Data transformations to improve the performance of health plan payment methods.

The conventional method for developing health care plan payment systems uses observed data to study alternative algorithms and set incentives for the health care system. In this paper, we take a different approach and transform the input data rather than the algorithm, so that the data used reflect the desired spending levels rather than the observed spending levels. We present a general economic model that incorporates the previously overlooked two-way relationship between health plan payment and insurer actions. We then demonstrate our systematic approach for data transformations in two Medicare applications: underprovision of care for individuals with chronic illnesses and health care disparities by geographic income levels. Empirically comparing our method to two other common approaches shows that the "side effects" of these approaches vary by context, and that data transformation is an effective tool for addressing misallocations in individual health insurance markets.

Classifying Stage IV Lung Cancer From Health Care Claims: A Comparison of Multiple Analytic Approaches.

PURPOSE: Cancer stage is a key determinant of outcomes; however, stage is not available in claims-based data sources used for real-world evaluations. We compare multiple methods for classifying lung cancer stage from claims data. METHODS: Our study used the linked SEER-Medicare data. The patient samples included fee-for-service Medicare beneficiaries diagnosed with lung cancer from 2010 to 2011 (development cohort) and 2012 to 2013 (validation cohort) who received chemotherapy. Classification algorithms considered Medicare Part A and B claims for care in the 3 months before and after chemotherapy initiation. We developed a clinical algorithm to predict stage IV (v I to III) cancer on the basis of treatment patterns (surgery, radiotherapy, chemotherapy). We also considered an ensemble of claims-based machine learning algorithms. Classification methods were trained in the development cohort, and performance was measured in both cohorts. The SEER data were the gold standard for cancer stage. RESULTS: Development and validation cohorts included 14,760 and 14,620 patients with lung cancer, respectively. Validation analyses assessed clinical, random forest, and simple logistic regression algorithms. The best performing classifier within the development cohort was the random forests, but this performance was not replicated in validation analysis. Logistic regression had stable performance across cohorts. Compared with the clinical algorithm, the 14-variable logistic regression algorithm demonstrated higher accuracy in both the development (77% v 71%) and validation cohorts (77% v 73%), with improved specificity for stage IV disease. CONCLUSION: Machine learning algorithms have potential to improve lung cancer stage classification but may be prone to overfitting. Use of ensembles, cross-validation, and external validation can aid generalizability. Degradation of accuracy between development and validation cohorts suggests the need for caution in implementing machine learning in research or care delivery.

Sample Selection for Medicare Risk Adjustment Due to Systematically Missing Data.

OBJECTIVE: To assess the issue of nonrepresentative sampling in Medicare Advantage (MA) risk adjustment. DATA SOURCES: Medicare enrollment and claims data from 2008 to 2011. DATA EXTRACTION: Risk adjustment predictor variables were created from 2008 to 2010 Part A and B claims and the Medicare Beneficiary Summary File. Spending is based on 2009-2011 Part A and B, Durable Medical Equipment, and Home Health Agency claims files. STUDY DESIGN: A propensity-score matched sample of Traditional Medicare (TM) beneficiaries who resembled MA enrollees was created. Risk adjustment formulas were estimated using multiple techniques, and performance was evaluated based on R2 , predictive ratios, and formula coefficients in the matched sample and a random sample of TM beneficiaries. PRINCIPAL FINDINGS: Matching improved balance on observables, but performance metrics were similar when comparing risk adjustment formula results fit on and evaluated in the matched sample versus fit on the random sample and evaluated in the matched sample. CONCLUSIONS: Fitting MA risk adjustment formulas on a random sample versus a matched sample yields little difference in MA plan payments. This does not rule out potential improvements via the matching method should reliable MA encounter data and additional variables become available for risk adjustment.

Computational health economics for identification of unprofitable health care enrollees.

Health insurers may attempt to design their health plans to attract profitable enrollees while deterring unprofitable ones. Such insurers would not be delivering socially efficient levels of care by providing health plans that maximize societal benefit, but rather intentionally distorting plan benefits to avoid high-cost enrollees, potentially to the detriment of health and efficiency. In this work, we focus on a specific component of health plan design at risk for health insurer distortion in the Health Insurance Marketplaces: the prescription drug formulary. We introduce an ensembled machine learning function to determine whether drug utilization variables are predictive of a new measure of enrollee unprofitability we derive, and thus vulnerable to distortions by insurers. Our implementation also contains a unique application-specific variable selection tool. This study demonstrates that super learning is effective in extracting the relevant signal for this prediction problem, and that a small number of drug variables can be used to identify unprofitable enrollees. The results are both encouraging and concerning. While risk adjustment appears to have been reasonably successful at weakening the relationship between therapeutic-class-specific drug utilization and unprofitability, some classes remain predictive of insurer losses. The vulnerable enrollees whose prescription drug regimens include drugs in these classes may need special protection from regulators in health insurance market design.

Classifying Lung Cancer Severity with Ensemble Machine Learning in Health Care Claims Data.

Research in oncology quality of care and health outcomes has been limited by the difficulty of identifying cancer stage in health care claims data. Using linked cancer registry and Medicare claims data, we develop a tool for classifying lung cancer patients receiving chemotherapy into early vs. late stage cancer by (i) deploying ensemble machine learning for prediction, (ii) establishing a set of classification rules for the predicted probabilities, and (iii) considering an augmented set of administrative claims data. We find our ensemble machine learning algorithm with a classification rule defined by the median substantially outperforms an existing clinical decision tree for this problem, yielding full sample performance of 93% sensitivity, 92% specificity, and 93% accuracy. This work has the potential for broad applicability as provider organizations, payers, and policy makers seek to measure quality and outcomes of cancer care and improve on risk adjustment methods.

Accountability across the Continuum: The Participation of Postacute Care Providers in Accountable Care Organizations.

OBJECTIVE: To examine the extent to which accountable care organizations (ACOs) formally incorporate postacute care providers. DATA SOURCES: The National Survey of ACOs (N = 269, response rate 66 percent). STUDY DESIGN: We report statistics on ACOs' formal inclusion of postacute care providers and the organizational characteristics and clinical capabilities of ACOs that have postacute care. PRINCIPAL FINDINGS: Half of ACOs formally include at least one postacute service, with inclusion at higher rates in ACOs with commercial (64 percent) and Medicaid contracts (70 percent) compared to ACOs with Medicare contracts only (45 percent). ACOs that have a formal relationship with a postacute provider are more likely to have advanced transition management, end of life planning, readmission prevention, and care management capabilities. CONCLUSIONS: Many ACOs have not formally engaged postacute care, which may leave room to improve service integration and care management.