to assess the quality of studies in systematic reviews, because different scales give divergent scores and rankings on one study (Greenland, 1994; Juni et al., 1999).

Results

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

The electronic search produced 130 references. A hand search of references cited in the key papers identified 15 additional papers (Appendix 3). These papers were obtained and reviewed, and the majority were subsequently excluded because they did not meet the review criteria: they were review articles (N = 20), editorials or letters (N = 12), reports (N = 20), not in English (N = 2), prevalence studies that did not estimate the TG (N = 19), trials of AEDs (N = 10), economic evaluation studies (N = 5), studies in developed countries (N = 28), or studies reporting the same results in different papers (N = 2).

Twenty-seven studies fulfilled the inclusion criteria. Twelve (44%) were prevalence studies that measured the TG, eight (30%) identified attributed causes of the TG, and the remaining seven (26%) reported the effect of interventions designed to address the attributed causes of the TG. Twelve (45%) of the studies were conducted in Africa, nine (33%) were conducted in Asia, and six (22%) were conducted in Latin America.

Magnitude of the TG

Out of the twelve studies identified, six (50%) were conducted in Latin America (Mendizabal & Salguero, 1996; Nicoletti et al., 1999; Noronha et al., 2004, 2007; Medina et al., 2005; Somoza et al., 2005), three (25%) were conducted in Africa (Coleman et al., 2002; Dent et al., 2005; Ndoye et al., 2005) and three (25%) were conducted in Asia (Aziz et al., 1997; Radhakrishnan et al., 2000; Wang et al., 2003). The majority, eleven (92%), were population-based cross-sectional surveys, while one (8%) used indirect method to estimate the gap. There was wide variability in the TG estimates among the studies that provided its magnitude (Fig. 1). The Cochrane χ2 statistic and measure of inconsistency were large (Q = 1331.5, df = 13, p < 0.0001; I2= 99%), suggesting substantial variation among the studies that was beyond sampling variation. The random effects mean of the TG prevalence for all of the studies was 56.0/100 (95% CI for true prevalence 31.1–100.0). When stratified by continent, the random effects mean of the TG prevalence for Latin America was 55.4/100 (95% CI 39.0–78.6), Asia was 64.0/100 (95% CI 24.3–100.0), while that of Africa was 49.0/100 (95% CI 14.0–100.0). The mean of the TG prevalence for urban settings was estimated at 46.8/100 (95% CI 34.1–64.8) and 73.3/100 (95% CI 49.5–100.0) for rural regions (Table 1). The TG estimate from Turkey was not included in this stratification because distinct figures were not provided for rural and urban regions (Aziz et al., 1997). The TG summary is outlined in Table 1, and details of individual studies are outlined in Appendix 4.

image

Figure 1. Forest plot for the epilepsy treatment gap prevalence.

Download figure to PowerPoint

Table 1.  Magnitude of the epilepsy treatment gap by region and location
Continent/locationNo. of studiesTG (%)Lower 95% CIUpper 95% CI
Latin America755.439.0 78.6
Asia464.324.3100.0
Africa348.914.3100.0
Urban746.834.1 64.2
Rural773.349.5100.0

Causes of the TG

Eight studies investigated causes of the TG. Half were conducted in Africa (Elechi, 1991; Preux et al., 2000; Bassili et al., 2002; El Sharkawy et al., 2006), and the other half were conducted in Asia (Pal et al., 2000; Asawavichienjinda et al., 2003; Mac et al., 2006; Das et al., 2007). No studies were indentified from Latin America. Two were qualitative, six were quantitative, and one study combined both methodologies. Most of the studies were small (fewer than 100 participants), except two for studies that interviewed 229 and 1450 PWE, respectively (Bassili et al., 2002; Das et al., 2007). All the eight studies reported that the cost associated with seeking epilepsy care contributed to the TG (Elechi, 1991; Pal et al., 2000; Preux et al., 2000; Bassili et al., 2002; Asawavichienjinda et al., 2003; El Sharkawy et al., 2006; Mac et al., 2006; Das et al., 2007). Attributed causes of the TG were multiple and overlapping in the two continents, as summarized in Table 2. These causes were also similar for rural and urban regions. The highest median (70%) was associated with inadequate skilled manpower, and the lowest median (18.5%) was associated with long distances to health facilities (Table 2). Nonadherence to AEDs, a factor that also contributes to TG was investigated in two studies (Elechi, 1991; Asawavichienjinda et al., 2003). Details of individual studies are shown in Appendix 5.

Table 2.  Cause of the epilepsy treatment gap expressed as median and range
Causes of ETGNo. of studiesMedian (%)Minimum (%)Maximum (%)
Cost of treatment8621190
Superstitions and cultural beliefs540 765
Unavailability of drugs5531844
Long distance to health facilities318.51819
Traditional treatment344 682
Inadequate skilled manpower3706476

Intervention strategies to address causes of the TG

Seven studies were identified that addressed attributed causes of the TG. Five (71%) were conducted in Africa (Feksi et al., 1991; Adamolekun et al., 1999, 2000; Olley et al., 2001; Berhanu et al., 2002), and two (29%) were conducted in Asia (Gourie-Devi et al., 2003; Liu et al., 2003). No studies were indentified from Latin America. Five interventions were solely educational (Adamolekun et al., 1999, 2000; Olley et al., 2001; Gourie-Devi et al., 2003; Liu et al., 2003), one provided AEDs (Feksi et al., 1991), and one combined education with provision of AEDs (Berhanu et al., 2002). The education interventions were modular in nature and were delivered verbally through workshops to PWE and health professionals. They covered different topics such as causes of epilepsy, epidemiology of epilepsy, diagnosis and management of epilepsy, psychosocial aspects and community-based care. These interventions led to an increase in knowledge among PWE and health professionals, as measured by the difference between assessment before and after the intervention (Adamolekun et al., 1999; Olley et al., 2001; Gourie-Devi et al., 2003; Liu et al., 2003). In addition, education led to an increase in patient recruitment (Adamolekun et al., 1999, 2000; Berhanu et al., 2002). Though information pamphlets led to improvements in knowledge and a reduced default rate, they did not have any effect on adherence, as measured by self-reports and serum AED levels (Adamolekun et al., 1999; Liu et al., 2003). However, verbal education and drug supply led to an increase in adherence (Feksi et al., 1991; Adamolekun et al., 1999, 2000; Liu et al., 2003). Only one study assessed psychosocial factors, which are known to affect quality of life among PWE. This study reported decreased levels of depression and neurotic disorders in the group receiving education (Olley et al., 2001). Details of individual studies are summarized in Appendix 6.

Discussion

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

Magnitude of the TG

A comprehensive search of the literature identified only twelve studies that estimated the magnitude of the TG. This paucity of studies substantiates Kale's findings that the TG as an outcome measure is not well studied in the developing world (Kale, 2002). The results of this review show that the pooled TG estimate of 56% is lower than the 90% that is widely quoted in many studies (Shorvon & Farmer, 1988; Scott et al., 2001; Diop et al., 2003, 2005; Dua et al., 2006) although the CI are wide. The higher estimate is not based upon systematic review of the data and does not provide CIs. To the best of our knowledge, this is the first study that comprehensively reviews the literature to assess the variability of the TG using a robust and reproducible method. The few narrative reviews that have been conducted in developing countries have addressed the epidemiology of epilepsy with a mention of the TG (Shorvon & Farmer, 1988; Ray et al., 2002; Sridharan, 2002; Bharucha, 2003; Rajbhandari, 2004; Mac et al., 2007).

A substantial amount of variation in the TG among studies was demonstrated by graphical display of the estimates, a statistical test of heterogeneity, and a measure of inconsistency. The pooled estimate, using a method that corrects for among study variation, provides a meaningful indication of the magnitude of the TG. When stratified by continent, studies conducted in Africa had the highest variability, whereas Latin America had the lowest. Variability was also higher in rural compared to urban areas (Table 1), although none of the studies in this review attempted to identify specific reasons for such rural/urban differences.

Though we did not investigate sources of heterogeneity, some variation in estimates may be explained by a failure to calculate the TG uniformly. Seven studies calculated it using active epilepsy as the denominator (Aziz et al., 1997; Radhakrishnan et al., 2000; Coleman et al., 2002; Wang et al., 2003; Dent et al., 2005; Medina et al., 2005; Noronha et al., 2007), and four used both active and passive epilepsy (Mendizabal & Salguero, 1996; Nicoletti et al., 1999; Ndoye et al., 2005; Somoza et al., 2005). In addition, studies defined active epilepsy differently: eight confined it to five years, according to the ILAE definition (Mendizabal & Salguero, 1996; Aziz et al., 1997; Nicoletti et al., 1999; Radhakrishnan et al., 2000; Coleman et al., 2002; Dent et al., 2005; Medina et al., 2005; Somoza et al., 2005), two limited it to one year (Wang et al., 2003; Ndoye et al., 2005), whereas one study used two years (Noronha et al., 2007). Other factors that might have contributed to heterogeneity include different study populations, unskilled manpower in rural settings, variable socioeconomic conditions, and diverse levels of health care development in the study regions (Jallon, 1997; Nicoletti et al., 1999). The study by Somoza et al., (2005), which had the widest CI (Fig. 1), consisted of school children, and only a small number failed to seek treatment. This could have been influenced by the high level of literacy and the study setting, which was described as a district where social, economic, and health indicators ranked among the country's highest and reached levels comparable to developed countries.

Causes of the TG

We found that attributed causes of the large TG in developing countries were multiple and overlapped between continents (Appendix 5). All eight studies included in the review reported that the cost of seeking epilepsy treatment was associated with the TG in developing countries (Elechi, 1991; Pal et al., 2000; Preux et al., 2000; Bassili et al., 2002; Asawavichienjinda et al., 2003; El Sharkawy et al., 2006; Mac et al., 2006; Das et al., 2007). The attributed causes with the highest medians were related to the health systems; mainly inadequate skilled manpower, cost of treatment, and unavailability of drugs. This indicates that health system issues are a major obstacle for TG. Though individual perceptions, such as cultural beliefs, traditional treatment, and distance to health facilities had lower medians, they greatly influence treatment seeking among PWE. The findings of this review show that superstitions and cultural beliefs influence PWE to seek treatment from traditional healers instead of allopathic practitioners (Pal et al., 2000; Preux et al., 2000; Asawavichienjinda et al., 2003; El Sharkawy et al., 2006; Das et al., 2007). Shorvon and Farmer also observed that many patients spent considerable amounts of money to obtain traditional cures, and it was common for patients to travel hundreds of miles or donate treasured items to a healer in return for antiepileptic treatment (Shorvon & Farmer, 1988). This negates the usefulness of the advances made in the diagnosis and treatment of epilepsy (Leppik, 1988; Das et al., 2007), hence the need for comprehensive programs to address these attributed causes.

Intervention strategies to address causes of the TG

The results of this review suggest that some attributed causes of the TG in developing countries can be addressed through educational interventions and supply of AEDs. Such interventions should target health providers (including traditional healers), PWE, and the wider community (Scott et al., 2001; Berhanu et al., 2002). The interventions should be tailored to the needs of each target group; those for health providers should be geared towards improving skills in diagnosis and management of epilepsy, whereas for PWE, emphasis should be on adherence, when and how to take AEDs, as well as how to live with epilepsy. They should also include psychosocial aspects of epilepsy that may lead to the development of a positive attitude towards PWE, which is essential in improving quality of life and treatment (Jallon, 1997). Educational interventions in developed and developing countries have been shown to improve epilepsy knowledge, AED adherence, seizure outcome, and self-esteem among PWE (Helgeson et al., 1990; Adamolekun et al., 1999; Clark et al., 2001; Olley et al., 2001; Berhanu et al., 2002; May & Pfafflin, 2002; Gourie-Devi et al., 2003; Helde et al., 2003, 2005; Liu et al., 2003; Snead et al., 2004; Wohlrab et al., 2007). Other studies have suggested that adequate drug supplies have to be provided for the success of any epilepsy management program (Watts, 1989; Feksi et al., 1991; Mani et al., 2001; Gourie-Devi et al., 2003). However, experience in developing countries with other major public health problems, particularly communicable diseases, has demonstrated that simply delivering drugs to these countries will not necessarily reduce the TG (Reynolds, 2000). This indicates that health system interventions are not sufficient on their own. There is need to accompany such interventions with nonpharmacological, community-based interventions in order to reduce the stigma of epilepsy and reduce barriers to effective care (Krishnamoorthy et al., 2003).

Limitations of the review

Studies may not have been identified at the search stage if they were not indexed in the three databases used or not published in mainstream journals. Studies on causes and interventions varied in the population studied, selection procedures, methods of ascertainment, study length and outcomes measured. Due to these variations, data from separate studies could not be statistically combined. Numeric estimates were not available for some quantitative outcomes in intervention studies, and therefore descriptive statistics were not calculated for this section of the review. Studies that investigated the TG were independent of those that instituted interventions; hence it was not possible to compare differences in the TG before and after an intervention. We did not investigate the potential sources of heterogeneity due to the small number of studies and inadequate variables in studies that estimated the TG. Though we included all languages during the search, we were not able to translate two Chinese papers reporting magnitude of the TG, although the abstracts indicated an estimate similar to the included studies.

Conclusion

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

This review provides a more accurate pooled estimate of the TG in developing countries with CI. Furthermore, it provides attributed causes of the gap and lists interventions that have been implemented to improve outcome in developing countries. Given the economic, social, political, and cultural context of the TG, there is need for future research to focus on well-planned and coordinated interventions. The findings of this review suggest that such interventions should consider the medical, developmental, and psychosocial needs of PWE as well as being financially, geographically, and culturally accessible. These interventions should also involve health system personnel as well as other personnel, such as traditional healers who incorporate cultural beliefs and provide more comprehensive care.

Acknowledgments

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

The Wellcome Trust-UK and Kenya Medical Research Institute supported this study. We thank Peter Odermatt of Swiss Tropical Institute for his comments on the manuscript. C.R.N. holds a Wellcome Trust Senior Fellowship in Clinical Tropical Medicine (No. 070114). This paper is published with the permission of the director of KEMRI.

Conflict of interest: 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. All the authors certify that they had no financial or personal interest, including advisory board affiliation, in any company or organization sponsoring the research.

References

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix

Appendix

  1. Top of page
  2. Methods
  3. Results
  4. Discussion
  5. Conclusion
  6. Acknowledgments
  7. References
  8. Appendix
Table Appendix 1..  Description of the search strategy
Search elementMEDLINEEMBASEPsycINFO
ExposureThesaurus terms explodedThesaurus terms explodedThesaurus terms exploded
AED interventionAED interventionAED intervention
Education interventionEducation interventionEducation intervention
OutcomeThesaurus terms explodedThesaurus terms explodedThesaurus terms exploded
Increased knowledgeIncreased knowledgeIncreased knowledge
AED adherenceAED adherenceAED adherence
Drug levelDrug levelDrug level
Seizure frequencySeizure frequencySeizure frequency
Clinic attendanceClinic attendanceClinic attendance
EpilepsyThesaurus terms explodedThesaurus terms explodedThesaurus terms exploded
Epilepsy:Epilepsy:Epilepsy:
ReflexAbsenceAbsence
AbsenceTemporal lobeGeneralized
RolandGeneralizedTonic–clonic
GeneralizedGrand malComplex partial
Tonic–clonicFrontal lobeSeizures
Frontal lobeMyoclonus 
Partial motorFocal 
Partial sensoryPetit mal 
PosttraumaticRolandic 
Temporal lobeReflex 
Benign neonatalStartle 
Complex partialIntractable 
Epilepsies:Seizures: 
MyoclonicAtonic 
PartialAudiogenic 
Myoclonic epilepsy:Clonic 
JuvenileGelastic 
ProgressiveNocturnal 
Seizures: 
Alcohol withdrawal seizures 
Febrile seizures 
Developing countriesResource poor countriesResource poor countriesResource poor countries
Third world countriesThird world countriesThird world countries
Majority world countriesMajority world countriesMajority world countries
South AmericaSouth AmericaSouth America
AfricaAfricaAfrica
AsiaAsiaAsia
Treatment gapTreatment gapTreatment gapTreatment gap
Treatment statusTreatment statusTreatment status
Antiepileptic drugsAnticonvulsantsAnticonvulsantsAnticonvulsants
AdherenceComplianceComplianceCompliance
Patient compliancePatient compliancePatient compliance
Treatment refusalTreatment refusalTreatment refusal
InterventionsProjectsProjectsProjects
ProgramsProgramsPrograms
PopulationAdults or childrenAdults or childrenAdults or children
LanguageAnyAnyAny
Table Appendix 2..  Domain of methodological quality of studies included in the review
Cross-sectional surveys
AuthorYearDescription of the study populationDescription of sampling procedureResponse rate (%)
Mendizabal1996PoorPoor  Good (97.3)
Aziz1997GoodGood  Good (100)
Nicoletti1999GoodGood  Good (98.3)
Radhakrishnan2000GoodGood  Good (98)
Coleman2002GoodGood  Good (99.8)
Wang2003Poor  Average  Good (94.6)
Ndoye2005GoodGood  Good (100)
Medina2005GoodGood  Good (86)
Somoza2005GoodGood  Good (83.1)
Dent2005GoodGood  Good (99)
Naronha2007GoodGood  Good (100)
Preux2000Good  Average  Average (72)
Bassili2002GoodGood  Good (100)
Thanin2003GoodGood  Good (89.2)
Mac2006GoodGood  Good (100)
Qualitative studies
  Description of setting and selection of subjectsDescription of data collection methodsDescription of data analysis and quality control measures
Gehane2006Good  GoodAverage
Pal2000AveragePoorPoor  
Case control studies
AuthorYearClear explanation on how cases were obtainedSelection of appropriate controlsIdentical data collection for cases and controls
Pal2000GoodGoodAverage
Prospective cohort studies
  Clear explanation of the study participantsAppropriateness of the control group if anyAdequate follow up
Elechi1991GoodN/AGood (2.5 years)
Das2007GoodN/AGood (1 year)
Adamolekun2000GoodN/AGood (2 years)
Berhanu2002GoodN/AGood (1.5 years)
Feksi1991GoodN/AGood (1 year)
Pre-post interventions studies
  Description of study participantsDescription of the interventionEvaluation interval
  1. N/A, not applicable.

Adamolekun1999GoodAverageGood (6 months)
Olley2001GoodGood  Poor (2 months)
Gourie-Devi2003GoodAverageAverage (3 months) 
Liu2003GoodAveragePoor (1 month)
Table Appendix 3..  Flow diagram of study selection process
inline image
Table Appendix 4..  Magnitude of the epilepsy treatment gap in developing countries
AuthorYearCountryStudy designAscertainmentAge yearsLength of studySample size Active epilepsyPrev/1000TG (%)L 95% CI U 95% CI
  1. Prev, prevalence; TG, treatment gap; L, lower; U, upper; CS, cross-sectional survey; N/A, not applicable; Q, questionnaire; R, rural; U, urban; SU, semiurban; C, Campinas city; S, Sao Jose do Rio Preto city.

Mendizabal1996GuatemalaCSQAllN/A1882 (R)115.868.841.389
Aziz1997PakistanCSQAllN/A8513 (R)126 (R)14.8 (R)98 (R)93.299.5
Turkey 15 617 (U)115 (U)7.4 (U)73 (U)6480.9
6680 (R)59(R)8.8 (R)70 
4817 (U)22 (U)4.5(U) 
115 (U) 
Nicoletti1999BoliviaCSQAll2 years9955 (R)11211.189.582.794.3
Radhakrishnan2000IndiaCSQAll5 months238 102 (SU)11754.93835.340.9
Coleman2002GambiaCSQAll9 months16 200 (R)694.34835.660.2
Wang2003ChinaCSQAllN/A55 000 (R)2574.66356.869.0
Ndoye2005SenegalCSQAllN/A4500 (SU)6414.223.413.835.7
Noronha2004BrazilIndirectFormulaN/AN/A6324 (S) (U)2591 (S)18.6 (S)59 (S)57.161.0
 methodn1-n2/n1×100 18600 (C) (U)8182 (C)18.6 (C)56 (C)5557.1
Medina2005HondurasCSQAll11 days6473 (R)10015.453.342.863.1
Somoza2005ArgentinaCSQChildren7 months31 615 (U)682.6 7.1 2.916.0
Dent2005TanzaniaCSQAll5 months4905 (R)428.695.883.899.4
Naronha2007BrazilCSQAll4 months96 300 (U)2905.43832.343.8
Table Appendix 5..  Causes of the epilepsy treatment gap in developing countries
AuthorYearCountryStudy designSample size AscertainmentAge yearsLength of of studyCauses of treatment gap
  1. PWE, people with epilepsy; CWE, children with epilepsy; THs, traditional healers; HCP, health care personnel; AEDs, antiepileptic drugs; R, rural; U, urban; SU, semiurban; N/A, not applicable.

Elechi1991NigeriaProspective cohort45 PWE (R)QuestionnaireAdults2–5 yearsNonavailability of drugs
Excessive cost of drugs
Continuity of seizures
Side effects of drugs
Long distance to health facility
Pal2000IndiaCase control 94 ChildrenQuestionnaire2–181 yearSeasonal income
-32 (Cases)Semistructured topic schedule Intercurrent illness in family members
-62 (Controls) Domestic commitment among women
Religious festivals
Qualitative 32 Parents (R) Local cultural customs and beliefs
Impassable roads during rainy season
Long distance to health facilities
Denial of diagnosis
Drug side effects
Alternative treatment
Continuing seizures
Lack of benefit from the treatment
Symptoms resolution
Preux2000CameroonCross-sectional survey33 PWEQuestionnaire≥184 monthsPoor health delivery infrastructure
26 Physicians Ignorance about causes of epilepsy
21 Pharmacists Uneven supply of drugs
 3 Distributors Superstitions and cultural beliefs
 8 THs (U) Alternative treatment
Unavailability of drugs
Long-term treatment
Expensive to consult physician
Bassili2002EgyptCross-sectional survey229 Children (U)Questionnaire1–154 monthsLack of access to neurologists
Lack of diagnostic services (EEG)
Defective health education of caregivers
Non-compliance to AEDs
Thanin2003ThailandCross-sectional survey72 PWE (R)Questionnaire≥155 monthsMisunderstanding need for long-term treatment
Forgetting to take drugs
Economic problems
Continuity of seizures despite treatment
Misbeliefs
No caregiver to take PWE to hospital
Mac2006VietnamCross-sectional survey33 Pharmacies (U)QuestionnaireAdults2 monthsNonavailability of AEDs
Cost of AEDs
Lack of skilled manpower
Drugs dispensed for short periods
Pharmacies concentrated around hospitals and main market
Gehane2006KenyaQualitative9 grandparentsChecklist of questionsAllN/AInadequate knowledge on causes epilepsy
16 parents Inadequate skilled manpower
19 HCP Superstitions and cultural beliefs
19 CWE (R) Severity of seizures
Distance and time to health facilities
Lack of finances
Continuity of seizures despite treatment
Inconsistent drug supply
Alternative treatment
Discrimination and stigma
Das2007IndiaProspective cohort1450 PWE (SU)QuestionnaireAdults1 yearLow annual income
Unemployment
Loss of jobs
Marital disharmony
Superstitions and beliefs
Nonavailability of drugs
Frustration and despair
Table Appendix 6..  Intervention strategies to mitigate causes of the epilepsy treatment gap
AuthorYearCountryStudy designSample sizeIntervention/ascertainmentAge yearsStudy lengthFindings
  1. PHCNs, primary health care nurses; EHTs, environmental health technicians; DMOs, district medical officers; Neulog, neurologists; SHA, state health administrators; DCs, developing countries; R, rural; U, urban; SU, semiurban.

Adamolekun1999ZimbabwePre-post31 PHCNsEducationAdults6 monthsImproved knowledge in diagnosis and management of epilepsy among PHCNs by 11% and EHTs by 9% as measured by pre- and posttest
24 EHTsQuestionnaire PHCNs and EHTs education led to a 35% increase in patient recruitment and drug compliance as measured by serum levels of AEDs
296 PWE (R)AEDs blood level Information pamphlets led to a reduction in patient default rate (56.3% in the control group and 22.3% in the experimental group) but did not influence drug compliance or seizure frequency
Adamolekun2000ZimbabweProspective cohort114 PWEEducation8–562 yearsNo cases of noncompliance as defined by undetectable serum AED levels. 26% out of 38 clinically indicated cases had normal levels, 58% were below therapeutic ranges, and 16% were above the range
(U)AEDs blood level 47% increase in PWE registered in epilepsy support group
Olley2001NigeriaPre-post30 PWEEducation21–652 monthsSignificant increase in knowledge about epilepsy in the education group as measured by the knowledge about illness schedule, p < 0.001
(R)Questionnaire Significant decrease in levels of depression in education group as measured by Becks Depression Inventory (BDI), p < 0.001
Significant decrease in neurotic disorders in the education group as measured by the Crown-Crisp Experiential Index (CCEI), p < 0.001
Berhanu2002EthiopiaProspective cohort813 PWEAEDs supply1–751.5 yearsRapid increase in the number of new patients attending the clinics due to awareness created by community leaders, community health workers, and the trained nurses
5–10 NursesEducation Regular and adequate supply of phenobarbitone to each health center
(R)Self-reports 
Feksi1991KenyaProspective cohort302 PWE (R, SU)AEDs supply6–651 year76% of 249 patients had detectable serum drug concentrations, and the AEDs were effective in reducing seizure frequency in 79% of the patients
AEDs blood level There was low drop-out, 17%, low rate of noncompliance, 6%, and good response to therapy, 53%
Gourie-Devi2003IndiaPre-post148 DMOsEducationAdults3 yearsIncrease of knowledge and skills in identification, diagnosis, investigation and management of epilepsy among DMOs as measured by pre- and posttest
21 SHAQuestionnaire Improved understanding of issues related to compliance, medication, counseling, psychosocial, and legal aspects among DMOs
28 Neulog Endorsement of the epilepsy control program by SHAs and neurologists which lead to initiation and strengthening of epilepsy services at the periphery
Liu2003TaiwanProspective cohort51 PWEEducation18–597 monthsProvision of drug leaflets improved medication knowledge by 30% among PWE, hence an increase in epilepsy medication score (EMA), p < 0.001
Pre-post Questionnaire Significant reduction in drug side effects after the awareness campaign, p < 0.05
AEDs blood level Nonsignificant improvement in AED compliance after provision of leaflets as measured by self-reports and serum AEDs levels