- Email: [email protected]
Cardiac rehabilitation for community-based patients with myocardial infarction
Journal of Clinical Epidemiology 54 (2001) 1025–1030
Cardiac rehabilitation for community-based patients with myocardial infarction: Factors predicting discharge recommendation and participation Kimberly Barbera,*, Manfred Stommelb, Jill Krollc, Margaret Holmes-Rovnerc, Barbara McIntoshd a
Saginaw Cooperative Hospitals, Inc., Department of Research and Sponsored Programs and Michigan State University/College of Human Medicine, East Lansing, MI b Michigan State University/College of Nursing, East Lansing, MI c Michigan State University/College of Human Medicine, Department of Medicine, East Lansing, MI d Michigan State University/College of Human Medicine and The Department of Internal Medicine, Hurley Medical Center, Flint, MI Received 12 September 2000; received in revised form 6 April 2001; accepted 6 April 2001
Abstract Although there is substantial evidence that cardiac rehabilitation is beneficial for post myocardial infarction (MI) patients, such programs are currently under utilized. This study examined systematic criteria predicting physician referral to and patients’ participation in cardiac rehabilitation programs. Patients discharged for MI were interviewed in-hospital and at 6-12 weeks post discharge to determine referral, participation, and completion. Stepwise logistic regression analyzed factors associated with rehabilitation. Factors associated with referral to rehab were catheterization (p 0.001), bypass surgery (p 0.01), cardiologist/cardiac surgeon appointment (p 0.02), and age (p 0.01). Participation was increased for those with bypass surgery (p 0.001), and referral to cardiologist or cardiac surgeon (p 0.001). Type of provider significantly influences referral to and participation in cardiac rehabilitation. This suggests that encouragement plays a strong role in attendance for rehabilitation. The same strong encouragement should be given to the broader range of MI patients who stand to benefit from cardiac rehabilitation. © 2001 Elsevier Science Inc. All rights reserved. Keywords: Cardiac rehabilitation; Myocardial infarction; Prevention; Cardiology; Coronary heart disease; Cardiovascular diseases
1. Introduction There is substantial evidence that attending physical exercise programs as part of cardiac rehabilitation is beneficial to post-myocardial infarction (MI) patients. However, participation rates in cardiac rehabilitation programs are generally low, with reported levels ranging from 17% to 21% [1,2]. This pattern even applies to patients who meet all the low-risk criteria outlined in the guidelines of the American Association of the Cardiovascular and Pulmonary Rehabilitation [3]. Yet, little is known about the reasons for the low patient participation in cardiac rehabilitation programs. An emerging literature has focused on patient-centered reasons for non-participation and non-compliance. It shows that affordability of service, possibly insurance coverage, social support from a spouse or other caregiver [4], gender-specific attitudes [5,6], and patient anxiety [7,8] all play a role * Corresponding author. Research and Sponsored Programs, Saginaw Cooperative Hospitals, Inc., Saginaw, MI 48602. Tel.: 989-583-6891; fax: 989- 583-6922. E-mail address: [email protected] (K. Barber).
in patient under utilization. Less well studied are physicianspecific predictors. Yet, anecdotal evidence suggests that physicians may under-appreciate the importance of physical exercise programs in cardiac rehabilitation. As a result, patient referral and encouragement to participate in cardiac rehabilitation may be less than optimal. Cardiac rehabilitation has been shown to be effective in improving clinical outcomes for patients following a heart attack event. Patients completing rehabilitation programs with exercise tend to achieve lower blood pressure, decreased weight, improved muscle tone, increased exercise capacity and tolerance, decreases in debilitating symptoms, and general improvement in cardiac functioning [9,10]. Together with dietary or pharmacological therapy, improvements in lipid levels have also been observed [11]. In combination with education and behavioral counseling, exercise increases the likelihood of smoking cessation and enhances measures of psychological functioning [12]. Finally, cardiac rehabilitation has been shown effective as secondary prevention in decreasing the risk of re-infarction and improving survival [13].
0895-4356/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved. PII: S0895-4356(01)00 3 7 5 - 4
1026
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
Currently, there is little evidence as to what impact physicians’ attitudes and decisions have on the willingness of patients to participate and follow through with a rehabilitation program. In a prospective study of 226 MI patients, 62 years and older, after controlling for the effects of demographic, medical, and psycho-social factors, the most powerful predictor of patient participation in a rehabilitation program was the primary physician’s recommendation [2]. This effect was strongest among the oldest patients and among women. The latter finding should not be surprising, since many cardiac patients after an MI are fearful and uncertain of the possible negative side effects of physical exercise [6]. Clearly, they will need encouragement and assurance that physical exercise, as a part of a comprehensive rehabilitation program not only has benefits, but also is safe and does not increase the risk of re-infarction. Other researchers have also shown that women and the elderly are less frequently referred to cardiac rehabilitation programs, even though program benefits have been observed in these groups [14,15]. Thus, available information on referral patterns seems to indicate a certain reluctance on the part of physicians to push for patient participation in physical rehabilitation programs [16]. Existing community studies show that fewer than one-third of MI patients receive information or counseling about cardiac rehabilitation before being discharged from the hospital. Contemporary trends toward shorter hospital stays may further reduce the likelihood of comprehensive counseling in this setting. There is a clear need to examine what systematic criteria, if any, predict referral and participation of MI patients in cardiac rehabilitation programs. The study of referral and participation patterns is complicated by the fact that different factors may influence: (1) the referral to cardiac rehabilitation by the provider, (2) the actual patient enrollment in a program, and (3) the completion of such a program. Employing data from a communitybased study, we focus primarily on two issues: (1) factors that predict referral of MI patients to cardiac rehabilitation programs at hospital discharge, and (2) factors that predict actual participation in cardiac rehabilitation 6 or 12 weeks after the hospital discharge.
