Predictor Of Health Care Costs Research Articles

Machine learning models for assessing risk factors affecting health care costs: 12-month exercise-based cardiac rehabilitation.

Exercise-based cardiac rehabilitation (ECR) has proven to be effective and cost-effective dominant treatment option in health care. However, the contribution of well-known risk factors for prognosis of coronary artery disease (CAD) to predict health care costs is not well recognized. Since machine learning (ML) applications are rapidly giving new opportunities to assist health care professionals' work, we used selected ML tools to assess the predictive value of defined risk factors for health care costs during 12-month ECR in patients with CAD. The data for analysis was available from a total of 71 patients referred to Oulu University Hospital, Finland, due to an acute coronary syndrome (ACS) event (75% men, age 61 ± 12 years, BMI 27 ± 4 kg/m2, ejection fraction 62 ± 8, 89% have beta-blocker medication). Risk factors were assessed at the hospital immediately after the cardiac event, and health care costs for all reasons were collected from patient registers over a year. ECR was programmed in accordance with international guidelines. Risk analysis algorithms (cross-decomposition algorithms) were employed to rank risk factors based on variances in their effects. Regression analysis was used to determine the accounting value of risk factors by entering first the risk factor with the highest degree of explanation into the model. After that, the next most potent risk factor explaining costs was added to the model one by one (13 forecast models in total). The ECR group used health care services during the year at an average of 1,624 ± 2,139€ per patient. Diabetes exhibited the strongest correlation with health care expenses (r = 0.406), accounting for 16% of the total costs (p < 0.001). When the next two ranked markers (body mass index; r = 0.171 and systolic blood pressure; r = - 0.162, respectively) were added to the model, the predictive value was 18% for the costs (p = 0.004). The depression scale had the weakest independent explanation rate of all 13 risk factors (explanation value 0.1%, r = 0.029, p = 0.811). Presence of diabetes is the primary reason forecasting health care costs in 12-month ECR intervention among ACS patients. The ML tools may help decision-making when planning the optimal allocation of health care resources.

Prevalence and impact of antipsychotic discontinuation among commercially-insured patients with bipolar disorder

BackgroundAntipsychotic discontinuation is common among patients with bipolar disorder, especially when psychotic symptoms are remitted. This analysis describes the prevalence, predictors, and economic impact of antipsychotic discontinuation among patients with bipolar disorder.MethodsA retrospective, observational study was conducted using administrative claims data in the IBM MarketScan Commercial Database. Patients with ≥1 claim with a diagnosis for bipolar disorder (manic or mixed) and newly-initiating antipsychotic therapy between 1 January 2011 and 30 June 2016 were included. Baseline characteristics were assessed in the 12 months prior to the initiation. Outcomes were assessed during a 24-month follow-up. Discontinuation of antipsychotic therapy was utilized as a predictor of healthcare costs in models adjusted for baseline characteristics. Using limited set of variables in the claims database, predictors of discontinuation were also assessed.ResultsA total of 18,259 commercially-insured patients were identified as initiators of antipsychotics. Common comorbidities among the cohorts included major depressive disorder and dyslipidemia. Discontinuation was very common among these patients (85%). Major depressive disorder, drug abuse, and other substance abuse/dependency were predictive of discontinuation. Controlling for differences in baseline characteristics, discontinuation was associated with 33% higher inpatient and emergency visit costs (p <.001) among those using these services, and 24% higher total healthcare costs (p <.001) for the overall cohort.ConclusionsMost patients with bipolar mania or mixed states discontinue antipsychotic treatment in less than 2 years. Antipsychotic discontinuation contributes to excess healthcare costs. Future research focusing on the reasons for discontinuation and tailoring disease management based on comorbidities may inform adherence improvement initiatives.

