Source Themes

Adjustment for prognostic baseline variables can reduce bias due to covariate imbalance and increase efficiency in randomized trials. While the use of covariate adjustment in late-phase trials is justified by favorable large-sample properties, it is seldom used in small, early-phase studies, due to uncertainty in which variables are prognostic and the potential for precision loss, type I error rate inflation, and undercoverage of confidence intervals. To address this problem, we consider adjustment for a valid negative control outcome (NCO), or an auxiliary post-randomization outcome believed to be completely unaffected by treatment but more highly correlated with the primary outcome than baseline covariates. We articulate the assumptions that permit adjustment for NCOs without producing post-randomization selection bias, and describe plausible data-generating models where NCO adjustment can improve upon adjustment for baseline covariates alone. In numerical experiments, we illustrate performance and provide practical recommendations regarding model selection and finite-sample variance corrections. We apply our methods to the reanalysis of two early-phase vaccine trials in HIV exposed uninfected (HEU) infants, where we demonstrate that adjustment for auxiliary post-baseline immunological parameters can enhance the precision of vaccine effect estimates relative to standard approaches that avoid adjustment or adjust for baseline covariates alone.

Autopsy studies of children dying of cerebral malaria (CM) have revealed that those with malarial retinopathy exhibited high levels of sequestration in the cerebral vasculature, whereas children with retinopathy-negative CM exhibited lower sequestration levels and possible nonmalarial causes of death. This suggests that children dying of retinopathy-negative CM have nonmalarial coma etiologies with concomitant incidental parasitemia, which is common in high malaria transmission areas. Subsequent studies have challenged this assertion, positing that retinopathy-negative CM and retinopathy-positive CM are variants of the same disease pathophysiology or host biology, both caused by acute malaria infection. We recently determined that electroencephalography (EEG) can be used to discriminate between a malarial coma (CM) and a nonmalarial coma. To better understand the contribution of acute malaria infection in the pathophysiology of retinopathy-negative CM, we compared qualitative and quantitative EEG findings from 30-minute EEG recordings of Malawian children aged 3 months to 14 years hospitalized at Queen Elizabeth Central Hospital with retinopathy-negative CM, retinopathy-positive CM, and nonmalarial coma. Neither qualitative nor quantitative EEG interpretation methods allow for the discrimination between children with retinopathy-positive CM and those with retinopathy-negative CM. Conversely, quantitative EEG readily differentiated children with retinopathy-negative CM from those with nonmalarial coma (area under the receiving operating characteristic [AUROC] curve of 0.83). When combining qualitative and quantitative EEG interpretation methods, the ability of EEG to distinguish retinopathy-negative CM from nonmalarial EEG increases (AUROC of 0.87). The EEGs of children with retinopathy-negative CM are similar to those of children with retinopathy-positive CM and significantly different from those of children with nonmalarial coma, supporting the hypothesis that acute malarial infection is pathophysiologically important in retinopathy-negative CM.

In an observational study, it is common to leverage known null effects to detect bias. One such strategy is to set aside a placebo sample – a subset of data immune from the hypothesized cause-and-effect relationship. Existence of an effect in the placebo sample raises concerns about unmeasured confounding bias while the absence of it helps corroborate the causal conclusion. This article describes a framework for using a placebo sample to detect and remove bias. We state the identification assumptions and develop estimation and inference methods based on outcome regression, inverse probability weighting, and doubly robust approaches. Simulation studies investigate the finite-sample performance of the proposed methods. We illustrate the methods using an empirical study of the effect of the earned income tax credit on infant health.

With limited access to mRNA COVID-19 vaccines in lower income countries, and people living with HIV (PLWH) largely excluded from clinical trials, Part A of the multicentre CoVPN 3008 (Ubuntu) study aimed to assess the safety of mRNA-1273, the relative effectiveness of hybrid versus vaccine immunity, and SARS-CoV-2 viral persistence among PLWH in East and Southern Africa during the omicron outbreak.Previously unvaccinated adults with HIV and/or other comorbidities associated with severe COVID-19 received either one (hybrid immunity) or two (vaccine immunity) 100-mcg doses of ancestral strain mRNA-1273 in the first month, depending on baseline evidence of prior SARS-CoV-2 infection. In a prospective cohort study design, we used covariate-adjusted Cox regression and counterfactual cumulative incidence methods to determine the hazard ratio and relative risk of COVID-19 and severe COVID-19 with hybrid versus vaccine immunity within six months. The ongoing Ubuntu study is registered on ClinicalTrials.gov (NCT05168813) and this work was conducted from December 2021 to March 2023.Between December 2021 and September 2022, 14,237 participants enrolled, and 14,002 (83% PLWH, 69% SARS-CoV-2 seropositive) were included in the analyses. Vaccinations were safe and well tolerated. Common adverse events were pain or tenderness at the injection site (26.7%), headache (20.4%), and malaise (20.3%). Severe adverse events were rare (0.8% of participants after the first and 1.1% after the second vaccination), and none were life-threatening or fatal. Among PLWH, the median CD4 count was 635 cells/μl and 18.5% had HIV viraemia. The six-month cumulative incidences in the hybrid immunity and vaccine immunity groups were 2.02% (95% confidence interval [CI] 1.61–2.44) and 3.40% (95% CI 2.30–4.49) for COVID-19, and 0.048% (95% CI 0.00–0.10) and 0.32% (95% CI 0.59–0.63) for severe COVID-19. Among all PLWH the hybrid immunity group had a 42% lower hazard rate of COVID-19 (hazard ratio [HR] 0.58; 95% CI 0.44–0.77; p < 0.001) and a 73% lower hazard rate of severe COVID-19 (HR 0.27; 95% CI 0.07–1.04; p = 0.056) than the vaccine immunity group, but this effect was not seen among PLWH with CD4 counts <350 cells/μl or HIV viraemia. Twenty PLWH had persistent SARS-CoV-2 virus at least 50 days.Hybrid immunity was associated with superior protection from COVID-19 compared to vaccine immunity with the ancestral mRNA-1273 vaccine. Persistent infections among immunocompromised PLWH may provide reservoirs for emerging variants.

Neutralizing antibody titer has been a surrogate endpoint for guiding COVID-19 vaccine approval and use, although the pandemic’s evolution and the introduction of variant-adapted vaccine boosters raise questions as to this surrogate’s contemporary performance. For 985 recipients of an mRNA second bivalent or monovalent booster containing various Spike inserts [Prototype (Ancestral), Beta, Delta, and/or Omicron BA.1 or BA.4/5] in the COVAIL trial (NCT05289037), titers against 5 strains were assessed as correlates of risk of symptomatic COVID-19 (“COVID-19”) and as correlates of relative (Pfizer-BioNTech Omicron vs. Prototype) booster protection against COVID-19 over 6 months of follow-up during the BA.2-BA.5 Omicron-dominant period. Consistently across the Moderna and Pfizer-BioNTech vaccine platforms and across all variant Spike inserts assessed, both peak and exposure-proximal (“predicted-at-exposure”) titers correlated with lower Omicron COVID-19 risk in individuals previously infected with SARS-CoV-2, albeit significantly less so in naïve individuals [e.g., exposure-proximal hazard ratio per 10-fold increase in BA.1 titer 0.74 (95% CI 0.59, 0.94) for naïve vs. 0.41 (95% CI 0.23, 0.64) for non-naïve; interaction p = 0.013]. Neutralizing antibody titer was a strong inverse correlate of Omicron COVID-19 in non-naïve individuals and a weaker correlate in naïve individuals, posing questions about how prior infection alters the neutralization correlate.

Many recent studies have probed status bias in the peer-review process of academic journals and conferences. In this article, we investigated the association between author metadata and area chairs’ final decisions (Accept/Reject) using our compiled database of 5,313 borderline submissions to the International Conference on Learning Representations from 2017 to 2022 under a matched observational study framework. We carefully defined elements in a cause-and-effect analysis, including the treatment and its timing, pre-treatment variables, potential outcomes (POs) and causal null hypothesis of interest, all in the context of study units being textual data and under Neyman and Rubin’s PO framework. We found some weak evidence that author metadata was associated with articles’ final decisions. We also found that, under an additional stability assumption, borderline articles from high-ranking institutions (top-30% or top-20%) were less favored by area chairs compared to their matched counterparts. The results were consistent in two different matched designs (odds ratio = 0.82 [95% confidence interval (CI): 0.67 to 1.00] in a first design and 0.83 [95% CI: 0.64 to 1.07] in a strengthened design) and most pronounced in the subgroup of articles with low ratings. We discussed how to interpret these results in the context of multiple interactions between a study unit and different agents (reviewers and area chairs) in the peer-review system.

