Participation in Cardiac Rehabilitation and Survival Following Coronary Artery Bypass Graft Surgery: A Community Based Study
Participation in Cardiac Rehabilitation and Survival Following Coronary Artery Bypass Graft Surgery: A Community Based Study Quinn R. Pack, Kashish Goel, Brian D. Lahr, Kevin L. Greason, Ray W. Squires, Francisco Lopez-Jimenez, Zixin Zhang and Randal J. Thomas Circulation. published online July 8, 2013; Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2013 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539 The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/early/2013/07/08/CIRCULATIONAHA.112.001365 Data Supplement (unedited) at: http://circ.ahajournals.org/content/suppl/2013/07/08/CIRCULATIONAHA.112.001365.DC1.html Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/ Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Participation in Cardiac Rehabilitation and Survival Following Coronary Artery Bypass Graft Surgery: A Community Based Study Running title: Pack et al.; Cardiac rehabilitation after CABG Quinn R. Pack, MD1; Kashish Goel, MD1,2; Brian D. Lahr, MS3; Kevin L. Greason, MD4; Ray W. Squires, PhD1; Francisco Lopez-Jimenez, MD, MS1; Zixin Zhang, MD1,5; Randal J. Thomas, MD, MS1 1 Divv of Cardiovascular Di Car ardi d ovvascu asc lar Diseases, Dept of Intern Internal rnal rn al Medicine; 3Di Divv off B Biomedical i medical Statistics and io Informatics; In nformatticcs; 4Di Div D v of C Cardiothoracic arrdi ardi d ot otho hora ho raci ra ciic Su S Surgery, rger rg ery, er y Ma Mayoo C Clinic, lini li nicc, Rochester, ni Roc oche hest he steer, st er, MN; MN N; 2Di Divv of IInternal nteerna nt erna nall Medicine, Me , De D Detroit troit Med M Medical edica call C Center/Wayne ent n err/W Waynee S State tatee U University, nive vers ve rsit rs ity,, De it Detroit, etrroit, MI M MI;; 5D Dept ept p of of Card C Cardiology, ardiiolo ogyy, F Firs First irs rstt Ho Hospit Hospital ital it a ooff Chin al C China hin ina Me M Medical edi dica di call Un ca Univ University, iver iv ersi sity si ty,, Sh ty Shenyang, hen nya yang ng,, Ch ng China hin inaa Address for Add f Correspondence: C d Randal Thomas, MD Division of Cardiovascular Disease and Internal Medicine Mayo Clinic 200 First Street SW Rochester, MN 55905 Tel: 507-774-4375. Fax: 507-266-7929. E-mail: [email protected] Journal Subject Codes: Treatment:[26] Exercise/exercise testing/rehabilitation, Cardiovascular (CV) surgery:[36] CV surgery: coronary artery disease, Treatment:[122] Secondary prevention, Ethics and policy:[100] Health policy and outcome research 1 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Abstract: Background—Cardiac rehabilitation (CR) is recommended for all patients following coronary artery bypass surgery (CABG), yet little is known about the long term mortality effects of CR in this population. Methods and Results—We performed a community-based analysis on residents of Olmsted County, Minnesota who underwent CABG between 1996 and 2007. We assessed the association between subsequent outpatient CR attendance and long-term survival. Propensity analysis was performed. Cox PH regression was then used to assess the association between CR attendance 846 and all-cause mortality adjusted for the propensity to attend CR. We identified 84 46 eeligible lig igib ig ible ib le patients (age 66 ± 11 years, 76% men, and 96% non-Hispanic whites) who survived at least 6 months attended mean mo mont nths nt hs after aft fter e ssurgery, er urrge gery ry, of whom 582 (69%) attend ded e CR. During a mea eaan ((± ± SD) follow-up of 9.0 3.7 mortality ± 3. .7 years, the he 110-year 0-ye yeaar aall-cause ll-c ll -cau -c au use m o tallity rrate or atte was wass 28% 288% (193 (1 193 deaths). deaath ths) s)). Adjusted Adju Adju ustted for forr the the h propensity attend CR, was with pr prop open op ensi en sity ty y tto o at atte teend nd C R,, pparticipation arti ar tici ti ciipa pati tioon ti on iin n CR w a aassociated as ssoci c at ated ed w i h a 10 it 110-year -year -ye ear relative rela rela lati tive ve rrisk issk redu rreduction eduucttio ion inn all-cause mortality mor orta taali lity ty of of 46% 46 6% (HR=0.54; (H HR= R 0. 0 54 4; 95% 95% CI, CI, I 0.40-0.74; 0.4400-0. 0 74 0. 74;; p<0.001), p<0. p< 0 0001) 1),, and and a 10-year 10-y 10 -y yea earr absolute risk reduction of 12.7% (NNT=8). There was no evidence of a differential effect of CR on mortality with respect to age (65 vs <65 yrs.), gender, diabetes, or prior myocardial infarction. Conclusions— Cardiac rehabilitation attendance is associated with a significant reduction in 10 year all-cause mortality following CABG. Our results strongly support national standards that recommend CR for this patient group. Key words: cardiac rehabilitation, mortality, secondary prevention, bypass surgery, Propensity, Participation 2 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Introduction Each year, more than 300,000 patients undergo Coronary Artery Bypass Grafting (CABG) surgery in the United States.1 Although the introduction of percutaneous coronary intervention (PCI) has decreased the utilization of CABG in the past decade,1, 2 CABG remains the most common cardiac surgery and is the standard of care for patients with either left main or severe 3vessel coronary artery disease. Following CABG, national guidelines strongly recommend cardiac rehabilitation (CR) for all patients.3, 4 Although a survival advantage has been well demonstrated with CR in patients with myocardial infarction (MI),5 and PCI,6, 7 surprisingly, only a small handful of studies have previously examined the mortality impact of CR following CABG. These studies ess w were e e ei er eith either ther th e er small mall and not statistically significant,8, 9 involved only older patients,10 utilized billing data,10 or di id no nott ac cco couunt fo forr pparticipation articipation bias.11 In additi addition, ion o , to our knowled knowledge, e gee, no randomized controlled ed did account trials riaals ls have ever er specifically speeciifi f caall llyy tested tested tes sted d CR CR following foollow wing CABG. win CABG. G Lastly, G. Lasstly stly,, based basedd upon base uponn the the results res esul ultts ul ts ooff one one recent patients with MI, ece cent nt controversial con ontr trov tr over erssiaal al ttrial riall iin rial n pa atiien ents ts w ith a re ith rrecent ece ceent n M I,1122 new new doubts dou o bt b s have h ve ha ve arisen ariise s n regarding reega gard rdiingg the the effectivenesss of CR CR inn the the h era era of of modern mode mo dern rn medical med ediica c l therapy ther th e ap er py fo forr co coro r na ro nary ry aartery rter rt eryy di er dise seas se ase. as e e. coronary disease. Consequently, the aim of this study was to determine the influence of CR participation on all-cause mortality following CABG in a contemporary, community-based, mixed-age cohort using detailed patient level data. Methods We utilized the database of the Division of Cardiovascular Surgery, Mayo Clinic, Rochester, MN to identify consecutive residents of Olmsted County who underwent CABG from January 1996 to December 2007 and were discharged alive. We excluded patients who were non-Olmsted 3 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 county residents, underwent a combined procedure, lacked a valid research consent (per Minnesota law), or were discharged to a long-term care facility. As Mayo Clinic is the only center in Olmsted County performing CABG and also has the only CR program in the county, this study closely approximates a community-based study.13 This study was approved by the Mayo Clinic Institutional Review Board. Standard definitions for risk factors and comorbidities were used according to the Society for Thoracic Surgery database.14 The Charlson index was calculated from 16 clinical variables known to be predictive of mortality.15 Renal failure was defined as a baseline creatinine 2.0 mg/dL, and renal complication as a creatinine 2.0 mg/dL with a doubling from baseline or need for temporary or permanent dialysis. Arrhythmia was defined as the presence of, f, orr pr prio prior io or treatment reatment for, atrial fibrillation, atrial flutter, ventricular fibrillation, ventricular tachycardia, or 3rd ddegree egrree eg ree he hea heart art bl blo block. oc Neurologic complicationn w ock. was as defined by a new w stro sstroke, tr ke, transient ischemic attack, at ttaack, or postpost-operative t-op op perat ativ i e de iv deli delirium liri li riuum ri um w with ith it h hhallucinations. alluuciinatiionns. Fu Func Functional ncctiion onaal al sstatus tatu ta tuss was was a co comb combination mbin mb inaatio in io on of New York Heart Association class Canadian Cardiovascular Society graded Ne ew Yo Y rk H eart ea rtt A ssoociiati ss tion n cla lass la ss aand nd dC anad an adia ad iann C ia ard dio ova vasc scul ular ul arr S ocie oc ieety y cclass, laasss, gr grad ad ded d oon n a I to to IV V scale. This iss database dat atab abas ab asee did as d d not di no have havee CR CR referral refe re ferr fe rral rr a iinformation al n or nf o ma mati tion ti on ffor or tthe hee m majority ajor aj orit or ityy of ppatients, it atie at ieent n s, nor did d it have sufficiently available and reliable morbidity events (subsequent MI, PCI, or stroke) to allow a proper analysis of these outcomes. In addition, we obtained socioeconomic factors known to affect CR attendance from the general medical record. Distance from home to CR was determined using an online web-based mileage calculator. Insurance status was classified in four groups: Medicare, Medicaid, commercial, or self-pay/uninsured. Medical “connectedness” was estimated by tabulating all outpatient physician visits in the year prior to CABG. These factors were used as important baseline socioeconomic variables in all analyses. 4 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Patients were considered to have participated in CR if they attended at least 1 outpatient session within 6 months of the index CABG surgery, as previously done.10 Attendance was ascertained by use of the Mayo Clinic CR database. All non-participants’ medical records were checked to assure they had not attended CR within 6 months following CABG surgery. All patients attending CR underwent conventional exercise training and lifestyle counseling according to American Association Cardiovascular and Pulmonary Rehabilitation guidelines. Most patients underwent a baseline 6 minute walk as part of an exercise prescription. Patients typically exercised for 30-45 minutes 3 times per week but were also encouraged to exercise for 30 minutes per day on days when not attending CR. This program also included evaluating and managing any active symptoms, arranging follow-up appointments appointmen nts and andd laboratory lab abor orat or atory esting, and managing risk factors such as hypertension, hyperlipidemia, and tobacco dependance testing, n ccoordination oord oo rdin rd inat in atio at ion wi io with t the patient’s primary care pphysician th hysician hys in Follow win ng CABG CA ABG G surgery, sur urge gery ry,, all all patients pattien pa ntss were werre encouraged en nco cour urag aged ag ed to to participate part pa rttic i ip ipat atte in a llong-term, ong-teerm ongm, Following nurs nu rsee ad rs aadministered, miini nist ster erred d, di iseeas asee-m mana mana nage geme ge meent pprogram. rogr ro gram gr am.166 P am Participating arrti ticcipa cipaati t ng ng ppatients a ients at nts were were r m more or llikely ore ikkely kely y too be b nurse disease-management CR participa ant ntss (d (due ue tto o en enco cour co u ag agem emen em e t fr en rom C R st taf aff) f , bu butt th this is w ass nnot ot uuniformly nifo ni form rmly rm ly tthe h case and he participants encouragement from CR staff), was was not recorded. In this program, follow-up typically occurred at 6, 9, 12, 18 and 24 months after surgery, and annually thereafter. Consequently, to evaluate the adherence patterns of our patients regarding healthcare follow-up after CABG, we tabulated all outpatient physician visits occurring between 6 and 24 months of follow-up. Similarly, we obtained low-density lipoprotein (LDL) cholesterol levels taken between 6 and 18 months of follow-up, and considered any LDL level <100 mg/dL controlled. The primary outcome was the 10-year incidence of all-cause mortality, which was ascertained using a two-stage approach. First, vital status was extracted from the Mayo 5 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 registration database, in partnership with the Minnesota State Death Tapes. For anyone not indicated as deceased according to Mayo records, death information was pulled from the Accurint® system.17 Patients with no record of death from either source were ultimately censored (indicated as alive) on October 25th, 2011, one month prior to the Accurint pull. Statistical Methods Cox proportional hazards (PH) regression analysis was used to test the association between CR attendance and long-term mortality. An individual was considered a CR participant if they attended at least one session within 6 months of surgery. We used a landmark approach such that subjects who died or were lost to follow-up within 6 months were excluded. Conditioning on 6month survival allowed all patients in the analysis to have the same opportunity to to aattend tten endd CR en CR,, and allowed testing of CR participation as a “baseline” factor in the Cox PH modeling. 18 To ccontrol ontrrol ffor or CR participation bias, propensity prop op pen nsity score methods met etho h ds18 were were employed. Using logistic ogiist s ic regress regression, ssio ionn, w io wee fi fitt a no non non-parsimonious n-paarssim n-pa moniouus us mu multivariable ultivarria riable ablee m model oddel ((including incl in c udin udin ng an anyy si sign significant gniffic ican an nt 22-way wa ay interactions) inte in t ra te ract ctio ct ions io ns)) to to assess asseess ess the th he influence infl in flue fl uenc ncce of all alll available avaaillab ble baseline basel asellin inee factors fac acto torrs on on the t e likelihood th like li kelliho ke hooodd ooff being a CR pparticipant. arti ar t ciipa ti pant ntt. Th The he pr ppropensity open op e si en s ty y sscore core co re iiss a function fun unct c io on of the the predicted pre reddic icte tedd pr te pprobability obab ob abil ab illit ityy of CR participation from this model. We controlled for the propensity score using three different approaches: 1) regression adjustment treating the propensity score as a covariate, 2) stratification by propensity score levels, and 3) caliper matching on propensity scores. All available variables were included in the propensity model with the exceptions of postoperative MI, resuscitation, shock, and postoperative angiogram. These factors were excluded due to low event rates, concerns about model convergence, and missing data. For the stratification approach, we examined the distribution of propensity scores from both groups and “trimmed” the non-overlapping tail ends, reducing the sample to those within a 6 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 common range of scores. The retained subjects were then divided into 5 equally-sized strata using propensity score quintile values. The influence of CR participation on mortality was assessed within each stratum using Kaplan-Meier methods and Cox PH regression modeling, as well as in the combined group via stratified Cox PH regression. To assess the effectiveness of the stratified propensity approach, CR participants were compared to non-participants within the five strata for each significant factor in the overall propensity model. To test the effect of the trimming, we also performed a sensitivity analysis including all patients. For the propensity matching approach, CR participants were individually matched to nonparticipants within the same caliper (width based on 0.10 of the standard deviation of propensity score) individuals core) and closest in terms of propensity score. This analysis included only indivi viiduualls wi with th a suitable uitable match and due to a relative shortage off non-participants in the overall cohort, the majority (362 of 406, or 89%) participants. majo ma jori jo rity ri ty of of those thosse unmatched th un 89% %) were %) were CR parti ici c pant ntss. nt s. For secondary seco cond ndarry analyses, nd an nal alys yses ys ess, logistic lo ogi gist s ic regression regrresssionn was wass used used to use to test test for for a linear lin inea earr trend ea tren tr endd in CR en CR participation rates over study period. The year surgery independent variable part pa rtic rt iccip ipat atio io on ra rate tees ov ove er tthe he stu tudy tu dy ttime imee pe im peri riod ri od. Th od he ye earr ooff su urg r ery ery wa was the the in inde d pe de pennden nden nt va var riab riab ble l and a patientt wa considered attended session wass co cons nssid derred tto o ha have ve aattended tten tt en nde dedd CR iiff tthey heey at atte tend te nded nd e ffor ed or a ssingle ingl in g e se gl ess ssio ionn anytime in io n the subsequent 6 months. Poisson regression was used to test for a difference in the annualized rate of outpatient physician follow-up visits between participants and non-participants. We also tested for group difference in the proportions with an LDL level drawn and controlled at 1-year follow-up using a Chi square test. A p value < 0.05 was considered statistically significant. All analyses were carried out using the SAS statistical software package (Version 9.2, SAS Institute Inc., Cary, NC). Results We identified 869 consecutive Olmsted county residents consent who underwent isolated CABG 7 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 during the study period. A total of 23 (2.6%) patients were excluded; 13 of whom died in the first 6 months (only 3 attended CR,) 7 in which the vital status could not be verified, and 3 due to unknown CR status. Consequently, a total of 846 subjects were used in all CR-related analyses. Baseline characteristics, socioeconomic factors, operative, postoperative, and surgical complications are reported in Table 1. During a mean follow-up of 9.0 ± 3.7 years, the 10-year all-cause mortality rate was 28% (193 deaths). The unadjusted 10-year cumulative incidence of all-cause mortality for the non-CR and CR groups were 45% (100 deaths) and 20% (93 deaths) respectively, p < 0.001. The majority of patients began CR within 1 month after hospital discharge (median [Q1, months. Median Q3] time to start CR was 10 [5-17] days) and all patients started CR within 6 mo onths nths h. M edia ed iann ia time ime (Q1, Q3) from first to last CR session was 55 (42 to 69) days. Overall attendance during the time was ime me frame fra rame me w as 669% 9% (annual rates ranged from 55% 9% 55% to 76%) andd the t e median th meddian number of CR me sessions patient There was trend esssio i ns (Q1, Q3 Q3) pe perr pa ati tien entt was en was 14 4 ((9 9 tto o 119.) 9..) T here wa w ass no ssignificant ig gnifi fica caant llinear in neaar tr tren e d fo en forr ch cchange haang a ng e in the study See S1 n CR CR attendance a te at tend ndan nd ance cee oover verr th ve he st tud u y ti time m fframe me raame by y uunivariate nivvari variiatte anal aanalysis nal alys ysis ys is ((p p = 0.79 00.79). .79 9). ) S eee Figure Figu Fi gu uree S 1 (online supplement.) online supp pleeme m nt.)) Table 2 lists the baseline variables and their effects on CR participation from a multivariable logistic regression model, from which the predicted probabilities were used to derive propensity scores. Specifically, smoking, renal failure, lack of insurance, and having a peri-operative neurologic complication were significantly associated with not attending CR (p<0.05 for each). Table 3 summarizes the differences across propensity quintiles for the prediction of attendance at CR. As seen, none of these factors differed significantly between the two groups in any of the five quintiles, suggesting adequate group balance to perform within-strata 8 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 comparisons of survival or to estimate an overall stratified effect of CR on survival. The associations between CR attendance and mortality are summarized in Table 4 based on unadjusted, propensity-adjusted, propensity-stratified, and propensity-matched analyses. Using the propensity score method of covariate adjustment, CR was associated with a 46% reduction in hazard of 10-year mortality (HR=0.54; 95% CI, 0.40-0.74; p<0.001), conditional on surviving the first 6 months. Both the stratified and matched propensity score methods produced a nearly identical effect for CR attendance as reflected by a 45% reduction in hazard of mortality (stratified: HR=0.55; 95% CI, 0.40-0.75; p<0.001; matched: HR=0.55; 95% CI, 0.36-0.84; p = 0.007.) Strata-specific effects of CR on long-term mortality are also reported in Table 4, and illustrated llustrated in Figure 1, for each of the five propensity score quintiles. Based on a 10-year 100-y yea earr adjusted mortality rate of 35.7% in non-participants and 23.0% in participants, the absolute risk reduction edu uct ctio ionn was io w s 12.7% wa 12 2.7 .7% 7% and the number needed to treat tre r at to prevent one onne death deat de athh over 10 years was 8. at Among associated with Amon Am o g those who who attended atte at teendded CR, CR, R, the the number num umberr of of sessions seess ssiionns ns aattended tten tt ende en dedd was wa no nnot ott as ssooci ciat atted w i h long it llongong ngterm trimmed erm m mortality, mor orta taali lity ty,, after ty afteer adjustment afte adju adju ust stme meent for for age, agee, gender gennder ge nder and and Charlson Chaarl rlsson son Index I de In dexx (p=0.41). (p p=0 0.4 41) 1). Wh When en tr rimme rim mmed patients (34 ttotal, were otal ot a , 4% al 4%)) we w re iincluded n lu nc lude dedd as ppart de artt of ar of tthe he ssensitivity e siiti en t vi v ty analysis, ana n ly ysi sis, s results s, res esul u tss were ul wer eree nearly n arly ne identical (HR, 0.54; 95% CI, 0.40 – 0.74; p<.001).” There was no significant differential effect of CR on mortality when tested by age ( or <65), p = 0. 87; gender, p = 0.86; prior MI, p = 0.74; or diabetes, p = 0.61. Compared to non-attendees, CR attendees were significantly more likely to have had an LDL level drawn at around 1-year following CABG, (83% vs. 52%, p <0.001) and to have that LDL <100 mg/dL (72% vs. 60%, p = 0.006). Furthermore, the annualized rate of total outpatient clinic visits in the two years after CABG was higher on average among CR attendees, (rate = 7.8 visits per person per year; 95% CI, 7.6 to 7.9) than non-attendees (rate = 6.2 visits per person per 9 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 year; 95% CI, 5.9 to 6.4; p<0.001). Discussion We found in our large contemporary, community-based, mixed age, post-CABG cohort that CR participation was significantly associated with an approximate 45% reduction in all-cause mortality. This finding was demonstrated using three different techniques to adjust for the propensity to attend cardiac rehabilitation, and was not significantly different by sex, age /< 65 years, presence/absence of diabetes or prior MI. The number needed to treat with CR to prevent one death following CABG at ten years was 8. These findings support national guidelines and coverage policies that recommend CR participation after CABG.3, 4 Our study adds to the small number of published studies that have assessed the relationship elaati tion onsh on ship sh ip bbetween etwe ween we en CR participation and mortal mortality aliityy following CA al CABG. ABG G. S Suaya u ya et al.10 ua ddemonstrated dem emonstrated mo a 55-year, -y yea earr, 228% 8% rrelative elat el atiive ri risk iskk red reduction ductio duc on uusing sin in ng a la larg large rgee Medi M Medicare edi diccar care re cclaims laim la imss da data database, taba basse, ba se, bu but this Hansen his study stu tudy dy was was as limited lim mited ited bby y lack lack k of of detailed dettail de tailled d ppatient atieent n inf iinformation nform rmat atio at ionn an io aand d an eelderly ldeerly erly y ccohort. o orrt.. H oh anse anse s n ett al.11 demonst demonstrated stra rate ra t d a 2te 2-ye 2-year, y arr, 87 ye 887% % re rrelative latiive la v rreduction educ ed ucti uc t on iin ti nm mortality orta or tali ta l ty rrisk li i k us is usin using ingg pa in ppatient tiien entt le level eve v l data, but participation bias was not accounted for. Hedback et al.8 found a 54% reduction in 10-year allcause mortality following CABG but this did not meet statistical significance likely due to small sample size, (n = 147, p = 0.06.) To our knowledge, there are no randomized controlled trials that have specifically tested CR in CABG, although these patients have generally been included in general post-MI trials. Among these trials, meta-analyses have demonstrated a 20-25% reduction in all-cause mortality seen in patients with either MI or PCI.19-21 Unfortunately, most randomized trials of CR are now more than 20 years old.21 In addition, one recent controversial trial12 suggests that CR following 10 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 an MI, may not be effective in the era of modern medical therapy for coronary artery disease. Furthermore, it appears unlikely that additional randomized trials of CR will be completed given the problematic ethical considerations of withholding this guideline-endorsed therapy. Consequently, carefully controlled cohort studies (such as this one) have an important role in evaluating the modern effects of comprehensive cardiac rehabilitation. When compared to the above studies, our results showed a smaller impact of CR on allcause mortality risk than was noted by Hansen et al.,11 similar to the results of Hedback et al.,8 larger than the results of the Suaya et al.,10 and larger than the meta-analyses of randomized controlled trials performed in mixed coronary artery disease populations.19-21 Furthermore, our findings are consistent with previously published cohort studies from Olmsted County, Count ount n y, y, Minnesota and Calgary, Alberta, Canada that found that CR participation was significantly as associated ssooci ciat ated at ed w with it a 40-50% ith 40 reduction in mortalityy risk r sk following MI, ri MI, I 13 P PCI, CI 6 or a new diagnosis of CI, coronary co oro ona n ry arter artery ry di ddisease. seas asse. e 22 Wh What, hat at,, then, then th en,, might en mighht explain expl ex p ai pl ainn our our findings? find ndin ingss? Two in Two possible pos ossi sibl si blee explanations ex xpllan anaatio at ons aare ree w worth ortth or consideration. n.. Fi F First, rst, rs t, w wee co contend ont nten e d that en t at th a our our u findings fin ndi ding ngss re ng refl reflect f ecct a true fl trrue aassociation ssoc ss ocia oc iati ia t on bbetween ti ettwe ween en C CR R and substantial reduced mortality risk, related to the direct effects of CR on the short- and long-term delivery of effective secondary prevention therapies in CABG patients. Previously published studies support this concept, showing that CR participation is associated with improvements in coronary heart disease risk factor control20 as well as long-term follow-up and adherence to secondary prevention medications.16, 23 We confirmed these concepts in our study by demonstrating improved rates of LDL measurement, LDL control, and outpatient follow-up in the first two years after CABG. We maintain that these results are a direct consequence of participation in a comprehensive CR program and one of several possible explanations for why 11 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 CR participants have improved outcomes. In addition, given that patients undergoing CABG typically have more severe coronary artery disease and risk factors, they likely derive a greater benefit from CR than patients with more limited coronary artery disease. A second possible explanation for our findings is the presence of a “healthy cohort” bias, such that CR participants are healthier and more motivated to maintain optimal health than are those who do not participate in CR. Certainly, our data show that CR participants were generally younger and healthier than those who did not participate in CR. To help address this concern, we utilized multiple propensity score techniques. Though the final result remained statistically significant, this adjustment substantially attenuated the effect of CR, likely reflecting the removal variable of such bias. Furthermore, we also assessed the potential role for instrumental va ari riab ablee aanalysis ab naly na lysi ly sis (i.e., i.e., adjusting for factors that are associated with “accessibility” of CR but that are not necessarily ne ece cess sssaril aril ilyy related reelate teed to t mortality risk), which in similar sim imiilar studies hass further im furth th her reduced reduced the point estimate. esti imate.10 Wee fo fou found, und, d hhowever, d, owev ow eveer, ev er, th that hatt tthe he inst he instrumental truumen ntal vvariables arria iabbles es w wee co coul could u d id ul identify den enttify tify ((driving driv dr ivin i g in distance dist di stan st ance an c to ce to CR aand ndd m medical edic edic i al a ““connectedness”) conn co nneected nn cted dne ness ss”) ss ”) ddid id d not not differ difffe ferr significantly siign gniifiica cant nttly y bbetween etwe et ween we en n tthose hosse se w who hoo participated iin n CR aand nd tthose hose ho se w who ho ddid id nnot, o , li ot likely ike kely ly y ddue ue to t tthe he sshort hort ho rt ddistance issta tanc ncee (< nc (<10 100 m miles) i es il es)) that most patients needed to travel to attend CR. Consequently we were unable to perform the planned instrumental analysis. Lastly, and most importantly, we found no differences in the key predictor variables for CR attendance when comparing our propensity quintiles across study groups as demonstrated in Table 3. In summary, even after adjusting for identifiable factors that were associated with a possible “healthy cohort” bias and “accessibility” of CR, we found a persistent significant association between CR participation and reduced all-cause mortality risk. We did not find an association between the numbers of CR sessions attended and reduced mortality in our cohort, which differs from recent findings suggesting an approximate 1% 12 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 reduction in mortality for each session of CR attended.22, 24 However, we believe this is primarily due to the fact that during the study period, Mayo Clinic CR individualized the recommended number of CR sessions per patient depending on progress towards goals. Healthier patients were allowed to graduate earlier, while more frail and sick patients were encouraged to complete 36 sessions. Therefore, a higher number of CR sessions attended may in part reflect poorer underlying health. Our CR participation rates were high (68%) and compare favorably to national participation rates of 20-30%.10 This implies that, nationally, higher participation rates are clearly achievable and that many more patients would attend CR if given the opportunity. In addition, our participation rates compared favorably to previously reported ratess from from fr o Olmsted om Olm lmst sted st ed County, Minnesota during a similar time period in persons recuperating from MI or PCI (55% an and nd 40 40% % re resp respectively). specti sp tiively vel ).6,13 This higher relative pa participation artiicipation for pa patients ati t en nts uundergoing ndergoing CABG when compared co ompared mp to MI orr PCI PC is likely lik ikel elyy due el due too increased increeassed disease diseaasee severity sever erit ityy and and the t e need th neeed for forr surgical sur u gica call revascularization, previously eva vasc sccul u ar ariz izat iz atio ionn, aand nd d hhas a pre as reevi viou ousl ou slyy been sl been nnoted. otted e . 100 Ho However, owe wevver, ver, iitt re rema remains main ma in ns un uncl unclear c ea cl earr wh whyy 32% 32% ooff not participate, par arti tici ti ciipa p tee, and and more m re importantly, mo imp mpor orta or tant ta ntly nt ly,, iff this ly thiis participation part pa rtic rt iccip ipat atio ionn rate io rate could cou uld d be be improved i proved im patients did not through additional quality improvement projects. Limitations Our study included data from only one county in Minnesota, and, therefore, may have limited generalizability. Subgroup analyses (such as those for individual propensity strata) may have been underpowered due to smaller sample sizes. Referral to CR was unknown and thus our results may reflect some element of referral bias. Our database did not include a number of behavioral factors and attitudes that might affect attendance at CR or adherence to other recommended therapies. Our follow-up system with nurse case management may have 13 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 favorably biased our results, but we note that the GOSPEL study tested a 3 year program of intensive followup after CR and found no difference in either cardiovascular or all-cause mortality.25 Our use of landmark analysis limits the generalizability of our results to patients that survive the first 6-months following CABG. Lastly, because our data were observational in nature, it is possible that our results may have been affected by bias in our population sample (e.g., “healthy cohort” bias). Despite our best efforts to adjust for bias in our population, it is possible that unidentified sources of potential bias still exist in our data. Our results, therefore, may overestimate the true association between CR participation and mortality risk following CABG. However, even if as much as half of our effect size was due to undetected residual bias, here would still be a substantial and important protective association between C R at aattendance teend ndan ance an c ce there CR and subsequent mortality following CABG. C on nc nclusions Conclusions We ffound ound ou nd iin n a co community-based, ommu mmuni unityity-ba baase sedd, m mixed ixed ix ed aage ge ccohort o ortt that oh thhat hat CR pparticipation a tici ar ticiipa pati tiionn aafter fter ft err C CABG ABG AB G wa was as associated w itth a si ssignificant gnif gn ific if i an nt re rreduction duct du c io ct ion in llong-term ongon g te gt rm m orrta tali lity li ty.. Ou ty Our re resu sult su ltss su lt supp pppor ortt th thee recently with mortality. results support released clinical practice guidelines that strongly recommend CR for all patients following CABG surgery. Acknowledgments: We would like to thank Judy Lenoch and Laurie Barr, BS for assisting with data collection and management. Funding Sources: Funding for statistical analysis was provided by the Mayo Clinic’s Department of Cardiovascular Disease. No other funding supported this work, including from design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. 14 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Conflict of Interest Disclosures: None. References: 1. Epstein AJ, Polsky D, Yang F, Yang L, Groeneveld PW. Coronary revascularization trends in the united states, 2001-2008. JAMA. 2011;305:1769-1776. 2. Hockenberry J, Lu X, Vaughan-Sarrazin MS, Snyder S, Peterson E, Cram P. Shifts in surgical revascularization and valve procedures among medicare beneficiaries. Med Care. 2011;49:686692. 3. Smith SC, Jr., Benjamin EJ, Bonow RO, Braun LT, Creager MA, Franklin BA, Gibbons RJ, Grundy SM, Hiratzka LF, Jones DW, Lloyd-Jones DM, Minissian M, Mosca L, Peterson ED, Sacco RL, Spertus J, Stein JH, Taubert KA. Aha/accf secondary prevention and risk reduction therapy for patients with coronary and other atherosclerotic vascular disease: 2011 update a guideline from the american heart association and american college of cardiology foundation endorsed by the world heart federation and the preventive cardiovascularr nurses association. asso as soci so ciat atio ion. n. J Am Coll Cardiol. 2011;58:2432-2446. 4. Hillis LD, Smith PK, Anderson JL, Bittl JA, Bridges CR, Byrne JG, Cigarroa JE, Disesa VJ, Hiratzka LF,, Hutter AM, Jr., Jessen ME, Keeley y EC, Lahey SJ, Langee RA, London MJ, Mack MJ,, Patel MJ Pate Pa tell MR, te MR Puskas Puuskas sk JD, Sabik JF, Selnes O, Shahian Shahhian DM, Trost stt JC, C, Winniford Win i niford MD. 2011 accf/aha guideline artery graft Executive ac ccff/aha /ah gui uiddeli line ne ffor or ccoronary orron onar aryy ar arte t ry te y bbypass y as yp asss gr graf afft ssurgery: urg rg ger ery: y: E xeecu cuti tive ve ssummary u ma um mary ry a rreport epor ep orrt off tthe h he american college practice am merican er colleegee of of cardiology caard dio iolo logy lo gy foundation/american fou ounnda ndatiion/am americcan am can heart heeart art association asso soci ciaatio ionn task task fforce orrce oon n pr prac a ti ac tice ce guidelines collaboration with american for gu uid del e ines ddeveloped evelop ev oped iin op n coll lllab borrat a ion o w ith tthe he am meric ican an aassociation ssoociaatioon fo ss or thoracic thoora raci cicc surgery, suurg gerry, society soccieety off cardiovascular carrdi d ov ovas asccula as cularr anesthesiologists, anessth anes hes esio olo l gi gist sts, st s, and and nd society soc ociiety ty of of thoracic thorac tho oraccic surgeons. sur urge ur geeon ns. s J Am Am Coll Col olll Cardiol. Carrdio Ca io ol.. 2011;58:2584-2614. 2011 11;5 ;588:25 2584 8 -2 -2614. 4. Taylor RS. 55. Heran Her eran an BS, BS Chen Che henn JM, JM Ebrahim Ebra Eb rahi him m S, S Moxham Mox oxha ham m T, T Oldridge Oldr drid idge ge N, N Rees Rees K, K Thompson Thom Th omps pson on DR, DR T aylo ay lorr RS Exercise-based cardiac rehabilitation for coronary heart disease. Cochrane Database Syst Rev. 2011:CD001800. 6. Goel K, Lennon RJ, Tilbury RT, Squires RW, Thomas RJ. Impact of cardiac rehabilitation on mortality and cardiovascular events after percutaneous coronary intervention in the community. Circulation. 2011;123:2344-2352. 7. Dendale P, Berger J, Hansen D, Vaes J, Benit E, Weymans M. Cardiac rehabilitation reduces the rate of major adverse cardiac events after percutaneous coronary intervention. Eur J Cardiovasc Nurs. 2005;4:113-116. 8. Hedback B, Perk J, Hornblad M, Ohlsson U. Cardiac rehabilitation after coronary artery bypass surgery: 10-year results on mortality, morbidity and readmissions to hospital. J Cardiovasc Risk. 2001;8:153-158. 9. Vanhees L, Fagard R, Thijs L, Amery A. Prognostic value of training-induced change in peak 15 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 exercise capacity in patients with myocardial infarcts and patients with coronary bypass surgery. Am J Cardiol. 1995;76:1014-1019. 10. Suaya JA, Stason WB, Ades PA, Normand SL, Shepard DS. Cardiac rehabilitation and survival in older coronary patients. J Am Coll Cardiol. 2009;54:25-33. 11. Hansen D, Dendale P, Leenders M, Berger J, Raskin A, Vaes J, Meeusen R. Reduction of cardiovascular event rate: Different effects of cardiac rehabilitation in cabg and pci patients. Acta Cardiol. 2009;64:639-644. 12. West RR, Jones DA, Henderson AH. Rehabilitation after myocardial infarction trial (RAMIT): Multi-centre randomised controlled trial of comprehensive cardiac rehabilitation in patients following acute myocardial infarction. Heart. 2012;98:637-644. 13. Witt BJ, Jacobsen SJ, Weston SA, Killian JM, Meverden RA, Allison TG, Reeder GS, Roger VL. Cardiac rehabilitation after myocardial infarction in the community. J Am Coll Cardiol. 2004;44:988-996. 14. The Society of Thoracic Surgeons Adult Cardiac Surgery Database. 2011; http://www.sts.org/national-database. Accessed March 29, 2012. 15. Charlson C arlson ME,, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic Ch comorbidity co omo morb rbid rb idit id ity it y in in longitudinal lon onggitudinal studies: Development Developmeent nt aand nd validation. n J Chronic Chhro roni n c Dis.1987;40:373383. 3883.. 16. Squires Montero-Gomez Allison TG, Thomas RJ. Long-term management 166. S quires RW RW,, M onnterro-G Gom mez A, mez A A llisonn T G,, T hom mas as R J. L onng-teerm m ddisease issea ease m anag an ageemen entt of en patients with coronary cardiac program staff. Cardiopulm Rehabil pati pa tien ti en nts t w ithh co it cor ronnary nary y ddisease isea is easse se bby y ca car rdiiacc rrehabilitation ehab ehab abil iliita itatio on pr pro ogra ogra ram m st taf aff. f. J Ca Card r io rd iopu pulm pu lm mR eh hab abiil Prev. Prev ev. 2008;28:180-186; 2008 20 08;2 ; 8: 8:18 1 00-18 186; quiz qui u z 187-188. 18718 7-18 1 8. 17. LexisNexis. LexisNexis Accurint. Accessed March 19th, 17 Le Lexi xisN sNex exis is Le Lexi xisN sNex exis is A ccur cc urin intt 22012; 012; 01 2; hhttp://www.accurint.com/. ttp: tt p:// //ww www w ac accu curi rint nt com om// A cces cc esse sedd Ma Marc rchh 19 19th th 2012. 18. D'Agostino RB, Jr. Propensity scores in cardiovascular research. Circulation. 2007;115:2340-2343. 19. Taylor RS, Brown A, Ebrahim S, Jolliffe J, Noorani H, Rees K, Skidmore B, Stone JA, Thompson DR, Oldridge N. Exercise-based rehabilitation for patients with coronary heart disease: Systematic review and meta-analysis of randomized controlled trials. Am J Med. 2004;116:682-692. 20. Lawler PR, Filion KB, Eisenberg MJ. Efficacy of exercise-based cardiac rehabilitation postmyocardial infarction: A systematic review and meta-analysis of randomized controlled trials. Am Heart J. 2011;162:571-584 e572. 21. Oldridge NB, Guyatt GH, Fischer ME, Rimm AA. Cardiac rehabilitation after myocardial infarction. Combined experience of randomized clinical trials. JAMA. 1988;260:945-950. 16 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 22. Martin BJ, Hauer T, Arena R, Austford LD, Galbraith PD, Lewin AM, Knudtson M, Ghali WA, Stone JA, Aggarwal S. Cardiac rehabilitation attendance and outcomes in coronary artery disease patients. Circulation. 2012;126:677-687. 23. Shah ND, Dunlay SM, Ting HH, Montori VM, Thomas RJ, Wagie AE, Roger VL. Longterm medication adherence after myocardial infarction: Experience of a community. Am J Med. 2009;122:961 e967-913. 24. Hammill BG, Curtis LH, Schulman KA, Whellan DJ. Relationship between cardiac rehabilitation and long-term risks of death and myocardial infarction among elderly medicare beneficiaries. Circulation. 2010;121:63-70. 25. Giannuzzi P, Temporelli PL, Marchioli R, Maggioni AP, Balestroni G, Ceci V, Chieffo C, Gattone M, Griffo R, Schweiger C, Tavazzi L, Urbinati S, Valagussa F, Vanuzzo D. Global secondary prevention strategies to limit event recurrence after myocardial infarction: Results of the gospel study, a multicenter, randomized controlled trial from the italian cardiac rehabilitation network. Arch Intern Med. 2008;168:2194-2204. 17 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Table 1. Baseline Characteristics Patient and Clinical Characteristics No CR (n=264) Age (yrs.) 68.3±11.0 Male Gender 192 (73%) Caucasian Race 245 / 254 (96%) Surgery Year^ 2001 (1997, 2004) Any Prior Cardiac Operations 9 / 180 (5%) Total # of Diseased Coronary Arteries 0 (0%) x 0 13 (5%) x 1 37 (14%) x 2 214 (81%) x 3 Left Main Disease >50% 82 (31%) Charlson Index (ordinal categories) 42 (16%) x 0 93 (35%) x 1-2 64 (24%) x 3-4 65 (25%) x 5 Ejection Fraction (%) 53.3±13.5 Functional Status Classification 11 (4%) x 1 2255 (9%) x 2 116 11 (44% (44%) %) x 3 112 11 12 (42% (42%) 2%)) 2% x 4 Preoperative P Preo r ope p rativee R Risk iskk F Factors acctors Body Bo ody Mass Mas asss Index Inde In dexx (BMI) (B BMI) MI) kg kg/m m2 229.4±5.6 9.44±5 5.66 BMI BM MI ca cate categories tego gorriess 5 ((19%) 51 1 %) 19 x <25 1099 (4 10 (41% (41%) 1%)) x 25 - <30 <30 62 (23%) x 30 - <35 42 (16%) x 35 Family History of Coronary Artery Disease 110 / 258 (43%) Diabetes 95 (36%) Hypercholesterolemia 223 (85%) Smoking 78 (30%) x Never 50 (19%) x Current 136 (52%) x Past Renal Failure 19 (7%) Hypertension 203 (77%) Peripheral Arterial Disease 62 (23%) Cerebrovascular Disease 47 (18%) Prior Cardiovascular Intervention 100 (38%) Prior Myocardial Infarction 137 (52%) Prior Congestive Heart Failure 33 (13%) Angina 232 (88%) CR (n=582) 64.4±10.3 456 (78%) 550 (95%) 2001 (1998, 2004) 5 / 415 (1%) 1 (<1%) 24 (4%) 119 (20%) 438 (75%) 179 (31%) 132 (23%) 235 (40%) 124 (21%) 91 (16%) 56.0±13.3 p-value <0.001 0.07 0.46 0.71 0.005 0.08 0.94 <0.001 00.010 .01 0100 0.11 25 (4%) (4%) 95 5 ((16%) 16%) 2333 (4 23 ((40%) 0% %) 229 22 29 ((39%) 39% 9% %) 29.5±5.2 29 9.55±5 ±5.2 .2 108 10 08 (1 (19% (19%) 9%)) 9% 2299 (3 22 (39% (39%) 9%)) 160 (27%) 85 (15%) 264 / 564 (47%) 157 (27%) 516 (89%) 200 (34%) 82 (14%) 300 (52%) 16 (3%) 417 (72%) 71 (12%) 81 (14%) 194 (33%) 257 (44%) 29 (5%) 528 (91%) 18 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 0.744 0.74 0.67 0. 67 0.26 0.008 0.10 0.13 0.003 0.11 <0.001 0.14 0.20 0.037 <0.001 0.20 DOI: 10.1161/CIRCULATIONAHA.112.001365 Cardiogenic Shock Resuscitation Arrhythmia Preoperative Beta Blocker Use Social Factors Type of insurance: . Medicare . Medicaid . Commercial/Mayo . Self-Pay/Uninsured Distance to CR facility (miles) ^ Distance to CR facility >10 miles # Visits in 1-year prior to CABG ^ Operative Characteristics Intraaortic Balloon Pump Used Cross Clamp Time (minutes) Elective Status Internal Mammary Artery Used as Graft Post-Operative Complications Length of hospitalization stay 1 week Post-operative Angiogram Reoperation for Bleeding Myocardial My yoccardial IInfarction n arrct nf ctio ion Neurologic Neur Ne urol ur olo ol ogic C ogic Complication ompl ompl pliicat ica ion Renal Re enaal Co C Complication mp pli lica cati tion on 6 (2%) 0 (0%) 49 (19%) 209 (79%) 12 (2%) 2 (<1%) 71 (12%) 491 (84%) 0.84 0.34 0.014 0.06 160 (61%) 13 (5%) 76 (29%) 15 (6%) 3.3 (1.8, 9.9) 65 (25%) 4 (1, 8) 275 (47%) 24 (4%) 279 (48%) 4 (1%) 3.6 (2.3, 7.1) 118 (20%) 5 (2, 8) 0.60 0.15 0.035 26 (10%) 45.6±21.3 179 (68%) 245 (93%) 26 (4%) 48.7±20.5 416 (71%) 561 (96%) 0.003 0.048 0.28 0.023 165 (63%) 1 / 211 (0%) 10 (4%) 0 (0 ((0%) %) 32 ((12%) 32 12%) 110 0 (4%) (4% %) 270 (46%) 11 / 487 (2%) 17 (3%) 2 (< ((<1%) 1%) 28 8 (5%) (5% 5 ) 11 (2%) (2% 2%)) <0.001 <0.0 <0 .0 001 0 00.10 .10 10 0.51 0.34 <0.001 0.10 0. 10 <0.001 ^ me median edi d an (Q1, Q3); Q3); p-value p--val allue ue from fro rom m Wilcoxon Wilc Wi lcox ox xon nR Rank ankk Su an Sum um Testt for with missing data, denominator subjects with N (%) (% %) reported for f r all fo all categorical cateego gorricaal variables; vaariab ble l s; fo or tthose hosee w ith any any mi miss ssin ss ingg da in ata, tthe hee deno ominaato or of o su ub ect ubje ctss w ith observed data compute the percentage ob bse s rv ved d dat a a is rreported epor ep orte tedd aand ndd us uused ed d too co comp mput mp utee th he pe erccen enta tage ge Cardiac Rehabilitation, Coronary Bypass Grafting Surgery, Body CR = C a di ar diac ac R eh hab abil ilit il itat it attio ionn, n, CABG CAB ABG G = Co Coro rona naary Artery Art rter eryy By er Bypa pass pa s G ss raft ra ftin ft ingg Su in S urg rg ger eryy, y, BMI BMI = B odyy Ma od Masss IIndex ndex nd ex 19 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Table 2. Propensity Factors for Attendance at Cardiac Rehabilitation Patient and Clinical Characteristics Age (per 10 years)† Hospitalization 1 week >1 week Male Gender Surgery Year Caucasian Race Any Prior Cardiac Operations Total # of Diseased Coronary Arteries Left main disease >50% Charlson Index (categorized) Ejection Fraction Functional Status Classification Preoperative Risk Factors BMI categories x <25 x 25 - <30 x 30 - <35 x 35 Family History of Coronary Artery Disease Hypercholesterolemia Smoking moking x Never x Current x Past Renal nal Failure Hypertension ypert rten ensi en sion si on Peripheral riph ri pher ph eral er al Art Arterial rter erria ial Di Dise Disease seas ase Cerebrovascular re rovascularr Disease rebr Dissea easse Prior or Cardiovascul C Cardiovascular a lar IIntervention ntterrve vent ntiion ion Prior or M Myocardial yocardial al IInfarction nfaarc arction Congestive ng ges esti tive ti ve Heart Heaart Failure Fai aillur lure re Angina ngiina Arrhythmia rhyt rh ythm yt hmia hm ia Preoperative eoperative Beta Bet etta Bl Bloc Blocker ocke oc kerr Us ke U Used e ed Operative perati tive Characteristics Charactteriisti Ch tics Intraaortic Balloon Pump Used Cross Clamp Time (min) Elective Status Internal Mammary artery used as graft Post-Operative Complications Reoperation for Bleeding Neurologic Complication Renal Complication Social Factors Insurance: x Medicare x Medicaid x Commercial/Mayo x Self-pay/Uninsured Distance of CR center >10 miles #Visits in prior 1-year‡ No diabetes Diabetes Multivariable Result 0.96 (0.69, 1.33) [0.786] 0.58 (0.43, 0.77) [<.001] 1.20 (0.79, 1.81) [0.396] 0.94 (0.88, 1.01) [0.094] 1.02 (0.43, 2.42) [0.966] 0.31 (0.09, 1.11) [0.072] 0.90 (0.65, 1.27) [0.559] 1.14 (0.79, 1.66) [0.485] 0.85 (0.69, 1.05) [0.123] 1.00 (0.99, 1.02) [0.531] 0.95 (0.72, 1.24) [0.683] F test p=0.473 1.0 (reference) 0.86 (0.54, 1.38) [0.541] 1.12 (0.67, 1.89) [0.660] 0.77 (0.43, 1.37) [0.371] 1.14 (0.77, 1.69) [0.500] 1.19 (0.72, 1.97) [0.499] F test stt pp=0.021 = .0021 =0 1.0 (reference) 0.46 (0.27, 0.80) [0.006] 0.82 (0.56, 1.22) [0.328] 0.43 (0.18, 0.99) [0.046] 0.96 (0.64, 1.44 1.44) 44 4) [0 [[0.844] .844] 0.64 (0.41, (0.441, 41 1.01 1.01) 01)) [0 [0.054] 1.20 1. 20 (0.74, (0. 0.74 74 4, 1. 11.93) 93)) [0 93 [0.4 [0.460] 460 0] 0.92 0. 2 (0.63, (0. 0.63 3, 1. 11.35) 35 5) [[0.678] 0.6 678 78]] 1.24 1. 4 (0.84, (0..84 4, 1.82 1.82) 2) [0 [0.2 [0.284] 284 4] 0.75 0. 5 (0.40, (0. 0.40 40,, 1. 11.41) 41 1) [0 [0.3 [0.370] .3 370 70]] 1.13 (0.61, 1.13 (0.61 61,, 2.12 61 22.12) .12 12)) [0 [0.6 [0.692] 692 2] 00.73 .73 73 (0.46, (0.46 46, 11.17) 46 .17 17)) [0 17 [0.196] .196 196 96]] 11.39 1. 39 (0.90, (0. 0 90 9 , 2. 22.14) 14)) [0 14 [0.1 [0.137] .1 137 37]] 0.64 (0.32, 1.28) [0.205] 1.01 (1.00, 1.01) [0.223] 0.83 (0.52, 1.32) [0.428] 1.64 (0.78, 3.47) [0.193] 0.84 (0.33, 2.08) [0.700] 0.52 (0.27, 0.98) [0.042] 1.06 (0.38, 2.97) [0.916] F test p=0.017 1.0 (ref) 0.97 (0.40, 2.34) [0.942] 1.14 (0.68, 1.91) [0.611] 0.15 (0.04, 0.53) [0.003] 0.92 (0.61, 1.38) [0.677] 1.28 (0.98, 1.68) [0.071] 2.14 (1.47, 3.12) [<.001] Results reported from logistic regression include: OR (95% CI) [p-value] † A significant interaction was detected between age and length of hospital stay; younger age was predictive of CR attendance among those hospitalized 1 week, but not in subjects whose hospital stay was >1 week. ‡ To satisfy regression assumptions, a log-transformation was applied to the total number of visits in the 1 year prior to CABG; a significant interaction was detected between # visits and diabetes, such that a higher [log-transformed] # visits was associated with non-attendance, though more so in diabetic subjects. CR = Cardiac Rehabilitation, CABG = Coronary Artery Bypass Grafting Surgery, BMI = Body Mass Index 20 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Table 3. Comparison of Patient and Clinical Predictors of Attendance at Cardiac Rehabilitation by Propensity Matched Quintiles Patient and Clinical Characteristics Age (years) Smoking Status Propensity Quintile 1 2 3 4 5 1 2 3 4 5 Diabetes Diab bet etes es PVD PV VD Hospital stay >1 week Renal Failure #Visits in 1-year prior^ Neurologic complication 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 Variable Subgroup Never Current Past Never Current Past Never Current Past Never Current Past Never Current Past - No CR 71.7±10.2 68.7±9.4 66.4±11.1 61.1±11.3 61.0±12.0 28 (27%) 21 (21%) 53 (52%) 22 (31%) 17 (24%) 31 (44%) 14 (31%) 6 (13%) 25 (56%) 5 (21%) 5 (21%) 14 (58%) 8 (40%) 0 (0%) 12 (60%) 42 (41%) 28 (40%) 16 6 ((36%) 3 %) 36 3 ((13%) 13%) 3%) 4 ((20%) 200%) 35 ((34%) 344%) 17 ((24%) 24 4%) %) (20%) 9 (2 (20% 0%)) 1 (4 (4%) %) 0 (0 (0%) %) 82 (80%) 46 (66%) 20 (44%) 7 (29%) 7 (35%) 13 (13%) 4 (6%) 2 (4%) 0 (0%) 0 (0%) 3.0 (1.0, 7.0) 5.0 (2.0, 9.0) 6.0 (2.0, 9.0) 2.0 (1.0, 5.5) 6.5 (2.5, 15.5) 21 (21%) 8 (11%) 1 (2%) 1 (4%) 0 (0%) CR 70.7±10.9 69.3±8.4 66.5±9.6 64.1±8.7 59.5±9.4 15 (25%) 15 (25%) 31 (51%) 21 (23%) 17 (18%) 54 (59%) 35 (30%) 21 (18%) 62 (53%) 44 (32%) 14 (10%) 80 (58%) %) (47%) 67 (47% %) 14 (10%) 61 (43%) 27 (44%) 332 2 (35%) (30%) 35 (3 (30% 0% %) 29 ((21%) 21%) 21 %) (23%) 32 2 (2 23%) (41%) 225 5 (4 (41% 1%)) (20%) 118 8 (2 (20% 0%)) 0% 16 ((14%) 14 4%) 7 ((5%) 5% %) 4 ((3%) 3%)) 3% 49 (80%) 62 (67%) 50 (42%) 39 (28%) 48 (34%) 8 (13%) 4 (4%) 0 (0%) 3 (2%) 1 (1%) 4.0 (1.0, 7.0) 4.0 (1.0, 8.0) 5.0 (2.0, 9.0) 4.0 (2.0, 7.0) 5.0 (3.0, 9.0) 15 (25%) 6 (7%) 4 (3%) 1 (1%) 1 (1%) 21 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 P-value 0.55 0.68 0.96 0.14 0.53 0.82 0.19 0.79 0.24 00.24* 0. 24** 24 0.70 0.50 0.47 0. 477 00.33 .33 33 0.80 0 00.39 0. 399 00.47 0. 47 00.31 .3 31 1.00* 11.00* .00 00* 0.99 0.82 0.81 0.93 0.92 0.95 0.69 0.07* 1.00* 1.00* 0.31 0.63 0.72 0.08 0.44 0.55 0.27 1.00* 0.28* 1.00* DOI: 10.1161/CIRCULATIONAHA.112.001365 Type of insurance 1 2 3 4 5 Medicare Medicaid Commercial Self-Pay Medicare Medicaid Commercial Medicare Medicaid Commercial Medicare Medicaid Commercial Medicare Medicaid Commercial 74 (73%) 6 (6%) 10 (10%) 12 (12%) 46 (66%) 3 (4%) 21 (30%) 25 (56%) 2 (4%) 18 (40%) 9 (38%) 0 (0%) 15 (63%) 6 (30%) 2 (10%) 12 (60%) 44 (72%) 3 (5%) 10 (16%) 4 (7%) 63 (68%) 5 (5%) 24 (26%) 71 (60%) 6 (5%) 41 (35%) 64 (46%) 3 (2%) 71 (51%) 31 (22%) 6 (4%) 105 (74%) ^ median (Q1, Q3); Wilcoxon Rank Sum Test * p-value obtained from Fisher’s Exact Test due to low cell counts CR = Cardiac Rehabilitation 22 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 0.48 0.83 0.82 0.70* 0.22* DOI: 10.1161/CIRCULATIONAHA.112.001365 Table 4. Effect of Cardiac Rehabilitation on All-Cause Mortality, by Propensity Approach Propensity Method Unadjusted Overall Adjustment Overall Stratification ification Overall erall Quintile uintile 1 Quintile uintile 2 Quintile uintile 3 Quintile uiint ntil ilee 4 il Quintile uint ui n ille 5 Matching h ng hi ng Overall era r lll (NCR+ / NCR-) N (#Deaths) (582 / 264) 846 (193) 20.3% 44.6% 0.36 (0.27-0.47) [<.001] (582 / 264) 846 (193) 23.0% 35.7% 0.54 (0.40-0.74) [<.001] (551 / 261) (61/102) (92/70) (118/45) (138/24) (138 (1 3 /24) (142/20) (142 (1 42/2 / 0) 812 (192) 163 (70) 162 (47) 163 (36) 162 (25) 162 162 (14) (14) 23.9% 40.9% 31.3% 17.4% 18.2% 1 .2% 18 2 10.5% 100.5% 5 36.5% 54.6% 38.9% 51.1% 30.9% 300.9 .9% 9% 19.9% 19.9 9.9% % 0.55 5 (0.40-0.75) 55 (0. 0 40 0-0 -0.7 0.7 .75) 5) [<.001] [<. 0.66 0.66 66 (0.40-1.10) (0.40 40-1 1.110) [0.113] [0. 0.76 (0.43-1.35) [0.348] [0. 0.27 (0.14-0.52) [<.001] [<. 0.54 (0.22-1.36) [0.191] [0. 0.50 0.5 (0.14-1.79) (0. 0.14 14-1 1.779) [0.288] [0.2 (220 ( 20 (2 2 / 220) 220) 0 440 4 0 (1 44 (130 (130) 30)) 26 26.9% 6.9% 9 42.7% 42.7% 7% 0.55 0 55 (0.36-0.84) 0. (0. 0 36 36-0.884)) [0.007] [0. 10-year Cumulative Incidence of Death (%)^ CR Participation Effect* HR (95% CI) [p-value] CR participants Non-participants ^Cumulative ulat ul a ive ve incidence rate ra rates ate tes obta obtained b ined from fro r m Kapl Kaplan-Meier plan a -Mei Me er m method; e ho et hod; d rates rat a es in in the h overall over e al a l group grrou o p for for all a l th al tthree ree prop propensity o ensity ty m methods ettho hods o rreflect efllect ad ef aadjustment justme ju u m nt nt for tthe h prop he propensity score *Effectt off CR CR participation part rti ticip icipat a ionn onn 10-year 100-ye year ar mortality morta tali lity was was assessed ass sses e sed via Cox Cox PH PH regression regr re gres essi sion conditional con ondi diti tion ti onal al oon n 666-month mont nthh surviva surv urviv iva CR = C Cardiac Rehabilitation. attended CR. attend CR. ardi ar diac di ac R ehab eh abil ab ilit il itat it atio at ionn. io n. + = at atte tend te nded nd ed C R. - = di R. didd no nott at atte tend te nd C R. R. 23 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 DOI: 10.1161/CIRCULATIONAHA.112.001365 Figure Legend: Figure 1. Association between Cardiac Rehabilitation and Survival in CABG Patients. 24 Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 Kaplan Meier survival plots by propensity score quintile and for strata-combined group, comparing rates over time between patients attending and not attending cardiac rehabilitation Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 Supplemental Material Participation in Cardiac Rehabilitation and Survival Following Coronary Artery Bypass Graft Surgery: A Community Based Study [Intended for online publication] Authors: Quinn R. Pack, MD; Kashish Goel, MD; Brian D. Lahr, MS; Kevin L. Greason, MD; Ray W. Squires, PhD; Francisco Lopez-Jimenez, MD, MS; Zixin Zhang, MD; Randal J.Thomas, MD, MS Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013 Figure S1. Time Trends in Cardiac Rehabilitation Participation among patients with Coronary Artery Bypass Graft Surgery in Olmsted County, MN. As seen, participation in CR remained consistently high for the duration of the study with only minimal year to year variability. Downloaded from http://circ.ahajournals.org/ at Mayo Clinic Libraries on July 9, 2013
© Copyright 2024