To understand a treatment’s potential impact at the individual level, it is crucial to explore whether the effect differs across patient subgroups and covariate values. Meta-analysis provides an important tool for detecting treatment–covariate interactions, as it can improve power compared to a single study. However, aggregation bias can occur when estimating individual-level treatment–covariate interactions in meta-analysis, due to trial-level confounding. This refers to when the association between the covariate and treatment effect across trials (at the aggregate level) differs from that observed within trials (at the individual level). It is, thus, recommended that heterogeneity in the treatment effect at the individual level should be disentangled from that at the trial level, ideally using an individual participant data (IPD) meta-analysis. Here, we explain this issue and provide new intuition about how trial-level confounding is impacted by differences in within-trial distributions of covariates and how this corresponds to asymmetry in subgroup-specific funnel plots in the case of categorical covariates. We then propose a sensitivity analysis to assess the robustness of interaction estimates to potential trial-level confounding. We illustrate these concepts using simulated and real data from an IPD meta-analysis of trials conducted on the TICO/ACTIV-3 platform, which assessed passive immunotherapy treatments for inpatients with COVID-19.

Ovine and equine protoscoleces of Echinococcus granulosus were cultured for 26 days with our without praziquantel and viability assessed, by eosin exclusion, for cultures in various drug concentrations (50, 250 and 500 μg/l) and periods of exposure (1, 3 or 7 days (d)) before removing/‘rescuing’ to drug-free medium. Drug efficacy was proportional to drug concentration and to length of exposure. At higher drug concentrations shorter exposures were required to produce the effect of continuous drug treatment, 1d therapy at 500 μg/l killing 96% ovine protoscoleces by day 14 whereas 7d therapy at 50 μg/l was required to produce a similar effect. Equine protoscoleces appeared marginally less susceptible than those of ovine origin. The relevance of the results in the need for peri-operative prophylaxis against spilled protoscoleces in man is discussed.

Praziquantel (500 mg/kg) administered orally to BALB/c mice with secondary equine E. granulosus daily for 21, 30 or 30 + 30 days without the drug resulted in the majority of cysts, using bench criteria of turgidity and eosin exclusion, being assessed as ‘alive’. Ultrastructural examination of 54 of these ‘alive’ cysts did not support this conclusion. They all showed increased vesiculation of the germinal layer leading, in many, to the loss of its integrity. Increased mitochondrial numbers occurred frequently. The longer drug treatments appeared to have greater effects on the germinal layer of ‘alive’ cysts and there was no detectable re-establishment of structural organization within 30 days after drug withdrawal. Subjectively, there was no substantial difference between cysts from 4-month and 9-month infections or between affected peritoneal and hepatic cysts. Tissue from collapsed cysts was necrotic. Peak serum levels of praziquantel (6430–6136 μg/l) occurred 5–10 min after drug administration (500 mg/kg) and dropped rapidly to less than 10 μg/l at 3 h. In an in vitro study at praziquantel concentrations of 1000 and 5000 μg/l over a 10-day period, most cysts were judged ‘alive’ by bench criteria but showed ultrastructurally a time- and concentration-dependent loss of integrity identical to that seen in vivo. Turgidity and eosin exclusion therefore underestimate the effect of praziquantel and the results indicate that in vitro experiments can fulfil a legitimate preliminary role in a hydatid chemotherapy programme.