For COVAIL recipients of a coronavirus disease 2019 (COVID-19) Sanofi booster vaccine, neutralizing antibody titers were assessed as a correlate of risk (CoR) of COVID-19. Peak and exposure-proximal titers were inverse CoRs with covariate-adjusted hazard ratios (95% confidence intervals) 0.30 (0.11, 0.78) and 0.25 (0.07, 0.85) per 10-fold increase in weighted average titer.

In the phase 3 Coronavirus Efficacy (COVE) trial (NCT04470427), post-dose two Ancestral Spike-specific binding (bAb) and neutralizing (nAb) antibodies were shown to be correlates of risk (CoR) and of protection against Ancestral-lineage COVID-19 in SARS-CoV-2 naive participants. In the SARS-CoV-2 Omicron era, Omicron subvariants with varying degrees of immune escape now dominate, seropositivity rates are high, and booster doses are administered, raising questions on whether and how these developments affect the bAb and nAb correlates. To address these questions, we assess post-boost BA.1 Spike-specific bAbs and nAbs as CoRs and as correlates of booster efficacy in COVE. For naive individuals, bAbs and nAbs inversely correlate with Omicron COVID-19: hazard ratios (HR) per 10-fold marker increase (95% confidence interval) are 0.16 (0.03, 0.79) and 0.31 (0.10, 0.96), respectively. In non-naive individuals the analogous results are similar: 0.15 (0.04, 0.63) and 0.28 (0.07, 1.08). For naive individuals, three vs two-dose booster efficacy correlates with predicted nAb titer at exposure, with estimates -8% (-126%, 48%), 50% (25%, 67%), and 74% (49%, 87%), at 56, 251, and 891 Arbitrary Units/ml. These results support the continued use of antibody as a surrogate endpoint.

Randomization inference is a powerful tool in early phase vaccine trials when estimating the causal effect of a regimen against a placebo or another regimen. Randomization-based inference often focuses on testing either Fisher’s sharp null hypothesis of no treatment effect for any participant or Neyman’s weak null hypothesis of no sample average treatment effect. Many recent efforts have explored conducting exact randomization-based inference for other summaries of the treatment effect profile, for instance, quantiles of the treatment effect distribution function. In this article, we systematically review methods that conduct exact, randomization-based inference for quantiles of individual treatment effects (ITEs) and extend some results to a special case where naïve participants are expected not to exhibit responses to highly specific endpoints. These methods are suitable for completely randomized trials, stratified completely randomized trials, and a matched study comparing two non-randomized arms from possibly different trials. We evaluate the usefulness of these methods using synthetic data in simulation studies. Finally, we apply these methods to HIV Vaccine Trials Network Study 086 (HVTN 086) and HVTN 205 and showcase a wide range of application scenarios of the methods.

Malaria is a significant obstacle to child health and survival. Plasmodium falciparum infections, especially in children under five, lead to high morbidity and mortality. Cerebral malaria (CM) is a life-threatening complication characterized by coma, and its diagnosis can be improved by observing malarial retinopathy in children. Monitoring vital signs is essential for managing patients with CM. To determine if changes in vital signs predict death in children with retinopathy positive cerebral malaria (RPCM). This was a retrospective case-control study using data collected from children admitted to the Paediatric Research Ward at Queen Elizabeth Central Hospital in Blantyre between 1997 and 2020. Patients who died 24 hours or more after admission were matched with control patients who survived. Linear regression analyses were used to assess the differential time trends of each vital sign in the survivor group and death group. Classification models were used to quantify various vital signs’ predictive power of death. Among the population that died, the estimated change in average respiratory rate per hour approaching death was 0.02 breaths per minute compared to -0.25 breaths per minute among those who survive (p < 0.001), and the estimated change in average BCS per hour approaching death was -0.01 compared to 0.06 among the survivors (p < 0.001). Changes in temperature and heart rate were not associated with clinical deterioration. Three models were developed, and the best receiver operating characteristic curve was 100% sensitive, the corresponding false positive rate was 75%. Changes in respiratory rate and BCS have prognostic significance in the final 24 hours before death in children with cerebral malaria. Extra attention should be paid to these two vital signs as they may help to identify children who are at increased risk of deteriorating.