Methodological assessment

Identification of Paragonimus in the literature was generally undertaken by microscopy, i.e., the microscopic identification of eggs or metacercariae. As only 4 studies used molecular analysis and/or serology [29,39,53,54], here identification was therefore accepted on microscopy only. This criterion was applied even if the result was contrary to the diagnosis in the paper, e.g., a positive case based on serology results or symptoms only was recorded here as unconfirmed. This assessment was made for consistency and to ensure that only current infections were identified. In cases where Paragonimus eggs were isolated from stool only, any available photographs were assessed, and evidence of positive identification of other parasite species checked to rule out the possibility of misidentification of another trematode.

In human surveys, publications were excluded from further analysis if the study population was selected from people pre-determined as likely infection candidates, for example hospital surveys of pulmonary clinic patients. A publication was also excluded if the total study population was not given. In some papers, study populations were identified and subsequently only those individuals presenting symptoms were recruited for the survey. These publications were included as the original study population total was also given. Both cross-sectional (community-wide) and school-based surveys were included. For mammal surveys also, studies were excluded if total study population was not provided.

All data were analysed in R, version 3.5.3 (2019-03-11) "Great Truth". The majority of the literature in the systematic review was highly heterogeneous and a qualitative review was considered the best approach for analysis. A subset of publications was also selected for a comparison of prevalence and intensity across studies, separately for each host category. Averages and standard deviation values were computed in R, using the tidyverse package [55]. Weighted means were computed, for example, egg measurement averages in a given study weighted by total eggs measured; average intensity, weighted by total number of infected hosts (per host-category) per study, and in prevalence, by total surveyed for the study (Table 1). The statistical significance of differences across studies between groups, for example in prevalence and intensity of infected hosts were tested with a Wilcoxon rank sum test, which does not assume a normal distribution, given the high heterogeneity and variance among studies. All statistical tests were conducted with the significance level α = 0.05 for rejecting null hypotheses.

Key: Countries: C = Cameroon, G = Gabon, L = Liberia, EG = Equatorial Guinea, Co = Côte d’Ivoire, Gh = Ghana, N = Nigeria, T = Tanzania, Cat. = Host Category: H: Human, C: Freshwater Crab, M: Mammal, Inf. = total infected, Int. = intensity, Host = Host species (or category for mammals), Prev. = prevalence, SD = standard deviation, N = total surveyed, left of bar: totals for prevalence, right of bar, totals for intensity data, Refs = total number of publications, Mammals* = species of mammal surveyed, those with individuals positive for infection in bold, those not infected, in brackets.

Prevalence levels were recorded for all hosts: human/other mammals, and intermediate freshwater crab hosts. Prevalence for crabs was calculated by species, excluding Callinectes marginatus, regarded as an unconfirmed host [56;NC, pers. comm.]. Prevalence differences by demographics (sex and age) were also examined. For sex, both overall total and proportion of female/male patients positive were recorded, and whether any difference was significant, if reported. For age data, the defined age categories, and proportion of prevalence per category were recorded (S2 Table).

Range and average for intensity of infection (egg/metacercariae count per host) were recorded, and the average over the study calculated if not reported (S3 Table). Generally, eggs were recorded per 5 ml of sputum (e/5g), or eggs per gram of stool (epg). In one case where egg counts were reported for 1ml sputum samples, this was recorded and counts adjusted to per 5ml to allow for comparability. In another publication, sputum samples of 1-5ml were collected, and this study was excluded from analysis. The sputum and stool sample preparation methods for human and mammal surveys, details on crab dissection and screening for infection, and egg/metacercariae counts per host were recorded. If multiple samples were collected per host and the average reported, this was recorded also. If intensity was reported only by level (in 5ml sputum), these totals were not included as direct comparison was not possible with average egg count totals. Egg count measurements from stool (in addition to sputum) as reported in 3 studies were excluded for consistency. No experimental infections were included in the subset analysis for intensity or prevalence, but measurements of eggs isolated from experimental hosts were included in egg measurement analysis and the host details recorded (S4 Table).