A Novel Machine Learning Algorithm for Creating Risk-Adjusted Payment Formulas.

Importance: Models predicting health care spending and other outcomes from administrative records are widely used to manage and pay for health care, despite well-documented deficiencies. New methods are needed that can incorporate more than 70 000 diagnoses without creating undesirable coding incentives. Objective: To develop a machine learning (ML) algorithm, building on Diagnostic Item (DXI) categories and Diagnostic Cost Group (DCG) methods, that automates development of clinically credible and transparent predictive models for policymakers and clinicians. Design, Setting, and Participants: DXIs were organized into disease hierarchies and assigned an Appropriateness to Include (ATI) score to reflect vagueness and gameability concerns. A novel automated DCG algorithm iteratively assigned DXIs in 1 or more disease hierarchies to DCGs, identifying sets of DXIs with the largest regression coefficient as dominant; presence of a previously identified dominating DXI removed lower-ranked ones before the next iteration. The Merative MarketScan Commercial Claims and Encounters Database for commercial health insurance enrollees 64 years and younger was used. Data from January 2016 through December 2018 were randomly split 90% to 10% for model development and validation, respectively. Deidentified claims and enrollment data were delivered by Merative the following November in each calendar year and analyzed from November 2020 to January 2024. Main Outcome and Measures: Concurrent top-coded total health care cost. Model performance was assessed using validation sample weighted least-squares regression, mean absolute errors, and mean errors for rare and common diagnoses. Results: This study included 35 245 586 commercial health insurance enrollees 64 years and younger (65 901 460 person-years) and relied on 19 clinicians who provided reviews in the base model. The algorithm implemented 218 clinician-specified hierarchies compared with the US Department of Health and Human Services (HHS) hierarchical condition category (HCC) model's 64 hierarchies. The base model that dropped vague and gameable DXIs reduced the number of parameters by 80% (1624 of 3150), achieved an R2 of 0.535, and kept mean predicted spending within 12% ($3843 of $31 313) of actual spending for the 3% of people with rare diseases. In contrast, the HHS HCC model had an R2 of 0.428 and underpaid this group by 33% ($10 354 of $31 313). Conclusions and Relevance: In this study, by automating DXI clustering within clinically specified hierarchies, this algorithm built clinically interpretable risk models in large datasets while addressing diagnostic vagueness and gameability concerns.

Development and Assessment of a New Framework for Disease Surveillance, Prediction, and Risk Adjustment: The Diagnostic Items Classification System.

Importance: Current disease risk-adjustment formulas in the US rely on diagnostic classification frameworks that predate the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Objective: To develop an ICD-10-CM-based classification framework for predicting diverse health care payment, quality, and performance outcomes. Design Setting and Participants: Physician teams mapped all ICD-10-CM diagnoses into 3 types of diagnostic items (DXIs): main effect DXIs that specify diseases; modifiers, such as laterality, timing, and acuity; and scaled variables, such as body mass index, gestational age, and birth weight. Every diagnosis was mapped to at least 1 DXI. Stepwise and weighted least-squares estimation predicted cost and utilization outcomes, and their performance was compared with models built on (1) the Agency for Healthcare Research and Quality Clinical Classifications Software Refined (CCSR) categories, and (2) the Health and Human Services Hierarchical Condition Categories (HHS-HCC) used in the Affordable Care Act Marketplace. Each model's performance was validated using R 2, mean absolute error, the Cumming prediction measure, and comparisons of actual to predicted outcomes by spending percentiles and by diagnostic frequency. The IBM MarketScan Commercial Claims and Encounters Database, 2016 to 2018, was used, which included privately insured, full- or partial-year eligible enrollees aged 0 to 64 years in plans with medical, drug, and mental health/substance use coverage. Main Outcomes and Measures: Fourteen concurrent outcomes were predicted: overall and plan-paid health care spending (top-coded and not top-coded); enrollee out-of-pocket spending; hospital days and admissions; emergency department visits; and spending for 6 types of services. The primary outcome was annual health care spending top-coded at $250 000. Results: A total of 65 901 460 person-years were split into 90% estimation/10% validation samples (n = 6 604 259). In all, 3223 DXIs were created: 2435 main effects, 772 modifiers, and 16 scaled items. Stepwise regressions predicting annual health care spending (mean [SD], $5821 [$17 653]) selected 76% of the main effect DXIs with no evidence of overfitting. Validated R 2 was 0.589 in the DXI model, 0.539 for CCSR, and 0.428 for HHS-HCC. Use of DXIs reduced underpayment for enrollees with rare (1-in-a-million) diagnoses by 83% relative to HHS-HCCs. Conclusions: In this diagnostic modeling study, the new DXI classification system showed improved predictions over existing diagnostic classification systems for all spending and utilization outcomes considered.

Diagnostic Category Prevalence in 3 Classification Systems Across the Transition to the International Classification of Diseases, Tenth Revision, Clinical Modification.

Importance: On October 1, 2015, the US transitioned to the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) for recording diagnoses, symptoms, and procedures. It is unknown whether this transition was associated with changes in diagnostic category prevalence based on diagnosis classification systems commonly used for payment and quality reporting. Objective: To assess changes in diagnostic category prevalence associated with the ICD-10-CM transition. Design, Setting, and Participants: This interrupted time series analysis and cross-sectional study examined level and trend changes in diagnostic category prevalence associated with the ICD-10-CM transition and clinically reviewed a subset of diagnostic categories with changes of 20% or more. Data included insurance claim diagnoses from the IBM MarketScan Commercial Database from January 1, 2010, to December 31, 2017, for more than 18 million people aged 0 to 64 years with private insurance. Diagnoses were mapped using 3 common diagnostic classification systems: World Health Organization (WHO) disease chapters, Department of Health and Human Services Hierarchical Condition Categories (HHS-HCCs), and Agency for Healthcare Research and Quality Clinical Classification System (AHRQ-CCS). Data were analyzed from December 1, 2018, to January 21, 2020. Exposures: US implementation of ICD-10-CM. Main Outcomes and Measures: Monthly rates of individuals with at least 1 diagnosis in a diagnostic classification category per 10 000 eligible members. Results: The analytic sample contained information on 2.1 billion enrollee person-months with 3.4 billion clinically assigned diagnoses; the mean (range) monthly sample size was 22.1 (18.4 to 27.1 ) million individuals. While diagnostic category prevalence changed minimally for WHO disease chapters, the ICD-10-CM transition was associated with level changes of 20% or more among 20 of 127 HHS-HCCs (15.7%) and 46 of 282 AHRQ-CCS categories (16.3%) and with trend changes of 20% or more among 12 of 127 of HHS-HCCs (9.4%) and 27 of 282 of AHRQ-CCS categories (9.6%). For HHS-HCCs, monthly rates of individuals with any acute myocardial infarction diagnosis increased 131.5% (95% CI, 124.1% to 138.8%), primarily because HHS added non-ST-segment-elevation myocardial infarction diagnoses to this category. The HHS-HCC for diabetes with chronic complications increased by 92.4% (95% CI, 84.2% to 100.5%), primarily from including new diabetes-related hypoglycemia and hyperglycemia codes, and the rate for completed pregnancy with complications decreased by 54.5% (95% CI, -58.7% to -50.2%) partly due to removing vaginal birth after cesarean delivery as a complication. Conclusions and Relevance: These findings suggest that the ICD-10-CM transition was associated with large prevalence changes for many diagnostic categories. Diagnostic classification systems developed using ICD-9-CM may need to be refined using ICD-10-CM data to avoid unintended consequences for disease surveillance, performance assessment, and risk-adjusted payments.

Persistent postoperative pain and healthcare costs associated with instrumented and non-instrumented spinal surgery: a case-control study.

PURPOSE: To compare rates of persistent postoperative pain (PPP) after lumbar spine surgery-commonly known as Failed Back Surgery Syndrome-and healthcare costs for instrumented lumbar spinal fusion versus decompression/discectomy. METHODS: The UK population-based healthcare data from the Hospital Episode Statistics (HES) database from NHS Digital and the Clinical Practice Research Datalink (CPRD) were queried to identify patients with PPP following lumbar spinal surgery. Rates of PPP were calculated by type of surgery (instrumented and non-instrumented). Total healthcare costs associated with the surgery and covering the 24-month period after index hospital discharge were estimated using standard methods for classifying health care encounters into major categories of health care resource utilization (i.e., inpatient hospital stays, outpatient clinic visits, accident and emergency attendances, primary care encounters, and medications prescribed in primary care) and applying the appropriate unit costs (expressed in 2013 GBP). RESULTS: Increasing the complexity of surgery with instrumentation was not associated with an increased rate of PPP. However, 2-year healthcare costs following discharge after surgery are significantly higher among patients who underwent instrumented surgery compared with decompression/discectomy. CONCLUSIONS: Although there is a not insubstantial risk of ongoing pain following spine surgery, with 1-in-5 patients experiencing PPP within 2 years of surgery, the underlying indications for surgical modality and related choice of surgical procedure do not, by itself, appear to be a driving factor.

Reoperation following lumbar spinal surgery: costs and outcomes in a UK population cohort study using the Clinical Practice Research Datalink (CPRD) and Hospital Episode Statistics (HES).

PURPOSE: To assess the likelihood of persistent postoperative pain (PPP) following reoperation after lumbar surgery and to estimate associated healthcare costs. METHODS: This is a retrospective cohort study using two linked UK databases: Hospital Episode Statistics and UK Clinical Practice Research Datalink. Costs and outcomes associated with reoperation were evaluated over a 2-year postoperative period using multivariate logistic regression for cases who underwent reoperation and controls who did not, based on demographics, index surgery type, smoking status, and pre-index comorbidities using propensity score matching. RESULTS: Risk factors associated with reoperation included younger age and the presence of diabetes with complications or rheumatic disease. The rate of PPP after reoperation was much higher than after index surgery, with 79 of 200 (39.5%; 95% CI 32.5%, 46.5%) participants experiencing ongoing pain compared with 983 of 5022 (19.5%; 95% CI 18.5%, 20.7%) after index surgery. Mean costs in the 2 years following reoperation were £1889 higher (95% CI £2, £3809) than for patients with PPP who did not undergo repeat surgery over an equivalent follow-up period. With the cost of reoperation itself included, the mean cost difference for patients who underwent reoperation compared with matched controls rose to £7221 (95% CI £5273, £9206). CONCLUSIONS: High rates of PPP and associated healthcare costs suggest that returning to the operating room is a complex and challenging decision. Spinal surgeons should review whether the potential benefits of additional surgery are justified when other approaches to managing and relieving chronic pain have demonstrated superior outcomes. These slides can be retrieved under Electronic Supplementary Material.

Short- and long-term cost and utilization of health care resources in Parkinson's disease in the UK.

BACKGROUND: There is currently no robust long-term data on costs of treating patients with Parkinson's disease. The objective of this study was to report levels of health care utilization and associated costs in the 10 years after diagnosis among PD patients in the United Kingdom. METHODS: We undertook a retrospective population-based cohort study using linked data from the UK Clinical Practice Research Datalink and Hospital Episode Statistics databases. Total health care costs of PD patients were compared with those of a control group of patients without PD selected using 1:1 propensity score matching based on age, sex, and comorbidity. RESULTS: Between 1994 and 2013, 7271 PD patients who met study inclusion criteria were identified in linked Clinical Practice Research Datalink-Hospital Episode Statistics; 7060 were matched with controls. The mean annual health care cost difference (at 2013 costs) between PD patients and controls was £2471 (US$3716) per patient in the first year postdiagnosis (P < 0.001), increasing to £4004 (US$6021) per patient (P < 0.001) 10 years following diagnosis because of higher levels of use across all categories of health care utilization. Costs in patients with markers of advanced PD (ie, presence of levodopa-equivalent daily dose > 1100 mg, dyskinesias, falls, dementia, psychosis, hospital admission primarily due to PD, or nursing home placement) were on average higher by £1069 (US$1608) per patient than those with PD without these markers. CONCLUSIONS: This study provides comprehensive estimates of health care costs in PD patients based on routinely collected data. Health care costs attributable to PD increase in the year following diagnosis and are higher for patients with indicators of advanced disease. © 2018 International Parkinson and Movement Disorder Society.

The incidence and healthcare costs of persistent postoperative pain following lumbar spine surgery in the UK: a cohort study using the Clinical Practice Research Datalink (CPRD) and Hospital Episode Statistics (HES).

OBJECTIVE: To characterise incidence and healthcare costs associated with persistent postoperative pain (PPP) following lumbar surgery. DESIGN: Retrospective, population-based cohort study. SETTING: Clinical Practice Research Datalink (CPRD) and Hospital Episode Statistics (HES) databases. PARTICIPANTS: Population-based cohort of 10 216 adults who underwent lumbar surgery in England from 1997/1998 through 2011/2012 and had at least 1 year of presurgery data and 2 years of postoperative follow-up data in the linked CPRD-HES. PRIMARY AND SECONDARY OUTCOMES MEASURES: Incidence and total healthcare costs over 2, 5 and 10 years attributable to persistent PPP following initial lumbar surgery. RESULTS: The rate of individuals undergoing lumbar surgery in the CPRD-HES linked data doubled over the 15-year study period, fiscal years 1997/1998 to 2011/2012, from 2.5 to 4.9 per 10 000 adults. Over the most recent 5-year period (2007/2008 to 2011/2012), on average 20.8% (95% CI 19.7% to 21.9%) of lumbar surgery patients met criteria for PPP. Rates of healthcare usage were significantly higher for patients with PPP across all types of care. Over 2 years following initial spine surgery, the mean cost difference between patients with and without PPP was £5383 (95% CI £4872 to £5916). Over 5 and 10 years following initial spine surgery, the mean cost difference between patients with and without PPP increased to £10 195 (95% CI £8726 to £11 669) and £14 318 (95% CI £8386 to £19 771), respectively. Extrapolated to the UK population, we estimate that nearly 5000 adults experience PPP after spine surgery annually, with each new cohort costing the UK National Health Service in excess of £70 million over the first 10 years alone. CONCLUSIONS: Persistent pain affects more than one-in-five lumbar surgery patients and accounts for substantial long-term healthcare costs. There is a need for formal, evidence-based guidelines for a coherent, coordinated management strategy for patients with continuing pain after lumbar surgery.

A Comparison of Estimation Methods for a Multi-unidimensional Graded Response IRT Model.

This study compared several parameter estimation methods for multi-unidimensional graded response models using their corresponding statistical software programs and packages. Specifically, we compared two marginal maximum likelihood (MML) approaches (Bock-Aitkin expectation-maximum algorithm, adaptive quadrature approach), four fully Bayesian algorithms (Gibbs sampling, Metropolis-Hastings, Hastings-within-Gibbs, blocked Metropolis), and the Metropolis-Hastings Robbins-Monro (MHRM) algorithm via the use of IRTPRO, BMIRT, and MATLAB. Simulation results suggested that, when the intertrait correlation was low, these estimation methods provided similar results. However, if the dimensions were moderately or highly correlated, Hastings-within-Gibbs had an overall better parameter recovery of item discrimination and intertrait correlation parameters. The performances of these estimation methods with different sample sizes and test lengths are also discussed.

Simulating Univariate and Multivariate Nonnormal Distributions through the Method of Percentiles.

This article derives a standard normal-based power method polynomial transformation for Monte Carlo simulation studies, approximating distributions, and fitting distributions to data based on the method of percentiles. The proposed method is used primarily when (1) conventional (or L) moment-based estimators such as skew (or L-skew) and kurtosis (or L -kurtosis) are unknown or (2) data are unavailable but percentiles are known (e.g., standardized test score reports). The proposed transformation also has the advantage that solutions to polynomial coefficients are available in simple closed form and thus obviates numerical equation solving. A procedure is also described for simulating power method distributions with specified medians, inter-decile ranges, left-right tail-weight ratios (skew function), tail-weight factors (kurtosis function), and Spearman correlations. The Monte Carlo results presented in this study indicate that the estimators based on the method of percentiles are substantially superior to their corresponding conventional product-moment estimators in terms of relative bias. It is also shown that the percentile power method can be modified for generating nonnormal distributions with specified Pearson correlations. An illustration shows the applicability of the percentile power method technique to publicly available statistics from the Idaho state educational assessment.

Implications of comorbidity on costs for patients with Alzheimer disease.

BACKGROUND: No prior studies have used a comprehensive clinical classification system to examine the effect of differences in overall illness burden and the presence of other diseases on costs for patients with Alzheimer disease (AD) when compared with demographically matched nondemented controls. STUDY DESIGN: Of a total of 627,775 enrollees who were eligible for medical and pharmacy benefits for 2003 and 2004 in the MarketScan Medicare Supplemental and Coordination of Benefits Database, we found 25,109 AD patients. For each case, 3 demographically matched nondemented controls were selected using propensity scores. Applying the diagnostic cost groups (DCGs) model to all enrollees, 2003 diagnoses were used to estimate prospective relative risk scores (RRSs) that predict 2004 costs from all illness other than AD. RRSs were then used to control for illness burden to estimate AD's independent effect on costs. RESULTS: Compared with the control group, the AD cohort has more comorbid conditions (8.1 vs. 6.5) and higher illness burden (1.23 vs. 1.04). Individuals with AD are more likely to have mental health conditions, neurologic conditions, cognitive disorders, cerebrovascular disease, diabetes with acute complications, and injuries. Annual costs for AD patients are $3567 (34%) higher than for controls. Excess costs attributable to AD, after controlling for non-AD illness burden, are estimated at $2307 per year with outpatient pharmacy being the key driver ($1711 in excess costs). CONCLUSIONS: AD patients are sicker and more expensive than demographically matched controls. Even after adjusting for differences in illness burden, costs remain higher for AD patients.

Healthcare costs and utilization for Medicare beneficiaries with Alzheimer's.

BACKGROUND: Alzheimer's disease (AD) is a neurodegenerative disorder incurring significant social and economic costs. This study uses a US administrative claims database to evaluate the effect of AD on direct healthcare costs and utilization, and to identify the most common reasons for AD patients' emergency room (ER) visits and inpatient admissions. METHODS: Demographically matched cohorts age 65 and over with comprehensive medical and pharmacy claims from the 2003-2004 MEDSTAT MarketScan Medicare Supplemental and Coordination of Benefits (COB) Database were examined: 1) 25,109 individuals with an AD diagnosis or a filled prescription for an exclusively AD treatment; and 2) 75,327 matched controls. Illness burden for each person was measured using Diagnostic Cost Groups (DCGs), a comprehensive morbidity assessment system. Cost distributions and reasons for ER visits and inpatient admissions in 2004 were compared for both cohorts. Regression was used to quantify the marginal contribution of AD to health care costs and utilization, and the most common reasons for ER and inpatient admissions, using DCGs to control for overall illness burden. RESULTS: Compared with controls, the AD cohort had more co-morbid medical conditions, higher overall illness burden, and higher but less variable costs ($13,936 s. $10,369; Coefficient of variation = 181 vs. 324). Significant excess utilization was attributed to AD for inpatient services, pharmacy, ER visits, and home health care (all p < 0.05). In particular, AD patients were far more likely to be hospitalized for infections, pneumonia and falls (hip fracture, syncope, collapse). CONCLUSION: Patients with AD have significantly more co-morbid medical conditions and higher healthcare costs and utilization than demographically-matched Medicare beneficiaries. Even after adjusting for differences in co-morbidity, AD patients incur excess ER visits and inpatient admissions.

Adaptive sliding-mode control for nonlinear systems with uncertain parameters.

This correspondence proposes a systematic adaptive sliding-mode controller design for the robust control of nonlinear systems with uncertain parameters. An adaptation tuning approach without high-frequency switching is developed to deal with unknown but bounded system uncertainties. Tracking performance is guaranteed. System robustness, as well as stability, is proven by using the Lyapunov theory. The upper bounds of uncertainties are not required to be known in advance. Therefore, the proposed method can be effectively implemented. Experimental results demonstrate the effectiveness of the proposed control method.

Sliding mode control with self-tuning law for uncertain nonlinear systems.

A robust sliding mode control that follows a self-tuning law for nonlinear systems possessing uncertain parameters is proposed. The adjustable control gain and a bipolar sigmoid function are on-line tuned to force the tracking error to approach zero. Control system stability is ensured using the Lyapunov method. Both simulation and experimental application of a planetary gear type inverted pendulum control system verify the effectiveness of the developed approach.