2. Methods 2.1. Sample Subjects were 1671 patients, who were diagnosed with acute myocardial infarction (AMI) at admission to one of five hospitals in two mid-Michigan communities. Enrollment in the Michigan State University Inter-institutional Collaborative Heart Study occurred in two different phases, between January 1994 and April 1995 (Phase I: n 1163) and between February 1 and September 30, 1997 (Phase II: n 508). The independent Institutional Review Boards of each hospital approved the study and procedures were in accordance with each institution’s guidelines. Informed con-
sent for in-hospital and post-discharge follow-up was obtained from each patient. Of the 1671 patients diagnosed with AMI, 1475 (86% of total MI population) survived to be discharged. All of the patients were approached for in-hospital interviews, but many were physically unable to endure an interview (20%), died right after discharge (12%), were discharged early or missed (15%), refused to be interviewed (20%) or did not participate for unknown reasons (7%).
Table 1 Sample characteristics
Characteristic Sex Male Female Age 44 and under 45 thru 54 55 thru 64 65 thru 74 75 Race White Black Other (incl. Hispanic) Missing Insurance Commercial Medicare Medicaid Self-pay/none Missing Comorbidity score 1 (no other diseases) 2–3 4 Smoking status at time of MI Current smoker Not current smoker Missing Previous MI Yes No Catheter Yes No PTCA/angio Yes No CABG/bypass Yes No Rehab on discharge plan? Yes No Attended rehab? Yes No Marital status Married Not married
Patients who survived to discharge n 1475
Patients with follow-up data available n 347
950 (64%) 525 (36%)
234 (67%) 113 (33%)
147 (10%) 282 (19%) 338 (23%) 383 (26%) 325 (22%)
37 (11%) 73 (21%) 98 (28%) 79 (23%) 60 (17%)
1255 (85%) 188 (13%) 27 (2%) 5
316 (91%) 25 (7%) 6 (2%) 0
571 (40%) 747 (52%) 54 (4%) 57 (4%) 46
167 (49%) 153 (45%) 9 (3%) 13 (4%) 5
496 (34%) 764 (52%) 215 (15%)
148 (43%) 165 (48%) 34 (10%)
635 (61%) 413 (39%) 427
137 (40%) 210 (60%) 0
227 (15%) 1248 (85%)
85 (25%) 258 (75%)
1188 (80%) 287 (20%)
307 (88%) 40 (12%)
473 (32%) 1002 (68%)
148 (43%) 199 (57%)
352 (24%) 1123 (76%)
83 (24%) 264 (76%)
160 (11%) 1315 (89%)
57 (16%) 290 (84%)
— —
91 (26%) 256 (74%) 235 (68%) 85 (32%)
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
Theoretically, the first two groups (unable or died, 32%) do not introduce bias, as they are unlikely to be eligible for cardiac rehabilitation. Early discharge or refusal may create selection bias if patients had a less severe (early discharges) or more severe (interview refusals) heart attack. A comparison of the total patients surviving to discharge with those followed post discharge indicates little bias occurred (Table 1). This left a total of 449 patients (30% of 1475) who were interviewed in the hospital and identified for a follow-up interview in the community after discharge. Due to resource changes Phase I patients received mail-survey follow-up and Phase II patients received interview follow-up. For Phase I patients, the follow-up in the community involved mailed questionnaires (n 273) 3 months after discharge from the hospital. For Phase II patients, the follow-up involved telephone interviews (n 176) 6 weeks after discharge. However, responses to questions about attendance of cardiac rehabilitation were obtained from 347 subjects (175 in phase I and 172 in phase II). Table 1 shows the characteristics of all 1475 patients discharged, as well as, the sub-sample for which follow-up data are available. 2.2. Procedures Patients were identified through elevated enzyme lists generated by each hospital daily, as well as on the basis of admission logs and hospital charts on floors occupied by cardiac patients. Patients who met symptom, enzyme and/or electrocardiogram criteria were approached for enrollment. Trained research nurses visited each hospital daily and obtained informed consent from English-speaking patients, unless disoriented, on ventilators, too ill to interview, unaware of the MI diagnosis, or refused. In some cases, patients were discharged before the research nurse identified them and attempted an interview. In addition to consenting to a medical record audit, a subset of patients were interviewed still in the hospital and were asked about their past medical history, family history, basic demographic information, lifestyle choices (diet, smoking, exercise, recreational substance use), and physical functioning. Six or 12 weeks following discharge, these patients were contacted and asked about attendance at a cardiac rehabilitation program. 2.3. Measures 2.3.1. Outcome measures Outcome measures included referral to a rehabilitation program for MI and participation in such a program. Information obtained from the hospital discharge plan established whether or not a referral had been made; 160 patients (10.8% of total) were referred. This included 57 (16%) of those interviewed with follow-up. At the follow-up interviews, patients were asked if they had attended a cardiac rehabilitation program. A total of 91 (26.2%) indicated that they were attending or had finished participation in a rehabilitation program at the time of the interview. (Only two patients reported program completion and were combined
1027
with those currently attending.) The higher percentage reporting attendance of a rehabilitation program than originally referred in the hospital reflects a delay in 34 of the 91 patients who were referred 3 to 6 weeks post discharge. 