Costs and effectiveness of cardiac rehabilitation for dialysis patients following coronary bypass
2008; Elsevier BV; Volume: 74; Issue: 8 Linguagem: Inglês
10.1038/ki.2008.381
ISSN1523-1755
AutoresYijian Huang, Rebecca Zhang, Steven D. Culler, Nancy G. Kutner,
Tópico(s)Heart Failure Treatment and Management
ResumoDialysis patients have a high risk of cardiovascular disease. Cardiac rehabilitation is recommended in the general population as a standard component of care and covered by Medicare for those who have undergone coronary artery bypass grafting (CABG). Here we determined the impact of cardiac rehabilitation on Medicare expenditures and its cost effectiveness in dialysis patients. A cohort of 4,324 patients with end-stage renal disease who began chronic hemodialysis and had undergone CABG over a seven year period were selected from the United States Renal Data System. Cardiac rehabilitation was defined by Current Procedural Terminology codes for monitored and non-monitored exercise in Medicare claims data. Medicare expenditures included in and outpatient claims adjusted to 1998 dollars. Over a 42-month follow-up, cardiac rehabilitation at baseline was associated with higher cumulative Medicare expenditures but this increase was not statistically significant. During the same period, cardiac rehabilitation was significantly associated with longer cumulative life, having an incremental benefit of 76 days. The incremental cost-effectiveness ratio of $13,887 per year of life saved suggests that cardiac rehabilitation is highly cost-effective in patients with end-stage renal disease following CABG. Dialysis patients have a high risk of cardiovascular disease. Cardiac rehabilitation is recommended in the general population as a standard component of care and covered by Medicare for those who have undergone coronary artery bypass grafting (CABG). Here we determined the impact of cardiac rehabilitation on Medicare expenditures and its cost effectiveness in dialysis patients. A cohort of 4,324 patients with end-stage renal disease who began chronic hemodialysis and had undergone CABG over a seven year period were selected from the United States Renal Data System. Cardiac rehabilitation was defined by Current Procedural Terminology codes for monitored and non-monitored exercise in Medicare claims data. Medicare expenditures included in and outpatient claims adjusted to 1998 dollars. Over a 42-month follow-up, cardiac rehabilitation at baseline was associated with higher cumulative Medicare expenditures but this increase was not statistically significant. During the same period, cardiac rehabilitation was significantly associated with longer cumulative life, having an incremental benefit of 76 days. The incremental cost-effectiveness ratio of $13,887 per year of life saved suggests that cardiac rehabilitation is highly cost-effective in patients with end-stage renal disease following CABG. In 2005, there were approximately 485,000 end-stage renal disease (ESRD) patients in the United States, for whom total Medicare costs reached 21.3 billion dollars.1U.S. Renal Data System USRDS 2007 Annual Data Report: Atlas of End-Stage Renal Disease in the United States. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD2007Google Scholar Given this economic burden, cost evaluation of medical therapies, in addition to standard effectiveness and safety assessment, is increasingly important for policy-makers and third-party payers. The leading cause of death among dialysis patients is cardiovascular disease (CVD), with cardiovascular mortality almost 40 times that in the general population.2.Collins A.J. Cardiovascular mortality in end-stage renal disease.Am J Med Sci. 2003; 325: 163-167Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar For patients having experienced coronary events, cardiac rehabilitation is considered an integral part of the contemporary care.3.Leon A.S. Franklin B.A. Costa F. et al.Cardiac rehabilitation and secondary prevention of coronary heart disease: an American Heart Association scientific statement from the Council on Clinical Cardiology (Subcommittee on Exercise, Cardiac Rehabilitation, and Prevention) and the Council on Nutrition, Physical Activity, and Metabolism (Subcommittee on Physical Activity), in collaboration with the American Association of Cardiovascular and Pulmonary Rehabilitation.Circulation. 2005; 111: 369-376Crossref PubMed Scopus (678) Google Scholar,4.Giannuzzi P. Saner H. Bjornstad H. et al.Secondary prevention through cardiac rehabilitation: position paper of the Working Group on Cardiac Rehabilitation and Exercise Physiology of the European Society of Cardiology.Eur Heart J. 2003; 24: 1273-1278Crossref PubMed Scopus (222) Google Scholar Medicare covers cardiac rehabilitation for patients who have undergone coronary artery bypass grafting (CABG) and for patients who have had an acute myocardial infarction (AMI) in the past 12 months or have stable angina.5Centers for Medicare & Medicaid Services (CMS) Medicare National Coverage Determinations Manual. 2007http://www.cms.hhs.gov/manuals/downloads/ncd103c1_Part1.pdfGoogle Scholar In the general population, cardiac rehabilitation has been shown to reduce coronary heart disease risk factors6.Thomas R.J. Miller N.H. Lamendola C. et al.National survey on gender differences in cardiac rehabilitation programs: patient characteristics and enrollment patterns.J Cardiopulm Rehabil. 1996; 16: 402-412Crossref PubMed Scopus (241) Google Scholar and mortality from coronary heart disease.7.Dalal H. Evans P.H. Campbell J.L. Recent developments in secondary prevention and cardiac rehabilitation after acute myocardial infarction.BMJ. 2004; 328: 693-697Crossref PubMed Scopus (56) Google Scholar,8.Belardinelli R. Georgiou D. Cianci G. et al.Randomized, controlled trial of long-term moderate exercise training in chronic heart failure: effects on functional capacity, quality of life, and clinical outcome.Circulation. 1999; 99: 1173-1182Crossref PubMed Scopus (896) Google Scholar,9.Hedback B. Perk J. Wodlin P. Long-term reduction of cardiac mortality after myocardial infarction: 10-year results of a comprehensive rehabilitation programme.Eur Heart J. 1993; 14: 831-835Crossref PubMed Scopus (143) Google Scholar,10.Lau A. Antman E.M. Jimenez-Silva J. et al.Cumulative meta-analysis of therapeutic trials for myocardial infarction.N Eng J Med. 1992; 327: 248-254Crossref PubMed Scopus (915) Google Scholar,11.Oldridge N.B. Guyatt G.H. Fischer M.E. et al.Cardiac rehabilitation after myocardial infarction: combined experience of randomized clinical trials.JAMA. 1988; 260: 945-950Crossref PubMed Scopus (923) Google Scholar Furthermore, a limited number of studies have suggested that it is cost effective and even cost saving.12.Yu C. Lau C. Chau J. et al.A short course of cardiac rehabilitation program is highly cost effective in improving long-term quality of life in patients with recent myocardial infarction or percutaneous coronary intervention.Arch Phys Med Rehabil. 2004; 85: 1915-1922Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar,13.Georgiou D. Chen Y. Appadoo S. et al.Cost-effectiveness analysis of long-term moderate exercise training in chronic heart failure.Am J Cardiol. 2003; 87: 984-988Abstract Full Text Full Text PDF Scopus (93) Google Scholar,14.Ades P. Pashkow F. Nestor J. Cost-effectiveness of cardiac rehabilitation after myocardial infarction.J Cardiopulm Rehabil. 1997; 17: 222-231Crossref PubMed Scopus (126) Google Scholar,15.Ades P. Huang D. Weaver S. Cardiac rehabilitation participation predicts lower rehospitalization costs.Am Heart J. 1992; 123: 916-921Abstract Full Text PDF PubMed Scopus (145) Google Scholar CABG may confer a lowered risk for death in dialysis patients compared with no revascularization. In an earlier report, we investigated the characteristics and long-term survival of dialysis patients who received cardiac rehabilitation after CABG, using United States Renal Data System files linked with Medicare inpatient and outpatient claims.16.Kutner N.G. Zhang R. Huang Y. et al.Cardiac rehabilitation and survival of dialysis patients after coronary bypass.J Am Soc Nephrol. 2006; 17: 1175-1180Crossref PubMed Scopus (31) Google Scholar We found that patients who received cardiac rehabilitation after CABG, compared to those who did not, had significantly improved survival. The purpose of this study was to assess the impact of dialysis patients' receipt of cardiac rehabilitation after CABG on Medicare expenditure and cost effectiveness. A total of 6040 eligible ESRD patients who initiated chronic hemodialysis (HD) and underwent CABG between 1998 and 2004 were identified from United States Renal Data System database. Of them, 635 patients were excluded because of being nonambulatory, having extended hospital stays for the CABG procedure, or incurring excessive Medicare expenditure during a 6-month entry period. In addition, 1081 patients were dropped from the study cohort as being in the lowest propensity quintile for receiving cardiac rehabilitation at baseline. See Materials and Methods section for details of these exclusion criteria. The final study cohort consisted of 4324 patients. Baseline demographic and clinical characteristics of the 4324 patients are reported in Table 1. A propensity score (C-index, 0.65) was calculated for receiving cardiac rehabilitation at baseline, and Table 1 is summarized by the cardiac rehabilitation status and the propensity score quartiles. Of the study cohort, 68% were 65 years of age or older, 72% were men, and 81% were white. Approximately 16.4% received cardiac rehabilitation in a period between CABG surgery and 6 months after CABG hospitalization discharge. The cardiac rehabilitation and noncardiac rehabilitation groups as a whole were fairly different in their baseline characteristics. However, these differences were minimized after stratification by the propensity score quartiles.Table 1Characteristics of the study cohortPropensity score quartileFirstSecondThirdFourthP-valueCardiac rehabilitation (yes/no)YNYNYNYNSize791001147934194888287794Age ≥65 years (%)57676869756872670.18Male (%)39507369827891930.90Race (%)1.00 Asian65332200 Black373016164722 Native American12123112 White5663797891899696 Other00110000Medicaid coverageaMeasured at HD start. (%)3329141555010.98No. of cardiovascular conditionsaMeasured at HD start. (mean)1.21.41.41.31.41.31.31.20.45HemoglobinaMeasured at HD start. (g/100 ml; mean)9.79.810.19.910.010.010.010.10.97Serum albuminaMeasured at HD start. (g/100 ml; mean)3.13.13.13.13.23.23.33.30.73Diabetic ESRDaMeasured at HD start. (%)61575655494730350.86AMI before CABG (%)41504342353621200.69COPDaMeasured at HD start. (%)6812859870.88Vintage (months; mean)17.413.813.014.512.314.414.415.10.29Hospital stays after CABG (days; mean)14.416.714.013.812.912.69.79.50.83Discharge to self-care/outpatient (%)35354642495259560.53Initial Medicare expenditurebAccumulated during the 6-month entry period. ($; mean)26,57927,34124,79724,71418,72919,28314,73614,5510.69ESRD, end-stage renal disease; CABG, coronary artery bypass grafting; HD, hemodialysis; AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease.A total of 4324 HD patients with CABG surgery as selected on the basis of baseline measures.a Measured at HD start.b Accumulated during the 6-month entry period. Open table in a new tab ESRD, end-stage renal disease; CABG, coronary artery bypass grafting; HD, hemodialysis; AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease. A total of 4324 HD patients with CABG surgery as selected on the basis of baseline measures. The follow-up of this study started 6 months after CABG hospitalization discharge. The mean follow-up time of the study cohort was 20.3 months, and 43.6% died or achieved 42 months of follow-up. The censoring was because of either the end of follow-up as of 31 December 2004 or renal transplantation; about 6% of the study cohort had renal transplantation during the 42-month follow-up period. The overall survival rate was 45.0% at 42 months. Cumulative Medicare expenditure and cumulative lifetime at 42 months were used as the primary cost and effectiveness outcomes, respectively. A cumulative measure at a follow-up time is the measure accumulated up to death or the follow-up time, whichever occurs earlier. The estimated mean cumulative Medicare expenditure and lifetime at 42 months for the study cohort were $105,133 and 29.8 months, respectively. Table 2 provides estimated covariate effects on cumulative Medicare expenditure and lifetime at 42 months, from multivariate regression models with stratification by the propensity score quartiles. Cardiac rehabilitation at baseline was associated with higher cumulative Medicare expenditure at 42 months, with an increase of $2904 (95% confidence interval (CI): -7028, 11,940), but this increase was not statistically significant. At the same time, cardiac rehabilitation at baseline was significantly associated with longer cumulative lifetime with an increase of 76 days (95% CI: 22, 129). The estimated incremental cost-effectiveness ratio was $13,887 per year of life saved at 42 months. Among other baseline measures included in the models, older patient age (>=65), male gender, and vintage were significantly associated with both shorter survival time and lower Medicare expenditure, whereas initial Medicare expenditure was significantly associated with shorter survival and higher Medicare expenditure. Number of cardiovascular conditions and chronic obstructive pulmonary disease (COPD) at HD start were significantly associated with shorter survival, whereas Asian race (relative to white) was significantly associated with longer survival. Finally, black race (relative to white), Medicaid coverage, higher serum albumin, primary diagnosis of diabetes, and AMI before CABG were significantly associated with higher Medicare expenditure at 42 months.Table 2Multivariate linear regression models predicting cumulative Medicare expenditure and cumulative lifetime at 42 months of 4324 HD patients with CABG surgeryRisk factor at baselineMedicare expenditure ($)Lifetime (days)Estimate95% CIEstimate95% CIAge ≥65 years−22,454−32,978−12,775−98−144−51Male gender−15,498−25,345−6636−53−107−5Race (reference=white) Asian22,908−14,96855,1301163214 Black18,683439432,11934−38100 Native American3844−24,74534,53389−45201 Other40,574−41,70565,778264−582491Medicaid coverageaMeasured at HD start.21,564610035,78028−49100No. of cardiovascular conditionsaMeasured at HD start.−1406−42741052−26−42−11HemoglobinaMeasured at HD start.−486−28611726−6−175Serum albuminaMeasured at HD start.634639011,69816−1548Diabetic ESRDaMeasured at HD start.11,623328419,77125−1968AMI before CABG917960717,1986−3847COPDaMeasured at HD start.130−10,79011,319−76−146−1Vintage−600−843−327−9−11−7Hospital stays after CABG−336−1022283−2−62Percentile of initial Medicare expenditurebAccumulated during the 6-month entry period.47,43728,68766,159−236−336−139Discharge to self care/outpatient1516−6483904818−1856Cardiac rehabilitation2904−702811,9407622129CABG, coronary artery bypass grafting; HD, hemodialysis; ESRD, end-stage renal disease; AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease.a Measured at HD start.b Accumulated during the 6-month entry period. Open table in a new tab CABG, coronary artery bypass grafting; HD, hemodialysis; ESRD, end-stage renal disease; AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease. To investigate the temporal trend of differences between the cardiac rehabilitation and noncardiac rehabilitation groups, the same models were applied to cumulative Medicare expenditure and lifetime at 6, 12, 18, 24, 30, and 36 months. The regression coefficients of cardiac rehabilitation along with their 95% CIs are plotted in Figure 1. The differences in both cumulative Medicare expenditure and lifetime were small during the first year. The survival benefit of cardiac rehabilitation increased steadily over time, reaching statistical significance after month 30. Meanwhile, by month 36, the cardiac rehabilitation group showed a statistically nonsignificant increase in cumulative Medicare expenditure. The estimated incremental cost-effectiveness ratio value for receiving cardiac rehabilitation at baseline was $13,887 per year of life saved among dialysis patients over a time horizon of 42 months. Although there is no universally accepted benchmark for the 'cost-effective' incremental cost-effectiveness ratio value, a familiar categorization scheme designates a value of $20,000 or less as highly cost effective, $20,000 to $40,000 as cost effective, $40,000 to $60,000 as borderline cost effective, and $60,000 or more as expensive.17.Kupersmith J. Holmes-Rovner M. Hogan A. et al.Cost-effectiveness analysis in heart disease, Part I: general principles.Prog Cardiovasc Dis. 1994; 37: 161-184Abstract Full Text PDF PubMed Scopus (50) Google Scholar,18.Kupersmith J. Holmes-Rovner M. Hogan A. et al.Cost-effectiveness analysis in heart disease, Part II: preventive therapies.Prog Cardiovasc Dis. 1995; 37: 243-271Abstract Full Text PDF PubMed Scopus (62) Google Scholar,19.Kupersmith J. Holmes-Rovner M. Hogan A. et al.Cost-effectiveness analysis in heart disease, Part III: ischemia, congestive heart failure, and arrhythmias.Prog Cardiovasc Dis. 1995; 37: 307-346Abstract Full Text PDF PubMed Scopus (68) Google Scholar In this realm, cardiac rehabilitation is regarded as highly cost effective in our dialysis population under investigation. To perform cost-effectiveness analysis, we adopted an approach to the analysis of survival time different from that in our previous investigation.16.Kutner N.G. Zhang R. Huang Y. et al.Cardiac rehabilitation and survival of dialysis patients after coronary bypass.J Am Soc Nephrol. 2006; 17: 1175-1180Crossref PubMed Scopus (31) Google Scholar The propensity score was used in the current analysis to define the study cohort and to conduct stratified analysis. In addition, the main survival time outcome was cumulative lifetime at 42 months and the multivariate linear regression model was adopted here, whereas the Cox proportional hazards model was employed in the previous investigation. Nevertheless, both analyses were highly consistent in showing that receiving cardiac rehabilitation was significantly associated with longer survival time in dialysis patients after CABG. Our study suggested a moderate increment in cumulative Medicare expenditure at 42 months associated with receiving cardiac rehabilitation. However, this increment was not likely because of the cardiac rehabilitation cost. By the start of follow-up, most baseline cardiac rehabilitation sessions were complete. Also, Figure 1 shows that the difference in cumulative Medicare expenditure between the cardiac rehabilitation and noncardiac rehabilitation groups was small during the first year. Repeating our analysis by considering hospitalization expenditure only (data not shown), we found that the increment of cumulative hospitalization expenditure was similar to that of cumulative Medicare expenditure in the cardiac rehabilitation group. Economic evaluations of cardiac rehabilitation have been limited even in the general population.20.Papadakis S. Oldridge N.B. Coyle D. et al.Economic evaluation of cardiac rehabilitation: a systematic review.Eur J Cardiovasc Prev Rehabil. 2005; 12: 513-520Crossref PubMed Scopus (77) Google Scholar,21.Oldridge N.B. Comprehensive cardiac rehabilitation: is it cost-effective?.Eur Heart J. 1998; 19: O42-O49PubMed Google Scholar Data have supported cost effectiveness of cardiac rehabilitation, with increased cost per year of life saved comparing favorably with other currently funded interventions. The majority of these investigations focused on patients who have had myocardial infarction. The strength of our study lies in the large population-based sample of dialysis patients who were followed for an extended period of time after CABG. Nevertheless, it is also important to recognize several potential limitations of this investigation. First, large differences in baseline characteristics were observed between patients who received cardiac rehabilitation and those who did not. We made extensive efforts to address this imbalance by adopting propensity score and multivariate regression analysis. Nevertheless, as with any observational studies, imbalance could still exist in unobserved patient characteristics that might account for the expenditure and survival differences. Second, our results were based on, and therefore applied to, healthier patients in the chronic HD population who underwent CABG. The observed rate of receiving cardiac rehabilitation among dialysis patients, approximately 13%, was lower than the estimated rate of approximately 23% in the general population after CABG.16.Kutner N.G. Zhang R. Huang Y. et al.Cardiac rehabilitation and survival of dialysis patients after coronary bypass.J Am Soc Nephrol. 2006; 17: 1175-1180Crossref PubMed Scopus (31) Google Scholar The cardiac rehabilitation rate for less healthy subgroups of chronic HD patients was even lower, and these subgroups were excluded from this investigation. Third, the time horizon for this study was 42 months after the 6-month entry period following CABG hospitalization discharge. As in any study with limited follow-up, the increments of cumulative Medicare expenditure and lifetime might not be the same when the time horizon is extended. At the same time, the findings of our study are consistent with literature in the general population that supports the efficacy and cost effectiveness of cardiac rehabilitation. It is likely that core aspects of cardiac rehabilitation services would be beneficial for virtually all patients with cardiovascular disease. In conclusion, this observational study suggests that cardiac rehabilitation is highly cost effective in ESRD patients following CABG. The data used in this analysis came from linking United States Renal Data System files to Medicare inpatient and outpatient claims. The former contained demographic and clinical information of dialysis patients. Medicare primary-pay ESRD patients who initiated chronic HD and underwent CABG while on chronic dialysis during the period between 1 January 1998 and 31 December 2004 were identified retrospectively from the United States Renal Data System database. For patients who were younger than 65 years, only those with CABG procedure dates that occurred day 90 or later of ESRD were included, because many of these patients did not become eligible for Medicare for up to 90 days after initiating dialysis and therefore might not have complete claims data. In addition, we did not include patients who, after the CABG procedure, left the hospital against medical advice or were discharged to home under care of an i.v. therapy provider or to hospice care. A preliminary investigation of this study cohort showed that less than 13% of CABG survivors received cardiac rehabilitation, and the rate varied substantially across subgroups. Those subgroups with minimal rate would provide little information on the association of cardiac rehabilitation with Medicare expenditure and survival. For this reason, we excluded the following patients: (1) patients who were nonambulatory at dialysis initiation, (2) patients whose CABG hospitalization exceeded 44 days following the surgery, or (3) patients who incurred Medicare expenditure of over $66,244 during the 6-month entry period. The cutoff points for the hospital stays and initial Medicare expenditure were determined as the 99th percentiles among patients who received cardiac rehabilitation at baseline. Physical limitation might be the main reason for the observed low rehabilitation rate among nonambulatory patients. Patients with longer hospital stays for the CABG procedure or with higher initial Medicare expenditure were more likely to experience serious complications post-surgery; they were overly represented among those who did not receive cardiac rehabilitation. After applying these exclusion criteria, a propensity score for receiving cardiac rehabilitation was calculated, and the patients in the lowest quintile of the propensity score distribution were excluded as well. The objective of this study was to assess the association between receiving cardiac rehabilitation at baseline and cumulative Medicare expenditure and survival time during a follow-up period of up to 42 months. The follow-up started after a 6-month entry period following the discharge from CABG hospitalization to allow for the determination of receiving cardiac rehabilitation at baseline. By design, all patients included in the analysis survived at least 6 months after discharge from CABG hospitalization. Medicare expenditure and survival time were followed up to 31 December 2004. A patient undergoing renal transplantation was treated as being censored at the time of transplantation. Cardiac rehabilitation was defined as outpatient cardiac rehabilitation without or with continuous electrocardiogram monitoring (Current Procedural Terminology codes 93797 and 93798). Medicare covers participation by eligible patients in a cardiac rehabilitation exercise program for up to 36 sessions, with three sessions a week in a single 12-week period of therapy. For this analysis, cardiac rehabilitation received after CABG and before the end of the 6-month entry period was regarded as cardiac rehabilitation at baseline. This time period included about 90% of the patients in our study population who ever received cardiac rehabilitation. Demographic characteristics of the patients included age, gender, and race. Medicaid coverage at dialysis initiation (yes/no) was used as an indicator of socioeconomic status (Medicaid is a federal/state health insurance entitlement program for low-income people). Clinical characteristics documented at dialysis initiation included primary diagnosis of diabetes, number of existing cardiovascular conditions (congestive heart failure, coronary artery disease, history of AMI, cardiac arrest, dysrhythmia, cerebral vascular disease, peripheral vascular disease), COPD, serum albumin (g/100 ml), and hemoglobin (g/100 ml). After dialysis initiation, AMI that occurred before CABG was also determined. Medicare expenditures included claims from institutional and physician/supplier data files. All costs were adjusted to 1998 dollars, using the Medical Care component of the Consumer Price Index for inflation. In addition, all cost value was discounted at an annual rate of 3%. The receipt of cardiac rehabilitation was not randomly determined in this investigation, as with any observational study. In fact, substantial clinical and socioeconomic differences between patients who received cardiac rehabilitation at baseline and those who did not were observed. Inadequately accounting for these differences could lead to biased assessment of the effect of cardiac rehabilitation. We employed the propensity score method22.Rosenbaum P.R. Rubin D.B. The central role of the propensity score in observational studies for causal effects.Biometrika. 1983; 70: 41-55Crossref Scopus (14775) Google Scholar,23.D'Agostino Jr, R. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.Statist Med. 1998; 17: 2265-2281Crossref PubMed Scopus (4086) Google Scholar to exclude subgroups of patients with minimal propensity for receiving cardiac rehabilitation and to carry out propensity score-stratified analyses. Stepwise logistic regression was used to compute the propensity score for receiving cardiac rehabilitation. Baseline measures used as candidate covariates included demographic characteristics (age, gender, race, and Medicaid coverage), clinical characteristics at the dialysis initiation (number of cardiovascular conditions, hemoglobin, serum albumin, primary diagnosis of diabetes, COPD), and before CABG (AMI), as well as measures associated with the CABG procedure (vintage defined as the dialysis duration at the time of CABG, hospital stays after CABG, discharge destination of the CABG hospitalization), and the decile ranking of Medicare expenditure during the 6-month entry period. After the patients in the lowest quintile of the estimated propensity score were excluded, the same stepwise logistic regression was re-run to refine the propensity scores for the study cohort. The refined propensity score was used in the subsequent stratified analyses. Demographic and baseline clinical characteristics were summarized using means for continuous variables and percentages for discrete variables, by the status of receiving cardiac rehabilitation at baseline and by the propensity score quartiles. Comparisons, as stratified by the propensity score quartiles, were made between cardiac rehabilitation and noncardiac rehabilitation groups, using linear regression for continuous variables and the Cochran–Mantel–Haenszel test for discrete variables. A multivariate linear regression model was used to examine the association between receiving cardiac rehabilitation at baseline and cumulative Medicare expenditure during the follow-up. Other dependent variables incorporated for adjustment included demographic characteristics and baseline clinical characteristics as well as percentiles of the entry-period Medicare expenditure. The analysis was stratified by the propensity score quartiles, that is, stratum-specific intercepts being incorporated in the model. To accommodate censoring because of either renal transplantation or the end of follow-up, we used a Horvitz–Thompson type approach to missing data.24.Little R.J.A. Rubin D.B. Statistical Analysis with Missing Data. 2nd edn. Wiley, New Jersey2002Crossref Scopus (11988) Google Scholar,25.Lin D.Y. Linear regression analysis of censored medical costs.Biostatistics. 2000; 1: 35-47Crossref PubMed Google Scholar An important component of this approach was to estimate the probability of being censored for every uncensored observation in the sample. A proportional hazards model was employed for the censoring time with all the aforementioned dependent variables. Subsequently, the uncensored observations, each weighted inversely by its associated probability, formed a pseudosample of complete cases to which the linear regression model could be applied. To carry out further cost-effectiveness analysis, a similar multivariate linear regression model coupled with the Horvitz–Thompson approach was employed for cumulative lifetime. For inference with these models, bootstrap of size 1000 was used to construct 95% percentile CIs. The incremental cost-effectiveness ratio was calculated for the cost-effectiveness assessment of receiving cardiac rehabilitation at baseline after CABG. The authors state no conflict of interest. This study was supported by National Institutes of Health contracts N01-DK-1-2471 and HHSN267200715004C, ADB no. N01-DK-7-5004 and by National Institutes of Health grant R01 CA090747.
Referência(s)