Adjustment for duration of employment in occupational epidemiology

PURPOSE

In occupational epidemiology, the healthy worker survivor effect can manifest as a time-dependent confounder because healthier workers can accrue greater amounts of exposure over longer periods of employment. For example, in occupational studies of radiation exposure that focus on cumulative annualized radiation dose, workers can accrue greater amounts of cumulative radiation exposure over longer periods of employment, while workers with longer periods of employment can transition into jobs with a reduced potential for annualized radiation exposure. The extent to which confounding arising from the healthy worker survivor effect impacts radiation risk estimates is unknown.

METHODS

We assessed the impact of the healthy worker survivor effect on estimates of radiation risk among nuclear workers in a Million Person Study cohort. In simulation studies, we contrasted the ability of marginal structural Cox models with inverse probability weighting and Cox proportional hazards models to account for time-dependent confounding arising from the healthy worker survivor effect.

RESULTS

Marginal structural Cox models and Cox proportional hazards models with flexible functional forms for duration of employment provided reliable results.

CONCLUSIONS

It is crucial to flexibly adjust for duration of employment to account for confounding arising from the healthy worker survivor effect in occupational epidemiology.

A non-linear connection between the total cholesterol to high-density lipoprotein cholesterol ratio and stroke risk: a retrospective cohort study from the China Health and Retirement Longitudinal Study

OBJECTIVE

The connection between total cholesterol to high-density lipoprotein cholesterol (TC/HDL-C) ratio and stroke risk is controversial. This study aims to examine the connection between the TC/HDL-C ratio and stroke in middle-aged and older individuals who are part of the China Health and Retirement Longitudinal Study (CHARLS).

METHODS

This study conducted a retrospective cohort analysis, enrolling a total of 10,184 participants who met the designated criteria from CHARLS between 2011 and 2012. We then used the Cox proportional-hazards regression model to analyze the relationship between the TC/HDL-C ratio and stroke risk. Using a Cox proportional hazards regression model with cubic spline functions and smooth curve fitting, we were able to identify the non-linear relationship between the TC/HDL-C ratio and stroke occurrence. The sensitivity and subgroup analyses were also performed to investigate the connection between TC/HDL-C ratio and stroke.

RESULTS

This study revealed a statistically significant association between the TC/HDL-C ratio and stroke risk in subjects aged 45 years or older after adjusting for risk factors (HR: 1.05, 95%CI 1.00-1.10, P = 0.0410). Furthermore, a non-linear connection between the TC/HDL-C ratio and stroke risk was detected, with a TC/HDL-C ratio inflection point of 3.71. We identified a significant positive connection between the TC/HDL-C ratio and stroke risk, when the TC/HDL-C ratio was less than 3.71 (HR: 1.25, 95%CI 1.07-1.45, P = 0.0039). However, their connection was not significant when the TC/HDL-C ratio exceeded 3.71 (HR: 1.00, 95%CI 0.94-1.06, P = 0.9232). The sensitivity analysis and subgroup analyses revealed that our findings were well-robust.

CONCLUSION

Our study demonstrated a positive, non-linear connection between the TC/HDL-C ratio and stroke risk in middle-aged and older individuals. There was a significant positive connection between the TC/HDL-C ratio and stroke risk, when the TC/HDL-C ratio was less than 3.71. The current research can be used as a guideline to support clinician consultation and optimize stroke prevention measures for middle-aged and older adults.

Elevated triglyceride-glucose-body mass index associated with lower probability of future regression to normoglycemia in Chinese adults with prediabetes: a 5-year cohort study

Objective

Despite the clear association of TyG-BMI with prediabetes and the progression of diabetes, no study to date has examined the relationship between TyG-BMI and the reversal of prediabetes to normoglycemia.