2.3.2. Predictors Race/ethnicity was obtained from the hospital charts. Given the skewed distribution (85% Caucasian, 13% African Americans, 2% Hispanic), patients were categorized as White or Non-White in all later analyses. Information on primary health insurance came from hospital charts and grouped into four categories: private insurance (all carriers), Medicare (with or without private insurance), Medicaid or no insurance, and no insurance information. However, information on specific policy coverage was unavailable and it remains unclear to what degree cardiac rehabilitation programs are covered. Chart information also was used to determine if a patient had undergone cardiac catheterization, angioplasty or cardiac bypass surgery. We hypothesized that use of various procedures in-hospital may affect recommendations for subsequent care. Comorbidity was assessed using the Charlson Index [17]. The index provides information on the patient’s history of 15 chronic diseases. We modified the index by excluding the MI as a co-morbidity, since all subjects experienced one. Based on Charlson et al. [18] the remaining chronic conditions yielded a weighted score, ranging from 0 to a maximum of 11. With few cases occupying the upper end of the scale, we re-coded the Charlson Index into three categories: MI only, comorbidity score of 1–2 and comorbidity score of 3–11. A separate dichotomous variable was employed to record a patient’s prior history of myocardial infarctions. From the hospital discharge plan, we also recorded whether post-discharge appointments were scheduled with: (1) a cardiologist or cardiac surgeon (the two groups were later combined after extensive probing showed them to have very similar referral patterns); (2) a primary care physician (internist, family practice, other); or (3) an unknown provider. However, while the data from the discharge records allowed us to identify the physician a patient was referred to, we did not have information on the physician managing the patient within the hospital. A total of 36 physicians were involved in the referrals. A dichotomous variable was employed to indicate whether a case came from Phase I with a 3-month follow-up interview or Phase II with a 6-week follow-up interview. Additional information was available from a subset of patients who participated in the in-hospital interview. Only marital status (married versus not married) proved a potent predictor and is included in the data analysis. 2.3.3. Statistical analysis After initial analysis to test for significant bivariate relationships between all the categorical variables, two hierarchical and one step-wise logistic regression models were employed. The logistic models were used to predict the multivariate, joint effects of all potential predictors on referral or participation in cardiac rehabilitation. The associated
1028
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
significance tests were based on the Wald-statistic or the log-likelihood ratio. For a stepwise logistic model, we employed probability values associated with the log-likelihood ratio (P .05 for entry, P .10 for removal) to compare models with and without the examined predictor variable.
3. Results Table 1 presents descriptive information on all variables used in the analysis for the discharge sample of 1475 MI patients and the subset of 347 patients interviewed about their participation in rehabilitation. There were no significant differences in the distribution of cases according to age groups, gender and bypass surgery. However, patients with followup were slightly less likely to be minority (9% vs. 15%, 2 8.5, P 0.014), and somewhat more likely to have commercial insurance (49% vs. 40%, 2 9.3, P 0.025). They also had fewer co-morbid conditions (2 12.2, P 0.002), were less likely to be smokers (40% vs. 61%, 2 47.0, P 0.001) and more likely to report a previous AMI (25% vs. 15%, 2 16.1, P 0.005). The total enrolled patients included repeat AMIs, several of whom were previously enrolled and interviewed but were not interviewed a second time while hospitalized for the current AMI. This explains the lower rate of reported previous MI among the total group versus the followed group. In terms of treatment, patients available for follow-up were more likely to receive a catheter (89% vs. 81%, 2 12.0, P 0.001) and PTCA (43% vs. 32%, 2 14.0, P 0.001). Bivariate relationships involving the referral to a rehabilitation program were examined and revealed only the following four significant predictors. Patients scheduled for an appointment with a cardiologist or cardiac surgeon were more likely to be referred at discharge than patients with appointment to other physicians (12.7% versus 6.2%; P .001). Patients with cardiac catheterization were more likely (12.5%) to be referred than those without (4.2%; P .001), as were non-smokers (13.9%) compared to smokers (9.4%; P . 032). Finally, patients who had cardiac bypass surgery were more likely to be referred to rehabilitation than those who did not (18.0% versus 10.2%, P .001). Among the 347 respondents who provided follow-up information about their attendance in rehabilitation programs, men were more likely to participate than women (30.6% versus 16.2%, P .01). Patients with a cardiologist or cardiac surgeon appointment were also more likely to participate (29.5% versus 9.7% of other patients, P .001). Again, patients with bypass surgery had higher participation rates (46.3%) than those without (19.5%, P .01). Some of the variables obtained from the hospital interviews were also related to participation in cardiac rehabilitation programs. Married patients were more likely to participate (28.3%) than not married patients (15.2%, P .02), and higher income was related to lower participation rates (P .001). Finally, referral to a rehabilitation program at hospital
discharge is a strong predictor of later participation (54.4% versus 20.3%, P .001). Table 2 shows the results of a multivariate logistic regression predicting the odds of referral to a cardiac rehabilitation program at the time of hospital discharge. Recall that 160 of the 1475 were referred. (Only 10 cases had missing information on some of the variables in the model.) Among significant predictors the strongest included: catheterization of the MI patient, the type of provider with whom the patient has an appointment, bypass surgery, and age of the patient. In particular, the odds of referral are 2.6 times larger among catheterized patients (P .001), and almost 1.7 times higher among patients who had bypass surgery (P .01). Patients scheduled for an appointment with a cardiologist or cardiac surgeon had more than twice the odds of being referred to cardiac rehabilitation (1/.43 2.33, P .02). Patients aged 45–59 had the greatest odds of referral, two times greater than the youngest MI patients (P .05). Table 3 examines the odds of participation in a cardiac rehabilitation program. Given the differences in the time interval for the follow-up interviews, it comes as no surprise that the odds of participation are only half as large (.52, P .02) in the Phase II study group. In addition, bypass surgery remains a powerful predictor of participation: patients with Table 2 Predictors of referral at hospital discharge Predictor Phase 0 Phase I, 1 Phase II Sex 0 male, 1 female Race 0 non-Hispanic White 1 minority Age (Reference: 26–44) 45–59 60–74 75–93 Insurance: (Reference: private) Medicare Medicaid No ins. Missing Charlson Index (Reference group: MI only) Comorbidities 1–2 Comorbidites: 2 Prior MI 0 no, 1 yes Catheterization 0 no, 1 yes Angioplasty 0 no, 1 yes Bypass surgery 0 no, 1 yes Provider: (Reference group: cardiologist) Family phys. Other provider
