Background – Consent rates for research studies in low- and lower middle-income countries (LMICs) are often greater than 90%. Uniformly high consent rates raise ethical concerns regarding the extent to which consent is both autonomous and informed. This study aimed to characterize consent rates for neonatal randomized controlled trials (RCTs) in LMICs compared to high-income countries (HICs). Methods – We searched MEDLINE, EMBASE and Cochrane for neonatal RCTS in LMICs or HICs published between 2013 and 2018. Given the disproportionately high number of HIC articles in the initial search, we used a random number generator to select a subset of HIC articles for screening. Our primary outcome was consent rate, defined as the proportion of eligible parents who consented amongst those who were approached. We extracted available study characteristics and consent rates from each article. In descriptive analysis, we compared consent rates in LMIC vs HIC trials overall, as well as by trial type classified by the intervention (drug/nutrition, device, other), control type (placebo, no placebo) and funding (private, public, both, none, not stated). We also compared consent rates between LMICs and HICs by trial type using a multivariate fractional logistic regression model. Results – Of 9270 articles identified, we screened 3523, yielding 300 articles for inclusion; 200 (67%) reported a consent rate. Reporting of consent rates was similar between LMIC and HIC trials (67.9% vs 59.8%; p=0.16). The median consent rate in LMIC trials was significantly higher than in HIC trials (95.6% [IQR 88.2%-98.9%] vs 82.7% [IQR 68.6%-93.0%]; p<0.001). Within HIC trials, consent rates differed by control type and funding, whereas consent rates in LMIC trials did not (Table 1). In regression modeling controlling for study characteristics, odds of consenting were significantly higher in LMIC vs HIC trials that evaluated drug/nutrition interventions both with a placebo control (OR 6.39; 95% CI 3.32 – 12.30) and without a placebo control (OR 3.67; 95% CI 1.87 – 7.20; Table 2). Odds of consenting in trials other than those using drug/nutrition interventions were not significantly higher in LMICs vs HICs. Conclusion – Consent rates in neonatal RCTs in LMICs are significantly higher than in HICs. In adjusted models, the odds of potential participants consenting to trials that involve drug/nutrition interventions are three to six times higher in LMICs than in HICs. These results raise ethical concerns about the consent process in LMICs. Future studies are needed to investigate whether factors such as inadequate research governance, coercive incentives, or cultural or economic factors influence autonomous decision making or result in poorly informed consent in LMICs.
(a) p-values from Kruskal-Wallis rank test; (b) p-values from Two-Sample Wilcoxon rank-sum test; LMIC=low- and lower middle-income country, HIC=high-income country, IQR=interquartile range
(a) p-values from Kruskal-Wallis rank test; (b) p-values from Two-Sample Wilcoxon rank-sum test; LMIC=low- and lower middle-income country, HIC=high-income country, IQR=interquartile range
Odds of consenting by trial type in LMIC vs HIC trials
(a) n=200 total observations, Wald Chi-Sq 64.01 (p-value <0.001), pseudo R-sq 0.063; (b) Fractional logistic regression adjusts for funding source, timing of consent, publication year and log number enrolled; n=199 total observations, Wald Chi-Sq 105.89 (p-value <0.001), pseudo R-sq 0.082; LMIC=low- and lower middle-income country, HIC=high-income country, OR=odds ratio, CI=confidence interval
Odds of consenting by trial type in LMIC vs HIC trials
(a) n=200 total observations, Wald Chi-Sq 64.01 (p-value <0.001), pseudo R-sq 0.063; (b) Fractional logistic regression adjusts for funding source, timing of consent, publication year and log number enrolled; n=199 total observations, Wald Chi-Sq 105.89 (p-value <0.001), pseudo R-sq 0.082; LMIC=low- and lower middle-income country, HIC=high-income country, OR=odds ratio, CI=confidence interval
Comments