Noncollapsibility and its role in quantifying confounding bias in logistic regression

The impact of social isolation on functional disability in older people: A multi-cohort study

OBJECTIVES

We assessed the relationship between social isolation and functional disability in older people.

DESIGN

Comparison of longitudinal cohort studies.

SETTING AND PARTICIPANTS

Harmonised longitudinal datasets from the United States, England, European countries, Japan, Korea, China and Hong Kong.

METHODS

Social isolation was operationalised as a composite score with five domains, such as marital status, living alone, and social contact with others. Functional disability was defined as whether the cohort participant had any difficulty in activities of daily living (ADL). In each dataset, we used robust Poisson regression models to obtain the relative risks (RRs) and the corresponding 95 % confidence intervals (CI). We combined the RRs to synthesize a pooled estimate using meta-analysis with random-effects models.

RESULTS

Overall, the social isolation composite score was not associated with ADL disability (pooled RR = 1.05, 95 % CI [0.97-1.14], n = 40,119). Subgroup analysis suggested social isolation composite score was associated with ADL disability in Asian regions (pooled RR = 1.09, 95 % CI [1.02, 1.16], but not in Western regions (pooled RR = 1.01, 95 % CI [0.96, 1.07]). The relationships between different domains of social isolation and ADL disability were heterogeneous, except that no participation in any social clubs or religious groups was consistently associated with ADL disability (pooled RR = 1.12, 95 % CI [1.04, 1.21]).

CONCLUSION

Targeting social isolation may prevent decline in functional abilities in older adults, providing an avenue to active and healthy ageing. Nonetheless, interventions tackling social isolation should tailor to the unique cultural and social underpinnings. A limitation of the study is that reverse causality could not be ruled out definitively.

unmconf : an R package for Bayesian regression with unmeasured confounders

The inability to correctly account for unmeasured confounding can lead to bias in parameter estimates, invalid uncertainty assessments, and erroneous conclusions. Sensitivity analysis is an approach to investigate the impact of unmeasured confounding in observational studies. However, the adoption of this approach has been slow given the lack of accessible software. An extensive review of available R packages to account for unmeasured confounding list deterministic sensitivity analysis methods, but no R packages were listed for probabilistic sensitivity analysis. The R package unmconf implements the first available package for probabilistic sensitivity analysis through a Bayesian unmeasured confounding model. The package allows for normal, binary, Poisson, or gamma responses, accounting for one or two unmeasured confounders from the normal or binomial distribution. The goal of unmconf is to implement a user friendly package that performs Bayesian modeling in the presence of unmeasured confounders, with simple commands on the front end while performing more intensive computation on the back end. We investigate the applicability of this package through novel simulation studies. The results indicate that credible intervals will have near nominal coverage probability and smaller bias when modeling the unmeasured confounder(s) for varying levels of internal/external validation data across various combinations of response-unmeasured confounder distributional families.

A practical guide to estimating treatment effects in patients with rheumatic diseases using real-world data

OBJECTIVE

Randomized controlled trials are considered the gold standard in study methodology. However, due to their study design and inclusion criteria, these studies may not capture the heterogeneity of real-world patient populations. In contrast, the lack of randomization and the presence of both measured and unmeasured confounding factors could bias the estimated treatment effect when using observational data. While causal inference methods allow for the estimation of treatment effects, their mathematical complexity may hinder their application in clinical research.

METHODS

We present a practical, nontechnical guide using a common statistical package (Stata) and a motivational simulated dataset that mirrors real-world observational data from patients with rheumatic diseases. We demonstrate regression analysis, regression adjustment, inverse-probability weighting, propensity score (PS) matching and two robust estimation methods.

RESULTS

Although the methods applied to control for confounding factors produced similar results, the commonly used one-to-one PS matching method could yield biased results if not thoroughly assessed.

CONCLUSION

The guide we propose aims to facilitate the use of readily available methods in a common statistical package. It may contribute to robust and transparent epidemiological and statistical methods, thereby enhancing effectiveness research using observational data in rheumatology.