Adjusted odds-ratio
95% CI for odds-ratio
P-value
.75
.50–1.14
.18
1.18
.82–1.71
.37
1.03
.63–1.70
.90
2.02 1.18 1.60
.99–4.13 .54–2.58 .66–3.88
.05 .68 .30
.98 1.01 .65 .18
.59–1.64 .41–2.54 .22–1.90 .02–1.37
.95 .98 .43 .10
1.29 1.34
.87–1.90 .75–2.38
.20 .32
1.13
.70–1.85
.61
2.60
1.34–5.07
.00
.80
.52–1.22
.29
1.67
1.11–2.50
.01
.43 .62
.22–.86 .34–1.11
.02 .11
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030 Table 4 Predictors of participation post discharge: stepwise results
Table 3 Predictors of participation 6 or 12 weeks post-discharge Predictor Phase 0 Phase I, 1 Phase II Sex 0 male, 1 female Race 0 non-Hispanic White 1 minority Age (Reference: 26–44) 45–59 60–74 75–93 Insurance: (Reference: private) Medicare Medicaid No insurance Missing Charlson Index (Reference: MI only) 1–2 Comorbidities 2 Comoridities Prior MI 0 no, 1 yes Catheterization 0 no, 1 yes Angioplasty 0 no, 1 yes Bypass surgery 0 no, 1 yes Provider: (Reference: cardiologist) Family physicians Other provider Discharge referral to rehab 0 no, 1 yes
1029
Adjusted 95% CI odds-ratio for odds-ratio P-value
Predictor Phase 0 3 months interview 1 6 weeks interview Marital status 0 not married 1 married Bypass surgery 0 no, 1 yes Discharge referral to cardiac rehabilitation 0 no, 1 yes
Adjusted odds-ratio
95% CI for odds-ratio
P-value
.52
.29–.92
.02
.57 1.15
.30–1.10 .42–3.16
.09 .78
1.83 1.01 1.42
.65–5.21 .32–3.15 .36–5.66
.25 .99 .62
1.00 1.22 .52 .22
.44–2.24 .21–7.06 .10–2.86 .02–2.96
.99 .83 .46 .25
.77 .41
.43–1.36 .13–1.31
.37 .13
.59
.30–1.19
.14
.81
.27–2.46
.71
1.36
.66–2.78
.40
4. Discussion
3.19
1.52–6.73
.00
.69 .31
.22–2.21 .09–1.09
.53 .07
4.07
2.07–8.04
.00
Despite the available evidence and clinical guidelines for MI patients published by such groups as the American Association of the Cardiovascular and Pulmonary Rehabilitation [3], overall participation in exercise-based cardiac rehabilitation programs remains low. Data from our study show only 16% of the MI patients were referred to a cardiac rehabilitation program at discharge, and only 26% of the patients later interviewed in the community actually participated in such a program. However, 54% of the patients who were referred at discharge did participate at the time of the interview and, in the multivariate analysis, referral emerged as the strongest predictor of participation. This finding supports those of Ades et al. [2] who showed that physicians’ recommendation are the most powerful predictors of patient participation in a cardiac rehabilitation program. We observed a “provider effect” in the referral of postMI patients. Merely scheduling an appointment with a community-based cardiologist or cardiac surgeon increased the likelihood of a patient being referred to rehabilitation at the time of discharge. While this finding attests to the fact that cardiac specialists have a greater awareness of the importance of cardiac rehabilitation programs, it is also troubling. The implication is that the less severely ill post-MI patients, treated by generalist physicians who appear to be less likely to refer for cardiac rehabilitation programs, are missing opportunities for cardio-health improvements. Our finding that married patients are more likely to participate in cardiac rehabilitation programs again underscores the importance of social support and encouragement [4,5]. A gender difference was suggested but not significant in the multivariate analysis. Other studies have shown evi-
bypass surgery had more than three times greater odds (3.19, P .001) of attending compared to patients without bypass surgery. However, while catheterization predicted referral to rehabilitation at the time of hospital discharge, it did not predict actual participation in such programs. Similarly, the “provider effect” (whether or not a patient goes to a cardiologist or primary care practitioner in the community) was absorbed once discharge referral was added as a predictor of actual participation. In fact (16.32 Wald statistic), referral at discharge had the strongest effect, with referred patients having more than 4 times greater odds of participation (P .001). The patterns did not differ when general internists and family practitioners were separated. Neither health insurance status nor comorbidity affected a person’s odds of participating in cardiac rehabilitation. In this sample, women had lower odds of participation (.57) than men. While this effect is not statistically significant (P .09), the results warrant further study. Because information from medical records was limited, we introduced additional measures from the in-hospital interview. There were 314 cases that yielded data from both their hospital and the follow-up interviews. The additional predictors included were patient marital status, household income, employment status and depression (Center for Epidemiologic Studies-
.61
.35–1.06
.08
2.38
1.15–4.94
.02
2.81
1.55–5.08
.00
4.03
2.03–7.75
.00
Depression scores obtained in the hospital interview). Given the number of variables and limited sample size, we opted for a step-wise approach to logistic modeling. Table 4 displays the final results. Four predictor variables were retained. Only marital status appears to have an additional effect on participation in cardiac rehabilitation: married patients had more than twice greater odds of participation (2.38, P .02).
1030
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
dence that women are less likely to be referred and to participate in cardiac rehabilitation programs [17]. Our data contain suggestions that this gender effect may, in part, be a case of under-referral of the elderly. We observed that patients in the younger age group of 45–59 were most likely to be referred, yet women generally have their first MI 10 years later than men. These data confirm the correlation between gender and age, with older MI patients much more likely to be female. As a result, the gender effect is no longer significant in the multivariate analysis, after the effects of age have been taken into account. Overall our evidence suggests the strong role that providers and their referral activity play in the actual participation of post-MI patients in cardiac rehabilitation programs. Patients who were scheduled to visit a cardiologist or cardiac surgeon post-discharge were more likely to have had catheterization and cardiac bypass surgery. They were more likely to be male, Caucasian, under the age of 75 years, and to have private insurance. Physician influence is evident in our data. While physician attitude may not be enough to overcome patient receptiveness, motivation, or compliance behavior, it is clear that post-MI patients followed up in the community by a cardiologist or cardiac surgeon have a much greater probability of getting involved in cardiac rehabilitation. It appears that stronger encouragement needs to be provided to the broader range of MI patients who stand to benefit from cardiac rehabilitation. Acknowledgments The Blue Cross/Blue Shield of Michigan Foundation, Michigan State University Foundation-College of Human Medicine at Michigan State University, and all hospitals involved in the study provided funding for this project. Investigators Participating in the Michigan State University Inter-Institutional Collaborative Heart Study: Firas Akhrass MD, Chris Colenda PhD, Francesca Dwamena MD, Hassan El-Tamimi MD, Joseph Gardinei PhD, Ken Gaines MD, Margaret Holmes-Rovner PhD, Jill Kroll PhD, Joel Kupersmith, MD, Raj Nfitra MD, David Rovner MD, Peter Vasilenko PhD, and Ralph Watson MD, (MSU, East Lansing); Aryeh Stein PhD, Emory University, Atlanta. Kimberly Barber MS, Sue Davis RN, Del DeHart MD, (Saginaw Cooperative Hospitals, Inc., Saginaw); Saginaw Cooperative Hospitals, Inc.; Jack Ferlinz MD, Aleda E. Lutz (VAMC, Saginaw). Michelle Debemardi RN, Barbara McIntosh MD, D. Kay Taylor PhD, (Hurley Medical Center, Flint); Susan Smith MD, (McLaren Regional Medical Center, Flint). Acknowledgement of Support Abin Achrekar MPH, Fawzia Ahmed MS, Usman Alam, Elisa Arfianti, Amy Cairns, Qin Chen, East Lansing; Peter
Fattel, Syed Fazal-Ur-Rehman, V. Kandallu, Anton Koinev, Annette McLane BS, Danutta Podgorska, Gennaro Polverino, Manfred Stommel, PhD (MSU, East Lansing). Leslie Franke BS and O. David Tenny, (Saginaw Cooperative Hospitals, Inc., Saginaw). Jeanne Minor, (Rochester). Kathy Sands, (Flint).
References [1] Campbell NC, Grimshaw JM, Ritchie LD, Rawles JM. Outpatient cardiac rehabilitation: are the potential benefits being realized? J R Coll Physicians Lond 1996;30:514–9. [2] Ades PA, Waldmann ML, Polk DM, Coflesky JT. Referral patterns and exercise response in the rehabilitation of female coronary patients aged 62 years. Am J Cardiol 1992;69:1422–5. [3] American Association of Cardiovascular and Pulmonary Rehabilitation. Guidelines for cardiac rehabilitation programs 2nd ed. Champaign, IL: Kinetics Books, 1995. [4] Wallace JP, Raglin JS, Jastremski CA. Twelve months adherence of adults who joined a fitness program with a spouse vs without a spouse. J Sports Med Phys Fitness 1995;35:206–13. [5] Ginzel AR. Women’s compliance with cardiac rehabilitation programs. Prog Cardiovasc Nurs 1996;11:30–5. [6] Moore SM. Women’s views of cardiac rehabilitation programs. J Cardiopulm Rehab 1996;16:123–9. [7] Crowe JM, Runions J, Ebbesen LS, Oldridge NB, Streiner DL. Anxiety and depression after acute myocardial infarction. Heart Lung 1996;25:98–107. [8] Oldridge NB, Furlong W, Feeny D, Torrance G, Guyatt G, Crowe J, Jones N. Economic evaluation of cardiac rehabilitation soon after acute myocardial infarction. Am J Cardiol 1993;72:154–61. [9] Leitch JW, Newling RP, Basta M, Inder K, Dear K, Fletcher PJ. Randomized trial of a hospital-based exercise training program after acute myocardial infarction: cardiac autonomic effects. J Am Coll Cardiol 1997;29:1263–9. [10] Adachi H, Koike A, Obayashi T, Umezawa S, Aonuma K, Inada M, Korenaga M, Niwa A, Marumo F, Hiroe M. Does appropriate endurance exercise training improve cardiac function in patients with prior myocardial infarction? Eur Heart J 1996;17:1511–21. [11] Heldol M, Sire S. Effects of intensive exercise training on lipid levels in high risk post-MI patients. Eur Heart J 1994;15:1362–7. [12] Wenger NK. Rehabilitation of the cardiac patient. Advance Cardiovasc Med 1996;3:1–7. [13] 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;14:831–5. [14] O’Connor GT, Buring JE, Yusuf S, Goldhaber SZ, Olmstead EM. An overview of randomized trials of rehabilitation with exercise after myocardial infarction. Circulation 1989;80:234–44. [15] Lavie CJ, Milani RV. Effects of cardiac rehabilitation and exercise training programs in patients or 75 years of age. Am J Cardiol 1996;78:675–7. [16] Balady GJ, Jette D, Scheer J, Downing J. Changes in exercise capacity following cardiac rehabilitation in patients stratified according to age and gender. Results of the Massachusetts Association of Cardiovascular and Pulmonary Rehabilitation Multicenter Database. J Cardiopulmonary Rehab 1996;16:38–46. [17] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. [18] Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1993;47:1245–51.
