Estimating Long-Term Drinking Patterns for People with Lifetime Alcohol Use Disorder

The Rising Costs of Alcohol-Associated Liver Disease in the United States

INTRODUCTION

Alcohol-associated liver disease (ALD) is rising in the United States because of an increase in high-risk drinking, but population-level ALD cost is unknown. Our aim was to project the direct and indirect costs associated with ALD in the US population through 2040.

METHODS

We used a previously validated microsimulation model of alcohol consumption and ALD with model parameters estimated from publicly available data sources, including the National Epidemiologic Survey Alcohol and Related Conditions-III, the Center for Disease Control and Prevention Wide-ranging Online Data for Epidemiologic Research, the Bureau of Labor Statistics, and published studies informing the impact of alcohol consumption on ALD severity in the United States resident population. The simulated scenario included current and projected ALD-associated costs.

RESULTS

From 2022 to 2040, the ALD is projected to cost $880 billion, $355 billion in direct healthcare-related costs, and $525 billion in lost labor and economic consumption. The annual cost of ALD is projected to increase from $31 billion in 2022 to $66 billion (118% increase) in 2040. Although the female population makes up 29% of these costs in 2022, by 2040 on a per annum basis, female costs would be 43% of the total annual expenditure.

DISCUSSION

Increased consumption of alcohol in the US population, especially in females, will cause a steep rise in the economic burden of ALD in the United States. These findings highlight the need for planners and policymakers to plan for the increased impact of liver disease in the United States.

Modeling the impact of a long-term horizon and multiple treatment episodes on estimates of the cost-effectiveness of alcohol treatment in the United States

BACKGROUND

Most clinical studies of alcohol use disorder (AUD) treatment have short follow-up periods, underestimating the full benefits of alcohol treatment. Furthermore, clinical studies only consider one treatment cycle and do not account for the need for multiple episodes to treat a chronic recurrent condition.

METHODS

A validated microsimulation model of the long-term drinking patterns of people with AUD in the United States simulated 10,000 individuals resembling those from a large clinical trial. The model was used to assess the impact of (1) 1-year, 5-year, and lifetime horizon on alcohol treatment cost-effectiveness estimates and (2) no, one, two, four, and unlimited additional treatment episodes on alcohol treatment cost-effectiveness estimates. Model outcomes included healthcare costs, crime costs, labor market productivity, life expectancy, quality-adjusted life years (QALYs), alcohol-related hospitalizations, and deaths. Cost-effectiveness analyses were conducted for two perspectives: a healthcare perspective that included costs from hospitalization and AUD treatment, and a broader societal perspective that also included crime costs and productivity.

RESULTS

The incremental cost per additional QALY gained for alcohol treatment compared with no treatment decreased from $55,590 after 1 year to $78 when healthcare costs and QALYs were tracked over the lifetime, that is, treatment became more cost effective. Treatment was cost saving for any time frame when the impacts on crime and labor productivity were also accounted for in a societal perspective. Access to multiple treatment episodes dominated (i.e., it was more effective and less costly) than no-treatment and one-episode scenarios. From a healthcare perspective, incremental costs per additional QALY for increasing from a maximum of two to four treatment episodes was $499 and from four to unlimited episodes was $5049. The unlimited treatment scenario dominated all others from a societal perspective. Results were robust in sensitivity analyses.

CONCLUSIONS

A long-term perspective and multiple episodes of alcohol treatment improve cost-effectiveness estimates. When societal impacts are included, alcohol treatment is cost saving. Results support the value of alcohol treatment.

Calibration and Validation of the Colorectal Cancer and Adenoma Incidence and Mortality (CRC-AIM) Microsimulation Model Using Deep Neural Networks

OBJECTIVES

Machine learning (ML)-based emulators improve the calibration of decision-analytical models, but their performance in complex microsimulation models is yet to be determined.

METHODS

We demonstrated the use of an ML-based emulator with the Colorectal Cancer (CRC)-Adenoma Incidence and Mortality (CRC-AIM) model, which includes 23 unknown natural history input parameters to replicate the CRC epidemiology in the United States. We first generated 15,000 input combinations and ran the CRC-AIM model to evaluate CRC incidence, adenoma size distribution, and the percentage of small adenoma detected by colonoscopy. We then used this data set to train several ML algorithms, including deep neural network (DNN), random forest, and several gradient boosting variants (i.e., XGBoost, LightGBM, CatBoost) and compared their performance. We evaluated 10 million potential input combinations using the selected emulator and examined input combinations that best estimated observed calibration targets. Furthermore, we cross-validated outcomes generated by the CRC-AIM model with those made by CISNET models. The calibrated CRC-AIM model was externally validated using the United Kingdom Flexible Sigmoidoscopy Screening Trial (UKFSST).

RESULTS

The DNN with proper preprocessing outperformed other tested ML algorithms and successfully predicted all 8 outcomes for different input combinations. It took 473 s for the trained DNN to predict outcomes for 10 million inputs, which would have required 190 CPU-years without our DNN. The overall calibration process took 104 CPU-days, which included building the data set, training, selecting, and hyperparameter tuning of the ML algorithms. While 7 input combinations had acceptable fit to the targets, a combination that best fits all outcomes was selected as the best vector. Almost all of the predictions made by the best vector laid within those from the CISNET models, demonstrating CRC-AIM's cross-model validity. Similarly, CRC-AIM accurately predicted the hazard ratios of CRC incidence and mortality as reported by UKFSST, demonstrating its external validity. Examination of the impact of calibration targets suggested that the selection of the calibration target had a substantial impact on model outcomes in terms of life-year gains with screening.

CONCLUSIONS

Emulators such as a DNN that is meticulously selected and trained can substantially reduce the computational burden of calibrating complex microsimulation models.

HIGHLIGHTS

Calibrating a microsimulation model, a process to find unobservable parameters so that the model fits observed data, is computationally complex.We used a deep neural network model, a popular machine learning algorithm, to calibrate the Colorectal Cancer Adenoma Incidence and Mortality (CRC-AIM) model.We demonstrated that our approach provides an efficient and accurate method to significantly speed up calibration in microsimulation models.The calibration process successfully provided cross-model validation of CRC-AIM against 3 established CISNET models and also externally validated against a randomized controlled trial.

Behavioral stability of alcohol consumption and socio-demographic correlates of change among a nationally representative cohort of US adults

AIMS

To estimate the probability of transitioning between different categories of alcohol use (drinking states) among a nationally representative cohort of United States (US) adults and to identify the effects of socio-demographic characteristics on those transitions.

DESIGN, SETTING AND PARTICIPANTS

Secondary analysis of data from the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC), a prospective cohort study conducted in 2001-02 and 2004-05; a US nation-wide, population-based study. Participants included 34 165 adults (mean age = 45.1 years, standard deviation = 17.3; 52% women).

MEASUREMENTS

Alcohol use was self-reported and categorized based on the grams consumed per day: (1) non-drinker (no drinks in past 12 months), (2) category I (women = ≤ 20; men = ≤ 40), (3) category II (women = 21-40; men = 41-60) and (4) category III (women = ≥ 41; men = ≥ 61). Multi-state Markov models estimated the probability of transitioning between drinking states, conditioned on age, sex, race/ethnicity and educational attainment. Analyses were repeated with alcohol use categorized based on the frequency of heavy episodic drinking.

FINDINGS

The highest transition probabilities were observed for staying in the same state; after 1 year, the probability of remaining in the same state was 90.1% [95% confidence interval (CI) = 89.7%, 90.5%] for non-drinkers, 90.2% (95% CI = 89.9%, 90.5%) for category I, 31.8% (95% CI = 29.7, 33.9%) category II and 52.2% (95% CI = 46.0, 58.5%) for category III. Women, older adults, and non-Hispanic Other adults were less likely to transition between drinking states, including transitions to lower use. Adults with lower educational attainment were more likely to transition between drinking states; however, they were also less likely to transition out of the 'weekly HED' category. Black adults were more likely to transition into or stay in higher use categories, whereas Hispanic/Latinx adults were largely similar to White adults.

CONCLUSIONS

In this study of alcohol transition probabilities, some demographic subgroups appeared more likely to transition into or persist in higher alcohol consumption states.

Modeling the impact of COVID-19 pandemic-driven increases in alcohol consumption on health outcomes and hospitalization costs in the United States

BACKGROUND AND AIMS

Alcohol consumption increased in the early phases of the COVID-19 pandemic in the United States. Alcohol use disorder (AUD) and risky drinking are linked to harmful health effects. This paper aimed to project future health and cost impacts of shifts in alcohol consumption during the COVID-19 pandemic.

DESIGN

An individual-level simulation model of the long-term drinking patterns for people with life-time AUD was used to simulate 10 000 individuals and project model outcomes to the estimated 25.9 million current drinkers with life-time AUD in the United States. The model considered three scenarios: (1) no change (counterfactual for comparison); (2) increased drinking levels persist for 1 year ('increase-1') and (3) increased drinking levels persist for 5 years ('increase-5').

SETTING

United States.

PARTICIPANTS

Current drinkers with life-time AUD.

MEASUREMENTS

Life expectancy [life-years (LYs)], quality-adjusted life-years (QALYs), alcohol-related hospitalizations and associated hospitalization costs and alcohol-related deaths, during a 5-year period.

FINDINGS

Short-term increases in alcohol consumption (increase-1 scenario) resulted in a loss of 79 000 [95% uncertainty interval (UI]) 26 000-201 000] LYs, a loss of 332 000 (104 000-604 000) QALYs and 295 000 (82 000-501 000) more alcohol-related hospitalizations, costing an additional $5.4 billion ($1.5-9.3 billion) over 5 years. Hospitalizations for cirrhosis of the liver accounted for approximately $3.0 billion ($0.9-4.8 billion) in hospitalization costs, more than half the increase across all alcohol-related conditions. Health and cost impacts were more pronounced for older age groups (51+), women and non-Hispanic black individuals. Increasing the duration of pandemic-driven increases in alcohol consumption in the increase-5 scenario resulted in larger impacts.

CONCLUSIONS

Simulations show that if the increase in alcohol consumption observed in the United States in the first year of the pandemic continues, alcohol-related mortality, morbidity and associated costs will increase substantially over the next 5 years.

Effect of increased alcohol consumption during COVID-19 pandemic on alcohol-associated liver disease: A modeling study

BACKGROUND AND AIMS

Alcohol consumption increased during the COVID-19 pandemic in 2020 in the United States. We projected the effect of increased alcohol consumption on alcohol-associated liver disease (ALD) and mortality.

APPROACH AND RESULTS

We extended a previously validated microsimulation model that estimated the short- and long-term effect of increased drinking during the COVID-19 pandemic in individuals in the United States born between 1920 and 2012. We modeled short- and long-term outcomes of current drinking patterns during COVID-19 (status quo) using survey data of changes in alcohol consumption in a nationally representative sample between February and November 2020. We compared these outcomes with a counterfactual scenario wherein no COVID-19 occurs and drinking patterns do not change. One-year increase in alcohol consumption during the COVID-19 pandemic is estimated to result in 8000 (95% uncertainty interval [UI], 7500-8600) additional ALD-related deaths, 18,700 (95% UI, 17,600-19,900) cases of decompensated cirrhosis, and 1000 (95% UI, 1000-1100) cases of HCC, and 8.9 million disability-adjusted life years between 2020 and 2040. Between 2020 and 2023, alcohol consumption changes due to COVID-19 will lead to 100 (100-200) additional deaths and 2800 (2700-2900) additional decompensated cirrhosis cases. A sustained increase in alcohol consumption for more than 1 year could result in additional morbidity and mortality.

CONCLUSIONS

A short-term increase in alcohol consumption during the COVID-19 pandemic can substantially increase long-term ALD-related morbidity and mortality. Our findings highlight the need for individuals and policymakers to make informed decisions to mitigate the impact of high-risk alcohol drinking in the United States.