Abstract
To confirm our earlier evidence of a temporal and dose–response relationship between onchocerciasis and epilepsy, we conducted another cohort study in a different setting in Cameroon. Individuals whose Onchocerca volvulus microfilarial density (Ov-MFD) was measured in 1992–1994 when they were children were revisited in 2019 to determine if they acquired epilepsy. With reference to individuals with no microfilariae in 1992–1994, the relative risks of acquiring epilepsy were 0.96, 2.76, 3.67, and 11.87 in subjects with initial Ov-MFD of 1–7, 8–70, 71–200, and > 200 microfilariae per skin snip, respectively. This study further demonstrates reproducibility using the Bradford Hill’s criteria for causality.
Keywords: onchocerciasis, epilepsy, causal relationship, Africa, cohort
Eighty percent of the 50–70 million people with epilepsy (PWE) worldwide are found in low- and middle-income countries [1, 2]. This over-representation of PWE in the general population is particularly patent in Central Africa, where the prevalence is estimated at 59.7 per 1000 people [3]. Furthermore, a meta-analysis showed that the median incidence rate of epilepsy worldwide is 50.4 per 100 000 persons-years, whereas values recorded in Sub-Saharan Africa range between 64 and 187 per 100 000 person-years [3].
A possible association between onchocerciasis and epilepsy was originally suggested in 1938 in Mexico [4]. Several studies and meta-analyses subsequently demonstrated a significant relationship between the 2 diseases [5, 6], even after adjusting for other risk factors and infections [7, 8]. Consequently, the concept of onchocerciasis-associated epilepsy (OAE) was proposed, and in 2015, 381 000 people were estimated to have OAE [9].
A longitudinal study conducted in 2017 in the Mbam valley onchocerciasis focus of Cameroon demonstrated that the incidence of epilepsy was positively correlated with the intensity of Onchocerca volvulus infection at a young age (5–10 years old) [10]. This first cohort study provided evidence for 2 of the main Bradford Hill criteria, supporting causality between onchocerciasis and epilepsy (temporality and biological gradient) but had to be replicated in another setting to provide evidence for another criterion, namely consistency (reproducibility). In the present study, we used a design similar to that used in the Mbam valley to evaluate whether the level of infection with O. volvulus measured in 1992–1994 in children aged 5–15 years living in the Lékié Division (Center Region, Cameroon) was associated with an increased risk of subsequently developing epilepsy.
METHODS
Initial Parasitological Surveys and Selection of Subjects for the 2019 Survey
Between 1992 and 1994, parasitological surveys were conducted in 18 villages of the Lékié Division to measure the levels of O. volvulus infection in individuals aged ≥5 years (Table 1; Supplementary Figure 1). Two skin snips were collected from each volunteer using a Holth-type corneoscleral punch and incubated in saline at room temperature for 24 hours. Emerged microfilariae (mf) were counted using a microscope, and the individuals’ microfilarial density (MFD), expressed as mf per snip, were calculated using the arithmetic mean of the counts. Community microfilarial load (CMFL), defined as the Williams geometric mean of the MFD in subjects aged ≥20 years, was calculated for each village. In addition, standardized thick blood smears were prepared between 10:00 and 16:00 to measure the participants’ Loa loa and Mansonella perstans MFD (mf/mL). All 18 communities surveyed in 1992–1994 were re-visited in 2019. Since the average age of first seizure in subjects with OAE is between 10 and 14 years old [11, 12], we sought information for all those 1258 individuals who were 5–15 years old during the baseline surveys.
Table 1. .
Data by Village
| Health District | Village | Subjects Targeted | Subjects Retrieved in 2019 | No. (%) SR With O. volvulus mfa | CMFLa | Loa loa prevalencea | No. (%) of SR with L. loa mfa | Mansonella perstans prevalencea | No. (%) SR with M. perstansa | PY | No. SCE | IR (per 100 000 PY) | 95% CI | CDTI Start Year |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ebebda | Djouniat | 40 | 37 | 34 (91.9) | 32.4 | 13.1 | 1 (2.7) | 30.3 | 2 (5.4) | 972.2 | 1 | 102.9 | 14.5–730.2 | 1999 |
| Ebebda | Eyene | 85 | 73 | 56 (76.7) | 61.8 | 10.3 | 1 (1.4) | 17.4 | 2 (2.7) | 1981.2 | 0 | 0.0 | N.A. | 1999 |
| Ebebda | Mbenega | 70 | 50 | 47 (94.0) | 53.1 | 4.0 | 1 (2.0) | 8.0 | 3 (6.0) | 1208.7 | 10 | 827.3 | 445.2–1537.6 | 1999 |
| Ebebda | Nega Lendong | 155 | 111 | 90 (81.1) | 29.3 | 21.1 | 8 (7.2) | 10.2 | 5 (4.5) | 2873.6 | 7 | 243.6 | 116.1–511.0 | 1999 |
| Elig Mfomo | Elig Mfomo | 31 | 23 | 11 (47.8) | 2.5 | 31.2 | 4 (17.4) | 32.6 | 2 (8.7) | 573.9 | 2 | 348.5 | 57.2–1393.4 | 1999 |
| Evodoula | Nkolakok | 42 | 35 | 26 (74.3) | 17.6 | 27.3 | 2 (5.7) | 28.1 | 2 (5.7) | 864.0 | 1 | 115.7 | 16.3–821.6 | 1994 |
| Evodoula | Nkolassa | 50 | 26 | 24 (92.3) | 53.8 | 15.7 | 2 (7.7) | 38.6 | 6 (23.1) | 612.3 | 4 | 653.3 | 245.2–1740.7 | 1994 |
| Evodoula | Nkolmeyos I | 35 | 27 | 24 (88.9) | 118.4 | 16.7 | 1 (3.7) | 40.5 | 5 (18.5) | 663.9 | 1 | 150.6 | 21.2–1069.3 | 1994 |
| Monatélé | Nkongmessa | 42 | 23 | 21 (91.3) | 81.3 | 9.6 | 1 (4.4) | 4.1 | 0 | 449.1 | 10 | 2226.5 | 1198.0–4138.1 | 1994 |
| Obala | Nkolfep | 82 | 51 | 4 (7.8) | 0.3 | 31.9 | 7 (13.7) | 3.5 | 0 | 1259.6 | 1 | 79.4 | 11.2–563.6 | 1999 |
| Okola | Ayos | 81 | 73 | 35 (48.0) | 5.7 | 24.9 | 11 (15.1) | 31.4 | 6 (8.2) | 1847.8 | 0 | 0.0 | N.A. | 1999 |
| Sa’a | Lebamzip I | 46 | 27 | 4 (14.8) | 1.3 | 33.1 | 2 (7.4) | 3.1 | 0 | 729.7 | 0 | 0.0 | N.A. | 1999 |
| Sa’a | Nkolbogo I | 70 | 65 | 26 (40.0) | 3.8 | 25.8 | 5 (7.7) | 28.1 | 10 (15.4) | 1735.5 | 0 | 0.0 | N.A. | 1999 |
| Sa’a | Nkolbogo II | 63 | 58 | 38 (65.5) | 22.8 | 26.1 | 8 (13.8) | 54.5 | 29 (50.0) | 1541.5 | 2 | 129.7 | 32.4–518.8 | 1999 |
| Sa’a | Nkolebassimbi | 73 | 53 | 40 (75.5) | 24.8 | 13.4 | 1 (1.9) | 11.4 | 2 (3.8) | 1419.1 | 1 | 70.5 | 9.9–500.3 | 1999 |
| Sa’a | Nkolntsa | 104 | 44 | 13 (29.6) | 2.3 | 28.5 | 6 (13.6) | 8.0 | 0 | 908.0 | 10 | 1101.3 | 592.6–2046.9 | 1999 |
| Sa’a | Nkolossang | 62 | 47 | 10 (21.3) | 3.3 | 26.4 | 5 (10.6) | 18.2 | 3 (6.4) | 1206.0 | 0 | 0.0 | N.A. | 1999 |
| Sa’a | Ntsan Mendouga | 127 | 99 | 52 (52.5) | 10.8 | 18.2 | 6 (6.1) | 13.8 | 3 (3.0) | 2441.3 | 3 | 112.9 | 39.6–381.0 | 1999 |
Abbreviations: CDTI, start year of community-directed treatment with ivermectin; CMFL, O. volvulus community microfilarial load (in mf/snip) in the village during the baseline parasitological survey; IR, incidence rate (number of cases per PY); Ov, Onchocerca volvulus; PY, person-years; SCE, suspected cases of epilepsy (number); SR, Subjects Retrieved.
aAssessed in 1992–1994.
Evaluation and Definition of Epilepsy
In November 2019, we investigated, with the help of key informants (village authorities, long-standing residents, health workers), whether the selected subjects were still alive and which ones had developed epilepsy. Once identified, we visited, with the assistance of local health workers or local authorities, the household of each selected individual or that of their relatives. If the targeted person was not at home, we asked his/her relatives whether he/she was still alive. If the subject had died, the year of death was recorded. For all individuals, a standardized 5-item questionnaire [13] was used to identify “suspected cases of epilepsy” (SCE). SCE was identified when a positive answer was given to at least 1 of the 5 questions. Interviewers had no information on the individuals’ MFD measured during the initial parasitological survey. When no information could be obtained from the families, we used the responses from the key informants to define the SCE status. This study was approved by the Cameroon National Ethics Committee for Research in Human Health (registration number 2018/12/1123/CE/CNERSH/SP).
Statistical Analyses
The variable of interest was SCE status. Independent variables were gender, age during the initial survey (5, 6–7, 8–9, 10–11 [reference group], 12–13, or 14–15 years), MFD (0 [reference group], 1–7, 8–70, 71–200, and > 200 mf/snip), the presence of blood L. loa mf (negative vs positive), the presence of M. perstans mf, CMFL measured in 1992–1994 in the subject’s village of residence (<4, 4–19, 20–29, and ≥30 mf/snip), and the start year of community-directed treatment with ivermectin (CDTI) for onchocerciasis control in the health district (HD; 1994 or 1999). HDs are health administrative units in charge of the implementation of health programs.
Data (individual duration of follow-up) concerning individuals who were not identified as SCE and who died between the initial survey and 2019 were censored by the year of death if it was known, or at half-time of the follow-up period if it was not known. Data concerning the SCE were censored at half-time of the follow-up period if the patient was alive or at half-time of the period between baseline and the year of death if declared dead. Incidence rates (IRs) were estimated by dividing the number of SCE by the total number of person-years of follow-up. Then, in order to assess individual risk factors associated with SCE (incidence rate ratios [IRRs]), we performed a multivariate Poisson regression model including all the independent variables mentioned above.
All possible and relevant interactions as well as random effects on the HD were assessed using likelihood ratio tests. All analyses were performed with Stata (version 14.0).
RESULTS
Population Interviewed and Incidence Rates
In 2019, information on SCE could be obtained for 922 of the 1258 targeted subjects (73.3%). The mean follow-up period for these 922 subjects was 25.2 years, and the number of person-years of follow-up was 23 287. Fifty-three SCE were identified, including 45 through questionnaires applied directly to the person or his/her relatives, and 8 were identified through questionnaires applied to key informants. The overall IR of epilepsy was 53/23 287 (227.6 per 100 000 person-years). The IR increased gradually with the initial MFD, with values ranging from 117.8 per 100 000 persons-years for individuals without skin mf to 952.8 per 100 000 persons-years for those with an MFD >200 mf/snip (P < .0001) (Table 2).
Table 2. .