Cardiac rehabilitation for community-based patients with myocardial infarction: Factors predicting discharge recommendation and participation Kimberly Barbera,*, Manfred Stommelb, Jill Krollc, Margaret Holmes-Rovnerc, Barbara McIntoshd a
Saginaw Cooperative Hospitals, Inc., Department of Research and Sponsored Programs and Michigan State University/College of Human Medicine, East Lansing, MI b Michigan State University/College of Nursing, East Lansing, MI c Michigan State University/College of Human Medicine, Department of Medicine, East Lansing, MI d Michigan State University/College of Human Medicine and The Department of Internal Medicine, Hurley Medical Center, Flint, MI Received 12 September 2000; received in revised form 6 April 2001; accepted 6 April 2001
Abstract Although there is substantial evidence that cardiac rehabilitation is beneficial for post myocardial infarction (MI) patients, such programs are currently under utilized. This study examined systematic criteria predicting physician referral to and patients’ participation in cardiac rehabilitation programs. Patients discharged for MI were interviewed in-hospital and at 6-12 weeks post discharge to determine referral, participation, and completion. Stepwise logistic regression analyzed factors associated with rehabilitation. Factors associated with referral to rehab were catheterization (p 0.001), bypass surgery (p 0.01), cardiologist/cardiac surgeon appointment (p 0.02), and age (p 0.01). Participation was increased for those with bypass surgery (p 0.001), and referral to cardiologist or cardiac surgeon (p 0.001). Type of provider significantly influences referral to and participation in cardiac rehabilitation. This suggests that encouragement plays a strong role in attendance for rehabilitation. The same strong encouragement should be given to the broader range of MI patients who stand to benefit from cardiac rehabilitation. © 2001 Elsevier Science Inc. All rights reserved. Keywords: Cardiac rehabilitation; Myocardial infarction; Prevention; Cardiology; Coronary heart disease; Cardiovascular diseases
1. Introduction There is substantial evidence that attending physical exercise programs as part of cardiac rehabilitation is beneficial to post-myocardial infarction (MI) patients. However, participation rates in cardiac rehabilitation programs are generally low, with reported levels ranging from 17% to 21% [1,2]. This pattern even applies to patients who meet all the low-risk criteria outlined in the guidelines of the American Association of the Cardiovascular and Pulmonary Rehabilitation [3]. Yet, little is known about the reasons for the low patient participation in cardiac rehabilitation programs. An emerging literature has focused on patient-centered reasons for non-participation and non-compliance. It shows that affordability of service, possibly insurance coverage, social support from a spouse or other caregiver [4], gender-specific attitudes [5,6], and patient anxiety [7,8] all play a role * Corresponding author. Research and Sponsored Programs, Saginaw Cooperative Hospitals, Inc., Saginaw, MI 48602. Tel.: 989-583-6891; fax: 989- 583-6922. E-mail address: [email protected] (K. Barber).
in patient under utilization. Less well studied are physicianspecific predictors. Yet, anecdotal evidence suggests that physicians may under-appreciate the importance of physical exercise programs in cardiac rehabilitation. As a result, patient referral and encouragement to participate in cardiac rehabilitation may be less than optimal. Cardiac rehabilitation has been shown to be effective in improving clinical outcomes for patients following a heart attack event. Patients completing rehabilitation programs with exercise tend to achieve lower blood pressure, decreased weight, improved muscle tone, increased exercise capacity and tolerance, decreases in debilitating symptoms, and general improvement in cardiac functioning [9,10]. Together with dietary or pharmacological therapy, improvements in lipid levels have also been observed [11]. In combination with education and behavioral counseling, exercise increases the likelihood of smoking cessation and enhances measures of psychological functioning [12]. Finally, cardiac rehabilitation has been shown effective as secondary prevention in decreasing the risk of re-infarction and improving survival [13].
