Propensity Score Methods Research Articles

Insomnia is related to long-term atrial fibrillation recurrence following radiofrequency ablation

Objective Atrial fibrillation (AF), the most common cardiac arrhythmia, presents significant health challenges, and the intricate connection between insomnia and AF has garnered substantial attention. This cohort study aims to investigate the relationship between insomnia and AF recurrences following radiofrequency ablation. Materials and Methods Data were retrieved from an electronic database of patients who underwent radiofrequency ablation for AF. The primary endpoint was AF recurrence. We utilized a multivariable Cox model, coupled with three propensity score methods, for analysis. Results Between January 1, 2017, and June 1, 2022, 541 patients who underwent radiofrequency ablation for AF were recorded in the database. After excluding 185 patients, the final cohort comprised 356 patients. Among them, 68 were afflicted by insomnia, while 288 were not. Over a median follow-up of 755 days, one patient died, and 130 (36.5%) experienced AF recurrence. Multivariate Cox regression analysis revealed that the insomnia group had a higher risk of AF recurrence compared to the non-insomnia group (HR: 1.83, 95% CI: 1.16–2.89). Further landmark analysis showed no significant difference in AF recurrence rates during the initial 1-year follow-up. However, beyond 1 year, the insomnia group demonstrated a significantly higher AF recurrence rate. As the number of insomnia symptoms increased, the risk of AF recurrence also rose significantly, indicating a dose-response relationship. Conclusion This study establishes a significant link between insomnia and long-term AF recurrence following radiofrequency ablation. It underscores the importance of identifying and addressing insomnia in patients with AF undergoing radiofrequency ablation.

Treatment effect estimation using the propensity score in clinical trials with historical control

BackgroundClinical trials assessing new treatment effects require a control group to compare the pure treatment effects. However, in clinical trials on regenerative medicine, rare diseases, and intractable diseases, it may be ethically difficult to assign participants to the control group. In recent years, the use of historical control data has attracted attention as a method for supplementing the number of participants in the control group. When combining historical control data with new randomized controlled trial (RCT) data, the assessment of heterogeneity using outcome data is not sufficient. Therefore, several statistical methods that consider participant outcomes and baseline characteristics, including the propensity score (PS) method have been proposed.MethodsWe propose a new method considering “information on whether the data are RCT data or not” in the PS model when combining the RCT and historical control data. The performance of the proposed method in estimating the treatment effect is evaluated using simulation data.ResultsWhen the distribution of covariates is similar between the RCT and historical control data, not much difference in performance is found between the proposed and conventional methods to estimate the treatment effect. On the other hand, when the distribution of covariates is not similar between the two kinds of data, the proposed method shows higher performance.ConclusionsEven when it is not known whether RCT and historical control data are similar, the proposed PS model is useful to estimate the treatment effect appropriately in RCTs using historical control data.

Effects of Non-Face-to-Face Chronic Care Management on Service Utilization and Outcomes Among US Medicare Beneficiaries with Diabetes.

Type 2 diabetes mellitus (T2DM) results in heavy economic and disease burdens in Louisiana. The Centers for Medicare and Medicaid Services has reimbursed non-face-to-face chronic care management (NFFCCM) for patients with two or more chronic conditions since 2015. To assess the impacts of NFFCCM on healthcare utilization and health outcomes. This retrospective cohort study included Medicare fee-for-service beneficiaries with T2DM and at least one additional chronic disease between 2014 and 2018. At least one record of NFFCCM Current Procedural Terminology codes. The health outcomes in the study included major adverse cardiovascular events (MACE), all-cause mortality, and heart failure. The monthly service utilization and continuity of care index for primary care were also included. The propensity score method was used to balance the baseline differences between the two groups. Weighted multivariate regression models were developed using propensity score weights to assess the impacts of NFFCCM on outcomes. During the 5years of study period, 8415 patients among the 118,643 Medicare beneficiaries received at least one NFFCCM. Patients receiving any NFFCCM had reduced healthcare utilization compared with patients not receiving NFFCCM, including 0.012 (95% CI - 0.014 to - 0.011; p < 0.001) fewer monthly hospital admissions, 0.017 (95% CI - 0.019 to - 0.016; p < 0.001) fewer monthly ED visits, and 0.399 (95% CI 0.375 to 0.423; p < 0.001) more monthly outpatient encounters. Patients receiving NFFCCM services had lower MACE event rates of 7.4% (95% CI 7.1 to 7.8%; p < 0.001), all-cause mortality rate of 7.8% (95% CI 7.4 to 8.1%; p < 0.001), and heart failure rate of 0.3% (95% CI 0.2 to 0.5%; p < 0.001), respectively. These findings suggest that reimbursement for NFFCCM was associated with the shifting high-cost utilization to lower-cost primary health care settings among patients with diabetes in Louisiana.