Population Study, Follow-up Data, Incidence Rates, and Incidence Rate Ratios
| No. Examined in Baseline Study (% of Total) | No. With Information Collected in 2017 (% of Total) | SCE, No. | PY | IR (per 100 000 PY) (95% CI) | P a | Model Without Random-Effect | Model With a Random-Effect at HD Level | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| aIRRb (95% CI) | P | aIRRc (95% CI) | P | ||||||||
| Total | 1258 | 922 | 53 | 23 287.4 | 227.6 (173.9–297.9) | ||||||
| Age, y | 5 | 95 (7.6) | 59 (6.4) | 7 | 1445.8 | 484.1 (230.8–1015.6) | .380 | 3.60 (2.19–5.92) | <.0001 | 3.39 (1.25–9.14) | .016 |
| 6–7 | 273 (21.7) | 182 (19.7) | 6 | 4673.4 | 128.4 (57.7–285.8) | 0.83 (0.37–1.85) | .645 | 0.74 (0.26–2.09) | .572 | ||
| 8–9 | 222 (17.7) | 169 (18.3) | 8 | 4285.5 | 186.7 (93.4–373.3) | 1.03 (0.35–3.31) | .956 | 1.07 (0.42–2.74) | .889 | ||
| 10–11 | 273 (21.7) | 212 (23.0) | 10 | 5410.4 | 184.8 (99.4–343.5) | Ref | |||||
| 12–13 | 235 (18.7) | 182 (19.7) | 12 | 4556.2 | 263.4 (149.6–463.8) | 1.06 (0.54–2.07) | .868 | 0.99 (0.42–2.36) | .991 | ||
| 14–15 | 160 (12.7) | 118 (12.8) | 10 | 2916.0 | 342.9 (184.5–637.4) | 1.19 (0.436–3.86) | .775 | 1.17 (0.47–2.90) | .735 | ||
| Sex | Female | 637 (50.6) | 449 (48.7) | 25 | 11 216.6 | 222.9 (150.6–329.9) | .892 | Ref | Ref | ||
| Male | 621 (49.4) | 473 (51.3) | 28 | 12 070.8 | 232.0 (160.2–336.0) | 0.75 (0.33–1.73) | .505 | 0.78 (0.44–1.38) | .388 | ||
| CMFL, mf/snip | <4 | 395 (31.4) | 257 (27.9) | 13 | 6412.7 | 202.7 (117.7–349.1) | .004 | Ref | Ref | ||
| 4–19 | 250 (19.9) | 207 (22.4) | 4 | 5153.1 | 77.6 (29.1–206.8) | 0.24 (0.04–1.36) | .106 | 0.29 (0.08–1.02) | .054 | ||
| 20–29 | 291 (23.1) | 222 (24.1) | 10 | 5834.1 | 171.4 (92.2–318.6) | 0.32 (0.07–1.47) | .143 | 0.20 (0.10–0.93) | .036 | ||
| ≥30 | 322 (25.6) | 236 (25.6) | 26 | 5887.5 | 441.6 (300.7–648.6) | 0.51 (0.09–2.86) | .447 | 0.49 (0.15–1.66) | .251 | ||
| Ov in skin snip | Negative | 524 (41.7) | 367 (39.8) | 11 | 9341.8 | 117.8 (65.2–212.6) | .004 | ||||
| Positive | 734 (58.43) | 555 (60.2) | 42 | 13 945.6 | 301.2 (222.6–407.5) | ||||||
| MFD, mf/snip | 0 | 524 (41.7) | 367 (39.8) | 11 | 9341.8 | 117.8 (65.2–212.6) | <.001 | Ref | Ref | ||
| 1–7 | 241 (19.2) | 182 (19.7) | 4 | 4638.1 | 86.2 (32.4–229.8) | 0.96 (0.38–2.43) | .937 | 0.93 (0.28–3.07) | .909 | ||
| 8–70 | 245 (19.5) | 178 (19.3) | 10 | 4558.8 | 219.4 (118.0–407.7) | 2.76 (1.20–6.35) | .017 | 2.66 (0.98–7.19) | .054 | ||
| 71–200 | 129 (10.3) | 100 (10.9) | 7 | 2544.8 | 275.1 (131.1–577.0) | 3.67 (1.86–7.21) | <.001 | 3.29 (1.01–10.71) | .048 | ||
| >200 | 119 (9.5) | 95 (10.3) | 21 | 2203.9 | 952.8 (621.3–1461.4) | 11.87 (5.56–25.33) | <.001 | 11.60 (3.89–34.61) | <.0001 | ||
| Loa loa, mf/mL | Negative | 1077 (85.6) | 796 (86.3) | 47 | 20 046.7 | 234.5 (176.2–312.0) | .484 | Ref | Ref | ||
| Positive | 99 (7.9) | 72 (7.8) | 2 | 1848.7 | 108.2 (27.1–432.6) | 0.46 (0.14–1.55) | .211 | 0.46 (0.11–1.96) | .292 | ||
| Missing | 82 (6.5) | 54 (5.9) | 4 | 1391.9 | 287.4 (107.9–765.7) | 1.46 (0.75–2.83) | .265 | 1.52 (0.48–4.77) | .476 | ||
| Mansonella perstans, mf/mL | Negative | 1074 (85.4) | 788 (85.5) | 43 | 19 857.7 | 216.5 (160.6–292.0) | .679 | Ref | Ref | ||
| Positive | 102 (8.1) | 80 (8.7) | 6 | 2037.7 | 294.4 (132.3–655.4) | 0.93 (0.31–2.75) | .896 | 1.15 (0.45–2.97) | .770 | ||
