Dataset Information

Incorporating machine learning and social determinants of health indicators into prospective risk adjustment for health plan payments.

ABSTRACT: BACKGROUND:Risk adjustment models are employed to prevent adverse selection, anticipate budgetary reserve needs, and offer care management services to high-risk individuals. We aimed to address two unknowns about risk adjustment: whether machine learning (ML) and inclusion of social determinants of health (SDH) indicators improve prospective risk adjustment for health plan payments. METHODS:We employed a 2-by-2 factorial design comparing: (i) linear regression versus ML (gradient boosting) and (ii) demographics and diagnostic codes alone, versus additional ZIP code-level SDH indicators. Healthcare claims from privately-insured US adults (2016-2017), and Census data were used for analysis. Data from 1.02 million adults were used for derivation, and data from 0.26 million to assess performance. Model performance was measured using coefficient of determination (R2), discrimination (C-statistic), and mean absolute error (MAE) for the overall population, and predictive ratio and net compensation for vulnerable subgroups. We provide 95% confidence intervals (CI) around each performance measure. RESULTS:Linear regression without SDH indicators achieved moderate determination (R2 0.327, 95% CI: 0.300, 0.353), error ($6992; 95% CI: $6889, $7094), and discrimination (C-statistic 0.703; 95% CI: 0.701, 0.705). ML without SDH indicators improved all metrics (R2 0.388; 95% CI: 0.357, 0.420; error $6637; 95% CI: $6539, $6735; C-statistic 0.717; 95% CI: 0.715, 0.718), reducing misestimation of cost by $3.5?M per 10,000 members. Among people living in areas with high poverty, high wealth inequality, or high prevalence of uninsured, SDH indicators reduced underestimation of cost, improving the predictive ratio by 3% (~$200/person/year). CONCLUSIONS:ML improved risk adjustment models and the incorporation of SDH indicators reduced underpayment in several vulnerable populations.

SUBMITTER: Irvin JA

PROVIDER: S-EPMC7195714 | biostudies-literature | 2020 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Incorporating machine learning and social determinants of health indicators into prospective risk adjustment for health plan payments.

Irvin Jeremy A JA Kondrich Andrew A AA Ko Michael M Rajpurkar Pranav P Haghgoo Behzad B Landon Bruce E BE Phillips Robert L RL Petterson Stephen S Ng Andrew Y AY Basu Sanjay S

BMC public health 20200501 1

<h4>Background</h4>Risk adjustment models are employed to prevent adverse selection, anticipate budgetary reserve needs, and offer care management services to high-risk individuals. We aimed to address two unknowns about risk adjustment: whether machine learning (ML) and inclusion of social determinants of health (SDH) indicators improve prospective risk adjustment for health plan payments.<h4>Methods</h4>We employed a 2-by-2 factorial design comparing: (i) linear regression versus ML (gradient ...[more]

PMID: 32357871

Similar Datasets

Project description:BackgroundIn 2014, a revision of the national medical tariffs for inpatient health care services took place in Iran, and a new hotline was set up to report informal payments. It was expected that such measures would eliminate or decrease informal payments prevalence. This study estimates the prevalence of informal payments for inpatient health care services in the post-reform period, explores factors associated with informal payments and examines patients' and healthcare providers' views regarding the causes of informal payments and possible practical solutions for their reduction.MethodsWe surveyed by phone patients who used inpatient health care services in seven Iranian hospitals in 2016. Descriptive and regression analyses were used to estimate the prevalence and determine factors associated with informal payments. We conducted a qualitative analysis through thematic analyses based on focus group discussions and in-depth interviews.ResultsOf 2696 respondents, 14% reported paying informally for inpatient services. Informal payments were reported more frequently among private hospital users, given more frequently to physicians in public teaching hospitals and 'other staff' in private hospitals, in the form of cash and voluntary. Being an adult, hospital or treatment type, being insured, and household head's education influenced the probability of paying informally. The amount paid informally was associated with being insured, the educational status of the household's head, household size, service, and hospital types. Based on qualitative findings, the leading causes of informal payments reported by patients and healthcare providers can be categorized into four groups - financing challenges; governance challenges; service delivery challenges; and actors and stakeholders. Modifying, adjusting and applying policy interventions; supervision, monitoring and evaluation; and actors and stakeholders were identified as possible solutions for tackling informal payment in the inpatient health care services.ConclusionThe prevalence of informal patient payments for inpatient services in the post-reform period seems to have reduced; however, they remain to be common. Regular monitoring, reviewing of payment policies to the physicians, informing patients, changing the behaviour of healthcare providers and patients, and developing ethical guidelines to prevent informal payments were suggested for reduction and elimination of informal payments in the Iranian healthcare sector.

Project description:ImportancePresent-on-admission (POA) indicators in administrative claims data allow researchers to distinguish between preexisting conditions and those acquired during a hospital stay. The impact of adding POA information to claims-based measures of hospital quality has not yet been investigated to better understand patient underlying risk factors in the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision setting.ObjectiveTo assess POA indicator use on Medicare claims and to assess the hospital- and patient-level outcomes associated with incorporating POA indicators in identifying risk factors for publicly reported outcome measures used by the Centers for Medicare & Medicaid Services (CMS).Design, setting, and participantsThis comparative effectiveness study used national CMS claims data between July 1, 2015, and June 30, 2018. Six hospital quality measures assessing readmission and mortality outcomes were modified to include POA indicators in risk adjustment models. The models using POA were then compared with models using the existing complications-of-care algorithm to evaluate changes in risk model performance. Patient claims data were included for all Medicare fee-for-service and Veterans Administration beneficiaries aged 65 years or older with inpatient hospitalizations for acute myocardial infarction, heart failure, or pneumonia within the measurement period. Data were analyzed between September 2019 and March 2020.Main outcomes and measuresChanges in patient-level (C statistics) and hospital-level (quintile shifts in risk-standardized outcome rates) model performance after including POA indicators in risk adjustment.ResultsData from a total of 6 027 988 index admissions were included for analysis, ranging from 491 366 admissions (269 209 [54.8%] men; mean [SD] age, 78.2 [8.3] years) for the acute myocardial infarction mortality outcome measure to 1 395 870 admissions (677 158 [48.5%] men; mean [SD] age, 80.3 [8.7] years) for the pneumonia readmission measure. Use of POA indicators was associated with improvements in risk adjustment model performance, particularly for mortality measures (eg, the C statistic increased from 0.728 [95% CI, 0.726-0.730] to 0.774 [95% CI, 0.773-0.776] when incorporating POA indicators into the acute myocardial infarction mortality measure).Conclusions and relevanceThe findings of this quality improvement study suggest that leveraging POA indicators in the risk adjustment methodology for hospital quality outcome measures may help to more fully capture patients' risk factors and improve overall model performance. Incorporating POA indicators does not require extra effort on the part of hospitals and would be easy to implement in publicly reported quality outcome measures.

Project description:BackgroundSocial determinants of health (SDoHs) and social risks (SRs) have been associated with adverse health and healthcare utilization and racial/ethnic disparities. However, there is limited information about the prevalence of SRs in non-"safety net" adult populations and how SRs differ by race/ethnicity, age, education, and income.MethodsWe analyzed weighted survey data for 16,247 White, 1861 Black, 2895 Latino, 1554 Filipino, and 1289 Chinese adults aged 35 to 79 who responded to the 2011 or 2014/2015 Kaiser Permanente Northern California Member Health Survey. We compared age-standardized prevalence estimates of SDoHs (education, household income, marital status) and SRs (financial worry, cost-related reduced medication use and fruit/vegetable consumption, chronic stress, harassment/discrimination, health-related beliefs) across racial/ethnic groups for ages 35 to 64 and 65 to 79.ResultsSDoHs and SRs differed by race/ethnicity and age group, and SRs differed by levels of education and income. In both age groups, Blacks, Latinos, and Filipinos were more likely than Whites to be in the lower income category and be worried about their financial situation. Compared to Whites, cost-related reduced medication use was higher among Blacks, and cost-related reduced fruit/vegetable consumption was higher among Blacks and Latinos. Younger adults were more likely than older adults to experience chronic stress and financial worry. Racial/ethnic disparities in income were observed within similar levels of education. Differences in prevalence of SRs by levels of education and income were wider within than across racial/ethnic groups.ConclusionsIn this non-"safety net" adult health plan population, Blacks, Latinos, and Filipinos had a higher prevalence of social risks than Whites and Chinese, and prevalence of social risks differed by age group. Our results support the assessment and EHR documentation of SDoHs and social risks and use of this information to understand and address drivers of racial/ethnic disparities in health and healthcare use.

Project description:ImportanceThe passage of the Creating High-Quality Results and Outcomes Necessary to Improve Chronic (CHRONIC) Care Act in 2018 allows Medicare Advantage (MA) plans, which enroll more than one-third of Medicare beneficiaries, greater flexibility to address members' social determinants of health (SDOH) through supplemental benefits.ObjectiveTo understand MA plan representatives' perspectives on the importance of addressing members' SDOH and their responses to the passage of the CHRONIC Care Act.Design, setting, and participantsThis semistructured qualitative interview study conducted via telephone from July 6, 2018, to November 7, 2018, included participants from 17 MA plans that collectively enrolled more than 13 million MA members (>65% of the total MA market). Data analysis was conducted from September 18, 2018, to December 13, 2018.Main outcomes and measuresAudio-recorded interviews were transcribed and then analyzed using a modified content analysis approach to identify major themes and subthemes.ResultsThirty-eight participants representing 17 MA plans varying in region, star rating, and size were interviewed. Analysis of interviews revealed 3 key themes. The first theme was that participants increasingly recognize the value of addressing members' SDOH. The second theme was that participants had different perspectives on whether MA plans should directly address SDOH and how to do so. While some reported that they were taking advantage of the increased flexibility provided by the CHRONIC Care Act to design new benefits or partner with community-based organizations, others indicated that it was outside of their purview to directly address members' SDOH. The third theme was that participants described complex decision-making around how to provide supplemental benefits, including a need for evidence, return on investment, strong community partnerships, and guidance from the US Centers for Medicare & Medicaid Services.Conclusions and relevanceThese findings suggest that the changes in MA plans' benefit packages in response to the CHRONIC Care Act and their efforts to address SDOH will vary. Therefore, it is likely that MA enrollees will be differentially affected by the implementation of the CHRONIC Care Act.

Dataset Information

Incorporating machine learning and social determinants of health indicators into prospective risk adjustment for health plan payments.

Publications

Incorporating machine learning and social determinants of health indicators into prospective risk adjustment for health plan payments.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure