Bias In Observational Studies Research Articles

Exposure misclassification: an “immortal time” bias in observational studies of training load and injury

Several observational studies of the relationship between training load and injury have found increased risks of injury at low loads. These associations are expected because load is often assessed at the end of the injury follow-up period. As such, athletes who get injured earlier in the follow-up period will have systematically lower loads than athletes who get injured later in the follow-up period. In this commentary, we identify this problem as a type of exposure misclassification occurring from the misalignment of exposure measurement and start of follow-up. This methodological issue has previously been recognized in other areas of epidemiology as “immortal time bias”. We discuss how this bias can be prevented by aligning the measurement of load with the start of follow-up for injury.

Application of Mendelian randomization analysis in investigating the genetic background of blood biomarkers for colorectal cancer.

Colorectal cancer (CRC), a malignancy affecting the colon and rectum, ranks as the third most common cancer worldwide and the second leading cause of cancer-related deaths. Early detection of CRC is crucial for preventing metastasis, reducing mortality, improving prognosis, and enhancing patients' quality of life. Genetic factors play a significant role in CRC development, accounting for up to 35% of the disease risk. Genome-wide association studies have identified several genetic loci associated with CRC risk. However, these studies often lack direct evidence of causality. While traditional blood biomarkers such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) are widely used for CRC diagnosis and monitoring, their sensitivity and accuracy in early diagnosis are limited. Thus, there is a pressing need to develop new biomarkers that reflect the genetic background of CRC to improve early detection and diagnostic accuracy. In addition, understanding the genetic mechanisms underlying these biomarkers is essential for elucidating CRC pathogenesis and developing precise personalized treatment strategies. Mendelian randomization (MR) analysis, as an emerging epidemiological tool, can accurately assess the causal relationship between genetic variations and diseases by reducing confounding biases in observational studies. MR analysis has been applied in evaluating the causal impact of various blood biomarkers on CRC risk, shedding lights on the potential causal relationships between these biomarkers and CRC pathogenesis in the context of genetic background. In this review, we summarize the applications of MR analysis in studies of blood biomarkers for CRC, aiming to enhance the early diagnosis and personalized treatment of CRC.

Are prophylactic antibiotics effective in preventing osteoradionecrosis after high-risk dental extractions?

A systematic review and meta-analysis. To assess and compare the effectiveness of different adjunctive treatments during tooth extraction in post-radiotherapy head and neck patients for preventing osteoradionecrosis (ORN). Two independent reviewers conducted an electronic search through PubMed, Embase, Cochrane Libraries, Scopus and Web of Science databases using relevant search terms. Additionally, references of relevant articles were manually searched for eligible studies. Included were full-length articles (randomized controlled trials (RCTs), clinical trials, or observational studies) published in English, reporting on ORN occurrence after dental extraction, with or without adjunctive interventions, in patients who had previous radiotherapy. Data extraction was similarly performed by two independent researchers. For quality assessment, bias in observational studies was evaluated using the ROBINS-I tool, while the RoB 2 tool was used for RCTs. One-arm meta-analyses estimated pooled ORN prevalence rates for different adjunctive treatments based on patient-level outcomes. For studies with direct comparisons between two modalities, pooled odds ratios for ORN were calculated using pairwise meta-analyses. This systematic review followed PRISMA guidelines and was registered in the PROSPERO database. A total of 29 studies with 1520 patients were included, comprising 4 RCTs, 4 prospective studies, and 21 retrospective observational studies. Five adjunctive treatment modalities around the time of dental extractions were identified: hyperbaric oxygen (HBO), pentoxifylline-tocopherol (PENTO), antibiotics (ABX), platelet-rich fibrin and photobiomodulation. Of the 25 observational studies included, 11 had a serious risk of bias, while all RCTs but 1 had low risk of bias. Egger's test showed no publication bias for the HBO and ABX groups (p > 0.05). Antibiotics were used in 17 studies, with significant variations in preoperative and postoperative protocols. Most studies utilized penicillin or clindamycin. The pooled prevalence of ORN was significantly lower in the HBO (4.6%), PENTO (3.4%), and ABX (3.8%) groups compared to the Control (17.6%), as indicated by the non-overlapping 95% CIs between the Control and any of the intervention groups. In studies with direct comparisons, HBO exhibited lower odds of developing ORN than the Control (odds ratio [OR] = 0.27) and ABX (OR = 0.57), although these differences were not statistically significant. HBO, PENTO, and ABX show similar reduction in ORN rates compared to no intervention. Given its cost-effectiveness and accessibility, ABX may be the most practical prophylactic modality at present.