Methods

25,279 participants with prediabetes who had physical examinations between 2010 and 2016 were enrolled in this retrospective cohort study. The relationship between baseline TyG-BMI and regression to normoglycemia from prediabetes was examined using the Cox proportional hazards regression model in this study. Additionally, the nonlinear association between TyG-BMI and the likelihood of regression to normoglycemia was investigated using the Cox proportional hazards regression with cubic spline function. Competing risk multivariate Cox regression analysis was conducted, with progression to diabetes as a competing risk for prediabetes reversal to normoglycemia. Furthermore, subgroup analyses and a series of sensitivity analyses were performed.

Results

After adjusting for covariates, the results showed that TyG-BMI was negatively associated with the probability of returning to normoglycemia (per 10 units, HR=0.970, 95% CI: 0.965, 0.976). They were also nonlinearly related, with an inflection point for TyG-BMI of 196.46. The effect size (HR) for TyG-BMI to the right of the inflection point (TyG-BMI ≥ 196.46) and the probability of return of normoglycemia was 0.962 (95% CI: 0.954, 0.970, per 10 units). In addition, the competing risks model found a negative correlation between TyG-BMI and return to normoglycemia (SHR=0.97, 95% CI: 0.96-0.98). Sensitivity analyses demonstrated the robustness of our results.

Conclusion

This study demonstrated a negative and nonlinear relationship between TyG-BMI and return to normoglycemia in Chinese adults with prediabetes. Through active intervention, the combined reduction of BMI and TG levels to bring TyG-BMI down to 196.46 could significantly increase the probability of returning to normoglycemia.

Association between high-density lipoprotein cholesterol and reversion to normoglycemia from prediabetes: an analysis based on data from a retrospective cohort study

The available evidence on the connection between high-density lipoprotein cholesterol (HDL-C) levels and the reversion from prediabetes (Pre-DM) to normoglycemia is currently limited. The present research sought to examine the connection between HDL-C levels and the regression from Pre-DM to normoglycemia in a population of Chinese adults. This historical cohort study collected 15,420 Pre-DM patients in China who underwent health screening between 2010 and 2016. The present research used the Cox proportional hazards regression model to investigate the connection between HDL-C levels and reversion from Pre-DM to normoglycemia. The Cox proportional hazards regression model with cubic spline functions and smooth curve fitting was employed to ascertain the nonlinear association between HDL-C and reversion from Pre-DM to normoglycemia. Furthermore, a set of sensitivity analyses and subgroup analyses were employed. Following the adjustment of covariates, the findings revealed a positive connection between HDL-C levels and the likelihood of reversion from Pre-DM to normoglycemia (HR 1.898, 95% CI 1.758-2.048, P < 0.001). Furthermore, there was a non-linear relationship between HDL-C and the reversion from Pre-DM to normoglycemia in both genders, and the inflection point of HDL-C was 1.540 mmol/L in males and 1.620 mmol/L in females. We found a strong positive correlation between HDL-C and the reversion from Pre-DM to normoglycemia on the left of the inflection point (Male: HR 2.783, 95% CI 2.373-3.263; Female: HR 2.217, 95% CI 1.802-2.727). Our sensitivity analysis confirmed the robustness of these findings. Subgroup analyses indicated that patients with SBP < 140 mmHg and ever smoker exhibited a more pronounced correlation between HDL-C levels and the reversion from Pre-DM to normoglycemia. In contrast, a less robust correlation was observed among patients with SBP ≥ 140 mmHg, current and never smokers. This study provides evidence of a positive and nonlinear association between HDL-C levels and the reversion from Pre-DM to normoglycemia in Chinese patients. Implementing intensified intervention measures to control the HDL-C levels of patients with Pre-DM around the inflection point may substantially enhance the likelihood of regression to normoglycemia.

Investigation of the structure and magnitude of time-varying uncontrolled confounding in simulated cohort data analyzed using g-computation

BACKGROUND

When estimating the effect of time-varying exposures on longer-term outcomes, the assumption of conditional exchangeability or no uncontrolled confounding extends beyond baseline confounding to include time-varying confounding. We illustrate the structures and magnitude of uncontrolled time-varying confounding in exposure effect estimates obtained from g-computation when sequential conditional exchangeability is violated.

METHODS

We used directed acyclic graphs (DAGs) to depict time-varying uncontrolled confounding. We performed simulations and used g-computation to quantify the effects of each time-varying exposure for each DAG type. Models adjusting all time-varying confounders were considered the true (bias-adjusted) estimate. The exclusion of time-varying uncontrolled confounders represented the biased effect estimate and an unmet 'no uncontrolled confounding' assumption. True and biased estimates were compared across DAGs, with different magnitudes of uncontrolled confounding.

RESULTS

Time-varying uncontrolled confounding can present in several scenarios, including relationships into subsequently measured exposure(s), outcome, unmeasured confounder(s) and other measured confounder(s). In simulations, effect estimates obtained from g-computation were more biased in DAGs when the uncontrolled confounders were directly related to the outcome. Complex DAGs that included relationships between uncontrolled confounders and other variables and relationships where exposures caused uncontrolled confounders at the next time point resulted in the most biased effect estimates. In these complex DAGs, excluding uncontrolled confounders affected the multiple effect estimates.

CONCLUSIONS

Time-varying uncontrolled confounding has the potential to substantially impact observed effect estimates. Given the importance of longitudinal studies in advising public health, the impact of time-varying uncontrolled confounding warrants more recognition and evaluation using quantitative bias analysis.

Triglyceride to high-density lipoprotein cholesterol ratio is associated with regression to normoglycemia from prediabetes in adults: a 5-year cohort study in China

OBJECTIVE

The current body of evidence on the association between the ratio of triglycerides to high-density lipoprotein cholesterol (TG/HDL-c) and the reversal of prediabetes to normoglycemia remains limited. The aim of this study is to investigate the association between TG/HDL-c and the reversion to normoglycemia in patients with prediabetes.

METHODS

This retrospective cohort study included 15,107 individuals with prediabetes from 32 Chinese districts and 11 cities who completed health checks from 2010 to 2016. The Cox proportional-hazards regression model examined baseline TG/HDL-c and reversion to normoglycemia from prediabetes. Cox proportional hazards regression with cubic spline functions and smooth curve fitting determined the non-linear connection between TG/HDL-c and reversion to normoglycemia. We also ran sensitivity and subgroup analysis. By characterizing progression to diabetes as a competing risk for the reversal of prediabetes to normoglycemic event, a multivariate Cox proportional hazards regression model with competing risks was created.

RESULTS

Upon adjusting for covariates, the findings indicate a negative association between TG/HDL-c and the likelihood of returning to normoglycemia (HR = 0.869, 95%CI:0.842-0.897). Additionally, a non-linear relationship between TG/HDL-c and the probability of reversion to normoglycemia was observed, with an inflection point of 1.675. The HR on the left side of the inflection point was 0.748 (95%CI:0.699, 0.801). The robustness of our results was confirmed through competing risks multivariate Cox's regression and a series of sensitivity analyses.

CONCLUSION

The present study reveals a negative and non-linear correlation between TG/HDL-c and the reversion to normoglycemia among Chinese individuals with prediabetes. The findings of this study are anticipated to serve as a valuable resource for clinicians in managing dyslipidemia in prediabetic patients. Interventions aimed at reducing the TG/HDL-c ratio through the reduction of TG or elevation of HDL-c levels may substantially enhance the likelihood of achieving normoglycemia in individuals with prediabetes.

Association between coffee and incident heart failure: A prospective cohort study from the UK Biobank

BACKGROUND AND AIMS

The relationship between coffee consumption and heart failure (HF) incidence is inconclusive. This study aimed to explore the association between time-varying coffee consumption and incident HF using a longitudinal study design.

