Isolating cost drivers in interstitial lung disease treatment using nonparametric Bayesian methods.

Mixture modeling is a popular approach to accommodate overdispersion, skewness, and multimodality features that are very common for health care utilization data. However, mixture modeling tends to rely on subjective judgment regarding the appropriate number of mixture components or some hypothesis about how to cluster the data. In this work, we adopt a nonparametric, variational Bayesian approach to allow the model to select the number of components while estimating their parameters. Our model allows for a probabilistic classification of observations into clusters and simultaneous estimation of a Gaussian regression model within each cluster. When we apply this approach to data on patients with interstitial lung disease, we find distinct subgroups of patients with differences in means and variances of health care costs, health and treatment covariates, and relationships between covariates and costs. The subgroups identified are readily interpretable, suggesting that this nonparametric variational approach to inference can discover valid insights into the factors driving treatment costs. Moreover, the learning algorithm we employed is very fast and scalable, which should make the technique accessible for a broad range of applications.

Identifying and interpreting subgroups in health care utilization data with count mixture regression models.

Inpatient care is a large share of total health care spending, making analysis of inpatient utilization patterns an important part of understanding what drives health care spending growth. Common features of inpatient utilization measures such as length of stay and spending include zero inflation, overdispersion, and skewness, all of which complicate statistical modeling. Moreover, latent subgroups of patients may have distinct patterns of utilization and relationships between that utilization and observed covariates. In this work, we apply and compare likelihood-based and parametric Bayesian mixtures of negative binomial and zero-inflated negative binomial regression models. In a simulation, we find that the Bayesian approach finds the true number of mixture components more accurately than using information criteria to select among likelihood-based finite mixture models. When we apply the models to data on hospital lengths of stay for patients with lung cancer, we find distinct subgroups of patients with different means and variances of hospital days, health and treatment covariates, and relationships between covariates and length of stay.

The effect of bariatric surgery on health care costs: A synthetic control approach using Bayesian structural time series.

Surgical measures to combat obesity are very effective in terms of weight loss, recovery from diabetes, and improvement in cardiovascular risk factors. However, previous studies found both positive and negative results regarding the effect of bariatric surgery on health care utilization. Using claims data from the largest health insurance provider in Germany, we estimated the causal effect of bariatric surgery on health care costs in a time period ranging from 2 years before to 3 years after bariatric intervention. Owing to the absence of a control group, we employed a Bayesian structural forecasting model to construct a synthetic control. We observed a decrease in medication and physician expenditures after bariatric surgery, whereas hospital expenditures increased in the post-intervention period. Overall, we found a slight increase in total costs after bariatric surgery, but our estimates include a high degree of uncertainty.

Tweedie distributions for fitting semicontinuous health care utilization cost data.

BACKGROUND: The statistical analysis of health care cost data is often problematic because these data are usually non-negative, right-skewed and have excess zeros for non-users. This prevents the use of linear models based on the Gaussian or Gamma distribution. A common way to counter this is the use of Two-part or Tobit models, which makes interpretation of the results more difficult. In this study, I explore a statistical distribution from the Tweedie family of distributions that can simultaneously model the probability of zero outcome, i.e. of being a non-user of health care utilization and continuous costs for users. METHODS: I assess the usefulness of the Tweedie model in a Monte Carlo simulation study that addresses two common situations of low and high correlation of the users and the non-users of health care utilization. Furthermore, I compare the Tweedie model with several other models using a real data set from the RAND health insurance experiment. RESULTS: I show that the Tweedie distribution fits cost data very well and provides better fit, especially when the number of non-users is low and the correlation between users and non-users is high. CONCLUSION: The Tweedie distribution provides an interesting solution to many statistical problems in health economic analyses.

Augmented Inverse Probability Weighting and the Double Robustness Property.

This article discusses the augmented inverse propensity weighted (AIPW) estimator as an estimator for average treatment effects. The AIPW combines both the properties of the regression-based estimator and the inverse probability weighted (IPW) estimator and is therefore a "doubly robust" method in that it requires only either the propensity or outcome model to be correctly specified but not both. Even though this estimator has been known for years, it is rarely used in practice. After explaining the estimator and proving the double robustness property, I conduct a simulation study to compare the AIPW efficiency with IPW and regression under different scenarios of misspecification. In 2 real-world examples, I provide a step-by-step guide on implementing the AIPW estimator in practice. I show that it is an easily usable method that extends the IPW to reduce variability and improve estimation accuracy.[Box: see text].

The effect of differential privacy on Medicaid participation among racial and ethnic minority groups.

