Marginal structural Cox models for estimating the association between β-interferon exposure and disease progression in a multiple sclerosis cohort

Risk of stroke admission after long-term exposure to PM1: Evidence from a large cohort in South China

BACKGROUND

Limited attention has been paid to the health effects of long-term PM1 exposure on stroke admission. Current investigations exploring the long-term PM exposure effect are largely based on observational studies, and PM generally is not allocated randomly to participants. Using traditional regression models might confuse messaging and hinder policy recommendations for pollution control and disease prevention policies.

METHODS

We conducted a cohort study among 36,271 adults from one of the largest cities in China in 2015 and followed up through 2020. Hazard ratios of stroke admissions following long-term PM1 exposure were estimated via a causal inference approach, marginal structural time-varying Cox proportional hazard model, accounting for multiple confounders. Additionally, several sensitivity analyses and impact modification analyses were carried out.

RESULTS AND DISCUSSION

Associations with 1 μg/m3 increase in long-term PM1 were identified for total (HR, 1.079; 95 %CI, 1.012-1.151) and ischemic stroke admissions (HR, 1.092; 95 %CI, 1.018-1.171). The harmful associations varied with exposure duration, initially increasing and then decreasing. The 2-3 years cumulative exposure was associated with a 3.3-5.4 % raised risk for total stroke. For every 1 μg/m³ increase in long-term PM1 exposure, females exhibited a higher risk of both total and ischemic stroke (13 % and 16 %) than men (4 % and 5 %). Low-exposure individuals (whose annual PM1 concentrations were under the third quartile among the annual concentrations for all the participants) exhibited greater sensitivity to PM1 effects (total stroke: 1.079 vs. 1.107; ischemic stroke: 1.092 vs. 1.116). The results underline the importance of safeguarding low-exposed people in highly polluted areas and suggest that long-term PM1 exposure may increase stroke admission risk, warranting attention to vulnerable groups.

Lipid control and stroke risk in atrial fibrillation patients treated with direct oral anticoagulants and statins

INTRODUCTION

The risk of ischemic stroke and intracerebral hemorrhage (ICH) with intensive lipid control by statins among patients with atrial fibrillation (AF) who require direct oral anticoagulants (DOAC) is unclear. We aimed to determine the risks of ischemic stroke and ICH in AF patients treated with DOAC and statins.

PATIENTS AND METHODS

In a population-based retrospective cohort study, we identified AF patients concurrently on DOAC and statins from 2015 to 2021 in Hong Kong. Primary outcome was ischemic stroke. Secondary outcomes were ICH and death. We correlated study outcomes with low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) as time-varying, continuous variables with restricted cubic spline. In secondary analyses, the risks of study outcomes with statin intensity (low, moderate, high) were determined by multivariable time-dependent marginal structural Cox models.

RESULTS

We identified 32,752 AF patients co-prescribed with DOAC and statins. Lower LDL-C (p < 0.001) and higher HDL-C (p < 0.001) levels were associated with lower risk of ischemic stroke but not significantly associated with ICH. LDL-C of <1.8 mmol/L (70 mg/dL) was not associated with mortality (19.6% vs 18.4%, difference 1.2% [95% CI -0.35 to 2.13]). High-intensity statin was associated with a lower risk of ischemic stroke compared with low-intensity statin (weighted Cox-specific hazard ratio [95% CI]: 0.82 [0.67-0.99], p = 0.040) independent of LDL-C levels. Similar associations were found in 11,444 AF patients with a history of ischemic stroke.

DISCUSSION AND CONCLUSION

Intensive lipid control by high-intensity statins was associated with a lower risk of ischemic stroke in AF patients who required DOACs and did not appear to increase the risk of ICH.

Methodological considerations for observational studies of treatment effectiveness in neurology: a clinician's guide

Data from cohorts, registries, randomised trials, electronic medical records and administrative claims databases have increasingly been used to inform the use of therapies for neurological diseases. While novel sophisticated methods are enabling us to use existing data to guide treatment decisions, the complexity of statistical methodology is making appraisal of clinical evidence increasingly demanding. In this narrative review, we provide a brief overview of the most commonly used methods for evaluation of treatment effectiveness in neurology. This primer discusses complementarity of randomised and non-randomised study designs, sources of observational data, different forms of bias and the appropriate mitigation strategies, statistical significance, Bayesian approaches and provides an overview of multivariable regression models, propensity score-based models, causal inference, mediation analysis and Mendelian randomisation.

Association between piperacillin/tazobactam use and acute kidney injury in critically ill patients: a retrospective multicentre cohort study

BACKGROUND

Piperacillin/tazobactam is one of the most common antibiotics prescribed in the ICU and the combination of piperacillin/tazobactam with vancomycin has been associated with acute kidney injury (AKI) in critically ill patients. However, data on the risk of AKI with piperacillin/tazobactam, despite vancomycin co-exposure, are lacking.

OBJECTIVES

To investigate the association of piperacillin/tazobactam with AKI and renal replacement therapy (RRT) among adult ICU patients.

METHODS

We analysed data from patients included in two open access databases (MIMIC-IV and eICU). Critically ill patients who received piperacillin/tazobactam or cefepime (a cephalosporin with similar broad-spectrum activity to piperacillin/tazobactam) during their first ICU stay were eligible for the study. Marginal structural Cox models, accounting for time-fixed covariates and time-dependent covariates were performed. The primary outcomes were AKI and need of RRT.

RESULTS

A total of 20 107 patients were included, with 11 213 in the piperacillin/tazobactam group and 8894 in the cefepime group. Exposure to piperacillin/tazobactam was associated with AKI (HR 1.77; 95% CI 1.51-2.07; P < 0.001) and with need of RRT (HR 1.31; 95% CI 1.08-1.57; P = 0.005). Tests for interaction were not statistically significant for occurrence of AKI and RRT in the subgroup of patients exposed to vancomycin or not (P = 0.26 and P = 0.6, respectively).

CONCLUSIONS

In critically ill patients, exposure to piperacillin/tazobactam was associated with increased risk of AKI and with increased risk of RRT, regardless of combination therapy with vancomycin.

Using instrumental variables to correct for bias in real-world cohort studies of the effects of disease-modifying treatment in MS

BACKGROUND

Estimating the effect of disease-modifying treatment of MS in observational studies is impaired by bias from unmeasured confounders, in particular indication bias.

OBJECTIVE

To show how instrumental variables (IVs) reduce bias.

METHODS

All patients with relapsing onset of MS 1996-2010, identified by the nationwide Danish Multiple Sclerosis Registry, were followed from onset. Exposure was treatment index throughout the first 12 years from onset, defined as a cumulative function of months without and with medium- or high-efficacy treatment, and outcomes were hazard ratios (HRs) per unit treatment index for sustained Expanded Disability Scale Score (EDSS) 4 and 6 adjusted for age at onset and sex, without and with an IV. We used the onset cohort (1996-2000; 2001-2005; 2006-2010) as an IV because treatment index increased across the cohorts.

RESULTS

We included 6014 patients. With conventional Cox regression, HRs for EDSS 4 and 6 were 1.15 [95% CI: 1.13-1.18] and 1.17 [1.13-1.20] per unit treatment index. Only with IVs, we confirmed a beneficial effect of treatment with HRs of 0.86 [0.81-0.91] and 0.82 [0.74-0.90].

CONCLUSION

The use of IVs eliminates indication bias and confirms that treatment is effective in delaying disability. IVs could, under some circumstances, be an alternative to marginal structural models.

Adverse Impact of Sodium Bicarbonate Administration on Multiple Outcomes in Acute Pancreatitis Patients With Hyperlactatemia

OBJECTIVE

Hyperlactatemia is likely to occur among patients with acute pancreatitis (AP). Sodium bicarbonate (SB) therapy could be applied to correct potential detrimental acidic disturbances, but the exact impact of SB treatment is unknown. This study aims to investigate the impact of SB on AP patients complicated with hyperlactatemia.

METHODS

The study was conducted based on the database named Medical Information Mart for Intensive Care-IV (MIMIC-IV). Propensity matching (PSM) and inverse probability weighting (IPTW) were used to balance the baseline differences. Multivariate regression and marginal structural Cox models were performed to investigate the association between SB and multiple outcomes.

RESULTS

Three hundred fifty-three AP patients with hyperlactatemia (initial serum lactate, >2.0 mmol/L) were extracted from the MIMIC-IV database. We found that SB treatment was significantly associated with worse multi-outcomes of AP patients with hyperlactatemia (in-hospital mortality: hazard ratio, 2.46; 95% confidence interval, 1.38-4.39; P < 0.01). Further analysis through marginal structural Cox models showed that SB had adverse impact on in-hospital prognosis of patients with severe lactic acidosis (pH < 7.15,lactate > 2.0 mmol/L).