Machine learning models in predicting health care costs in patients with a recent acute coronary syndrome: A prospective pilot study

Health care budgets are limited, requiring the optimal use of resources. Machine learning (ML) methods may have an enormous potential for effective use of health care resources. We assessed the applicability of selected ML tools to evaluate the contribution of known risk markers for prognosis of coronary artery disease to predict health care costs for all reasons in patients with a recent acute coronary syndrome (n = 65, aged 65 ± 9 years) for 1-year follow-up. Risk markers were assessed at baseline, and health care costs were collected from electronic health registries. The Cross-decomposition algorithms were used to rank the considered risk markers based on their impacts on variances. Then regression analysis was performed to predict costs by entering the first top-ranking risk marker and adding the next-best markers, one by one, to build up altogether 13 predictive models. The average annual health care costs were €2601 ± €5378 per patient. The Depression Scale showed the highest predictive value (r = 0.395), accounting for 16% of the costs (P = .001). When the next 2 ranked markers (LDL cholesterol, r = 0.230; and left ventricular ejection fraction, r = -0.227, respectively) were added to the model, the predictive value was 24% for the costs (P = .001). Higher depression score is the primary variable forecasting health care costs in 1-year follow-up among acute coronary syndrome patients. The ML tools may help decision-making when planning optimal utilization of treatment strategies.

Function and Frailty: Value Added in Medicare

EditorialsApril 2023Function and Frailty: Value Added in MedicareJulie P.W. Bynum, MD, MPHJulie P.W. Bynum, MD, MPHDepartment of Internal Medicine, University of Michigan Medical School, Ann Arbor, MichiganSearch for more papers by this authorAuthor, Article, and Disclosure Informationhttps://doi.org/10.7326/M23-0563 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail With Medicare’s nearly $690 billion dollars of spending in 2021 (1), getting payment right is a high-stakes business. Risk adjustment, the process of assigning a score to the health status of individuals to predict health care costs, is a central component used to determine payment in value-based models of care and Medicare Advantage. The same measures are also used by health systems to segment their population into distinct groups based on the type of care they need. Typically, morbidity, measured by number and types of diseases, is used for risk measures. Because most chronic diseases are age related, older adults, ...

Association of the Kidney Failure Risk Equation With High Health Care Costs

The Kidney Failure Risk Equations (KFRE) are accurate and validated to predict the risk of kidney failure in individuals with chronic kidney disease (CKD), but their potential to predict health care costs in the US health care system is unknown. We assessed the association of kidney failure risk from the 4-variable and 8-variable 2-year KFRE models with monthly health care costs in US patients with CKD stages G3 and G4. This was an ancillary study to a larger observational, retrospective cohort study examining the association between serum bicarbonate and adverse kidney outcomes. Monthly medical costs were calculated from individual health care insurance claims. Generalized linear regression models were used to examine the association of KFRE score with health care costs. A total of 1721 patients qualified for the study (1475 and 246 with CKD stages G3 and G4, respectively). For 8-variable KFRE, each 1% (absolute) increase in risk was associated with 13.5% (P< 0.001) and 4.1% (P< 0.001) higher monthly costs for patients with CKD stage G3 and G4, respectively. For 4-variable KFRE, a 1% increase in risk was associated with 6.7% (P= 0.016) and 2.9% (P= 0.014) increase in monthly costs for patients with CKD stage G3 and G4, respectively. Higher risks of kidney failure as predicted by the 4-variable or 8-variable KFRE were associated with higher 2-year medical costs for patients with CKD stages G3 and G4. The KFRE may be a useful tool to anticipate medical costs and target cost-reducing interventions for patients at risk of kidney failure.

Prediction of total healthcare cost following total shoulder arthroplasty utilizing machine learning.