Explanatory and modifying factors of the association between sex and depression onset during adolescence: An exploratory study

BACKGROUND

The prevalence of Major Depressive Disorder (MDD) is twice as high in women as in men and this difference already emerges during adolescence. Because the mechanisms underlying this sex-difference remain poorly understood, we took a bottom-up approach to identify factors explaining the sex-MDD relationship.

METHODS

Data came from the TRacking Adolescents' Individual Lives Survey (TRAILS), a population study investigating youths' development from age 11 into adulthood. We assessed multiple baseline covariates (e.g., demographic, social and psychological) at ages 11-13 years and MDD onset at ages 19 and 25 years. In regression analyses, each covariate's role in the sex-MDD association as an effect modifier or confounder/explanatory variable was investigated. Replicability was evaluated in an independent sample.

RESULTS

The analyses identified no effect-modifiers. Baseline internalizing problems, behavioral inhibition, dizziness, comfort in classroom, physical complaints, attention problems, cooperation, self/effortful control, interpersonal life events and computer use partially explained the association between sex and MDD at age 19. The association between sex and MDD at age 25 was explained by largely the same variables, but also by shyness, acne, antisocial behavior, aggression, affection from peers and time spent shopping. The explanatory roles of internalizing problems, behavioral inhibition, negative events involving gossip/rumors and leisure-time spending (computer-use/shopping) were replicated.

LIMITATIONS

Potentially important baseline variables were not included or had low response rates. Gender roles or identification were not considered. Baseline MDD was not adjusted for.

CONCLUSION

The sex-MDD association is partially explained by sex differences in symptoms and vulnerability factors already present in early adolescence.

Estimation of the causal effect of sex on neonatal intensive care unit outcomes among very low birth weight infants

OBJECTIVE

Estimate the causal effect of sex on outcomes in the neonatal intensive care unit (NICU) among very low birth weight (VLBW) infants.

STUDY DESIGN

Retrospective cohort study using Vermont Oxford Network data to compare NICU outcomes for VLBW males versus females. Odds ratios (OR) for outcomes that differed significantly by sex were computed using standard unweighted analysis and inverse probability weighted (IPW) analysis to correct for selection bias.

RESULTS

Using standard analysis, males were significantly more likely to die before discharge and experience six other adverse outcomes. From IPW analysis, male sex caused a 56% increase in the odds of death before discharge (OR = 1.56, 95% confidence interval: 1.18-1.94). Standard unweighted results were significantly biased towards increased risk of adverse outcomes for males (p = 0.005) compared to IPW results for which three outcomes were no longer significantly associated with male sex.

CONCLUSION

Standard statistical methods generally overestimate the casual effect of sex among VLBW infants.

Comparing the performance of two-stage residual inclusion methods when using physician's prescribing preference as an instrumental variable: unmeasured confounding and noncollapsibility

Aim: The first objective is to compare the performance of two-stage residual inclusion (2SRI), two-stage least square (2SLS) with the multivariable generalized linear model (GLM) in terms of the reducing unmeasured confounding bias. The second objective is to demonstrate the ability of 2SRI and 2SPS in alleviating unmeasured confounding when noncollapsibility exists. Materials & methods: This study comprises a simulation study and an empirical example from a real-world UK population health dataset (Clinical Practice Research Datalink). The instrumental variable (IV) used is based on physicians' prescribing preferences (defined by prescribing history). Results: The percent bias of 2SRI in terms of treatment effect estimates to be lower than GLM and 2SPS and was less than 15% in most scenarios. Further, 2SRI was found to be robust to mild noncollapsibility with the percent bias less than 50%. As the level of unmeasured confounding increased, the ability to alleviate the noncollapsibility decreased. Strong IVs tended to be more robust to noncollapsibility than weak IVs. Conclusion: 2SRI tends to be less biased than GLM and 2SPS in terms of estimating treatment effect. It can be robust to noncollapsibility in the case of the mild unmeasured confounding effect.

