LAUSR

2190 www.thelancet.com Vol 389 June 3, 2017 inverse variance­weighted generalised linear models, significant heterogeneity still existed in these quasi­experimental studies without controls, leading to wide confidence intervals around odds ratios (ORs). The baseline prevalence seems to have an opposite effect on prevalence reduction of roundworms (OR 2·7; 95% CI 0·03–239·7) and hookworms (0·07; 0·01–0·77). Besides, because low response rate was the most common potential deficiency assessed in terms of risk of bias, we felt that it was inappropriate to include studies with very low coverage. Given the facts, results from cluster­randomised controlled trials are urgently required. Second, the concerns about whether anthelmintic resistance is likely to occur in human STH infections and whether alternative strategies (eg, novel anthelmintics or combination ther apy) have been explored also need serious reassessment. The scarce literature reporting reduced efficacy of anthelmintics in human STH infection does not yet provide conclusive evidence of drug resistance; however, the warning signs are clear. Exclusive reliance on benzimidazole anthelmintics might put mass control programmes at particular risk through rising levels of drug resistance. Last, safety issues were under­ estimated for mass drug administration programmes. When using albendazole or mebendazole treatment, health­ care providers should be aware of potential liver injury and bone marrow suppression.