Visualizing External Validity: Graphical Displays to Inform the Extension of Treatment Effects from Trials to Clinical Practice.

BACKGROUND: In the presence of effect measure modification, estimates of treatment effects from randomized controlled trials may not be valid in clinical practice settings. The development and application of quantitative approaches for extending treatment effects from trials to clinical practice settings is an active area of research. METHODS: In this article, we provide researchers with a practical roadmap and four visualizations to assist in variable selection for models to extend treatment effects observed in trials to clinical practice settings and to assess model specification and performance. We apply this roadmap and visualizations to an example extending the effects of adjuvant chemotherapy (5-fluorouracil vs. plus oxaliplatin) for colon cancer from a trial population to a population of individuals treated in community oncology practices in the United States. RESULTS: The first visualization screens for potential effect measure modifiers to include in models extending trial treatment effects to clinical practice populations. The second visualization displays a measure of covariate overlap between the clinical practice populations and the trial population. The third and fourth visualizations highlight considerations for model specification and influential observations. The conceptual roadmap describes how the output from the visualizations helps interrogate the assumptions required to extend treatment effects from trials to target populations. CONCLUSIONS: The roadmap and visualizations can inform practical decisions required for quantitatively extending treatment effects from trials to clinical practice settings.

Comparing Trial and Real-world Adjuvant Oxaliplatin Delivery in Patients With Stage III Colon Cancer Using a Longitudinal Cumulative Dose.

Importance: Delivery of adjuvant chemotherapy can differ substantially between trial and real-world populations. Adherence metrics like relative dose intensity (RDI) cannot capture the timing of modifications and mask differences in the total amount of chemotherapy received. Objective: To compare oxaliplatin delivery between MOSAIC trial participants and patients treated in the US Oncology Network with stage III colon cancer using a longitudinal cumulative dose (LCD). Design, Setting, and Participants: This cohort study used secondary data from the MOSAIC trial, an international randomized clinical trial (concluded in 2004), and electronic health records from US Oncology (2009-2018), a network of community oncology practices in the US. It included participants in MOSAIC with stage III colon cancer who were randomized to receive treatment with oxaliplatin and fluorouracil/leucovorin (n = 663) and US Oncology patients with stage III colon cancer who were treated with a modified FOLFOX-6 regimen (n = 2523). Exposures: Oxaliplatin and fluorouracil/leucovorin. Outcomes and Measures: We evaluated RDI and LCD over time and at the end of treatment in the MOSAIC and US Oncology populations. We used bootstrapping to estimate 95% confidence bands for LCD differences between the populations. Results: The 663 MOSAIC participants (296 women [44.7%]) and 2523 US Oncology patients (1245 women [49.4%]) were generally similar with respect to demographic characteristics. Median RDI was lower in US Oncology (80% in MOSAIC vs 70% in US Oncology). The LCD also suggested differences in the total amount of oxaliplatin received between populations; the final median LCD in US Oncology was 10.2% lower than in MOSAIC, equivalent to receiving 1.2 fewer treatment cycles less of oxaliplatin. This difference only began 133 days into treatment and persisted after accounting for covariates, likely in terms of more frequent oxaliplatin treatment discontinuation in US Oncology patients than their MOSAIC counterparts. Conclusions and Relevance: The study results suggest that real-world patients in community practice in the US treated with modified FOLFOX 6 received less oxaliplatin than their historical counterparts in the MOSAIC trial, with differences manifesting late in the treatment course. The LCD allowed us to identify the amount and extent of these differences, the timing of which was unclear when using RDI alone. Trial Registration: ClinicalTrials.gov identifier: NCT00275210.

Practice patterns among oncologists participating in the oncology care model after three years.