Given the increase in demand in treatment of glenohumeral arthritis with anatomic total (aTSA) and reverse shoulder arthroplasty (RTSA), it is imperative to improve quality of patient care while controlling costs as private and federal insurers continue its gradual transition toward bundled payment models. Big data analytics with machine learning shows promise in predicting health care costs. This is significant as cost prediction may help control cost by enabling health care systems to appropriately allocate resources that help mitigate the cause of increased cost. The Nationwide Readmissions Database (NRD) was accessed in 2018. The database was queried for all primary aTSA and RTSA by International Classification of Diseases, Tenth Revision (ICD-10) procedure codes: 0RRJ0JZ and 0RRK0JZ for aTSA and 0RRK00Z and 0RRJ00Z for RTSA. Procedures were categorized by diagnoses: osteoarthritis (OA), rheumatoid arthritis (RA), avascular necrosis (AVN), fracture, and rotator cuff arthropathy (RCA). Costs were calculated by utilizing the total hospital charge and each hospital's cost-to-charge ratio. Hospital characteristics were included, such as volume of procedures performed by the respective hospital for the calendar year and wage index, which represents the relative average hospital wage for the respective geographic area. Unplanned readmissions within 90 days were calculated using unique patient identifiers, and cost of readmissions was added to the total admission cost to represent the short-term perioperative health care cost. Machine learning algorithms were used to predict patients with immediate postoperative admission costs greater than 1 standard deviation from the mean, and readmissions. A total of 49,354 patients were isolated for analysis, with an average patient age of 69.9±9.6 years. The average perioperative cost of care was $18,843±$10,165. In total, there were 4279 all-cause readmissions, resulting in an average cost of $13,871.00±$14,301.06 per readmission. Wage index, hospital volume, patient age, readmissions, and diagnosis-related group severity were the factors most correlated with the total cost of care. The logistic regression and random forest algorithms were equivalent in predicting the total cost of care (area under the receiver operating characteristic curve=0.83). After shoulder arthroplasty, there is significant variability in cumulative hospital costs, and this is largely affected by readmissions. Hospital characteristics, such as geographic area and volume, are key determinants of overall health care cost. When accounting for this, machine learning algorithms may predict cases with high likelihood of increased resource utilization and/or readmission.

Prediction of future healthcare expenses of patients from chest radiographs using deep learning: a pilot study

Our objective was to develop deep learning models with chest radiograph data to predict healthcare costs and classify top-50% spenders. 21,872 frontal chest radiographs were retrospectively collected from 19,524 patients with at least 1-year spending data. Among the patients, 11,003 patients had 3 years of cost data, and 1678 patients had 5 years of cost data. Model performances were measured with area under the receiver operating characteristic curve (ROC-AUC) for classification of top-50% spenders and Spearman ρ for prediction of healthcare cost. The best model predicting 1-year (N = 21,872) expenditure achieved ROC-AUC of 0.806 [95% CI 0.793–0.819] for top-50% spender classification and ρ of 0.561 [0.536–0.586] for regression. Similarly, for predicting 3-year (N = 12,395) expenditure, ROC-AUC of 0.771 [0.750–0.794] and ρ of 0.524 [0.489–0.559]; for predicting 5-year (N = 1779) expenditure ROC-AUC of 0.729 [0.667–0.729] and ρ of 0.424 [0.324–0.529]. Our deep learning model demonstrated the feasibility of predicting health care expenditure as well as classifying top 50% healthcare spenders at 1, 3, and 5 year(s), implying the feasibility of combining deep learning with information-rich imaging data to uncover hidden associations that may allude to physicians. Such a model can be a starting point of making an accurate budget in reimbursement models in healthcare industries.

Multivariable prediction models for health care spending using machine learning: a protocol of a systematic review

BackgroundWith rising cost pressures on health care systems, machine-learning (ML)-based algorithms are increasingly used to predict health care costs. Despite their potential advantages, the successful implementation of these methods could be undermined by biases introduced in the design, conduct, or analysis of studies seeking to develop and/or validate ML models. The utility of such models may also be negatively affected by poor reporting of these studies. In this systematic review, we aim to evaluate the reporting quality, methodological characteristics, and risk of bias of ML-based prediction models for individual-level health care spending.MethodsWe will systematically search PubMed and Embase to identify studies developing, updating, or validating ML-based models to predict an individual’s health care spending for any medical condition, over any time period, and in any setting. We will exclude prediction models of aggregate-level health care spending, models used to infer causality, models using radiomics or speech parameters, models of non-clinically validated predictors (e.g., genomics), and cost-effectiveness analyses without predicting individual-level health care spending. We will extract data based on the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies (CHARMS), previously published research, and relevant recommendations. We will assess the adherence of ML-based studies to the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement and examine the inclusion of transparency and reproducibility indicators (e.g. statements on data sharing). To assess the risk of bias, we will apply the Prediction model Risk Of Bias Assessment Tool (PROBAST). Findings will be stratified by study design, ML methods used, population characteristics, and medical field.DiscussionOur systematic review will appraise the quality, reporting, and risk of bias of ML-based models for individualized health care cost prediction. This review will provide an overview of the available models and give insights into the strengths and limitations of using ML methods for the prediction of health spending.

