Association of copayments with healthcare utilization and expenditures among Medicaid enrollees with a substance use disorder

BACKGROUND

The purpose of this study was to examine the association between copayments and healthcare utilization and expenditures among Medicaid enrollees with substance use disorders.

METHODS

This study used claims data (2020-2021) from a private insurer participating in Arkansas's Medicaid expansion. We compared service utilization and expenditures for enrollees in different Medicaid program structures with varying copayments. Enrollees with incomes above 100 % FPL (N = 10,240) had copayments for substance use treatment services while enrollees below 100 % FPL (N = 2478) did not. Demographic, diagnostic, utilization, and cost information came from claims and enrollment information. The study identified substance use and clinical comorbidities using claims from July through December 2020 and evaluated utilization and costs in 2021. Generalized linear models (GLM) estimated outcomes using single equation and two-part modeling. A gamma distribution and log link were used to model expenditures, and negative binomial models were used to model utilization. A falsification test comparing behavioral health telemedicine utilization, which had no cost sharing in either group, assessed whether differences in the groups may be responsible for observed findings.

RESULTS

Substance use enrollees with copayments were less likely to have a substance use or behavioral health outpatient (-0.04 PP adjusted; p = 0.001) or inpatient visit (-0.04 PP; p = 0.001) relative to their counterparts without copayments, equal to a 17 % reduction in substance use or behavioral health outpatient services and a nearly 50 % reduction in inpatient visits. The reduced utilization among enrollees with a copayment was associated with a significant reduction in total expenses ($954; p = 0.001) and expenses related to substance use or behavioral health services ($532; p = 0.001). For enrollees with at least one behavioral health visit, there were no differences in outpatient or inpatient utilization or expenditures between enrollees with and without copayments. Copayments had no association with non-behavioral health or telemedicine services where neither group had cost sharing.

CONCLUSION

Copayments serve as an initial barrier to substance use treatment, but are not associated with the amount of healthcare utilization conditional on using services. Policy makers and insurers should consider the role of copayments for treatment services among enrollees with substance use disorders in Medicaid programs.

Differentially Expressed Heterogeneous Overdispersion Genes Testing for Count Data

The mRNA-seq data analysis is a powerful technology for inferring information from biological systems of interest. Specifically, the sequenced RNA fragments are aligned with genomic reference sequences, and we count the number of sequence fragments corresponding to each gene for each condition. A gene is identified as differentially expressed (DE) if the difference in its count numbers between conditions is statistically significant. Several statistical analysis methods have been developed to detect DE genes based on RNA-seq data. However, the existing methods could suffer decreasing power to identify DE genes arising from overdispersion and limited sample size. We propose a new differential expression analysis procedure: heterogeneous overdispersion genes testing (DEHOGT) based on heterogeneous overdispersion modeling and a post-hoc inference procedure. DEHOGT integrates sample information from all conditions and provides a more flexible and adaptive overdispersion modeling for the RNA-seq read count. DEHOGT adopts a gene-wise estimation scheme to enhance the detection power of differentially expressed genes. DEHOGT is tested on the synthetic RNA-seq read count data and outperforms two popular existing methods, DESeq and EdgeR, in detecting DE genes. We apply the proposed method to a test dataset using RNAseq data from microglial cells. DEHOGT tends to detect more differently expressed genes potentially related to microglial cells under different stress hormones treatments.

Interpreting Interaction Effects in Generalized Linear Models of Nonlinear Probabilities and Counts

Psychology research frequently involves the study of probabilities and counts. These are typically analyzed using generalized linear models (GLMs), which can produce these quantities via nonlinear transformation of model parameters. Interactions are central within many research applications of these models. To date, typical practice in evaluating interactions for probabilities or counts extends directly from linear approaches, in which evidence of an interaction effect is supported by using the product term coefficient between variables of interest. However, unlike linear models, interaction effects in GLMs describing probabilities and counts are not equal to product terms between predictor variables. Instead, interactions may be functions of the predictors of a model, requiring nontraditional approaches for interpreting these effects accurately. Here, we define interactions as change in a marginal effect of one variable as a function of change in another variable, and describe the use of partial derivatives and discrete differences for quantifying these effects. Using guidelines and simulated examples, we then use these approaches to describe how interaction effects should be estimated and interpreted for GLMs on probability and count scales. We conclude with an example using the Adolescent Brain Cognitive Development Study demonstrating how to correctly evaluate interaction effects in a logistic model.

Estimating completeness of national and subnational death reporting in Brazil: application of record linkage methods

Background

In Brazil, both the Civil Registry (CR) and Ministry of Health (MoH) Mortality Information System (SIM) are sources of routine mortality data, but neither is 100% complete. Deaths from these two sources can be linked to facilitate estimation of completeness of mortality reporting and measurement of adjusted mortality indicators using generalized linear modeling (GLM).

Methods

The 2015 and 2016 CR and SIM data were linked using deterministic methods. GLM with covariates of the deceased’s sex, age, state of residence, cause of death and place of death, and municipality-level education decile and population density decile, was used to estimate total deaths and completeness nationally, subnationally and by population sub-group, and to identify the characteristics of unreported deaths. The empirical completeness method and Global Burden of Disease (GBD) 2017 estimates were comparators at the national and state level.

Results

Completeness was 98% for SIM and 95% for CR. The vast majority of deaths in Brazil were captured by either system and 94% were reported by both sources. For each source, completeness was lowest in the north. SIM completeness was consistently high across all sub-groups while CR completeness was lowest for deaths at younger ages, outside facilities, and in the lowest deciles of municipality education and population density. There was no clear municipality-level relationship in SIM and CR completeness, suggesting minimal dependence between sources. The empirical completeness method model 1 and GBD completeness estimates were each, on average, less than three percentage points different from GLM estimates at the state level. Life expectancy was lowest in the northeast and 7.5 years higher in females than males.

Conclusions

GLM using socio-economic and demographic covariates is a valuable tool to accurately estimate completeness from linked data sources. Close scrutiny of the quality of variables used to link deaths, targeted identification of unreported deaths in poorer, northern states, and closer coordination of the two systems will help Brazil achieve 100% death reporting completeness. The results also confirm the validity of the empirical completeness method.

Predicting Tests Ordered in Hospital Laboratories using Generalized Linear Modeling

Laboratory services in healthcare systems play a vital role in inpatient care. Most hospital laboratories are facing the challenge of reducing cost and improving service quality. The author focuses on identifying test order patterns in a laboratory for a large urban hospital. The data collected from this facility consists of all tests ordered over a three-month time frame and contains test orders for approximately 17,500 patients. Poisson and negative binomial regression models are used to determine how well patient characteristics (patient length of stay and the medical units in which patients are placed) will predict the number of tests being ordered. The test order prediction model developed in this study will aid the management and phlebotomists in the hospital laboratory in securing methods to satisfy the test order demand. By implementing the recommendations of this study, hospital laboratories should see significant improvements in phlebotomist productivity and resource utilization, implementation of which could result in cost savings.