0895-4356/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved. PII: S0895-4356(01)00 3 7 5 - 4
1026
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
Currently, there is little evidence as to what impact physicians’ attitudes and decisions have on the willingness of patients to participate and follow through with a rehabilitation program. In a prospective study of 226 MI patients, 62 years and older, after controlling for the effects of demographic, medical, and psycho-social factors, the most powerful predictor of patient participation in a rehabilitation program was the primary physician’s recommendation [2]. This effect was strongest among the oldest patients and among women. The latter finding should not be surprising, since many cardiac patients after an MI are fearful and uncertain of the possible negative side effects of physical exercise [6]. Clearly, they will need encouragement and assurance that physical exercise, as a part of a comprehensive rehabilitation program not only has benefits, but also is safe and does not increase the risk of re-infarction. Other researchers have also shown that women and the elderly are less frequently referred to cardiac rehabilitation programs, even though program benefits have been observed in these groups [14,15]. Thus, available information on referral patterns seems to indicate a certain reluctance on the part of physicians to push for patient participation in physical rehabilitation programs [16]. Existing community studies show that fewer than one-third of MI patients receive information or counseling about cardiac rehabilitation before being discharged from the hospital. Contemporary trends toward shorter hospital stays may further reduce the likelihood of comprehensive counseling in this setting. There is a clear need to examine what systematic criteria, if any, predict referral and participation of MI patients in cardiac rehabilitation programs. The study of referral and participation patterns is complicated by the fact that different factors may influence: (1) the referral to cardiac rehabilitation by the provider, (2) the actual patient enrollment in a program, and (3) the completion of such a program. Employing data from a communitybased study, we focus primarily on two issues: (1) factors that predict referral of MI patients to cardiac rehabilitation programs at hospital discharge, and (2) factors that predict actual participation in cardiac rehabilitation 6 or 12 weeks after the hospital discharge.
2. Methods 2.1. Sample Subjects were 1671 patients, who were diagnosed with acute myocardial infarction (AMI) at admission to one of five hospitals in two mid-Michigan communities. Enrollment in the Michigan State University Inter-institutional Collaborative Heart Study occurred in two different phases, between January 1994 and April 1995 (Phase I: n 1163) and between February 1 and September 30, 1997 (Phase II: n 508). The independent Institutional Review Boards of each hospital approved the study and procedures were in accordance with each institution’s guidelines. Informed con-
sent for in-hospital and post-discharge follow-up was obtained from each patient. Of the 1671 patients diagnosed with AMI, 1475 (86% of total MI population) survived to be discharged. All of the patients were approached for in-hospital interviews, but many were physically unable to endure an interview (20%), died right after discharge (12%), were discharged early or missed (15%), refused to be interviewed (20%) or did not participate for unknown reasons (7%).
Table 1 Sample characteristics
Characteristic Sex Male Female Age 44 and under 45 thru 54 55 thru 64 65 thru 74 75 Race White Black Other (incl. Hispanic) Missing Insurance Commercial Medicare Medicaid Self-pay/none Missing Comorbidity score 1 (no other diseases) 2–3 4 Smoking status at time of MI Current smoker Not current smoker Missing Previous MI Yes No Catheter Yes No PTCA/angio Yes No CABG/bypass Yes No Rehab on discharge plan? Yes No Attended rehab? Yes No Marital status Married Not married
Patients who survived to discharge n 1475
Patients with follow-up data available n 347
950 (64%) 525 (36%)
234 (67%) 113 (33%)
147 (10%) 282 (19%) 338 (23%) 383 (26%) 325 (22%)
37 (11%) 73 (21%) 98 (28%) 79 (23%) 60 (17%)
1255 (85%) 188 (13%) 27 (2%) 5
316 (91%) 25 (7%) 6 (2%) 0
571 (40%) 747 (52%) 54 (4%) 57 (4%) 46
167 (49%) 153 (45%) 9 (3%) 13 (4%) 5
496 (34%) 764 (52%) 215 (15%)
148 (43%) 165 (48%) 34 (10%)
635 (61%) 413 (39%) 427
137 (40%) 210 (60%) 0
227 (15%) 1248 (85%)
85 (25%) 258 (75%)
1188 (80%) 287 (20%)
307 (88%) 40 (12%)
473 (32%) 1002 (68%)
148 (43%) 199 (57%)
352 (24%) 1123 (76%)
83 (24%) 264 (76%)
160 (11%) 1315 (89%)
57 (16%) 290 (84%)
— —
91 (26%) 256 (74%) 235 (68%) 85 (32%)
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
Theoretically, the first two groups (unable or died, 32%) do not introduce bias, as they are unlikely to be eligible for cardiac rehabilitation. Early discharge or refusal may create selection bias if patients had a less severe (early discharges) or more severe (interview refusals) heart attack. A comparison of the total patients surviving to discharge with those followed post discharge indicates little bias occurred (Table 1). This left a total of 449 patients (30% of 1475) who were interviewed in the hospital and identified for a follow-up interview in the community after discharge. Due to resource changes Phase I patients received mail-survey follow-up and Phase II patients received interview follow-up. For Phase I patients, the follow-up in the community involved mailed questionnaires (n 273) 3 months after discharge from the hospital. For Phase II patients, the follow-up involved telephone interviews (n 176) 6 weeks after discharge. However, responses to questions about attendance of cardiac rehabilitation were obtained from 347 subjects (175 in phase I and 172 in phase II). Table 1 shows the characteristics of all 1475 patients discharged, as well as, the sub-sample for which follow-up data are available. 2.2. Procedures Patients were identified through elevated enzyme lists generated by each hospital daily, as well as on the basis of admission logs and hospital charts on floors occupied by cardiac patients. Patients who met symptom, enzyme and/or electrocardiogram criteria were approached for enrollment. Trained research nurses visited each hospital daily and obtained informed consent from English-speaking patients, unless disoriented, on ventilators, too ill to interview, unaware of the MI diagnosis, or refused. In some cases, patients were discharged before the research nurse identified them and attempted an interview. In addition to consenting to a medical record audit, a subset of patients were interviewed still in the hospital and were asked about their past medical history, family history, basic demographic information, lifestyle choices (diet, smoking, exercise, recreational substance use), and physical functioning. Six or 12 weeks following discharge, these patients were contacted and asked about attendance at a cardiac rehabilitation program. 2.3. Measures 2.3.1. Outcome measures Outcome measures included referral to a rehabilitation program for MI and participation in such a program. Information obtained from the hospital discharge plan established whether or not a referral had been made; 160 patients (10.8% of total) were referred. This included 57 (16%) of those interviewed with follow-up. At the follow-up interviews, patients were asked if they had attended a cardiac rehabilitation program. A total of 91 (26.2%) indicated that they were attending or had finished participation in a rehabilitation program at the time of the interview. (Only two patients reported program completion and were combined