Bayesian network-based Mendelian randomization for variant prioritization and phenotypic causal inference

Mendelian randomization is a powerful method for inferring causal relationships. However, obtaining suitable genetic instrumental variables is often challenging due to gene interaction, linkage, and pleiotropy. We propose Bayesian network-based Mendelian randomization (BNMR), a Bayesian causal learning and inference framework using individual-level data. BNMR employs the random graph forest, an ensemble Bayesian network structural learning process, to prioritize candidate genetic variants and select appropriate instrumental variables, and then obtains a pleiotropy-robust estimate by incorporating a shrinkage prior in the Bayesian framework. Simulations demonstrate BNMR can efficiently reduce the false-positive discoveries in variant selection, and outperforms existing MR methods in terms of accuracy and statistical power in effect estimation. With application to the UK Biobank, BNMR exhibits its capacity in handling modern genomic data, and reveals the causal relationships from hematological traits to blood pressures and psychiatric disorders. Its effectiveness in handling complex genetic structures and modern genomic data highlights the potential to facilitate real-world evidence studies, making it a promising tool for advancing our understanding of causal mechanisms.

Inclusion of binary proxy variables in logistic regression improves treatment effect estimation in observational studies in the presence of binary unmeasured confounding variables

We present a simulation study and application that shows inclusion of binary proxy variables related to binary unmeasured confounders improves the estimate of a related treatment effect in binary logistic regression. The simulation study included 60,000 randomly generated parameter scenarios of sample size 10,000 across six different simulation structures. We assessed bias by comparing the probability of finding the expected treatment effect relative to the modeled treatment effect with and without the proxy variable. Inclusion of a proxy variable in the logistic regression model significantly reduced the bias of the treatment or exposure effect when compared to logistic regression without the proxy variable. Including proxy variables in the logistic regression model improves the estimation of the treatment effect at weak, moderate, and strong association with unmeasured confounders and the outcome, treatment, or proxy variables. Comparative advantages held for weakly and strongly collapsible situations, as the number of unmeasured confounders increased, and as the number of proxy variables adjusted for increased.

Who prefers what? Correlates of preferences for next-generation HIV prevention products among a national U.S. sample of young men who have sex with men

INTRODUCTION

Pre-exposure prophylaxis (PrEP) has been available for young people for over a decade, yet only ∼15% of young people in the United States with indications for PrEP have a prescription for it. Next-generation PrEP modalities may address some of the challenges of daily oral PrEP. However, preferences for these products are unknown.

METHODS

From October 2020 to June 2021, we conducted an online survey of 737 cisgender, young men who have sex with men (age 15-24 years) without HIV across the United States who reported same-sex attraction or consensual sex with another man in the past 6 months. Participants completed a conjoint experiment comparing daily oral pills, event-driven oral pills, event-driven rectal douches, intramuscular injections, intravenous broadly neutralizing antibody (bnAb) infusions and subcutaneous implants. Participants ranked the products from most to least preferred. Exploded logit models examined the association between ranked preferences of PrEP modalities and socio-demographic and behavioural characteristics.

RESULTS

Participants' mean age was 21 years (SD = 2.3), and 56% identified as White. Nineteen percent were currently taking daily oral PrEP, and another 9% had previously taken it. Participants prioritized efficacy, absence of side effects and costs in the conjoint analyses. Daily oral PrEP had the highest preference ranking, followed by event-driven oral (OR = 0.89, p = 0.058), injectable (OR = 0.83, p = 0.005), implant (OR = 0.48, p < 0.0001), bnAb infusions (OR = 0.38, p < 0.0001) and rectal douches (OR = 0.24, p < 0.0001). There were differences in PrEP preferences across age, insurance status, sexual behaviour, PrEP use history, HIV and sexually transmitted infection (STI) testing history, and STI diagnoses (omnibus tests: p < 0.05). Participants also provided reasons for selecting their top product choice: ease of use for those who chose daily oral (99%) and daily event-driven (98.5%); feel more protected against HIV for those who chose injectable (95.4%) and implants (100%); not worrying about forgetting to take it for those who chose bnAbs (93.8%); and being able to stop taking it when they want for those who chose rectal douche (90.9%).

CONCLUSIONS

Next-generation modalities were less likely to be preferred over daily oral PrEP, with differences in the magnitude by socio-demographic and behavioural characteristics. Given the low uptake of daily oral PrEP, end-users' preferences for and concerns about PrEP products must be understood to ensure high acceptability and penetration.

Frameworks for estimating causal effects in observational settings: comparing confounder adjustment and instrumental variables

To estimate causal effects, analysts performing observational studies in health settings utilize several strategies to mitigate bias due to confounding by indication. There are two broad classes of approaches for these purposes: use of confounders and instrumental variables (IVs). Because such approaches are largely characterized by untestable assumptions, analysts must operate under an indefinite paradigm that these methods will work imperfectly. In this tutorial, we formalize a set of general principles and heuristics for estimating causal effects in the two approaches when the assumptions are potentially violated. This crucially requires reframing the process of observational studies as hypothesizing potential scenarios where the estimates from one approach are less inconsistent than the other. While most of our discussion of methodology centers around the linear setting, we touch upon complexities in non-linear settings and flexible procedures such as target minimum loss-based estimation and double machine learning. To demonstrate the application of our principles, we investigate the use of donepezil off-label for mild cognitive impairment. We compare and contrast results from confounder and IV methods, traditional and flexible, within our analysis and to a similar observational study and clinical trial.

Leveraging electronic health records to identify risk factors for recurrent pregnancy loss across two medical centers: a case-control study

Recurrent pregnancy loss (RPL), defined as 2 or more pregnancy losses, affects 5-6% of ever-pregnant individuals. Approximately half of these cases have no identifiable explanation. To generate hypotheses about RPL etiologies, we implemented a case-control study comparing the history of over 1,600 diagnoses between RPL and live-birth patients, leveraging the University of California San Francisco (UCSF) and Stanford University electronic health record databases. In total, our study included 8,496 RPL (UCSF: 3,840, Stanford: 4,656) and 53,278 Control (UCSF: 17,259, Stanford: 36,019) patients. Menstrual abnormalities and infertility-associated diagnoses were significantly positively associated with RPL in both medical centers. Age-stratified analysis revealed that the majority of RPL-associated diagnoses had higher odds ratios for patients <35 compared with 35+ patients. While Stanford results were sensitive to control for healthcare utilization, UCSF results were stable across analyses with and without utilization. Intersecting significant results between medical centers was an effective filter to identify associations that are robust across center-specific utilization patterns.

Subgroup State Prediction under Different Noise Levels Using MODWT and XGBoost

In medical states prediction, the observations of different individuals are generally assumed to follow an identical distribution, whereas precision medicine has a rigorous requirement for accurate subgroup analysis. In this research, an aggregated method is proposed by means of combining the results generated from different subgroup models and is compared with the original method for different denoising levels as well as the prediction gaps. The results using real data demonstrate the effectiveness of the aggregated method exhibiting superior performance such as 0.95 in AUC, 0.87 in F1, and 0.82 in sensitivity, particularly for the denoising level that is set to be 2. With respect to the variable importance, it is shown that some variables such as heart rate and lactate arterial become more important when the denoising level increases.

Metformin is associated with reduced COVID-19 severity in patients with prediabetes

AIMS

Studies suggest that metformin is associated with reduced COVID-19 severity in individuals with diabetes compared to other antihyperglycemics. We assessed if metformin is associated with reduced incidence of severe COVID-19 for patients with prediabetes or polycystic ovary syndrome (PCOS), common diseases that increase the risk of severe COVID-19.

METHODS

This observational, retrospective study utilized EHR data from 52 hospitals for COVID-19 patients with PCOS or prediabetes treated with metformin or levothyroxine/ondansetron (controls). After balancing via inverse probability score weighting, associations with COVID-19 severity were assessed by logistic regression.

RESULTS

In the prediabetes cohort, when compared to levothyroxine, metformin was associated with a significantly lower incidence of COVID-19 with "mild-ED" or worse (OR [95% CI]: 0.636, [0.455-0.888]) and "moderate" or worse severity (0.493 [0.339-0.718]). Compared to ondansetron, metformin was associated with lower incidence of "mild-ED" or worse severity (0.039 [0.026-0.057]), "moderate" or worse (0.045 [0.03-0.069]), "severe" or worse (0.183 [0.077-0.431]), and "mortality/hospice" (0.223 [0.071-0.694]). For PCOS, metformin showed no significant differences in severity compared to levothyroxine, but was associated with a significantly lower incidence of "mild-ED" or worse (0.101 [0.061-0.166]), and "moderate" or worse (0.094 [0.049-0.18]) COVID-19 outcome compared to ondansetron.

CONCLUSIONS

Metformin use is associated with less severe COVID-19 in patients with prediabetes or PCOS.

Examining socioeconomic status disparities in facility-based childbirth in Kenya: role of perceived need, accessibility, and quality of care

Background

Giving birth in health facilities with skilled birth attendants (SBAs) is one of the key efforts promoted to reduce preventable maternal deaths in sub-Saharan Africa. However, research has revealed large socioeconomic status (SES) disparities in facility-based childbirth. We seek to extend the literature on the factors underlying these SES disparities. Drawing on the Disparities in Skilled Birth Attendance (DiSBA) framework, we examined the contribution of three proximal factors—perceived need, accessibility, and quality of care—that influence the use of SBAs.

Methods

We used data from a survey conducted in Migori County, Kenya in 2016, among women aged 15–49 years who gave birth nine weeks before the survey (N = 1020). The primary outcome is facility-based childbirth. The primary predictors are wealth, measured in quintiles calculated from a wealth index based on principal component analysis of household assets, and highest education level attained. Proposed mediating variables include maternal perceptions of need, accessibility (physical and financial), and quality of care (antenatal services received and experience of care). Logistic regression with mediation analysis was used to investigate the mediating effects.

Results

Overall, 85% of women in the sample gave birth in a health facility. Women in the highest wealth quintile were more likely to give birth in a facility than women in the lowest quintile, controlling for demographic factors (adjusted odds ratio [aOR]: 2.97, 95% CI: 1.69–5.22). College-educated women were five times more likely than women with no formal education or primary education to give birth in a health facility (aOR: 4.96; 95% CI: 1.43–17.3). Women who gave birth in health facilities had higher perceived accessibility and quality of care than those who gave birth at home. The five mediators were estimated to account for between 15% and 48% of the differences in facility births between women in the lowest and higher wealth quintiles.

Conclusion

Our results confirm SES disparities in facility-based childbirth, with the proximal factors accounting for some of these differences. These proximal factors – particularly perceived accessibility and quality of care – warrant attention due to their relationship with facility-birth overall, and their impact on inequities in this care.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-022-05111-1.

Metformin is Associated with Reduced COVID-19 Severity in Patients with Prediabetes

Background

With the continuing COVID-19 pandemic, identifying medications that improve COVID-19 outcomes is crucial. Studies suggest that use of metformin, an oral antihyperglycemic, is associated with reduced COVID-19 severity in individuals with diabetes compared to other antihyperglycemic medications. Some patients without diabetes, including those with polycystic ovary syndrome (PCOS) and prediabetes, are prescribed metformin for off-label use, which provides an opportunity to further investigate the effect of metformin on COVID-19.

Participants

In this observational, retrospective analysis, we leveraged the harmonized electronic health record data from 53 hospitals to construct cohorts of COVID-19 positive, metformin users without diabetes and propensity-weighted control users of levothyroxine (a medication for hypothyroidism that is not known to affect COVID-19 outcome) who had either PCOS (n = 282) or prediabetes (n = 3136). The primary outcome of interest was COVID-19 severity, which was classified as: mild, mild ED (emergency department), moderate, severe, or mortality/hospice.