Association of serum uric acid with male sexual hormones and erectile dysfunction: a bidirectional 2-sample Mendelian randomization analysis.

Observational studies indicated that serum uric acid (SUA) was associated with male sexual hormones and erectile dysfunction (ED). However, their relationship was still heterogeneous. This study conducted 2-sample univariate mendelian randomization (UVMR) and multivariate mendelian randomization (MVMR) to explore the causal relationship between SUA and sexual hormones as well as ED. Genetic variants associated with SUA were derived from the UK Biobank database (N = 437 354). Outcomes from the IEU Open GWAS and summary data sets were sexual hormones (sex hormone-binding globulin [SHBG], testosterone, estradiol [E2], follicle-stimulating hormone, luteinizing hormone) and ED, with 3301 to 625 650 participants. UVMR analysis primarily utilized the inverse variance weighted method, complemented by MVMR analysis. Thorough sensitivity analyses were carried out to ensure the reliability of results. Moreover, mediation analysis was conducted to estimate the mediated effect between SUA and outcomes. The primary outcomes included results of UVMR and MVMR analysis and mediation analysis, along with sensitivity analyses involving the Cochran Q test, the MR Egger intercept test, leave-1-out analysis, and the MR-PRESSO method (mendelian randomization pleiotropy residual sum and outlier). UVMR analysis revealed that an elevated SUA level could decrease levels of SHBG (β = -0.10, P = 1.70 × 10-7) and testosterone (β = -0.10, P = 5.94 × 10-3) and had a positive causal effect on ED (odds ratio,1.10; P = .018). According to reverse mendelian randomization results, increased levels of SHBG (β = -0.06, P = 4.82 × 10-4) and E2 (β = -0.04, P = .037) could also reduce SUA levels. As shown by MVMR analysis, SUA had a negative effect on SHBG and testosterone levels (P <.05), while the significant causal relationship between SUA and ED disappeared. Furthermore, SHBG mediated 98.1% of the effect of SUA on testosterone levels. Results of other mendelian randomization analyses were not statistically significant. No pleiotropy was found by sensitivity analysis in this study. Given the causal relationship between SUA and sexual hormones, we must focus on SUA and E2 levels in men, especially patients with hypogonadism and ED. This study evaluated the causal effect of SUA on male sexual hormones and ED genetically for the first time, clarifying the common biases in observational studies and confirming the negative relationship between SUA and testosterone level. Limitations include a population based on European ancestry, some crossover of the samples, and unobserved confounding factors. Genetic studies provide evidence for the causal relationship between SUA and male sexual hormones (SHBG, testosterone, E2), while the relationship between SUA and ED should be further evaluated.

Comparative effectiveness of treatment approaches for early invasive breast cancer

BackgroundRandomized controlled trials (RCTs) suggested breast conserving therapy (BCT) and mastectomy have similar survival for early-stage breast cancer, whereas observational studies reported survival advantage for BCT. We aimed to address biases in observational studies to compare the effect of BCT and mastectomy on survival. MethodsWe emulated a target trial using institutional cancer registry. We included adult women diagnosed with early invasive first primary breast tumors ≤ 5 cm between July 2011 and December 2017. We used cloning, censoring, and weighting to estimate risk differences (RDs) and risk ratios (RRs) for all-cause mortality and recurrence or all-cause mortality between BCT and mastectomy (reference). ResultsOur study population comprised 534 observations with breast cancer. Median age was 56 years and 65 % were racial/ethnic minorities. The 8-year RD was 1.5 % (95 % confidence limits [CL]: −7.0 %, 9.8 %) and RR was 1.1 (95 % CL: 0.57, 2.2) for all-cause mortality. Results for recurrence or mortality were similar. ConclusionsOur results suggest that target trial emulation to mitigate selection and immortal-time biases in observational studies may generate estimates that are more compatible with RCTs when comparing the effects of BCT and mastectomy on survival. Studies with longer follow-up and more events are needed to confirm our findings.

