A new casemix adjustment index for hospital mortality among patients with congestive heart failure

Causes and Predictors of In-Hospital Mortality in Patients Admitted with or for Heart Failure at a Tertiary Hospital in Brazil

Background

Although heart failure (HF) has high morbidity and mortality, studies in Latin America on causes and predictors of in-hospital mortality are scarce. We also do not know the evolution of patients with compensated HF hospitalized for other reasons.

Objective

To identify causes and predictors of in-hospital mortality in patients hospitalized for acute decompensated HF (ADHF), compared to those with HF and admitted to the hospital for non-HF related causes (NDHF).

Methods

Historical cohort of patients hospitalized in a public tertiary hospital in Brazil with a diagnosis of HF identified by the Charlson Comorbidity Index (CCI).

Results

A total of 2056 patients hospitalized between January 2009 and December 2010 (51% men, median age of 71 years, length of stay of 15 days) were evaluated. There were 17.6% of deaths during hospitalization, of which 58.4% were non-cardiovascular (63.6% NDHF vs 47.4% ADHF, p = 0.004). Infectious causes were responsible for most of the deaths and only 21.6% of the deaths were attributed to HF. The independent predictors of in-hospital mortality were similar between the groups and included: age, length of stay, elevated potassium, clinical comorbidities, and CCI. Renal insufficiency was the most relevant predictor in both groups.

Conclusion

Patients hospitalized with HF have high in-hospital mortality, regardless of the primary reason for hospitalization. Few deaths are directly attributed to HF; Age, renal function and levels of serum potassium, length of stay, comorbid burden and CCI were independent predictors of in-hospital death in a Brazilian tertiary hospital.

Co-morbidities in elderly patients with hip fracture: recommendations of the ISFR-IOF hip fracture outcomes working group

INTRODUCTION

Hip fractures are the second leading cause of hospitalization in the aged and by 2041, epidemiologists forecast an increase in economic cost to $2.4 billion. The hip patient population often presents with comorbidities causing these patients to receive less aggressive medical treatment and have a low quality of life. We believe that physical function is a patient-important outcome for many medical and surgical interventions. The functional co-morbidity index (FCI), unlike prior co-morbidity indices, was developed with physical function as an outcome instead of being designed for administrative purposes or to predict mortality. Our objective was to evaluate the perceptions of practitioners in hip fracture care about the impact of comorbidities on physical function as primary outcome.

METHODS

We piloted and then distributed a self-administered survey to members of the International Society for Fracture Repair hip fracture outcomes working group. For each of the 18 diagnoses included in the FCI index, we asked in our survey whether the presence of the co-morbidity and whether the severity of the co-morbidity was perceived to impact physical function in patients following a hip fracture.

RESULTS

Seventeen out of 20 respondents completed the questionnaire. The presence and severity of arthritis was 'strongly' believed to predict physical function in those with hip fracture (69 and 85.7 %, respectively). Respondents 'agreed' (range 53-73 %) that 10/18 diagnoses would predict changes in physical function following hip fracture treatment. Whereas, 63 % of the practitioners'strongly disagreed' that diabetes types I and II would change physical function scores. Furthermore, dementia was listed as an additional diagnosis that would affect physical function.

CONCLUSION

The FCI may provide a useful instrument to predict functional outcome after hip fracture; however, the index may need to be modified for this specific population.

Systematic review of comorbidity indices for administrative data

BACKGROUND

Adjustment for comorbidities is common in performance benchmarking and risk prediction. Despite the exponential upsurge in the number of articles citing or comparing Charlson, Elixhauser, and their variants, no systematic review has been conducted on studies comparing comorbidity measures in use with administrative data. We present a systematic review of these multiple comparison studies and introduce a new meta-analytical approach to identify the best performing comorbidity measures/indices for short-term (inpatient + ≤ 30 d) and long-term (outpatient+>30 d) mortality.

METHODS

Articles up to March 18, 2011 were searched based on our predefined terms. The bibliography of the chosen articles and the relevant reviews were also searched and reviewed. Multiple comparisons between comorbidity measures/indices were split into all possible pairs. We used the hypergeometric test and confidence intervals for proportions to identify the comparators with significantly superior/inferior performance for short-term and long-term mortality. In addition, useful information such as the influence of lookback periods was extracted and reported.

RESULTS

Out of 1312 retrieved articles, 54 articles were eligible. The Deyo variant of Charlson was the most commonly referred comparator followed by the Elixhauser measure. Compared with baseline variables such as age and sex, comorbidity adjustment methods seem to better predict long-term than short-term mortality and Elixhauser seems to be the best predictor for this outcome. For short-term mortality, however, recalibration giving empirical weights seems more important than the choice of comorbidity measure.

CONCLUSIONS

The performance of a given comorbidity measure depends on the patient group and outcome. In general, the Elixhauser index seems the best so far, particularly for mortality beyond 30 days, although several newer, more inclusive measures are promising.

Risk stratification in acute heart failure: rationale and design of the STRATIFY and DECIDE studies

BACKGROUND

A critical challenge for physicians facing patients presenting with signs and symptoms of acute heart failure (AHF) is how and where to best manage them. Currently, most patients evaluated for AHF are admitted to the hospital, yet not all warrant inpatient care. Up to 50% of admissions could be potentially avoided and many admitted patients could be discharged after a short period of observation and treatment. Methods for identifying patients that can be sent home early are lacking. Improving the physician's ability to identify and safely manage low-risk patients is essential to avoiding unnecessary use of hospital beds.

METHODS

Two studies (STRATIFY and DECIDE) have been funded by the National Heart Lung and Blood Institute with the goal of developing prediction rules to facilitate early decision making in AHF. Using prospectively gathered evaluation and treatment data from the acute setting (STRATIFY) and early inpatient stay (DECIDE), rules will be generated to predict risk for death and serious complications. Subsequent studies will be designed to test the external validity, utility, generalizability and cost-effectiveness of these prediction rules in different acute care environments representing racially and socioeconomically diverse patient populations.

RESULTS

A major innovation is prediction of 5-day as well as 30-day outcomes, overcoming the limitation that 30-day outcomes are highly dependent on unpredictable, post-visit patient and provider behavior. A novel aspect of the proposed project is the use of a comprehensive cardiology review to correctly assign post-treatment outcomes to the acute presentation.

CONCLUSIONS

Finally, a rigorous analysis plan has been developed to construct the prediction rules that will maximally extract both the statistical and clinical properties of every data element. Upon completion of this study we will subsequently externally test the prediction rules in a heterogeneous patient cohort.

Biomarker changes during acute heart failure treatment

Biomarker changes may provide physicians with objective evidence of treatment efficacy in patients with acute decompensated heart failure (ADHF) and facilitate early hospital discharge. The authors hypothesize that mid-regional-pro-adrenomedullin (MR-proADM), C-terminal-pro-endothelin-1 (CT-pro-ET-1), and mid-regional-pro-atrial natriuretic peptide (MR-proANP) change during the first 24 hours of ADHF therapy. Eligible patients had an emergency department diagnosis of ADHF and fulfilled modified Framingham criteria. Clinical data, serum, and plasma values were collected at enrollment, 2 to 4 hours, and 12 to 24 hours after treatment. Changes in biomarker concentrations from baseline to 2 to 4 hours, baseline to 12 to 24 hours, and 2 to 4 to 12 to 24 hours were calculated. Fisher exact and Kruskal-Wallis tests were used for comparisons. Forty-eight patients were included. The median age was 62 years (range 40-88), 54% were men and 50% were white. More patients had changes in MR-pro-ANP levels in the first 2 to 4 hours after ADHF therapy compared with MR-proADM or CT-pro-ET-1 (36% vs 16% and 24%). However, 12 to 24 hours after therapy, similar proportions of patients had changes in MR-proANP, MR-proADM, and CT-proET-1 levels (47%, 41%, and 49%). In this preliminary study, patients with ADHF had measurable changes in MR-proANP, MR-proADM, and CT-pro-ET-1 24 hours after initial therapy. A study of association with clinical course and outcomes to determine the role of these markers in risk-stratification is warranted.

Psychiatric comorbidity and mortality among veterans hospitalized for congestive heart failure

A Behavioral Model of Health Services Utilization approach was used to examine the impact of comorbid mental illness on mortality of veterans admitted to Veterans Affairs medical centers in fiscal year 2001 with a primary diagnosis of congestive heart failure (n = 15,497). Thirty percent had a psychiatric diagnosis, 4.7% died during the index hospitalization, and 11.5% died during the year following discharge. Among those with mental illness, 23.6% had multiple psychiatric disorders. Multivariable logistic regression models found dementia to be positively associated with inpatient mortality. Depression alone (excluding other psychiatric disorders) was positively associated with one-year mortality. Primary care visits were associated with a reduced likelihood of both inpatient and one-year mortality. Excepting dementia, VA patients with a mental illness had comparable or higher levels of primary care visits than those having no mental illness. Patients with multiple psychiatric disorders had more outpatient care than those with one psychiatric disorder.

Maximising the value of hospital administrative datasets

Mandatory and standardised administrative data collections are prevalent in the largely public-funded acute sector. In these systems the data collections are used for financial, performance monitoring and reporting purposes. This paper comments on the infrastructure and standards that have been established to support data collection activities, audit and feedback. The routine, local and research uses of these datasets are described using examples from Australian and international literature. The advantages of hospital administrative datasets and opportunities for improvement are discussed under the following headings: accessibility, standardisation, coverage, completeness, cost of obtaining clinical data, recorded Diagnostic Related Groups and International Classification of Diseases codes, linkage and connectivity. In an era of diminishing resources better utilisation of these datasets should be encouraged. Increased study and scrutiny will enhance transparency and help identify issues in the collections. As electronic information systems are increasingly embraced, administrative data collections need to be managed as valuable assets and powerful operational and patient management tools.

Acoustic cardiography S3 detection use in problematic subgroups and B-type natriuretic peptide "gray zone": secondary results from the Heart failure and Audicor technology for Rapid Diagnosis and Initial Treatment Multinational Investigation

BACKGROUND

Dyspneic emergency department (ED) patients present a diagnostic dilemma. The S3, although highly specific for acute heart failure (AHF) and predicting death and readmission, is often difficult to auscultate. The HEart failure and Audicor technology for Rapid Diagnosis and Initial Treatment (HEARD-IT) multinational trial evaluated the S3 via acoustic cardiography (Audicor). Our goal in this secondary analysis was to determine if the strength of the S3 can provide diagnostic/prognostic information in problematic heart failure subgroups.

METHODS

Dyspneic ED patients older than 40 years and not on dialysis were prospectively enrolled. A gold standard AHF diagnosis was determined by 2 cardiologists blinded to acoustic cardiography results. The S3 strength parameter was delineated on a scale of 0 to 10. This secondary analysis of subgroups from the HEARD-IT database used univariate/multivariate regression to determine the diagnostic/prognostic ability of the S3 strength.

RESULTS

In the 995 patients enrolled, S3 strength was a significant prognosticator in univariate analysis for adverse events but not in a multivariable model. In patients with "gray zone" B-type natriuretic peptide (BNP) levels (100-499 pg/mL), acoustic cardiography increased diagnostic accuracy of AHF from 47% to 69%. Acoustic cardiography improved S3 detection sensitivity in obese patients when compared to auscultation.

CONCLUSION

The strength of the S3 gallop provides rapid results that assist with identification of AHF in selected populations. S3 detection complements the use of BNP in the gray zone, and its diagnostic/prognostic ability is largely unaffected by body mass index and renal function. S3 strength shows promise as a diagnostic and prognostic tool in problematic HF subgroups.

The Effect of Age upon Care and Outcomes in Patients Hospitalized for Congestive Heart Failure in Alberta, Canada

We describe the age-specific outcomes for patients hospitalized with newly diagnosed congestive heart failure using administrative hospital abstracts from Alberta, Canada, from April 1, 1994, to March 31, 2000. Seniors (aged 65 years and older) constituted about 85 per cent of the 16,162 patients. Both co-morbidity and severity of illness tended to increase with age. The use of special care unit admissions, coronary artery diagnostic services (cardiac catheterization), and revascularization procedures (percutanenous transluminal coronary angioplasty/stenting, coronary artery bypass surgery) peaked in the 50-to 64-year age group and decreased with increasing age. Specialist/sub-specialist care, prescriptions of beta blockers and angiotensin-converting enzyme inhibitors / angiotensin receptor blockers decreased with age in seniors. Adjusted in-hospital, 1-year mortality and crude, age-specific 5-year mortality were significantly greater in those 75 years and older. Outcomes and process of care in patients with newly diagnosed congestive heart failure were not uniformly distributed with age. The elderly had greater mortality but received less therapy.

Emergency physicians' risk attitudes in acute decompensated heart failure patients

OBJECTIVES

Despite the existence of various clinical prediction rules, no data exist defining what frequency of death or serious nonfatal outcomes comprises a realistic "low-risk" group for clinicians. This exploratory study sought to identify emergency physicians' (EPs) definition of low-risk acute decompensated heart failure (ADHF) emergency department (ED) patients.

METHODS

Surveys were mailed to full-time physicians (n = 88) in a multihospital EP group in southwestern Pennsylvania between December 2004 and February 2005. Participation was voluntary, and each EP was asked to define low risk (low risk of all-cause 30-day death and low risk of either hospital death or other serious medical complications) and choose a risk threshold at which they might consider outpatient management for those with ADHF. A range of choices was offered (<0.5, <1, <2, <3, <4, and <5%), and demographic data were collected.

RESULTS

The response rate was 80%. Physicians defined low risk both for all-cause 30-day death and for hospital death or other serious complications, at <1% (38.8 and 40.3%, respectively). The decision threshold to consider outpatient therapy was <0.5% risk both for all-cause 30-day death (44.6%) and for hospital death or serious medical complications (44.4%).

CONCLUSIONS

Emergency physicians in this exploratory study define low-risk ADHF patients as having less than a 1% risk of 30-day death or inpatient death or complications. They state a desire to have and use an ADHF clinical prediction rule that can identify low-risk ADHF patients who have less than a 0.5% risk of 30-day death or inpatient death or complications.

Covariate balance in a Bayesian propensity score analysis of beta blocker therapy in heart failure patients

Regression adjustment for the propensity score is a statistical method that reduces confounding from measured variables in observational data. A Bayesian propensity score analysis extends this idea by using simultaneous estimation of the propensity scores and the treatment effect. In this article, we conduct an empirical investigation of the performance of Bayesian propensity scores in the context of an observational study of the effectiveness of beta-blocker therapy in heart failure patients. We study the balancing properties of the estimated propensity scores. Traditional Frequentist propensity scores focus attention on balancing covariates that are strongly associated with treatment. In contrast, we demonstrate that Bayesian propensity scores can be used to balance the association between covariates and the outcome. This balancing property has the effect of reducing confounding bias because it reduces the degree to which covariates are outcome risk factors.

The rationale for an acute heart failure syndromes clinical trials network

BACKGROUND

Clinical trials involving novel therapies treating acute heart failure syndromes (AHFS) have shown limited success with regard to both efficacy and safety. As a direct result, outcomes have changed little over time and AHFS remains a disease process associated with largely no change in hospitalization rates (80%), hospital length of stay (median 4.5 days), and in-hospital (4-7%) and 60-day mortality (10%). Despite extensive emergency department (ED) involvement during the initial phase of AHFS management, clinical trials have enrolled patients after the ED phase of management, up to 48 hours after initial therapy, long after many patients have experienced significant beneficial effects of standard therapy. As standard therapy has provided symptomatic improvement in up to 70% of patients in these trials, it is not surprising that investigational agents started after 24 to 48 hours of standard therapy have shown limited clinical efficacy when compared with standard therapy.

METHODS AND RESULTS

The ability to screen, enroll, and randomize in the emergency setting is fundamental. The unique environment, the ethical complexities of enrollment in emergency-based research, and the need for rapid and standardized study-compliant care represent key challenges to active recruitment in AHFS studies. Specifically, the ability to identify and enroll a large cohort of AHFS patients early (<6 hours) in their presentation has been cited as the primary barrier to the appropriate design of clinical trials that includes this early window.

CONCLUSIONS

In response, we have created a network of dedicated academic physicians with experience in clinical trials and acute management of heart failure who together can surmount this barrier and provide a framework for conducting early trials in AHFS.

Differences in initial severity of illness between black and white emergency department patients hospitalized with heart failure

BACKGROUND

Black patients hospitalized for heart failure have better reported short-term survival than white patients for unknown reasons. We sought to determine if initial severity of illness differed between black and white emergency department (ED) patients hospitalized for heart failure.

METHODS

We analyzed 1,408 black and 7,260 white randomly selected patients in one state hospitalized from an ED during 2003 and 2004 and with a discharge diagnosis of heart failure. We used three validated clinical prediction rules to estimate severity of illness on admission.

RESULTS

Black patients were younger than white patients (65.8 +/- 14.8 vs 77.4 +/- 11.5 years, P < .01) and were assigned to lower risk classes by all 3 prediction rules more frequently than white patients (P < .01). The odds ratio (95% CI) for classification of black versus white patients into the lowest risk class within the three rules ranged from 1.16 (1.00-1.33) to 4.30 (3.75-4.94). After adjusting for hospital clustering, the odds ratio (95% CI) for black versus white patient hospital death and complications was 0.75 (0.60-0.95) and, for 30-day death, was 0.34 (0.27-0.48).

CONCLUSIONS

Black ED patients hospitalized with heart failure are younger, less severely ill on admission and less likely to experience short-term fatal and nonfatal outcomes than white patients. Our findings suggest a varying opportunity between black and white patients when considering alternative initial treatment strategies and sites of care.

Use of instrumental variables in the analysis of generalized linear models in the presence of unmeasured confounding with applications to epidemiological research

A major, often unstated, concern of researchers carrying out epidemiological studies of medical therapy is the potential impact on validity if estimates of treatment are biased due to unmeasured confounders. One technique for obtaining consistent estimates of treatment effects in the presence of unmeasured confounders is instrumental variables analysis (IVA). This technique has been well developed in the econometrics literature and is being increasingly used in epidemiological studies. However, the approach to IVA that is most commonly used in such studies is based on linear models, while many epidemiological applications make use of non-linear models, specifically generalized linear models (GLMs) such as logistic or Poisson regression. Here we present a simple method for applying IVA within the class of GLMs using the generalized method of moments approach. We explore some of the theoretical properties of the method and illustrate its use within both a simulation example and an epidemiological study where unmeasured confounding is suspected to be present. We estimate the effects of beta-blocker therapy on one-year all-cause mortality after an incident hospitalization for heart failure, in the absence of data describing disease severity, which is believed to be a confounder.

Avaliação do índice de comorbidade de Charlson em internações da região de Ribeirão Preto, São Paulo, Brasil

O objetivo deste artigo foi avaliar o uso do índice de comorbidade de Charlson (ICC) para predizer óbito hospitalar em internações da região de Ribeirão Preto, São Paulo, Brasil. Foram analisadas 54.680 hospitalizações entre janeiro de 1996 e dezembro de 1997. Duas adaptações do ICC para a Classificação Internacional de Doenças (CID) foram comparadas e as trinta condições clínicas avaliadas por Charlson foram revistas. A regressão logística foi utilizada para avaliar a capacidade dos modelos de predizer o óbito hospitalar. O modelo de base incluiu: idade, sexo e diagnóstico principal. Diferenças na adaptação para a CID-9 pouco impactaram a capacidade de discriminação dos modelos. A revisão das trinta condições clínicas aumentou a capacidade de discriminação do modelo de predição de óbito (estatística C = 0,73) quando comparado ao modelo com o ICC original (estatística C = 0,72). Todos os modelos testados tiveram efeito reduzido sobre a capacidade discriminativa do modelo de base (estatística C = 0,70). Os resultados apontam a importância de se dispor no país de um sistema de informação que permita uma descrição completa da morbidade hospitalar para o monitoramento do desempenho dos serviços.

Validation of the Acute Heart Failure Index

STUDY OBJECTIVE

Validate a clinical prediction rule prognostic of short-term fatal and inpatient nonfatal outcomes for heart failure patients admitted through the emergency department.

METHODS

We retrospectively studied a random cohort of 8,384 adult patients admitted to Pennsylvania hospitals in 2003 and 2004 with a diagnosis of heart failure as defined by primary discharge diagnosis codes. We reported the proportions of inpatient death, serious medical complications before discharge, and 30-day death in the patients identified as low risk by the prediction rule.

RESULTS

The prediction rule classified 1,609 (19.2%) of the patients as low risk. Within this subgroup, there were 12 (0.7%; 95% confidence interval [CI] 0.3% to 1.2%) inpatient deaths, 28 (1.7%; 95% CI 1.1% to 2.4%) patients survived to hospital discharge after a serious complication, and 47 (2.9%; 95% CI 2.1% to 3.7%) patients died within 30 days of the index hospitalization.

CONCLUSION

This prediction rule identifies a group of admitted heart failure patients at low risk of inpatient mortal and nonmortal complications. Our validation findings suggest the rule could assist physicians in making site-of-care decisions for this patient population and aid in analyzing presenting illness burden in study populations.

