Perceived Stigma in Epilepsy Patients in a Low Income Country; Ethiopia: A Review and Meta-Analysis of the Prevalence and Related Factors.


 Background: The presence of perceived stigma hinders epilepsy patients from help-seeking behavior and diminishes adherence to medical follow up. Despite this, the limitation of aggregate evidence regarding the prevalence and associated factors of perceived stigma existed in Ethiopia. Methods: Our search for eligible articles was accompanied in the databases of PubMed, Scopus, EMBASE, and African Index Medicus. An additional manual search by the enumeration of the reference list of other articles was done. We used stata-11 and random-effects models to obtain the average prevalence of perceived stigma. Heterogeneity was assured with the Higgs I2 test and Cochran's Q- statistic. Sensitivity analysis and sub-grouping analyses were done. Qualitative checks for symmetry in funnel plot and quantitative description of Egger's test were done to check a publication bias.Results: Of 1524 articles identified by the electronic and manual search, only 10 articles fulfilled the eligibility criteria and included in the analysis. The average prevalence of perceived stigma was found to be 43.79% (95% CI: 31.84, 55.75). The prevalence of perceived stigma varies across the measurement tool used, study location, and year of publication. The average prevalence of perceived stigma in studies measured with the modified FIS scale was 76.3%; higher than those studies measured with the Kilifi stigma scale; 33.63% and three-item stigma scale; 39.5%. Besides, the average prevalence of stigma was higher in Amhara (53.2 %) and the Oromia region (52.2%). Studies published before 2016 provide higher perceived stigma (56.26 %) than studies published after 2016 (35.47%). Our qualitative analysis revealed that rural residence, low level of education and income, age groups of 18 to 24 years, frequent seizure occurrence, contagion believe regarding epilepsy, duration of epilepsy from 1-10 years, depression and anxiety co-morbidity were among the associated factors with perceived stigma in people with epilepsy.Conclusion: The average prevalence of perceived stigma in people with epilepsy was high and varies across study location, measurement tool, and study's publication year. Management of epilepsy should focus on the stigma component besides the biological intervention.

A review study in Asia reported that rural residence, lower-income, and education, being female, not witnessing individuals having the disorder, contagious belief regarding epilepsy and mental illness co-morbidity were among the associated factors with perceived stigma (35,36). In another review of literature in Nigeria, the psychopathology associated with epilepsy is the main factor for perceived stigma in people with epilepsy (37). In Ethiopia too rural residence, poor income and education, young age, frequent seizure occurrence, contagion believe regarding epilepsy, longer duration of epilepsy, and presence of psychopathology was associated with perceived stigma (20,27,29,32,33).
The World Health Organization (WHO) has designed a Global Campaign against Epilepsy and its related stigma entitled "Out of the Shadows" initiative (38) so that the larger and wider population should stand against epilepsy and the related stigma. In Ethiopia however, there is a limitation of evidence on perceived stigma in people with epilepsy. Even, the little evidence obtained regarding perceived stigma is from cross-sectional studies that might not be strong evidence for policy recommendation. Therefore, looking for aggregate evidence on the prevalence and the associated factors of perceived stigma in people with epilepsy are signi cant and con dently useful for the design of programs for the management of perceived stigma and minimize the negative consequences. Thus, this meta-analysis study was conducted to (1) determine the average prevalence of perceived stigma in people with epilepsy in Ethiopia and (2) provide a narrative synthesis regarding the associated factors for perceived stigma in Ethiopia.

Methods
The Preferred Reporting Items for Systematic Reviews and Meta-analysis Protocols (PRISMA-P) guideline (39) was used as a reference during the study process.

Search strategy
We applied both an electronic as well as manual assessment for eligible articles. The search databases addressed in the electronic search were Scopus, PubMed, and EMBASE and African Index Medicus. These databases were explored for the existence of evidence on the subject of perceived stigma AND/OR associated factors for people living with epilepsy. Our search in the pub-med database was applied exhausting the keywords and headings of (prevalence OR Epidemiology OR Incidence OR Magnitude AND "stigma" OR perceived stigma AND epilepsy OR epileptic OR seizure OR seizure disorder AND "risk factor" OR determinant OR factor OR associated factors AND Ethiopia). The above search strategy was developed considering the PICO organization; extended using the MeSH terms and nally joined one another with the Boolean operators. The electronic search for data in EMBASE, Psych-Info, African Index Medicus, and Scopus was also accompanied in line with database-speci c searching strategies. A manual search for the reference lists of the included studies was also done. Grey literature was also addressed over the mentioned search databases and Google scholar. During the search process, we did not put time restrictions on the inclusion of articles.

