Prompt access to outpatient care post-incarceration among adults with a history of substance use: Predisposing, enabling, and need-based factors

INTRODUCTION

As expanded Medicaid coverage reduces financial barriers to receiving health care among formerly incarcerated adults, more information is needed to understand the factors that predict prompt use of health care after release among insured adults with a history of substance use. This study's aim was to estimate the associations between characteristics suggested by the Andersen behavioral model of health service use and measures of health care use during the immediate reentry period and in the presence of Medicaid coverage.

METHODS

In this retrospective cohort study, we linked individual-level data from multiple Wisconsin agencies. The sample included individuals aged 18-64 released from a Wisconsin State Correctional Facility between April 2014 and June 2017 to a community in the state who enrolled in Medicaid within one month of release and had a history of substance use. We grouped predictors of outpatient care into variable domains within the Andersen model: predisposing- individual socio-demographic characteristics; enabling characteristics including area-level socio-economic resources, area-level health care supply, and characteristics of the incarceration and release; and need-based- pre-release health conditions. We used a model selection algorithm to select a subset of variable domains and estimated the association between the variables in these domains and two outcomes: any outpatient visit within 30 days of release from a state correctional facility, and receipt of medication for opioid use disorder within 30 days of release.

RESULTS

The size and sign of many of the estimated associations differed for our two outcomes. Race was associated with both outcomes, Black individuals being 12.1 p.p. (95 % CI, 8.7-15.4, P < .001) less likely than White individuals to have an outpatient visit within 30 days of release and 1.3 p.p. (95 % CI, 0.48-2.1, P = .002) less likely to receive MOUD within 30 days of release. Chronic pre-release health conditions were positively associated with the likelihood of post-release health care use.

CONCLUSIONS

Conditional on health insurance coverage, meaningful differences in post-incarceration outpatient care use still exist across adults leaving prison with a history of substance use. These findings can help guide the development of care transition interventions including the prioritization of subgroups that may warrant particular attention.

Heteroscedasticity of residual spending after risk equalization: a potential source of selection incentives in health insurance markets with premium regulation

Many community-rated health insurance markets include risk equalization (also known as risk adjustment) to mitigate risk selection incentives for competing insurers. Empirical evaluations of risk equalization typically quantify selection incentives through predictable profits and losses net of risk equalization for various groups of consumers (e.g. the healthy versus the chronically ill). The underlying assumption is that absence of predictable profits and losses implies absence of selection incentives. This paper questions this assumption. We show that even when risk equalization perfectly compensates insurers for predictable differences in mean spending between groups, selection incentives are likely to remain. The reason is that the uncertainty about residual spending (i.e., spending net of risk equalization) differs across groups, e.g., the risk of substantial losses is larger for the chronically ill than for the healthy. In a risk-rated market, insurers are likely to charge a higher profit mark-up (to cover uncertainty in residual spending) and a higher safety mark-up (to cover the risk of large losses) to chronically ill than to healthy individuals. When such differentiation is not allowed, insurers face incentives to select in favor of the healthy. Although the exact size of these selection incentives depends on contextual factors, our empirical simulations indicate they can be non-trivial. Our findings suggest that - in addition to the equalization of differences in mean spending between the healthy and the chronically ill - policy measures might be needed to diminish (or compensate insurers for) heteroscedasticity of residual spending across groups.

Comparing risk-adjusted inpatient fall rates internationally: validation of a risk-adjustment model using multicentre cross-sectional data from hospitals in Switzerland and Austria

BACKGROUND

Inpatient falls in hospitals are an acknowledged indicator of quality of care. International comparisons could highlight quality improvement potential and enable cross-national learning. Key to fair cross-national comparison is the availability of a risk adjustment model validated in an international context. This study aimed to 1) ascertain that the variables of the inpatient fall risk adjustment model do not interact with country and thus can be used for risk adjustment, 2) compare the risk of falling in hospitals between Switzerland and Austria after risk adjustment.

METHODS

The data on inpatient falls from Swiss and Austrian acute care hospitals were collected on a single measurement day in 2017, 2018 and 2019 as part of an international multicentre cross-sectional study. Multilevel logistic regression models were used to screen for interaction effects between the patient-related fall risk factors and the countries. The risks of falling in hospital in Switzerland and in Austria were compared after applying the risk-adjustment model.

RESULTS

Data from 176 hospitals and 43,984 patients revealed an inpatient fall rate of 3.4% in Switzerland and 3.9% in Austria. Two of 15 patient-related fall risk variables showed an interaction effect with country: Patients who had fallen in the last 12 months (OR 1.49, 95% CI 1.10-2.01, p = 0.009) or had taken sedatives/psychotropic medication (OR 1.40, 95% CI 1.05-1.87, p = 0.022) had higher odds of falling in Austrian hospitals. Significantly higher odds of falling were observed in Austrian (OR 1.38, 95% CI 1.13-1.68, p = 0.002) compared to Swiss hospitals after applying the risk-adjustment model.

CONCLUSIONS

Almost all patient-related fall risk factors in the model are suitable for a risk-adjusted cross-country comparison, as they do not interact with the countries. Further model validation with additional countries is warranted, particularly to assess the interaction of risk factors "fall in the last 12 months" and "sedatives/psychotropic medication intake" with country variable. The study underscores the crucial role of an appropriate risk-adjustment model in ensuring fair international comparisons of inpatient falls, as the risk-adjusted, as opposed to the non-risk-adjusted country comparison, indicated significantly higher odds of falling in Austrian compared to Swiss hospitals.

A risk-adjusted cumulative sum analysis of the progression from a novice to an expert surgeon at a single institution

BACKGROUND/OBJECTIVE

Laparoscopic surgery for rectal cancer is challenging for novice surgeons because it requires a sharp dissection in a narrow pelvis with visual limitations. Therefore, this study aimed to analyze the learning curve and clinical outcomes of laparoscopic surgery for rectal cancer performed by a novice surgeon en route to becoming an expert.

METHODS

In total, 119 patients who underwent laparoscopic surgery for rectal cancer performed by a single surgeon between June 2010 and December 2019 were analyzed. A single hybrid model based on the operative time, open conversion, complications, and resection margin involvement was generated to assess the success of laparoscopic surgery. Furthermore, the learning curve was evaluated using the risk-adjusted cumulative sum (RA-CUSUM) method.

RESULTS

The learning period was categorized into three phases according to the RA-CUSUM method (phase 1, 1st-33rd cases; phase 2, 34th-84th cases; and phase 3, 85th-119th cases). Tumor size (p = 0.004), distal resection margin (p = 0.003), and the number of harvested lymph nodes (p < 0.001) significantly increased with the learning period. The time to tolerable soft diet became shorter according to the learning period (p = 0.017). Advanced T stage (p = 0.024) and adjuvant chemotherapy (p = 0.012) were more common in phase 3.

CONCLUSIONS

This study suggested that the initial technical competence of laparoscopic surgery for rectal cancer was acquired in the 33rd case. Technical mastery was achieved in the 84th case. After mastering the technique, the surgeon tended to challenge more advanced cases, however, the complication rates did not increase.

Risk adjustment for regional healthcare funding allocations with ensemble methods: an empirical study and interpretation

We experiment with recent ensemble machine learning methods in estimating healthcare costs, utilizing Finnish data containing rich individual-level information on healthcare costs, socioeconomic status and diagnostic data from multiple registries. Our data are a random 10% sample (553,675 observations) from the Finnish population in 2017. Using annual healthcare cost in 2017 as a response variable, we compare the performance of Random forest, Gradient Boosting Machine (GBM) and eXtreme Gradient Boosting (XGBoost) to linear regression. As machine learning methods are often seen as unsuitable in risk adjustment applications because of their relative opaqueness, we also introduce visualizations from the machine learning literature to help interpret the contribution of individual variables to the prediction. Our results show that ensemble machine learning methods can improve predictive performance, with all of them significantly outperforming linear regression, and that a certain level of interpretation can be provided for them. We also find individual-level socioeconomic variables to improve prediction accuracy and that their effect is larger for machine learning methods. However, we find that the predictions used for funding allocations are sensitive to model selection, highlighting the need for comprehensive robustness testing when estimating risk adjustment models used in applications.

Adjusting for Confounders in Outcome Studies Using the Korea National Health Insurance Claim Database: A Review of Methods and Applications

OBJECTIVES

Adjusting for potential confounders is crucial for producing valuable evidence in outcome studies. Although numerous studies have been published using the Korea National Health Insurance Claim Database, no study has critically reviewed the methods used to adjust for confounders. This study aimed to review these studies and suggest methods and applications to adjust for confounders.

METHODS

We conducted a literature search of electronic databases, including PubMed and Embase, from January 1, 2021 to December 31, 2022. In total, 278 studies were retrieved. Eligibility criteria were published in English and outcome studies. A literature search and article screening were independently performed by 2 authors and finally, 173 of 278 studies were included.

RESULTS

Thirty-nine studies used matching at the study design stage, and 171 adjusted for confounders using regression analysis or propensity scores at the analysis stage. Of these, 125 conducted regression analyses based on the study questions. Propensity score matching was the most common method involving propensity scores. A total of 171 studies included age and/or sex as confounders. Comorbidities and healthcare utilization, including medications and procedures, were used as confounders in 146 and 82 studies, respectively.

CONCLUSIONS

This is the first review to address the methods and applications used to adjust for confounders in recently published studies. Our results indicate that all studies adjusted for confounders with appropriate study designs and statistical methodologies; however, a thorough understanding and careful application of confounding variables are required to avoid erroneous results.

Hospital frailty risk score is superior to legacy comorbidity indices for risk adjustment of in-hospital cirrhosis cases

Background & Aims

The hospital frailty risk score (HFRS) identifies older patients at risk of poor outcomes and may have value in cirrhosis. We compared the Charlson (CCI), Elixhauser (ECI), and cirrhosis (CirCom) comorbidity indices with the HFRS in predicting outcomes for cirrhosis hospitalisations.

Methods

Using the National Inpatient Sample (quarter 4 of 2015-2019), we analysed cirrhosis hospitalisations. For each index, we described the prevalence of comorbid conditions and inpatient mortality. We compared the ability of CCI, ECI, CirCom, and HFRS to predict inpatient mortality. Raw and adjusted models predicting inpatient mortality were compared using the area under the receiver operating characteristic curve and the Akaike information criterion.

Results

The cohort's (N = 626,553) median age was 61 years (IQR 52-68 years), 60% were male, cirrhosis was caused by alcohol in 43%, and 38% had ascites. The median comorbidity scores are as follows: ECI 4 (IQR 3-6), CCI 5 (IQR 4-8), and HFRS 5.6 (IQR 3.0-8.6). The most common CirCom score was 0 + 0 (44%). Across the range of values of each index, we observed different mortality ranges: CCI 1.9-13.1%, ECI 3.2-8.7%, CirCom 4.9-13.8%, and HFRS 1.0-15.2%. An adjusted model with HFRS had the highest area under the receiver operating characteristic curve in predicting mortality (HFRS 0.782 vs. ECI 0.689, CCI 0.695, and CirCom 0.692). We observed substantial variation in mortality with HFRS within each level of CCI, ECI, and CirCom. For example, for ECI 4, mortality increased from 0.6 to 16.4%, as HFRS increased from 0 to 15.

Conclusions

Comorbidity indices predict inpatient cirrhosis mortality, but HFRS performs better than CCI, ECI, and CirCom. HFRS is an ideal tool for measuring comorbidity burden and disease severity risk adjustment in cirrhosis-related administrative database studies.

Impact and Implications

We compared commonly used comorbidity indices to a more recently described risk score (hospital frailty risk score [HFRS]) in patients with cirrhosis using a national sample of hospital records. Comorbid conditions are common in hospitalised patients with cirrhosis. There is significant variability in mortality across the range of each index. HFRS outperforms the Charlson comorbidity index, Elixhauser comorbidity index, and CirCom (cirrhosis-specific comorbidity scoring system) in predicting inpatient mortality. HFRS is a valuable index for risk adjustment in inpatient administrative database studies.

Using the Elixhauser risk adjustment model to predict outcomes among patients hospitalized in internal medicine at a large, tertiary-care hospital in Israel

BACKGROUND

In Israel, internal medicine admissions are currently reimbursed without accounting for patient complexity. This is at odds with most other developed countries and has the potential to lead to market distortions such as avoiding sicker patients. Our objective was to apply a well-known, freely available risk adjustment model, the Elixhauser model, to predict relevant outcomes among patients hospitalized on the internal medicine service of a large, Israeli tertiary-care hospital.

METHODS

We used data from the Shaare Zedek Medical Center, a large tertiary referral hospital in Jerusalem. The study included 55,946 hospitalizations between 01.01.2016 and 31.12.2019. We modeled four patient outcomes: in-hospital mortality, escalation of care (intensive care unit (ICU) transfer, mechanical ventilation, daytime bi-level positive pressure ventilation, or vasopressors), 30-day readmission, and length of stay (LOS). We log-transformed LOS to address right skew. As is usual with the Elixhauser model, we identified 29 comorbid conditions using international classification of diseases codes, clinical modification, version 9. We derived and validated the coefficients for these 29 variables using split-sample derivation and validation. We checked model fit using c-statistics and R2, and model calibration using a Hosmer-Lemeshow test.

RESULTS

The Elixhauser model achieved acceptable prediction of the three binary outcomes, with c-statistics of 0.712, 0.681, and 0.605 to predict in-hospital mortality, escalation of care, and 30-day readmission respectively. The c-statistic did not decrease in the validation set (0.707, 0.687, and 0.603, respectively), suggesting that the models are not overfitted. The model to predict log length of stay achieved an R2 of 0.102 in the derivation set and 0.101 in the validation set. The Hosmer-Lemeshow test did not suggest issues with model calibration.

CONCLUSION

We demonstrated that a freely-available risk adjustment model can achieve acceptable prediction of important clinical outcomes in a dataset of patients admitted to a large, Israeli tertiary-care hospital. This model could potentially be used as a basis for differential payment by patient complexity.

Reformulating provider profiling by grouping providers treating similar patients prior to evaluating performance

Standard approaches to comparing health providers' performance rely on hierarchical logistic regression models that adjust for patient characteristics at admission. Estimates from these models may be misleading when providers treat different patient populations and the models are misspecified. To address this limitation, we propose a novel profiling approach that identifies groups of providers treating similar populations of patients and then evaluates providers' performance within each group. The groups of providers are identified using a Bayesian multilevel finite mixture of general location models. To compare the performance of our proposed profiling approach to standard methods, we use patient-level data from 119 skilled nursing facilities in Massachusetts. We use simulated and observed outcome data to explore the performance of these profiling methods in different settings. In simulations, our proposed method classifies providers to groups with similar patients' admission characteristics. In addition, in the presence of limited overlap in patient characteristics across providers and misspecifications of the outcome model, the provider-level estimates obtained using our approach identified providers that under- and overperformed compared to the standard regression-based approaches more accurately.

Monitoring the performance of a dedicated weaning unit using risk-adjusted control charts for the weaning rate in prolonged mechanical ventilation

BACKGROUND

Weaning rate is an important quality indicator of care for patients with prolonged mechanical ventilation (PMV). However, diverse clinical characteristics often affect the measured rate. A risk-adjusted control chart may be beneficial for assessing the quality of care.

METHODS

We analyzed patients with PMV who were discharged between 2018 and 2020 from a dedicated weaning unit at a medical center. We generated a formula to estimate monthly weaning rates using multivariate logistic regression for the clinical, laboratory, and physiologic characteristics upon weaning unit admission in the first two years (Phase I). We then applied both multiplicative and additive models for adjusted p-charts, displayed in both non-segmented and segmented formats, to assess whether special cause variation existed.

RESULTS

A total of 737 patients were analyzed, including 503 in Phase I and 234 in Phase II, with average weaning rates of 59.4% and 60.3%, respectively. The p-chart of crude weaning rates did not show special cause variation. Ten variables from the regression analysis were selected for the formula to predict individual weaning probability and generate estimated weaning rates in Phases I and II. For risk-adjusted p-charts, both multiplicative and additive models showed similar findings and no special cause variation.

CONCLUSION

Risk-adjusted control charts generated using a combination of multivariate logistic regression and control chart-adjustment models may provide a feasible method to assess the quality of care in the setting of PMV with standard care protocols.

ECMO PAL: using deep neural networks for survival prediction in venoarterial extracorporeal membrane oxygenation

PURPOSE

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is a complex and high-risk life support modality used in severe cardiorespiratory failure. ECMO survival scores are used clinically for patient prognostication and outcomes risk adjustment. This study aims to create the first artificial intelligence (AI)-driven ECMO survival score to predict in-hospital mortality based on a large international patient cohort.

METHODS

A deep neural network, ECMO Predictive Algorithm (ECMO PAL) was trained on a retrospective cohort of 18,167 patients from the international Extracorporeal Life Support Organisation (ELSO) registry (2017-2020), and performance was measured using fivefold cross-validation. External validation was performed on all adult registry patients from 2021 (N = 5015) and compared against existing prognostication scores: SAVE, Modified SAVE, and ECMO ACCEPTS for predicting in-hospital mortality.

RESULTS

Mean age was 56.8 ± 15.1 years, with 66.7% of patients being male and 50.2% having a pre-ECMO cardiac arrest. Cross-validation demonstrated an inhospital mortality sensitivity and precision of 82.1 ± 0.2% and 77.6 ± 0.2%, respectively. Validation accuracy was only 2.8% lower than training accuracy, reducing from 75.5% to 72.7% [99% confidence interval (CI) 71.1-74.3%]. ECMO PAL accuracy outperformed the ECMO ACCEPTS (54.7%), SAVE (61.1%), and Modified SAVE (62%) scores.

CONCLUSIONS

ECMO PAL is the first AI-powered ECMO survival score trained and validated on large international patient cohorts. ECMO PAL demonstrated high generalisability across ECMO regions and outperformed existing, widely used scores. Beyond ECMO, this study highlights how large international registry data can be leveraged for AI prognostication for complex critical care therapies.

Risk stratification models for predicting preventable hospitalization in commercially insured late middle-aged adults with depression

BACKGROUND

A significant number of late middle-aged adults with depression have a high illness burden resulting from chronic conditions which put them at high risk of hospitalization. Many late middle-aged adults are covered by commercial health insurance, but such insurance claims have not been used to identify the risk of hospitalization in individuals with depression. In the present study, we developed and validated a non-proprietary model to identify late middle-aged adults with depression at risk for hospitalization, using machine learning methods.

METHODS

This retrospective cohort study involved 71,682 commercially insured older adults aged 55-64 years diagnosed with depression. National health insurance claims were used to capture demographics, health care utilization, and health status during the base year. Health status was captured using 70 chronic health conditions, and 46 mental health conditions. The outcomes were 1- and 2-year preventable hospitalization. For each of our two outcomes, we evaluated seven modelling approaches: four prediction models utilized logistic regression with different combinations of predictors to evaluate the relative contribution of each group of variables, and three prediction models utilized machine learning approaches - logistic regression with LASSO penalty, random forests (RF), and gradient boosting machine (GBM).

RESULTS

Our predictive model for 1-year hospitalization achieved an AUC of 0.803, with a sensitivity of 72% and a specificity of 76% under the optimum threshold of 0.463, and our predictive model for 2-year hospitalization achieved an AUC of 0.793, with a sensitivity of 76% and a specificity of 71% under the optimum threshold of 0.452. For predicting both 1-year and 2-year risk of preventable hospitalization, our best performing models utilized the machine learning approach of logistic regression with LASSO penalty which outperformed more black-box machine learning models like RF and GBM.

CONCLUSIONS

Our study demonstrates the feasibility of identifying depressed middle-aged adults at higher risk of future hospitalization due to burden of chronic illnesses using basic demographic information and diagnosis codes recorded in health insurance claims. Identifying this population may assist health care planners in developing effective screening strategies and management approaches and in efficient allocation of public healthcare resources as this population transitions to publicly funded healthcare programs, e.g., Medicare in the US.

Improving diagnosis-based cost groups in the Dutch risk equalization model: the effects of a new clustering method and allowing for multimorbidity

Health insurance markets with community-rated premiums typically use risk equalization (RE) to compensate insurers for predictable profits on people in good health and predictable losses on those with a chronic disease. Over the past decades RE models have evolved from simple demographic models to sophisticated health-based models. Despite the improvements, however, non-trivial predictable profits and losses remain. This study examines to what extent the Dutch RE model can be further improved by redesigning one key morbidity adjuster: the Diagnosis-based Cost Groups (DCGs). This redesign includes (1) revision of the underlying hospital diagnoses and treatments ('dxgroups'), (2) application of a new clustering procedure, and (3) allowing multi-qualification. We combine data on spending, risk characteristics and hospital claims for all individuals with basic health insurance in the Netherlands in 2017 (N = 17 m) with morbidity data from general practitioners (GPs) for a subsample (N = 1.3 m). We first simulate a baseline RE model (i.e., the RE model of 2020) and then modify three important features of the DCGs. In a second step, we evaluate the effect of the modifications in terms of predictable profits and losses for subgroups of consumers that are potentially vulnerable to risk selection. While less prominent results are found for subgroups derived from the GP data, our results demonstrate substantial reductions in predictable profits and losses at the level of dxgroups and for individuals with multiple dxgroups. An important takeaway from our paper is that smart design of morbidity adjusters in RE can help mitigate selection incentives.

Do we care about high-cost patients? Estimating the savings on health spending by integrated care

A recent integrated health care initiative in Belgium supports 12 regional pilot projects scattered across the country and representing 21% of the population. As in shared savings programs, part of the estimated savings in health spending are paid out to the projects to reinvest in new actions. Short-term savings are expected in particular from cost reductions among high-cost patients. We estimate the effect of the projects on spending using a difference-in-difference model. The sensitivity of the results to the right-skewness of spending is commonly addressed by removing or top-coding high-cost cases. However, this leads to an underestimation of realized savings at the top end of the distribution, therefore, lowering incentives for cost reduction. We show that this trade-off can be weakened by an alternative approach in which cost categories that fall out of the scope of the projects' interventions are excluded from the dependent variable. We find that this approach leads to improvements in precision and model fit that are of the same magnitude as excluding high-cost cases altogether. At the same time, it sharpens the incentives for cost reduction because the model better reflects the costs that projects can affect.

Performance and analysis of four pediatric mortality prediction scores among critically ill children: A multicenter prospective observational study in four PICUs

OBJECTIVE

We aimed to evaluate and compare the prognostic performance of common pediatric mortality scoring systems (the Pediatric Index of Mortality 2 [PIM2], PIM3, Pediatric Risk of Mortality [PRISM], and PRISM4 scores) to determine which is the most applicable score in our pediatric study cohort.

METHODS

This prospective observational multicenter cohort study was conducted in four tertiary-care pediatric intensive care units (PICUs) in Turkey. All children, between 1 month and 16 years old, admitted to the participating PICUs between October 1, 2019, and March 31, 2020, were included in the study. Discrimination between death and survival was assessed by area under the receiver operating characteristic plot (AUC) for each model. The Hosmer-Lemeshow goodness-of-fit (GOF) test was used to assess the calibration of the models, RESULTS: A total of 570 patients (median age 35 months) were enrolled in the study. The observed mortality rate was 8.2% (47/570). The standardized mortality ratio (SMR) of PIM2, PIM3, PRISM, and PRISM4 with 95% confidence interval (CI) were 0.94 (0.68-1.23), 1.27 (0.93-1.68), 0.86 (0.63-1.13), and 1.5 (1.10-1.97), respectively. The AUC with 95% CI was 0.934 (0.91-0.96) for PIM2, 0.934 (0.91-0.96) for PIM3, 0.917 (0.88-0.95) for PRISM, and 0.926 (0.88-0.97) for PRISM4 models. The Hosmer-Lemeshow test showed that the difference between observed and predicted mortality by PIM3 (p = 0.003) and PRISM4 (p = 0.008) was statistically significant whereas PIM2 (p = 0.28) and PRISM (p = 0.62) showed good calibration.

CONCLUSION

The overall performance of (both discrimination and calibration) PRISM and PIM2 scoring systems in Turkish pediatric patients aged 1 month to 16 years was accurate and had the best fit for risk groups according to our study. Although PIM3 and PRISM4 have good discriminatory power, their calibration was very poor in our study cohort.

Interpretability, credibility, and usability of hospital-specific template matching versus regression-based hospital performance assessments; a multiple methods study

BACKGROUND

Hospital-specific template matching (HS-TM) is a newer method of hospital performance assessment.

OBJECTIVE

To assess the interpretability, credibility, and usability of HS-TM-based vs. regression-based performance assessments.

RESEARCH DESIGN

We surveyed hospital leaders (January-May 2021) and completed follow-up semi-structured interviews. Surveys included four hypothetical performance assessment vignettes, with method (HS-TM, regression) and hospital mortality randomized.

SUBJECTS

Nationwide Veterans Affairs Chiefs of Staff, Medicine, and Hospital Medicine.

MEASURES

Correct interpretation; self-rated confidence in interpretation; and self-rated trust in assessment (via survey). Concerns about credibility and main uses (via thematic analysis of interview transcripts).

RESULTS

In total, 84 participants completed 295 survey vignettes. Respondents correctly interpreted 81.8% HS-TM vs. 56.5% regression assessments, p < 0.001. Respondents "trusted the results" for 70.9% HS-TM vs. 58.2% regression assessments, p = 0.03. Nine concerns about credibility were identified: inadequate capture of case-mix and/or illness severity; inability to account for specialized programs (e.g., transplant center); comparison to geographically disparate hospitals; equating mortality with quality; lack of criterion standards; low power; comparison to dissimilar hospitals; generation of rankings; and lack of transparency. Five concerns were equally relevant to both methods, one more pertinent to HS-TM, and three more pertinent to regression. Assessments were mainly used to trigger further quality evaluation (a "check oil light") and motivate behavior change.

CONCLUSIONS

HS-TM-based performance assessments were more interpretable and more credible to VA hospital leaders than regression-based assessments. However, leaders had a similar set of concerns related to credibility for both methods and felt both were best used as a screen for further evaluation.

Donor-recipient risk assessment tools in heart transplant recipients: the Bad Oeynhausen experience

AIMS

Some risk assessment tools have been developed to categorize mortality risk in heart transplant recipients, but it is unclear whether these tools can be used interchangeable in different transplant regions.

METHODS AND RESULTS

We performed a retrospective single-centre study in 1049 adult German heart transplant recipients under jurisdiction of Eurotransplant. Univariable and multivariable Cox regression analysis was used to generate a risk scoring system. C-statistics were used to compare our score with a US score and a French score regarding their ability to discriminate between 1 year survivors and non-survivors within our study cohort. Of 38 parameters assessed, seven recipient-specific parameters [age, height, dilated cardiomyopathy (DCM), ischaemic cardiomyopathy (ICM), total bilirubin, extracorporeal membrane oxygenation (ECMO), and biventricular assist device/total artificial heart (BVAD/TAH) implant], one donor-specific parameter (cold ischaemic time), and one recipient-independent and donor-independent other parameter (late transplant era) were statistically significant in predicting 1 year mortality. The initial score was generated by using the regression coefficients from the multivariable analysis as follows: 1.70 * ln age - 4.0 * ln height - 0.9 * diagnosis (= 1 if diagnosis = DCM) - 0.67 * diagnosis (= 1 if diagnosis = ICM) + 0.33 * ln total bilirubin + 1.74 * ln cold ischaemic time + 0.98 * mechanical circulatory support (MCS) implant (= 1 if MCS implant = ECMO) + 0.47 * MCS implant (= 1 of MCS implant = BVAD/TAH) - 0.66 * transplant era (= 1 if transplant era = 2017-2018). The initial score was converted into the Bad Oeynhausen (BO) score as a positive integer variable by means of the following formula: BO score = (initial score + 8) * 3. In patients scoring 2 to <7 points (n = 112), 7 to <11 points (n = 580), 11 to <15 points (n = 339), and 15 to 20 points (n = 18), 1 year survival was 93.1%, 84.2%, 66.9%, and 27.8%, respectively. The c-index of our score was 0.73 [95% confidence interval (CI): 0.69-0.77]. Values were in our cohort for the US and French scores 0.66 (95% CI: 0.62-0.70) and 0.63 (95% CI: 0.59-0.67), respectively.

CONCLUSIONS

Data indicate that our score, but also risk assessment tools from other transplant regions, may be used as a reliable support for risk-adjusted organ allocation and potentially help to improve outcomes in heart transplantation. Further developments will have to include as yet unaccounted risk factors for even more reliable predictions.

Impact of postnatal age on neonatal intensive care unit bloodstream infection reporting

Due to their short- and long-term impact on patients in the neonatal intensive care unit (NICU), bloodstream infections are a closely monitored quality measure. NICU infection rates are risk-adjusted for birth weight, but not postnatal age. Our findings suggest that infection rates are not constant over time in neonates with long NICU lengths of stay and adjusting for postnatal age in addition to birth weight may improve unit comparisons.

How the COVID-19 pandemic can distort risk adjustment of health plan payment

The COVID-19 pandemic has led to disruptions in healthcare utilization and spending. While some changes might persist (e.g. substitution of specialist visits by online consultations), others will be transitory (e.g. fewer surgical procedures due to cancellation of treatments). This paper discusses the implications of transitory changes in healthcare utilization and spending for risk adjustment of health plan payment. In practice, risk adjustment methodologies typically consist of two steps: (1) calibration of payment weights for a given set of risk adjusters and (2) calculation of payments to insurers by combining the calibrated weights with enrollee characteristics. In this paper, we first introduce a simple conceptual framework for analyzing the (potential) distortions from the pandemic for both steps and then provide a hypothetical illustration of how these distortions can lead to under- or overpayment of insurers. The size of these under-/overpayments depends on (1) the impact of the pandemic on patterns in utilization and spending, (2) the distribution of risk types across insurers, (3) the extent to which insurers are disproportionately affected by the pandemic, and (4) features of the risk adjustment system.

Measuring patient satisfaction in acute care hospitals: nationwide monitoring in Switzerland

The National Association for Quality Development in Hospitals and Clinics (ANQ) has conducted patient satisfaction measurements in the inpatient sector in Switzerland since 2009. Specifically designed for this measurement, an instrument consisting of five questions was evaluated on an 11-point rating scale. Nevertheless, the instrument showed substantial ceiling effects, which did not allow for hospital discrimination. Therefore, ANQ initiated a revision testing different scales in a pilot study. The results showed that a 5-point verbal scale displayed good psychometric properties. Compared to the 7- or 11-point scales, the 5-point verbal scale exhibited reduced ceiling effects, which was more appropriate to compare hospitals. For the national public reporting of hospitals and clinics, risk adjustment by age and self-reported health status was recommended, which was not the case for gender, principal diagnosis, type of admission and insurance status.

Risk-adjusted zero-inflated Poisson CUSUM charts for monitoring influenza surveillance data

Background

The influenza surveillance has been received much attention in public health area. For the cases with excessive zeroes, the zero-inflated Poisson process is widely used. However, the traditional control charts based on zero-inflated Poisson model, ignore the association between influenza cases and risk factors, and thus may lead to unexpected mistakes when implementing monitoring charts.

Method

In this paper, we proposed risk-adjusted zero-inflated Poisson cumulative sum control charts, in which the risk factors were put to adjust the risk of influenza and the adjustment was made by zero-inflated Poisson regression. We respectively proposed the control chart monitoring the parameters individually and simultaneously.

Results

The performance of our proposed risk-adjusted zero-inflated Poisson cumulative sum control chart was evaluated and compared with the unadjusted standard cumulative sum control charts in simulation studies. The results show that for different distribution of impact factors and different coefficients, the risk-adjusted cumulative sum charts can generate much less false alarm than the standard ones. Finally, the influenza surveillance data from Hong Kong is used to illustrate the application of the proposed chart.

Conclusions

Our results suggest that the adjusted cumulative sum control chart we proposed is more accurate and credible than the unadjusted standard control charts because of the lower false alarm rate of the adjusted ones. Even the unadjusted control charts may signal a little faster than the adjusted ones, the alarm they raise may have low credibility since they also raise alarm frequently even the processes are in control. Thus we suggest using the risk-adjusted cumulative sum control charts to monitor the influenza surveillance data to alert accurately, credibly and relatively quickly.

Assessing the performance of a method for case-mix adjustment in the Korean Diagnosis-Related Groups (KDRG) system and its policy implications

Background

To evaluate the performance of the patient clinical complexity level (PCCL) mechanism, which is the patient-level complexity adjustment factor within the Korean Diagnosis-Related Groups (KDRG) patient classification system, in explaining the variation in resource consumption within age adjacent diagnosis-related groups (AADRGs).

Methods

We used the inpatient claims data from a public hospital in Korea from 1 January 2017 to 30 June 2019, with 18 846 claims and 138 AADRGs. The differences in the total average payment between the four PCCL levels for each AADRG was tested using ANOVA and Duncan’s post hoc test. The three patterns of differences with R-squared were as follows: the PCCL reflected the complexity well (valid); the average payment for PCCL 2, 3, and 4 was greater than PCCL 0 (partially valid); the PCCL did not reflect the complexity (not valid).

Results

There were 9 (6.52%), 26 (18.84%), and 103 (74.64%) ADRGs included in the valid, partially valid, and not valid categories, respectively. The average R-squared values were 32.18, 40.81, and 35.41%, respectively, with an average R-squared for all patterns of 36.21%.

Conclusions

Adjustment using the PCCL in the KDRG classification system exhibited low performance in explaining the variation in resource consumption within AADRGs. As the KDRG classification system is used for reimbursement under the new DRG-based prospective payment system (PPS) pilot project, with plans for expansion, there should be an overall review of the validity of the complexity and rationality of using the KDRG classification system.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12961-021-00739-5.

[Opportunities and limitations of risk adjustment of quality indicators based on inpatient administrative health data - a workshop report]

INTRODUCTION

The quality indicators of the Initiative Qualitätsmedizin e. V. (IQM) have been developed as triggers to examine treatment processes for opportunities for improvement. Published quality results have partly been used for external quality comparisons in the media. Therefore, member hospitals of IQM demanded to investigate if methods of risk adjustment should be applied in the calculation of the quality indicators. After a hearing of experts had been held, a task force was founded to conduct test calculations on risk adjustment methods.

METHODS

Specific risk adjustment models for mortality in myocardial infarction, heart failure, stroke, pneumonia, and colectomy in colorectal cancer were developed in the database of national German DRG data of the year 2016. These models were used to calculate standardized mortality ratios (SMR) per indicator in a sample of 172 member hospitals of IQM based on the data of the year 2018. Median SMR per indicator were compared to median SMR based on a standardization by age and gender, which is the standard procedure in IQM. Correlations between the different SMR were calculated. Quality of care was judged by two different approaches: a) a descriptive discrepancy of |0.1| from the SMR value of 1, and b) a significant discrepancy from 1 using the 95% confidence limits. The effect of using the specific risk adjustment in relation to the standard procedure was investigated for both approaches (a and b).

RESULTS

The specific risk adjustment methods showed an area under the curve between 0.72 and 0.84. The median differences between the SMR based on standardization by age and gender and the SMR based on specific risk adjustment were small (between 0 and 0.4); Spearman's correlations were between 0.90 and 0.99. Changes in the judgement of quality of care in comparison to the national average occurred in 3.9% (mortality from pneumonia) to 20.6% of the hospitals (mortality from heart failure) in descriptive comparisons. When the judgement was based on confidence limits changes were observed in 1.6% (mortality after colectomy) to 17.4% of the hospitals (mortality from heart failure).

DISCUSSION

Implementing specific risk adjustment models had only minor effects on the distribution of risk-adjusted mortality compared to the standard procedure, but the judgement of quality of care could change for a fifth of the hospitals in individual indicators. Concerning methodological and practical reasons, the task force recommends further development of risk adjustment methods for selected indicators. This should be accompanied by studies on the validity of inpatient administrative data for quality management as well as by efforts to improve the usefulness of these data for such purposes.

Risk-adjusted rates between hospitals for adverse outcomes of very-low-birth-weight infants

BACKGROUND/PURPOSE

To analyze the amount of variation in these risk-adjusted adverse outcomes corresponding to the care of premature births. In addition, hospitals were ranked according to their unadjusted and adjusted rates, and we assessed the degree of concordance between these rankings. Finally, the correlations of hospital-adjusted adverse outcomes were also tested.

METHODS

The study utilized the 5-year Taiwan Premature Infant Follow-up Network (TPFN) database in Taiwan from 2014 to 2018, and the sample size was 6482. We calculated the "observed over expected" (OE) ratio every year to form the risk-adjusted adverse outcome rate for each hospital.

RESULTS

There was a larger variation in the risk-adjusted rate for NEC and the second-largest variation for IVH. Regarding the concordances between the unadjusted and adjusted ranks, the ranks for mortality had the lowest concordance (coefficient of concordance 0.64), and only a few of the risk-adjusted rates between outcomes were significantly correlated.

CONCLUSION

The results of the TPFN show that there is room to improve performance in terms of large variations in NEC and IVH. Furthermore, risk adjustment is important, especially for mortality, since the ranks for mortality have the lowest concordance. Finally, we cannot generate a conclusion regarding whether a hospital is high in quality if we only take 1 or 2 adverse outcomes as profiling measures because only a few of the risk-adjusted rates between outcomes were significantly correlated.

Variability in outcomes and quality-of-care indicators across clinics participating in a large smoking-cessation program

BACKGROUND

The effectiveness of care for substance-related problems varies across providers. Best-known treatments are rarely universally applied, and various process differences can affect participant outcomes. Measuring and understanding this variability can suggest changes that will improve system performance.

METHODS

We measure variability in 7-day cigarette abstinence at a six-month follow-up; return for a second clinical visit; and receipt of combination nicotine replacement therapy across 223 primary care clinics participating in the Smoking Treatment for Ontario Patients program, a large smoking cessation initiative in Ontario, Canada. We include 41,992 enrolments from April 11, 2016 and May 31, 2019. We risk adjust for demographic and clinical case-mix differences and characterize variability using funnel plots and measures based on clinic-level variance explained. The abstinence outcome is missing for 38% of participants. We adjust for missingness using multiple imputation and inverse probability weighting.

RESULTS

Abstinence was achieved by 28.0% (95% CI = 27.5%-28.5%) of participants, 63.2% (62.8%-63.7%) received combination NRT, and 72.9% (72.4%-73.3%) returned for a second clinical visit. Variability was moderate for abstinence (median odds ratio (MOR) = 1.16) and pronounced for return visit (MOR = 1.29) and combination therapy (MOR = 1.89).

CONCLUSION

Outcomes and processes vary significantly across clinics within a program with shared guidelines and standards. Differences across providers may be greater in other contexts. Results underscore the importance of measuring and understanding variability, and of ongoing maintenance and improvement. The existence of high outliers holds out the possibility of identifying practices that might be more widely adopted.

Very high and low residual spenders in private health insurance markets: Germany, The Netherlands and the U.S. Marketplaces

We study the extremely high and low residual spenders in individual health insurance markets in three countries. A high (low) residual spender is someone for whom the residual-spending less payment (from premiums and risk adjustment)-is high (low), indicating that the person is highly underpaid (overpaid). We begin with descriptive analysis of the top and bottom 1% and 0.1% of residuals building to address the question of the degree of persistence in membership at the extremes. Common findings emerge among the countries. First, the diseases found among those with the highest residual spending are also disproportionately found among those with the lowest residual spending. Second, those at the top of the residual spending distribution (where spending exceeds payments the most) account for a massively high share of the unexplained variance in the predictions from the risk adjustment model. Third, in terms of persistence, we find that membership in the extremes of the residual spending distribution is highly persistent, raising concerns about selection-related incentives targeting these individuals. As our results show, the one-in-a-thousand people (on both sides of the residual distribution) play an outsized role in creating adverse incentives associated with health plan payment systems. In response to the observed importance of the extremes of the residual spending distribution, we propose an innovative combination of risk-pooling and reinsurance targeting the predictively undercompensated group. In all three countries, this form of risk sharing substantially improves the overall fit of payments to spending. Perhaps surprisingly, by reducing the burden on diagnostic indicators to predict high payments, our proposed risk sharing policy reduces the gap between payments and spending not only for the most undercompensated individuals but also for the most overcompensated people.

Validation of the geriatric trauma outcome scores in predicting outcomes of elderly trauma patients

BACKGROUND

Using three patient characteristics, including age, Injury Severity Score (ISS) and transfusion within 24 h of admission (yes vs. no), the Geriatric Trauma Outcome Score (GTOS) and Geriatric Trauma Outcome Score II (GTOS II) have been developed to predict mortality and unfavourable discharge (to a nursing home or hospice facility), of those who were ≥65 years old, respectively.

OBJECTIVES

This study aimed to validate the GTOS and GTOS II models. For the nested-cohort requiring intensive care, we compared the GTOS scores with two ICU prognostic scores - the Acute Physiology and Chronic Health Evaluation (APACHE) III and Australian and New Zealand Risk of Death (ANZROD).

METHODS

All elderly trauma patients admitted to the State Trauma Unit between 2009 and 2019 were included. The discrimination ability and calibration of the GTOS and GTOS II scores were assessed by the area under the receiver-operating-characteristic (AUROC) curve and a calibration plot, respectively.

RESULTS

Of the 57,473 trauma admissions during the study period, 15,034 (26.2%) were ≥65 years-old. The median age and ISS of the cohort were 80 (interquartile range [IQR] 72-87) and 6 (IQR 2-9), respectively; and the average observed mortality was 4.3%. The ability of the GTOS to predict mortality was good (AUROC 0.838, 95% confidence interval [CI] 0.821-0.855), and better than either age (AUROC 0.603, 95%CI 0.581-0.624) or ISS (AUROC 0.799, 95%CI 0.779-0.819) alone. The GTOS II's ability to predict unfavourable discharge was satisfactory (AUROC 0.707, 95%CI 0.696-0.719) but no better than age alone. Both GTOS and GTOS II scores over-estimated risks of the adverse outcome when the predicted risks were high. The GTOS score (AUROC 0.683, 95%CI 0.591-0.775) was also inferior to the APACHE III (AUROC 0.783, 95%CI 0.699-0.867) or ANZROD (AUROC 0.788, 95%CI 0.705-0.870) in predicting mortality for those requiring intensive care.

CONCLUSIONS

The GTOS scores had a good ability to discriminate between survivors and non-survivors in the elderly trauma patients, but GTOS II scores were no better than age alone in predicting unfavourable discharge. Both GTOS and GTOS II scores were not well-calibrated when the predicted risks of adverse outcome were high.

Comparing Charlson and Elixhauser comorbidity indices with different weightings to predict in-hospital mortality: an analysis of national inpatient data

Background

Understanding how comorbidity measures contribute to patient mortality is essential both to describe patient health status and to adjust for risks and potential confounding. The Charlson and Elixhauser comorbidity indices are well-established for risk adjustment and mortality prediction. Still, a different set of comorbidity weights might improve the prediction of in-hospital mortality. The present study, therefore, aimed to derive a set of new Swiss Elixhauser comorbidity weightings, to validate and compare them against those of the Charlson and Elixhauser-based van Walraven weights in an adult in-patient population-based cohort of general hospitals.

Methods

Retrospective analysis was conducted with routine data of 102 Swiss general hospitals (2012–2017) for 6.09 million inpatient cases. To derive the Swiss weightings for the Elixhauser comorbidity index, we randomly halved the inpatient data and validated the results of part 1 alongside the established weighting systems in part 2, to predict in-hospital mortality. Charlson and van Walraven weights were applied to Charlson and Elixhauser comorbidity indices. Derivation and validation of weightings were conducted with generalized additive models adjusted for age, gender and hospital types.

Results

Overall, the Elixhauser indices, c-statistic with Swiss weights (0.867, 95% CI, 0.865–0.868) and van Walraven’s weights (0.863, 95% CI, 0.862–0.864) had substantial advantage over Charlson’s weights (0.850, 95% CI, 0.849–0.851) and in the derivation and validation groups. The net reclassification improvement of new Swiss weights improved the predictive performance by 1.6% on the Elixhauser-van Walraven and 4.9% on the Charlson weights.

Conclusions

All weightings confirmed previous results with the national dataset. The new Swiss weightings model improved slightly the prediction of in-hospital mortality in Swiss hospitals. The newly derive weights support patient population-based analysis of in-hospital mortality and seek country or specific cohort-based weightings.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-020-05999-5.

Prediction model for prolonged length of stay in patients with community-acquired pneumonia based on Japanese administrative data

BACKGROUND

The length of hospital stay in community-acquired pneumonia patients is closely associated with medical costs, the burden of which is increasing in aging societies. Herein, we developed and validated models for predicting prolonged length of stay in community-acquired pneumonia patients to support efficient care in these patients.

METHODS

We obtained data of 32,916 patients hospitalized for pneumonia who were discharged between 2012 and 2013 from 304 acute care hospitals in Japan. Logistic regression models were developed with prolonged length of stay as the outcome and patient characteristics as predictors. The models were internally validated using bootstrapping and externally validated using pneumonia patients discharged in 2014.

RESULTS

The median length of stay was 11 (interquartile range, 8-17) days. The following were significant predictors of prolonged length of stay (odds ratio >1.6): age ≥75 years, Barthel index score ≤6, fraction of inspired oxygen ≥35%, Japan Coma Scale score of 100-300, anemia, muscle wasting and atrophy, bedsores, dysphasia, and methicillin-resistant Staphylococcus aureus infection. Our validation models had a c-statistic of 0.78 (95% confidence interval, 0.77-0.79) and a calibration slope of 0.98.

CONCLUSIONS

Our prediction models may help policymakers in developing strategies for the optimal management of community-acquired pneumonia patients with a focus on patients at a high risk of prolonged length of stay.

Impact Of Medicare Readmissions Penalties On Targeted Surgical Conditions

The Hospital Readmissions Reduction Program, announced in 2010 to penalize excess readmissions for patients with selected medical diagnoses, was expanded in 2013 to include targeted surgical diagnoses, beginning with hip and knee replacements. Whether these surgical penalties reduced procedure-specific readmissions is not well understood. Using Medicare claims, we evaluated the penalty announcements' effects on risk-adjusted readmission rates, episode payments, lengths-of-stay, and observation status use. Risk-adjusted readmission rates declined for both procedures from 7.6 percent in 2008 to 5.5 percent in 2016. These rates were decreasing before the program was announced, but the rate of reductions doubled after the announcement of medical penalties in March 2010 (from -0.05 percentage points to -0.10 percentage points per quarter). After targeted surgical penalties were announced in August 2013, readmission reductions returned to near the baseline trend. During the same time period, mean episode payments and lengths-of-stay decreased substantially, and trends in observation status were unchanged. This suggests that medical readmission penalties led to readmission reductions for surgical patients as well, that targeted surgical penalties did not have an additional effect, and that readmission reductions are approaching a "floor" below which further reductions may be unlikely.

Improving risk-equalization in Switzerland: Effects of alternative reform proposals on reallocating public subsidies for hospitals

The Swiss healthcare financing system is on the verge of one of its largest reforms. The Swiss parliament is currently debating how to reallocate about 20 % of total health expenditures. Swiss cantons make substantial tax-funded contributions to health expenditures by paying 55 % of hospital inpatient costs. As health insurers are fully responsible for all outpatient costs, the present system may provide unintended incentives to treat patients in inpatient settings. This paper presents and evaluates three alternative reform proposals for the reallocation of the cantonal contribution. Two proposals are currently under consideration in the Swiss parliament, suggesting either partial cost-sharing (20 %) of all healthcare costs or inclusion of cantonal contributions into the risk-equalization fund. A third option is developed in this paper, which proposes using the cantonal funds to pay a share of insurer's expenses above a high-cost threshold. The high-cost risk-sharing alternative is clearly superior: it mitigates the incentive to discriminate against sicker individuals, improves incentives for cost control, and reduces risk of loss for insurers. The paper adds results from Switzerland to an international literature on the properties of adding high-cost risk sharing to a risk equalization model.

Costs Are Higher For Marketplace Members Who Enroll During Special Enrollment Periods Compared With Open Enrollment

More than 20 percent of Affordable Care Act (ACA) exchange market (Marketplace) members insured by a large national insurer in 2015 and 2016 enrolled during a special enrollment period (SEP), defined as any enrollment outside the annual open enrollment period. These members were younger and had approximately 34 percent higher average monthly total costs than members who enrolled during open enrollment. SEP members had 69-114 percent higher inpatient costs and 11-19 percent higher emergency department costs than open enrollment members. Higher costs, especially among a slightly younger population, may suggest potential adverse selection among SEP members, which could contribute to increased premiums and insurer exit from ACA Marketplaces. Although SEP members had a shorter average enrollment length per calendar year, they were more likely than open enrollment members to stay insured through the end of the calendar year and to renew in a Marketplace plan offered by the insurer in the following year. However, renewing SEP and open enrollment members were older, sicker, and costlier than nonrenewing members of both enrollee types, which suggests that healthier members are switching carriers or leaving the market over time. Additional research is urgently needed to inform evidence-based policy regarding Marketplace risk adjustment and SEP eligibility rules and to improve outreach to people who are eligible for SEP enrollment.

Comparison of individual and neighbourhood socioeconomic status in case mix adjustment of hospital performance in primary total hip replacement in Sweden: a register-based study

BACKGROUND

Case mix adjustment is a pre-requisite for valid measurement of healthcare performance and socioeconomic status (SES) is important to account for. Lack of information on individual-level SES has led to investigations into using a proxy for SES based on patient area of residence. The objective of this study was to use neighbourhood SES for case mix adjustment of performance indicators in total hip replacement (THR) in Sweden, and to compare with use of individual SES.

METHODS

Data from patient administrative systems and the Swedish Hip Arthroplasty Register were extracted for all patients undergoing THR in four Swedish regions. For each subject, individual data and neighbourhood data on country of birth, educational level, and income were provided by Statistics Sweden. Three variables were selected for analysis of performance; EQ-5D, hip pain and length of stay (LoS). In addition to socioeconomic information, several important clinical characteristics were used as case mix factors. Regression analysis was used to study each variable's impact on the three outcome variables and model fit was evaluated using mean squared error.

RESULTS

A total of 27,121 patients operated between 2010 and 2016 were included in the study. Both educational level and income were higher when based on neighbourhood information than individual information, while proportion born in Sweden was similar. Higher SES was generally found to be associated with better outcomes and lower LoS, albeit with certain differences between the different measures of SES. The predictive ability of the models was increased when adding information on SES to the clinical characteristics. The increase in predictive ability was higher for individual SES compared to neighbourhood SES. When analysing performance for the two providers with most diverging case mix in terms of SES, the inclusion of SES altered the relative performance using individual as well as neighbourhood SES.

CONCLUSIONS

Incorporating SES improves case mix adjustment marginally compared to using only clinical information. In this patient group, geographically derived SES was found to improve case mix adjustment compared to only clinical information but not to the same extent as actual individual-level SES.

Population segments as a tool for health care performance reporting: an exploratory study in the Canadian province of British Columbia

Background

Primary care serves all age groups and individuals with health states ranging from those with no chronic conditions to those who are medically complex, or frail and approaching the end of life. For information to be actionable and guide planning, there must be some population disaggregation based on differences in expected needs for care. Promising approaches to segmentation in primary care reflect both the breadth and severity of health states, the types and amounts of health care utilization that are expected, and the roles of the primary care provider. The purpose of this study was to assess population segmentation as a tool to create distinct patient groups for use in primary care performance reporting.

Methods

This cross-sectional study used administrative data (patient characteristics, physician and hospital billings, prescription medicines data, emergency department visits) to classify the population of British Columbia (BC), Canada into one of four population segments: low need, multiple morbidities, medically complex, and frail. Each segment was further classified using socioeconomic status (SES) as a proxy for patient vulnerability. Regression analyses were used to examine predictors of health care use, costs and selected measures of primary care attributes (access, continuity, coordination) by segment.

Results

Average annual health care costs increased from the low need ($ 1460) to frail segment ($10,798). Differences in primary care cost by segment only emerged when attributes of primary care were included in regression models: accessing primary care outside business hours and discontinuous primary care (≥5 different GP’s in a given year) were associated with higher health care costs across all segments and higher continuity of care was associated with lower costs in the frail segment (cost ratio = 0.61). Additionally, low SES was associated with higher costs across all segments, but the difference was largest in the medically complex group (cost ratio = 1.11).

Conclusions

Population segments based on expected need for care can support primary care measurement and reporting by identifying nuances which may be lost when all patients are grouped together. Our findings demonstrate that variables such as SES and use of regression analyses can further enhance the usefulness of segments for performance measurement and reporting.

Toward risk adjustment in mental health in Israel: calculation of risk adjustment rates from large outpatient and inpatient databases

BACKGROUND

In 2015, mental health services were added to the Israeli National Health Insurance package of services. As such, these services are financed by the budget which is allocated to the Health Plans according to a risk adjustment scheme. An inter-ministerial team suggested a formula by which the mental health budget should be allocated among the Health Plans. Our objective in this study was to develop alternative rates based on individual data, and to evaluate the ones suggested.

METHODS

The derivation of the new formula is based on our previous study of all psychiatric inpatients in Israel in the years 2012-2013 (n = 27,446), as well as outpatients in one psychiatric clinic in the same period (n = 6115). Based on Ministry of Health and clinic data we identified predictors of mental health services consumption. Age, gender, marital status and diagnosis were used as risk adjusters to calculate the capitation rates for outpatient care and inpatient care, respectively. All prices of services were obtained from the Ministry of Health tariffs. These rates were modified to include non-users using restricted models.

RESULTS

The mental health capitation scales are typically "humped" with regard to age. The rates for ambulatory care varied from a minimum 0.19 of the average consumption for males above the age of 85 to a maximum of 1.93 times the average for females between the ages of 45-54. For inpatient services the highest rate was 409.25 times the average for single, male patients with schizophrenia spectrum diagnoses, aged 45-54. The overall mental health scale ranges from 2.347 times the average for men aged 45-54, to 0.191 for women aged 85+. The modified scale for the entire post-reform package of benefits (including both mental health care and physical health care) is increasing with age to 4.094 times the average in men aged over 85. The scale is flatter than the pre-reform scale.

CONCLUSIONS

The risk adjustment rates calculated for outpatient care are substantially different from the ones suggested by the inter-ministerial team. The inpatient rates are new, and indicate that for patients with schizophrenia, a separate risk-sharing arrangement might be desirable. Adopting the rates developed in this analysis would decrease the budget shares of Clalit and Leumit with their relatively older populations, and increase Maccabi and Meuhedet's shares. Future research should develop a risk-adjustment scheme which covers directly both mental and physical care provided by the Israeli Health Plans, using their data.

Hospital Volume Threshold Associated with Higher Survival after Endovascular Recanalization Therapy for Acute Ischemic Stroke

BACKGROUND AND PURPOSE

Endovascular recanalization therapy (ERT) is becoming increasingly important in the management of acute ischemic stroke (AIS). However, the hospital volume threshold for optimal ERT remains unknown. We investigated the relationship between hospital volume of ERT and risk-adjusted patient outcomes.

METHODS

From the National Health Insurance claims data in Korea, 11,745 patients with AIS who underwent ERT from July 2011 to June 2016 in 111 hospitals were selected. We measured the hospital's ERT volume and patient outcomes, including the 30-day mortality, readmission, and postprocedural intracranial hemorrhage (ICH) rates. For each outcome measure, we constructed risk-adjusted prediction models incorporating demographic variables, the modified Charlson comorbidity index, and the stroke severity index (SSI), and validated them. Risk-adjusted outcomes of AIS cases were compared across hospital quartiles to confirm the volume-outcome relationship (VOR) in ERT. Spline regression was performed to determine the volume threshold.

RESULTS

The mean AIS volume was 14.8 cases per hospital/year and the unadjusted means of mortality, readmission, and ICH rates were 11.6%, 4.6%, and 8.6%, respectively. The VOR was observed in the risk-adjusted 30-day mortality rate across all quartile groups, and in the ICH rate between the first and fourth quartiles (P<0.05). The volume threshold was 24 cases per year.

CONCLUSIONS

There was an association between hospital volume and outcomes, and the volume threshold in ERT was identified. Policies should be developed to ensure the implementation of the AIS volume threshold for hospitals performing ERT.

Risk Adjustment, Mode Adjustment, and Nonresponse Bias Analysis on Quality Measures From a Long-Term Care Hospital Experience of Care Survey

OBJECTIVES

To develop a patient risk adjustment model for experience of care (EOC) quality measures for long-term care hospitals (LTCHs) that includes mode of survey administration. To assess presence of nonresponse bias in the adjusted facility-level scores.

DESIGN

We tested 3 modes of collecting the EOC data: mail-only, mixed (ie, mail with telephone follow-up), and in-facility. This study used sequential modeling and impact analysis, specified a risk and mode adjustment model, and evaluated presence of nonresponse after adjustment.

SETTING

LTCHs.

PARTICIPANTS

Patients (N=1364) and 69 LTCHs.

INTERVENTION

Not applicable.

MAIN OUTCOME MEASURES

Risk and mode adjusted responses to 28 survey questions and 6 facility-level scores derived from survey responses.

RESULTS

Mode of data collection and patient risk variables (age, sex, overall health, overall mental health, marital status, education, race, and whether a proxy responded) were included in the model. Clinical variables were not significant. The in-facility mode was associated with significantly higher performance scores than the other modes. When the recommended risk and mode adjustment model was applied, nonresponse bias was not observed in any mode.

CONCLUSIONS

LTCH EOC data should be adjusted for patient risk variables including mode of data collection.

County Rankings Have Limited Utility When Predicting Liver Transplant Outcomes

BACKGROUND

Evidence of geographical differences in liver transplantation (LT) outcomes has been proposed as a reason to include community characteristics in risk adjustment of transplant quality metrics. However, consistency and utility of rankings in LT outcomes for counties have not been demonstrated.

AIMS

We sought to evaluate the utility of county rankings (county socioeconomic status (SES) or county health scores (CHS)) on outcomes after LT.

METHODS

Using the United Network for Organ Sharing Registry, adults ≥ 18 years of age undergoing LT between 2002 and 2014 were identified. County-specific 1-year survival was calculated using the Kaplan-Meier method for counties with ≥ 5 LT performed during this period. Agreement between high-risk designation by 1-year mortality rate and county ranking was calculated using the Spearman correlation coefficient.

RESULTS

The analysis included 47,769 LT recipients in 1092 counties. County 1-year mortality rates were not correlated with county CHS (Spearman ρ = 0.01, p = 0.694) or county SES (Spearman ρ = - 0.01, p = 0.734). After controlling for individual-level covariates, a statistically significant variability in mortality hazards across counties (p < 0.001) persisted. Although both CHS and SES measures improved the model fit (p = 0.004 and p = 0.048, respectively), an unexplained residual variation in mortality hazard across counties continued.

CONCLUSIONS

There is poor agreement between county rankings on various socioeconomic indicators and LT outcomes. Although there is variability in outcomes across counties, this appears not to be due to county-level socioeconomic indices.

Is it feasible to pay specialty substance use disorder treatment programs based on patient outcomes?

BACKGROUND

Some US payers are starting to vary payment to providers depending on patient outcomes, but this approach is rarely used in substance use disorder (SUD) treatment.

PURPOSE

We examine the feasibility of applying a pay-for-outcomes approach to SUD treatment.

METHODS

We reviewed several relevant literatures: (1) economic theory papers that describe the conditions under which pay-for-outcomes is feasible in principle; (2) description of the key outcomes expected from SUD treatment, and the measures of these outcomes that are available in administrative data systems; and (3) reports on actual experiences of paying SUD treatment providers based on patient outcomes.

RESULTS

The economics literature notes that when patient outcomes are strongly influenced by factors beyond provider control and when risk adjustment performs poorly, pay-for-outcomes will increase provider financial risk. This is relevant to SUD treatment. The literature on SUD outcome measurement shows disagreement on whether to include broader outcomes beyond abstinence from substance use. Good measures are available for some of these broader constructs, but the need for risk adjustment still brings many challenges. Results from two past payment experiments in SUD treatment reinforce some of the concerns raised in the more conceptual literature.

CONCLUSION

There are special challenges in applying pay-for-outcomes to SUD treatment, not all of which could be overcome by developing better measures. For SUD treatment it may be necessary to define outcomes more broadly than for general medical care, and to continue conditioning a sizeable portion of payment on process measures.

Are current DRG-based bundled payment models for lumbar fusions risk-adjusting adequately? An analysis of Medicare beneficiaries

BACKGROUND CONTEXT

Current bundled payment programs in spine surgery, such as the bundled payment for care improvement rely on the use of diagnosis-related groups (DRG) to define payments. However, these DRGs may not be adequate enough to appropriately capture the large amount of variation seen in spine procedures. For example, DRG 459 (spinal fusion except cervical with major comorbidity or complication) and DRG 460 (spinal fusion except cervical without major comorbidity or complication) do not differentiate between the type of fusion (anterior or posterior), the levels/extent of fusion, the use of interbody/graft/BMP, indication of surgery (primary vs. revision) or even if the surgery was being performed for a vertebral fracture.

PURPOSE

We carried out a comprehensive analysis to report the factors responsible for cost-variation in a bundled payment model for spinal fusions.

STUDY DESIGN

Retrospective review of a 5% national sample of Medicare claims from 2008 to 2014 (SAF5).

OUTCOME MEASURES

To understand the independent marginal cost impact of various patient-level, geographic-level, and procedure-level characteristics on 90-day costs for patients undergoing spinal fusions under DRG 459 and 460.

METHODS

The 2008 to 2014 Medicare 5% standard analytical files (SAF) were used to retrieve patients undergoing spinal fusions under DRG 459 and DRG 460 only. Patients with missing gender, age, and/or state-level data were excluded. Only those patients who had complete data, with regard to payments/costs/reimbursements, starting from day 0 of surgery up to 90 days postoperatively were included to prevent erroneous collection. Multivariate linear regression models were built to assess the independent marginal cost impact (decrease/increase) of each patient-level, state-level, and procedure-level characteristics on the average 90-day cost while controlling for other covariates.

RESULTS

A total of 21,367 patients (DRG-460=20,154; DRG-459=1,213) were included in the study. The average 90-day cost for all lumbar fusions was $31,716±$18,124, with the individual 90-day payments being $54,607±$30,643 (DRG-459) and $30,338±$16,074 (DRG-460). Increasing age was associated with significant marginal increases in 90-day payments (70-74 years: +$2,387, 75-79 years: +$3,389, 80-84 years: +$2,872, ≥85: +$1,627). With regards to procedure-level factors-undergoing an anterior fusion (+$3,118), >3 level fusion (+$5,648) vs. 1 to 3 level fusion, use of interbody device (+$581), intraoperative neuromonitoring (+$1,413), concurrent decompression (+$768) and undergoing surgery for thoracolumbar fracture (+$6,169) were associated with higher 90-day costs. Most individual comorbidities were associated with higher 90-day costs, with malnutrition (+$12,264), CVA/stroke (+$5,886), Alzheimer's (+$4,968), Parkinson's disease (+$4,415), and coagulopathy (+$3,810) having the highest marginal 90-day cost-increases. The top five states with the highest marginal cost-increase, in comparison to Michigan (reference), were Maryland (+$12,657), Alaska (+$11,292), California (+$10,040), Massachusetts (+$8,800), and the District of Columbia (+$8,315).

CONCLUSIONS

Under the proposed DRG-based bundled payment model, providers would be reimbursed the same amount for lumbar fusions regardless of the surgical approach (posterior vs. anterior), the extent of fusion (1-3 level vs. >3 level), use of adjunct procedures (decompressions) and cause/indication of surgery (fracture vs. degenerative pathology), despite each of these factors having different resource utilization and associated costs. When defining and developing future bundled payments for spinal fusions, health-policy makers should strive to account for the individual patient-level, state-level, and procedure-level variation seen within DRGs to prevent the creation of a financial dis-incentive in taking care of sicker patients and/or performing more extensive complex spinal fusions.

Service- and practitioner-level variation in non-consensual dropout from child mental health services

Non-attendance of mental health service appointments is an international problem. In the UK, for example, the estimated cost of non-attendance in child mental health services is over £45 million (US dollar 60.94 million) per annum. The objective of this study was to examine whether there were service- and practitioner-level variation in non-consensual dropout in child mental health services. This was an analysis of routinely collected data. Service-level variation (as services covered different geographic areas) and practitioner-level variation were examined in N = 3622 children (mean age 12.70 years; SD 3.62, 57% female, 50% white or white British) seen by 896 practitioners across 39 services. Overall, 35% of the variation in non-consensual dropout was explained at the service level and 15% at the practitioner level. Children were almost four times more likely to drop out depending on which service they attended (median odds ratio = 3.92) and were two-and-a-half times more likely to drop out depending on which practitioner they saw (median odds ratio = 2.53). These levels of variation were not explained by levels of deprivation in areas covered by services or by children's demographic and case characteristics. The findings of the present research may suggest that, beyond service-level variation, there is also practitioner-level variation in non-consensual dropout in child mental health services.

Over view of major traumatic injury in Australia--Implications for trauma system design

BACKGROUND

Trauma registries are known to drive improvements and optimise trauma systems worldwide. This is the first reported comparison of the epidemiology and outcomes at major centres across Australia.

METHODS

The Australian Trauma Registry was a collaboration of 26 major trauma centres across Australia at the time of this study and currently collects information on patients admitted to these centres who die after injury and/or sustain major trauma (Injury Severity Score (ISS) > 12). Data from 1 July 2016 to 30 June 2017 were analysed. Primary endpoints were risk adjusted length of stay and mortality (adjusted for age, cause of injury, arrival Glasgow coma scale (GCS), shock-index grouped in quartiles and ISS).

RESULTS

There were 8423 patients from 24 centres included. The median age (IQR) was 48 (28-68) years. Median (IQR) ISS was 17 (14-25). There was a predominance of males (72%) apart from the extremes of age. Transport-related cases accounted for 45% of major trauma, followed by falls (35.1%). Patients took 1.42 (1.03-2.12) h to reach hospital and spent 7.10 (3.64-15.00) days in hospital. Risk adjusted length of stay and mortality did not differ significantly across sites. Primary endpoints across sites were also similar in paediatric and older adult (>65) age groups.

CONCLUSION

Australia has the capability to identify national injury trends to target prevention and reduce the burden of injury. Quality of care following injury can now be benchmarked across Australia and with the planned enhancements to data collection and reporting, this will enable improved management of trauma victims.

Developing an adapted Charlson comorbidity index for ischemic stroke outcome studies

Background

The Charlson comorbidity index (CCI) is commonly used to adjust for patient casemix. We reevaluated the CCI in an ischemic stroke (IS) cohort to determine whether the original seventeen comorbidities and their weights are relevant.

Methods

We identified an IS cohort (N = 6988) from the Ontario Stroke Registry (OSR) who were discharged from acute hospitals (N = 100) between April 1, 2012 and March 31, 2013. We used hospital discharge ICD-10-CA data to identify Charlson comorbidities. We developed a multivariable Cox model to predict one-year mortality retaining statistically significant (P < 0.05) comorbidities with hazard ratios ≥1.2. Hazard ratios were used to generate revised weights (1–6) for the comorbid conditions. The performance of the IS adapted Charlson comorbidity index (ISCCI) mortality model was compared to the original CCI using the c-statistic and continuous Net Reclassification Index (cNRI).

Results

Ten of the 17 Charlson comorbid conditions were retained in the ISCCI model and 7 had reassigned weights when compared to the original CCI model . The ISCCI model showed a small but significant increase in the c-statistic compared to the CCI for 30-day mortality (c-statistic 0.746 vs. 0.732, p = 0.009), but no significant increase in c-statistic for in-hospital or one-year mortality. There was also no improvement in the cNRI when the ISCCI model was compared to the CCI.

Conclusions

The ISCCI model had similar performance to the original CCI model. The key advantage of the ISCCI model is it includes seven fewer comorbidities and therefore easier to implement in situations where coded data is unavailable.

Risk-adjusted cUSUM control charts for shared frailty survival models with application to hip replacement outcomes: a study using the NJR dataset

Background

Continuous monitoring of surgical outcomes after joint replacement is needed to detect which brands’ components have a higher than expected failure rate and are therefore no longer recommended to be used in surgical practice. We developed a monitoring method based on cumulative sum (CUSUM) chart specifically for this application.

Methods

Our method entails the use of the competing risks model with the Weibull and the Gompertz hazard functions adjusted for observed covariates to approximate the baseline time-to-revision and time-to-death distributions, respectively. The correlated shared frailty terms for competing risks, corresponding to the operating unit, are also included in the model. A bootstrap-based boundary adjustment is then required for risk-adjusted CUSUM charts to guarantee a given probability of the false alarm rates. We propose a method to evaluate the CUSUM scores and the adjusted boundary for a survival model with the shared frailty terms. We also introduce a unit performance quality score based on the posterior frailty distribution. This method is illustrated using the 2003-2012 hip replacement data from the UK National Joint Registry (NJR).

Results

We found that the best model included the shared frailty for revision but not for death. This means that the competing risks of revision and death are independent in NJR data. Our method was superior to the standard NJR methodology. For one of the two monitored components, it produced alarms four years before the increased failure rate came to the attention of the UK regulatory authorities. The hazard ratios of revision across the units varied from 0.38 to 2.28.

Conclusions

An earlier detection of failure signal by our method in comparison to the standard method used by the NJR may be explained by proper risk-adjustment and the ability to accommodate time-dependent hazards. The continuous monitoring of hip replacement outcomes should include risk adjustment at both the individual and unit level.

A machine learning approach for predictive models of adverse events following spine surgery

BACKGROUND

Rates of adverse events following spine surgery vary widely by patient-, diagnosis-, and procedure-related factors. It is critical to understand the expected rates of complications and to be able to implement targeted efforts at limiting these events.

PURPOSE

To develop and evaluate a set of predictive models for common adverse events after spine surgery.

STUDY DESIGN

A retrospective cohort study.

PATIENT SAMPLES

We extracted 345,510 patients from the Truven MarketScan (MKS) and MarketScan Medicaid Databases and 760,724 patients from the Centers for Medicare and Medicaid Services (CMS) Medicare database who underwent spine surgeries between 2009 and 2013.

OUTCOME MEASURES

Overall adverse event (AE) occurrence and types of AE occurrence during the 30-day postoperative follow-up.

METHODS

We applied a least absolute shrinkage and selection operator regularization method and a logistic regression approach for predicting the risks of an overall AE and the top six most commonly observed AEs. Predictors included patient demographics, location of the spine procedure, comorbidities, type of surgery performed, and preoperative diagnosis.

RESULTS

The median ages of MKS and CMS patients were 49 years and 69, respectively. The most frequent individual AE was a cardiac dysfunction in CMS (10.6%) patients and a pulmonary complication (4.7%) in MKS. The area under the curve (AUC) of a prediction model for an overall AE was 0.7. Among the six individual prediction models, the model for predicting the risk of a pulmonary complication showed the greatest accuracy (AUC 0.76), and the range of AUC for these six models was 0.7 and 0.76. Medicaid status was one of the most important factors in predicting the occurrences of AEs; Medicaid recipients had increased odds of AEs by 20%-60% compared with non-Medicaid patients (odds ratios 1.28-1.6; p<10^-10). Logistic regression showed higher AUCs than least absolute shrinkage and selection operator across these different models.

CONCLUSIONS

We present a set of predictive models for AEs following spine surgery that account for patient-, diagnosis-, and procedure-related factors which can contribute to patient-counseling, accurate risk adjustment, and accurate quality metrics.

Refining risk adjustment for bundled payment models in cervical fusions-an analysis of Medicare beneficiaries

BACKGROUND CONTEXT

The current Bundled Payment for Care Improvement model relies on the use of "Diagnosis Related Groups" (DRGs) to risk-adjust reimbursements associated with a 90-day episode of care. Three distinct DRG groups exist for defining payments associated with cervical fusions: (1) DRG-471 (cervical fusions with major comorbidity/complications), (2) DRG-472 (with comorbidity/complications), and (3) DRG-473 (without major comorbidity/complications). However, this DRG system may not be entirely suitable in controlling the large amounts of cost variation seen among cervical fusions. For instance, these DRGs do not account for area/location of surgery (upper cervical vs. lower cervical), type of surgery (primary vs. revision), surgical approach (anterior vs. posterior), extent of fusion (1-3 level vs. >3 level), and cause/indication of surgery (fracture vs. degenerative pathology).

PURPOSE

To understand factors responsible for cost variation in a 90-day episode of care following cervical fusions.

STUDY DESIGN

Retrospective study of a 5% national sample of Medicare claims from 2008 to 2014 5% Standard Analytical Files (SAF5).

OUTCOME MEASURES

To calculate the independent marginal cost impact of various patient-level, geographic-level, and procedure-level characteristics on 90-day reimbursements for patients undergoing cervical fusions under DRG-471, DRG-472, and DRG-473.

METHODS

The 2008 to 2014 Medicare SAF5 was queried using DRG codes 471, 472, and 473 to identify patients receiving a cervical fusion. Patients undergoing noncervical fusions (thoracolumbar), surgery for deformity/malignancy, and/or combined anterior-posterior fusions were excluded. Patients with missing data and/or those who died within 90 days of the postoperative follow-up period were excluded. Multivariate linear regression modeling was performed to assess the independent marginal cost impact of DRG, gender, age, state, procedure-level factors (including cause/indication of surgery), and comorbidities on total 90-day reimbursement.

RESULTS

Following application of inclusion/exclusion criteria, a total of 12,419 cervical fusions were included. The average 90-day reimbursement for each DRG group was as follows: (1) DRG-471=$54,314±$32,643, (2) DRG-472=$28,535±$17,271, and (3) DRG-473=$18,492±$10,706. The risk-adjusted 90-day reimbursement of a nongeriatric (age <65) female, with no major comorbidities, undergoing a primary 1- to 3-level anterior cervical fusion for degenerative cervical spine disease was $14,924±$753. Male gender (+$922) and age 70 to 84 (+$1,007 to +$2,431) was associated with significant marginal increases in 90-day reimbursements. Undergoing upper cervical surgery (-$1,678) had a negative marginal cost impact. Among other procedure-level factors, posterior approach (+$3,164), >3 level fusion (+$2,561), interbody (+$667), use of intra-operative neuromonitoring (+$1,018), concurrent decompression/laminectomy (+$1,657), and undergoing fusion for cervical fracture (+$3,530) were associated higher 90-day reimbursements. Severe individual comorbidities were associated with higher 90-day reimbursements, with malnutrition (+$15,536), CVA/stroke (+$6,982), drug abuse/dependence (+$5,059), hypercoagulopathy (+$5,436), and chronic kidney disease (+$4,925) having the highest marginal cost impacts. Significant state-level variation was noted, with Maryland (+$8,790), Alaska (+$6,410), Massachusetts (+$6,389), California (+$5,603), and New Mexico (+$5,530) having the highest reimbursements and Puerto Rico (-$7,492) and Iowa (-$3,393) having the lowest reimbursements, as compared with Michigan.

CONCLUSIONS

The current cervical fusion bundled payment model fails to employ a robust risk adjustment of prices resulting in the large amount of cost variation seen within 90-day reimbursements. Under the proposed DRG-based risk adjustment model, providers would be reimbursed the same amount for cervical fusions regardless of the surgical approach (posterior vs. anterior), the extent of fusion, use of adjunct procedures (decompressions), and cause/indication of surgery (fracture vs. degenerative pathology), despite each of these factors having different resource utilization and associated reimbursements. Our findings suggest that defining payments based on DRG codes only is an imperfect way of employing bundled payments for spinal fusions and will only end up creating major financial disincentives and barriers to access of care in the healthcare system.

Texas Arguments Lead Heavy ACA Action

Arguments in Texas raise questions about the ACA's fate as enrollment holds steady and new state reinsurance waivers are approved.

Spatial risk adjustment between health insurances: using GWR in risk adjustment models to conserve incentives for service optimisation and reduce MAUP

This paper presents a new approach to deal with spatial inequalities in risk adjustment between health insurances. The shortcomings of non-spatial and spatial fixed effects in risk adjustment models are analysed and opposed against spatial kernel estimators. Theoretical and empirical evidence suggests that a reasonable choice of the spatial kernel could limit the spatial uncertainty of the modifiable area unit problem under heavy-tailed claims data, leading to more precise predictions and economically positive incentives on the healthcare market. A case study of the German risk adjustment shows a spatial risk spread of 86 Euro p.c., leading to incentives for spatial risk selection. The proposed estimator eliminates this issue and conserves incentives for services optimisation.