CONCLUSION

Sodium bicarbonate might not be an appropriate treatment for AP patients with hyperlactatemia (lactate > 2.0 mmol/L) or with severe lactic acidosis (pH < 7.15, lactate > 2.0 mmol/L).

Impact of early heparin therapy on outcomes in patients with solid malignancy associated sepsis: a marginal structural model causal analyse

Background: Previous studies documented that heparin can inhibit the invasion and metastasis of tumors, but its role on outcomes in patients with solid malignancy complicated sepsis remains unclear. Methods: A retrospective cohort study was conducted in critically ill patients with solid malignancy associated sepsis from the Medical Information Mart for Intensive Care (MIMIC)-IV database. The primary endpoint was intensive care unit (ICU) mortality, secondary outcomes were thrombosis and hospital mortality. Propensity score matching (PSM), marginal structural Cox model (MSCM), cox proportional hazards model, stratification analysis and E-value were used to account for baseline differences, time-varying confounding and unmeasured variables. Results: A total of 1,512 patients with solid malignancy complicated sepsis were enrolled, of which 683 in the heparin group with intensive care unit mortality, thrombosis rate and hospital mortality were 9.7%, 5.4%, 16.1%, and 829 in the non-heparin group with ICU mortality, thrombosis rate and hospital mortality were 14.6%, 12.5%, 22.6%. Similar results were observed on outcomes for patients with PSM (ICU mortality hazard ratio [HR] 0.61, 95% confidence interval [CI] 0.41-0.92), thrombosis rate (HR 0.42, 95% confidence interval 0.26-0.68); hospital mortality HR 0.70, 95% CI 0.50-0.99). marginal structural Cox model further reinforced the efficacy of heparin in reducing ICU mortality (HR 0.48, 95% CI 0.34-0.68). Logistic regression and Cox regression model showed heparin use also markedly reduced thrombosis (HR 0.42; 95% CI 0.26-0.68; p < 0.001) and hospital mortality (HR 0.70; 95% CI 0.50-0.99; p = 0.043). Stratification analysis with the MSCM showed an effect only those with digestive system cancer (HR 0.33, 95% CI 0.16-0.69). Conclusion: Early heparin therapy improved outcomes in critically ill patients with solid malignancy complicated sepsis. These results are evident especially in those with digestive system cancer. A prospective randomized controlled study should be designed to further assess the relevant findings.

Effect of sodium bicarbonate infusion on hospital mortality in acute kidney injury patients with metabolic acidosis

Background

Physicians usually consider that sodium bicarbonate (SB) infusion can be used for metabolic acidosis; however, there is little evidence available to assess its effect on hospital mortality in large AKI cohorts. Here, we investigated the effect of SB infusion in patients with AKI complicated by metabolic acidosis.

Method

Patients with AKI complicated by metabolic acidosis were screened from the MIMIC-IV database. A propensity score analysis (PSA) was used to decrease baseline differences in the probability of receiving SB. The marginal structural Cox model (MSCM) was employed to adjust for both baseline and time-varying confounding factors.

Results

A total of 1853 patients with AKI complicated with metabolic acidosis were included in our study. A total of 390 pairs of patients were divided into an SB infusion group and a non-SB infusion group. The SB infusion group had more serious and worse laboratory indicators, including lower pH [7.19 (0.11) vs. 7.26 (0.07)] and bicarbonate concentration (BC) [12.36 (4.26) vs. 15.96 (3.25) mmol/l]. While there was no significant effect on overall hospital mortality in AKI patients complicated with metabolic acidosis (p = 0.056), SB infusion was observed to have beneficial correlation on hospital mortality in patients with high AG acidosis (AG > 18 mmol/L) (p = 0.012). Similar results were replicated with the MSCM.

Conclusion

We found that SB infusion in AKI patients with metabolic acidosis is not beneficial for hospital mortality. However, SB infusion for AKI patients and high AG metabolic acidosis significantly improved hospital mortality. Further larger randomized controlled trials are needed to confirm these results.

Air Pollution and Cardiovascular and Thromboembolic Events in Older Adults With High-Risk Conditions

Little epidemiologic research has focused on pollution-related risks in medically vulnerable or marginalized groups. Using a nationwide 50% random sample of 2008-2016 Medicare Part D-eligible fee-for-service participants in the United States, we identified a cohort with high-risk conditions for cardiovascular and thromboembolic events (CTEs) and linked individuals with seasonal average zip-code-level concentrations of fine particulate matter (particulate matter with an aerodynamic diameter ≤ 2.5 μm (PM2.5)). We assessed the relationship between seasonal PM2.5 exposure and hospitalization for each of 7 CTE-related causes using history-adjusted marginal structural models with adjustment for individual demographic and neighborhood socioeconomic variables, as well as baseline comorbidity, health behaviors, and health-service measures. We examined effect modification across geographically and demographically defined subgroups. The cohort included 1,934,453 individuals with high-risk conditions (mean age = 77 years; 60% female, 87% White). A 1-μg/m3 increase in PM2.5 exposure was significantly associated with increased risk of 6 out of 7 types of CTE hospitalization. Strong increases were observed for transient ischemic attack (hazard ratio (HR) = 1.039, 95% confidence interval (CI): 1.034, 1.044), venous thromboembolism (HR = 1.031, 95% CI: 1.027, 1.035), and heart failure (HR = 1.019, 95% CI: 1.017, 1.020). Asian Americans were found to be particularly susceptible to thromboembolic effects of PM2.5 (venous thromboembolism: HR = 1.063, 95% CI: 1.021, 1.106), while Native Americans were most vulnerable to cerebrovascular effects (transient ischemic attack: HR = 1.093, 95% CI: 1.030, 1.161).

Potential causal links between long-term ambient particulate matter exposure and cerebrovascular mortality: Insights from a large cohort in southern China

Cohort studies conducted in North America and Europe have linked cerebrovascular mortality to long-term exposure to particulate matter (PM). However, limited evidence from large cohorts in high-exposure areas and the traditional approach of association assessment may cause residual confounding issues. In this study, we aimed to investigate the causal links between cerebrovascular mortality and long-term exposure to PM_2.5, PM₁₀, and PM_2.5-10 in an ongoing cohort study with 580,757 participants in southern China. Using satellite-based estimates of PM concentration at a 1-km² spatial resolution, we assigned exposure levels to each participant and used the marginal structural Cox model to assess the association between PM exposure and cerebrovascular mortality while accounting for time-varying covariates. We also explored the potential modification effects of sociodemographic and behavioral factors on the PM-health associations. Adjusted hazard ratios (HR) for overall cerebrovascular mortality were 1.041 (95% confidence interval (CI): 1.034-1.049) and 1.032 (95% CI: 1.026-1.038) for each 1 μg/m³ increase in PM_2.5, and PM₁₀, respectively. Similar trends were observed in the mortality risk from stroke and ischemic stroke, with HRs ranging from 1.040 to 1.069 and 1.025 to 1.052, respectively, across 2 p.m. exposures. The impact of PM exposure was generally more apparent among women, participants with primary school diplomas and below, and the subgroup under low-exposure. Multiple sensitivity analyses confirmed the robustness of the results. In conclusion, this sizable prospective cohort study hypothesizes causal links between long-term PM exposure and cerebrovascular mortality, particularly among vulnerable participants, supporting the rationale for reducing PM concentration in China to reduce cerebrovascular mortality.

Effectiveness of early heparin therapy on outcomes in critically ill patients with sepsis-induced coagulopathy

Background: This study aimed to investigate whether early unfractionated heparin (UFH) administration provides a survival advantage for patients with sepsis-induced coagulopathy (SIC). Methods: Patients hospitalized with sepsis-induced coagulopathy from the Medical Information Mart for Intensive Care (MIMIC)-IV database were identified. Patients were divided into two groups, who received unfractionated heparin (UFH) subcutaneously within 24 h after intensive care unit (ICU) admission, and the control group, who received not. The primary endpoint was intensive care unit mortality, the secondary outcomes were 7, 14, and 28-day and hospital mortality. Propensity score matching (PSM) the marginal structural Cox model (MSCM) and E-value analysis were used to account for baseline differences, time-varying and unmeasured confounding factors. Results: A total of 3,377 patients with sepsis-induced coagulopathy were enrolled in the study, of which 815 in unfractionated heparin group and 2,562 in control group. There was significant effect on primary and secondary outcomes with unfractionated heparin after propensity score matching (intensive care unit mortality, hazard ratio [HR] 0.69, 95% confidence interval [CI] 0.52-0.92; 7-day, HR 0.70, 95% CI 0.49-0.99; 14-day, HR 0.68.95% CI 0.50-0.92; 28-day, HR 0.72, 95% CI 0.54-0.96; hospital mortality, HR 0.74, 95% CI 0.57-0.96), marginal structural Cox model manifested unfractionated heparin associated with decreased intensive care unit mortality in all populations (HR 0.64, 95% CI 0.49-0.84), and stratification with the marginal structural Cox model indicated analysis further indicated the survival advantage only among patients with an sepsis-induced coagulopathy score of 4 (HR 0.56, 95% CI 0.38-0.81). Further analysis showed that treatment with 6,250-13750 IU/day of unfractionated heparin associated with a decreased risk of intensive care unit mortality. Similar results were replicated in subgroup analysis with propensity score matching only for patients with an sepsis-induced coagulopathy score of 4 (intensive care unit mortality, HR 0.51, 95% CI 0.34-0.76). Conclusion: This study found early unfractionated heparin therapy to patients with sepsis-induced coagulopathy appears to be associated with improved outcomes. Subgroup analysis further demonstrates heparin therapy decreased intensive care unit mortality primarily in patients only with SIC score of 4.

Potential causal links between long-term ambient particulate matter exposure and cardiovascular mortality: New evidence from a large community-based cohort in South China

BACKGROUND

Cardiovascular disease (CVD) mortality is associated with long-term particulate matter (PM) exposure. However, evidence from large, highly-exposed population cohort and observational-data-based causal inference approaches remains limited.

AIMS

We examined the potential causal links between PM exposure and the CVD mortality in South China.

METHODS

580,757 participants were recruited during 2009-2015 and followed up through 2020. Satellite-based annual concentrations of PM_2.5, PM₁₀, and PM_coarse (i.e., PM₁₀ - PM_2.5) at 1 km² spatial resolution were estimated and assigned to each participant. Marginal structural Cox models with time-varying covariates, adjusted using inverse probability weighting, were developed to evaluate the association between prolonged PM exposure and CVD mortality.

RESULTS

For overall CVD mortality, the hazard ratios and 95% confidence interval for each 1 μg/m³ increase in the annual average concentration of PM_2.5, PM₁₀, and PM_coarse were 1.033 (1.028-1.037), 1.028 (1.024-1.032), and 1.022 (1.012-1.033), respectively. All three PMs were linked to a higher mortality risk for myocardial infarction and ischemic heart disease (IHD). The mortality risk of chronic IHD and hypertension was linked to PM_2.5 and PM₁₀. Significant association between PM_coarse and other heart disease mortality was also observed. The older, women, less-educated participants, or inactive participants exhibited particularly higher susceptibility. Participants who were generally exposed to PM₁₀ concentrations below 70 μg/m³ were more vulnerable to PM_2.5-, PM₁₀- and PM_coarse-CVD mortality risks.

CONCLUSION

This large cohort study provides evidence for the potential causal links between increased CVD mortality and ambient PM exposure, as well as socio-demographics linked to the highest vulnerability.

Variability of the response to immunotherapy among subgroups of patients with multiple sclerosis

BACKGROUND AND PURPOSE

This study assessed the effect of patient characteristics on the response to disease-modifying therapy (DMT) in multiple sclerosis (MS).

METHODS

We extracted data from 61,810 patients from 135 centers across 35 countries from the MSBase registry. The selection criteria were: clinically isolated syndrome or definite MS, follow-up ≥ 1 year, and Expanded Disability Status Scale (EDSS) score ≥ 3, with ≥1 score recorded per year. Marginal structural models with interaction terms were used to compare the hazards of 12-month confirmed worsening and improvement of disability, and the incidence of relapses between treated and untreated patients stratified by their characteristics.

RESULTS

Among 24,344 patients with relapsing MS, those on DMTs experienced 48% reduction in relapse incidence (hazard ratio [HR] = 0.52, 95% confidence interval [CI] = 0.45-0.60), 46% lower risk of disability worsening (HR = 0.54, 95% CI = 0.41-0.71), and 32% greater chance of disability improvement (HR = 1.32, 95% CI = 1.09-1.59). The effect of DMTs on EDSS worsening and improvement and the risk of relapses was attenuated with more severe disability. The magnitude of the effect of DMT on suppressing relapses declined with higher prior relapse rate and prior cerebral magnetic resonance imaging activity. We did not find any evidence for the effect of age on the effectiveness of DMT. After inclusion of 1985 participants with progressive MS, the effect of DMT on disability mostly depended on MS phenotype, whereas its effect on relapses was driven mainly by prior relapse activity.

CONCLUSIONS

DMT is generally most effective among patients with lower disability and in relapsing MS phenotypes. There is no evidence of attenuation of the effect of DMT with age.

Instrumental variable estimation of the marginal structural Cox model for time-varying treatments

Robins (1998) introduced marginal structural models, a general class of counterfactual models for the joint effects of time-varying treatments in complex longitudinal studies subject to time-varying confounding. Robins (1998) established the identification of marginal structural model parameters under a sequential randomization assumption, which rules out unmeasured confounding of treatment assignment over time. The marginal structural Cox model is one of the most popular marginal structural models for evaluating the causal effect of time-varying treatments on a censored failure time outcome. In this paper, we establish sufficient conditions for identification of marginal structural Cox model parameters with the aid of a time-varying instrumental variable, in the case where sequential randomization fails to hold due to unmeasured confounding. Our instrumental variable identification condition rules out any interaction between an unmeasured confounder and the instrumental variable in its additive effects on the treatment process, the longitudinal generalization of the identifying condition of Wang & Tchetgen Tchetgen (2018). We describe a large class of weighted estimating equations that give rise to consistent and asymptotically normal estimators of the marginal structural Cox model, thereby extending the standard inverse probability of treatment weighted estimation of marginal structural models to the instrumental variable setting. Our approach is illustrated via extensive simulation studies and an application to estimating the effect of community antiretroviral therapy coverage on HIV incidence.

Clinical effectiveness of sodium bicarbonate therapy on mortality for septic patients with acute moderate lactic acidosis

Objective: This study aimed to explore the effectiveness of sodium bicarbonate (SB) administration on mortality in septic patients with acute moderate lactic acidosis (MLA). Methods: The large ICU database (MIMIC-IV) was retrospectively analyzed. Patients with sepsis and acute MLA were identified. Propensity score analysis (PSA) was performed to explain baseline differences in the probability of accepting or not accepting SB. The edge structured cox model (MSCM) was used to adjust for baseline and time-varying confounding variables. The primary outcome was the ICU and hospital mortality. The subgroup of septic shock was also investigated. Results: A total of 512 septic patients with acute MLA were identified in this study, including 160 in the SB group and 352 in the non-SB group. In the PSA, SB administration was associated with reduced ICU (HR .58, 95% CI 0.38-.89; p <.05) and hospital (HR .67, 95% CI 0.47-.95; p <.05) mortality in septic patients with acute MLA. In the subgroup, the results were similar with septic patients. In the MSCM, SB administration could also improve the ICU (HR .35, 95% CI 0.16-.75; p <.01) and (HR .50, 95% CI 0.28-.88; p <.05) mortality on septic patients. However, In the subgroup, SB administration could only be found associated with improved hospital (HR .44, 95% CI 0.20-.97; p <.05) survival in septic shock. Conclusion: SB administration treatment could reduce ICU and hospital mortality of septic patients with acute MLA. Meanwhile, it could also improve hospital survival in the subgroup of septic shock patients with acute MLA.

Using marginal structural joint models to estimate the effect of a time-varying treatment on recurrent events and survival: An application on arrhythmogenic cardiomyopathy

In many clinical applications to evaluate the effect of a treatment, randomized control trials are difficult to carry out. On the other hand, clinical observational registries are often available and they contain longitudinal data regarding clinical parameters, drug therapies, and outcomes. In the past, much research has addressed causal methods to estimate treatment effects from observational studies. In the context of time-varying treatments, marginal structural models are often used. However, most analyses have focused on binary outcomes or time-to-the-first event analyses. The novelty of our approach is to combine the marginal structural methodology with the case where correlated recurrent events and survival are the outcomes of interest. Our work focuses on solving the nontrivial problem of defining the measures of effect, specifying the model for the time-dependent weights and the model to estimate the outcome, implementing them, and finally estimating the final treatment effects in this life-history setting. Our approach provides a strategy that allows obtaining treatment effect estimates both on the recurrent events and the survival with a clear causal and clinical interpretation. At the same time, the strategy we propose is based on flexible modeling choices such as the use of joint models to capture the correlation within events from the same subject and the specification of time-dependent treatment effects. The clinical problem which motivated our work is the evaluation of the treatment effect of beta-blockers in arrhythmogenic right ventricular cardiomyopathy (ARVC/D), and the dataset comes from the Trieste Heart Muscle Disease Registry.