Association Between Healthy Behaviors and Health Care Resource Use With Subsequent Positive Airway Pressure Therapy Adherence in OSA

BackgroundThe healthy adherer effect (HAE) has gained increasing attention as potential source of bias in observational studies examining the association of positive airway pressure (PAP) adherence with health outcomes in obstructive sleep apnea (OSA). Research questionIs adherence to PAP associated with healthy behaviors and healthcare resource use prior to device prescription? MethodsData from the IRSR Pays de la Loire Sleep Cohort were linked to health administrative data to identify proxies of heathy behaviors (HB) including adherence to cardiovascular (CV) drugs (medication possession ratio, [MPR]), cancer screening tests, influenza vaccination, alcohol and smoking consumption, and drowsiness-related road accidents during the two years preceding PAP onset in OSA patients. Multivariable regression analyses were conducted to evaluate the association of HB with subsequent PAP adherence. Healthcare resource use was evaluated according to subsequent PAP adherence. FindingsWe included 2,836 patients who had started PAP therapy between 2012 and 2018 (65% of whom were PAP adherent with mean daily use ≥4h/night). Being adherent to CV active drugs (MPR≥80%) and non-smoker were associated with a higher likelihood of PAP adherence (odds ratio, OR [95% confidence interval]: 1.43 [1.15; 1.77] and 1.37 [1.10; 1.71] respectively). Patients with no history of drowsiness-related road accidents were more likely to continue PAP (OR: 1.39 [1.04; 1.87]). PAP adherent patients used less healthcare resources 2 years before PAP initiation, than non-adherents (mean number of outpatient consultations: 19.0 vs 17.2, P=.003; hospitalization days: 5.7 vs 5.0, P=.04; emergency room visit: 30.7 vs 24.0% P=.0002). InterpretationPatients who adhere to PAP therapy of OSA were more health seeking and less healthcare users prior to device initiation than non-adherent patients. Until the HAE associated with PAP adherence is better understood, caution is warranted when interpreting the association of PAP adherence with CV health outcomes and healthcare resource use in non-randomized cohorts.

A comparison of random forest-based missing imputation methods for covariates in propensity score analysis.

Propensity score analysis (PSA) is a prominent method to alleviate selection bias in observational studies, but missing data in covariates is prevalent and must be dealt with during propensity score estimation. Through Monte Carlo simulations, this study evaluates the use of imputation methods based on multiple random forests algorithms to handle missing data in covariates: multivariate imputation by chained equations-random forest (Caliber), proximity imputation (PI), and missForest. The results indicated that PI and missForest outperformed other methods with respect to bias of average treatment effect regardless of sample size and missing mechanisms. A demonstration of these five methods with PSA to evaluate the effect of participation in center-based care on children's reading ability is provided using data from the Early Childhood Longitudinal Study, Kindergarten Class of 2010-2011. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Resampling-based confidence intervals and bands for the average treatment effect in observational studies with competing risks

The g-formula can be used to estimate the treatment effect while accounting for confounding bias in observational studies. With regard to time-to-event endpoints, possibly subject to competing risks, the construction of valid pointwise confidence intervals and time-simultaneous confidence bands for the causal risk difference is complicated, however. A convenient solution is to approximate the asymptotic distribution of the corresponding stochastic process by means of resampling approaches. In this paper, we consider three different resampling methods, namely the classical nonparametric bootstrap, the influence function equipped with a resampling approach as well as a martingale-based bootstrap version, the so-called wild bootstrap. For the latter, three sub-versions based on differing distributions of the underlying random multipliers are examined. We set up a simulation study to compare the accuracy of the different techniques, which reveals that the wild bootstrap should in general be preferred if the sample size is moderate and sufficient data on the event of interest have been accrued. For illustration, the resampling methods are further applied to data on the long-term survival in patients with early-stage Hodgkin’s disease.

Is there a correlation between peri-implantitis and systemic inflammation?

This systematic review investigates the association between peri-implantitis, an infectious/inflammatory disease sharing clinical and radiographic characteristics with periodontitis, and systemic inflammation. This study, adhering to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, systematically reviewed available evidence up to February 9, 2023. Searches were carried out across eight electronic databases (Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, Web of Science, Dentistry & Oral Sciences Source, Scopus, LILACS, and China Online), ClinicalTrials.gov, WHO International Clinical Trials Registry Platform (ICTRP), and gray literature. The review encompasses human studies, including randomised controlled trials, non-randomised intervention studies, cohort studies, case-control, and cross-sectional studies, yielded 27 full-text articles, with 11 clinical studies meeting inclusion criteria, and 9 articles included in the meta-analysis. Two independent reviewers carried out data extraction using the PECOS /PICOS tool (Patients, Exposure, Comparison, Outcomes, Study designs). The evaluation of the quality and risk of bias in observational studies, randomised controlled trials, and non-randomised studies of interventions was conducted utilising the Newcastle-Ottawa Scale (NOS), the revised Cochrane tool (RoB 2), and the ROBINS-I tool, respectively. Qualitative and quantitative analyses, including weighted mean differences (WMDs) and standard mean differences (SMD), were conducted using Stata/MP 17.0. Heterogeneity was assessed with the Q-statistic method. Pooled estimates, addressing potential inter-study heterogeneity, were calculated with random effects models. Significance criteria were set at p < 0.05. Publication bias was examined via funnel plot and Egger's test. Sensitivity analyses were predefined. Meta-analyses adhered to GRADE approach for inflammatory biomarkers/outcomes evaluation. Patients with peri-implantitis exhibited elevated levels of serum C-reactive protein (CRP) (standard mean difference (SMD): 4.68, 98.7% CI: 2.12 to 7.25), interleukin-6 (IL-6) (weighted mean difference (WMD): 6.27 pg/mL, 0% CI: 5.01 to 7.54), and white blood cell counts (WMD: 1.16 * 103/mL, 0% CI: 0.61 to 1.70) compared to those without peri-implantitis. The findings underscore a significant link between peri-implantitis and heightened systemic inflammation, emphasising the need for further research to elucidate the precise nature of this association.

Association between Ureaplasma urealyticum colonization and bronchopulmonary dysplasia in preterm infants: a systematic review and meta-analysis.

Bronchopulmonary dysplasia (BPD) is the most prevalent chronic lung disease in preterm infants. Studies have shown that Ureaplasma urealyticum (UU) infection is linked to its pathogenesis. However, it remains controversial whether UU colonization in preterm infants increases the risk of developing BPD. This study aimed to conduct a systematic review and meta-analysis to summarize the correlation between UU and BPD. We searched PubMed, Embase, the Cochrane Library, Web of Science, Wanfang Database, Chinese National Knowledge Infrastructure Database, Chinese Science and Technique Journal Database, and the China Biology Medicine disc from their inception to March 15, 2024. We included cohort and case-control studies investigating the association between UU infections and BPD in preterm infants, adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The Newcastle-Ottawa Scale was used for quality assessment. The outcome was defined as the continued need for oxygen or respiratory support at 28 days after birth (BPD28) or at 36 weeks postmenstrual age (BPD36). Considering the potential publication bias in observational studies, we used a random-effects meta-analysis model, assessed heterogeneity (I2), performed subgroup analyses, evaluated publication bias, and graded the quality of evidence. The meta-analysis included 36 cohort studies encompassing 5,991 participants. Among these, 20 reported on BPD28, 13 on BPD36, and 3 on both. The results indicated a significant association between UU colonization and BPD28 (odds ratio (OR): 2.26, 95% confidence interval (CI): 1.78-2.85, P < 0.00001, 23 studies, very low certainty of evidence) and BPD36 (OR: 2.13, 95% CI: 1.47-3.07, P < 0.0001, 16 studies, very low certainty of evidence). There is a correlation between UU colonization and the development of BPD in preterm infants. Future research should prioritize well-designed, large-scale, high-quality randomized controlled trials (RCTs) to comprehensively assess the risk of BPD in neonates following UU infection and to provide stronger evidence for clinical screening and prevention strategies to improve the prognosis of affected newborns. https://www.crd.york.ac.uk/, identifier (CRD42024524846).

Preventing immortal time bias in observational studies: a matter of design.

Preventing immortal time bias in observational studies: a matter of design.

Is there a sex difference in mortality rates in paediatric intensive care units?: a systematic review.

Mortality rates in infancy and childhood are lower in females than males. However, for children admitted to Paediatric Intensive Care Units (PICU), mortality has been reported to be lower in males, although males have higher admission rates. This female mortality excess for the subgroup of children admitted in intensive care is not well understood. To address this, we carried out a systematic literature review to summarise the available evidence. Our review studies the differences in mortality between males and females aged 0 to <18 years, while in a PICU, to examine whether there was a clear difference (in either direction) in PICU mortality between the two sexes, and, if present, to describe the magnitude and direction of this difference. Any studies that directly or indirectly reported the rates of mortality in children admitted to intensive care by sex were eligible for inclusion. The search strings were based on terms related to the population (those admitted into a paediatric intensive care unit), the exposure (sex), and the outcome (mortality). We used the search databases MEDLINE, Embase, and Web of Science as these cover relevant clinical publications. We assessed the reliability of included studies using a modified version of the risk of bias in observational studies of exposures (ROBINS-E) tool. We considered estimating a pooled effect if there were at least three studies with similar populations, periods of follow-up while in PICU, and adjustment variables. We identified 124 studies of which 114 reported counts of deaths by males and females which gave a population of 278,274 children for analysis, involving 121,800 (44%) females and 156,474 males (56%). The number of deaths and mortality rate for females were 5,614 (4.61%), and for males 6,828 (4.36%). In the pooled analysis, the odds ratio of female to male mortality was 1.06 [1.01 to 1.11] for the fixed effect model, and 1.10 [1.00 to 1.21] for the random effects model. Overall, males have a higher admission rate to PCU, and potentially lower overall mortality in PICU than females. www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=203009, identifier (CRD42020203009).

Unintentional unblinding in rheumatic disease trials.

The practice of blinding treatment assignment in randomised controlled trials mitigates important biases in observational studies. Unblinding, whereby study participants or investigators become aware of treatment assignments, is an important threat to the validity of trial results. Rheumatology studies might be particularly susceptible to unblinding because rheumatic disease therapies often cause high rates of idiosyncratic side-effects and frequently rely on subjective endpoints. Despite this susceptibility, the degree to which unblinding occurs in randomised controlled trials in rheumatic diseases has rarely been assessed during trials or acknowledged as a limitation. Rheumatologists should be aware of this important threat to the validity of trial results, assessments of unblinding should be undertaken, and strategies to prevent unblinding should be deployed when feasible.

Identifying Optimal Methods for Addressing Confounding Bias When Estimating the Effects of State-level Policies.

Policy evaluation studies that assess how state-level policies affect health-related outcomes are foundational to health and social policy research. The relative ability of newer analytic methods to address confounding, a key source of bias in observational studies, has not been closely examined. We conducted a simulation study to examine how differing magnitudes of confounding affected the performance of 4 methods used for policy evaluations: (1) the two-way fixed effects difference-in-differences model; (2) a 1-period lagged autoregressive model; (3) augmented synthetic control method; and (4) the doubly robust difference-in-differences approach with multiple time periods from Callaway-Sant'Anna. We simulated our data to have staggered policy adoption and multiple confounding scenarios (i.e., varying the magnitude and nature of confounding relationships). Bias increased for each method: (1) as confounding magnitude increases; (2) when confounding is generated with respect to prior outcome trends (rather than levels), and (3) when confounding associations are nonlinear (rather than linear). The autoregressive model and augmented synthetic control method had notably lower root mean squared error than the two-way fixed effects and Callaway-Sant'Anna approaches for all scenarios; the exception is nonlinear confounding by prior trends, where Callaway-Sant'Anna excels. Coverage rates were unreasonably high for the augmented synthetic control method (e.g., 100%), reflecting large model-based standard errors and wide confidence intervals in practice. In our simulation study, no single method consistently outperformed the others, but a researcher's toolkit should include all methodologic options. Our simulations and associated R package can help researchers choose the most appropriate approach for their data.

