Addressing selection bias in the UK Biobank neurological imaging cohort

The UK Biobank is a national prospective study of half a million participants between the ages of 40 and 69 at the time of recruitment between 2006 and 2010, established to facilitate research on diseases of aging. The imaging cohort is a subset of UK Biobank participants who have agreed to undergo extensive additional imaging assessments. However, Fry et al (2017) find evidence of "healthy volunteer bias" in the UK Biobank -- participants are less likely to smoke, be obese, consume alcohol daily than the target population of UK adults. Here we examine selection bias in the UK Biobank imaging cohort. We address two common misconceptions: first, that study size can compensate for bias in data collection, and second that selection bias does not affect estimates of associations, which are the primary interest of the UK Biobank. We introduce inverse probability weighting (IPW) as an approach commonly used in survey research that can be used to address selection bias in volunteer health studies like the UK Biobank. We discuss 6 such methods -- five existing and one novel --, assess relative performance in simulation studies, and apply them to the UK Biobank imaging cohort. We find that our novel method, BART for predicting the probability of selection combined with raking, performs well relative to existing methods, and helps alleviate selection bias in the UK Biobank imaging cohort.

Asociación entre calcificaciones de la glándula pineal y enfermedad aterosclerótica intracraneal en adultos mayores

Background: This study assesses whether pineal gland calcification (PGC) – a surrogate for reduced endogenous melatonin production – is associated with significant stenosis of large intracranial arteries – a biomarker of intracranial atherosclerotic disease (ICAD). Methods: Individuals aged ≥60 years enrolled in the Three Villages Study received head CT to assess PGC and MRA to estimate stenosis of large intracranial arteries. Multivariate logistic regression models were fitted to assess the association between PGC and ICAD, after adjusting for relevant confounders. Inverse probability of exposure weighting was used to estimate the effect of PGC on ICAD. Results: A total of 581 individuals were enrolled. PGC and ICAD were associated in a fully-adjusted logistic regression model (p=0.032). Inverse probability of exposure weighting showed an estimate for the proportion of ICAD among those without PGC of 3.7% and the adjusted-effect coefficient was 5.7% higher among those with PGC (p=0.031). Conclusions: PGC is associated with ICAD. Study results provide grounds for evaluating the role of melatonin deficiency in ICAD progression. Keywords: Pineal gland calcification, intracranial atherosclerosis, stenosis of large intracranial arteries, melatonin; population study, older adults

Additional effect of azithromycin over β-lactam alone for severe community-acquired pneumonia-associated acute respiratory distress syndrome: a retrospective cohort study

Background Community-acquired pneumonia (CAP) is the most common cause of acute respiratory distress syndrome (ARDS). Although previous studies have suggested that macrolide therapy is beneficial for ARDS, its benefit for severe CAP-associated ARDS remains uncertain. Previous studies were limited in that they had a small sample size and included patients with non-pulmonary ARDS and those with pulmonary ARDS. This study aimed to investigate the additional effect of azithromycin when used with β-lactam compared with the effect of β-lactam alone in mechanically ventilated patients with CAP-associated ARDS. Methods We identified mechanically ventilated patients with CAP-associated ARDS between July 2010 and March 2015 using data in the Diagnosis Procedure Combination database, a Japanese nationwide inpatient database. We performed propensity score matching analysis to assess 28-day mortality and in-hospital mortality in mechanically ventilated patients with CAP-associated ARDS who received β-lactam with and without azithromycin within hospital 2 days after admission. The inverse probability of treatment weighting analysis was also conducted. Results Eligible patients (n = 1257) were divided into the azithromycin group (n = 226) and the control group (n = 1031). The one-to-four propensity score matching analysis included 139 azithromycin users and 556 non-users. No significant difference was observed between the groups with respect to 28-day mortality (34.5% vs. 37.6%, p = 0.556) or in-hospital mortality (46.0% vs. 49.1%, p = 0.569). The inverse probability of treatment weighting analysis showed similar results. Conclusions Compared with treatment with β-lactam alone, treatment with azithromycin plus β-lactam had no significant additional effect on 28-day mortality or in-hospital mortality in mechanically ventilated patients with CAP-associated ARDS. To the best of our knowledge, this study is the first to determine the effect of azithromycin in mechanically ventilated patients with CAP-associated ARDS.

