Calum Davey

Re-analysis of health and educational impacts of a school-based deworming programme in western Kenya: a pure replication

Background: Helminth (worm) infections cause morbidity among poor communities worldwide. An influential study conducted in Kenya in 1998–99 reported that a school-based drug-and-educational intervention had benefits regarding worm infections and school attendance. Effects were seen among children treated with deworming drugs, untreated children in intervention schools and children in nearby non-intervention schools. Combining these effects, the intervention was reported to increase school attendance by 7.5% in treated children. Effects on other outcomes (worm infections, anaemia, nutritional status and examination performance) were also investigated. Methods: In this pure replication, we used data provided by the original authors to re-analyse the study according to their methods. We compared these results against those presented in the original paper. Results: Although most results were reproduced as originally reported, we identified discrepancies of several types between the original findings and re-analysis. For worm infections, re-analysis showed reductions similar to those originally reported. For anaemia prevalence, in contrast to the original findings, re-analysis found no evidence of benefit. For nutritional status, both original findings and re-analysis described modest evidence for a small improvement. For school attendance, re-analysis showed benefits similar to those originally found in intervention schools for both children who did and those who did not receive deworming drugs. However, after correction of coding errors, there was little evidence of an indirect effect on school attendance among children in schools close to intervention schools. Combining these effects gave a total increase in attendance of 3.9% among treated children, which was no longer statistically significant. As in the original results, re-analysis found no effect of the intervention on examination performance. Conclusions: Re-applying analytical approaches originally used, but correcting various errors, we found little evidence for some previously-reported indirect effects of a deworming intervention. Effects on worm infections, nutritional status, examination performance and school attendance on children in intervention schools were largely unchanged.

Analysis and reporting of stepped wedge randomised controlled trials: synthesis and critical appraisal of published studies, 2010 to 2014

BackgroundStepped wedge cluster randomised trials introduce interventions to groups of clusters in a random order and have been used to evaluate interventions for health and wellbeing. Standardised guidance for reporting stepped wedge trials is currently absent, and a range of potential analytic approaches have been described.MethodsWe systematically identified and reviewed recently published (2010 to 2014) analyses of stepped wedge trials. We extracted data and described the range of reporting and analysis approaches taken across all studies. We critically appraised the strategy described by three trials chosen to reflect a range of design characteristics.ResultsTen reports of completed analyses were identified. Reporting varied: seven of the studies included a CONSORT diagram, and only five also included a diagram of the intervention rollout. Seven assessed the balance achieved by randomisation, and there was considerable heterogeneity among the approaches used. Only six reported the trend in the outcome over time. All used both ‘horizontal’ and ‘vertical’ information to estimate the intervention effect: eight adjusted for time with a fixed effect, one used time as a condition using a Cox proportional hazards model, and one did not account for time trends. The majority used simple random effects to account for clustering and repeat measures, assuming a common intervention effect across clusters. Outcome data from before and after the rollout period were often included in the primary analysis. Potential lags in the outcome response to the intervention were rarely investigated. We use three case studies to illustrate different approaches to analysis and reporting.ConclusionsThere is considerable heterogeneity in the reporting of stepped wedge cluster randomised trials. Correct specification of the time-trend underlies the validity of the analytical approaches. The possibility that intervention effects vary by cluster or over time should be considered. Further work should be done to standardise the reporting of the design, attrition, balance, and time-trends in stepped wedge trials.

Re-analysis of health and educational impacts of a school-based deworming programme in western Kenya: a statistical replication of a cluster quasi-randomized stepped-wedge trial

