The assumption that comparative effectiveness research will provide timely, relevant evidence rests on changing the current framework for assembling evidence. In this commentary, we provide the background of how coverage decisions for new medical technologies are currently made in the United States. We focus on the statistical issues regarding how to use the ensemble of information for inferring comparative effectiveness. It is clear a paradigm shift in how clinical information is integrated in real-world settings to establish effectiveness is required.
evidentiary evaluation; multiple outcomes and comparisons; hierarchical Bayesian models; Bayes factors; posterior predictive probability
To describe physician practices, ranging from solo and two-physician practices to large medical groups, in three geographically diverse parts of the country with strong managed care presences.
Data Sources/Study Design
Surveys of medical practices in three managed care markets conducted in 2000–2001.
We administered questionnaires to all medical practices affiliated with two large health plans in Boston, MA, and Portland, OR, and to all practices providing primary care for cardiovascular disease patients admitted to five large hospitals in Minneapolis, MN. We offer data on how physician practices are structured under managed care in these geographically diverse regions of the country with a focus on the structural characteristics, financial arrangements, and care management strategies adopted by practices.
A two-staged survey consisting of an initial telephone survey that was undertaken using CATI (computerized assisted telephone interviewing) techniques followed by written modules triggered by specific responses to the telephone survey.
We interviewed 468 practices encompassing 668 distinct sites of care (overall response rate 72 percent). Practices had an average of 13.9 member physicians (range: 1–125). Most (80.1 percent) medium- (four to nine physicians) and large-size (10 or more physicians) groups regularly scheduled meetings to discuss resource utilization and referrals. Almost 90 percent of the practices reported that these meetings occurred at least once per month. The predominant method for paying practices was via fee-for-service payments. Most other payments were in the form of capitation. Overall, 75 percent of physician practices compensated physicians based on productivity, but there was substantial variation related to practice size. Nonetheless, of the practices that did not use straight productivity methods (45 percent of medium-sized practices and 54 percent of large practices), most used arrangements consisting of combinations of salary and productivity formulas.
We found diversity in the characteristics and capabilities of medical practices in these three markets with high managed care involvement. Financial practices of most practices are geared towards rewarding productivity, and care management practices and capabilities such as electronic medical records remain underdeveloped.
Physician groups; managed care; financial incentives
To examine whether availability of cardiac services at the admitting hospital affects case-selection for angiography and one-year survival following angiography, within groups of patients who have similar clinical need for angiography according to published criteria.
Elderly Medicare beneficiaries (37,788) discharged with a diagnosis of acute myocardial infarction (AMI) from hospitals in seven U.S. states between February 1994 and July 1995. We focused on patients who were eligible to receive angiography 12 or more hours after symptom onset.
Data were abstracted from patient's medical records, Medicare National Claims Standard Analytic Files, Health Care Financing Administration (HCFA) Provider of Service File and Health Insurance Master File.
Admitting hospitals were classified as offering no cardiac services, angiography only, or revascularization. Case-selection differences across these three types of hospitals were examined by comparing relative risk of receiving angiography for various patient and hospital characteristics. Relative differences in one-year survival rate, comparing patients who received angiography to those who did not, were estimated within each hospital type and clinical need category (necessary, appropriate, or uncertain) after matching on propensity to receive angiography.
Compared to patients for whom angiography was deemed necessary, the relative risk of receiving angiography among those for whom it was deemed of uncertain benefit was 0.58, 0.79, and 0.92 (p-value of homogeneity test < 0.001) at hospitals offering no cardiac services, angiography only, and revascularization, respectively. There was no significant difference in survival following angiography across hospital types, overall as well as within clinical need categories.
Despite increased case selection at hospitals with on-site cardiac services, there was no evidence of increase in the survival rate associated with angiography use at these hospitals.
Coronary angiography; practice guidelines; classification trees; survival
The impact of parity coverage on the quantity of behavioral health services used by enrollees and on the prices of these services was examined in a set of Federal Employees Health Benefit (FEHB) Program plans. After parity implementation, the quantity of services used in the FEHB plans declined in five service categories, compared with plans that did not have parity coverage. The decline was significant for all service types except inpatient care. Because a previous study of the FEHB Program found that total spending on behavioral health services did not increase after parity implementation, it can be inferred that average prices must have increased over the period. The finding of a decline in service use and increase in prices provides an empirical window on what might be expected after implementation of the federal parity law and the parity requirement under the health care reform law.
Randomized trials of implantable cardioverter defibrillators (ICDs) for primary prevention predominantly employed single chamber devices. In clinical practice, patients often receive dual chamber ICDs, even without clear indications for pacing. The outcomes of dual versus single chamber devices are uncertain.
Compare outcomes of single and dual chamber ICDs for primary prevention of sudden cardiac death.
Design, Setting, and Participants
Retrospective cohort study. Admissions in the National Cardiovascular Data Registry’s (NCDR®) ICD Registry™ from 2006–2009 that could be linked to CMS fee for service Medicare claims data were identified. Patients were included if they received an ICD for primary prevention and did not have a documented indication for pacing.
Main Outcome Measures
Adjusted risks of 1-year mortality, all-cause readmission, HF readmission and device-related complications within 90 days were estimated with propensity-score matching based on patient, clinician and hospital factors.
Among 32,034 patients, 38% (n=12,246) received a single chamber device and 62% (n=19,788) received a dual chamber device. In a propensity-matched cohort, rates of complications were lower for single chamber devices (3.5% vs. 4.7%; p<0.001; risk difference −1.20; 95% CI −1.72, −0.69), but device type was not significantly associated with mortality or hospitalization outcomes (unadjusted rate 9.9% vs. 9.8%; HR 0.99, 95% CI 0.91–1.07; p=0.792 for 1-year mortality; unadjusted rate 43.9% vs. 44.8%; HR 1.00, 95% CI 0.97–1.04; p=0.821 for 1-year all-cause hospitalization; unadjusted rate 14.7% vs. 15.4%; HR 1.05, 95% CI 0.99–1.12; p=0.189 for 1-year HF hospitalization).
Conclusions and Relevance
Among patients receiving an ICD for primary prevention without indications for pacing, the use of a dual chamber device compared with a single chamber device was associated with a higher risk of device-related complications but not with different risks for mortality or hospitalization. Further studies should be performed to determine if other benefits of dual chamber devices exist, such as reduced device therapy or improved quality of life, to justify their use in this context.
In disparities models, researchers adjust for differences in “clinical need,” including indicators of comorbidities. We reconsider this practice, assessing (1) if and how having a comorbidity changes the likelihood of recognition and treatment of mental illness; and (2) differences in mental health care disparities estimates with and without adjustment for comorbidities.
Longitudinal data from 2000 to 2007 Medical Expenditure Panel Survey (n = 11,083) split into pre and postperiods for white, Latino, and black adults with probable need for mental health care.
First, we tested a crowd-out effect (comorbidities decrease initiation of mental health care after a primary care provider [PCP] visit) using logistic regression models and an exposure effect (comorbidities cause more PCP visits, increasing initiation of mental health care) using instrumental variable methods. Second, we assessed the impact of adjustment for comorbidities on disparity estimates.
We found no evidence of a crowd-out effect but strong evidence for an exposure effect. Number of postperiod visits positively predicted initiation of mental health care. Adjusting for racial/ethnic differences in comorbidities increased black–white disparities and decreased Latino–white disparities.
Positive exposure findings suggest that intensive follow-up programs shown to reduce disparities in chronic-care management may have additional indirect effects on reducing mental health care disparities.
Access/demand/utilization of services; mental health; racial/ethnic differences in health and health care
To assess the frequency and predictors of vascular closure device (VCD) deployment failure, and its association with vascular complications of three commonly used VCDs.
VCDs are commonly used following percutaneous coronary intervention (PCI) on the basis of studies demonstrating reduced time to ambulation, increased patient comfort, and possible reduction in vascular complications as compared to manual compression. However, limited data are available on the frequency and predictors of VCD failure, and the association of deployment failure with vascular complications.
From a de-identified dataset provided by Massachusetts Department of Health, 23,813 consecutive interventional coronary procedures that used either a collagen plug-based (n=18,533) or nitinol clip-based (n=2,284) or suture-based (n=2,996) VCD between 06/2005 and 12/2007 were identified. We defined VCD failure as unsuccessful deployment or failure to achieve immediate access site hemostasis.
Among 23,813 procedures, VCD failed in 781 (3.3%) procedures (2.1% of collagen plug-based, 6.1% of suture-based, 9.5% of nitinol clip-based). Patients with VCD failure had excess risk of ‘any’ (7.7% versus 2.8%; P<0.001), major (3.3% versus 0.8%; P<0.001), or minor (5.8% versus 2.1%; P<0.001) vascular complications compared with successful VCD deployment. In a propensity-score adjusted analysis, when compared with collagen plug-based VCD (Reference OR =1.0), nitinol clip-based VCD had 2-fold increased risk (OR 2.0, 95% CI: 1.8–2.3, p<0.001) and suture-based VCD had 1.25-fold increased risk (OR 1.25, 95% CI: 1.2–1.3, p<0.001) for VCD failure. VCD failure was a significant predictor of subsequent vascular complications for both collagen plug-based VCD and nitinol clip-based VCD, but not for suture-based VCD.
VCD failure rates vary depending upon the types of VCD used and are associated with significantly higher vascular complications as compared to deployment successes.
Angio-Seal; complications; Perclose; StarClose; vascular closure device
Safety net hospitals remain under financial strain, possibly affecting quality of care, and face uncertain financial consequences under the Patient Protection and Affordable Care Act. We compared risk-standardized mortality and readmission rates among fee-for-service Medicare beneficiaries admitted for acute myocardial infarction, heart failure, or pneumonia to urban hospitals within metropolitan statistical areas containing at least one safety net and non-safety net hospital. There was substantial variation in both mortality and readmission rates among safety-net and non-safety net hospitals for all three conditions, although safety-net hospitals had marginally worse outcomes. Herein we discuss the clinical and policy implications of these findings.
Safety net hospitals; Vulnerable Populations; Quality of Care
Effectiveness trials have confirmed the superiority of clozapine in the treatment of schizophrenia, but little is known about whether the drug’s superiority holds across racial-ethnic groups. This study examined the effect of race-ethnicity on the effectiveness of clozapine relative to other antipsychotics among patients in maintenance antipsychotic treatment.
Black, Latino, and white Florida Medicaid beneficiaries with schizophrenia receiving maintenance treatment with clozapine or other antipsychotic medications during 7/1/00-6/30/05 were identified. Cox proportional hazard regression models were used to estimate associations of clozapine, race-ethnicity, and their interaction, with time to discontinuation for any cause, our primary measure of effectiveness.
The study cohort included 20,122 episodes of treatment with clozapine (3.7%) and other antipsychotics (96.3%), with 23% black and 36% Latino. Unadjusted analyses suggested that Latinos continue on clozapine longer than whites, while they and blacks discontinue other antipsychotics sooner than whites. Adjusted analyses using 749 propensity score matched sets of clozapine and other antipsychotic users indicated that risk of discontinuation was lower for clozapine users (RR = .45, 95% CI = .39 – .52), an effect that was not moderated by race-ethnicity. Times to discontinuation were longer for clozapine users. Overall risk of antipsychotic discontinuation was higher for blacks (RR =1.56, CI = 1.27 – 1.91), and Latinos (RR = 1.23, CI = 1.02 – 1.48).
This study confirmed clozapine’s superior effectiveness and did not find evidence that race-ethnicity modifies this effect. These findings heighten the need for efforts to increase clozapine use, particularly among minority groups.
Whether recent declines in ischemic heart disease and its risk factors have been accompanied by declines in heart failure (HF) hospitalization and mortality is not known.
To examine changes in HF hospitalization rate and 1-year mortality rate in the U.S., nationally and by state/territory.
Design, Setting, and Participants
55,097,390 fee-for-service Medicare beneficiaries hospitalized between 1998 and 2008 in acute care hospitals in the U.S. and Puerto Rico admitted with a principal discharge diagnosis code for HF.
Main Outcome Measures
Changes in patient demographics and comorbidities, HF hospitalization rates, and 1-year mortality rates.
The HF hospitalization rate adjusted for age, sex, and race declined from 2,845 per 100,000 person-years in 1998 to 2,007 per 100,000 person-years in 2008 (p<0.001), a relative decline of 29.5%. Age-adjusted HF-hospitalization rates declined over the study period for all race-sex categories. Black men had the lowest rate of decline (4,142 to 3,201 per 100,000 person-years) among all race-sex categories which persisted after adjusting for age (incidence rate ratio=0.81, 95% confidence interval [CI] 0.79 to 0.84). HF hospitalization rates declined significantly faster than the national mean in 16 states, and significantly slower in 3 states. Risk-adjusted 1-year mortality fell from 31.7% in 1999 to 29.6% in 2008 (p<0.001), a relative decline of 6.6%. 1-year mortality rates declined significantly in 4 states, but increased in 5 states.