Comparative effectiveness of mRNA-1273 and BNT162b2 COVID-19 vaccines in immunocompromised individuals: a systematic review and meta-analysis using the GRADE framework.

Despite representing only 3% of the US population, immunocompromised (IC) individuals account for nearly half of the COVID-19 breakthrough hospitalizations. IC individuals generate a lower immune response after vaccination in general, and the US CDC recommended a third dose of either mRNA-1273 or BNT162b2 COVID-19 vaccines as part of their primary series. Influenza vaccine trials have shown that increasing dosage could improve effectiveness in IC populations. The objective of this systematic literature review and pairwise meta-analysis was to evaluate the clinical effectiveness of mRNA-1273 (50 or 100 mcg/dose) vs BNT162b2 (30 mcg/dose) in IC populations using the GRADE framework. The systematic literature search was conducted in the World Health Organization COVID-19 Research Database. Studies were included in the pairwise meta-analysis if they reported comparisons of mRNA-1273 and BNT162b2 in IC individuals ≥18 years of age; outcomes of interest were symptomatic, laboratory-confirmed SARS-CoV-2 infection, SARS-CoV-2 infection, severe SARS-CoV-2 infection, hospitalization due to COVID-19, and mortality due to COVID-19. Risk ratios (RR) were pooled across studies using random-effects meta-analysis models. Outcomes were also analyzed in subgroups of patients with cancer, autoimmune disease, and solid organ transplant. Risk of bias was assessed using the Newcastle-Ottawa Scale for observational studies. Evidence was evaluated using the GRADE framework. Overall, 17 studies were included in the pairwise meta-analysis. Compared with BNT162b2, mRNA-1273 was associated with significantly reduced risk of SARS-CoV-2 infection (RR, 0.85 [95% CI, 0.75-0.97]; P=0.0151; I2=67.7%), severe SARS-CoV-2 infection (RR, 0.85 [95% CI, 0.77-0.93]; P=0.0009; I2=0%), COVID-19-associated hospitalization (RR, 0.88 [95% CI, 0.79-0.97]; P<0.0001; I2=0%), and COVID-19-associated mortality (RR, 0.63 [95% CI, 0.44-0.90]; P=0.0119; I2=0%) in IC populations. Results were consistent across subgroups. Because of sample size limitations, relative effectiveness of COVID-19 mRNA vaccines in IC populations cannot be studied in randomized trials. Based on nonrandomized studies, evidence certainty among comparisons was type 3 (low) and 4 (very low), reflecting potential biases in observational studies. This GRADE meta-analysis based on a large number of consistent observational studies showed that the mRNA-1273 COVID-19 vaccine is associated with improved clinical effectiveness in IC populations compared with BNT162b2.

Introduction of a tool to assess Risk of Bias in Non-randomized Studies-of Exposure (2022)

This article introduces the contents of the latest edition Risk of Bias in Non-randomized Studies-of Exposure (ROBINS-E) published in June 2022 [ROBINS-E (2022)], and gives some examples about its usage. ROBINS-E is a tool for assessing the risk of bias in non-randomized studies-of exposure. Compared with ROBINS-E (2019), ROBINS-E (2022) adds more bias for observational studies, covers a more comprehensive range of bias, and adds the assessment of the external authenticity of the study. ROBINS-E (2022) adds a preliminary evaluation process to improve the efficiency of evaluation. In addition, ROBINS-E (2022) visualizes and instrumentalizes the use of signal problems in the form of path graph, making it more convenient to use. ROBINS-E (2022), although more consideration has been given to the issue of co-exposure, still does not address the problem of effect modification in co-exposure, and there is still room to expand the applicable research.

Development of the TrAnsparent ReportinG of observational studies Emulating a Target trial (TARGET) guideline

BackgroundObservational studies are increasingly used to inform health decision-making when randomised trials are not feasible, ethical or timely. The target trial approach provides a framework to help minimise common biases...

Mendelian randomization shows depression increases the risk of type 2 diabetes.