Assessing the Added Value of Blood Pressure Information Derived from Electronic Health Records in Predicting Health Care Cost and Utilization.

Health care providers are increasingly using clinical measures derived from electronic health records (EHRs) for risk stratification and predictive modeling. EHR-specific data elements such as prescriptions, laboratory results, and vital signs have been shown to improve risk prediction models. In this study, the value of EHR-based blood pressure (BP) values was assessed in predicting health care costs (ie, total, medical, and pharmacy) and key utilization end points (ie, hospitalization, emergency department use, and being among the highest utilizers). The study population included 37,451 patients of a large integrated delivery system in the mid-western United States with complete EHR data files, who were 18-64 years old, had continuous insurance at an affiliated health plan, and had eligible BP records. Both EHRs and insurance claims of the study population were used to extract the predictors (ie, demographics, diagnosis, and BP values) and outcomes (ie, costs and utilizations). Predictors were extracted from 2012 data, whereas concurrent and prospective outcomes were extracted from 2012 to 2013 data. Three base models (BMs) were constructed to predict each of the outcomes. The first BM no. 1 used demographics. The second BM no. 2 added the Charlson comorbidity index to BM no. 1, whereas the third BM no. 3 added the Adjusted Clinical Group Dx-PM case-mix score to BM no. 1. BP was specified as means, ranges, and classes. Adding BP ranges to BM no. 1 and BM no. 2 showed the greatest improvements when predicting costs and utilization. More specifically, adjusted R2 and area under the curve of BM no. 2 improved by 32.9% and 14.1% when BP ranges were added to predict concurrent total cost and hospitalization, respectively. The effect of BP measures on improving the risk stratification models was diminished when predicting prospective outcomes after adding the measures to BM no. 3 (ie, the more comprehensive diagnostic model), specifically when represented as BP means. Given the increasing availability of BP information, this research suggests that these data should be integrated into provider-based population health analytic activities. Future research should focus on subpopulations that benefit the most from incorporating vital signs such as BP measures in risk stratification models.

The Kidney Failure Risk Equation as a Predictor of Healthcare Costs

The Kidney Failure Risk Equation as a Predictor of Healthcare Costs

Are Lipid Profiles and Circulating miRNAs Able to Predict Health Costs?—The Health Survey of São Paulo, Brazil (ISA-Capital)

Abstract Objectives To evaluate associations between circulating miRNAs and lipid profiles with household income, medication and health care expenditures per capita in residents of Sao Paulo municipality, Brazil. Methods As part of the Health Survey of São Paulo (ISA-Capital), a population-based study, we collected blood samples of 573 healthy individuals aged 12 years or more to measure the lipid profile and levels of 21 circulating miRNA, which have been previously implicated in dyslipidemia, obesity, insulin resistance and inflammation. Data on household income, medication and health care expenditures per capita was assessed by a questionnaire administered in the individual's households. Considering that the variables were not normally distributed, their correlations were estimated using the Spearman's coefficient test under a 0.05 significance level. Data analysis was performed using the software R version 4.0.2. Results The correlation coefficients between circulating miRNAs and plasma lipid profile with household income, medication and health care expenditures per capita were not statistically significant (P-value &amp;gt; 0.05), and their point estimates were very close to zero. These results show that circulating miRNAs and lipid profile are not informative for the prediction of expenditures on health care or drug intake and household income per capita in Sao Paulo residents. Conclusions This study showed that there is no relationship between the dyad miRNA-lipid levels and health care expenditures. Further strategies to identify potential predictors of health care costs, especially other epigenetic and genetic markers related to dyslipidemia are necessary and might help decreasing the economic burden in the health system. Funding Sources São Paulo Municipal Health Department, Research Support Foundation of the State of São Paulo (FAPESP), and National Council for Scientific and Technological Development (CNPq)