BACKGROUND: The Oncology Care Model (OCM) incentivized care coordination and cost-efficiency and has been associated with short-term care reductions, but its multi-year associations are less well-studied. METHODS: We used monthly provider-level claims data spanning nearly five years between July 1 st, 2014 and May 30th, 2019 from a large community oncology practice network where roughly half of the practices participated in the OCM. The key outcome measures were monthly mean office visits, costs, and buy-and-bill drug costs among prostate, colon, breast, and lung cancers. We conducted two quasi-experimental analyses: an event study, which measures the monthly association of providing care in an OCM relative to a non-participating practice, and a difference-in-differences model, which summarizes the event study results into post-launch average estimates. We controlled for mean differences between practices, providers, and patient. RESULTS: The event study analysis shows similar pre-period estimates and trends for each cancer. The difference-in-differences estimates for office visits are statistically significant for each cancer: 33 percentage point reductions in prostate cancer (95 % CI: -0.66 to 0.00; p = 0.05), 22 percentage point reductions in colon cancer (95 % CI: -0.48 to 0.04; p = 0.09), 21 percentage point reductions in breast cancer (95 % CI: -0.45 to 0.02; p = 0.08), and 24 percentage point reductions in lung cancer (95 % CI: -0.49 to 0.00; p = 0.05). Monthly prostate cancer costs also reduced by $505 (95 % CI: -$1108 to $99; p = 0.10). CONCLUSION: Our results suggest that the OCM was associated with relative reductions in office visits and, for prostate cancer, in overall costs too. These associations generally decreased within the first year of launch and sustained through roughly two years. POLICY SUMMARY: Novel payment models that incentivize care coordination and cost-efficiency like the OCM may modestly yet sustainably reduce office visits and overall costs.

Clinical outcomes among unresectable, locally advanced, and metastatic cutaneous squamous cell carcinoma patients treated with systemic therapy.

Prior studies of conventional chemotherapy or epidermal growth factor receptor inhibitors for advanced (ie, locally advanced cutaneous squamous cell carcinoma [laCSCC] or metastatic [mCSCC]) cutaneous squamous cell cancer enrolled ≤ 40 patients. This retrospective, observational study assessed real-world treatment patterns and clinical outcomes in patients with unresectable laCSCC or mCSCC using electronic health records of patients who initiated first-line (1L) systemic treatment from 1 January 2008 to 31 December 2015, with follow-up to 30 September 2017. The median duration of follow-up from 1L treatment was 10.1 months (range 0.03-67.6 months). Duration of therapy (DOT) and overall survival (OS) were assessed using Kaplan-Meier analysis. Response rate was calculated as the proportion of patients who achieved physician-assessed-response. Eighty-two patients were identified (17 laCSCC and 65 mCSCC). Median age at 1L treatment initiation was 75 years; 85% were male, 88% had an Eastern Cooperative Oncology Group performance status of 1, and 84% had received radiotherapy. The most common 1L regimens were carboplatin + paclitaxel (27%) and cetuximab monotherapy (24%). The median 1L DOT was 4.1 months for laCSCC and 2.3 months for mCSCC. The physician-assessed response rate for 1L therapy was 17.6% for laCSCC, and 18.5% for mCSCC. The median OS from 1L treatment initiation was 16.2 months for laCSCC, and 15.3 months for mCSCC. Only 24 patients (29%) received second-line therapy. This is the largest retrospective data set regarding patients with advanced CSCC treated with anticancer systemic therapy prior to approval of the anti-programmed cell death-1 antibody, cemiplimab. Efficacy was low in both laCSCC and mCSCC. These data provide historic benchmarks for outcomes in patients with advanced CSCC prior to Food and Drug Administration approval of cemiplimab-rwlc.

Evaluation of Practice Patterns Among Oncologists Participating in the Oncology Care Model.