Introduction: Helminth (worm) infections cause morbidity among poor communities worldwide. An influential study conducted in Kenya in 1998–99 reported that a school-based drug-and-educational intervention had benefits for worm infections and school attendance. Methods: In this statistical replication, we re-analysed data from this cluster quasi-randomized stepped-wedge trial, specifying two co-primary outcomes: school attendance and examination performance. We estimated intention-to-treat effects using year-stratified cluster-summary analysis and observation-level random-effects regression, and combined both years with a random-effects model accounting for year. The participants were not blinded to allocation status, and other interventions were concurrently conducted in a sub-set of schools. A protocol guiding outcome data collection was not available. Results: Quasi-randomization resulted in three similar groups of 25 schools. There was a substantial amount of missing data. In year-stratified cluster-summary analysis, there was no clear evidence for improvement in either school attendance or examination performance. In year-stratified regression models, there was some evidence of improvement in school attendance [adjusted odds ratios (aOR): year 1: 1.48, 95% confidence interval (CI) 0.88–2.52, P = 0.147; year 2: 1.23, 95% CI 1.01–1.51, P = 0.044], but not examination performance (adjusted differences: year 1: −0.135, 95% CI −0.323–0.054, P = 0.161; year 2: −0.017, 95% CI −0.201–0.166, P = 0.854). When both years were combined, there was strong evidence of an effect on attendance (aOR 1.82, 95% CI 1.74–1.91, P < 0.001), but not examination performance (adjusted difference −0.121, 95% CI −0.293–0.052, P = 0.169). Conclusions: The evidence supporting an improvement in school attendance differed by analysis method. This, and various other important limitations of the data, caution against over-interpretation of the results. We find that the study provides some evidence, but with high risk of bias, that a school-based drug-treatment and health-education intervention improved school attendance and no evidence of effect on examination performance.

Designing a stepped wedge trial: three main designs, carry-over effects and randomisation approaches

BackgroundThere is limited guidance on the design of stepped wedge cluster randomised trials. Current methodological literature focuses mainly on trials with cross-sectional data collection at discrete times, yet many recent stepped wedge trials do not follow this design. In this article, we present a typology to characterise the full range of stepped wedge designs, and offer guidance on several other design aspects.MethodsWe developed a framework to define and report the key characteristics of a stepped wedge trial, including cluster allocation and individual participation. We also considered the relative strengths and weaknesses of trials according to this framework. We classified recently published stepped wedge trials using this framework and identified illustrative case studies. We identified key design choices and developed guidance for each.ResultsWe identified three main stepped wedge designs: those with a closed cohort, an open cohort, and a continuous recruitment short exposure design. In the first two designs, many individuals experience both control and intervention conditions. In the final design, individuals are recruited in continuous time as they become eligible and experience either the control or intervention condition, but not both, and then provide an outcome measurement at follow-up. While most stepped wedge trials use simple randomisation, stratification and restricted randomisation are often feasible and may be useful. Some recent studies collect outcome information from individuals exposed a long time before or after the rollout period, but this contributes little to the primary analysis. Incomplete designs should be considered when the intervention cannot be implemented quickly. Carry-over effects can arise in stepped wedge trials with closed and open cohorts.ConclusionsStepped wedge trial designs should be reported more clearly. Researchers should consider the use of stratified and/or restricted randomisation. Trials should generally not commit resources to collect outcome data from individuals exposed a long time before or after the rollout period. Though substantial carry-over effects are uncommon in stepped wedge trials, researchers should consider their possibility before conducting a trial with closed or open cohorts.

Stepped wedge randomised controlled trials: systematic review of studies published between 2010 and 2014

BackgroundIn a stepped wedge, cluster randomised trial, clusters receive the intervention at different time points, and the order in which they received it is randomised. Previous systematic reviews of stepped wedge trials have documented a steady rise in their use between 1987 and 2010, which was attributed to the design’s perceived logistical and analytical advantages. However, the interventions included in these systematic reviews were often poorly reported and did not adequately describe the analysis and/or methodology used. Since 2010, a number of additional stepped wedge trials have been published. This article aims to update previous systematic reviews, and consider what interventions were tested and the rationale given for using a stepped wedge design.MethodsWe searched PubMed, PsychINFO, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), the Web of Science, the Cochrane Library and the Current Controlled Trials Register for articles published between January 2010 and May 2014. We considered stepped wedge randomised controlled trials in all fields of research. We independently extracted data from retrieved articles and reviewed them. Interventions were then coded using the functions specified by the Behaviour Change Wheel, and for behaviour change techniques using a validated taxonomy.ResultsOur review identified 37 stepped wedge trials, reported in 10 articles presenting trial results, one conference abstract, 21 protocol or study design articles and five trial registrations. These were mostly conducted in developed countries (n = 30), and within healthcare organisations (n = 28). A total of 33 of the interventions were educationally based, with the most commonly used behaviour change techniques being ‘instruction on how to perform a behaviour’ (n = 32) and ‘persuasive source’ (n = 25). Authors gave a wide range of reasons for the use of the stepped wedge trial design, including ethical considerations, logistical, financial and methodological. The adequacy of reporting varied across studies: many did not provide sufficient detail regarding the methodology or calculation of the required sample size.ConclusionsThe popularity of stepped wedge trials has increased since 2010, predominantly in high-income countries. However, there is a need for further guidance on their reporting and analysis.

Engagement with HIV prevention treatment and care among female sex workers in Zimbabwe: a respondent driven sampling survey.

Objective(S) To determine the HIV prevalence and extent of engagement with HIV prevention and care among a representative sample of Zimbabwean sex workers working in Victoria Falls, Hwange and Mutare. Design Respondent driven sampling (RDS) surveys conducted at each site. Methods Sex workers were recruited using respondent driven sampling with each respondent limited to recruiting 2 peers. Participants completed an interviewer-administered questionnaire and provided a finger prick blood sample for HIV antibody testing. Statistical analysis took account of sampling method. Results 870 women were recruited from the three sites. HIV prevalence was between 50 and 70%. Around half of those confirmed HIV positive were aware of their HIV status and of those 50-70% reported being enrolled in HIV care programmes. Overall only 25-35% of those with laboratory-confirmed HIV were accessing antiretroviral therapy. Among those reporting they were HIV negative, 21-28% reported having an HIV test in the last 6 months. Of those tested HIV negative, most (65-82%) were unaware of their status. Around two-thirds of sex workers reported consistent condom use with their clients. As in other settings, sex workers reported high rates of gender based violence and police harassment. Conclusions This survey suggests that prevalence of HIV is high among sex workers in Zimbabwe and that their engagement with prevention, treatment and care is sub-optimal. Intensifying prevention and care interventions for sex workers has the potential to markedly reduce HIV and social risks for sex workers, their clients and the general population in Zimbabwe and elsewhere in the region.

Five questions to consider before conducting a stepped wedge trial

Researchers should consider five questions before starting a stepped wedge trial.Why are you planning one? Researchers sometimes think that stepped wedge trials are useful when there is little doubt about the benefit of the intervention being tested. However, if the primary reason for an intervention is to measure its effect, without equipoise there is no ethical justification for delaying implementation in some clusters. By contrast, if you are undertaking pragmatic research, where the primary reason for rolling out the intervention is for it to exert its benefits, and if phased implementation is inevitable, a stepped wedge trial is a valid option and provides better evidence than most non-randomized evaluations.What design will you use? Two common stepped wedge designs are based on the recruitment of a closed or open cohort. In both, individuals may experience both control and intervention conditions and you should be concerned about carry-over effects. In a third, continuous-recruitment, short-exposure design, individuals are recruited as they become eligible and experience either control or intervention condition, but not both.How will you conduct the primary analysis? In stepped wedge trials, control of confounding factors through secular variation is essential. ‘Vertical’ approaches preserve randomization and compare outcomes between randomized groups within periods. ‘Horizontal’ approaches compare outcomes before and after crossover to the intervention condition. Most analysis models used in practice combine both types of comparison. The appropriate analytic strategy should be considered on a case-by-case basis.How large will your trial be? Standard sample size calculations for cluster randomized trials do not accommodate the specific features of stepped wedge trials. Methods exist for many stepped wedge designs, but simulation-based calculations provide the greatest flexibility. In some scenarios, such as when the intracluster correlation coefficient is moderate or high, or the cluster size is large, a stepped wedge trial may require fewer clusters than a parallel cluster trial.How will you report your trial? Stepped wedge trials are currently challenging to report using CONSORT principles. Researchers should consider how to demonstrate balance achieved by randomization and how to describe trends for outcomes in both intervention and control clusters.

The HIV care cascade among female sex workers in Zimbabwe: results of a population-based survey from the Sisters Antiretroviral therapy Programme for Prevention of HIV, an Integrated Response (SAPPH-IRe) Trial.

Introduction: Female sex workers (FSW) in sub-Saharan Africa have a higher prevalence of HIV than other women of reproductive age. Social, legal, and structural barriers influence their access to care. Little is known about the HIV diagnosis and care cascade in most countries in Southern Africa. We aimed to describe the HIV diagnosis and care cascade among FSW in Zimbabwe. Methods: We conducted cross-sectional respondent driven sampling (RDS) surveys of FSW in 14 sites across Zimbabwe as the baseline for a cluster-randomised controlled trial investigating a combination HIV prevention and care package. We administered a questionnaire, tested women for HIV and measured viral load. We report the mean, minimum, and maximum respondent-driven sampling-2 weighted site values. Results: The survey included 2722 women, approximately 200 per site. The mean HIV prevalence was 57.5% (42.8–79.2 site minimum and maximum). Of HIV-positive women, 64.0% (51.6–73.7) were aware of their status, 67.7% (53.4–84.1) of these reported taking antiretroviral therapy, and 77.8% (64.4–90.8) of these had a suppressed HIV viral load (<1000 copies/mL). Among all HIV-positive women, 49.5% had a viral load < 1000 copies/mL. Conclusions: Although most HIV-positive women aware of their status are accessing antiretroviral therapy, 36.0% of HIV-positive women are unaware of their status and 29.3% of all FSW have an unsuppressed HIV viral load. Investigation and investment into models of testing, treatment, and care are necessary to reach UNAIDS targets for HIV elimination.

Finger Prick Dried Blood Spots for HIV Viral Load Measurement in Field Conditions in Zimbabwe

Background In the context of a community-randomized trial of antiretrovirals for HIV prevention and treatment among sex workers in Zimbabwe (the SAPPH-IRe trial), we will measure the proportion of women with HIV viral load (VL) above 1000 copies/mL (“VL>1000”) as our primary endpoint. We sought to characterize VL assay performance by comparing results from finger prick dried blood spots (DBS) collected in the field with plasma samples, to determine whether finger prick DBS is an acceptable sample for VL quantification in the setting. Methods We collected whole blood from a finger prick onto filter paper and plasma samples using venipuncture from women in two communities. VL quantification was run on samples in parallel using NucliSENS EasyQ HIV-1 v2.0. Our trial outcome is the proportion of women with VL>1000, consistent with WHO guidelines relating to regimen switching. We therefore focused on this cut-off level for assessing sensitivity and specificity. Results were log transformed and the mean difference and standard deviation calculated, and correlation between VL quantification across sample types was evaluated. Results A total of 149 HIV-positive women provided DBS and plasma samples; 56 (63%) reported being on antiretroviral therapy. VL ranged from undetectable-6.08 log10 using DBS and undetectable-6.40 log10 using plasma. The mean difference in VL (plasma-DBS) was 0.077 log10 (95%CI = 0.025–0.18 log10; standard deviation = 0.63 log10,). 78 (52%) DBS and 87 (58%) plasma samples had a VL>1000. Based on plasma ‘gold-standard’, DBS sensitivity for detection of VL>1000 was 87.4%, and specificity was 96.8%. Conclusion There was generally good agreement between DBS and plasma VL for detection of VL>1000. Overall, finger prick DBS appeared to be an acceptable sample for classifying VL as above or below 1000 copies/mL using the NucliSENS assay.

Trends in socioeconomic inequalities in HIV prevalence among young people in seven countries in eastern and southern Africa.

Background In Eastern and Southern Africa, HIV prevalence was highest among higher socioeconomic groups during the 1990s. It has been suggested that this is changing, with HIV prevalence falling among higher-educated groups while stable among lower-educated groups. A multi-country analysis has not been undertaken. Methods We analysed data on socio-demographic factors and HIV infection from 14 nationally representative surveys of adults aged 15-24 (seven countries, two surveys each, 4-8 years apart). Sample sizes ranged from 2,408-12,082 (72,135 total). We used logistic regression to assess gender-stratified associations between highest educational level attended and HIV status in each survey, adjusting for age and urban/rural setting. We tested for interactions with urban/rural setting and age. Our primary hypothesis was that higher education became less of a risk factor for HIV over time. We tested for interaction between survey-year and the education-HIV association in each country and all countries pooled. Findings In Ethiopia and Malawi, HIV prevalence was higher in more educated women in both surveys. In Lesotho, Kenya and Zimbabwe, HIV prevalence was lower in higher educated women in both surveys. In Ethiopia, HIV prevalence fell among no and secondary educated women only (interaction p<0·01). Only among young men in Tanzania there was some evidence that the association between education and HIV changed over time (p=0·07). Pooled analysis found little evidence for an interaction between survey year and the education-HIV association among men (p=0·60) or women (p=0·37). Interpretation The pattern of prevalent HIV infection among young adults by level of education in different sub-Saharan African countries was heterogeneous. There was little statistical evidence that this pattern changed between 2003-5 and 2008-12. Explanations for the social epidemiology of HIV in Africa will need to account for time-trends and inter-country differences.