PMH54 Impact of Bipolar Disorder Treatment Discontinuation on Healthcare Utilization and Costs in the Real-World Setting

To describe the real-world antipsychotic (AP) treatment discontinuation patterns and the association between treatment discontinuation and healthcare costs among bipolar disorder patients. A retrospective observational study was conducted using administrative claims data the IBM® MarketScan Commercial Database. Patients with at least 1 medical claim with a diagnosis of bipolar disorder (manic or mixed sub-type) and newly initiating AP therapy between 1/1/11-6/30/16 were included. Baseline patient characteristics were assessed in the 12 months prior to the AP initiation. Total healthcare costs and healthcare utilization were assessed during a 24-month follow-up. Discontinuation of index AP therapy, defined as a gap of at least 60 days without the medication on-hand, was utilized as a predictor of healthcare costs and utilization outcomes in generalized linear and logistic regression models adjusted for baseline characteristics. Predictors of discontinuation were also assessed via logistic regression. A total of 18,259 bipolar disorder patients were identified as new initiators of APs. The cohort was 61% female and had an average age of 39 years. The most commonly initiated AP was quetiapine (33.6%), followed by aripiprazole (29.3%). A total of 15,460 patients (85%) discontinued their index regimen. Patients with baseline substance use disorders, depression, and other anxiety disorders were significantly more likely to discontinue. More than half of the cohort utilized emergency room services (53%) and 33% had at least one inpatient admission during the follow-up. The median total healthcare costs were $24,861 during the 24-month follow-up. Controlling for differences in baseline characteristics, discontinuation of AP therapy was associated with 24% higher total healthcare costs (Cost ratio: 1.24; p<0.001) and 33% higher inpatient+ER costs (1.33; p<0.001). Discontinuation of AP therapy among bipolar disorder patients is high and contributes to excess healthcare costs and high-cost services. Future research focusing on the reasons for discontinuation may inform medication adherence initiatives.

Use of Data-Driven Methods to Predict Long-term Patterns of Health Care Spending for Medicare Patients

Current approaches to predicting health care costs generally rely on a single composite value of spending and focus on short time horizons. By contrast, examining patients' spending patterns using dynamic measures applied over longer periods may better identify patients with different spending and help target interventions to those with the greatest need. To classify patients by their long-term, dynamic health care spending patterns using a data-driven approach and assess the ability to predict spending patterns, particularly using characteristics that are potentially modifiable through intervention. This cohort study used a retrospective cohort design from a random nationwide sample of Medicare fee-for-service administrative claims data to identify beneficiaries aged 65 years or older with continuous eligibility from 2011 to 2013. Statistical analysis was performed from August 2018 to December 2019. Group-based trajectory modeling was applied to the claims data to classify the Medicare beneficiaries by their total health care spending patterns over a 2-year period. The ability to predict membership in each trajectory spending group was assessed using generalized boosted regression, a data mining approach to model building and prediction, with split-sample validation. Models were estimated using (1) prior-year predictors and (2) prior-year predictors potentially modifiable through intervention measured in the claims data. These models were evaluated using validated C-statistics. The relative influence of individual predictors in the models was evaluated. Among the 329 476 beneficiaries, the mean (SD) age was 76.0 (7.2) years and 190 346 (57.8%) were female. This final 5-group model included a minimal-user group (group 1, 37 572 individuals [11.4%]), a low-cost group (group 2, 48 575 individuals [14.7%]), a rising-cost group (group 3, 24 736 individuals [7.5%]), a moderate-cost group (group 4, 83 338 individuals [25.3%]), and a high-cost group (group 5, 135 255 individuals [41.2%]). Potentially modifiable characteristics strongly predicted these patterns (C-statistics range: 0.68-0.94). For groups with progressively increasing spending in particular, the most influential factors were number of medications (relative influence: 29.2), number of office visits (relative influence: 30.3), and mean medication adherence (relative influence: 33.6). Using a data-driven approach, distinct spending patterns were identified with high accuracy. The potentially modifiable predictors of membership in the rising-cost group represent important levers for early interventions that may prevent later spending increases. This approach could be adapted by organizations to target quality improvement interventions, particularly because numerous health care organizations are increasingly using these routinely collected data.