Importance: Health insurers reimburse clinicians in many ways, including the ubiquitous fee-for-service model and the emergent shared-savings models. Evidence on the effects of these emergent models in oncological treatment remains limited. Objectives: To analyze the early use and cost associations of a recent Medicare payment program, the Oncology Care Model (OCM), which included a shared savings-like component. Design, Setting, and Participants: This nonrandomized controlled study used a difference-in-differences approach on 2 years of data, from July 1, 2015, to June 30, 2017-1 year before and 1 year after launch of the OCM-to compare the differences between participating and nonparticipating practices, controlling for patient, clinician, and practice factors. Participation in the OCM began on July 1, 2016. Associations of participation with care use and cost were estimated for care directly managed by clinicians from a large network within their Medicare populations for breast, lung, colon, and prostate cancers. Data were analyzed from September 2019 to March 2020. Exposures: Participating practices were paid a monthly management fee of $160 per beneficiary and a potential risk-adjusted performance-based payment for eligible patients who received chemotherapy treatment, in addition to standard fee-for-service payments. Main Outcomes and Measures: Office visits, drug administrations, patient hydrations, drug costs, and total costs. Results: Monthly means data at the physician-level were evaluated for 11 869 physician-months for breast cancers, 11 135 physician-months for lung cancers, 8592 physician-months for colon cancers, and 9045 physician-months for prostate cancers. Patients at OCM practices had a mean (SD) age of 63.4 (3.1) years, and a mean (SD) of 59% (7 percentage points) of their patients were women. Participation in the OCM was associated with less physician-administered prostate cancer drug use (difference, 0.29 [95% CI, -0.47 to -0.11] percentage points, or 24.0%) translating to a mean of $706 (95% CI, -$1383 to -$29) less in drug costs per month. Monthly drug costs were also lower, at $558 (95% CI, -$1173 to $58) less for treatment for lung cancer. Total costs were lower by 9.7% or $233 (95% CI, -$495 to $30) for breast cancer, 9.9% or $337 (95% CI, -$618 to -$55) for lung cancer, 14.2% or $385 (95% CI, -$780 to $10) for colon cancer, and 29.2% or $610 (95% CI, -$1095 to -$125) for prostate cancer; however, these differences were largely offset by program costs. Clinician visits were also lower by 11.2% or 0.11 (95% CI, -0.20 to -0.01) percentage points among patients with breast cancer and by 14.4% or 0.19 (95% CI, -0.37 to -0.02) among patients with colon cancer. Conclusions and Relevance: These findings suggest that payment models with shared-savings components can be associated with fewer visits and lower costs in certain cancer settings in the first year, but the savings can be modest given the costs of program administration.

Treatment Patterns and Clinical Outcomes in Patients With Metastatic Gastroenteropancreatic Neuroendocrine Tumors Treated in the Community Practice Setting in the United States.

OBJECTIVE: This study was conducted to understand treatment patterns and clinical outcomes in metastatic gastroenteropancreatic neuroendocrine tumor patients treated in a large community oncology network. METHODS: This retrospective study used the McKesson Specialty Health/US Oncology Network iKnowMed electronic health record database with supplemental chart review. Eligibility criteria included a metastatic neuroendocrine tumor diagnosis between January 1, 2008, and to December 31, 2012; at least 2 US Oncology Network visits; and age at least 18 years. Follow-up was through October 31, 2014. RESULTS: Among the 229 patients identified, median age was 64.0 years, 52.4% were male, 69.4% were white, and 62.9% were overweight/obese. Primary tumor sites included small bowel (47.6%), pancreas (31.4%), and stomach/colorectum (21.0%). There were 16.2% under observation without treatment, 52.4% received only somatostatin analogs (SSAs), and 31.4% received chemotherapy/targeted therapy during treatment. In the first-line setting (n = 192), 77% received SSAs, 12% received chemotherapy, and 10.9% received targeted therapy. Fifty percent of patients receiving octreotide had a relative dose intensity of less than 85%, and 16.7% received above-label dose. Toxicities of SSAs included diarrhea (18.2%), abdominal pain (16.9%), and fatigue (13.5%). Median overall survival from diagnosis was 68.0 months (95% confidence interval, 57.1 to not reached). CONCLUSIONS: Most metastatic gastroenteropancreatic neuroendocrine tumor patients received systemic treatment with SSAs. Patient treatment used an individualized dosing approach. Overall survival and toxicity were consistent with the published literature.

Pembrolizumab Utilization and Outcomes for Advanced Melanoma in US Community Oncology Practices.

The programmed death-1 inhibitor pembrolizumab has demonstrated efficacy and safety in clinical trials for treating advanced (unresectable/metastatic) melanoma. We investigated the real-world utilization of pembrolizumab and associated patient outcomes for advanced melanoma in US community oncology practices. This retrospective, observational study used deidentified data from electronic health records for adult patients with advanced melanoma who received pembrolizumab at The US Oncology Network sites from September 2014 through December 2015, with follow-up through September 2016. Patients enrolled in clinical trials were excluded. Overall survival (OS) and physician-stated progression-free survival (PFS) were analyzed from pembrolizumab initiation using Kaplan-Meier, and associations between pembrolizumab therapy and OS/PFS, using multivariable Cox regression. Of 168 patients studied, 110 (65%) were male; the median age was 66 years (range, 26-over 90). Pembrolizumab was prescribed as first-line, second-line, and third-line/later for 39 (23%), 87 (52%), and 42 (25%) patients, respectively. In total, 41 patients (24%) had brain metastases. At pembrolizumab initiation, 21/129 (16%) had Eastern Cooperative Oncology Group performance status (ECOG PS) >1; 51/116 (44%) had elevated lactate dehydrogenase. Median follow-up was 10.5 months (range, 0-25.1); median OS was 19.4 months (95% confidence interval, 14.0-not reached); median PFS was 4.2 months (95% confidence interval, 2.9-5.3). Brain metastases, ECOG PS>1, elevated lactate dehydrogenase, and third-line/later (vs. first-line) pembrolizumab were significant predictors (P<0.01) of decreased survival. Treatment-related toxicity was a discontinuation reason for 25% (29/117) of patients, and for 10 of these 29 patients (6% of the full-study cohort) treatment-related toxicity was the only reported reason. The real-world effectiveness and safety of pembrolizumab for advanced melanoma are consistent with clinical trial findings.

Depression diagnoses following the identification of bipolar disorder: costly incongruent diagnoses.

BACKGROUND: Previous research has documented that the symptoms of bipolar disorder are often mistaken for unipolar depression prior to a patient's first bipolar diagnosis. The assumption has been that once a patient receives a bipolar diagnosis they will no longer be given a misdiagnosis of depression. The objectives of this study were 1) to assess the rate of subsequent unipolar depression diagnosis in individuals with a history of bipolar disorder and 2) to assess the increased cost associated with this potential misdiagnosis. METHODS: This study utilized a retrospective cohort design using administrative claims data from 2002 and 2003. Patient inclusion criteria for the study were 1) at least 2 bipolar diagnoses in 2002, 2) continuous enrollment during 2002 and 2003, 3) a pharmacy benefit, and 4) age 18 to 64. Patients with at least 2 unipolar depression diagnoses in 2003 were categorized as having an incongruent diagnosis of unipolar depression. We used propensity scoring to control for selection bias. Utilization was evaluated using negative binomial models. We evaluated cost differences between patient cohorts using generalized linear models. RESULTS: Of the 7981 patients who met all inclusion criteria for the analysis, 17.5% (1400) had an incongruent depression diagnosis (IDD). After controlling for background differences, individuals who received an IDD had higher rates of inpatient and outpatient psychiatric utilization and cost, on average, an additional $1641 per year compared to individuals without an IDD. CONCLUSIONS: A strikingly high proportion of bipolar patients are given the differential diagnosis of unipolar depression after being identified as having bipolar disorder. Individuals with an IDD had increased acute psychiatric care services, suggesting higher levels of relapses, and were at risk for inappropriate treatment, as antidepressant therapy without a concomitant mood-stabilizing medication is contraindicated in bipolar disorder. Further prospective research is needed to validate the findings from this retrospective administrative claims-based analysis.

Discharge destination's effect on bounce-back risk in Black, White, and Hispanic acute ischemic stroke patients.

OBJECTIVE: To determine whether racial and ethnic effects on bounce-back risk (ie, movement to settings of higher care intensity within 30 d of hospital discharge) in acute stroke patients vary depending on initial posthospital discharge destination. DESIGN: Retrospective analysis of administrative data. SETTING: Four hundred twenty-two hospitals, southern/eastern United States. PARTICIPANTS: All Medicare beneficiaries 65 years or more with hospitalization for acute ischemic stroke within one of the 422 target hospitals during the years 1999 or 2000 (N=63,679). INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Adjusted predicted probabilities for discharge to and for bouncing back from each initial discharge site (ie, home, home with home health care, skilled nursing facility [SNF], or rehabilitation center) by race (ie, black, white, and Hispanic). Models included sociodemographics, comorbidities, stroke severity, and length of stay. RESULTS: Blacks and Hispanics were significantly more likely to be discharged to home health care (blacks=21% [95% confidence interval (CI), 19.9-22.8], Hispanic=19% [17.1-21.7] vs whites=16% [15.5-16.8]) and less likely to be discharged to SNFs (blacks=26% [95% CI, 23.6-29.3], Hispanics=28% [25.4-31.6] vs whites=33% [31.8-35.1]) than whites. However, blacks and Hispanics were significantly more likely to bounce back when discharged to SNFs than whites (blacks=26% [95% CI, 24.2-28.6], Hispanics=28% [24-32.6] vs whites=21% [20.3-21.9]). Hispanics had a lower risk of bouncing back when discharged home than either blacks or whites (Hispanics=14% [95% CI, 11.3-17] vs blacks=20% [18.4-22.2], whites=18% [16.8-18.3]). Patients discharged to home health care or rehabilitation centers demonstrated no significant differences in bounce-back risk. CONCLUSIONS: Racial/ethnic bounce-back risk differs depending on initial discharge destination. Additional research is needed to fully understand this variation in effect.