METHODS AND RESULTS

Data were obtained from the UK Biobank, comprising 497,503 adults (age, 56.5 ± 8.1 years; 54.6% women) who were free from HF at baseline in 2006-2010. The median follow-up time for the HF incidence was 11.9 years. Marginal structural models (MSM) were employed to adjust for potential time-varying confounders and account for bias caused by loss of follow-up. Furthermore, we used a restricted cubic spline to test and describe the nonlinear relationship between coffee consumption and HF risk. At baseline, 70.5% of participants reported drinking ≥1 cups/d coffee and 2.7% participants developed HF. After adjusting for potential confounders, we identified a nonlinear J-shaped association between coffee consumption and HF risk (P < 0.001). Compared with drinking coffee <1 cups/d, 1-2 cups/d (HR = 0.878; 95% CI: 0.838-0.920), 3-4 cups/d (HR = 0.920; 95% CI: 0.869-0.974) may be associated with a reduced risk of HF, while >6 cups/d (HR = 1.209; 95% CI: 1.056-1.385) may be associated with a higher risk of HF. However, sensitive analyses stratified by gender and smoking status indicated that >6 cups/d does not significantly increase the risk of HF. Additionally, the type of coffee was found to significant impact on the incidence of HF (P < 0.05).

CONCLUSION

In this large cohort of UK adults, moderate coffee consumption may reduce risk of HF incidence.

Association between body mass index and reversion to normoglycemia from impaired fasting glucose among Chinese adults: a 5-year cohort study

Objective

Evidence regarding the relationship between body mass index (BMI) and reversion to normoglycemia from prediabetes is still limited. The purpose of our study is to survey the link of BMI on reversion to normoglycemia among patients with impaired fasting glucose (IFG).

Methods

This study, a retrospective cohort, covered 32 regions and 11 cities in China and collected 258,74 IFG patients who underwent a health check from 2010 to 2016. We investigated the association between baseline BMI and reversion to normoglycemia in patients with IFG using the Cox proportional-hazards regression model. The nonlinear relationship between BMI and reversion to normoglycemia was determined using a Cox proportional hazards regression with cubic spline functions and smooth curve fitting. In addition, we also performed a series of sensitivity analyses and subgroup analyses. A competing risk multivariate Cox regression was performed using progression to diabetes as a competing risk for reversal of normoglycemic events.

Results

After adjusting covariates, the results showed that BMI was negatively related to the probability of reversion to normoglycemia (HR=0.977, 95%CI:0.971-0.984). Compared with participants with normal BMI(<24kg/m²), overweight (BMI:24-28kg/m²) participants with IFG had a 9.9% lower probability of returning to normoglycemia (HR=0.901,95%CI:0.863-0.939), while obese patients (BMI ≥ 28kg/m²) had a 16.9% decreased probability of reverting from IFG to normoglycemia (HR=0.831,95%CI:0.780-0.886). There was also a nonlinear relationship between them, and the inflection point of BMI was 21.7kg/m². The effect sizes (HR) on the left sides of the inflection point were 0.972(95%CI:0.964-0.980). The competing risks multivariate Cox's regression and sensitivity analysis demonstrated the robustness of our results.

Conclusion

This study demonstrates a negative and nonlinear relationship between BMI and reversion to normoglycemia in Chinese patients with IFG. Minimizing BMI to 21.7 kg/m² in patients with IFG through aggressive intervention may significantly increase the probability of returning to normoglycemia.

Two-phase analysis and study design for survival models with error-prone exposures

Increasingly, medical research is dependent on data collected for non-research purposes, such as electronic health records data. Health records data and other large databases can be prone to measurement error in key exposures, and unadjusted analyses of error-prone data can bias study results. Validating a subset of records is a cost-effective way of gaining information on the error structure, which in turn can be used to adjust analyses for this error and improve inference. We extend the mean score method for the two-phase analysis of discrete-time survival models, which uses the unvalidated covariates as auxiliary variables that act as surrogates for the unobserved true exposures. This method relies on a two-phase sampling design and an estimation approach that preserves the consistency of complete case regression parameter estimates in the validated subset, with increased precision leveraged from the auxiliary data. Furthermore, we develop optimal sampling strategies which minimize the variance of the mean score estimator for a target exposure under a fixed cost constraint. We consider the setting where an internal pilot is necessary for the optimal design so that the phase two sample is split into a pilot and an adaptive optimal sample. Through simulations and data example, we evaluate efficiency gains of the mean score estimator using the derived optimal validation design compared to balanced and simple random sampling for the phase two sample. We also empirically explore efficiency gains that the proposed discrete optimal design can provide for the Cox proportional hazards model in the setting of a continuous-time survival outcome.

Assessing the impact of unmeasured confounders for credible and reliable real-world evidence

PURPOSE

We review statistical methods for assessing the possible impact of bias due to unmeasured confounding in real world data analysis and provide detailed recommendations for choosing among the methods.

METHODS

By updating an earlier systematic review, we summarize modern statistical best practices for evaluating and correcting for potential bias due to unmeasured confounding in estimating causal treatment effect from non-interventional studies.

RESULTS

We suggest a hierarchical structure for assessing unmeasured confounding. First, for initial sensitivity analyses, we strongly recommend applying a recently developed method, the E-value, that is straightforward to apply and does not require prior knowledge or assumptions about the unmeasured confounder(s). When some such knowledge is available, the E-value could be supplemented by the rule-out or array method at this step. If these initial analyses suggest results may not be robust to unmeasured confounding, subsequent analyses could be conducted using more specialized statistical methods, which we categorize based on whether they require access to external data on the suspected unmeasured confounder(s), internal data, or no data. Other factors for choosing the subsequent sensitivity analysis methods are also introduced and discussed, including the types of unmeasured confounders and whether the subsequent sensitivity analysis is intended to provide a corrected causal treatment effect.

CONCLUSION

Various analytical methods have been proposed to address unmeasured confounding, but little research has discussed a structured approach to select appropriate methods in practice. In providing practical suggestions for choosing appropriate initial and, potentially, more specialized subsequent sensitivity analyses, we hope to facilitate the widespread reporting of such sensitivity analyses in non-interventional studies. The suggested approach also has the potential to inform pre-specification of sensitivity analyses before executing the analysis, and therefore increase the transparency and limit selective study reporting.

How unmeasured confounding in a competing risks setting can affect treatment effect estimates in observational studies

Background

Analysis of competing risks is commonly achieved through a cause specific or a subdistribution framework using Cox or Fine & Gray models, respectively. The estimation of treatment effects in observational data is prone to unmeasured confounding which causes bias. There has been limited research into such biases in a competing risks framework.

Methods

We designed simulations to examine bias in the estimated treatment effect under Cox and Fine & Gray models with unmeasured confounding present. We varied the strength of the unmeasured confounding (i.e. the unmeasured variable’s effect on the probability of treatment and both outcome events) in different scenarios.

Results

In both the Cox and Fine & Gray models, correlation between the unmeasured confounder and the probability of treatment created biases in the same direction (upward/downward) as the effect of the unmeasured confounder on the event-of-interest. The association between correlation and bias is reversed if the unmeasured confounder affects the competing event. These effects are reversed for the bias on the treatment effect of the competing event and are amplified when there are uneven treatment arms.

Conclusion

The effect of unmeasured confounding on an event-of-interest or a competing event should not be overlooked in observational studies as strong correlations can lead to bias in treatment effect estimates and therefore cause inaccurate results to lead to false conclusions. This is true for cause specific perspective, but moreso for a subdistribution perspective. This can have ramifications if real-world treatment decisions rely on conclusions from these biased results. Graphical visualisation to aid in understanding the systems involved and potential confounders/events leading to sensitivity analyses that assumes unmeasured confounders exists should be performed to assess the robustness of results.

Electronic supplementary material

The online version of this article (10.1186/s12874-019-0808-7) contains supplementary material, which is available to authorized users.