Impact of Area Deprivation Index on the Performance of Claims-Based Risk-Adjustment Models in Predicting Health Care Costs and Utilization.

Traditionally, risk-adjustment models do not address the characteristics of minority populations, such as race or socioeconomic status. This study aimed to evaluate the added value of place-based social determinants on risk-adjustment models in explaining health care costs and utilization. Statewide commercial claims from the Maryland Medical Care Database were used, including 1,150,984 Maryland residents aged 18 to 63 with ≥6 months enrollment in 2013 and 2014. Area Deprivation Index (ADI) was assigned to individuals through zip code. The authors examined the addition of ADI to predictive models of concurrent and prospective costs and utilization; linear regression was adopted for costs and logistic regression for utilization markers. Performance measures included R2 for costs (total, pharmacy, and medical costs) and the area under the curve (AUC) for utilization (being top 5% top users, having any hospitalization, having any emergency room [ER] visit, having any avoidable ER visit, and having any readmission). All performance measures were derived from the bootstrapping analysis with 200 iterations. Study subjects were ∼48% male with a mean age of ∼41 years. Adding ADI to the demographics or claims-based models generally did not improve performance except in predicting the probability of having any ER or any avoidable ER visit; for example, AUC of avoidable ER visits increased significantly from .610 to .613 when using ADI rank deciles in claims-based models. Future research should focus on patients with a higher need for social services, assess more granular place-based determinants (eg, Census block group), and evaluate the added value of individual social variables.

Using biomarkers to predict healthcare costs: Evidence from a UK household panel

We investigate the extent to which healthcare service utilisation and costs can be predicted from biomarkers, using the UK Understanding Society panel. We use a sample of 2314 adults who reported no history of diagnosed long-lasting health conditions at baseline (2010/11), when biomarkers were collected. Five years later, their GP, outpatient (OP) and inpatient (IP) utilisation was observed. We develop an econometric technique for count data observed within ranges and a method of combining administrative reference cost data with the survey data without exact individual-level matching. Our composite biomarker index (allostatic load) is a powerful predictor of costs: for those with a baseline allostatic load of at least one standard deviation (1-s.d.) above mean, a 1-s.d. reduction reduces GP, OP and IP costs by around 18%.

Prediction of Health Care Costs by Dental Health Care Costs and Periodontal Status

Reducing heath care costs is an important issue in Japan. The aim of this study was to analyze the contribution of oral health to health care costs and to predict health care costs by statistical modeling. Data from 46 individuals (29 men and 17 women; mean age of 44.6 ± 1.7 years) on health care costs, dental health care costs, and the results of the salivary levels of lactate dehydrogenase (LD) over two years were provided by the association. Multilayer perceptron neural networks were applied to predict the health care costs from data from the previous year and included health care costs, dental health care costs, and salivary levels of LD. Nonlinear relationships were observed between medical health care costs, dental health care costs, and periodontal conditions. The health care costs from the previous year were the most important predictor of health care costs. The simulation results showed that health care costs decreased with the increase in dental health care costs from the previous year. Health care costs increased with increasing salivary levels of LD from the previous year. Improvements in periodontal conditions and dental health care may play some roles in reducing health care costs.

