Cohort Of Medicare Patients Research Articles

Not There Yet: Using Data‐Driven Methods to Predict Who Becomes Costly Among Low‐Cost Patients with Type 2 Diabetes

Research ObjectiveDiabetes is a leading cause of Medicare spending, and predicting which individuals are likely to have high health care costs is essential for targeting interventions. Current approaches to predicting spending generally focus on composite measures, short time horizons, or patients who are already high utilizers and, therefore, whose costs may be harder to modify. In contrast, dynamic approaches that model monthly spending over time may help discriminate between low‐cost patients with diabetes who do and do not subsequently become costly.Study DesignWe identified Medicare beneficiaries with type 2 diabetes whose spending was in the bottom 90% of spending on diabetes care during a one‐year baseline period, a threshold for low‐cost patients used in prior studies. We used group‐based trajectory modeling, a data‐driven method, to classify unique clusters of these low‐cost patients by their diabetes spending patterns over the subsequent two‐year follow‐up period. Prediction models for these spending groups were then estimated with generalized boosted regression, a nonparametric machine learning method, using sets of all baseline predictors, diabetes predictors, and predictors that are potentially‐modifiable through interventions. For clinical context, we explored the relative influence of each predictor from the regression models to provide insight into baseline factors that may help distinguish patients who may become costly later. We evaluated each model through discrimination (C‐statistic) measures and split‐sample validation.Population StudiedWe identified Medicare fee‐for‐service beneficiaries with type 2 diabetes from a random, nationwide sample from 2011 to 2014 and used their enrollment, medical, and prescription claims data.Principal FindingsBaseline median diabetes spending for the 33 789 Medicare beneficiaries classified as low cost was $4153. We identified five distinct diabetes spending patterns among these individuals during the two‐year follow‐up period, including those in three groups with different, yet consistent, spending levels (68.1% of beneficiaries), a group with spending that rose quickly (25.3%), and a group with spending that rose progressively (6.6%). The ability to predict these spending groups was moderate to strong, with validated C‐statistics ranging from 0.63 to 0.87. The most influential factors for those in the group with progressively rising spending, in particular, were their age, generosity of Medicare insurance coverage, prior spending on diabetes care, and average adherence to medications.ConclusionsUsing data‐driven approaches, we identified distinct diabetes spending patterns among a nationally‐representative cohort of Medicare patients who were initially low utilizers. These patterns could be predicted using baseline characteristics, including factors that may be potentially modifiable.Implications for Policy or PracticeMany health care organizations use claims data to identify and predict patients for cost‐containment interventions. The approach we describe could inform the design and timing of cost‐containment interventions and target them to those at greatest need among patients with type 2 diabetes or other costly, chronic diseases.Primary Funding SourceNational Institute for Health Care Management (NIHCM) Foundation.

Identifying Potentially Avoidable Hospitalizations in Medicare Patients With Prostate Cancer: A Retrospective Analysis.

If identifiable, potentially avoidable hospitalizations (PAHs) can serve as an important target for cost containment efforts in oncology. PAHs among a cohort of Medicare patients with prostate cancer were identified using a two-stage consensus-driven review process. In stage 1, two clinicians independently evaluated admissions records using a case review form, which we modified from a previous study to assess for PAHs. In stage 2, any admissions that the reviewers disagreed on or were unsure of were re-examined in a larger group of clinicians to yield a consensus determination regarding avoidability. Univariable and multivariable regression analyses were performed to identify factors predictive of PAH. There were 160 admissions among this cohort of 210 patients from January 2012 to June 2015, of which 99 were evaluable. Consensus-driven clinical review yielded an overall PAH rate of 28.3%. Factors associated with increased PAH risk were admission for symptoms related to cancer (odds ratio [OR], 7.33; P < .001), presence of a social contributor to admission (OR, 4.40; P = .014), and history of alcohol or drug abuse (OR, 4.93; P = .025). Admission for a noncancer condition was associated with decreased PAH risk (OR, 0.32; P = .011). On multivariable analysis, presence of a social contributor to admission (OR, 9.35; P = .002) and admission as a result of a noncancer condition (OR, 0.16; P = .038) remained predictive of PAH risk. A significant proportion of hospitalizations among patients with prostate cancer are potentially avoidable. Understanding factors predictive of risk for PAH can help inform programs aimed at avoiding such admissions to improve overall care quality and value.

Evaluation of the Yale New Haven Readmission Risk Score for Pneumonia in a General Hospital Population

BackgroundThe Yale New Haven Readmission Risk Score (YNHRRS) for pneumonia is a clinical prediction tool developed to assess risk for 30-day readmission. This tool was validated in a cohort of Medicare patients; generalizability to a broader patient population has not been evaluated. In addition, it lacks indicators of functional status or social support, which have been shown in other studies to be predictors of readmission. The objective of this study was to evaluate the generalizability of the YNHRRS for pneumonia in a general population of hospitalized patients, and assess the impact of incorporating measures of functional status and social support on its predictive value. MethodsThis retrospective chart review comprised all patients admitted to a 563-bed academic medical center with a primary diagnosis of pneumonia between March 2014 and March 2015. Abstraction of clinical variables allowed calculation of the YNHRRS and additional indicators of functional status and social support. The primary outcome was 30-day readmission rate. We created a logistic regression model to predict readmission using the YNHRRS, functional status, and social support as covariates. ResultsAmong 270 discharges with pneumonia, the observed readmission rate was 23%. The YNHRRS was a significant predictor of readmission in our multivariate model, with an odds ratio of 2.20 (95% confidence interval, 1.29-3.73) for each 10% increase in calculated risk. Indicators of functional status and social support were not significant predictors of readmission. ConclusionsThe YNHRRS can be applied to an unselected population as a tool to predict patients with pneumonia at risk for readmission.

Adherence to the American Society of Radiation Oncology Choosing Wisely Campaign Recommendations for Treatment of Bone Metastasis: Analysis of Medicare Data at 12 Centers in the Southeast US (SA527C)

•Understand adherence to Choosing Wisely guidelines from a radiation oncology perspective.•Appreciate the impact of patient, practitioner, and system on choice of duration of treatment. The American Society of Therapeutic Radiation Oncology (ASTRO) Choosing Wisely (CW) campaign recommends avoidance of extended fractionation (>10 treatments) for palliation of bone metastasis with radiation therapy (RT). The purpose of this study is to analyze palliative fractionation schemes in a cohort of Medicare patients. Data were for Medicare patients > 65 with bone metastasis diagnosis from 12 cancer community networks affiliated with an academic center in Alabama, Mississippi, Georgia, Tennessee, and Florida. Stage IV cancer patients treated with radiation for bone metastasis from Quarter 1, 2012 to Quarter 1, 2015 were eligible and identified using ICD-9(198.5) and CPT/HCPCS codes. Adherence was measured by treatment duration ≤ 14 vs >14 calendar days. 2194 patients with bone metastasis were identified, 427 (19.5%) of which were treated with RT: 16.6% (2012), 37.5% (2013), 27.4% (2014), 18.5% (2015). Median age at diagnosis was 69.2 (IQR 9.0), and 1928 (87.8%) patients were Caucasian. In the 427 patients treated with RT, 25.8%, 37.0%, and 33.0% were treated in 2012, 2013, and 2014. The top 4 primary sites were: genitourinary (22.7%), breast (20.4%), gastrointestinal (15.7%), and lung (10.8%). Overall guideline adherence was 65.1%: 1 fraction in 33.3%, <14 calendar days in 31.9%, and > 14 calendar days in 34.9%. Average adherence by upper, middle, and lower tertile of sites was 52.0%, 64.5%, 71.5%. Adherence by primary cancer type was similar: breast (58.6%), lung (65.2%), gastrointestinal (67.2), and genitourinary (72.5%). In patients treated < 14 days, 4% were treated with either 1 or 2-5 fractions of stereotactic body radiotherapy. While compliance was high overall with ASTRO CW recommendations, there remained a significant portion (36%) of patients who received a prolonged course of palliative radiation.

Intensity Modulated Radiation Therapy in Breast Cancer: Analysis of Medicare Data at 12 Institutions

The Choosing Wisely campaign recommends avoidance of intensity modulated radiation therapy (IMRT) to deliver whole breast radiation therapy as part of breast conservation therapy. The purpose of this project is to analyze IMRT usage in a cohort of Medicare patients in 5 states in the southeastern United States. Data were for Medicare patients > 65 years of age with breast cancer diagnosis from 12 cancer community networks affiliated with an academic center in Alabama, Mississippi, Georgia, Tennessee, and Florida. Eligible patients with AJCC 7 stage 0-III breast cancer who were treated with lumpectomy and subsequent radiation therapy (RT) from Quarter 1, 2012 to Quarter 1, 2015 were identified using ICD-9 and CPT/HCPCS codes. Adherence to guidelines was defined by utilization or omission of IMRT. Cost of treatment was calculated per patient per quarter over the first year from breast cancer diagnosis. A total of 1,235 patients treated with lumpectomy were identified, 802 (64.9%) of which were treated with radiation therapy. Median age at diagnosis was 71.2 (8.0 IQR) and 701 (87.4%) patients were white. Stage at diagnosis was as follows: 0: 15.3%, I: 51.5%, II: 19.3%, III: 2.5%. Median time from lumpectomy to start of radiation was 97 days (IQR 65 days). Average adherence at all sites to guideline recommendations was 80.4%. Average adherence by upper, middle, and lower tertile of sites was 97.3%, 84.1%, 51.4% respectively. Guideline adherence per year was relatively stable: 84.8% (2012), 75.8% (2013), and 82.3% (2014). IMRT use by stage was as follows: 0: 17.1%, I: 18.2%, II and III: 21.7%. Patients treated with IMRT had ∼29% increased cost ($8741 vs $6798/patient/quarter) over the first year after diagnosis. Compliance with guidelines was relatively high, but there was significant variability by sites. A disproportionate number of patients with Stage 0-I disease were treated with IMRT, possibly resulting in increased healthcare costs. Future studies would benefit from analysis of specific patient and physician factors driving use of more expensive modalities in the setting of early stage disease known to have good outcomes. The project described was supported by Grant Number 1C1CMS331023 from the Department of Health and Human Services, Centers for Medicare and Medicaid Services. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. Department of Health and Human Services or any of its agencies. The research presented here was conducted by the awardee. Findings might or might not be consistent with or confirmed by the independent evaluation contractor.

Development and validation of a simple risk score to predict 30-day readmission after percutaneous coronary intervention in a cohort of medicare patients.

To develop a risk model that can be used to identify PCI patients at higher risk of readmission who may benefit from additional resources at the time of discharge. A high proportion of patients undergoing PCI are readmitted within 30 days of discharge. The sample comprised patients aged ≥65 years who underwent PCI at a CathPCI Registry®-participating hospital and could be linked with 100% Medicare fee-for-service claims between 01/2007 and 12/2009. The sample (n = 388,078) was randomly divided into risk score development (n = 193,899) and validation (n = 194,179) cohorts. We did not count as readmissions those associated with staged revascularization procedures. Multivariable logistic regression models using stepwise selection models were estimated to identify variables independently associated with all-cause 30-day readmission. The mean 30-day readmission rates for the development (11.36%) and validation (11.35%) cohorts were similar. In total, 19 variables were significantly associated with risk of 30-day readmission (P < 0.05), and model c-statistics were similar in the development (0.67) and validation (0.66) cohorts. The simple risk score based on 14 variables identified patients at high and low risk of readmission. Patients with a score of ≥13 (15.4% of sample) had more than an 18.5% risk of readmission, while patients with a score ≤6 (41.9% of sample) had less than an 8% risk of readmission. Among PCI patients, risk of readmission can be estimated using clinical factors present at the time of the procedure. This risk score may guide clinical decision-making and resource allocation for PCI patients at the time of hospital discharge. © 2016 Wiley Periodicals, Inc.

Identifying patterns of breast cancer care provided at high volume hospitals: A classification and regression tree analysis.

69 Background: There is a growing body of literature linking hospital volume to outcomes in breast cancer. However, the mechanism through which volume influences outcome is poorly understood. The purpose of this study was to examine the relationship between hospital volume of breast cancer cases and patterns of processes of care in a population-based cohort of Medicare patients. Methods: A previously described and validated algorithm was applied to Medicare claims for newly diagnosed breast cancer cases in 2003 to identify potential subjects. Breast cancer patients were recruited to participate in a survey study examining breast cancer outcomes, and data was merged with Medicare claims and state tumor registries. Hospital volume was divided into tertiles. A Classification and Regression Tree (CART) model was performed to look for statistically significant relationships between patterns of processes of care and hospital volume. Results: Using CART analysis, eight patterns of care were identified that differentiated breast cancer care at high versus low volume hospitals. Sentinel lymph node dissection (SLND) was the single process of care that demonstrated the greatest differentiation across hospitals with differing volumes. Four patterns of care significantly predicted that a patient was less likely to be treated at a high volume hospital. Conclusions: Our study demonstrates differences in patterns of processes of care between low and high volume hospitals. Hospital volume was associated with several patterns of care that reflect the most current standards of care, particularly SLND. Greater adoption of these patterns by low volume hospitals could improve the overall quality of care for breast cancer.

Unruptured cerebral aneurysm clipping: association of combined open and endovascular expertise with outcomes

BackgroundIt is often questioned if one physician can conduct both open and endovascular techniques successfully and safely.ObjectiveTo investigate the association of combined open and endovascular expertise with the outcomes of...

Surgical clipping versus endovascular coiling for elderly patients presenting with subarachnoid hemorrhage

BackgroundThe comparative effectiveness of the two treatment options (surgical clipping and endovascular coiling) for ruptured cerebral aneurysms has not been studied in real-world practice in the USA. We investigated the...

P-951 - Medicare Part D, generic prescribing and adherence to antidepressant therapy

Introduction In the U.S., Medicare Part D coverage gap poses concern for antidepressant adherence, but wide availability of generic antidepressants may mitigate the effect. Objectives and aims To assess the effects of coverage gap and generic prescribing on antidepressant adherence among a cohort of Medicare patients experiencing a new episode of depression. Methods We used 2007 Medicare Part D drug event file to examine antidepressant use over the 6 months following antidepressant initiation. We estimated Cox proportional hazard models to assess the effects of the two factors on the risk of antidepressant treatment disruption. We estimated mixed-effects linear models to assess the effects on monthly days of antidepressant possession. All analyses were stratified by Part D low-income subsidy (LIS) status and Medicare entitlement (aged vs. disabled). Results Receiving a generic (vs. brand) antidepressant on the first script was associated with a significantly reduced risk of treatment disruption across all four populations: hazard ratios ranged from 0.66 (p = 0.006) among disabled patients not receiving LIS to 0.88 (p = 0.020) among aged patients receiving LIS. For patients not receiving LIS, experience of coverage gap was associated with increased risk of disruption only among the disabled (hazard ratio = 2.15, p = 0.006). Analysis of monthly antidepressant possession produced consistent results and indicated that the beneficial effect of generic first script was comparable with or outweighed the detrimental effect of coverage gap. Conclusions Part D coverage gap adversely affected antidepressant adherence by disabled patients not receiving LIS. Increased generic prescribing may help improve adherence among all patients.

Abstract P119: Comparison of Treatment of Acute Coronary Syndrome in a Cohort of Medicare Patients With and Without Diabetes Mellitus

Objective: This study compared real-world treatment patterns for ACS patients with and without DM in a cohort of Medicare population. Methods: A retrospective cohort study was conducted using the MarketScan Medicare claims database. Patients aged ≥65 years, hospitalized with a primary diagnosis of ACS between 01/01/2007 and 12/31/2007 were categorized into 2 groups: with and without DM. Patient characteristics, treatment patterns during the index hospitalization, and cardiovascular medications use 12 months pre- and post-index hospitalization were analyzed. (All p&lt;.01 unless otherwise stated.) Results: Of 8,666 ACS patients identified, 2,407 (28%) were diabetic and 6,259 (72%) were non-diabetic. Patients with DM were more likely to be younger (77.6 vs. 79.6), male (58.8% vs. 54.6%), and have higher rates of previous MI (17.2% vs. 12.0%), congestive heart failure (28% vs. 16.1%), PVD (19.2% vs. 12.8%), TIA/stroke (10.0% vs. 7.3%), hypertension (54.4% vs. 49.6%), and renal diseases (17.1% vs. 7.0%). Patients with DM were significantly more likely to present with non-ST-segment elevation myocardial infarction at the index hospitalization compared to those without DM (66.4% vs. 61.2%). During the index hospitalization, DM patients underwent less PCI (27.8% vs. 34.6%) but more CABG (9.7% vs. 7.6%), and were more likely to have medical management during the index hospitalization compared to non-DM patients. Compared to their non-DM counterparts, patients with DM utilized more statins (66.9% vs. 48.0%), beta blockers (53.0% vs .42.4%), thienopyridine (24.5% vs. 13.2%), and ACE inhibitors (42.3% vs. 28.6%) during the 12-month pre-hospitalization period, while during the 12-month post-hospitalization period, they utilized slightly more statins (85.8% vs. 82.1%) and had similar level of beta blockers (86.1% vs. 85.6%, p=0.57) and thienopyridine (62.5% vs. 64.7%, p=0.12) and ACE inhibitors use (54.8% vs. 56.1%, p=0.37). Conclusion: ACS patients with DM presented additional clinical risks than those without DM. Observed treatment patterns varied correspondingly between the 2 groups. Differences in cardiovascular medications usage between the 2 groups narrowed after the ACS hospitalization.

Abstract P236: Health Care Costs and Resource Utilizations of Acute Coronary Syndrome in a Cohort of Medicare Patients With and Without Diabetes Mellitus

Objective: Diabetes is an independent prognostic factor for increased risk for ischemic heart disease. This study evaluates differences in resource utilization and health care costs for acute coronary syndrome (ACS) patients with and without diabetes mellitus (DM). Methods: A retrospective cohort study was conducted using a large US administrative claims database. Patients, ≥65, hospitalized with a primary diagnosis of ACS between 01/01/2007 and 12/31/2007 were identified and categorized into 2 groups: 1) with DM (ACS patients had ≥ 2 outpatient office visits or 1 hospitalization with the DM diagnosis during the 1 year before/within the index hospitalization) and 2) without DM. Patients with complete insurance coverage 1-year pre- and post- index hospitalization were included. Annual health care costs and resource utilization were compared usingChi-square tests for categorical variables and Wilcoxon rank-sum tests for continuous variables (all p&lt;.05 unless otherwise stated). Results: Of 5,840 identified patients, 1,609 (28%) were diabetic and 4,231 (72%) were non-diabetic. A higher percentage of diabetic patients had at least one all-cause rehospitalization event (44.4% vs. 35.7%) or CV-related rehospitalization event (27.0% vs. 19.5%). Mean length of stay (LOS) was longer for DM patients during the index hospitalization (6.3 days vs. 5.5 days), as well as during the CV-related rehospitalization (5.7 days vs. 4.8 days). In addition, patients with DM had more physician's office visits (18.3 vs. 15.0), ER visits (0.6 vs. 0.5), and outpatient hospital visits (7.4 vs. 6.1) during the 12-month follow-up period. Both cohorts had similar index ACS hospitalization costs ($18,532 vs. $17,609) but DM patients incurred higher rehospitalization costs (all-cause: $21,064 vs. $13,821; CV-related: $11,086 vs. $7,260), outpatient costs (all-cause: $28,167 vs. $19,815; CV-related: $7,193 vs. $5,686), and pharmacy costs (all-cause: $5,537 vs. $3,815; CV-related: $3,522 vs. $3,143). One-year follow-up health care costs were significantly higher for patients with DM compared with those without DM ($54,768 vs. $37,451). Conclusion: The presence of DM is significantly associated with greater resource utilization and health care costs for ACS patients.

All-Cause Readmission and Repeat Revascularization After Percutaneous Coronary Intervention in a Cohort of Medicare Patients

The purpose of this study was to report on the all-cause readmission and repeat revascularization rates after percutaneous coronary intervention (PCI). Although PCIs are frequently performed, 30-day rates of readmission and repeat revascularization after PCI are not known. Retrospective analysis of a cohort of Medicare fee-for-service admissions associated with a PCI in 2005. Primary outcomes were 30-day all-cause readmission rates and 30-day readmission rates associated with a revascularization procedure. A total of 315,241 PCI procedures performed at 1,108 hospitals were included in the analysis. The all-cause 30-day readmission rate was 14.6%, and the all-cause 30-day mortality rate was 1.0%. All-cause 30-day mortality among readmitted patients was higher than patients who were not readmitted (3.6% vs. 0.6%; p < 0.001). The 30-day readmission rate of acute myocardial infarction (AMI) patients was significantly higher than that of non-AMI patients (AMI 17.5%, non-AMI 13.6%, p < 0.001). Among all patients readmitted within 30 days after the index PCI, 27.5% had an associated revascularization procedure (PCI 25.8%, coronary artery bypass grafting 1.7%). The median readmission rates varied across hospitals, from 8.9% in the lowest decile to 22.0% in the highest decile. A substantial proportion of PCI patients are readmitted within 30 days of discharge, and readmission rates vary widely across hospitals. Readmissions within 30 days of an index PCI procedure were associated with a significantly higher 30-day mortality rate, and more than one-quarter of such readmissions resulted in a repeat revascularization procedure. These findings warrant further attention to determine whether these readmissions are preventable.