A sensitivity analysis using information about measured confounders yielded improved uncertainty assessments for unmeasured confounding

OBJECTIVE

In the analysis of observational data, the argument is sometimes made that if adjustment for measured confounders induces little change in the treatment-outcome association, then there is less concern about the extent to which the association is driven by unmeasured confounding. We quantify this reasoning using Bayesian sensitivity analysis (BSA) for unmeasured confounding. Using hierarchical models, the confounding effect of a binary unmeasured variable is modeled as arising from the same distribution as that of measured confounders. Our objective is to investigate the performance of the method compared to sensitivity analysis, which assumes that there is no relationship between measured and unmeasured confounders.

STUDY DESIGN AND SETTING

We apply the method in an observational study of the effectiveness of beta-blocker therapy in heart failure patients.

RESULTS

BSA for unmeasured confounding using hierarchical prior distributions yields an odds ratio (OR) of 0.72, 95% credible interval (CrI): 0.56, 0.93 for the association between beta-blockers and mortality, whereas using independent priors yields OR=0.72, 95% CrI: 0.45, 1.15.

CONCLUSION

If the confounding effect of a binary unmeasured variable is similar to that of measured confounders, then conventional sensitivity analysis may give results that overstate the uncertainty about bias.

Beyond pulmonary edema: diagnostic, risk stratification, and treatment challenges of acute heart failure management in the emergency department

The majority of heart failure hospitalizations in the United States originate in the emergency department (ED). Current strategies for acute heart failure syndromes have largely been tailored after chronic heart failure guidelines and care. Prospective ED-based acute heart failure syndrome trials are lacking, and current guidelines for disposition are based on either little or no evidence. As a result, the majority of ED acute heart failure syndrome patients are admitted to the hospital. Recent registry data suggest there is a significant amount of heterogeneity in acute heart failure syndrome ED presentations, and diagnostics and therapeutics may need to be individualized to the urgency of the presentation, underlying pathophysiology, and acute hemodynamic characteristics. A paradigm shift is necessary in acute heart failure syndrome guidelines and research: prospective trials need to focus on diagnostic, therapeutic, and risk-stratification algorithms that rely on readily available ED data, focusing on outcomes more proximate to the ED visit (5 days). Intermediate outcomes (30 days) are more dependent on inpatient and outpatient care and patient behavior than ED management decisions. Without these changes, the burden of acute heart failure syndrome care is unlikely to change. This article proposes such a paradigm shift in acute heart failure syndrome care and discusses areas of further research that are necessary to promote this change in approach.

A single health status question had important prognostic value among outpatients with chronic heart failure

OBJECTIVE

Health status is an important marker of the impact of disease on function among patients with chronic heart failure (CHF). However, the prognostic value of CHF-specific health status on long-term mortality has not been adequately evaluated. Our objective was to assess CHF-specific health status and 5-year mortality among outpatients with CHF.

STUDY DESIGN AND SETTING

We analyzed data from 494 Veterans Affairs outpatients with diagnoses of CHF and objective evidence of left ventricular dysfunction who enrolled in a quality improvement intervention. We extracted information about comorbid diagnoses, severity of illness (Charlson index), health care utilization, drug therapy, laboratory, and vital sign data along with generic and CHF-specific health status. We then identified multivariate correlates of subsequent mortality at 5 years.

RESULTS

Five-year mortality was 44%. Age (chi2=26.1, hazard ratio [HR]=1.63, confidence interval [CI]: 1.35, 1.97; P<0.0001) and Charlson index (chi2=12.9, HR=1.39, CI: 1.16, 1.67; P=0.0003) were significantly associated with 5-year mortality. Controlling for clinical, lab, medication, and administrative data, a single-item assessing change in CHF-specific health status was independently associated with 5-year mortality (chi2=11.4, HR=0.87, CI: 0.80, 0.94, P=0.0007).

CONCLUSIONS

Given the strength of the association with mortality, health care providers should routinely assess this single-item change in health status among outpatients with CHF to identify higher risk patients and guide therapy.

Bayesian sensitivity analysis for unmeasured confounding in observational studies

We consider Bayesian sensitivity analysis for unmeasured confounding in observational studies where the association between a binary exposure, binary response, measured confounders and a single binary unmeasured confounder can be formulated using logistic regression models. A model for unmeasured confounding is presented along with a family of prior distributions that model beliefs about a possible unknown unmeasured confounder. Simulation from the posterior distribution is accomplished using Markov chain Monte Carlo. Because the model for unmeasured confounding is not identifiable, standard large-sample theory for Bayesian analysis is not applicable. Consequently, the impact of different choices of prior distributions on the coverage probability of credible intervals is unknown. Using simulations, we investigate the coverage probability when averaged with respect to various distributions over the parameter space. The results indicate that credible intervals will have approximately nominal coverage probability, on average, when the prior distribution used for sensitivity analysis approximates the sampling distribution of model parameters in a hypothetical sequence of observational studies. We motivate the method in a study of the effectiveness of beta blocker therapy for treatment of heart failure.

Comparison of four clinical prediction rules for estimating risk in heart failure

STUDY OBJECTIVE

We examine the performance of 4 clinical prediction rules prognostic of short-term fatal and hospital-based nonfatal outcomes in heart failure patients.

METHODS

We used a retrospective cohort of 33,533 adult patients admitted to Pennsylvania hospitals in 1999 with a diagnosis of heart failure. We stratified patients into risk categories defined by each clinical prediction rule. We assessed prognostic accuracy according to sensitivity and specificity and compared discriminatory power according to area under the receiver operating characteristic (ROC) curves. The outcomes were inpatient death, 30-day mortality, and death or serious medical complications before hospital discharge.

RESULTS

The 4 rules each created risk groups of various proportions and frequencies of outcomes. The proportion of patients assigned to the lowest risk group ranged from 13.3% to 73.0%. The rates of inpatient death or complications in the lowest risk group ranged from 6.7% to 9.2%, and 30-day death rates varied from 1.7% to 6.0%. Patients categorized at the highest risk of death or complication demonstrated similar variability. The area under the ROC curve for inpatient death and complications differed only slightly among rules (0.58 to 0.62). The area under the ROC curve for fatal outcomes tended to be higher and differed among rules (0.59 to 0.74)

CONCLUSION

Current acute heart failure prediction rules offer varying ability to predict short-term death or serious outcomes. Although each creates a risk gradient, differences in risk-group proportions and outcome frequencies should drive rule selection or use in clinical practice.

Relation between cardiac troponin I and mortality in acute decompensated heart failure

BACKGROUND

Troponin level elevations are common in patients with acute decompensated heart failure (ADHF), yet their prognostic value above and beyond traditional predictors of outcomes in heart failure is uncertain.

METHODS

In the EFFECT study, we determined the association between cardiac troponin I and all-cause mortality in 2025 patients hospitalized for heart failure in Ontario, Canada, between April 1, 1999, and March 31, 2001.

RESULTS

Cardiac troponin I levels >0.5 microg/L (median 1.7 microg/L, interquartile range 0.9-4.8 microg/L) occurred in 699 (34.5%) patients and was an independent predictor of mortality (adjusted hazard ratio 1.49, 95% CI 1.25-1.77, P < .001). Furthermore, we observed a dose-response relationship between cardiac troponin I and mortality that persisted after adjustment for potential confounding factors (adjusted hazard ratio 1.10 per 1 microg/L increase, 95% CI 1.05-1.15, P < .001). The association between cardiac troponin I and mortality was similar for patients with and without other features of acute ischemia on presentation (P > .05 for interaction).

CONCLUSIONS

In patients hospitalized for ADHF who had cardiac troponin levels measured during the course of clinical practice, cardiac troponin I was an independent predictor of all-cause mortality. Cardiac troponin testing is easily accessible, has predictive value above and beyond traditional clinical predictors of mortality, and may help guide medical decision making in patients with ADHF.

Maternal race, procedures, and infant birth weight in type 2 and gestational diabetes

OBJECTIVE

To examine the relation between race and cesarean delivery, episiotomy, and low birth weight infants in pregnancies with type 2 and gestational diabetes mellitus and to identify factors that might explain racial differences.

METHODS

Population-based, cross-sectional study of 1999-2004 Maryland hospital discharge data. Hospitalizations for delivery of pregnancies with type 2 and gestational diabetes mellitus were identified and matched to infants. The independent variable was maternal race. Dependent variables were cesarean delivery, episiotomy, and low infant birth weight. Stepwise logistic regression models were developed to estimate the independent effect of race on use of each procedure and infant birth weight, after adjusting for sociodemographic, hospital, and clinical factors.

RESULTS

We examined 6,310 deliveries for pregnancies with type 2 (15%) and gestational (85%) diabetes. Before adjustment, black race was associated with a higher odds of cesarean delivery (odds ratio [OR] 1.40, 95% confidence interval [CI] 1.24-1.58) and low birth weight infants (OR 1.94, 95% CI 1.57-2.40) compared with white race. Adjustment for racial differences in preeclampsia and fetal heart rate abnormalities accounted for a modest degree of the racial variation in outcomes. With full adjustment, black race was still associated with a higher odds of cesarean delivery (OR 1.38, 95% CI 1.20-1.60) and low birth weight (OR 1.81, 95% CI 1.41-2.34) and a lower odds of episiotomy (OR 0.45, 95% CI 0.36-0.57).

CONCLUSION

In pregnancies with diabetes, adjustment for sociodemographic, hospital, and clinical factors only partially explains racial differences in procedure use and infant low birth weight.

A Simple Clinically Based Predictive Rule for Heart Failure In-Hospital Mortality

BACKGROUND

Scarce data are available to predict in-hospital mortality for decompensated heart failure (HF) in South American populations.

METHODS AND RESULTS

We evaluated 779 consecutive HF admissions defined by the Boston criteria in a tertiary care hospital. Stepwise logistic regression was used to determine independent correlates of in-hospital mortality, derived from 83 potential predictors collected on hospital admission. A clinical score rule (HF Revised Score) was created using the regression coefficient estimates derived from multivariate modeling. During hospital stay, 77 (10%) deaths occurred and 6 clinical characteristics were independently associated with in-hospital mortality: presence of cancer (P < .001), systolic blood pressure < or =124 mm Hg (P < .001), serum creatinine >1.4 mg/dL (P = .02), blood urea nitrogen >37 mg/dL (P = .03), serum sodium <136 mEq/L (P = .03), and age >70 years old (P = .03). Both the Acute Decompensated Heart Failure National Registry stratification algorithm and the proposed HF Revised Score performed adequately to predict in-hospital mortality ("c" statistics = 0.71 and 0.76, respectively). The newly proposed score, however, discriminated a very low-risk group (101 [13%]) in whom all patients were discharged home, representing patients admitted with none of the 6 predictors of risk.

CONCLUSION

HF risk stratification can be accurately accomplished during the first day of admission with simple and easily obtained clinical variables.

Assessment of Long-Term Physical Function in Acute Respiratory Distress Syndrome (ARDS) Patients

OBJECTIVE

It is often important to adjust for the effect of comorbid diseases on patient outcomes. This study compares the association between physical function in acute respiratory distress syndrome patients with scores on two comorbidity indices, the Charlson Comorbidity Index, designed to predict mortality, and the Functional Comorbidity Index (FCI), which was designed to predict physical function.

DESIGN

This is a prospective, longitudinal, observational study. A total of 73 survivors of acute respiratory distress syndrome were contacted at 3, 6, and 12 mos. Patient comorbidity was evaluated with the Charlson Comorbidity Index and the FCI. Physical function was measured using the Medical Outcomes Study 36-Item Short Form Health Survey Physical Function Subscale and the Physical Component Subscale scores.

RESULT

Mean FCI and Charlson Comorbidity Index scores correlated fairly strongly (Spearman rho = 0.62, P < 0.001). FCI, but not the Charlson Comorbidity Index, scores correlated with the Physical Function Subscale and Physical Component Subscale scores. After controlling for other potentially confounding variables such as age and severity of illness through regression analysis, the FCI score was still significantly associated with both Physical Function Subscale and Physical Component Subscale scores.

CONCLUSIONS

The FCI is a better method of measuring comorbidity with physical function as the outcome. This study illustrates the importance of choosing the most appropriate comorbidity index for the outcome of interest.

Validation of a Risk Score for Dying Within 1 Year of an Admission for Heart Failure

BACKGROUND

Development of heart failure greatly reduces life expectancy. Accurate estimates of the risk of dying are needed in clinical practice and for risk adjustment in observational studies. A relatively simple risk score has been developed to determine the risk of dying within 1-year of an admission for heart failure. We wanted to evaluate the risk score's predictive validity.

METHODS AND RESULTS

Data were abstracted from the electronic medical records of 769 patients admitted to the Minneapolis Veterans Administration medical center with a primary diagnosis of heart failure. Mortality within 1 year of admission was 25%. The c-index for the risk score was 0.71 (95% confidence interval 0.67-0.76). Similar to the original derivation cohort, mortality in risk score groups was 7% for a score lower than 60 (n = 44), 14% for 61 to 90 (n = 246), 26% for 91 to 120 (n = 222), 51% for 121 to 150 (n = 106), and 50% for scores greater than 150 (n = 8).

CONCLUSION

A previously developed risk score for 1-year mortality after an admission for heart failure provided a moderate degree of discrimination in a validation cohort from a different setting. Mortality in risk score groups was consistent with the original patient cohort. These results support the validity of the risk score and its application to a different patient population.

Evaluation of comorbidity indices for inpatient mortality prediction models

BACKGROUND AND OBJECTIVES

The objectives of the current study were: to compare the predictive capacity of the original Charlson comorbidity index (CCI), the CCI with new assigned diagnostic codes and estimated weights, and a new developed comorbidity index in a Brazilian population; and to study the effect of the number of comorbidity diseases recorded on the predictive capacity of the comorbidity indices.

MATERIALS AND METHODS

The study was limited to the Ribeirão Preto region in the State of São Paulo, Brazil, from January 1996 to December 1998. We included only admissions in which the principal diagnoses were respiratory and circulatory diseases.

RESULTS

Evaluation of the CCI indicates that revision of the clinical conditions studied by Charlson, as well as their weights, increased mortality model predictive capacity. The C statistic was 0.72 for the original CCI, and increased to 0.74 for the CCI with new weights and 0.76 for the new index. The C statistic increases in all the comorbidity indices with the utilization of more diagnostic information. This impact is greater when a second secondary diagnosis is added.

CONCLUSIONS

The results of the validity analysis for comorbidity indices favor the utilization of empirically developed indices. However, the increase in predictive capacity was weak. In addition, age and principal diagnosis are the most important predictors of inpatient mortality.

Health outcomes in decompensated congestive heart failure: a comparison of tertiary hospitals in Brazil and United States

BACKGROUND

Few international studies prospectively compared evidence-based practices and health outcomes among congestive heart failure (CHF) cohorts from countries with different cultural and economic backgrounds.

METHODS

Patients consecutively admitted with congestive heart failure to tertiary care teaching hospitals in Brazil and in the United States (U.S.) were systematically evaluated using a structured data form. Follow-up data 3 months after discharge were obtained using chart review and telephone interviews.

RESULTS

U.S. patients were older (p < 0.01), had higher prevalence of ischemic etiology (p < 0.01) and less previous hospitalizations for congestive heart failure (p = 0.03) than Brazilian patients, but similar Charlson comorbidity scores (p = 0.54) and left ventricular (LV) function (p = 0.45). Prescription of angiotensin-converting enzyme inhibitors at discharge was lower at the U.S. hospital (57% vs. 68%; p = 0.03), but beta-blockers prescription was higher (37% vs. 10%; p < 0.01). Length-of-stay was significantly shorter (5 [interquartile range, 3-9] vs. 11 [6-19] days; p < 0.001) and in-hospital mortality was lower (2.4% vs. 13%; p < 0.001) in the U.S. cohort, but fewer clinical events within 3 months after discharge were observed in Brazilian patients (42% vs. 54%; p = 0.02). Combined clinical outcomes within 3 months, including overall mortality and hospital readmission, were similar between cohorts (57% vs. 55%; p = 0.80). In multivariate analysis, hospital site remained significantly associated with health outcomes.

CONCLUSIONS

Medical practice and health-related outcomes were different between U.S. and Brazilian congestive heart failure patients. In order to improve management worldwide, potential factors (structural, cultural or disease-related) that might be associated with these differences need to be evaluated in future studies.

Emergency department observation of heart failure: preliminary analysis of safety and cost

Emergency-department (ED)-based observation-unit treatment has been shown to reduce inpatient admissions, hospital bed-hours, and costs without adversely affecting outcomes for several conditions. A sequential group design study compared risk-matched, acute decompensated heart failure patients admitted directly to the inpatient setting with those admitted to an ED observation unit for up to 23 hours before ED disposition. Outcomes were 30-day readmissions or repeat ED visits for heart failure or 30-day mortality. Estimates of bed-hours and charges between the groups were compared. Sixty-four patients were enrolled with 36 inpatient admissions and 28 observation unit patients. No patients died within 30 days. Observation unit patients had no significant difference in outcomes, a decrease in time from ED triage to discharge, a saving in mean bed-hours, and less total charges. This pilot trial provides preliminary data that suggest admitted, low-risk heart failure patients may be safely and cost-effectively managed in an ED-based observation unit. These findings need to be further evaluated in a randomized clinical trial.

Variation by race in factors contributing to heart failure hospitalizations

BACKGROUND

Previous studies have shown a paradox of lower mortality in black compared with white patients after hospitalization for heart failure, in contrast to the overall higher mortality reported for nonhospitalized black patients with heart failure. We examined racial differences in factors contributing to hospitalization and in severity of illness in a cohort of black and white patients with heart failure who were hospitalized within a financially "equal access" health care system.

METHODS AND RESULTS

We performed a retrospective cohort study on 100 black or white male veterans admitted with heart failure to a VA Medical Center (black, n=52; white, n=48). Severity of illness as measured by the APACHE II score, a generic severity score, was similar between black and white patients (P=.72). However, using a recently developed heart failure-specific risk score, we found that white patients had higher severity of illness (P=.03). White patients had a higher number of coexisting comorbidities than black patients (P=.01), while black patients more frequently had uncontrolled hypertension at the time of admission (P=.004). Nonclinical factors contributing to hospitalization-such as nonadherence with medications or diet, inadequate outpatient follow-up, poor social support, and substance abuse-were documented more frequently for black patients compared with white patients.

CONCLUSIONS

At the time of hospitalization for heart failure, black patients may have an overall lower burden of disease and may more frequently have heart failure exacerbation precipitated by nonclinical factors. These findings may partly account for better long-term survival after hospitalization in black patients compared with white patients.

Association between physician specialty and volumes of treated patients and mortality among patients hospitalized for newly diagnosed heart failure

PURPOSE

To assess the effects of hospital care by a specialist or nonspecialist physician, and by volume of treated patients, on mortality among hospitalized patients with newly diagnosed heart failure.

METHODS

Data describing heart failure patients in Alberta, Canada, from April 1, 1994, to March 31, 2000, were extracted from hospital abstracts and analyzed using hierarchical regression, with adjustment for patient demographic characteristics, comorbid conditions, physician volume, physician specialty, and hospital volume.

RESULTS

There were 16,162 hospital discharges for heart failure. Nonspecialist physicians were predominantly in the two lowest-volume quartiles (93%) and specialists were predominantly in the two highest-volume quartiles (68%). Considering the effects of volume alone and after adjustment for comorbidity, for each 10 additional hospital patients treated by a physician, the odds ratio for in-hospital mortality was 0.97 (95% confidence interval [CI]: 0.95 to 0.98), and the odds ratio for 1-year mortality was 0.99 (95% CI: 0.98 to 0.999). In analyses that considered both volume and specialty, the odds of in-hospital mortality decreased by 4% for each 10 additional in-hospital patients treated by a physician (odds ratio [OR] = 0.96; 95% CI: 0.95 to 0.98). In these same analyses, the odds ratio for in-hospital mortality was 1.32 (95% CI: 1.13 to 1.53) for general practitioners with specialist consultation and 1.32 (95% CI: 1.08 to 1.61) for specialists compared with general practitioners without specialist consultations. At 1 year, mortality was not associated significantly with the volume of in-hospital patients treated, or with the specialty of the treating physician.

CONCLUSION

Treatment by high-volume physicians during hospitalization for newly diagnosed heart failure was associated with a decrease in mortality, but these benefits did not persist at 1 year. The increased mortality noted in patients treated by specialists may be due to residual confounding or unmeasured comorbidity.

Angiotensin-converting enzyme inhibitor therapy in patients with heart failure enrolled in a managed care organization: effect on costs and probability of hospitalization

STUDY OBJECTIVE

To evaluate the effect of angiotensin-converting enzyme (ACE) inhibitor therapy on risk of hospitalization and resource utilization in patients with heart failure enrolled in a managed care organization.

DESIGN

Retrospective medical and pharmacy claims analysis.

PATIENTS

One thousand five hundred seventy-three patients with heart failure enrolled in a managed care organization.

MEASUREMENTS AND MAIN RESULTS

Medical and pharmacy claims from January 1, 1997-December 31, 1999, from a managed care organization covering approximately 350,000 individuals were analyzed. Patients aged 35 years or older with a diagnostic code for heart failure and 18 months of continuous eligibility were selected. From this group (1573 patients), two cohorts were selected based on exposure to an ACE inhibitor. Dependent variables of interest were all-cause hospitalization and total direct medical costs during the 12-month study period. A logistic regression model and an ordinary least-squares model adjusting for patient demographics, comorbidities, and concomitant drug therapy were used to analyze the risk of all-cause hospitalization and total direct medical costs, respectively. Therapy with an ACE inhibitor for 180 days was associated with a decreased risk of all-cause hospitalization (odds ratio 0.65, p<0.0001) and lower total costs (mean dollars 2397, p<0.001) compared with no ACE inhibitor therapy.

CONCLUSION

In patients with a diagnosis of heart failure, exposure to ACE inhibitor therapy is associated with fewer hospitalizations and lower total costs than no ACE inhibitor exposure.

The province of Alberta, Canada avoids the hospitalization epidemic for congestive heart failure patients

BACKGROUND

We assessed the incidence and prevalence of congestive heart failure (CHF) in patients diagnosed at the time of hospitalization and patients diagnosed in specialists offices without prior hospitalization in order to compare the trends in Canada with previously published trends in the USA and other industrialized countries.

METHODS

Administrative data for Alberta, Canada from 1 April 1994 to 31 March 2000.

RESULTS

There was a small but statistically significant decline in the age-sex incident and prevalent hospitalization rates for CHF between 1994/1995 (incidence per 1000 of 1.59; 99% CI 1.51, 1.66: prevalence per 1000 of 2.31; 99% CI 2.22, 2.40) and the year 1999/2000 (incidence per 1000 of 1.24; 99% CI 1.18, 1.30: prevalence per 1000 of 1.97; 99% CI 1.89, 2.05). Crude hospitalization rate per 1000 also demonstrated a small but statistically significant decline between 1994/1995 (2.98; 99% CI 2.88, 3.08) and 1999/2000 (2.55; 99% CI 2.46, 2.64). The age-sex incident rates of ambulatory diagnosis of CHF were similar throughout the 1994/1995-1999/2000 time period (0.88; 99% CI 0.82, 0.94 during 1994/1995 and 0.84; 99% CI 0.79, 0.89 during 1999/2000). The crude mortality percentage for incident hospitalization for CHF were similar throughout the 1994/1995-1999/2000 time period (31.0%; 99% CI 28.7, 33.3 during 1994/1995 and 28.6%; 99% CI 26.3, 30.9 during 1999/2000).

CONCLUSIONS

We noted a small decrease in the incident, prevalent, and total hospitalizations for CHF in the time period 1994/1995-1999/2000. The decrease was not the result of a substituted increase in ambulatory diagnosis for CHF.

Are there race/ethnicity differences in outpatient congestive heart failure management, hospital use, and mortality among an insured population?

OBJECTIVE

The objective of this study was to assess the quality of outpatient care received by patients with congestive heart failure (CHF) and whether differences in care and outcomes exist by race/ethnicity.

BACKGROUND

Appropriate outpatient CHF management can improve patient well-being and reduce the need for costly inpatient care. Yet, little is known regarding outpatient CHF management or whether differences in this care exist by race/ethnicity.

METHODS

Using automated data sources, we identified a cohort of insured patients seen in an outpatient setting for CHF between September 1992 and August 1993. Medical record abstraction was used to confirm diagnosis of CHF. Patients (N = 566) were followed until September 1998. Race/ethnicity differences in outpatient management and medical care utilization were assessed using generalized estimating equations. Differences in mortality and hospitalization for CHF, controlling for patient characteristics and outpatient management, were assessed using Cox and Andersen-Gill models, respectively.

RESULTS

With the exception of beta blocker use and primary care visit frequency, few differences by race/ethnicity in patient characteristics and CHF management were found. However, older black patients had more hospital use both at baseline and during follow up. These differences persisted after adjusting for patient characteristics and clinical management. No race/ethnicity differences were found in mortality.

CONCLUSIONS

In an insured population, older black patients with CHF have substantially more hospital use than older white patients. This increased use was not explained by differences in CHF outpatient management. Further research is needed to understand why race/ethnicity differences in hospital use are observed among older patients with CHF.

Mortalidade hospitalar e tempo de permanência: comparação entre hospitais públicos e privados na região de Ribeirão Preto, São Paulo, Brasil

A avaliação de desempenho dos serviços de saúde é essencial. A comparação de indicadores de desempenho requer o uso de estratégias de ajuste de risco. O objetivo deste artigo é avaliar variações no desempenho clínico, mensurado pela mortalidade e pelo tempo de permanência, entre hospitais públicos e privados, levando em conta diferenças nas características dos pacientes tratados. Este estudo é limitado à região de Ribeirão Preto, São Paulo, Brasil. Entre os anos de 1996 e 1998, 32.906 pacientes admitidos com diagnósticos cardiovasculares e respiratórios foram estudados. As variáveis usadas para o ajuste de risco dos indicadores de desempenho foram: sexo, idade, diagnóstico principal e medidas de gravidade baseada em comorbidade. Os resultados mostraram que o desempenho clínico dos hospitais públicos, mensurado pela mortalidade hospitalar ajustada (razão de chance = 0,41), é superior ao dos privados. Os hospitais públicos e privados não foram estatisticamente diferentes com relação ao tempo de permanência dos pacientes. Ainda que problemas conceituais e metodológicos devam ser resolvidos, taxa de mortalidade e outros indicadores de desempenho ajustados devem ser considerados como instrumentos úteis para identificar problemas de desempenho dos serviços de saúde.

Beta-blockers and angiotensin-converting enzyme inhibitors/receptor blockers prescriptions after hospital discharge for heart failure are associated with decreased mortality in Alberta, Canada

OBJECTIVES

We sought to evaluate the common utilization of beta-blockers and angiotensin-converting enzyme (ACE) inhibitors or receptor blockers (RBs) in congestive heart failure (CHF).

BACKGROUND

We assessed the association between prescriptions of beta-blockers and ACE inhibitors or RBs within three months after hospitalization and mortality for newly diagnosed CHF in Alberta, Canada seniors (age 65 years and older).

METHODS

Administrative hospital discharge abstracts and drug data during October 1, 1994, to December 31, 1999, were analyzed.

RESULTS

There were 11854 hospitalizations for newly diagnosed CHF. The use of beta-blockers within three months after hospitalization increased from 7.3% in 1994-1995 to 20.9% in 1999-2000. The use of ACE inhibitor or RBs within three months after hospitalization increased from 31.0% in 1994-1995 to 44.3% in 1999-2000. Adjusted one-year mortality was lower in seniors with prescriptions for beta-blockers (18.2%; 95% confidence interval [CI] 14.2 to 22.2), ACE inhibitors/RBs (22.3%; 95% CI 20.9 to 23.7), or both (16.6%; 95% CI 13.3 to 20.0), compared with those with no prescriptions (29.9%; 95% CI 28.8 to 31.0). Absolute adjusted risk reduction comparing no prescription with prescription of both beta-blockers or ACE inhibitors/RBs was 13.3% for a relative adjusted risk reduction of 44%.

CONCLUSIONS

This study of incident CHF hospitalizations among seniors demonstrates an association between decreased mortality and the use of beta-blockers, ACE inhibitors/RBs, or combination of both. The effectiveness of beta-blockers and ACE inhibitors/RBs for CHF should be more broadly tested in clinical trials that recruit older patients and those with diastolic dysfunction.

Rural and urban outcomes after hospitalization for congestive heart failure in Alberta, Canada

OBJECTIVES

We compare the hospitalization rate, duration, cost, and mortality for newly diagnosed congestive heart failure in patients admitted to rural and metropolitan hospitals in one Canadian province.

METHODS

Administrative data for Alberta, Canada, from April 1, 1994, to March 31, 2000.

RESULTS

Hospitalizations (16,162) for newly diagnosed congestive heart failure constituted 50% of all hospitalizations for congestive heart failure. Hospitals were distributed as follows: rural with less than 200 cases (21% of hospitalizations), rural with 204 to 646 cases (21% of hospitalizations), regional (13% of hospitalizations), metropolitan with angiography capability (24% of hospitalizations), and metropolitan without angiography capability (21% of hospitalizations). The hospitalization rate per 1000 population was lower for residents of metropolitan regions (1.01; 95% confidence interval [CI] 0.97 to 1.05) compared with residents of rural (1.70; 95% CI 1.65 to 1.75) and regional (1.95; 95% CI 1.90 to 2.00) health regions. Patient comorbidity and severity scores were lower in rural hospitals. Special care unit admissions and cardiac catheterizations were more frequent in patients admitted to metropolitan hospitals. After adjustment, the length of stay and mortality were similar amongst all hospital types. Adjusted hospital total costs were about 23% (900 Canadian dollars) greater in metropolitan hospitals with angiography capability compared to rural hospitals.

CONCLUSION

Hospital admission rates for newly diagnosed congestive heart failure were lower for metropolitan residents compared to non-metropolitan residents. Cost per admission was greatest in metropolitan hospitals with angiography capability compared to other hospital types.

Is heart failure survival improving? Evidence from 2323 elderly patients hospitalized between 1989-2000

BACKGROUND

While drug therapy and medical management improved markedly over the last decade, the basic clinical characteristics of the heart failure patient population treated at the study hospital changed little. This offers an excellent opportunity to study potential heart failure survival improvements for a general patient population.

METHODS

Vital status follow-up through 2001 was obtained from the Social Security Death Index for all 2323 patients aged >or=65 years at the time of an initial, medically managed heart failure hospitalization between October 1989 and March 2000. Kaplan Meier survival probabilities were compared across 4 time periods in the 1990s. A Cox proportional hazards model was used to estimate age, sex, race and comorbidity-adjusted differences in survival among patients admitted in 1989-1991 and 3 subsequent multi-year periods.

RESULTS

There was an increase in the proportion of older female patients with more chronic conditions. Compared with patients admitted in 1989-1991, survival probabilities for patients admitted in 1999-2000 had improved about 5% at 30 days (to 95%) and 10% at 1 year in 1999-2001 (to 73.5%). For those admitted between 1989-1998, there was a 9% improvement over 1989-1991 at 5 years (to 36%). Hazards model results indicated that patients admitted in 1999-2000 had a relative risk of death only 66% that of patients admitted in 1989-1991 (P <.0001).

CONCLUSIONS

These findings provide evidence of modest but significant short-term survival improvements, particularly after 1998, when drug therapy had became optimal in the inpatient setting, patient education and discharge planning became better documented, and inpatient mortality rates had declined substantially.

Heart failure in patients admitted to hospital: mortality is still high

BACKGROUND: Two separate cohorts of consecutive patients admitted to hospital with a primary diagnosis of heart failure were studied, the first in 1986 in Rochdale, and the second in 1995 in Brighton. METHODS: We observed the clinical profile, treatment and mortality during hospital admission and reviewed their status at 6 months. There were 132 patients in the Rochdale cohort and 223 in the Brighton cohort. RESULTS: The Rochdale cohort was characterised by a lower mean age and longer hospital stay. Significant differences were also observed in co-morbidity and the use of ACE inhibitors, but hospital mortality was almost identical (25% in Rochdale and 24% in Brighton). A low systolic blood pressure, hyponatraemia, hyperkalaemia and a raised blood urea at presentation were independent adverse prognostic factors. In contrast, prior treatment with ACE inhibitors in patients with congestive cardiac failure led to a more favourable hospital outcome. Age, gender and co-morbidity did not affect mortality apart from patients with acute myocardial infarction. Follow-up of these cohorts showed that mortality of the two groups remained high at 180 days after admission (40% in Rochdale and 39% in Brighton). There were marked differences in the use of ACE inhibitors in survivors, but target doses of ACE inhibitors (enalapril 20 mg/day or equivalent) were only achieved in 31%, despite direct communication between the hospital and primary care physicians. CONCLUSIONS: Although clinical and treatment profiles differed between the two periods studied, the hospital and 6-month mortality of patients with heart failure remained high. More emphasis needs to be given to optimising ACE inhibitor use in primary care.

Assessing comorbidity using claims data: an overview

Comorbidity, additional disease beyond the condition under study that increases a patient's total burden of illness, is one dimension of health status. For investigators working with observational data obtained from administrative databases, comorbidity assessment may be a useful and important means of accounting for differences in patients' underlying health status. There are multiple ways of measuring comorbidity. This paper provides an overview of current approaches to and issues in assessing comorbidity using claims data, with a particular focus on established indices and the SEER-Medicare database. In addition, efforts to improve measurement of comorbidity using claims data are described, including augmentation of claims data with medical record, patient self-report, or health services utilization data; incorporation of claims data from sources other than inpatient claims; and exploration of alternative conditions, indices, or ways of grouping conditions. Finally, caveats about claims data and areas for future research in claims-based comorbidity assessment are discussed. Although the use of claims databases such as SEER-Medicare for health services and outcomes research has become increasingly common, investigators must be cognizant of the limitations of comorbidity measures derived from these data sources in capturing and controlling for differences in patient health status. The assessment of comorbidity using claims data is a complex and evolving area of investigation.

How to adjust for comorbidity in survival studies in ESRD patients: a comparison of different indices

BACKGROUND

Many patients with end-stage renal disease (ESRD) have additional comorbid conditions. Differences in the presence and severity of these comorbid conditions can bias comparisons between treatment groups. Adjustment for prognostic factors can statistically counterbalance these differences. For this purpose, appropriate weighting of comorbid conditions is necessary. We evaluated three existing methods to score comorbidity in patients with ESRD and compared their ability to predict survival: the Khan, Davies, and Charlson indices. In addition, these three indices were compared with a new index that explicitly incorporates the severity grading of a number of comorbid diseases.

METHODS

In a large Dutch prospective multicenter study (Netherlands Co-operative Study on the Adequacy of Dialysis-2), new patients with ESRD were included. Comorbidity was assessed at the start of dialysis therapy. Patient data were randomly allocated to a modeling or testing set. The new index was developed in the modeling set. All indices were evaluated in the testing set.

RESULTS

We obtained data for 1,205 patients. Of the three existing indices, the Charlson index had the best discriminating features, with a concordance c statistic of 0.71. The addition of severity grading of several comorbid conditions did not improve discrimination. After combining the comorbidity indices with age, all c statistics improved. These final values ranged from 0.72 to 0.75.

CONCLUSION

We conclude that the Khan, Davies, and Charlson scores are appropriate for expressing the prognostic impact of comorbidity on mortality risk in patients with ESRD provided sufficient adjustment for age is performed. Adding the severity grading of several comorbid conditions will not lead to improved prognostic power.