The overall HF hospitalization rate declined substantially from 1998 to 2008, but at a lower rate for black men. The overall 1-year mortality rate declined slightly over the past decade, but remains high. Changes in HF hospitalization and 1-year mortality rates were uneven across states.
heart failure; hospitalization; mortality; epidemiology
We examine the use of fixed-effects and random-effects moment-based meta-analytic methods for analysis of binary adverse event data. Special attention is paid to the case of rare adverse events which are commonly encountered in routine practice. We study estimation of model parameters and between-study heterogeneity. In addition, we examine traditional approaches to hypothesis testing of the average treatment effect and detection of the heterogeneity of treatment effect across studies. We derive three new methods, simple (unweighted) average treatment effect estimator, a new heterogeneity estimator, and a parametric bootstrapping test for heterogeneity. We then study the statistical properties of both the traditional and new methods via simulation. We find that in general, moment-based estimators of combined treatment effects and heterogeneity are biased and the degree of bias is proportional to the rarity of the event under study. The new methods eliminate much, but not all of this bias. The various estimators and hypothesis testing methods are then compared and contrasted using an example dataset on treatment of stable coronary artery disease.
Case management–based interventions aimed at improving quality of care have the potential to narrow racial and ethnic disparities among people with chronic illnesses. The aim of this study was to assess the equity effects of assertive community treatment (ACT), an evidence-based case management intervention, among homeless adults with severe mental illness.
This study used baseline, three-, and 12-month data for 6,829 black, Latino, and white adults who received ACT services through the ACCESS study (Access to Community Care and Effective Services and Support). Zero-inflated Poisson random regression models were used to estimate the adjusted probability of use of outpatient psychiatric services and, among service users, the intensity of use. Odds ratios and rate ratios (RRs) were computed to assess disparities at baseline and over time.
No disparities were found in probability of use at baseline or over time. Compared with white users, baseline intensity of use was lower for black users (RR=.89; 95% confidence interval [CI]=.83–.96) and Latino users (RR=.65; CI=.52–.81]). Intensity did not change over time for whites, but it did for black and Latino users. Intensity increased for blacks between baseline and three months (RR=1.11, CI=1.06–1.17]) and baseline and 12 months (RR=1.17, CI=1.11–1.22]). Intensity of use dropped for Latinos between baseline and three months (RR=.83, CI=.70–.98).
Receipt of ACT was associated with a reduction in service use disparities for blacks but not for Latinos. Findings suggest that ACT’s equity effects differ depending on race-ethnicity.
Biomedical research often involves the measurement of multiple outcomes in different scales (continuous, binary and ordinal). A common approach for the analysis of such data is to ignore the potential correlation among the outcomes and model each outcome separately. This can lead not only to loss of efficiency but also to biased estimates in the presence of missing data. We address the problem of missing data in the context of multiple non-commensurate outcomes. The consequences of missing data when using likelihood and quasi-likelihood methods are described, and an extension of these methods to the situation of missing observations in the outcomes is proposed. Two real data examples illustrate the methodology.
mixed outcomes; multivariate; latent variable; non-commensurate; missing data; maximum likelihood; direct maximization; weighted generalized estimating equations
The use of coronary artery bypass grafting (CABG) surgery in China is growing, but little is known about hospital-level performance. We sought to characterize the variation in performance across hospitals participating in a national registry in China.
Methods and results
The study sample was drawn from the Chinese Cardiac Surgery Registry (CCSR), a national multi-center database that includes 43 hospitals across 13 provinces and 4 direct-controlled municipalities in China. We assessed consecutive patients undergoing isolated CABG surgery during the period of January 1, 2007 through December 31, 2008. Hierarchical generalized linear models were used to estimate hospital-level risk-standardized in-hospital all-cause mortality rates (RSMR) and major complication rates (RSMCR), which included death, myocardial infarction, reoperation for bleeding, mediastinal infection, stroke, re-intubation, and renal failure. Among 8739 patients who underwent isolated CABG surgery, the mean age was 62.2 years (Standard Deviation [SD]=9.2) and 78% were male. Observed in-hospital mortality and complication rates were 2.2% (95% Confidence Interval [CI], 1.9%–2.5%) and 6.6% (95% CI, 6.1%–7.1%) respectively. The mean RSMR was 1.9% (SD=1.1) with a range of 0.7% to 5.8%, and the mean RSMCR was 6.4% (SD=1.5) with a range of 3.8% to 10.1%. The odds of dying and the odds of having a complication after CABG surgery at a hospital one SD below the average relative to a hospital one SD above the average were 2.06 (95% CI, 1.40–3.04) and 1.53 (95% CI, 1.31–1.79) respectively. The Eastern region had the lowest RSMR and RSMCR (1.6% and 5.8%, respectively), whereas the Central region had the highest RSMR (2.5%) and the Southern region had the highest RSMCR (7.7%).
Mortality and complication rates after CABG surgery in the Chinese Cardiac Surgery Registry are generally low but vary by hospital and region within China. These results suggest that there are opportunities to improve outcomes in some CABG facilities.
CABG; outcomes research
Improvements in prevention have led to declines in incidence and mortality of MI in selected populations. However, no studies have examined regional differences in recent trends in MI incidence, and few have examined whether known regional disparities in MI care have narrowed over time.
Methods and Results
We compared trends in incidence rates of MI, associated procedures and mortality for all U.S. Census Divisions (regions) in Medicare fee-for-service patients between 2000 and 2008 (292,773,151 patient-years). Two-stage hierarchical models were used to account for patient characteristics and state-level random effects. To assess trends in geographical disparities, we calculated changes in between-state variance for outcomes over time. While the incidence of MI declined in all regions (P < 0.001 for trend for each) between 2000 and 2008, adjusted rates of decline varied by region (annual declines ranging from 2.9% to 6.1%). Widening geographical disparities, as measured by percent change of between-state variance from 2000 to 2008, were observed for MI incidence (37.6% increase, P = 0.03) and PCI rates (31.4% increase, P = 0.06). Significant declines in risk-adjusted 30-day mortality were observed in all regions, with the fastest declines observed in states with higher baseline mortality rates.
In a large contemporary analysis of geographic trends in MI epidemiology, the incidence of MI and associated mortality declined significantly in all U.S. Census Divisions between 2000 and 2008. While geographical disparities in MI incidence may have increased, regional differences in MI-associated mortality have narrowed.
myocardial infarction; Medicare; trends; disparities
Evidence suggests that minority populations have lower levels of attendance and retention in mental health care than non-Latino whites. Patient activation and empowerment interventions may be effective in increasing minority patients’ attendance and retention.
This study developed and evaluated a patient self-reported activation and empowerment strategy in mental health care.
The Right Question Project–Mental Health (RQP-MH) trainings consisted of 3 individual sessions using a pre/post test comparison group design with patients from 2 community mental health clinics. The RQP-MH intervention taught participants to identify questions that would help them consider their role, process and reasons behind a decision; and empowerment strategies to better manage their care.
A total of 231 participated, completing at least the pretest interview (n = 141 intervention site, 90 comparison site).
Four main outcomes were linked to the intervention: changes in self-reported patient activation; changes in self-reported patient empowerment; treatment attendance; and retention in treatment.
Findings show that intervention participants were over twice as likely to be retained in treatment and over 3 times more likely than comparison participants to have scheduled at least 1 visit during the 6-month follow-up period. Similarly, intervention participants demonstrated 29% more attendance to scheduled visits than comparison patients. There was no evidence of an effect on self-reported patient empowerment, only on self-reported patient activation.
Results demonstrate the intervention’s potential to increase self-reported patient activation, retention, and attendance in mental health care for minority populations. By facilitating patient-provider communication, the RQP-MH intervention may help minorities effectively participate in mental health care.
retention; attendance; mental health; patient activation; ethnic minorities
In-hospital mortality measures, which are widely used to assess hospital quality, are not based on a standardized follow-up period and may systematically favor hospitals with shorter lengths of stay (LOS).
To assess the agreement between performance measures of U.S. hospitals using risk-standardized in-hospital and 30-day mortality rates.
U.S. acute care non-federal hospitals with at least 30 admissions for acute myocardial infarction (AMI), heart failure (HF) and pneumonia in 2004–2006.
Medicare fee-for-service patients admitted for AMI, HF, and pneumonia from 2004–2006.
The primary outcomes are in-hospital and 30-day risk-standardized mortality rates.
There were 718,508 AMI admissions to 3,135 hospitals, 1,315,845 HF admissions to 4,209 hospitals, and 1,415,237 pneumonia admissions to 4,498 hospitals. The hospital-level mean patient LOS in days varied across hospitals for each condition, ranging (min-max) for AMI, HF and pneumonia from 2–13, 3–11, and 3–14 days, respectively. The mean risk-standardized mortality rate differences (30-day minus inhospital) were 5.3% (SD=1.3) for AMI, 6.0% (SD=1.3) for HF, and 5.7% (SD=1.4) for pneumonia, with wide distributions across hospitals. Hospital performance classifications differed between in-hospital and 30-day models for 257 hospitals (8.2%) for AMI, 456 (10.8%) for HF, and 662 (14.7%) for pneumonia. Hospital mean LOS was positively correlated with in-hospital RSMR for all three conditions.
Our study uses Medicare claims data for risk adjustment.
In-hospital mortality measures provide a different assessment of hospital performance than 30-day mortality and are biased in favor of hospitals with shorter LOS.
Substantial hospital-level variation in the risk of readmission after hospitalization for heart failure (HF) or acute myocardial infarction (AMI) has been reported. Prior studies have documented considerable state-level variation in rates of discharge to skilled nursing facilities (SNFs) but evaluation of hospital-level variation in SNF rates and its relationship to hospital-level readmission rates is limited.
Hospital-level 30-day all-cause risk-standardized readmission rates (RSRRs) were calculated using claims data for fee-for-service Medicare patients hospitalized with a principal diagnosis of HF or AMI from 2006-2008. Medicare claims were used to calculate rates of discharge to SNF following HF-specific or AMI-specific admissions in hospitals with ≥25 HF or AMI patients, respectively. Weighted regression was used to quantify the relationship between RSRRs and SNF rates for each condition.
Mean RSRR following HF admission among 4,101 hospitals was 24.7%, and mean RSRR after AMI admission among 2,453 hospitals was 19.9%. Hospital-level SNF rates ranged from 0% to 83.8% for HF and from 0% to 77.8% for AMI. No significant relationship between RSRR after HF and SNF rate was found in adjusted regression models (p=0.15). RSRR after AMI increased by 0.03 percentage point for each 1 absolute percentage point increase in SNF rate in adjusted regression models (p=0.001). Overall, HF and AMI SNF rates explained <1% and 4% of the variation for their respective RSRRs.
SNF rates after HF or AMI hospitalization vary considerably across hospitals, but explain little of the variation in 30-day all-cause readmission rates for these conditions.
heart failure; acute myocardial infarction; skilled nursing facilities; readmission; hospitals; health services research
To investigate the feasibility, reliability, and validity of comprehensively assessing physician-level performance in ambulatory practice.
Data Sources/Study Setting
Ambulatory-based general internists in 13 states participated in the assessment.
We assessed physician-level performance, adjusted for patient factors, on 46 individual measures, an overall composite measure, and composite measures for chronic, acute, and preventive care. Between- versus within-physician variation was quantified by intraclass correlation coefficients (ICC). External validity was assessed by correlating performance on a certification exam.
Data Collection/Extraction Methods
Medical records for 236 physicians were audited for seven chronic and four acute care conditions, and six age- and gender-appropriate preventive services.
Performance on the individual and composite measures varied substantially within (range 5–86 percent compliance on 46 measures) and between physicians (ICC range 0.12–0.88). Reliabilities for the composite measures were robust: 0.88 for chronic care and 0.87 for preventive services. Higher certification exam scores were associated with better performance on the overall (r = 0.19; p <.01), chronic care (r = 0.14, p = .04), and preventive services composites (r = 0.17, p = .01).
Our results suggest that reliable and valid comprehensive assessment of the quality of chronic and preventive care can be achieved by creating composite measures and by sampling feasible numbers of patients for each condition.
Comprehensive assessment; quality of care; primary care; composite measures
In disparities models, researchers adjust for differences in “clinical need,” including indicators of comorbidities. We reconsider this practice, assessing 1) if and how having a comorbidity changes the likelihood of recognition and treatment of mental illness; and 2) differences in mental health (MH) care disparities estimates with and without adjustment for comorbidities.
Longitudinal data from the 2000-2007 Medical Expenditure Panel Survey (n=11,083) split into pre- and post-periods for White, Latino, and Black adults with probable need for MH care.
First, we tested a crowd-out effect (comorbidities decrease initiation of MH care after a PCP visit) using logistic regression models and an exposure effect (comorbidities cause more PCP visits, increasing initiation of MH care) using instrumental variable methods. Second, we assessed the impact of adjustment for comorbidities on disparity estimates.
We found no evidence of a crowd-out effect but strong evidence for an exposure effect. Number of post-period visits positively predicted initiation of MH care. Adjusting for racial/ethnic differences in comorbidities increased Black-White disparities and decreased Latino-White disparities.
Positive exposure findings suggest that intensive follow-up programs shown to reduce disparities in chronic care management may have additional indirect effects on reducing MH care disparities.
In clinical trials multiple outcomes are often used to assess treatment interventions. This paper presents an evaluation of likelihood-based methods for jointly testing treatment effects in clinical trials with multiple continuous outcomes. Specifically, we compare the power of joint tests of treatment effects obtained from joint models for the multiple outcomes with univariate tests based on modelling the outcomes separately. We also consider the power and bias of tests when data are missing, a common feature of many trials, especially in psychiatry. Our results suggest that joint tests capitalize on the correlation of multiple outcomes and are more powerful than standard univariate methods, especially when outcomes are missing completely at random. When outcomes are missing at random, test procedures based on correctly specified joint models are unbiased, while standard univariate procedures are not. Results of a simulation study are reported, and the methods are illustrated in an example from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) for schizophrenia.
joint tests; multiple outcomes; power; missing data; psychiatry
National attention has increasingly focused on readmission as a target for quality improvement. We present the development and validation of a model approved by the National Quality Forum and used by the Centers for Medicare & Medicaid Services for hospital-level public reporting of risk-standardized readmission rates for patients discharged from the hospital after an acute myocardial infarction.
Methods and Results
We developed a hierarchical logistic regression model to calculate hospital risk-standardized 30-day all-cause readmission rates for patients hospitalized with acute myocardial infarction. The model was derived using Medicare claims data for a 2006 cohort and validated using claims and medical record data. The unadjusted readmission rate was 18.9%. The final model included 31 variables and had discrimination ranging from 8% observed 30-day readmission rate in the lowest predictive decile to 32% in the highest decile and a C statistic of 0.63. The 25th and 75th percentiles of the risk-standardized readmission rates across 3890 hospitals were 18.6% and 19.1%, with fifth and 95th percentiles of 18.0% and 19.9%, respectively. The odds of all-cause readmission for a hospital 1 SD above average were 1.35 times that of a hospital 1 SD below average. Hospital-level adjusted readmission rates developed using the claims model were similar to rates produced for the same cohort using a medical record model (correlation, 0.98; median difference, 0.02 percentage points).
This claims-based model of hospital risk-standardized readmission rates for patients with acute myocardial infarction produces estimates that are excellent surrogates for those produced from a medical record model.
myocardial infarction; health policy; quality of health care
This study investigated the impact of adding novel elements to models predicting in-hospital mortality following percutaneous coronary interventions (PCIs).
Massachusetts (MA) mandated public reporting of hospital-specific PCI mortality in 2003. In 2006, a physician advisory group recommended adding to the prediction models three attributes not collected by the National Cardiovascular Data Registry instrument. These “compassionate use” (CU) features included coma on presentation, active hemodynamic support during PCI, and cardiopulmonary resuscitation at PCI initiation.
From October 2005 through September 2007, PCI was performed during 29,784 admissions in MA non-federal hospitals. Of these, 5,588 involved patients with ST segment elevation myocardial infarction or cardiogenic shock. Cases with CU criteria identified were adjudicated by trained physician reviewers. Regression models with and without the CU composite variable (presence of any of the 3 features) were compared using areas under the receiver operator characteristic curves (AUC).
Unadjusted mortality in this high-risk subset was 5.7%. Among these admissions, 96 (1.7%) had at least one CU feature, with 69.8% mortality. The adjusted odds ratio for in-hospital death for CU PCIs (vs. no CU criteria) was 27.3 (95% CI 14.5–47.6). Discrimination of the model improved after including CU, with AUC increasing from 0.87 to 0.90 (p<0.01), while goodness of fit was preserved.
A small proportion of patients at extreme risk for post-PCI mortality can be identified using pre-procedural factors not routinely collected, but that heighten predictive accuracy. Such improvements in model performance may result in greater confidence in reporting of risk-adjusted PCI outcomes.
Percutaneous coronary intervention (PCI); predictive models; Hierarchical risk prediction models; ACC-NCDR CathPCI