Azithromycin extended release vs amoxicillin/clavulanate: symptom resolution in acute sinusitis.

OBJECTIVE: The aim of the study was to compare early symptom resolution with a single 2-g dose of azithromycin extended release or 10 days of amoxicillin/clavulanate 875 mg/125 mg every 12 hours in patients with acute sinusitis. MATERIALS AND METHODS: This was a prospective, randomized, open-label, observational study to mimic "real-world" conditions, including patients with symptoms of acute bacterial sinusitis lasting between 7 and 30 days. Key symptoms were assessed twice daily by patient diary, and patients were interviewed by telephone at 12 and 28 days. The primary end point was symptom resolution at 5 days, defined as reporting "no problem" with at least 3 of 4 diary symptoms in 2 consecutive measures in the per-protocol population. Secondary end points included additional antibiotic use, sinusitis-related quality of life, and treatment satisfaction. RESULTS: Three hundred seventy-eight patients were randomized to a single dose of azithromycin extended release and 371 to 10 days of amoxicillin/clavulanate. In the per-protocol population at day 5, 70/236 patients (29.7%) in the azithromycin extended release arm and 45/238 patients (18.9%) in the amoxicillin/clavulanate arm had resolution of symptoms (difference = 10.8%; 95% confidence interval [CI], 3.1-18.4%). By day 28, 26/236 patients (11.0%) in the azithromycin extended release arm and 27/238 patients (11.3%) in the amoxicillin/clavulanate arm had used additional antibiotics (difference = -0.4%; 95% CI: -6.1% to 5.3%). Additional physician visits, quality of life, and overall satisfaction were similar between groups. CONCLUSIONS: More patients randomized to azithromycin extended release experienced symptom resolution at day 5 than those randomized to amoxicillin/clavulanate, without experiencing differences in second antibiotic use at 28 days.

Estimation of economic costs associated with transfusion dependence in adults with MDS.

OBJECTIVE: To examine the economic burden of myelodysplastic syndromes (MDS) and the incremental cost of transfusion dependence. RESEARCH DESIGN AND METHODS: Adults with evidence of MDS were identified between 05/01/2000 and 09/30/2003 from a longitudinal, retrospective claims database for a large, geographically diverse US health plan and their medical histories were followed for at least 6 months. Patients were classified as transfusion-dependent (MDS-TD) or transfusion-independent (MDS-TI). MAIN OUTCOME MEASURES: Variables were categorized as demographic, health status, utilization, or cost. Utilization (inpatient hospitalizations, outpatient facility visits, emergency department visits, and physician office visits) is reported as the mean and median numbers of each specified encounter per subject. Costs were measured as the sum of patient and plan liability. All variables were analyzed descriptively, and appropriate statistical tests were used to compare the MDS-TD and MDS-TI cohorts. Pharmacy, medical, and total health care costs, adjusted for demographics and comorbidity, were estimated using gamma regression with a log link. RESULTS: The MDS-TI cohort consisted of 2864 patients, and the MDS-TD cohort comprised 336 patients. Mean age for the entire study sample was 70.2 years. The MDS-TI cohort tended to receive most of its medical care at physicians' offices, whereas the MDS-TD cohort received nearly as much medical care at outpatient facilities (e.g., infusion clinics, hospital outpatient clinics) as it did in physicians' offices. The MDS-TD cohort had significantly higher mean annual costs: pharmacy, $4457 vs. $2926; medical, $50,663 vs. $17,469; total, $51,066 vs. $19,811 (p < 0.001 for all comparisons). Thus, transfusion dependence was associated with an incremental cost of $31,255 per patient per year. Some limitations inherent to using claims data and diagnosis codes for research apply to this study. CONCLUSIONS: This study demonstrated that an important consequence of transfusion dependence for MDS patients was markedly greater use of, and consequently higher costs associated with, inpatient and outpatient services. Continued research and efforts to develop biologic and pharmaceutical therapies may help more patients achieve transfusion independence, thereby reducing the financial burden of MDS.