Income diversification patterns and their impact on bank risk

AbstractThis study focuses on the crucial topic of income diversification and its implications for bank risk, using the context of India's public and private Sector banks. The research spans the years 2005–2021, investigating the influence of varying degrees of non‐interest income on the risk profiles of Indian banks. Employing a program impact evaluation methodology, the study employs the generalised propensity score‐fractional dose–response function technique to ascertain the impact of different levels of non‐interest income on bank risk. The outcomes underscore the diverse effects of diversification contingent upon factors like bank size and ownership type, including large and small banks as well as public and private sector institutions. Notably, the aftermath of the Global Financial Crisis introduced distinctive dynamics to income diversification's impact. By segmenting the study timeline into two periods—2005–2013 and 2014–2021—it emerges that income diversification lowered bank risk in the earlier period, while conversely amplifying risk in the latter phase. These insights bear significance as an early warning mechanism for bank executives, policymakers, and regulatory authorities, enabling them to navigate strategic transitions associated with non‐interest income. Employing a novel approach, this research contributes to the literature by evaluating the interplay between income diversification and bank risk across diverse levels of non‐interest income mix. In contrast to conventional econometric methods, the use of the generalised propensity score method stands out as a robust technique to mitigate confounding biases inherent in non‐experimental inquiries.

Immune-Enhancing Treatment among Acute Necrotizing Pancreatitis Patients with Metabolic Abnormalities: A Post Hoc Analysis of a Randomized Clinical Trial.

Metabolic syndrome is common in patients with acute pancreatitis and its components have been reported to be associated with infectious complications. In this post hoc analysis, we aimed to evaluate whether metabolic abnormalities impact the effect of immune-enhancing thymosin alpha-1 (Tα1) therapy in acute necrotizing pancreatitis (ANP) patients. All data were obtained from the database for a multicenter randomized clinical trial that evaluated the efficacy of Tα1 in ANP patients. Patients who discontinued the Tα1 treatment prematurely were excluded. The primary outcome was 90-day infected pancreatic necrosis (IPN) after randomization. Three post hoc subgroups were defined based on the presence of hyperglycemia, hypertriglyceridemia, or both at the time of randomization. In each subgroup, the correlation between Tα1 and 90-day IPN was assessed using the Cox proportional-hazards regression model. Multivariable propensity-score methods were used to control potential bias. Overall, 502 participants were included in this post hoc analysis (248 received Tα1 treatment and 254 received matching placebo treatment). Among them, 271 (54.0%) had hyperglycemia, 371 (73.9%) had hypertriglyceridemia and 229 (45.6%) had both. Tα1 therapy was associated with reduced incidence of IPN among patients with hyperglycemia (18.8% vs 29.7%: hazard ratio, 0.80; 95% confidence interval, 0.37 to 0.97; p=0.03), but not in the other subgroups. Additional multivariate regression models using three propensity-score methods yielded similar results. Among ANP patients with hyperglycemia, immune-enhancing Tα1 treatment was associated with a reduced risk of IPN (ClinicalTrials.gov, Registry number: NCT02473406).

Vegetarian diets and risk of nonalcoholic fatty liver disease: An observational study of National Health and Nutrition Examination Survey2005-2018 using propensity score methods.

Studies on the association between vegetarian diets and nonalcoholic fatty liver disease (NAFLD) are limited and have inconsistent results. This study aims to explore the association between vegetarian diets and NAFLD and compare the stage of fibrosis between vegetarians and nonvegetarians in a US representative sample. Cross-sectional data from 23,130 participants aged ≥20 years were obtained from the National Health and Nutrition Examination Survey, 2005-2018. Vegetarian status was classified based on two 24-h dietary recalls. We examined the association between vegetarian diets and the risk of NAFLD using the propensity score weighting method. Vegetarian diets were significantly associated with decreases in hepatic steatosis index (HSI), US fatty liver index and nonalcoholic fatty liver disease fibrosis score with mean differences of -2.70 (95% confidence interval [CI]: -3.69, -1.70), -3.03 (95% CI: -7.15, -0.91) and -0.12 (95% CI: -0.26, -0.01), respectively. While modelling the risk of NAFLD, we estimated that vegetarians were 53% less likely to have NAFLD assessed by HSI (odds ratios [OR]: 0.47; 95% CI: 0.34, 0.65). The effect of vegetarian diets was higher among individuals with lower waist circumferences (OR: 0.20) than among those with higher waist circumferences (OR: 0.53, = 0.004). However, the association was largely attenuated after adjusting for body mass index and diabetes status. No significant association was identified between vegetarian diets and advanced fibrosis. Vegetarian diets were associated with a lower prevalence of NAFLD among US adults, and the association appeared to be stronger in people with lower waist circumferences. Further studies are warranted to replicate our findings.

Risk factors and an early predictive model for Kawasaki disease shock syndrome in Chinese children

BackgroundKawasaki disease shock syndrome (KDSS), though rare, has increased risk for cardiovascular complications. Early diagnosis is crucial to improve the prognosis of KDSS patients. Our study aimed to identify risk factors and construct a predictive model for KDSS.MethodsThis case-control study was conducted from June, 2015 to July, 2023 in two children’s hospitals in China. Children initially diagnosed with KDSS and children with Kawasaki disease (KD) without shock were matched at a ratio of 1:4 by using the propensity score method. Laboratory results obtained prior to shock syndrome and treatment with intravenous immunoglobulin were recorded to predict the onset of KDSS. Univariable logistic regression and forward stepwise logistic regression were used to select significant and independent risk factors associated with KDSS.ResultsAfter matching by age and gender, 73 KDSS and 292 KD patients without shock formed the development dataset; 40 KDSS and 160 KD patients without shock formed the validation dataset. Interleukin-10 (IL-10) > reference value, platelet counts (PLT) < 260 × 109/L, C-reactive protein (CRP) > 80 mg/ml, procalcitonin (PCT) > 1ng/ml, and albumin (Alb) < 35 g/L were independent risk factors for KDSS. The nomogram model including the above five indicators had area under the curves (AUCs) of 0.91(95% CI: 0.87–0.94) and 0.90 (95% CI: 0.71–0.86) in the development and validation datasets, with a specificity and sensitivity of 80% and 86%, 66% and 77%, respectively. Calibration curves showed good predictive accuracy of the nomogram. Decision curve analyses revealed the predictive model has application value.ConclusionsThis study identified IL-10, PLT, CRP, PCT and Alb as risk factors for KDSS. The nomogram model can effectively predict the occurrence of KDSS in Chinese children. It will facilitate pediatricians in early diagnosis, which is essential to the prevention of cardiovascular complications.

Efficacy and safety of Chinese herbal medicine as adjunctive therapy in sepsis patients with bloodstream infection: a propensity-matched analysis.

To investigate the efficacy and safety of Chinese herbal medicine in treating sepsis patients with bloodstream infection. A 6-year retrospective study was carried out at a university hospital in China. Adult sepsis patients with bloodstream infection were included. The primary outcome was 28-day mortality after admission. Propensity score method was used to adjust for possible confounding. 28-day mortality was estimated by Kaplan-Meier analysis and compared using the log-rank test. Cox regression analysis was carried out to identify factors impacting in-hospital mortality outcomes. Following the application of the propensity score method, a total of 176 patients were included. The all-cause 28-day mortality in the control group and Chinese herbal medicine group was 21.6% and 14.8%, respectively. Kaplan-Meier survival analysis showed that Chinese herbal medicine was associated with a lower hazard ratio (HR) in all-cause 28-day death compared with the control group [HR = 0.44, 95% CI(0.22, 0.90), P < 0.05]. The complications were similar between the two groups (P >0.05). Blood-activating and stasis-eliminating herb administration was associated with reduced in-hospital mortality among sepsis patients with bloodstream infection [HR = 0.54, 95% CI(0.34, 0.94), P < 0.05]. Chinese herbal medicine, especially the blood-activating and stasis-eliminating herb, might have certain efficacy and safety in treating sepsis patients with bloodstream infection. Clinicians should prescribe blood-activating and stasis-eliminating herb in treating these two coalescent critical diseases as long as no contraindications exist. However, further studies are needed to validate our results.

Abstract TP73: Outcomes in Mexican American Informal Stroke Caregivers and Differences by Ethnicity

Background: Informal home care is the most prevalent type of post-acute care (PAC) in Mexican American (MA) stroke survivors, but data on the impact on caregivers is not available. The objective was to investigate strain, positive aspects of caregiving, familial obligations/support, and depression in stroke caregivers and ethnic differences in a population-based study. Methods: Data are from Brain Attack Surveillance in Corpus Christi PAC Study. Community-dwelling stroke survivors and their caregivers were recruited (n=307). Caregiver eligibility included: 1) residence in study county, 2) not cognitively impaired, 3) not providing care pre-stroke. Caregiver interviews at ~90 days included sociodemographics, dyad characteristics, Modified Caregiver Strain Index (MCSI range 0-26, higher more), PAC scale by Tarlow et al (range 5-45, higher more), familial obligations/support scales by Sabogal et al (range 1-5, higher greater), and PHQ-8 (range 0-30, higher worse). Patient data were from interviews and medical records. Propensity score (PS) methods were used to balance caregiver and patient factors among MA and non-Hispanic white (NHW) caregivers by fitting a model with ethnicity of caregiver as the outcome and predictors being caregiver sociodemographics, patient-caregiver dyad characteristics, and patient sociodemographics and disability. PS were then included as a covariate in regression models considering the association between ethnicity and outcomes. Results: MA caregivers were younger, more likely female, and more likely a child than NHWs. Median MCSI, PAC scale, PHQ-8 were 11 (IQR:6-17), 36 (IQR:30-40), and 11 (IQR:9-15) in MAs and 13.5 (IQR:9-18.3), 30 (IQR:25-36.3) and 13 (IQR:10-16) in NHWS. MA ethnicity was associated with less caregiver strain (β=-2.22, 95% CI:-4.19, -0.26) and depression (β=-2.05, 95% CI:-3.48, -0.63) and more positive aspects of caregiving (β=2.96, 95% CI:0.91, 5.01), but not associated with familial obligations/support. Conclusions: Informal stroke caregivers report moderate levels of strain and depressive symptoms, with NHW caregivers reporting higher levels. MA caregivers report greater positive aspects of caregiving suggesting a possible intervention target to improve outcomes in MAs.

Platelet Transfusion and Death or Neurodevelopmental Impairment in Children Born Extremely Preterm

Infants born extremely preterm receive transfusions at higher platelet count thresholds than older children and adults due to concerns for intracranial hemorrhage. A recent randomized trial comparing 2 platelet transfusion thresholds showed the higher threshold was associated with increased risk of long-term adverse neurodevelopmental outcomes. To evaluate the association of platelet transfusion exposure with death and severe neurodevelopmental impairment (NDI) at 2 years' corrected age in a cohort of infants born extremely preterm. An observational cohort study and secondary analysis of the Preterm Erythropoietin Neuroprotection Trial, a randomized, placebo-controlled clinical trial of erythropoietin neuroprotection in neonates born extremely preterm, was conducted in 30 neonatal intensive care units in the US from December 1, 2013, to September 31, 2016. This analysis included 819 infants born extremely preterm at 24 to 27 completed weeks of gestation who had a documented outcome (death or neurodevelopmental assessment). Analysis was performed in April 2023. Any platelet transfusion during neonatal intensive care unit hospitalization. The primary composite outcome was death or severe NDI evaluated at 2 years' corrected age using the Bayley Scales of Infant Development-Third Edition (BSID-III) and the Gross Motor Function Classification System and was defined as the presence of severe cerebral palsy or a BSID-III composite motor or cognitive score 2 SDs below the mean. Confounding by indication for platelet transfusion was addressed with covariate adjustment and propensity score methods. Of the 819 infants included in the analysis (429 [52.4%] male; mean [SD] gestational age, 25.5 [1.1] weeks), 245 (30.0%) received at least 1 platelet transfusion during their initial hospitalization. The primary outcome occurred in 46.5% (114 of 245) of infants exposed to a platelet transfusion and 13.9% (80 of 574) of nonexposed infants with a corresponding odds ratio of 2.43 (95% CI, 1.24-4.76), adjusted for propensity score, gestational age at birth, and trial treatment group. The individual components of death and severe NDI were directionally consistent with the overall composite outcome. The findings of this study suggest that platelet transfusion in infants born extremely preterm may be associated with an increased risk of death or severe NDI at 2 years' corrected age, although the possibility of residual confounding by indication cannot be excluded.

Impact of post-disaster federal assistance on the national flood insurance program

This research tested the hypothesis that post-disaster federal relief program crowds out the demand for flood insurance in the U.S. To do that, this research has collected historical data between 2016 and 2020, and conducted propensity score-based analysis to understand the causality between post-disaster federal assistance and flood insurance enrollment in the flood affected counties in the U.S. In the first part, the treatment variable has been considered as binary to compare the effect of the availability of federal assistance to that of non-availability of federal assistance in a flood affected county by using propensity score matching method. It was found that the availability of the federal payout in a county in a year increased the number of flood insurance policies by 5.2 % and the total insured value of the policies by 4.6 % in the following year. Next, the treatment variable was considered continuous and was used in a generalized propensity score method to develop a dose response function, i.e., a function that depicts the changes in the outcome variables based on different levels of treatment. It was found that for each 1000 households in a county that received federal assistance, the percentage increase in the number of flood insurance policies was 3.41 %. On the other hand, for each million-dollar of federal assistance in a county, the total insured value of the flood insurance policies increased by 1.96 %. Therefore, this research has concluded that, contrary to the hypothesis, the availability of post-disaster federal assistance increased the flood insurance enrollment in the U.S.

Impaired humoral immunity following COVID-19 vaccination in HTLV-1 carriers

BackgroundWhether human T-lymphotropic virus type 1 (HTLV-1) carriers can develop sufficient humoral immunity after coronavirus disease 2019 (COVID-19) vaccination is unknown.MethodsTo investigate humoral immunity after COVID-19 vaccination in HTLV-1 carriers, a multicenter, prospective observational cohort study was conducted at five institutions in southwestern Japan, an endemic area for HTLV-1. HTLV-1 carriers and HTLV-1-negative controls were enrolled for this study from January to December 2022. During this period, the third dose of the COVID-19 vaccine was actively administered. HTLV-1 carriers were enrolled during outpatient visits, while HTLV-1-negative controls included health care workers and patients treated by participating institutions for diabetes, hypertension, or dyslipidemia. The main outcome was the effect of HTLV-1 infection on the plasma anti-COVID-19 spike IgG (IgG-S) titers after the third dose, assessed by multivariate linear regression with other clinical factors.ResultsWe analyzed 181 cases (90 HTLV-1 carriers, 91 HTLV-1-negative controls) after receiving the third dose. HTLV-1 carriers were older (median age 67.0 vs. 45.0 years, p < 0.001) and more frequently had diabetes, hypertension, or dyslipidemia than did HTLV-1-negative controls (60.0% vs. 27.5%, p < 0.001). After the third dose, the IgG-S titers decreased over time in both carriers and controls. Multivariate linear regression in the entire cohort showed that time since the third dose, age, and HTLV-1 infection negatively influenced IgG-S titers. After adjusting for confounders such as age, or presence of diabetes, hypertension, or dyslipidemia between carriers and controls using the overlap weighting propensity score method, and performing weighted regression analysis in the entire cohort, both time since the third dose and HTLV-1 infection negatively influenced IgG-S titers.ConclusionsThe humoral immunity after the third vaccination dose is impaired in HTLV-1 carriers; thus, customized vaccination schedules may be necessary for them.

Comparing quality indicator rates for home care clients receiving palliative and end-of-life care before and during the Covid-19 pandemic

BackgroundThe consensus among Canadians with regards to end-of-life preferences is that with adequate support the majority prefer to live and die at home.PurposeTo compare quality indicator (QI) rates for home care clients receiving palliative and end-of-life care prior to and after the onset of the COVID-19 pandemic.MethodsA retrospective population-based cohort design was used. Sixteen QIs informed by existing literature and a preliminary set of QIs recently evaluated by a modified Delphi panel were compared. Data were obtained from the interRAI Palliative Care instrument for Ontario home care clients for two separate cohorts: the pre-COVID (January 14, 2019 to March 16, 2020) and COVID cohort (March 17, 2020 to May 18, 2021). A propensity score analysis was used to match (using nearest neighbour matching) on 21 covariates, resulting in a sample size of 2479 unique interRAI Palliative Care assessments in each cohort. Alternative propensity score methods were explored as part of a sensitivity analysis.ResultsAfter matching the pre-COVID and COVID cohorts, five of the 16 QIs had statistically significant differences in the QI rates (change from pre-COVID to COVID): decrease in prevalence of severe or excruciating daily pain (p = 0.03, effect size=-0.08), decrease in prevalence of caregiver distress (p = 0.02, effect size=-0.06), decrease in prevalence of negative mood (p = 0.003, effect size=- 0.17), decrease in prevalence of a delirium-like syndrome (p = 0.001, effect size=-0.25) and decrease in prevalence of nausea or vomiting (p = 0.04, effect size=-0.06). While the alternative propensity score methods produced slightly different results, no clinically meaningful differences were seen between the cohorts when effect sizes were examined. All methods were in agreement regarding the highest QI rates, which included the prevalence of shortness of breath with activity, no advance directives, and fatigue.ConclusionThis study is the first to examine differences in QI rates for home care clients receiving palliative and end-of-life care before and during COVID in Ontario. It appears that QI rates did not change over the course of the pandemic in this population. Future work should be directed to understanding the temporal variation in these QI rates, risk-adjusting the QI rates for further comparison among jurisdictions, provinces, and countries, and in creating benchmarks for determining acceptable rates of different QIs.

Clinical outcomes of sentinel node navigation surgery in patients with preoperatively estimated stage IA endometrial cancer and evaluation of validity for continuing sentinel node navigation surgery based on dispersion of recurrence probability.

To evaluate the feasibility of the use and continuation of sentinel lymph node navigation surgery (SNNS) as an alternative to pelvic lymph node dissection (PLND) for patients with preoperatively estimated stage IA endometrial cancer. This retrospective study selected the electronic medical records of all patients who had received CT scans and MRI imaging before surgery from April 1, 2009 to March 31, 2021. Sentinel lymph nodes (SLNs) were detected by administrating 99mTc-phytate and/or indocyanine green into the cervix, and the clinical outcomes of the patients who underwent SNNS or PLND were evaluated. Furthermore, in case of nodal recurrence, a new procedure to determine whether the facility should continue with SNNS or not was developed that compares the maximum likelihood hypothesis and an alternative one based on recurrence rates. Among 137 patients, SLN biopsies with ultrastaging were performed on 91 patients. The SLN detection rate was 95.6%. Over a 59-month median observation period, no statistically significant differences were shown in overall survival, disease-specific survival and disease-free survival between the SNNS and PLND groups when introducing the propensity score method (p-values: 0.06, 0.153, and 0.625, respectively). Our procedure demonstrated that, in our department without recurrence up to the 65th attempt, it was possible to continue SNNS if a recurrence occurs at the 66th attempt. This study suggests the validity of SNNS as an alternative to PLND. Even in the absence of evidence from randomized controlled trials, we can confirm the validity of continuing SNNS using our procedure.

Which employers pay a higher college wage premium?

PurposeThe authors' work aims to identify the employer-specific drivers of the college (or university) wage gap, which has been identified as one of the major determinants of the dynamics of overall wage and income inequality in the past decades. The authors focus on three employer-level features that can be associated with asymmetries in the employment relation orientation adopted for college and non-college-educated employees: (1) size, (2) the share of standard employment and (3) the pervasiveness of incentive pay schemes.Design/methodology/approachThe authors' establishment-level analysis (data from the Basic Survey on Wage Structure (BSWS), 2005–2018) focusses on Japan, an economy characterised by many unique economic and institutional features relevant to the aims of the authors' analysis. The authors use an adjusted measure of firm-specific college wage premium, which is not biased by confounding individual and establishment-level factors and reflects unobservable characteristics of employees that determine the payment of a premium. The authors' empirical methods account for the complexity of the relationships they investigate, and the authors test their baseline outcomes with econometric approaches (propensity score methods) able to address crucial identification issues related to endogeneity and reverse causality.FindingsThe authors' findings indicate that larger establishment size, a larger share of regular workers and more pervasive implementation of IPSs for college workers tend to increase the college wage gap once all observable workers, job and establishment characteristics are controlled for. This evidence corroborates the authors' hypotheses that a larger establishment size, a higher share of regular workers and a more developed set-up of performance pay schemes for college workers are associated with a better capacity of employers to attract and keep highly educated employees with unobservable characteristics that justify a wage premium above average market levels. The authors provide empirical evidence on how three relevant establishment-level characteristics shape the heterogeneity of the (adjusted) college wage observed across organisations.Originality/valueThe authors' contribution to the existing knowledge is threefold. First, the authors combine the economics and management/organisation literature to develop new insights that underpin the authors' testable empirical hypotheses. This enables the authors to shed light on employer-level drivers of wage differentials (size, workforce composition, implementation of performance-pay schemes) related to many structural, institutional and strategic dimensions. The second contribution lies in the authors' measure of the “adjusted” college wage gap, which is calculated on the component of individual wages that differs between observationally identical workers in the same establishment. As such, the metric captures unobservable workers' characteristics that can generate a wage premium/penalty. Third, the authors provide empirical evidence on how three relevant establishment-level characteristics shape the heterogeneity of the (adjusted) college wage observed across organisations.