| Missing | 82 (6.5) | 54 (5.9) | 4 | 1391.9 | 287.4 (107.9–765.7) | N.A. | N.A. | ||||
| CDTI | 1994 | 169 (13.4) | 111 (12.0) | 16 | 5589.4 | 617.9 (378.6–1008.6) | <.001 | Ref | Ref | ||
| 1999 | 1089 (86.6) | 811 (98.0) | 37 | 20 698.0 | 178.8 (129.5–246.7) | 0.63 (0.11–3.80) | .617 | 0.49 (0.13–1.75) | .270 |
Abbreviations: aIRR, adjusted incidence rate ratio; CDTI, start year of community-directed treatment with ivermectin; CI, confidence interval; CMFL, community microfilarial load (in mf/snip); HD, health district; IR, incidence rate (number of cases per PY); MFD, individual microfilarial density (in mf/snip); Ov, Onchocerca volvulus mf; PY, persons-years; SCE, suspected cases of epilepsy (number).
a P values were calculated within each variable and assessed using the log-rank test for sex, skin snip positivity, Loa loa microfilariae positivity, Mansonella perstans positivity, CDTI, and the trends modified log-rank test for age, CMFL, and skin snip in 5 categories of variables.
bMultivariate logistic model with a cluster-robust standard errors to account for possible intra-community clustering.
cMultivariate logistic model with a random effect at the HD level.
Individual Risk Factors Associated With SCE
The risk of being identified as an SCE in 2019 was higher for those who were 5 years old during the initial surveys than for the 10-year-olds (adjusted IRR [aIRR], 3.60; P < .0001). There was no difference between genders or between CMFL categories (after adjustment for individual MFD). The risk increased gradually with the individual MFD: aIRRs for individuals with 1–7, 8–70, 71–200, and >200 mf/snip were 0.96 (P = .937), 2.76 (P = .017), 3.67 (P < .001), and 11.87 (P < .001), respectively. No significant association was found between SCE status and presence of L. loa or M. perstans microfilaremia, or start year of CDTI. No interactions between the covariates were found. Inclusion of a random effect at the HD level (P = .090) did not affect the strength and significance of the effect of the MFD on SCE status.
DISCUSSION
This study confirms the results obtained in the neighbouring Mbam division [10]: a temporal relationship between onchocerciasis and epilepsy and a dose–effect relationship with a risk of developing epilepsy increasing gradually with the MFD during childhood. This study also confirms the absence of any gender effect but an increased risk for the younger children, all other parameters being equal. Lowering the minimal age of inclusion in CDTI (presently 5 years) should be considered.
Considering that all villages in a given HD benefitted from similar CDTI-related activities, the random effect was evaluated at the HD level (not the village level). As this random effect was close to significance, the risk of developing epilepsy could vary slightly between the HDs.
Unexpectedly, the start year of CDTI was not associated with SCE incidence. However, one should consider that the HDs where CDTI started in 1994 had the highest onchocerciasis endemicity levels. Although there is a possible lack of statistical power, one may consider that the first years of CDTI in these HDs had little impact on the intensity of transmission and/or that children were less treated than adults during the first years of CDTI implementation. As individual history of treatment is lacking, it is impossible to support one hypothesis or the other.