The quality of diabetes care following hospitalization for ischemic stroke.

BACKGROUND: Follow-up is critically important for stroke survivors with diabetes, yet there is limited research about the quality of diabetes care that these patients receive. We investigated performance on diabetes quality of care indicators for stroke survivors overall and by race. METHODS: Claims data was extracted for 1,460 Medicare beneficiaries with preexisting diabetes who survived hospitalization for acute ischemic stroke in 2000. Adjusted probabilities of receiving HbA1c, LDL and dilated eye exams were estimated using logistic regression. RESULTS: 53% had a dilated eye exam, 60% received an LDL check, 73% percent had their HbA1c checked at least once and only 51% received two or more HbA1c checks. In the unadjusted results, blacks were significantly less likely than whites to receive these tests. CONCLUSIONS: Care of stroke survivors, particularly blacks, shows gaps according to guidelines.

Predictors of hospice utilization among acute stroke patients who died within thirty days.

BACKGROUND: Hospice is considered to be underutilized, particularly among patients with noncancer diagnoses such as stroke. The highest mortality among stroke patients occurs within the first 30 days; however, we know little about the hospice enrollment decision for this population during this critical time frame. OBJECTIVES: To determine hospice enrollment rates and to describe sociodemographic and clinical predictors of hospice utilization among patients who die within 30 days of their stroke. DESIGN: Retrospective analysis of administrative data. SUBJECTS: Medicare beneficiaries 65 years and older discharged with ischemic stroke from 422 hospitals and 11 metropolitan regions during the year 2000 who died within 30 days of their stroke. MEASURES: Hospice utilization within 30 days. RESULTS: The overall hospice enrollment rate in our study was 23%. Using multivariable logistic regression, factors predicting increased hospice enrollment included older age, female gender, health management organization (HMO) membership, length of stay more than 3 days, and dementia. Factors predicting decreased enrollment included African American race, mechanical ventilation, gastrostomy tube placement, uncomplicated diabetes mellitus, and valvular disease. When in-hospital deaths were excluded, overall enrollment increased to 44%, and mechanical ventilation and dementia ceased to predict enrollment. CONCLUSIONS: Hospice enrollment rates among patients who die within the first 30 days of their stroke, particularly among those who survive to discharge, are much higher than prior estimates suggest. Although overall enrollment rates were higher than anticipated, there remain important sociodemographic and clinical characteristics unique to this population that predict low hospice utilization that should serve as targets for further research and intervention.

Health service utilization among Alzheimer's disease patients: evidence from managed care.

BACKGROUND: The objective of this study was to assess the disease burden of Alzheimer's disease (AD) in a commercial managed care setting by comparing direct health care costs and adverse event outcomes between patients with AD and without AD. METHODS: The study design used eligibility, medical, and pharmacy claims data from a large, national, geographically diverse, fee-for-service U.S. managed health plan. Commercially insured patients aged 65 years and older with a pharmacy benefit with evidence of AD (n = 4,450) and a control group without AD (n = 13,650) were matched by age, gender, plan location, and length of enrollment. Adverse event outcomes, comorbid conditions, and annualized health care costs were compared. Incremental costs were calculated by using a two-part model to estimate the burden of illness; incremental cost confidence intervals were estimated by bootstrap analysis. RESULTS: Patients with AD had generally higher health care costs and higher risk of acute adverse outcomes than the control cohort. Annual adjusted total health care costs per patient were approximately $1,418 greater for the AD cohort. Patients with AD had an unadjusted fracture risk of 14.6% versus 6.2% in the matched cohort and accidental injury/falls risk of 27.4% versus 11.4%. CONCLUSIONS: Few studies have examined the disease burden of AD in commercial managed care settings. Similar to results of comparative studies with Medicare data, the disease burden is greater for patients with AD compared with a matched control cohort, with a different mix and a greater number of comorbid health care conditions partially accounting for this difference. As membership in commercial and Medicare managed care plans increases, plans will need to develop effective mechanisms to manage the health care of high-risk, high-cost patients with AD.

The price of bouncing back: one-year mortality and payments for acute stroke patients with 30-day bounce-backs.

OBJECTIVES: To examine 1-year mortality and healthcare payments of stroke patients experiencing zero, one and two or more bounce-backs within 30 days of discharge. DESIGN: Retrospective analysis of administrative data. SETTING: Four hundred twenty-two hospitals in the southern and eastern United States. PARTICIPANTS: Eleven thousand seven hundred twenty-nine Medicare beneficiaries aged 65 and older surviving at least 30 days with acute ischemic stroke in 2000. MEASUREMENTS: One-year mortality and predicted total healthcare payments were calculated using log-normal parametric survival analysis and quantile regression, respectively. Models included sociodemographics, prior medical history, stroke severity, length of stay, and discharge site. RESULTS: Crude survival at 1 year for the zero, one and two or more bounce-back groups was 83%, 67%, and 55%, respectively. The one bounce-back group had 49% shorter (time ratio (TR)=0.51, 95% confidence interval (CI)=0.46-0.56) and the two or more bounce-backs group had 68% shorter (TR=0.32, 95% CI=0.27-0.38) adjusted 1-year survival time than the zero bounce-back group. For high- and low-cost patients, adjusted predicted payments were greater with each additional bounce-back experienced. CONCLUSION: Acute stroke patients experiencing bounce-backs within 30 days have strikingly poorer survival and higher healthcare payments over the subsequent year than their counterparts with no bounce-backs. Bounce-backs may serve as a simple predictor for identifying stroke patients at extremely high risk for poor outcomes.

Diagnosis of unipolar depression following initial identification of bipolar disorder: a common and costly misdiagnosis.

BACKGROUND: Bipolar disorder is challenging to diagnose in medical practice. OBJECTIVES: Our objectives were (1) to determine the rate of depression misdiagnosis in patients previously diagnosed with bipolar disorder in administrative claims, (2) to determine the resulting increased treatment costs, and (3) to verify the misdiagnoses in the medical charts for a subset of patients. METHOD: We employed cohort analysis using claims from a large, commercial, U.S. health plan from January 2001 through December 2003. Inclusion criteria included 2 bipolar disorder diagnoses (ICD-9-CM criteria), continuous enrollment for 1 year before and after initial bipolar disorder diagnosis, age 18-64 years, and a pharmacy benefit. Propensity scoring was used to control for differences between patients with and without 2 depression diagnoses in the year following their bipolar disorder diagnosis. Medical charts were obtained for 100 patients, including 76 with a bipolar disorder diagnosis chart from one provider and a depression diagnosis chart from a second provider. RESULTS: Of 3119 bipolar disorder patients meeting inclusion criteria, 857 (27.5%) had subsequent depression misdiagnoses during the follow-up year. These patients had 1.82 times more psychiatric hospitalizations and 2.47 times more psychiatric emergency room visits. For 673 patients (78.5%), a different provider gave the depression misdiagnosis. Annual per-patient treatment costs were significantly higher (p < .001) for those diagnosed with depression ($12,594) than for those not ($9405). In the chart review, both the bipolar disorder and subsequent depression diagnoses were confirmed for 65.8% (50/76) of the patients who had charts from 2 different providers. CONCLUSIONS: More than one quarter of individuals diagnosed with bipolar disorder received an ostensible depression misdiagnosis during the follow-up period. Significant (p = .001) increases in psychiatric inpatient hospitalization suggest that improvements in the continuity of care could improve outcomes and reduce costs.

Bouncing-back: rehospitalization in patients with complicated transitions in the first thirty days after hospital discharge for acute stroke.

BACKGROUND: "Bounce-backs" (movements from a less intensive to a more intensive care setting) soon after hospital discharge are common, but reasons for bouncing-back remain unknown. OBJECTIVE: To examine how the primary diagnosis for first rehospitalization relates to thirty-day bounce-back number and initial discharge destination in acute stroke. POPULATION: Administrative data from 5,250 Medicare beneficiaries > or = 65 years discharged with acute ischemic stroke in 1998-2000 to a rehabilitation center, skilled nursing facility or home with home health care and with at least one thirty day rehospitalization. ANALYSIS: Probability of thirty-day bounce-back was calculated using multivariate models. RESULTS: Infections and aspiration pneumonitis were the most common reasons for rehospitalization, regardless of initial discharge site. CONCLUSIONS: Efforts addressing aspirations and infections, the preventable complications of immobility, will be critical in decreasing acute stroke bounce-backs.

The complexity of medication regimens and test ordering for patients with diabetes from 1995 to 2003.

OBJECTIVE: Diabetes care has become increasingly complex. We set out to quantify recent trends in the complexity of medication regimens and test ordering for diabetes patients continuously enrolled in health plans affiliated with a large, regional US health maintenance organization, with representation in the South and Midwest. RESEARCH DESIGN AND METHODS: We provide descriptive trends analysis of overall diabetes care complexity (number of components [i.e., glucose, blood pressure, cholesterol control], number of medications/tests) from 1995 to 2003 for adults with diabetes (N = 304,233). MAIN OUTCOME MEASURES: The main outcomes were (1) the proportion of patients receiving diabetes-related medications (blood glucose, blood pressure, and cholesterol control agents), (2) the average number of medications, (3) the proportion of patients receiving diabetes-related tests (glycosylated hemoglobin [HbA1c], urine microalbumin, and serum cholesterol), (4) and the average number of tests ordered within the first year that a patient had any indication of diabetes. RESULTS: The proportion of patients on cholesterol lowering drugs (18% --> 39%, p < 0.01) and blood pressure lowering drugs (51% --> 62%, p = 0.04) rose significantly, while the proportion on glucose lowering drugs fell (76% --> 47%, p < 0.01). Among patients prescribed medications, the average total number of diabetes-related medications rose from 2.96 to 3.70 medications (p < 0.01) with smaller increases seen for glucose lowering (1.45 --> 1.65, p < 0.01) and blood pressure lowering regimens (2.14 --> 2.51, p < 0.01), and no change for cholesterol lowering drugs (1.23 --> 1.19, p = 0.19). For laboratory tests, the proportion receiving cholesterol (40% --> 58%), and urine microalbumin (4% --> 18%) (all p's < 0.01) rose significantly, while the testing rates for HbA(1c) remained unchanged. The average total number of tests ordered per year increased significantly from 3.34 to 4.10 (p < 0.01) with more modest increases observed for individual tests. LIMITATION: Trends analyses are unadjusted for many clinical characteristics that might influence the complexity of diabetes care. CONCLUSIONS: Diabetes care grew more complex with the largest change in the number of patients receiving multi-component diabetes care. While the use of blood pressure and cholesterol lowering drugs rose overall, the proportion of patients using glucose lowering drugs declined and the average number of prescribed glucose lowering drugs did not increase in a clinically significant manner.

A claims database analysis of persistence with alendronate therapy and fracture risk in post-menopausal women with osteoporosis.

OBJECTIVE: To explore the relationship between persistence with alendronate therapy and fracture rates in women with postmenopausal osteoporosis. RESEARCH DESIGN AND METHODS: Claims data from a large US health plan database were used to examine persistence with therapy in postmenopausal women followed for 24 months. Persistence was defined as the time (in days) from the date of first fill to the run-out date of the last prescription with no lapses > 30 days after completion of the previous refill. A persistent cohort (length of persistence > or = 182 days) and a nonpersistent cohort (length of persistence < 182 days) were defined. The number of patients with a fracture claim in each cohort was determined. Cox-proportional hazards regression (HR) analysis was used to determine significant differences in fracture rates between the two cohorts. RESULTS: 4769 patients were followed for 24 months. Patients in the persistent cohort were significantly more likely to receive a treatment (vs. prevention) dose of alendronate (p = 0.03) and to be older than 65 years (p = 0.04). There was a trend toward more fractures in the non-persistent (4.9%) than in the persistent cohort (3.9%; p = 0.09). When controlled for other significant factors (including age and previous fractures) patients in the persistent cohort were 26% less likely to have a fracture diagnosis claim during the study period than those in the non-persistent cohort (HR = 0.74; 95% CI, 0.549-0.996; p = 0.045). Prescription fill data are an indirect measure of medication-taking behavior. The use of claims data to estimate persistence and identify fracture events prohibits the establishment of causality between these two variables. CONCLUSION: Study results demonstrated that non-persistence with therapy, along with previous fracture and increasing age, was associated with a greater risk of fracture.