Table 3. The relative effect of different mosquito sampling strategies on molecular xenomonitoring prevalence and evaluations of the certainty of the evidence.
| Comparison | Relative effect, prevalence ratio (95% CI) | No. of mosquitoes (studies) | Quality of the evidence | Comments |
|---|---|---|---|---|
| Comparison 1: collection methods | ||||
| Fed versus gravid collection methods | 1.54 (0.46–5.16) | 12 711 (3) | Very lowa,b,c | Downgraded for imprecision, inconsistency and indirectness |
| Fed versus unfed collection methods | 3.53 (1.52–8.24) | 5 167 (2) | Very lowd,e | Downgraded for indirectness and risk of bias |
| Gravid versus unfed collection methods | 0.20 (0.01–3.40) | 5 927 (1) | Very lowf,g | Downgraded for serious imprecision and indirectness |
| Comparison 2: sampling intensity | ||||
| 300 versus 150 trapping locations | 0.89 (0.54–1.46) | 29 797 (1) | Very lowg | Downgraded for indirectness |
| 300 versus 75 trapping locations | 0.90 (0.54–1.51) | 28 624 (1) | Very lowg | Downgraded for indirectness |
| 150 versus 75 trapping locations | 1.01 (0.60–1.68) | 27 921 (1) | Very lowg | Downgraded for indirectness |
| Comparison 3: mosquito genera | ||||
| Anopheles versus Culex mosquitoes: | ||||
| Anopheles areas | 6.91 (1.73–27.52) | 28 974 (6) | Very lowb,h | Downgraded for inconsistency and imprecision |
| Culex areas | 2.68 (0.08–94.93) | 53 610 (2) | Very lowa | Downgraded for imprecision |
| Aedes versus Culex mosquitoes | 1.07 (0.52–2.19) | 64 705 (2) | Very lowa,i | Downgraded for imprecision and indirectness |
CI: confidence interval.
a Downgraded for imprecision: the 95% CIs include both no difference and a large difference.
b Downgraded for inconsistency: there was substantial heterogeneity that could not be explained by subgroup analyses.
c Downgraded for indirectness: all studies were conducted in areas where the primary filariasis vectors belonged to the Anopheles gambiae species complex; the findings may not apply to other transmission areas.
d Downgraded for indirectness: over 92% of the weight of the meta-analysis came from a single study; the findings may not apply to other transmission areas.
e Downgraded for risk of bias: the sampling methods for the main study contributing to the analysis were judged to have a high risk bias as mosquitoes collected using fed traps were screened in smaller pool sizes than those collected by gravid traps; this may favour higher molecular xenomonitoring prevalence for mosquitoes from fed collection methods.
f Downgraded twice for serious imprecision: the 95% CIs include large differences in favour of each type of collection method.
g Downgraded for indirectness: a single study contributed to the meta-analysis; the findings may therefore not apply to other transmission areas.
h Downgraded for imprecision: although the 95% CIs are extremely wide and include both a small difference and a very large difference.
i Downgraded for indirectness: studies were conducted in American Samoa and Samoa; the findings may not apply to other transmission areas.
Note: We used the Grading of Recommendations Assessment, Development and Evaluation approach for grading the quality of the evidence.22