OBJECTIVE: To investigate how county and state-level estimates of Medicaid enrollment among the total, non-Hispanic White, non-Hispanic Black or African American, and Hispanic or Latino/a population are affected by Differential Privacy (DP), where statistical noise is added to the public decennial US census data to protect individual privacy. DATA SOURCES: We obtained population counts from the final version of the US Census Bureau Differential Privacy Demonstration Products from 2010 and combined them with Medicaid enrollment data. STUDY DESIGN: We compared 2010 county and state-level population counts released under the traditional disclosure avoidance techniques and the ones produced with the proposed DP procedures. DATA COLLECTION/EXTRACTION METHODS: Not applicable. PRINCIPAL FINDINGS: We find the DP method introduces errors up to 10% into counts and proportions of Medicaid participation rate accuracy at the county level, especially for small subpopulations and racial and ethnic minority groups. The effect of DP on Medicaid participation rate accuracy is only small and negligible at the state level. CONCLUSIONS: The implementation of DP in the 2020 census can affect the analyses of health disparities and health care access and use among different subpopulations in the United States. The planned implementation of DP in other census-related surveys such as the American Community Survey can misrepresent Medicaid participation rates for small racial and ethnic minority groups. This can affect Medicaid funding decisions.

The causal impact of sugar taxes on soft drink sales: evidence from France and Hungary.

Sugar-sweetened beverages (SSBs) are associated with increased body weight and obesity, which induce a wide array of health impairments such as diabetes or cardiovascular disorders. Excise taxes have been introduced to counteract SSB consumption. We investigated the effect of sugar taxes on SSB sales in Hungary and France using a synthetic control approach. For France, we found a slight decrease in SSB sales after tax implementation while overall soft drink sales increased. For Hungary, there was only a short-term decrease in SSB sales which disappeared after 2 years, leading to an overall increase in SSB sales. However, both effects are characterized by great uncertainty.

The Effect of BMI and Type 2 Diabetes on Socioeconomic Status: A Two-Sample Multivariable Mendelian Randomization Study.

OBJECTIVE: To assess the independent causal effect of BMI and type 2 diabetes (T2D) on socioeconomic outcomes by applying two-sample Mendelian randomization (MR) analysis. RESEARCH DESIGN AND METHODS: We performed univariable and multivariable two-sample MR to jointly assess the effect of BMI and T2D on socioeconomic outcomes. We used overlapping genome-wide significant single nucleotide polymorphisms for BMI and T2D as instrumental variables. Their causal impact on household income and regional deprivation was assessed using summary-level data from the UK Biobank. RESULTS: In the univariable analysis, higher BMI was related to lower income (marginal effect of 1-SD increase in BMI [ß = -0.092; 95% CI -0.138; -0.047]) and higher deprivation (ß = 0.051; 95% CI 0.022; 0.079). In the multivariable MR, the effect of BMI controlling for diabetes was slightly lower for income and deprivation. Diabetes was not associated with these outcomes. CONCLUSIONS: High BMI, but not diabetes, shows a causal link with socioeconomic outcomes.

Application of Mendelian Randomization to Investigate the Association of Body Mass Index with Health Care Costs.

Causal effect estimates for the association of obesity with health care costs can be biased by reversed causation and omitted variables. In this study, we use genetic variants as instrumental variables to overcome these limitations, a method that is often called Mendelian randomization (MR). We describe the assumptions, available methods, and potential pitfalls of using genetic information and how to address them. We estimate the effect of body mass index (BMI) on total health care costs using data from a German observational study and from published large-scale data. In a meta-analysis of several MR approaches, we find that models using genetic instruments identify additional annual costs of €280 for a 1-unit increase in BMI. This is more than 3 times higher than estimates from linear regression without instrumental variables (€75). We found little evidence of a nonlinear relationship between BMI and health care costs. Our results suggest that the use of genetic instruments can be a powerful tool for estimating causal effects in health economic evaluation that might be superior to other types of instruments where there is a strong association with a modifiable risk factor.

A greedy stacking algorithm for model ensembling and domain weighting.

OBJECTIVE: Because it is impossible to know which statistical learning algorithm performs best on a prediction task, it is common to use stacking methods to ensemble individual learners into a more powerful single learner. Stacking algorithms are usually based on linear models, which may run into problems, especially when predictions are highly correlated. In this study, we develop a greedy algorithm for model stacking that overcomes this issue while still being very fast and easy to interpret. We evaluate our greedy algorithm on 7 different data sets from various biomedical disciplines and compare it to linear stacking, genetic algorithm stacking and a brute force approach in different prediction settings. We further apply this algorithm on a task to optimize the weighting of the single domains (e.g., income, education) that build the German Index of Multiple Deprivation (GIMD) to be highly correlated with mortality. RESULTS: The greedy stacking algorithm provides good ensemble weights and outperforms the linear stacker in many tasks. Still, the brute force approach is slightly superior, but is computationally expensive. The greedy weighting algorithm has a variety of possible applications and is fast and efficient. A python implementation is provided.