Effectiveness of sodium bicarbonate infusion on mortality for elderly septic patients with acute metabolic acidosis

Objective: This study aimed to explore the effectiveness of sodium bicarbonate (SB) infusion on mortality in elderly septic patients with acute metabolic acidosis (MA) and in other subgroups.

Methods: Retrospective analysis of a large ICU database (MIMIC-IV) was performed. Elderly septic patients with acute MA were identified from MIMIC-IV. Propensity score analysis (PSA) was performed to explain for the baseline differences in the probability to receive SB or not. The marginal structural Cox model (MSCM) was developed to adjust for both baseline and time-varying confounding variables. The primary outcome was the ICU and hospital mortality.

Results: A total of 869 elderly septic patients with acute MA were identified in this study, including 361 in the SB group and 508 in the non-SB group. In the PSA, SB infusion was not associated with reduced ICU (HR 0.82, 95% CI 0.62–1.10; p = 0.19) or hospital (HR 0.94, 95% CI 0.74–1.19; p = 0.60) mortality in overall elderly septic patients with acute MA. In the subgroup of severe metabolic acidosis, SB infusion could not improve the ICU (HR 0.82, 95% CI 0.62–1.10; p =0.19) and hospital (HR 0.94, 95% CI 0.74–1.19; p =0.60) mortality on elderly septic patients. However, In the subgroup of moderate metabolic acidosis, SB infusion could be found associated with improved ICU (HR 0.64, 95% CI 0.43–0.95; p &lt;0.05) and hospital (HR 0.70, 95% CI 0.50–0.99; p &lt;0.05) survival in elderly septic patients. In the MSCM, the results were similar with PSA.

Conclusion: SB infusion could improve both ICU and hospital survival for elderly septic patients with acute metabolic acidosis.

Superior effects of natalizumab versus other DMTs on patient-reported outcomes in people with multiple sclerosis

BACKGROUND

Little is known about the comparative effectiveness of multiple sclerosis (MS) disease-modifying therapies (DMTs) on patient-reported outcomes in MS. We compared the effects of natalizumab to other DMTs in relation to MS symptom severity, quality of life, disability, disease progression and employment outcomes using real-world data.

METHODS

We included 2817 observations in 2015, 2016 and 2017 from 1382 participants in the Australian MS Longitudinal Study. Information on treatment, health and employment outcomes was prospectively collected by questionnaires. Marginal structural models with interaction terms for DMT×time were used to compare natalizumab and other comparator treatment groups.

RESULTS

Natalizumab was associated with improvements over time, or general trends of improvement, in the severity of many symptoms and work productivity loss. Compared with any other DMTs, natalizumab was associated with superior effects over time for 8 of 23 patient-reported outcomes, with similar directions of effect observed for another 6, demonstrating consistency. There were no differences in effect for spasticity, fatigue, pain, feelings of depression, disability, European quality of life five dimension index, presenteeism and work status. Natalizumab did not perform significantly worse over time compared with any other DMTs for any of the outcomes.

CONCLUSIONS

Natalizumab was associated with superior outcomes over time for many patient-reported health and employment outcomes when compared with other DMTs in this large prospective cohort study. These findings may influence treatment selection in clinical practice and future treatment cost-effectiveness analyses.

Impact of methodological choices in comparative effectiveness studies: application in natalizumab versus fingolimod comparison among patients with multiple sclerosis

Background

Natalizumab and fingolimod are used as high-efficacy treatments in relapsing–remitting multiple sclerosis. Several observational studies comparing these two drugs have shown variable results, using different methods to control treatment indication bias and manage censoring. The objective of this empirical study was to elucidate the impact of methods of causal inference on the results of comparative effectiveness studies.

Methods

Data from three observational multiple sclerosis registries (MSBase, the Danish MS Registry and French OFSEP registry) were combined. Four clinical outcomes were studied. Propensity scores were used to match or weigh the compared groups, allowing for estimating average treatment effect for treated or average treatment effect for the entire population. Analyses were conducted both in intention-to-treat and per-protocol frameworks. The impact of the positivity assumption was also assessed.

Results

Overall, 5,148 relapsing–remitting multiple sclerosis patients were included. In this well-powered sample, the 95% confidence intervals of the estimates overlapped widely. Propensity scores weighting and propensity scores matching procedures led to consistent results. Some differences were observed between average treatment effect for the entire population and average treatment effect for treated estimates. Intention-to-treat analyses were more conservative than per-protocol analyses. The most pronounced irregularities in outcomes and propensity scores were introduced by violation of the positivity assumption.

Conclusions

This applied study elucidates the influence of methodological decisions on the results of comparative effectiveness studies of treatments for multiple sclerosis. According to our results, there are no material differences between conclusions obtained with propensity scores matching or propensity scores weighting given that a study is sufficiently powered, models are correctly specified and positivity assumption is fulfilled.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-022-01623-8.

Adherence to a priori-Defined Diet Quality Indices Throughout the Early Disease Course Is Associated With Survival in Head and Neck Cancer Survivors: An Application Involving Marginal Structural Models

No studies, to date, have scrutinized the role of a priori dietary patterns on prognosis following a head and neck squamous cell carcinoma (HNSCC) diagnosis. The purpose of this analysis was to evaluate the associations between adherence to six a priori defined diet quality indices (including AHEI-2010, aMED, DASH, and three low-carbohydrate indices) throughout the first 3 years of observation and all-cause and cancer-specific mortalities in 468 newly diagnosed HNSCC patients from the University of Michigan Head and Neck Specialized Program of Research Excellence (UM-SPORE). The dietary intake data were measured using a food frequency questionnaire administered at three annual time points commencing at study entry. Deaths and their causes were documented throughout the study using various data sources. Marginal structural Cox proportional hazards models were used to evaluate the role of diet quality, as a time-varying covariate, on mortality. There were 93 deaths from all causes and 74 cancer-related deaths adjudicated throughout the observation period. There was a strong inverse association between adherence to the AHEI-2010, all-cause mortality (HR_Q5–Q1:0.07, 95% CI:0.01–0.43, p_trend:0.04), and cancer-specific mortality (HR_Q5–Q1:0.15, 95% CI:0.02–1.07, p_trend:0.04). Other more modest associations were noted for the low-carbohydrate indices. In sum, higher adherence to the AHEI-2010 and a plant-based low-carbohydrate index throughout the first 3 years since diagnosis may bolster survival and prognosis in newly diagnosed patients with HNSCC.

Estimating the joint effect of diabetes and subsequent depressive symptoms on mortality among older latinos

PURPOSE

Diabetes and depression are risk factors for cardiovascular disease, but the evidence about their interaction effect on long-term health outcomes among Latinos is lacking. We aimed to investigate the joint association of diabetes and subsequent depressive symptoms with mortality among older Latinos, an understudied racial/ethnic group with high prevalence of diabetes.

METHODS

This study included 1,495 adults from the Sacramento Area Latino Study on Aging. We employed Cox proportional-hazards models to estimate the adjusted hazard ratios [aHRs] for cardiovascular and all-cause mortality according to diabetes status at enrollment and depressive symptoms a year after the enrollment. We used marginal structural models to adjust for time-varying confounders.

RESULTS

The mean age (standard deviation) of participants was 70 (6.6) years. Over follow-up (median 7.7 years), diabetes and depressive symptoms were individually associated with increased risk of cardiovascular mortality (diabetes, aHR[95% CI]=2.13[1.60-2.84]; depressive symptoms, aHR[95% CI]=1.62[1.09-2.39]) and all-cause mortality (diabetes, aHR[95% CI]=1.92[1.53-2.41]; depressive symptoms, aHR[95% CI]=1.41[1.02-1.94]). After adjusting for time-varying confounders, we found a multiplicative interaction between diabetes and subsequent depressive symptoms for cardiovascular mortality (aHR[95% CI]=2.94[1.07-8.39]), but not all-cause mortality (aHR[95% CI]=1.80[0.81-4.35]).

CONCLUSIONS

Using a longitudinal cohort of community-dwelling older Latinos, we found that diabetes and subsequent depressive symptoms were jointly associated with increased risk of cardiovascular mortality.

Stratified Restricted Mean Survival Time Model for Marginal Causal Effect in Observational Survival Data

Time to event outcomes is commonly encountered in epidemiologic research. Multiple papers have discussed the inadequacy of using the hazard ratio as a causal effect measure due to its noncollapsibility and the time-varying nature. In this paper, we further clarified that the hazard ratio might be used as a conditional causal effect measure, but it is generally not a valid marginal effect measure, even under randomized design. We proposed to use the restricted mean survival time (RMST) difference as a causal effect measure, since it essentially measures the mean difference over a specified time horizon and has a simple interpretation as the area under survival curves. For observational studies, propensity score adjustment can be implemented with RMST estimation to remove observed confounding bias. We proposed a propensity score stratified RMST estimation strategy, which performs well in our simulation evaluation and is relatively easy to implement for epidemiologists in practice. Our stratified RMST estimation includes two different versions of implementation, depending on whether researchers want to involve regression modeling adjustment, which provides a powerful tool to examine the marginal causal effect with observational survival data.

Immune checkpoint inhibitors at any treatment line in advanced NSCLC: Real-world overall survival in a large Italian cohort

OBJECTIVES

To estimate the average treatment effect of immune checkpoint inhibitors in any line of treatment in a 2016-2018 population-based cohort of patients with advanced non-small-cell lung cancer (NSCLC).

MATERIALS AND METHODS

The cohort, and information on the tumor, were derived from the cancer registry of the Agency for Health Protection of Milan, Italy. Inclusion criteria were adult age, microscopically confirmed NSCLC, stage IIIB or IV at diagnosis, and having received at least one line of treatment. Treatment with all licensed anti PD-1/PD-L1 inhibitors was derived from inpatients and outpatients' pharmaceutical databases of the ATS and vital status at 31 December 2019 from the health registry office of the Lombardy region. We investigated, with a causal approach, the relationship between survival and anti PD-1/PD-L1 treatment at any line constructing a directed acyclic graph and fitting a Marginal Structural Cox Model (MSCM).

RESULTS

Of 1673 subjects, 324 received anti PD-1/PD-L1 at any treatment line. Overall, one-year survival was 61.1% (95 %CI, 55.6-66.2%) in the group treated with anti PD-1/PD-L1 at any line and 31.1% (95 %CI, 28.6-33.5%) among not treated. One-year hazard ratio (HR) of death for not treated vs. treated was 2.15 (95 %CI, 1.91-2.41), decreasing to 1.23 (95 %CI, 1.03-1.46) at two years and reaching one in the third year.

CONCLUSION

In un unselected population-based cohort with advanced lung cancer, treatment with anti PD-1/PD-L1 at any line lowered the hazard of death up to two-years from date of diagnosis, confirming the efficacy of immunotherapy outside clinical trials.

A non-parametric propensity score for estimating the effect of interferon-beta or glatiramer acetate on long-term outcomes of multiple sclerosis

BACKGROUND

The few observational studies that investigated the long-term effects of interferon-beta and glatiramer acetate were usually focused on progression to irreversible disability and other outcomes such as number of relapses and transition to secondary-progressive multiple sclerosis (SPMS) have been rarely studied. The objective of this paper is to estimate the effect of interferon-beta/glatiramer acetate on progression to irreversible disability, transition from relapsing-remitting multiple sclerosis (RRMS) to SPMS and the rate of relapses over 10 years.

METHODS

Analyses included 2498 patients with confirmed diagnosis of RRMS followed in Montréal from 1977 to 2016. Marginal structural models with propensity score for treatment and censoring were used to account for potential confounding and attrition. Specifically, we used pooled logistic regression for progression to irreversible disability and transition to SPMS, and Poisson models for the rate of relapses.

RESULTS

77% of subjects were female and the median age at RRMS diagnosis was 35 years. The hazard of progression to irreversible disability was lower among treated patients than untreated patients (HR=0.73, 95% CI [0.57-0.94]). We did not find evidence of an association between interferon-beta/glatiramer acetate and the rate of transition to SPMS either over the 3-month intervals or for the duration of treatment. Patients treated for >5 years had a lower rate of relapses compared to those untreated (HR=0.70, 95% CI [0.57-0.86]).

CONCLUSION

Treatment with interferon-beta/glatiramer acetate suggests a beneficial effect on progression to irreversible disability and rate of relapses, but not on transition to SPMS.

Dealing With Treatment-Confounder Feedback and Sparse Follow-up in Longitudinal Studies: Application of a Marginal Structural Model in a Multiple Sclerosis Cohort

The beta-interferons are widely prescribed platform therapies for patients with multiple sclerosis (MS). We accessed a cohort of patients with relapsing-onset MS from British Columbia, Canada (1995-2013), to examine the potential survival advantage associated with beta-interferon exposure using a marginal structural model. Accounting for potential treatment-confounder feedback between comorbidity, MS disease progression, and beta-interferon exposure, we found an association between beta-interferon exposure of at least 6 contiguous months and improved survival (hazard ratio (HR) = 0.63, 95% confidence interval 0.47, 0.86). We also assessed potential effect modifications by sex, baseline age, or baseline disease duration, and found these factors to be important effect modifiers. Sparse follow-up due to variability in patient contact with the health system is one of the biggest challenges in longitudinal analyses. We considered several single-level and multilevel multiple imputation approaches to deal with sparse follow-up and disease progression information; both types of approach produced similar estimates. Compared to ad hoc imputation approaches, such as linear interpolation (HR = 0.63), and last observation carried forward (HR = 0.65), all multiple imputation approaches produced a smaller hazard ratio (HR = 0.53), although the direction of effect and conclusions drawn concerning the survival advantage remained the same.

Comparing New-User Cohort Designs: The Example of Proton Pump Inhibitor Effectiveness in Idiopathic Pulmonary Fibrosis

The prevalent new-user cohort design is useful for assessing the effectiveness of a medication in the absence of an active comparator. Alternative approaches, particularly in the presence of informative censoring, include a variant of this design based on never users of the study drug and the marginal structural Cox model approach. We compared these approaches in assessing the effectiveness of proton pump inhibitors (PPIs) in reducing mortality among patients with idiopathic pulmonary fibrosis (IPF) using a cohort of IPF patients identified in the United Kingdom's Clinical Practice Research Datalink and diagnosed between 2003 and 2016. The cohort included 2,944 IPF patients, 1,916 of whom initiated use of PPIs during follow-up. There were 2,136 deaths (mortality rate = 25.8 per 100 person-years). Using the conventional prevalent new-user design, we found a hazard ratio for death associated with PPI use compared with nonuse of 1.07 (95% confidence interval (CI): 0.94, 1.22). The variant of the prevalent new-user design comparing PPI users with never users found a hazard ratio of 0.82 (95% CI: 0.73, 0.91), while the marginal structural Cox model found a hazard ratio of 1.08 (95% CI: 0.85, 1.38). The marginal structural model and the conventional prevalent new-user design, both accounting for informative censoring, produced similar results. However, the prevalent new-user design variant based on never users introduced selection bias and should be avoided.

Effect of Disease-Modifying Therapy on Disability in Relapsing-Remitting Multiple Sclerosis Over 15 Years

OBJECTIVE

To test the hypothesis that immunotherapy prevents long-term disability in relapsing-remitting multiple sclerosis (MS), we modeled disability outcomes in 14,717 patients.

METHODS

We studied patients from MSBase followed for ≥1 year, with ≥3 visits, ≥1 visit per year, and exposed to MS therapy, and a subset of patients with ≥15-year follow-up. Marginal structural models were used to compare the cumulative hazards of 12-month confirmed increase and decrease in disability, Expanded Disability Status Scale (EDSS) step 6, and the incidence of relapses between treated and untreated periods. Marginal structural models were continuously readjusted for patient age, sex, pregnancy, date, disease course, time from first symptom, prior relapse history, disability, and MRI activity.

RESULTS

A total of 14,717 patients were studied. During the treated periods, patients were less likely to experience relapses (hazard ratio 0.60, 95% confidence interval [CI] 0.43-0.82, p = 0.0016), worsening of disability (0.56, 0.38-0.82, p = 0.0026), and progress to EDSS step 6 (0.33, 0.19-0.59, p = 0.00019). Among 1,085 patients with ≥15-year follow-up, the treated patients were less likely to experience relapses (0.59, 0.50-0.70, p = 10^-9) and worsening of disability (0.81, 0.67-0.99, p = 0.043).

CONCLUSION

Continued treatment with MS immunotherapies reduces disability accrual by 19%-44% (95% CI 1%-62%), the risk of need of a walking aid by 67% (95% CI 41%-81%), and the frequency of relapses by 40-41% (95% CI 18%-57%) over 15 years. This study provides evidence that disease-modifying therapies are effective in improving disability outcomes in relapsing-remitting MS over the long term.

CLASSIFICATION OF EVIDENCE

This study provides Class IV evidence that, for patients with relapsing-remitting MS, long-term exposure to immunotherapy prevents neurologic disability.

The use and quality of reporting of propensity score methods in multiple sclerosis literature: A review

Background:

Propensity score (PS) analyses are increasingly used in multiple sclerosis (MS) research, largely owing to the greater availability of large observational cohorts and registry databases.

Objective:

To evaluate the use and quality of reporting of PS methods in the recent MS literature.

Methods:

We searched the PubMed database for articles published between January 2013 and July 2019. We restricted the search to comparative effectiveness studies of two disease-modifying therapies.

Results:

Thirty-nine studies were included in the review, with most studies (62%) published within the past 3 years. All studies reported the list of covariates used for the PS model, but only 21% of studies mentioned how those covariates were selected. Most studies used PS matching (72%), followed by PS adjustment (18%), weighting (15%), and stratification (3%), with some overlap. Most studies using matching or weighting reported checking post-PS covariate imbalance (91%), although about 45% of these studies relied on p values from various statistical tests. Only 25% of studies using matching reported calculating robust standard errors for the PS analyses.

Conclusions:

The quality of reporting of PS methods in the MS literature is sub-optimal in general, and in some cases, inappropriate methods are used.

Transfusion-related Epstein-Barr virus (EBV) infection: A multicenter prospective cohort study among pediatric recipients of hematopoietic stem cell transplants (TREASuRE study)

BACKGROUND

Epstein-Barr virus (EBV) is carried in the blood of most adults, and transfusion-related infections have been reported. EBV is particularly deleterious in immunosuppressed transplant patients. The aim was to determine if EBV transmission occurred through leukodepleted blood product transfusion in pediatric recipients of hematopoietic stem cell transplants (HSCT).

STUDY DESIGN AND METHODS

This prospective Canadian multi-center cohort study includes 156 allogeneic HSCT pediatric recipients. The association between EBV and transfusion was analyzed using Cox regressions. EBV infection, defined by a PCR+ test in the blood of seronegative recipients of an EBV-negative graft, was monitored in order to correlate the recipient EBV strain with that of the blood donors. EBV genotypes were determined by PCR amplification followed by DNA sequencing at two loci (EBNA3b and LMP1).

RESULTS

No statistically significant associations were found between transfusions and EBV. One case of post-transplant EBV infection was identified among the 21 EBV-seronegative recipients receiving an EBV-negative graft. A total of 22 blood donors were retraced to determine whether the recipient's EBV strain matched that of a donor. One donor strain showed 100% sequence homology at the EBNA3b locus, but differed by one or two point mutations and by a 132-bp deletion at the LMP1 locus. The blood donor in question was alone among the 22 donors to show amplifiable virus in plasma. Blood from this donor readily produced an immortalized lymphoblastoid cell line in culture.

CONCLUSION

While considered a rare event, EBV transmission through transfusion may occur in the context of severe immunosuppression.

Comparison of the efficacy and safety of biologic agents between elderly-onset and young-onset RA patients: the ANSWER cohort study

The objective of the study was to compare the efficacy and safety of biological disease-modifying antirheumatic drugs (bDMARDs) between elderly-onset rheumatoid arthritis (EORA) and young-onset rheumatoid arthritis (YORA) patients. Patients with rheumatoid arthritis (RA) aged ≧18 years enrolled in a Japanese multicenter observational registry between 2009 and 2018 who had moderate or high disease activity when initiating bDMARDs were included. EORA was defined as RA with onset at 60 or over. After propensity score weighting for differences in confounding factors, generalized estimating equations were used to assess the relationship between the age of RA onset and bDMARD clinical effectiveness at 48 weeks after starting a bDMARD. Among a total of 7183 patients in the registry, 2815 (39.2%) were identified as EORA. The proportion of patients on bDMARDs was lower in the EORA as compared to the YORA (18.3% vs 28.0%, p < 0.001). Of the 989 bDMARD initiators, 364 (36.8%) were identified as EORA. The median follow-up duration was 48 weeks both in the EORA and in the YORA. After adjusting for differences in baseline characteristics between the two age groups, there was no significant difference in Clinical Disease Activity Index scores at 48 weeks (mean difference 1.01, 95% CI = - 0.62 to 2.64, p = 0.22). There was a non-significant trend toward lower remission in EORA (OR = 0.52, 95% CI = 0.24-1.14, p = 0.10), and low disease activity/remission was similar (OR = 0.86, 95% CI = 0.29-2.52, p = 0.77). Drug retention (HR = 0.95, 95% CI = 0.55-1.35, p = 0.78) and discontinuations due to adverse events (HR = 0.78, 95% CI = 0.38-1.18, p = 0.22) were similar between the two age groups after adjustment for confounders. In RA patients initiating bDMARDs, improvements in clinical disease at 48 weeks were similar between EORA and YORA. Drug retention and adverse events discontinuation were similar between the two age groups.

Estimating long-term pollution exposure effects through inverse probability weighting methods with Cox proportional hazards models

BACKGROUND

Fine particulate matter (PM2.5) is associated with negative health outcomes in both the short and long term. However, the cohort studies that have produced many of the estimates of long-term exposure associations may fail to account for selection bias in pollution exposure as well as covariate imbalance in the study population; therefore, causal modeling techniques may be beneficial.

METHODS

Twenty-nine years of data from the National Health Interview Survey (NHIS) was compiled and linked to modeled annual average outdoor PM2.5 concentration and restricted-use mortality data. A series of Cox proportional hazards models, adjusted using inverse probability weights, yielded causal risk estimates of long-term exposure to ambient PM2.5 on all-cause and cardiopulmonary mortality.

RESULTS

Covariate-adjusted estimated relative risks per 10 μg/m3 increase in PM2.5 exposure were estimated to be 1.117 (1.083, 1.152) for all-cause mortality and 1.232 (1.174, 1.292) for cardiopulmonary mortality. Inverse probability weighted Cox models provide relatively consistent and robust estimates similar to those in the unweighted baseline multivariate Cox model, though they have marginally lower point estimates and higher standard errors.

CONCLUSIONS

These results provide evidence that long-term exposure to PM2.5 contributes to increased mortality risk in US adults and that the estimated effects are generally robust to modeling choices. The size and robustness of estimated associations highlight the importance of clean air as a matter of public health. Estimated confounding due to measured covariates appears minimal in the NHIS cohort, and various distributional assumptions have little bearing on the magnitude or standard errors of estimated causal associations.

All-cause mortality following low-dose aspirin treatment for patients with high cardiovascular risk in remote Australian Aboriginal communities: an observational study

OBJECTIVES

To evaluate the benefit and risk of low-dose acetylsalicylic acid (aspirin) in patients from remote Aboriginal communities in the Northern Territory, Australia.

DESIGN

Retrospective cohort study using primary care and hospital data routinely used for healthcare. Aspirin users and non-users were compared before and after controlling confounders by matching. Marginal structural models (MSM) were applied to ascertain the benefit and risk.

SETTING

The benefit and harm of aspirin were investigated in patients aged ≥18 years from 54 remote Aboriginal communities.

PARTICIPANTS

None had a previous cardiovascular event or major bleeds. Patients on anticoagulants or other antiplatelets were excluded.

INTERVENTION

Aspirin at a dose of 75-162 mg/day.

OUTCOME MEASURES

Endpoints were all-cause, cardiovascular mortality and incidences of cardiovascular events and major bleeds.

RESULTS

8167 predominantly Aboriginal adults were included and followed between July 2009 and June 2017 (aspirin users n=1865, non-users n=6302, mean follow-up 4 years with hospitalisations 6.4 per person). Univariate analysis found material differences in demographics, prevalence of chronic diseases and outcome measures between aspirin users and non-users before matching. After matching, aspirin was significantly associated with reduced all-cause mortality (HR=0.45: 95% CI 0.34 to 0.60; p<0.001), but not bleeding (HR=1.13: 95% CI 0.39 to 3.26; p=0.820). After using MSMs to eliminate the effects of confounders, loss of follow-up and time dependency of treatment, aspirin was associated with reduced all-cause mortality (HR=0.60: 95% CI 0.47 to 0.76; p<0.001), independent of age (HR=1.06; p<0.001), presence of diabetes (HR=1.42; p<0.001), hypertension (HR=1.61; p<0.001) and alcohol abuse (HR=1.81; p<0.001). No association between aspirin and major bleeding was found (HR=1.14: 95% CI 0.48 to 2.73; p=0.765). Sensitivity analysis suggested these findings were unlikely to have been the result of unmeasured confounding.

CONCLUSION

Aspirin was associated with reduced all-cause mortality. Bleeding risk was less compared with survival benefits. Aspirin should be considered for primary prevention in Aboriginal people with high cardiovascular risk.

Leveraging real-world data to investigate multiple sclerosis disease behavior, prognosis, and treatment

Randomized controlled clinical trials and real-world observational studies provide complementary information but with different validity. Some clinical questions (disease behavior, prognosis, validation of outcome measures, comparative effectiveness, and long-term safety of therapies) are often better addressed using real-world data reflecting larger, more representative populations. Integration of disease history, clinician-reported outcomes, performance tests, and patient-reported outcome measures during patient encounters; imaging and biospecimen analyses; and data from wearable devices increase dataset utility. However, observational studies utilizing these data are susceptible to many potential sources of bias, creating barriers to acceptance by regulatory agencies and the medical community. Therefore, data standardization and validation within datasets, harmonization across datasets, and application of appropriate analysis methods are important considerations. We review approaches to improve the scope, quality, and analyses of real-world data to advance understanding of multiple sclerosis and its treatment, as an example of opportunities to better support patient care and research.

Hazard of complex regional pain syndrome following human papillomavirus vaccination among adolescent girls in the United States: a case-cohort analysis of insurance claims data

Objectives: Complex regional pain syndrome (CRPS) cases have followed human papillomavirus (HPV) vaccination, but no causal link has been established.Methods: Using insurance claims, the authors observed unvaccinated 11-year-old girls for CRPS diagnoses. The authors used time-dependent Cox regression to identify health-related CRPS predictors using diagnosis codes. Next, the authors identified HPV vaccinations using procedural codes. HPV vaccination and CRPS predictors were considered time-dependent covariates to estimated adjusted hazard ratios (HR) and 95% confidence intervals (CI) for CRPS, 30, 90, and 180-days post-vaccination.Results: 1,232,572 girls received 563 unique CRPS diagnoses. In a 10% sub-cohort of 123,981 girls accounting for potential confounders and predisposing risk factors (i.e. injury, infection, mental illness, primary care use), CRPS hazard was not significantly elevated 30 days (HR: 0.90, 95% CI: 0.46, 1.73), 90 days (HR: 1.17, 95% CI: 0.83, 1.65), or 180-days post-vaccination (HR: 1.11, 95% CI: 0.83, 1.47).Conclusion: The results support the safety and continued administration of HPV vaccines to adolescents.

Association Between the Use of Sodium Bicarbonate and Mortality in Acute Kidney Injury Using Marginal Structural Cox Model

OBJECTIVE

Acute kidney injury with metabolic acidosis is common in critically ill patients. This study assessed the associations between the use of IV sodium bicarbonate and mortality of patients with acute kidney injury and acidosis.

DESIGN

The study was conducted by using data from Beth Israel Deaconess Medical Center, which included several ICUs such as coronary care unit, cardiac surgery recovery unit, medical ICU, surgical ICU, and trauma-neuro ICU. Marginal structural Cox model was used to assess the relationship between receipt of sodium bicarbonate and hospital mortality, allowing pH, PaCO2, creatinine, and bicarbonate concentration as time-varying predictors of sodium bicarbonate exposure while adjusting for baseline characteristics of age, gender, Sequential Organ Failure Assessment score, acute kidney injury stage, Elixhauser score, quick Sequential Organ Failure Assessment, and Simplified Acute Physiology Score II.

SETTING

A large U.S.-based critical care database named Medical Information Mart for Intensive Care.

PATIENTS

Patients with Kidney Disease: Improving Global Outcomes acute kidney injury stage greater than or equal to 1 (> 1.5 (Equation is included in full-text article.)baseline creatinine) and one measurement of acidosis (pH ≤ 7.2). Baseline creatinine was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Of the 3,406 eligible patients, 836 (24.5%) had received sodium bicarbonate treatment. Patients who received sodium bicarbonate treatment had a higher Sequential Organ Failure Assessment (9 vs 7; p < 0.001), lower pH (7.16 vs 7.18; p < 0.001), and bicarbonate concentration (16.51 ± 7.04 vs 20.57 ± 6.29 mmol/L; p < 0.001) compared with those who did not receive sodium bicarbonate. In the marginal structural Cox model by weighing observations with inverse probability of receiving sodium bicarbonate, sodium bicarbonate treatment was not associated with mortality in the overall population (hazard ratio, 1.16; 95% CI, 0.98-1.42; p = 0.132), but it appeared to be beneficial in subgroups of pancreatitis (hazard ratio, 0.53; 95% CI, 0.28-0.98; p = 0.044) and severe acidosis (pH < 7.15; hazard ratio, 0.75; 95% CI, 0.58-0.96; p = 0.024). Furthermore, sodium bicarbonate appeared to be beneficial in patients with severe bicarbonate deficit (< -50 kg·mmol/L).

CONCLUSIONS

In the analysis by adjusting for potential confounders, there is no evidence that IV sodium bicarbonate is beneficial for patients with acute kidney injury and acidosis. Although the study suggested potential beneficial effects in some highly selected subgroups, the results need to be validated in experimental trials.

Assessing the long-term effectiveness of interferon-beta and glatiramer acetate in multiple sclerosis: final 10-year results from the UK multiple sclerosis risk-sharing scheme

BACKGROUND

Because multiple sclerosis (MS) is a chronic disease causing disability over decades, it is crucial to know if the short-term effects of disease-modifying therapies reported in randomised controlled trials reduce long-term disability. This 10-year prospective observational study of disability outcomes (Expanded Disability Status Scale (EDSS) and utility) was set up, in conjunction with a risk-sharing agreement between payers and producers, to investigate this issue.

METHODS

The outcomes of the UK treated patients were compared with a modelled untreated control based on the British Columbia MS data set to assess the long-term effectiveness of these treatments. Two complementary analysis models were used: a multilevel model (MLM) and a continuous Markov model.

RESULTS

4862 patients with MS were eligible for the primary analysis (mean and median follow-up times 8.7 and 10 years). EDSS worsening was reduced by 28% (MLM), 7% (Markov) and 24% time-adjusted Markov in the total cohort, and by 31% (MLM) and 14% (Markov) for relapsing remitting patients. The utility worsening was reduced by 23%-24% in the total cohort and by 24%-31% in the RR patients depending on the model used. All sensitivity analyses showed a treatment effect. There was a 4-year (CI 2.7 to 5.3) delay to EDSS 6.0. An apparent waning of treatment effect with time was seen. Subgroup analyses suggested better treatment effects in those treated earlier and with lower EDSS scores.

CONCLUSIONS

This study supports a beneficial effect on long-term disability with first-line MS disease-modifying treatments, which is clinically meaningful. However the waning effect noted requires further study.

Can We Train Machine Learning Methods to Outperform the High-dimensional Propensity Score Algorithm?

The use of retrospective health care claims datasets is frequently criticized for the lack of complete information on potential confounders. Utilizing patient's health status-related information from claims datasets as surrogates or proxies for mismeasured and unobserved confounders, the high-dimensional propensity score algorithm enables us to reduce bias. Using a previously published cohort study of postmyocardial infarction statin use (1998-2012), we compare the performance of the algorithm with a number of popular machine learning approaches for confounder selection in high-dimensional covariate spaces: random forest, least absolute shrinkage and selection operator, and elastic net. Our results suggest that, when the data analysis is done with epidemiologic principles in mind, machine learning methods perform as well as the high-dimensional propensity score algorithm. Using a plasmode framework that mimicked the empirical data, we also showed that a hybrid of machine learning and high-dimensional propensity score algorithms generally perform slightly better than both in terms of mean squared error, when a bias-based analysis is used.

Effectiveness of sodium bicarbonate infusion on mortality in septic patients with metabolic acidosis

OBJECTIVE

Although sodium bicarbonate (SB) solution has been widely used in clinical practice, its effect on mortality when administered to a large population of patients with acidosis is not known. The study aimed to investigate the effectiveness of SB infusion in septic patients with metabolic acidosis.

METHODS

Septic patients with metabolic acidosis were identified from the Medical Information Mart for Intensive Care (MIMIC)-III database. Propensity score (PS) was used to account for the baseline differences in the probability to receive SB or not. The marginal structural Cox model (MSCM) was employed to adjust for both baseline and time-varying confounding factors.

MAIN RESULTS

A total of 1718 septic patients with metabolic acidosis were enrolled in the study, including 500 in the SB group and 1218 in the non-SB group. Both pH [7.16 (standard deviation (SD): 0.10) vs. 7.22 (SD: 0.07); p < 0.001] and bicarbonate concentration (BC) [11.84 (SD: 3.63) vs. 14.88 (SD: 3.36) mmol/l; p < 0.001] were significantly lower in the SB than that in the non-SB group. While there was no significant mortality effect in the overall population [hazard ratio (HR): 1.04; 95% CI 0.86-1.26; p = 0.67], SB was observed to be beneficial in patients with acute kidney injury (AKI) stage 2 or 3 and pH < 7.2 (HR 0.74; 95% CI 0.51-0.86; p = 0.021). Similar results were replicated with the MSCM.

CONCLUSION

Our study observed that SB infusion was not associated with improved outcome in septic patients with metabolic acidosis, but it was associated with improved survival in septic patients with AKI stage 2 or 3 and severe acidosis. The results need to be verified in randomized controlled trials.

'Progressive MS - macro views': The need for novel clinical trial paradigms to enable drug development for progressive MS

This article outlines the principal challenges to establish a standard phase-2 approach for progressive multiple sclerosis (PMS) and presents referring strategies to accelerate the registration process via a guidance approved by regulatory agencies. Accordingly, the contribution of 'big datasets' for a better understanding of the natural history of primary-progressive multiple sclerosis (PPMS) and secondary-progressive multiple sclerosis (SPMS) and of their prognostic factors and the value of novel biomarkers are discussed. The establishment of new industry-academic initiatives, such as independent consortia under the umbrella of Progressive MS Alliance (PMSA), with the endorsement of MS organizations and Scientific Societies (e.g. European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS)) may be crucial to overcome some of the current challenges. Within this frame, the continuous interaction with regulatory agencies is instrumental for the formal validation of the many developments suitable to improve clinical trialling in PMS.

Fractional anisotropy of white matter, disability and blood iron parameters in multiple sclerosis

Multiple sclerosis (MS) is a disorder related to myelin damage, which can be investigated by neuroimaging techniques such as fractional anisotropy (FA), a measure of microstructural white matter properties. The objectives of this study were to investigate (1) the relationship between FA and disability using an extremes of outcome approach, and (2) whether blood iron parameters were associated with FA and/or disability. Patients diagnosed with MS (n = 107; 14 males and 93 females) had iron parameter tests and disability determinations using the Expanded Disability Status Scale (EDSS). FA was recorded in 48 white matter tracts in 11 of the female patients with MS and 12 female controls.

RESULTS

In patients with high disability scores the mean FA was significantly lower (0.34 ± 0.067) than in the control group (0.45 ± 0.036; p = 0.04), while patients with low disability had mean FA values (0.44 ± 0.014) similar to controls (p = 0.5). Positive associations were found between FA and the iron parameters serum iron, ferritin and percentage transferrin saturation (%Tfsat) in all the white matter tracts. For % Tfsat, the associations were highly significant in 14 tracts (p < 0.01; r-values 0.74-0.84) and p < 0.001 (r = 0.83) in the superior fronto occipital fasciculus (LH). In the whole patient group a trend was found towards an inverse association between the EDSS and the %Tfsat (r = -0.26, p = 0.05) after excluding male gender and smoking as confounders, suggesting reduced disability in the presence of higher blood iron parameters. Additionally, significant inverse associations between disease duration and haemoglobin (p = 0.04) as well as %Tfsat (p = 0.02) suggested that patients with MS may experience a decrease in blood iron concentrations over time.

Methods for time-varying exposure related problems in pharmacoepidemiology: An overview

Purpose

Lack of control for time‐varying exposures can lead to substantial bias in estimates of treatment effects. The aim of this study is to provide an overview and guidance on some of the available methodologies used to address problems related to time‐varying exposure and confounding in pharmacoepidemiology and other observational studies. The methods are explored from a conceptual rather than an analytical perspective.

Methods

The methods described in this study have been identified exploring the literature concerning to the time‐varying exposure concept and basing the search on four fundamental pharmacoepidemiological problems, construction of treatment episodes, time‐varying confounders, cumulative exposure and latency, and treatment switching.

Results

A correct treatment episodes construction is fundamental to avoid bias in treatment effect estimates. Several methods exist to address time‐varying covariates, but the complexity of the most advanced approaches—eg, marginal structural models or structural nested failure time models—and the lack of user‐friendly statistical packages have prevented broader adoption of these methods. Consequently, simpler methods are most commonly used, including, for example, methods without any adjustment strategy and models with time‐varying covariates. The magnitude of exposure needs to be considered and properly modelled.

Conclusions

Further research on the application and implementation of the most complex methods is needed. Because different methods can lead to substantial differences in the treatment effect estimates, the application of several methods and comparison of the results is recommended. Treatment episodes estimation and exposure quantification are key parts in the estimation of treatment effects or associations of interest.

Estimating inverse probability weights using super learner when weight-model specification is unknown in a marginal structural Cox model context

Correct specification of the inverse probability weighting (IPW) model is necessary for consistent inference from a marginal structural Cox model (MSCM). In practical applications, researchers are typically unaware of the true specification of the weight model. Nonetheless, IPWs are commonly estimated using parametric models, such as the main-effects logistic regression model. In practice, assumptions underlying such models may not hold and data-adaptive statistical learning methods may provide an alternative. Many candidate statistical learning approaches are available in the literature. However, the optimal approach for a given dataset is impossible to predict. Super learner (SL) has been proposed as a tool for selecting an optimal learner from a set of candidates using cross-validation. In this study, we evaluate the usefulness of a SL in estimating IPW in four different MSCM simulation scenarios, in which we varied the specification of the true weight model specification (linear and/or additive). Our simulations show that, in the presence of weight model misspecification, with a rich and diverse set of candidate algorithms, SL can generally offer a better alternative to the commonly used statistical learning approaches in terms of MSE as well as the coverage probabilities of the estimated effect in an MSCM. The findings from the simulation studies guided the application of the MSCM in a multiple sclerosis cohort from British Columbia, Canada (1995-2008), to estimate the impact of beta-interferon treatment in delaying disability progression. Copyright © 2017 John Wiley & Sons, Ltd.

Extended-hours hemodialysis is associated with lower mortality risk in patients with end-stage renal disease

Extended-hours hemodialysis offers substantially longer treatment time compared to conventional hemodialysis schedules and is associated with improved fluid and electrolyte control and favorable cardiac remodeling. However, whether extended-hours hemodialysis improves survival remains unclear. Therefore, we determined the association between extended-hours compared to conventional hemodialysis and the risk of all-cause mortality in a nationally representative cohort of patients initiating maintenance dialysis in the United States from 2007 to 2011. Survival analyses using causal inference modeling with marginal structural models were performed to compare mortality risk among 1206 individuals undergoing thrice weekly extended-hours hemodialysis or 111,707 patients receiving conventional hemodialysis treatments. The average treatment time per session for extended-hours hemodialysis was 399 minutes compared to 211 minutes for conventional therapy. The crude mortality rate with extended-hours hemodialysis was 6.4 deaths per 100 patient-years compared with 14.7 deaths per 100 patient-years for conventional hemodialysis. In the primary analysis, patients treated with extended-hours hemodialysis had a 33% lower adjusted risk of death compared to those who were treated with a conventional regimen (95% confidence interval: 7% to 51%). Additional analyses accounting for analytical assumptions regarding exposure and outcome, facility-level confounders, and prior modality history were similar. Thus, in this large nationally representative cohort, treatment with extended-hours hemodialysis was associated with a lower risk for mortality compared to treatment with conventional in-center therapy. Adequately powered randomized clinical trials comparing extended-hours to conventional hemodialysis are required to confirm these findings.

Personalised medicine for multiple sclerosis care

Treatments with a range of efficacy and risk of adverse events have become available for the management of multiple sclerosis (MS). However, now the heterogeneity of clinical expression and responses to treatment pose major challenges to improving patient care. Selecting and managing the drug best balancing benefit and risk demands a new focus on the individual patient. Personalised medicine for MS is based on improving the precision of diagnosis for each patient in order to capture prognosis and provide an evidence-based framework for predicting treatment response and personalising patient monitoring. It involves development of predictive models involving the integration of clinical and biological data with an understanding of the impact of disease on the lives of individual patients. Here, we provide a brief, selective review of challenges to personalisation of the management of MS and suggest an agenda for stakeholder engagement and research to address them.

Comparison of statistical approaches dealing with time-dependent confounding in drug effectiveness studies

In longitudinal studies, if the time-dependent covariates are affected by the past treatment, time-dependent confounding may be present. For a time-to-event response, marginal structural Cox models are frequently used to deal with such confounding. To avoid some of the problems of fitting marginal structural Cox model, the sequential Cox approach has been suggested as an alternative. Although the estimation mechanisms are different, both approaches claim to estimate the causal effect of treatment by appropriately adjusting for time-dependent confounding. We carry out simulation studies to assess the suitability of the sequential Cox approach for analyzing time-to-event data in the presence of a time-dependent covariate that may or may not be a time-dependent confounder. Results from these simulations revealed that the sequential Cox approach is not as effective as marginal structural Cox model in addressing the time-dependent confounding. The sequential Cox approach was also found to be inadequate in the presence of a time-dependent covariate. We propose a modified version of the sequential Cox approach that correctly estimates the treatment effect in both of the above scenarios. All approaches are applied to investigate the impact of beta-interferon treatment in delaying disability progression in the British Columbia Multiple Sclerosis cohort (1995-2008).