Hospital outcomes of community-acquired COVID-19 versus influenza: Insights from the Swiss hospital-based surveillance of influenza and COVID-19

Background Since the onset of the COVID-19 pandemic, the disease has frequently been compared with seasonal influenza, but this comparison is based on little empirical data. Aim This study compares in-hospital outcomes for patients with community-acquired COVID-19 and patients with community-acquired influenza in Switzerland. Methods This retrospective multi-centre cohort study includes patients > 18 years admitted for COVID-19 or influenza A/B infection determined by RT-PCR. Primary and secondary outcomes were in-hospital mortality and intensive care unit (ICU) admission for patients with COVID-19 or influenza. We used Cox regression (cause-specific and Fine-Gray subdistribution hazard models) to account for time-dependency and competing events with inverse probability weighting to adjust for confounders. Results In 2020, 2,843 patients with COVID-19 from 14 centres were included. Between 2018 and 2020, 1,381 patients with influenza from seven centres were included; 1,722 (61%) of the patients with COVID-19 and 666 (48%) of the patients with influenza were male (p < 0.001). The patients with COVID-19 were younger (median 67 years; interquartile range (IQR): 54–78) than the patients with influenza (median 74 years; IQR: 61–84) (p < 0.001). A larger percentage of patients with COVID-19 (12.8%) than patients with influenza (4.4%) died in hospital (p < 0.001). The final adjusted subdistribution hazard ratio for mortality was 3.01 (95% CI: 2.22–4.09; p < 0.001) for COVID-19 compared with influenza and 2.44 (95% CI: 2.00–3.00, p < 0.001) for ICU admission. Conclusion Community-acquired COVID-19 was associated with worse outcomes compared with community-acquired influenza, as the hazards of ICU admission and in-hospital death were about two-fold to three-fold higher.

A potential outcomes approach to defining and estimating gestational age-specific exposure effects during pregnancy

Many studies seek to evaluate the effects of potentially harmful pregnancy exposures during specific gestational periods. We consider an observational pregnancy cohort where pregnant individuals can initiate medication usage or become exposed to a drug at various times during their pregnancy. An important statistical challenge involves how to define and estimate exposure effects when pregnancy loss or delivery can occur over time. Without proper consideration, the results of standard analysis may be vulnerable to selection bias, immortal time-bias, and time-dependent confounding. In this study, we apply the “target trials” framework of Hernán and Robins in order to define effects based on the counterfactual approach often used in causal inference. This effect is defined relative to a hypothetical randomized trial of timed pregnancy exposures where delivery may precede and thus potentially interrupt exposure initiation. We describe specific implementations of inverse probability weighting, G-computation, and Targeted Maximum Likelihood Estimation to estimate the effects of interest. We demonstrate the performance of all estimators using simulated data and show that a standard implementation of inverse probability weighting is biased. We then apply our proposed methods to a pharmacoepidemiology study to evaluate the potentially time-dependent effect of exposure to inhaled corticosteroids on birthweight in pregnant people with mild asthma.

Effects of Reversal Technique for Neuromuscular Paralysis on Time to Recovery of Bowel Function after Craniotomy

Background Non-depolarizing neuromuscular blockade can be reversed with neostigmine/glycopyrrolate or sugammadex. We test the hypothesis that sugammadex is associated with earlier postoperative recovery of bowel function (first bowel movement, BM). Methods In adult patients undergoing craniotomy from 2016 to 2019, we identified time of first postoperative BM after receiving neostigmine/glycopyrrolate or sugammadex to reverse neuromuscular blockade. Logistic and proportional hazard regression, with and without inverse probability of treatment weighting (IPTW), were used to assess whether sugammadex is associated with earlier recovery of bowel function. Results Seven hundred and thirty-one patients underwent craniotomy, 323 (44.2%) received neostigmine/glycopyrrolate, and 408 (55.8%) sugammadex. From logistic regression analysis, the proportion of patients having a BM within the first 24 and 48 hours was higher in sugammadex group (unadjusted OR [95% CI]) 1.79 [1.16 to 2.77] P = .009; and 1.45 [1.08 to 1.94] P = .014; IPTW adjusted OR [95% CI]) 1.58 [.95, 2.61] P = .078; and 1.38 [.95 to 2.02] P = .095 for 24 and 48 h, respectively). From proportional hazards regression, sugammadex was associated with improved bowel function recovery (unadjusted hazard ratio (HR) [95% CI] 1.35 [1.08, 1.68], P = .008; IPTW adjusted HR 1.29 [.97 to 1.71], P = .076). Conclusion Patients undergoing craniotomy who had neuromuscular blockade reversed with sugammadex may have earlier recovered bowel function compared to patients reversed with neostigmine/glycopyrrolate.

A Comparison of Covariate Adjustment Approaches Under Model Misspecification In Individually Randomized Trials

Adjustment for baseline covariates in randomized trials has been shown to lead to gains in power and can protect against chance imbalances in covariates. For continuous covariates, there is a risk that the the form of the relationship between the covariate and outcome is misspecified when taking an adjusted approach. Using a simulation study focusing on small to medium-sized individually randomized trials, we explore whether a range of adjustment methods are robust to misspecification, either in the covariate-outcome relationship or through an omitted covariate-treatment interaction. Specifically, we aim to identify potential settings where G-computation, Inverse Probability of Treatment Weighting ( IPTW ), Augmented Inverse Probability of Treatment Weighting ( AIPTW ) and Targeted Maximum Likelihood Estimation ( TMLE ) offer improvement over the commonly used Analysis of Covariance ( ANCOVA ). Our simulations show that all adjustment methods are generally robust to model misspecification if adjusting for a few covariates, sample size is 100 or larger, and there are no covariate-treatment interactions. When there is a non-linear interaction of treatment with a skewed covariate and sample size is small, all adjustment methods can suffer from bias; however, methods that allow for interactions (such as G-computation with interaction and IPTW ) show improved results compared to ANCOVA . When there are a high number of covariates to adjust for, ANCOVA retains good properties while other methods suffer from under- or over-coverage. An outstanding issue for G-computation, IPTW and AIPTW in small samples is that standard errors are underestimated; development of small sample corrections is needed.

HIV and Hodgkin Lymphoma Survival: A Prospective Study in Botswana

PURPOSE People living with HIV (PLWH) experience increased risk of Hodgkin lymphoma (HL) despite effective initiation of antiretroviral therapy (ART). In high-income countries, outcomes following HIV HL have been reported to be non-differential, or inferior for PLWH. We sought to assess the effect of HIV on HL survival in Botswana, an African country with a generalized HIV epidemic and high ART coverage, to describe a context more reflective of global HIV populations. PATIENTS AND METHODS In the Thabatse Cancer Cohort, consenting participants initiating treatment for HL at one of four cancer centers in Botswana were enrolled from 2010 to 2020. Patients were followed quarterly for up to 5 years. The impact of HIV on survival following treatment initiation was assessed using an inverse probability–weighted Cox marginal structural model adjusted for age, performance status, and disease stage. RESULTS Seventy-eight new HL cases were enrolled, 47 (60%) were PLWH and 31 (40%) were HIV-uninfected. Baseline characteristics were similar between groups. The majority (61%) of patients presented with regional disease (stage I or II) with no statistically significant difference by HIV status ( P = .38). Nearly all (87%) PLWH participants were on ART before their HL diagnosis (median ART duration 42 months), and median CD4 count was 413 cells/μL (interquartile range 253-691). Survival, in unadjusted analyses, was lower among patients without HIV compared with PLWH (log rank P = .021). In adjusted analysis, HIV infection was not significantly associated with survival in inverse probability–weighted Cox model (hazard ratio 0.43; 95% CI, 0.16 to 1.16; P = .094). CONCLUSION In this cohort of patients treated for HL in Botswana, survival in PLWH (87% on long-standing ART) was at least as good as in individuals without HIV.

Association of Surge Conditions with Mortality Among Critically Ill Patients with COVID-19

Objective To determine whether surge conditions were associated with increased mortality. Design Multicenter cohort study. Setting U.S. ICUs participating in STOP-COVID. Patients Consecutive adults with COVID-19 admitted to participating ICUs between March 4 and July 1, 2020. Interventions None Measurements and Main Results The main outcome was 28-day in-hospital mortality. To assess the association between admission to an ICU during a surge period and mortality, we used two different strategies: (1) an inverse probability weighted difference-in-differences model limited to appropriately matched surge and non-surge patients and (2) a meta-regression of 50 multivariable difference-in-differences models (each based on sets of randomly matched surge- and non-surge hospitals). In the first analysis, we considered a single surge period for the cohort (March 23 – May 6). In the second, each surge hospital had its own surge period (which was compared to the same time periods in matched non-surge hospitals). Our cohort consisted of 4342 ICU patients (average age 60.8 [sd 14.8], 63.5% men) in 53 U.S. hospitals. Of these, 13 hospitals encountered surge conditions. In analysis 1, the increase in mortality seen during surge was not statistically significant (odds ratio [95% CI]: 1.30 [0.47-3.58], p = .6). In analysis 2, surge was associated with an increased odds of death (odds ratio 1.39 [95% CI, 1.34-1.43], p < .001). Conclusions Admission to an ICU with COVID-19 in a hospital that is experiencing surge conditions may be associated with an increased odds of death. Given the high incidence of COVID-19, such increases would translate into substantial excess mortality.