Introduction: Type 2 diabetes (T2D) is associated with severe mental illnesses (SMIs), such as schizophrenia, bipolar disorder, and depression. However, causal relationships between SMIs and T2D remain unclear owing to potential bias in observational studies. We aimed to characterize the causal effect of SMI liability on T2D using two-sample Mendelian randomization (MR). Methods: The causality between liability to SMI and T2D was investigated using the inverse-variance weighted (IVW), MREgger, MR-Egger with a simulation extrapolation, weighted median, and the MR pleiotropy residual sum and outlier method. Similarly, we performed additional MR which can detect the reverse causation effect by switching exposure and outcome for T2D liability for SMI. To further consider pleiotropic effects between SMIs, multivariable MR analysis was performed after accounting for the other traits. Results: In the univariable IVW method, depression showed a causal effect on T2D (odds ratio [OR]: 1.128, 95% confidence interval [CI]: 1.024-1.245, p = 0.014). Multinomial MR more strongly supported these results (IVW OR: 1.197, 95% CI: 1.069, 1.340, p = 0.002; MR-Egger OR: 1.198, 95% CI: 1.062, 1.349, p = 0.003). Bidirectional MR showed absence of reversecausality between depression and T2D. However, causal relationship of bipolar and schizophrenia on T2D was not detected. Discussion: Careful attention is needed for patients with depression regarding T2D prevention and treatment.

Essential Concepts for Reducing Bias in Observational Studies.

Randomized controlled trials (RCTs) are the gold standard study design for clinical research, as prospective randomization, at least in theory, balances any differences that can exist between groups (including any differences not measured as part of the study) and isolates the studied treatment effect. Any remaining imbalances after randomization are attributable to chance. However, there are many barriers to conducting RCTs within pediatric populations, including lower disease prevalence, high costs, inadequate funding, and additional regulatory requirements. Researchers thus frequently use observational study designs to address many research questions. Observational studies, whether prospective or retrospective, do not involve randomization and thus have more potential for bias when compared with RCTs because of imbalances that can exist between comparison groups. If these imbalances are associated with both the exposure of interest and the outcome, then failure to account for these imbalances may result in a biased conclusion. Understanding and addressing differences in sociodemographic and/or clinical characteristics within observational studies are thus necessary to reduce bias. Within this Method/ology submission we describe techniques to minimize bias by controlling for important measurable covariates within observational studies and discuss the challenges and opportunities in addressing specific variables.

Multi-morbidity and its association with common cancer diagnoses: a UK Biobank prospective study

BackgroundWhilst multi-morbidity is known to be a concern in people with cancer, very little is known about the risk of cancer in multi-morbid patients. This study aims to investigate the risk of being diagnosed with lung, colorectal, breast and prostate cancer associated with multi-morbidity.MethodsWe investigated the association between multi-morbidity and subsequent risk of cancer diagnosis in UK Biobank. Cox models were used to estimate the relative risks of each cancer of interest in multi-morbid participants, using the Cambridge Multimorbidity Score. The extent to which reverse causation, residual confounding and ascertainment bias may have impacted on the findings was robustly investigated.ResultsOf the 436,990 participants included in the study who were cancer-free at baseline, 21.6% (99,965) were multi-morbid (≥ 2 diseases). Over a median follow-up time of 10.9 [IQR 10.0–11.7] years, 9,019 prostate, 7,994 breast, 5,241 colorectal, and 3,591 lung cancers were diagnosed. After exclusion of the first year of follow-up, there was no clear association between multi-morbidity and risk of colorectal, prostate or breast cancer diagnosis. Those with ≥ 4 diseases at recruitment had double the risk of a subsequent lung cancer diagnosis compared to those with no diseases (HR 2.00 [95% CI 1.70–2.35] p for trend < 0.001). These findings were robust to sensitivity analyses aimed at reducing the impact of reverse causation, residual confounding from known cancer risk factors and ascertainment bias.ConclusionsIndividuals with multi-morbidity are at an increased risk of lung cancer diagnosis. While this association did not appear to be due to common sources of bias in observational studies, further research is needed to understand what underlies this association.