Handling of outcome missing data dependent on measured or unmeasured background factors in micro-randomized trial: Simulation and application study.

Micro-randomized trials (MRTs) enhance the effects of mHealth by determining the optimal components, timings, and frequency of interventions. Appropriate handling of missing values is crucial in clinical research; however, it remains insufficiently explored in the context of MRTs. Our study aimed to investigate appropriate methods for missing data in simple MRTs with uniform intervention randomization and no time-dependent covariates. We focused on outcome missing data depending on the participants' background factors. We evaluated the performance of the available data analysis (AD) and the multiple imputation in generalized estimating equations (GEE) and random effects model (RE) through simulations. The scenarios were examined based on the presence of unmeasured background factors and the presence of interaction effects. We conducted the regression and propensity score methods as multiple imputation. These missing data handling methods were also applied to actual MRT data. Without the interaction effect, AD was biased for GEE, but there was almost no bias for RE. With the interaction effect, estimates were biased for both. For multiple imputation, regression methods estimated without bias when the imputation models were correct, but bias occurred when the models were incorrect. However, this bias was reduced by including the random effects in the imputation model. In the propensity score method, bias occurred even when the missing probability model was correct. Without the interaction effect, AD of RE was preferable. When employing GEE or anticipating interactions, we recommend the multiple imputation, especially with regression methods, including individual-level random effects.

Why Are We Weighting? Understanding the Estimates From Propensity Score Weighting and Matching Methods.

Propensity score methods are used in observational studies to compensate for the lack of random allocation by balancing measured baseline characteristics between treated and untreated patients. We sought to explain the treatment effect estimates derived from different propensity score methods. We performed a retrospective analysis of long-term mortality after single internal mammary artery versus bilateral internal mammary artery (BIMA) conduit in 47 984 index isolated coronary artery bypass grafting procedures from 1992 to 2014 in the Northern New England Cardiovascular Disease Study Group registry using multivariable Cox regression, 1:1 propensity score matching, inverse probability weighting (IPW) among the treated, and IPW among the overall population treatment estimates. The mean duration of follow-up was 13.2 (interquartile range, 7.4-17.7) years. In multivariable Cox regression, the adjusted hazard ratio for mortality was 0.83 (95% CI, 0.75-0.92) in patients receiving BIMA compared with a single internal mammary artery. The 1:1 propensity matched (hazard ratio, 0.79 [95% CI, 0.69-0.91]) and IPW among the treated (hazard ratio, 0.83 [95% CI, 0.75-0.92]) estimates showed a protective treatment effect of BIMA use on mortality. However, the IPW estimate of treatment effect for the overall population showed an increased risk of mortality after BIMA that was not statistically significant (hazard ratio, 1.08 [95% CI, 0.94-1.24]). While the multivariable Cox regression, 1:1 propensity matching, and IPW treatment effect in the treated estimates demonstrate that BIMA was associated with a statistically significantly decreased risk of mortality, the IPW treatment effect in the average study population showed an increased risk of mortality associated with BIMA that was not statistically significant. This is attributed to the different populations (weighted to look like the overall study population versus treated group) represented by the 2 IPW approaches. Determining how the study population is balanced is a large driver of the treatment effect. Ultimately, the treatment effect estimate desired should drive the choice of the propensity score method.