Abdominal aortic calcification (AAC) and ankle-brachial index (ABI) predict health care costs and utilization in older men, independent of prevalent clinical cardiovascular disease and each other

Background and aimsAbdominal aortic calcification (AAC) and low ankle-brachial index (ABI) are markers of multisite atherosclerosis. We sought to estimate their associations in older men with health care costs and utilization adjusted for each other, and after accounting for CVD risk factors and prevalent CVD diagnoses. MethodsThis was an observational cohort study of 2393 community-dwelling men (mean age 73.6 years) enrolled in the Osteoporotic Fractures in Men (MrOS) study and U.S. Medicare Fee for Service (FFS). AAC was scored on baseline lateral lumbar spine X-rays using a 24-point scale. ABI was measured as the lowest ratio of arm to right or left ankle blood pressure. Health care costs, hospital stays, and SNF stays were identified from Medicare FFS claims over 36 months following the baseline visit. ResultsMen with AAC score ≥9 (n = 519 [21.7% of analytic cohort]) had higher annualized total health care costs of $1473 (95% C.I. 293, 2654, 2017 U S. dollars) compared to those with AAC score 0–1, after multivariable adjustment. Men with ABI <0.90 (n = 154 [6.4% of analytic cohort]) had higher annualized total health care costs of $2705 (95% CI 634, 4776) compared to men with normal ABI (≥0.9 and < 1.4), after multivariable adjustment. ConclusionsHigh levels of AAC and low ABI in older men are associated with higher subsequent health care costs, after accounting for clinical CVD risk factors, prevalent CVD diagnoses, and each other. Further investigations of whether preventing progression of peripheral vascular disease and calcification reduces subsequent health care costs are warranted.

Social Restrictions Versus Testing Campaigns in the COVID-19 Crisis: A Predictive Model Based on the Spanish Case

Background: The global COVID-19 spread has forced countries to implement different nonpharmacological interventions (NPI) to preserve their health systems. Spain was one of the first and most severely impacted countries, both clinically and economically. In an effort to support policy decision-making, we aimed to assess the impacts of different NPI on COVID-19 epidemiology, healthcare costs and Gross Domestic Product (GDP). Methods: A modified Susceptible-Exposed-Infectious-Removed epidemiological model was created to simulate the pandemic evolution. Its output was then used to populate an economic model to predict healthcare costs and Q4 2020 GDP decline, estimated through a multiple linear regression model based on NPIs and GDP decline from 42 countries during Q2 2020. Thirteen scenarios combining different NPI were consecutively simulated in the epidemiological and economic models. Findings: With no changes in testing rate or stringency (lockdown), 2·38 million new exposed cases in the subsequent 90 days were estimated in Spain. Both increased testing and stringency could reduce cases, hospitalizations and deaths. While policies based on increased testing rates led to higher healthcare costs, increased stringency led to greater GDP declines with differences of up to 4·4% points. Increased test sensitivity may lead to a reduction of cases, hospitalizations and deaths and to the implementation of pooling techniques that can increase throughput testing capacity. Interpretation: Alternative strategies for controlling the COVID-19 spread lead to differing economic outcomes. Decision-makers may utilize this tool to identify the most suitable strategy considering epidemiological and economic outcomes. Funding: This study is funded by Grifols S.A.Declaration of Interests: EV is employee from Grifols S.A. FJC, JS-R-M, RR, PB and MMC received honoraria from Grifols S.A. for their participation in the study as subject experts, within the Fair Market Value range. DC and AK are employees from IQVIA, which was the Clinical Research Organization hired by Grifols S.A. for the implementation of this study and medical writing support. MCR is employee from Grifols S.S.N.A.

Predicting the Cost of Health Care Services: A Comparison of Case-mix Systems and Comorbidity Indices That Use Administrative Data.

Case-mix systems and comorbidity indices aggregate clinical information about patients over time and are used to characterize need for health care services. These tools were validated for their original purpose, but those purposes are varied, and they have not been compared directly in the context of predicting costs of health care services. To compare predictions of next-year health care service costs across 4 tools, including: the Johns Hopkins Adjusted Clinical Groups (ACG), the Elixhauser Comorbidity Index, Charlson-Deyo Comorbidity Index, and the Canadian Institute for Health Information (CIHI) population grouper. British Columbia administrative data from fiscal years 2012-2013 were used to generate case-mix variables and the comorbidity indices. Outcome variables include next-year (2013-2014) total, physician, acute care, and pharmaceutical costs, Outcomes were modeled using 2-part models. Performance was compared using adjusted R, root mean squared error, and mean absolute error using the predicted and the actual next-year cost. Models including the CIHI grouper (239 conditions) and ACG system had similar performance in most cost categories and slightly better fit than Charlson Comorbidity Index (CCI) and Elixhauser Comorbidity Index (ECI). Adding a dummy variable for nonusers in the models for CCI and ECI increased R values slightly. All these systems have empirical support for use in predicting health care costs, despite in some cases being developed for other purposes. No system is particularly effective at predicting next-year acute care cost, likely because acute events are often by definition unexpected. The freely available ECI and CCI comorbidity indices implemented using the highest-performing methods developed here may be a good choice in many circumstances.

94. Cost analysis of primary single-level lumbar discectomies

BACKGROUND CONTEXT Improving value in surgical care requires a detailed understanding of the true costs of surgical and operating room resources. PURPOSE In this study, we sought to identify predictors of health care costs and operating room resource use for patients undergoing initial single-level discectomies using a unique institutional cost dataset from the value driven outcomes (VDO) program developed by the University of Utah. STUDY DESIGN/SETTING A retrospective cost analysis of primary single-level lumbar discectomies. PATIENT SAMPLE A total of 622 patients undergoing initial single-level lumbar discectomy between January 2014 to April 2016 at the University of Utah Orthopaedic Center. OUTCOME MEASURES Health care costs, clinical length of stay, and operating room time. METHODS Univariate and multivariate generalized linear models (GLM) performed to identify predictors of health care costs, clinical length of stay (LOS, days) and operating room time (OR, minutes) using a unique costing tool. Modified park test procedure was implemented to determine GLM error distribution specifications and standard errors were clustered by provider. Cost outcomes were normalized using mean costs for a patient with normal body mass index (BMI) and a healthy American Society of Anesthesiologists (ASA) classification system. Average marginal effects in multivariate analysis were reported as percentage of normalized costs. RESULTS Advanced age, male gender, Hispanic, Black, unemployment, and being retired were significant positive predictors of costs in univariate analysis. Asian, Native American, outpatient procedures, and being a student were significant negative predictors of costs. Univariate analysis also showed that obesity, higher ASA class and insurance status were also positive and significant predictors of costs. In multivariate analysis, we found that obesity led to higher average marginal total direct (9%), total facility (15%), and facility OR costs (22%), as well as an average marginal effect of 24 more OR minutes per surgery. Conversely, being underweight was associated with lower average marginal total direct (-23%), total non-facility (-54%), total facility (-8%), facility OR costs (-15%), as well as lower clinical LOS (-.42 days), and 18 less OR minutes. While being overweight was not significantly associated with greater total direct costs, it was associated with higher total facility (8%), and facility OR costs (12%), with an average marginal effect of 11 more OR minutes per surgery. Age was associated with a longer LOS but not with OR costs in multivariate analysis. As expected, outpatient surgical costs, LOS and OR time were significantly lower than inpatient procedures. Patients with incapacitating disease ASA Class had the highest predicted OR time in GLM estimates with an average marginal effect of 28 minutes, while severe systematic disease was associated with greater total and nonfacility costs. In addition, Medicare patients had higher facility costs (14%) compared to privately insured patients. CONCLUSIONS Although many of our observed variables were significant predictors of costs in univariate analysis, multivariate GLM estimates showed that the main variables determining cost and operating room time and resource use were obesity (BMI), general health (ASA class), Medicare status and advanced age, and whether the surgery was an outpatient or inpatient procedure. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs.