Medicare Cohort Research Articles

One-year costs of medical admissions with and without a 30-day readmission and enhanced risk adjustment

BackgroundTo overcome the limitations of administrative data in adequately adjusting for differences in patients’ risk of readmissions, recent studies have added supplemental data from patient surveys and other sources (e.g., electronic health records). However, judging the adequacy of enhanced risk adjustment for use in assessment of 30-day readmission as a hospital quality indicator is not straightforward. In this paper, we evaluate the adequacy of risk adjustment by comparing the one-year costs of those readmitted within 30 days to those not after excluding the costs of the readmission.MethodsIn this two-step study, we first used comprehensive administrative and survey data on a nationally representative Medicare cohort of hospitalized patients to compare patients with a medical admission who experienced a 30-day readmission to patients without a readmission in terms of their overall Medicare payments during 12 months following the index discharge. We then examined the extent to which a series of enhanced risk adjustment models incorporating code-based comorbidities, self-reported health status and prior healthcare utilization, reduced the payment differences between the admitted and not readmitted groups.ResultsOur analytic cohort consisted 4684 index medical hospitalization of which 842 met the 30-day readmission criteria. Those readmitted were more likely to be older, White, sicker and with higher healthcare utilization in the previous year. The unadjusted subsequent one-year Medicare spending among those readmitted ($56,856) was 60% higher than that among the non-readmitted ($35,465). Even with enhanced risk adjustment, and across a variety of sensitivity analyses, one-year Medicare spending remained substantially higher (46.6%, p < 0.01) among readmitted patients.ConclusionsEnhanced risk adjustment models combining health status indicators from administrative and survey data with previous healthcare utilization are unable to substantially reduce the cost differences between those medical admission patients readmitted within 30 days and those not. The unmeasured patient severity that these cost differences most likely reflect raises the question of the fairness of programs that place large penalties on hospitals with higher than expected readmission rates.

Abstract WP384: Impact of Comorbidities on All-cause Mortality among Patients Hospitalized with Acute Ischemic Stroke - a Medicare Cohort Study

Introduction: Limited studies examined the prevalence and impact of comorbidities on all-cause mortality longitudinally among older patients with acute ischemic stroke (AIS). Methods: We included Medicare fee-for-service beneficiaries diagnosed with AIS from 2007 to 2010 and followed up to 2015. We calculated prevalence of 12 comorbidities, including atrial fibrillation (AF), congestive heart failure (CHF), chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), diabetes, hypertension, hyperlipidemia, ischemic heart disease (IHD), obesity, peripheral vascular disease (PVD), cancer, and depression at baseline. We used Cox proportional hazard models to examine the associations with all-cause mortality, and calculated population attributable fraction (PAF) contributing to mortality. Results: Among 306,142 Medicare beneficiaries 65-74 years of age hospitalized with AIS, we documented 139,228 deaths (45.5%) during a median 5.8 years of follow-up. Six comorbidities had a prevalence ≥15% and were associated with ≥20% increased risk of death. Prevalence and adjusted hazard ratio (AHR) were 28.5% and 1.75 for CKD, 30.9% and 1.67 for CHF, 26.9% and AHR 1.52 for COPD, 47.7% and 1.33 for diabetes, 22.1% and 1.29 for depression, and 16.1% and 1.23.for AF. Beneficiaries with ≥4 of 6 prevalent comorbidities had a 5.8 fold increased risk of death as compared to those without any comorbidities. The all-cause mortality among AIS beneficiaries would be reduced by about 50% (PAF) if all six comorbidities could be prevented. Beneficiaries with cancer, IHD, and PVD were also more likely to die as compared to those without. However, beneficiaries with hyperlipidemia (AHR 0.54), hypertension (AHR 0.81), and obesity (AHR 0.85) had a significantly lower risk of death than those without. Conclusion: We identified several common comorbidities associated with a significantly increased risk of all-cause mortality. Our findings provide useful information to better target prevention and treatment strategies to enhance overall survival.

Relation of Admission Blood Pressure to In-hospital and 90-Day Outcomes in Patients Presenting With Transient Ischemic Attack

The association between admission blood pressure (BP) and outcomes in patients with transient ischemic attack (TIA) is not well defined. Patients in the United States national Get With The Guidelines-Stroke registry with a TIA were included. Admission systolic and diastolic BP was used to compute mean arterial pressure and pulse pressure (PP). A subset of this cohort was linked to Centers for Medicare and Medicaid claims data for postdischarge outcomes. The in-hospital outcomes of interest were: mortality, not discharged home, and inability to ambulate independently at discharge. Postdischarge, 30-day and 90-day outcomes of interest were mortality, readmission for stroke, and readmission for major cardiovascular event-composite of death, cerebrovascular, or cardiovascular readmission. Among the 218,803 patients with TIA, lower admission systolic blood pressure (SBP) was associated with worse in-hospital outcomes. Compared with patients with SBP of 150 mm Hg, a lower SBP of 120 mm Hg was associated with higher risk of in-hospital death (adjusted OR = 1.79; 95% CI = 1.50 to 2.12), not being discharged home (adjusted OR = 1.31; 95% CI = 1.27 to 1.36), or inability to ambulate independently at discharge (adjusted OR = 1.27; 95% CI = 1.23 to 1.31). Similarly, among the 64,352 patients in the Centers for Medicare and Medicaid-linked cohort, an inverse association between systolic BP and postdischarge mortality (p <0.0001), and major cardiovascular event (p = 0.0001) was observed at 30-days and at 90-days postdischarge. However, there was no relation between SBP and readmission for stroke either at 30-days (p = 0.35) or at 90-days (p = 0.11). Results were largely similar for diastolic BP, mean arterial pressure, PP, and outcomes. In conclusion, in patients with a transient ischemic attack, a BP paradox was observed, with higher admission BP associated with improved in-hospital, 30-day, and 90-day postdischarge outcomes.

Opioid Overdose Hospitalizations among Medicare-Disability Beneficiaries.

As the opioid epidemic progresses, a better understanding of those at elevated risk of opioid overdose is needed, particularly for populations whose growing risk may be overlooked. Medicare recipients under age 65 (Medicare-disability beneficiaries [MDBs]) are one such population. We sought to analyze characteristics of opioid-overdose hospitalizations among MDBs and quantify the contribution of this population to opioid-overdose hospitalizations overall. This retrospective cohort study included patients hospitalized for opioid overdose in the National/Nationwide Inpatient Sample from 1998 to 2013. The primary outcome measurements were number and characteristics of discharges, including patient sex, age, race, prescription opioid versus heroin overdose, and comorbidities. MDBs constituted 11.7% of US opioid overdose hospitalizations among those under 65 years of age in 1998; this proportion grew to 24.5% by 2013 (P < .0001). The proportion of female patients grew markedly among this cohort (P < .0001) and were disproportionately represented among MDBs (P < .0001). Prescription opioid overdose accounted for a larger proportion of opioid overdose hospitalizations among MDBs than among non-Medicare-insured patients under 65 years old (P < .0001). MDBs generally exhibited greater comorbidity burden versus non-Medicare-insured patients under age 65; however, chronic drug and alcohol abuse were less commonly documented among the Medicare cohort (P < .0001). MDBs constitute a substantial and growing proportion of opioid overdose hospitalizations in the United. To prevent opioid overdoses among MDBs, care must be taken to address the unique needs of this population.

Investigating the Association Between Advanced Practice Providers and Chemotherapy-Related Adverse Events in Women With Breast Cancer: A Nested Case-Control Study.

The effect of advanced practice provider (APP) involvement in oncology care on cancer-specific outcomes is unknown. We examined the association between team-based APP-physician care during chemotherapy and chemotherapy-related adverse events (AEs) among women with breast cancer. We performed separate nested case-control analyses in two national cohorts of women who received chemotherapy for incident breast cancer. Cohorts were identified from Medicare (≥ 65 years of age) and MarketScan (18 to 64 years of age) data. Cases experienced a chemotherapy-related AE (emergency room visit and/or hospitalization). Controls were matched 1:1 on the basis of each patient's age, comorbidities, census region, state's APP scope of practice regulations, and observation period from chemotherapy initiation to first AE. APP exposure (any outpatient claim billed by an APP during the observation period) was assessed for each matched pair member. Among the 1,948 cases in the Medicare cohort, 225 (12%) had APP exposure before the first chemotherapy-related AE, compared with 213 controls (11%; P = .54). Among the 725 cases in the MarketScan cohort, 52 (7%) had APP exposure compared with 65 controls (9%; P = .21). In the matched case-control analysis, there was no association between outpatient APP exposure during chemotherapy and AEs in either cohort (Medicare: OR, 1.06 [95% CI, 0.87 to 1.30]; MarketScan: OR, 0.76 [95% CI, 0.50 to 1.14]). Our results suggest that team-based APP-physician care that includes an APP who is billing independently, at least for certain patients receiving chemotherapy, may be a viable strategy to safely leverage the scarce oncology workforce to increase access and delivery of cancer care.

A Marginal Estimate of PM2.5 Effects on Mortality Using Propensity Scores and Exposure Randomization by Moving

Introduction: Multiple observational studies have found an association between ambient PM2.5 and mortality, and while they all adjusted for confounding, inferring causality is always challenging. In this study, we leverage information on changing residential address (moving) to design an observational study that will be less sensitive to confounding bias.Methods: In the Medicare cohort aged 65 and older, we identified about 10 million movers and 52 million non-movers and then used a previously validated exposure prediction model to assign mean annual PM2.5 exposure. We checked if moving randomized subsequent exposure by comparing the standardized differences in confounders by exposure in movers to non-movers. We controlled for potential individual and ecological confounders and Zip code of origin using propensity score models to generate inverse probability weights (IPWs) among white and black movers separately and applied these weights in Cox proportional hazards models to estimate the risk of all-cause mortality.Results: All standardized differences were smaller among white movers compared to non-movers indicating randomization to exposure. Among black movers, all standardized differences were smaller except for Medicaid eligibility and % home ownership at the Zip code level (these differences were slight). Using our IPWs, per 10 µg/m3 of PM2.5 exposure we found a hazard ratio of 1.15 (95% CI: 1.14 to 1.16) among whites and 1.23 (1.21-1.26) among blacks. Hazard ratios held even when restricting these analyses to movers exposed to levels below the current US EPA standard of 12 µg/m3.Conclusion: We used a pseudo-randomization approach (moving) to obtain protection against unmeasured confounders combined with a propensity score approach, which makes exposure independent of measured covariates, and found a highly significant effect of PM2.5 on death rates, which is likely causal and continues below current standards.

Type 2 Diabetes Remission—Analysis of Three Population-Level Historical Cohorts

Objective: To analyze the characteristics of individuals who went into remission for type 2 diabetes mellitus (T2DM) in the absence of bariatric surgery in three large population level datasets. Methods: Data from three large T2DM cohorts were used in this analysis: a) 10,059 subjects from a longitudinal Medicare cohort (n=2.4 Million); b) 4,997 subjects, Continuous NHANES (n=82,091); and c) 562 subjects, NHANES III (n=33,994). A Medicare T2DM individual with no T2DM related claims continuously for more than 12 months was considered to be in remission. In the NHANES data T2DM remission was defined as an HbA1c ≤ 5.7 without T2DM related medications. Cox proportional hazards was used to identify significant variables associated with T2DM “remission” in the Medicare data. NHANES data were analyzed with multiple logistic regression with T2DM remission as the dependent variable. Results: In the Medicare dataset 4.97% met the criteria for T2DM remission in a five-year longitudinal study, while 2.76% and 1.97% of the continuous NHANES and the NHANES III subjects, respectively, achieved remission in the cross-sectional cohorts. The factors associated with T2DM remission were no statin use; no insulin use; no hypertension; no hyperlipidemia; no neuropathy; no retinopathy; female gender; non-African American ethnicity; decreased waist circumference and lower systolic blood pressure (Bonferroni adjusted p≤0.05). The rate of T2DM remission appears to be approximately 1.97 to 4.97%. Conclusions: T2DM remission triggers exist (Lancet, Dec 2017). Of the factors associated with T2DM remission, only ethnicity and gender are non-modifiable risk factors. Therefore, diabetes prevention and management programs should concentrate on lowering body weight, lipids, and blood pressure. Management of these factors is associated with T2DM remission. Identification of other factors associated with T2DM may lead to more efficacious programmatic management of the disease and attenuation of serious co-morbidities. Disclosure S. Tangelloju: Employee; Self; Humana Inc.. G. Vu: None. H. Chavis-Blakely: None. B.B. Little: None.

Increased Risk of Infectious Complications in Older Patients With Indolent Non-Hodgkin Lymphoma Exposed to Bendamustine.

Bendamustine is a potent chemotherapy agent increasingly used to treat indolent non-Hodgkin lymphoma (iNHL). While effective, it causes significant T-cell lymphopenia, which may increase risk of infection. We examined infectious complications associated with bendamustine-containing regimens among older patients with iNHL. For this Surveillance, Epidemiology, and End Results (SEER)-Medicare cohort study, we identified 9395 patients with iNHL (follicular, marginal zone, Waldenström macroglobulinemia) treated with chemotherapy from 2006 to 2013. Thirteen percent received bendamustine-containing regimens. We compared baseline characteristics and infection incidence rates between patients treated with and without bendamustine. We conducted multivariate Cox proportional hazards regression (adjusting for demographics, comorbidities, disease and treatment characteristics, risk factors for infection, and antimicrobial prophylaxis) to determine infectious risks associated with bendamustine. Bendamustine was associated with an increased risk of both common infections such as bacterial pneumonia (hazard ratio [HR], 1.50 [95% confidence interval {CI}, 1.21-4.85]) and opportunistic infections such as cytomegalovirus (HR, 3.98 [95% CI, 1.40-11.26]), varicella zoster virus (HR, 1.49 [95% CI, 1.18-1.89]), histoplasmosis (HR, 3.55 [95% CI, 1.10-11.42]), and Pneumocystis jirovecii pneumonia (when administered as third-line therapy: HR, 3.32 [95% CI, 1.00-11.11]). Risk of infections was more prominent in patients receiving bendamustine as part of later (third-line and above) regimens, and independently associated with well-established factors such as neutropenia and corticosteroid exposure. Bendamustine is associated with an increased risk of common and opportunistic infections in patients with iNHL. Further prospective investigation into the potential role of antimicrobial prophylaxis is needed in these patients.

Healthcare Costs for Insured Older U.S. Adults with Hearing Loss.

To measure 18-month healthcare cost difference attributable to hearing loss (HL) in older adults. Matched cohort. Retrospective analysis of administrative healthcare bills of insured older Americans. Older U.S. adults with health insurance in 3 cohorts matched based on HL diagnosis using propensity score methods (N=904,750). Comparison groups were defined as those with and without HL diagnosis using International Classification of Diseases, Ninth Revision, diagnosis codes from billing records. Outcomes measured include 18-month total healthcare payments and healthcare payments broken down according to payment type (inpatient, outpatient, prescription, hearing services). Three comparison cohorts were examined for these outcomes: individuals covered by Medicare plus supplemental private insurance (Cohort 1, n=782,216), those covered only under Medicare (Cohort 2, n=105,296), and those within the Medicare sample that were dually eligible for Medicare and Medicaid coverage (Subcohort 3, n=17,238). The quality of the propensity score match was examined using standardized differences in means or proportions between all matched covariates, and cost outcomes were analyzed using multiple generalized linear regression models. Fully adjusted models showed significantly higher 18-month healthcare payments for individuals with a diagnosis of HL without indication of use of hearing services than in those without a HL diagnosis in the 3 samples (payment differences: $3,587 Cohort 1, $3,779 Cohort 2, $4,657 Subcohort 3; all p<.001). Payment differences were also found between individuals with HL and indications of hearing services and those without HL. We observed more than 20% higher total healthcare payments over 18 months for a group of insured individuals with HL regardless of insurance type or hearing services use, indicating that negative health-related effects of HL may increase healthcare use unrelated to HL. Thus, clinical care to ameliorate HL may improve overall health.

Association of Electronic Health Record Use With Quality of Care and Outcomes in Heart Failure: An Analysis of Get With The Guidelines-Heart Failure.

BackgroundAdoption of electronic health record (EHR) systems has increased significantly across the nation. Whether EHR use has translated into improved quality of care and outcomes in heart failure (HF) is not well studied.Methods and ResultsWe examined participants from the Get With The Guidelines—HF registry who were admitted with HF in 2008 (N=21 222), using various degrees of EHR implementation (no EHR, partial EHR, and full EHR). We performed multivariable logistic regression to determine the relation between EHR status and several in‐hospital quality metrics and outcomes. In a substudy of Medicare participants (N=8421), we assessed the relation between EHR status and rates of 30‐day mortality, readmission, and a composite outcome. In the cohort, the mean age was 71±15 years, 49% were women, and 64% were white. The mean ejection fraction was 39±17%. Participants were admitted to hospitals with no EHR (N=1484), partial EHR (N=13 473), and full EHR (N=6265). There was no association between EHR status and several quality metrics (aside from β blocker at discharge) or in‐hospital outcomes on multivariable adjusted logistic regression (P>0.05 for all comparisons). In the Medicare cohort, there was no association between EHR status and 30‐day mortality, readmission, or the combined outcome.ConclusionsIn a large registry of hospitalized patients with HF, there was no association between degrees of EHR implementation and several quality metrics and 30‐day postdischarge death or readmission. Our results suggest that EHR may not be sufficient to improve HF quality or related outcomes.

Cancer and non-cancer pain opioid utilization in Medicare and Medicaid populations.

139 Background: Limited data exist comparing opioid use patterns in Medicare and Medicaid patients with cancer-related (CP) and non-cancer-related pain (NCP). Methods: A retrospective analysis of Medicare and Medicaid claims data (MarketScan Research Databases) investigated opioid use patterns in patients with CP and NCP. Adults (age ≥18 yr) with ≥1 pharmacy claim for an opioid (index date), continuous plan eligibility for 6 months pre- and 12 months post-index date, and duration of opioid use of ≥4 weeks were identified. CP patients were identified by medical claim for a cancer diagnosis within 30 days before index date. Results: A total of 4,009 Medicare and 551 Medicaid patients with CP and 98,631 Medicare and 25,163 Medicaid patients with NCP were analyzed. The most common cancer diagnoses were breast, lung, prostate, and colorectal. Medicare patients with CP and NCP had similar mean age; in the Medicaid cohort, patients with CP were older than those with NCP. In the Medicare cohort, NCP patients were more likely to be women; sex distribution was similar among Medicaid patients. Higher rates of comorbidity in the CP cohorts were observed in both datasets. Median index and post-index opioid doses were consistent between the CP and NCP cohorts. The post-index pattern of change in opioid dose was consistent between CP and NCP in both Medicare and Medicaid patients. The most common pattern observed was up to a doubling of index dose. Conclusions: Similar opioid utilization patterns in Medicare and Medicaid populations, including dose escalation, were observed regardless of pain etiology (cancer or non-cancer). [Table: see text]

MEDICARE-VALIDATED READMISSION SCORE MAINTAINS PROGNOSTIC VALUE IN REAL-WORLD HEART FAILURE

We previously developed and validated a regional heart failure (HF) readmission risk tool utilizing national Center for Medicare and Medicaid Services’ (CMS) model variables and region-specific variables in a southeastern United States Medicare cohort. We aim to measure the readmission prediction

Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting

BackgroundThis study compares early and late outcomes in patients undergoing coronary artery bypass grafting with and without preoperative atrial fibrillation in a contemporary, nationally representative Medicare cohort. MethodsIn the Medicare-Linked Society of Thoracic Surgeons database, 361,138 patients underwent isolated coronary artery bypass from 2006 to 2013, of whom 37,220 (10.3%) had preoperative atrial fibrillation; 13,161 (35.4%) were treated with surgical ablation and were excluded. Generalized estimating equations were used to compare 30-day mortality and morbidity. Long-term survival was summarized using Kaplan-Meier curves and Cox regression models. Stroke and systemic embolism incidence was modeled using the Fine-Gray model and the CHA2DS2-VASc score was used to analyze stroke risk. Median follow-up was 4 years. ResultsPreoperative atrial fibrillation was associated with a higher adjusted in-hospital mortality (odds ratio [OR], 1.5; P < .0001) and combined major morbidity including stroke, renal failure, prolonged ventilation, reoperation, and deep sternal wound infection (OR, 1.32; P < .0001). Patients with preoperative atrial fibrillation experienced a higher adjusted long-term risk of all-cause mortality and cumulative risk of stroke and systemic embolism compared to those without atrial fibrillation. At 5 years, the survival probability in the preoperative atrial fibrillation versus no atrial fibrillation groups stratified by CHA2DS2-VASc scores was 74.8% versus 86.3% (score 1-3), 56.5% versus 73.2% (score 4-6), and 41.2% versus 57.2% (score 7-9; all P < .001). ConclusionsPreoperative atrial fibrillation is independently associated with worse early and late postoperative outcomes. CHA2DS2-VASc stratifies risk, even in those without preoperative atrial fibrillation.

Defining Multiple Chronic Conditions for Quality Measurement.

Patients with multiple chronic conditions (MCCs) are a critical but undefined group for quality measurement. We present a generally applicable systematic approach to defining an MCC cohort of Medicare fee-for-service beneficiaries that we developed for a national quality measure, risk-standardized rates of unplanned admissions for Accountable Care Organizations. To define the MCC cohort we: (1) identified potential chronic conditions; (2) set criteria for cohort conditions based on MCC framework and measure concept; (3) applied the criteria informed by empirical analysis, experts, and the public; (4) described "broader" and "narrower" cohorts; and (5) selected final cohort with stakeholder input. Subjects were patients with chronic conditions. Participants included 21.8 million Medicare fee-for-service beneficiaries in 2012 aged 65 years and above with ≥1 of 27 Medicare Chronic Condition Warehouse condition(s). In total, 10 chronic conditions were identified based on our criteria; 8 of these 10 were associated with notably increased admission risk when co-occurring. A broader cohort (2+ of the 8 conditions) included 4.9 million beneficiaries (23% of total cohort) with an admission rate of 70 per 100 person-years. It captured 53% of total admissions. The narrower cohort (3+ conditions) had 2.2 million beneficiaries (10%) with 100 admissions per 100 person-years and captured 32% of admissions. Most stakeholders viewed the broader cohort as best aligned with the measure concept. By systematically narrowing chronic conditions to those most relevant to the outcome and incorporating stakeholder input, we defined an MCC admission measure cohort supported by stakeholders. This approach can be used as a model for other MCC outcome measures.

Functional Status Is Associated With 30-Day Potentially Preventable Readmissions Following Skilled Nursing Facility Discharge Among Medicare Beneficiaries

ObjectivesThe objectives of this study were to determine the association between patients’ functional status at discharge from skilled nursing facility (SNF) care and 30-day potentially preventable hospital readmissions, and to examine common reasons for potentially preventable readmissions. DesignRetrospective cohort study. SettingSNFs and acute care hospitals submitting claims to Medicare. ParticipantsNational cohort of Medicare fee-for-service beneficiaries discharged from SNF care between July 15, 2013, and July 15, 2014 (n = 693,808). Average age was 81.4 (SD 8.1) years, 67.1% were women, and 86.3% were non-Hispanic white. MeasurementsFunctional items from the Minimum Data Set 3.0 were categorized into self-care, mobility, and cognition domains. We used specifications for the SNF potentially preventable 30-day postdischarge readmission quality metric to identify potentially preventable readmissions. ResultsThe overall observed rate of 30-day potentially preventable readmissions following SNF discharge was 5.7% (n = 39,318). All 3 functional domains were independently associated with potentially preventable readmissions in the multivariable models. Odds ratios for the most dependent category versus the least dependent category from multilevel models adjusted for patients’ sociodemographic and clinical characteristics were as follows: mobility, 1.54 (95% confidence interval [CI] 1.49–1.59); self-care, 1.50 (95% CI 1.44–1.55); and cognition, 1.12 (95% CI 1.04–1.20). The 5 most common conditions were congestive heart failure (n = 7654, 19.5%), septicemia (n = 7412, 18.9%), urinary tract infection/kidney infection (n = 4297, 10.9%), bacterial pneumonia (n = 3663, 9.3%), and renal failure (n = 3587, 9.1%). Across all 3 functional domains, septicemia was the most common condition among the most dependent patients and congestive heart failure among the least dependent. ConclusionsPatients with functional limitations at SNF discharge are at increased risk of hospital readmissions considered potentially preventable. Future research is needed to determine whether improving functional status reduces risk of potentially preventable readmissions among this vulnerable population.

Short-term Morbidity and Readmissions Increase With Skilled Nursing Facility Discharge After Total Joint Arthroplasty in a Medicare-Eligible and Skilled Nursing Facility–Eligible Patient Cohort

BackgroundThe benefits of discharge to a skilled nursing facility (SNF) in Medicare-eligible patients after total joint arthroplasty (TJA) have recently been scrutinized. The purpose of this study was to determine short-term complication and readmission rates for SNF versus home discharge in patients eligible for Medicare and SNF discharge. MethodsPatients who underwent TJA between 2012 and 2013 were identified in the National Surgical Quality Improvement Project database. Patients over 65 years and who discharged at or after postoperative day 3, and thus SNF eligible by Medicare rule, were included. Patient demographics and comorbidities were compared in the 2 cohorts (home versus SNF), and subsequent univariate and multivariate analyses were used to determine risk factors for short-term complications. ResultsWe identified 34,610 Medicare- and SNF-eligible TJA patients; 54.8% discharged home. Patients with SNF discharge were older, had higher rates of comorbidities, and were more frequently American Society of Anesthesiologists class 3 or 4 (P < .001). Univariate analysis revealed that patients with SNF discharge had higher rates of any complication (7.9% vs. 4.7%, P < .001) and readmission (5.3% vs. 3.3%, P < .001). Multivariate regression analysis identified SNF discharge (adjusted odds ratio 1.9, 95% confidence interval 1.7-2.0) as an independent risk factor for a 30-day complication and readmission. ConclusionsIn a cohort of Medicare- and SNF-eligible patients, SNF discharge was the strongest predictor of 30-day complication after TJA. SNF discharge was also an independent predictor of readmission after TJA.

Influence of Nonindex Hospital Readmission on Length of Stay and Mortality.

Hospitals and health care systems face increasing accountability for postdischarge outcomes of patients, but it is unclear how frequently hospital readmissions in particular occur at a different hospital than the index hospitalization and whether this is associated with worse outcomes. Describe the prevalence of nonindex 30-day readmissions in a nationally representative sample of all payers and associations with outcomes. Secondary retrospective analysis of the 2013 Nationwide Readmissions Database. Nonfederal hospitals from 21 states representing half of hospitalizations in the United States annually. Our overall sample included all adults discharged alive from an inpatient stay with 30 days of follow-up; we also created 3 additional cohorts: patients with Medicare as the payer (Medicare cohort), patients discharged to home health or skilled nursing facilities after discharge (postacute care cohort), and Medicare patients with any of the current Hospital Readmission Reduction Program's penalized conditions (readmission penalty cohort). Readmission within 30 days to "index" hospital (where index stay occurred) or "nonindex" hospital. In-hospital mortality and length of stay during the readmission. The weighted overall sample included 22,884,505 hospital discharges from 2004 unique hospitals. The overall 30-day readmission rate was 11.9%, of these, 22.5% occurred at a nonindex hospital. Readmissions to nonindex facilities were associated with increased odds of in-hospital mortality (odds ratio, 1.21; 95% confidence interval, 1.17-1.25) and longer hospital length of stay (hazard ratio for hospital discharge, 0.87; 95% confidence interval, 0.86-0.88) in the overall sample and in the 3 cohorts. Nonindex readmissions are common and associated with worse outcomes; the common findings across cohorts highlight the importance for hospitals and care systems participating in value-based payment models. Hospitals and care systems should invest in improved methods for real-time identification and intervention for these patients.

Characterizing Autoimmune Disease-associated Diffuse Large B-cell Lymphoma in a SEER–Medicare Cohort

BackgroundSevere immune dysregulation such as seen in autoimmune (AI) disease is known to act as a significant risk factor for diffuse large B-cell lymphoma (DLBCL). However, little is known about the demographics or clinical outcomes of DLBCL that arises in the setting of AI disease. Patients and MethodsWe used the Surveillance, Epidemiology, and End Results (SEER) database for patients with a diagnosis from 1999 to 2009 linked to their Medicare claims data through 2011 to characterize the presentation, treatment, and survival patterns in DLBCL patients, including those with rheumatoid arthritis, systemic lupus erythematosus (SLE), Sjögren syndrome, and other B-cell–mediated AI diseases. We examined the baseline clinical characteristics for patients with B-cell–mediated AI disease, plotted the overall survival and lymphoma-related survival (LRS) for these groups, and compared the median survival times. ResultsPatients with DLBCL and AI disease were more commonly female. However, patients with DLBCL and rheumatoid arthritis, SLE, Sjögren syndrome, or other B-cell AI diseases did not differ from other DLBCL patients in any other baseline presenting features and received similar first-line treatment. A trend toward decreased LRS was seen in patients with SLE and DLBCL compared with all other groups, but this difference was not statistically significant in this cohort. ConclusionIn the present retrospective claims-based cohort of older patients with DLBCL, concomitant AI disease was uncommon and was more likely to occur in female DLBCL patients, which likely reflects the greater incidence of AI disease in women. The possibility of lower LRS for SLE patients should be explored in future studies.

Herpes Zoster as a Risk Factor for Incident Giant Cell Arteritis.

Histopathologic studies have implicated herpes zoster (HZ) as a causative organism of giant cell arteritis (GCA). The purpose of this study was to assess the epidemiologic association of HZ events with incident GCA. We performed a retrospective cohort study in 2 large independent US administrative data sets: Medicare 5% and Truven Health Analytics MarketScan. Eligible subjects had 12 months of continuous coverage, were >50 years old, and had no history of GCA or polymyalgia rheumatica. HZ events (complicated and uncomplicated) and GCA were identified by the presence of International Classification of Diseases, Ninth Revision, Clinical Modification codes from physician visit or hospital discharge records. Antiviral therapies and vaccinations were identified from prescription claims and drug codes. Risk of incident GCA was calculated using multivariable Cox proportional hazards regression. Among 16,686,345 subjects, a total of 5,942 GCA cases occurred, with 3.1% (MarketScan) and 6.0% (Medicare) having preceding HZ events. Unadjusted GCA incidence rates were highest in the groups with complicated and uncomplicated HZ. After multivariable adjustment, complicated HZ was associated with an increased risk of GCA (hazard ratio [HR] 1.99 [95% confidence interval (95% CI) 1.32-3.02] in the Medicare cohort and 2.16 [95% CI 1.46-3.18] in the MarketScan cohort), as was uncomplicated HZ (HR 1.42 [95% CI 1.02-1.99] and HR 1.45 [95% CI 1.05-2.01] in the respective cohorts). Vaccination and antiviral treatment were not consistently associated with GCA risk, although antiviral treatment was marginally associated with a decreased risk of GCA in the Medicare cohort (HR 0.67 [95% CI 0.46-0.99]). HZ is associated with an increased risk of GCA. The infrequency of HZ in GCA patients suggests that it is only one potential trigger for GCA. Antivirals and vaccination did not consistently mitigate this risk.

Adaptation of the Acute Organ Failure Score for Use in a Medicare Population.

Without widely available physiologic data, a need exists for ICU risk adjustment methods that can be applied to administrative data. We sought to expand the generalizability of the Acute Organ Failure Score by adapting it to a commonly used administrative database. Retrospective cohort study. One hundred fifty-one hospitals in Pennsylvania. A total of 90,733 ICU admissions among 77,040 unique patients between January 1, 2009, and December 1, 2009, in the Medicare Provider Analysis and Review database. We used multivariable logistic regression on a random split cohort to predict 30-day mortality, and to examine the impact of using different comorbidity measures in the model and adding historical claims data. Overall 30-day mortality was 17.6%. In the validation cohort, using the original Acute Organ Failure Score model's β coefficients resulted in poor discrimination (C-statistic, 0.644; 95% CI, 0.639-0.649). The model's C-statistic improved to 0.721 (95% CI, 0.711-0.730) when the Medicare cohort was used to recalibrate the β coefficients. Model discrimination improved further when comorbidity was expressed as the COmorbidity Point Score 2 (C-statistic, 0.737; 95% CI, 0.728-0.747; p < 0.001) or the Elixhauser index (C-statistic, 0.748; 95% CI, 0.739-0.757) instead of the Charlson index. Adding historical claims data increased the number of comorbidities identified, but did not enhance model performance. Modification of the Acute Organ Failure Score resulted in good model discrimination among a diverse population regardless of comorbidity measure used. This study expands the use of the Acute Organ Failure Score for risk adjustment in ICU research and outcomes reporting using standard administrative data.