The Causal Cookbook: Recipes for Propensity Scores, G-Computation, and Doubly Robust Standardization

Recent developments in the causal-inference literature have renewed psychologists’ interest in how to improve causal conclusions based on observational data. A lot of the recent writing has focused on concerns of causal identification (under which conditions is it, in principle, possible to recover causal effects?); in this primer, we turn to causal estimation (how do researchers actually turn the data into an effect estimate?) and modern approaches to it that are commonly used in epidemiology. First, we explain how causal estimands can be defined rigorously with the help of the potential-outcomes framework, and we highlight four crucial assumptions necessary for causal inference to succeed (exchangeability, positivity, consistency, and noninterference). Next, we present three types of approaches to causal estimation and compare their strengths and weaknesses: propensity-score methods (in which the independent variable is modeled as a function of controls), g-computation methods (in which the dependent variable is modeled as a function of both controls and the independent variable), and doubly robust estimators (which combine models for both independent and dependent variables). A companion R Notebook is available at github.com/ArthurChatton/CausalCookbook. We hope that this nontechnical introduction not only helps psychologists and other social scientists expand their causal toolbox but also facilitates communication across disciplinary boundaries when it comes to causal inference, a research goal common to all fields of research.

Impact of Masticatory Performance and the Tongue-Lip Motor Function on Incident Adverse Health Events in Patients with Metabolic Disease.

The present study aimed to determine whether decreased masticatory performance and tongue-lip motor function are associated with an increased incidence of adverse health events in patients with metabolic disease. One thousand patients with metabolic diseases including diabetes, dyslipidemia, hypertension, and hyperuricemia were recruited. Masticatory performance was assessed using a gummy jelly test, wherein glucose elution from chewed gummy jelly was measured. The tongue-lip motor function was measured using repeatedly pronounced syllables per second. Their association with the incidence of adverse health events (a composite of all-cause death, cardiovascular disease, bone fracture, malignant neoplasm, pneumonia, and dementia) was investigated using the generalized propensity score (GPS) method. During a median follow-up period of 36.6 (interquartile range, 35.0-37.7) months, adverse health events were observed in 191 patients. The GPS adjusted dose-response function demonstrated that masticatory performance was inversely associated with the incidence of adverse health events. The 3-year incidence rate was 22.8% (95% confidence interval, 19.0-26.4%) for the lower quartile versus 13.6% (10.5-16.7%) for the upper quartile (P＜0.001). Similarly, the tongue-lip motor function was inversely associated with the incidence of adverse health events, with a 3-year incidence rate of 23.6% (20.0-27.0%) for the lower quartile versus 13.2% (10.4-15.9%) for the upper quartile (P＜0.001). Decreased masticatory performance and tongue-lip motor function were associated with an increased incidence of adverse health events in patients with metabolic disease.

Comparing Machine Learning and Advanced Methods with Traditional Methods to Generate Weights in Inverse Probability of Treatment Weighting: The INFORM Study.

Observational research provides valuable insights into treatments used in patient populations in real-world settings. However, confounding is likely to occur if there are differences in patient characteristics associated with both the exposure and outcome between the groups being evaluated. One approach to reduce confounding and facilitate unbiased comparisons is inverse probability of treatment weighting (IPTW) using propensity scores. Machine learning (ML) and entropy balancing can potentially be used in generating propensity scores for IPTW, but there is limited literature on this application. We aimed to assess the feasibility of applying these methods for reducing confounding in observational studies. These methods were assessed in a study comparing cardiovascular outcomes in adults with type 2 diabetes and established atherosclerotic cardiovascular disease taking once-weekly glucagon-like peptide-1 receptor agonists or dipeptidyl peptidase-4 inhibitors. We applied advanced methods to generate the propensity scores compared to the original logistic regression method in terms of covariate balance. After calculating weights, a weighted Cox proportional hazards model was used to calculate the sample average treatment effect. Support Vector Classification, Support Vector Regression, XGBoost, and LightGBM were the ML models used. Entropy balancing was also performed on features identified in the original cardiovascular outcomes study. Accuracy (range: 0.71 to 0.73), area under the curve (0.77 to 0.79), precision (0.53 to 0.60), recall (0.66 to 0.68), and F1 score (0.60 to 0.64) were similar between all of the advanced propensity score methods and traditional logistic regression. Among ML models, only XGBoost achieved balance in all measured baseline characteristics between the two treatment groups, closely approximating the performance of the original logistic regression. Entropy balancing weights provided the best performance among all models in balancing baseline characteristics, achieving near perfect balancing. Among the advanced methods examined, entropy balancing weights performed the best for optimizing balancing and can produce similar results compared to traditional logistic regression.