Eligibility criteria's
We included articles published in the English language, focused on determining the prevalence of perceived stigma AND/OR the associated factors in people with epilepsy. Articles included were also expected to be conducted in the context of Ethiopia and must be case-control, cohort, and crosssectional in design. But by chance, all of the included studies were found to be cross-sectional. The studies included must also be conducted in adulthood population with an age of 18 years and above. We did not limit the year of publications of articles during our search procedure. We excluded earlier reviews and meta-analysis studies, studies in which the subjects were not human, editorials, and articles published in a language other than the English language. We performed our search in May 2020 and the search result was exported and stored in the endnote reference manager. Then the removal of duplication articles was done. Next to this, MN and MB independently screened the titles and abstracts of articles stored using the above-mentioned eligibility criteria. Again, these authors read the full text of the articles that were not excluded in the rst step and independently decided on the articles that had to be included for nal meta-analysis. Any variances in the notion concerning eligibility criteria between the above review authors were resolved by agreement and communication with the nal reviewer (SG).

Data extraction and quality assessment
After deciding the articles to be included in the study, MN and MB extracted the essential data from the 10 included articles in line with a welldeveloped, standardized data extraction template organized as suggested by PRISMA guidelines (39). The components of the data extraction template were author and year of the study, location of the study, sample size, study design, outcome variable, target population, measurement tool used, response rate, and sampling techniques used in the study. Using these templates the data was extracted from included studies and summarized in the form of a table. The modi ed Newcastle-Ottawa Scale (NOS) (40) had been used for the quality appraisal of included studies.
Statistical quality, ascertainment of cases representativeness of the sample, and comparability between participants were the dimensions of the NOS scale.

Data synthesis and analysis
The average prevalence of perceived stigma in people with epilepsy with its 95% CI was computed with Meta-XL version 5.3 (41) and Stata-11 Meta-prop package (42). The Cochran Q-statistics and Higgs I 2 statistics (43) were employed and interpreted to identify the between-study variance. The Higgs I 2 value of 50% and above during analysis indicates the existence of potential heterogeneity (43). In the present meta-analysis, we obtained a signi cant difference between the included studies. Therefore, it was necessary to mitigate the source of heterogeneity. Likewise, we did a subgrouping analysis based on the regional location of the study, year of publication, and the measurement tools used to quantity perceived stigma. Further screening for a single in uential study was also performed by leaving a single study out from the analysis at a time. We did also a screening for the presence of publication bias in the current study using a qualitative eyeball test (44) and quantitatively using the Eggers publication bias test. In this meta-analysis study a P-value < 0.05 was interpreted as statistically signi cant value.

Identi cation of studies
Our systematic electronic as well as manual investigation for articles regarding the prevalence AND/OR associated factors of perceived stigma yields a total of 1524 articles. However, only 10 of the studies have been included in the nal analysis after the remaining 1514 articles were excluded at the different stages of screening (Fig. 1).
Considering the measurement tool, one study used Jacoby stigma scale (32), two studies used modi ed Family Interview Schedule (FIS) (20,27), another two used three-item stigma scales (28,34), and four utilized Kili stigma scale (29)(30)(31)33) to measure perceived stigma in people with epilepsy. However, the assessment tool was not speci ed in one of the included studies. Among studies included; three (20,26,28) have no report regarding the response rate of the study participants. Eight of the included studies (27-34) used a systematic sampling technique and one study implemented a simple random sampling (26) and no sampling technique was reported in the remaining one study (20) (Table 1).

Quality of Included Studies
Overall the quality score of the 10 included studies ranges from 6 (26) to 10 (27). There of the included studies (26,29,33) had moderate quality, and the remaining six of the included studies (20,27,28,(30)(31)(32)34) had a good quality score. A study with poor quality was not present in this study (Additional le 1).

The average prevalence of perceived stigma in people living with epilepsy in Ethiopia
Ten cross-sectional studies (20,(26)(27)(28)(29)(30)(31)(32)(33)(34) that reported the prevalence of perceived stigma in people with epilepsy in Ethiopia were included in the current meta-analysis. The stated prevalence of perceived stigma among the included studies ranges from 31.1% in SNNP (26) to 81% in the Oromia region (20). The average estimated prevalence of perceived stigma among people with epilepsy in Ethiopia was 43.79% (95% CI: 31.84, 55.75). This average prevalence was subjected to a high degree of heterogeneity (I 2 = 99%, p-value < 0.001) from the difference between the involved studies (Fig. 2).
The prevalence of perceived stigma in people with epilepsy in Ethiopia based on the measurement tool, the regional setting of the study, and the year of publication.
Since the present meta-analysis study was with potential heterogeneity from the difference between included studies, our next step was performing a subgroup analysis to obtain the source for between-study differences. Therefore, we accomplished a subgroup analysis centered on measurement tools utilized to screen perceived stigma, study location, and year of publication of the studies.
Subgroup analysis based on the measurement tool used to screen perceived stigma Our sub-group analysis with measurement tool used to screen perceived stigma among patients with epilepsy revealed that the average prevalence of perceived stigma in studies that screened perceived stigma with modi ed FIS scale (20,27)  Subgroup analysis based on the regional setting where the study was conducted

Sensitivity analysis
We also wander for the source attributable to the difference between studies variance by employing one study leave out at a time sensitivity analysis. The result obtained showed that the average prevalence of perceived stigma ranges between 39.66% and 45.20% when every 10 studies were excluded at a time.
This was in between the 95% con dence interval of pooled prevalence of perceived stigma when all 10 studies were pooled together. Therefore, our meta-analysis result was not out-weighed by a single in uential study (Table 4).

Publication bias
Our qualitative visual assessment on funnel plot for the logit event rate of prevalence of perceived stigma among people with epilepsy against its standard error as illustrated in (Fig. 4) revealed that the slight presence of publication bias. This was, however, contrary to the quantitative statistical result of the Eggers publication bias test which showed the absence of potential publication bias (B= -74.9, SE = 1.94, P = 0.088).
A narrative description of the associated Factors for perceived stigma in people with epilepsy Five of the included studies had a report concerning the associated factors for perceived stigma among people living with epilepsy in Ethiopia (20,27,29,32,33). One of the studies from Addis Ababa reported that being divorced/widowed, living in rural areas, low-income level, and frequent seizure occurrence were signi cantly associated with perceived stigma (32). Another study from the same setting also reported age groups of 18 to 24 years, contagion believes regarding epilepsy and seizure-related injury were associated signi cantly with perceived stigma (29). However, in the Oromia region, a low level of education, duration of epilepsy from 1 to 10 years, depression, and anxiety co-morbidity were the associated factors with perceived stigma in people with epilepsy (33). On the other hand, duration of epilepsy 2 up to 5 years, duration of epilepsy ≥ s11 years, frequency of seizure; 1 up to 11 per year and ≥ 11 per year were the associated factors for perceived stigma in the Amhara region (27) ( Table 2).

Discussion
To date, the researchers of the current study assured that this review and meta-analysis on the average prevalence of perceived stigma and its associated factors in people living with epilepsy is the rst of its kind in the context of Ethiopia. Therefore, the result obtained from this metaanalysis on the prevalence of perceived stigma in people living with epilepsy will be signi cant evidence to clinicians, future researchers, and any other interested bodies planning to design interventional as well as administrative policy in the eld.
The average prevalence of perceived stigma in people living with epilepsy in Ethiopia was substantial (43.79%). This was lower than the prevalence of perceived stigma reported by a study that assessed more than 5,000 people living with epilepsy in 15 European countries where the reported prevalence of perceived stigma was 51% (14). The result of the present study was also lower than a study on systematic review and meta-analysis on the prevalence, associated factors, and impacts of personal stigma in patients with schizophrenia spectrum disorders whereby 64.5% of patients had perceived stigma (45). The long time difference between the present and earlier studies might be responsible for this. Besides, the loss of judgment and insight in most people with schizophrenia spectrum disorders might affect the higher prevalence of perceived stigma.
The average prevalence of perceived stigma in this study was however higher than the prevalence of perceived stigma in ten European countries where the overall prevalence of perceived stigma was 17% (14). Differences in cultural, economic, and associated attitudinal factors where the European countries are economically advanced and health care utilization in this setting might be high and reduces the perception of stigma.
The average prevalence of perceived stigma in people living with epilepsy varies according to the measurement tool used to assess' perceived stigma, the location of study, and year of publication.
A sub-group analysis of perceived stigma with measurement tool used to screen perceived stigma revealed that the average prevalence of perceived stigma in studies measured with modi ed FIS scale (20,27) was signi cantly higher than the average prevalence of perceived stigma in studies measured with Kili stigma scale and three-item stigma scale.
The modi ed FIS scale is a WHO-approved tool for screening perceived stigma. Also, its Amharic version is validated in Ethiopia so that that it can sensitively detect epilepsy patients with perceived stigma but in most of the included studies that utilized Kili stigma scale (29)(30)(31)33), the 66th percentile is used as the cut-off point that might not be as such sensitive to detect perceived stigma. Moreover, the small number of studies included in the sub-group of the FIS scale might result in an overestimation of the result and gives a higher prevalence of perceived stigma than studies assessed with the Kili stigma scale.
Furthermore, the average prevalence of perceived epilepsy stigma was signi cantly higher in studies from the Amhara region (27,30) and Oromia region (20,33,34) than the pooled prevalence of perceived epilepsy stigma in Addis Ababa (28,29,32). The population living in Addis Ababa, the capital city of Ethiopia is relatively developed in terms of knowledge, social, and cultural views. Therefore they might have a better attitude regarding chronic illness in general and epilepsy, in particular that could probably reduce the average perceived stigma than the population in the Amhara and Oromia region. Further to this, both Amhara and Oromia regions are the two most populated regions in Ethiopia so that the health care coverage of these populous regions is still low which might contribute to the high average perceived stigma in the regions.
We also did a segmental analysis of the prevalence of perceived stigma on the ground of year of publication of the included studies and studies published before 2016 (20,27,29,34) provide higher average perceived stigma (56.26%) than studies published in and after 2016 (26,28,(30)(31)(32)(33) (35.47%). This might be because increased awareness of people towards epilepsy is expected as the country's economic development is advancing through time that lowers the perceived stigma. Further to this the relatively few studies included in the sub-group of studies published before 2016 as compared to the sub-group of studies published in and after 2016 could minimize the precision and result in an overestimation of perceived stigma.
Relating to the factors associated with perceived stigma in people living with epilepsy, a narrative analysis revealed that being divorced/widowed, living in rural areas, low level of education, low-income level, age groups of 18 to 24 years, frequent seizure occurrence, contagion believe regarding epilepsy, seizure-related injury, duration of epilepsy from 1 to 10 years, depression and anxiety co-morbidity were the associated factors with perceived stigma in people with epilepsy in Ethiopia.
This was supported by earlier review studies abroad, whereby rural residence, lower-income, and educational status, being female, and single, not witnessing individuals having the disorder, contagious believe regarding epilepsy and mental illness co-morbidity were reported to be associated with perceived stigma in epilepsy (35,36). However, a review of literature in Nigeria had shown that people with epilepsy are felt stigmatized due to the psychopathology associated with epilepsy (37).

The difference between studies included in the meta-analysis
This review and meta-analysis study was prejudiced by a great heterogeneity from the difference between analyzed studies. Therefore consideration of sub-group and sensitivity analysis was instigated. The analysis of the sub-group was done under the assessment instrument for perceived stigma, regional setting, and year of publication of studies. All the above three dimensions of subgroup provide subtle evidence from where was the difference between studies arise. Even though a sensitivity analysis was done, there was no implication of a single study outweighing the average prevalence of perceived stigma in people with epilepsy in Ethiopia.
In using the results of the present meta-analysis study, we should take into consideration its shortcomings. Primarily, a few numbers of studies have been pooled in the subgroup analysis based on measurement tool, study location, and publication years with the possibility of in uence on estimate precision. Further to this, the occurrence of great heterogeneity in the pooled prevalence of perceived stigma was also a constraint to the plausibility of the study. Finally, a sole narrative analysis of the associated factors for perceived stigma in individuals living with epilepsy still requires further exploration of the average effect size of associated factors by future researchers. The shortage of earlier meta-analysis studies in people living with epilepsy has also made the comparison the results of the present study di cult and should be taken into consideration.

Conclusion
The pooled prevalence of perceived stigma in people living with epilepsy in Ethiopia was substantial (43.79%) and was under the effect of the high difference between studies analyzed. A sub-group analysis discovered that the average prevalence of perceived stigma was signi cantly higher in studies measured with a modi ed FIS scale than the average prevalence of perceived stigma in studies measured with the Kili stigma scale and three-item stigma scales. In addition to this, the average prevalence of perceived epilepsy stigma was much higher in studies from Amhara and Oromia regions than the prevalence of perceived epilepsy stigma in Addis Ababa and SNNP.
Moreover, studies published earlier to 2016 provide higher pooled perceived stigma than studies published in and after 2016. Even though a sensitivity analysis was done, none of the included studies was outweighing the overall study result. Qualitative analysis revealed that being divorced/widowed, living in rural areas, low level of education, low-income level, age groups of 18 to 24 years, frequent seizure occurrence, contagion believe regarding epilepsy and seizure-related injury, duration of epilepsy from 1 to 10 years, depression and anxiety co-morbidity were among the associated factors with perceived stigma in people with epilepsy in Ethiopia. This pooled evidences regarding perceived stigma in people living with epilepsy in Ethiopia plays a signi cant role as it disclosed substantial evidence for future interventional accomplishments in the eld. Great emphasis should, therefore, be given to the aforementioned ndings in the management procedures for people with epilepsy.