Results

In the prediabetes cohort, metformin use was associated with a lower rate of COVID-19 with severity of mild ED or worse (OR: 0.630, 95% CI 0.450 - 0.882, p < 0.05) and a lower rate of COVID-19 with severity of moderate or worse (OR: 0.490, 95% CI 0.336 - 0.715, p < 0.001). In patients with PCOS, we found no significant association between metformin use and COVID-19 severity, although the number of patients was relatively small.

Conclusions

Metformin was associated with less severe COVID-19 in patients with prediabetes, as seen in previous studies of patients with diabetes. This is an important finding, since prediabetes affects between 19 and 38% of the US population, and COVID-19 is an ongoing public health emergency. Further observational and prospective studies will clarify the relationship between metformin and COVID-19 severity in patients with prediabetes, and whether metformin usage may reduce COVID-19 severity.

Green space, air pollution, traffic noise and mental wellbeing throughout adolescence: Findings from the PIAMA study

BACKGROUND

Green space, air pollution and traffic noise exposure may be associated with mental health in adolescents. We assessed the associations of long-term exposure to residential green space, ambient air pollution and traffic noise with mental wellbeing from age 11 to 20 years.

METHODS

We included 3059 participants of the Dutch PIAMA birth cohort who completed the five-item Mental Health Inventory (MHI-5) at ages 11, 14, 17 and/or 20 years. We estimated exposure to green space (the average Normalized Difference Vegetation Index (NDVI) and percentages of green space in circular buffers of 300 m, 1000 m and 3000 m), ambient air pollution (particulate matter (PM₁₀ and PM_2.5), nitrogen dioxide, PM_2.5 absorbance and the oxidative potential of PM_2.5) and road traffic and railway noise (L_den) at the adolescents' home addresses at the times of completing the MHI-5. Associations with poor mental wellbeing (MHI-5 score ≤ 60) were assessed by generalized linear mixed models with a logit link, adjusting for covariates.

RESULTS

The odds of poor mental wellbeing at age 11 to 20 years decreased with increasing exposure to green space in a 3000 m buffer (adjusted odds ratio (OR) 0.78 [95% CI 0.68-0.88] per IQR increase in the average NDVI; adjusted OR 0.77 [95% CI 0.67-0.88] per IQR increase in the total percentage of green space). These associations persisted after adjustment for air pollution and road traffic noise. Relationships between mental wellbeing and green space in buffers of 300 m and 1000 m were less consistent. Higher air pollution exposure was associated with higher odds of poor mental wellbeing, but these associations were strongly attenuated after adjustment for green space in a buffer of 3000 m, traffic noise and degree of urbanization. Traffic noise was not related to mental wellbeing throughout adolescence.

CONCLUSIONS

Residential exposure to green space may be associated with a better mental wellbeing in adolescents.

Early changes in laboratory tests predict liver function damage in patients with moderate coronavirus disease 2019: a retrospective multicenter study

Background

Most patients with coronavirus disease 2019 demonstrate liver function damage. In this study, the laboratory test data of patients with moderate coronavirus disease 2019 were used to establish and evaluate an early prediction model to assess the risk of liver function damage.

Methods

Clinical data and the first laboratory examination results of 101 patients with moderate coronavirus disease 2019 were collected from four hospitals’ electronic medical record systems in Jilin Province, China. Data were randomly divided into training and validation sets. A logistic regression analysis was used to determine the independent factors related to liver function damage in patients in the training set to establish a prediction model. Model discrimination, calibration, and clinical usefulness were evaluated in the training and validation sets.

Results

The logistic regression analysis showed that plateletcrit, retinol-binding protein, and carbon dioxide combining power could predict liver function damage (P < 0.05 for all). The receiver operating characteristic curve showed high model discrimination (training set area under the curve: 0.899, validation set area under the curve: 0.800; P < 0.05). The calibration curve showed a good fit (training set: P = 0.59, validation set: P = 0.19; P > 0.05). A decision curve analysis confirmed the clinical usefulness of this model.

Conclusions

In this study, the combined model assesses liver function damage in patients with moderate coronavirus disease 2019 performed well. Thus, it may be helpful as a reference for clinical differentiation of liver function damage.

Trial registration retrospectively registered.