As in the previous study [10], our current study presents these possible biases: (i) the possible misclassification of SCE for individuals with a history of provoked seizures, (ii) the absence of control for other possible risk factors (eg, cysticercosis), (iii) and the absence of confirmation of epilepsy by detailed neurological examination of the SCE.
Nevertheless, this study provides significant new findings, investigating for the first time the possible role of L. loa and M. perstans in inducing epilepsy and demonstrating that it was not the case.
In conclusion, this study supports the previous findings and consequently adds the reproducibility principle to the Bradford Hill’s criteria [14], supporting the causal nature of the relationship between O. volvulus infection and epilepsy.
Supplementary Data
Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Acknowledgments
We are grateful to local authorities for the help and their support. We are also grateful to the populations for their invaluable help to collect the information.
Financial support. The collection of baseline data was funded by Helen Keller International. The 2019 survey was funded by the “Institut de Recherche pour le Développement.” The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Potential conflicts of interest. The authors declare no conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1. Ngugi AK, Bottomley C, Kleinschmidt I, et al. . Estimation of the burden of active and life-time epilepsy: a meta-analytic approach. Epilepsia 2010; 51:883–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Newton CR, Garcia HH. Epilepsy in poor regions of the world. Lancet 2012; 380:1193–201. [DOI] [PubMed] [Google Scholar]
- 3. Ba-Diop A, Marin B, Druet-Cabanac M, et al. . Epidemiology, causes, and treatment of epilepsy in sub-Saharan Africa. Lancet Neurol 2014; 13:1029–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Reddy SG, Sacre CG, Das L, Pani S. El sindrome epileptico y su relacion con onchocercosis. Bol Salub Mex 1938; 1:11–31. [Google Scholar]
- 5. Kaiser C, Pion SD, Boussinesq M. Case-control studies on the relationship between onchocerciasis and epilepsy: systematic review and meta-analysis. PLoS Negl Trop Dis 2013; 7:e2147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Pion SD, Kaiser C, Boutros-Toni F, et al. . Epilepsy in onchocerciasis endemic areas: systematic review and meta-analysis of population-based surveys. PLoS Negl Trop Dis 2009; 3:e461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Kamuyu G, Bottomley C, Mageto J, et al. ; Study of Epidemiology of Epilepsy in Demographic Sites (SEEDS) group Exposure to multiple parasites is associated with the prevalence of active convulsive epilepsy in sub-Saharan Africa. PLoS Negl Trop Dis 2014; 8:e2908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Robertson D, Ngugi AK, Bottomley C, et al. . Prevalence of active convulsive epilepsy in sub-Saharan Africa and associated risk factors: cross-sectional and case-control studies. Lancet Neurol 2013; 12:253–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Vinkeles Melchers NVS, Mollenkopf S, Colebunders R, et al. . Burden of onchocerciasis-associated epilepsy: first estimates and research priorities. Infect Dis Poverty 2018; 7:101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Chesnais CB, Nana-Djeunga HC, Njamnshi AK, et al. . The temporal relationship between onchocerciasis and epilepsy: a population-based cohort study. Lancet Infect Dis 2018; 18:1278–86. [DOI] [PubMed] [Google Scholar]
- 11. Boullé C, Njamnshi AK, Dema F, et al. . Impact of 19 years of mass drug administration with ivermectin on epilepsy burden in a hyperendemic onchocerciasis area in Cameroon. Parasit Vectors 2019; 12:1– 13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Siewe JFN, Ngarka L, Tatah G, et al. . Clinical presentations of onchocerciasis-associated epilepsy (OAE) in Cameroon. Epilepsy Behav 2019; 90:70–8. [DOI] [PubMed] [Google Scholar]
- 13. Diagana M, Preux P, Tuillas M, Ould Hamady A, Druet-Cabanac M. Dépistage de l’épilepsie en zones tropicales: validation d’un questionnaire en Mauritanie. Bull Soc Pathol Exot 2006; 99:103–7. [PubMed] [Google Scholar]
- 14. Hill AB. The environment and disease: association or causation? J Roy Soc Med 1965; 58:295–300. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.