1027
with those currently attending.) The higher percentage reporting attendance of a rehabilitation program than originally referred in the hospital reflects a delay in 34 of the 91 patients who were referred 3 to 6 weeks post discharge. 2.3.2. Predictors Race/ethnicity was obtained from the hospital charts. Given the skewed distribution (85% Caucasian, 13% African Americans, 2% Hispanic), patients were categorized as White or Non-White in all later analyses. Information on primary health insurance came from hospital charts and grouped into four categories: private insurance (all carriers), Medicare (with or without private insurance), Medicaid or no insurance, and no insurance information. However, information on specific policy coverage was unavailable and it remains unclear to what degree cardiac rehabilitation programs are covered. Chart information also was used to determine if a patient had undergone cardiac catheterization, angioplasty or cardiac bypass surgery. We hypothesized that use of various procedures in-hospital may affect recommendations for subsequent care. Comorbidity was assessed using the Charlson Index [17]. The index provides information on the patient’s history of 15 chronic diseases. We modified the index by excluding the MI as a co-morbidity, since all subjects experienced one. Based on Charlson et al. [18] the remaining chronic conditions yielded a weighted score, ranging from 0 to a maximum of 11. With few cases occupying the upper end of the scale, we re-coded the Charlson Index into three categories: MI only, comorbidity score of 1–2 and comorbidity score of 3–11. A separate dichotomous variable was employed to record a patient’s prior history of myocardial infarctions. From the hospital discharge plan, we also recorded whether post-discharge appointments were scheduled with: (1) a cardiologist or cardiac surgeon (the two groups were later combined after extensive probing showed them to have very similar referral patterns); (2) a primary care physician (internist, family practice, other); or (3) an unknown provider. However, while the data from the discharge records allowed us to identify the physician a patient was referred to, we did not have information on the physician managing the patient within the hospital. A total of 36 physicians were involved in the referrals. A dichotomous variable was employed to indicate whether a case came from Phase I with a 3-month follow-up interview or Phase II with a 6-week follow-up interview. Additional information was available from a subset of patients who participated in the in-hospital interview. Only marital status (married versus not married) proved a potent predictor and is included in the data analysis. 2.3.3. Statistical analysis After initial analysis to test for significant bivariate relationships between all the categorical variables, two hierarchical and one step-wise logistic regression models were employed. The logistic models were used to predict the multivariate, joint effects of all potential predictors on referral or participation in cardiac rehabilitation. The associated
1028
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
significance tests were based on the Wald-statistic or the log-likelihood ratio. For a stepwise logistic model, we employed probability values associated with the log-likelihood ratio (P .05 for entry, P .10 for removal) to compare models with and without the examined predictor variable.
3. Results Table 1 presents descriptive information on all variables used in the analysis for the discharge sample of 1475 MI patients and the subset of 347 patients interviewed about their participation in rehabilitation. There were no significant differences in the distribution of cases according to age groups, gender and bypass surgery. However, patients with followup were slightly less likely to be minority (9% vs. 15%, 2 8.5, P 0.014), and somewhat more likely to have commercial insurance (49% vs. 40%, 2 9.3, P 0.025). They also had fewer co-morbid conditions (2 12.2, P 0.002), were less likely to be smokers (40% vs. 61%, 2 47.0, P 0.001) and more likely to report a previous AMI (25% vs. 15%, 2 16.1, P 0.005). The total enrolled patients included repeat AMIs, several of whom were previously enrolled and interviewed but were not interviewed a second time while hospitalized for the current AMI. This explains the lower rate of reported previous MI among the total group versus the followed group. In terms of treatment, patients available for follow-up were more likely to receive a catheter (89% vs. 81%, 2 12.0, P 0.001) and PTCA (43% vs. 32%, 2 14.0, P 0.001). Bivariate relationships involving the referral to a rehabilitation program were examined and revealed only the following four significant predictors. Patients scheduled for an appointment with a cardiologist or cardiac surgeon were more likely to be referred at discharge than patients with appointment to other physicians (12.7% versus 6.2%; P .001). Patients with cardiac catheterization were more likely (12.5%) to be referred than those without (4.2%; P .001), as were non-smokers (13.9%) compared to smokers (9.4%; P . 032). Finally, patients who had cardiac bypass surgery were more likely to be referred to rehabilitation than those who did not (18.0% versus 10.2%, P .001). Among the 347 respondents who provided follow-up information about their attendance in rehabilitation programs, men were more likely to participate than women (30.6% versus 16.2%, P .01). Patients with a cardiologist or cardiac surgeon appointment were also more likely to participate (29.5% versus 9.7% of other patients, P .001). Again, patients with bypass surgery had higher participation rates (46.3%) than those without (19.5%, P .01). Some of the variables obtained from the hospital interviews were also related to participation in cardiac rehabilitation programs. Married patients were more likely to participate (28.3%) than not married patients (15.2%, P .02), and higher income was related to lower participation rates (P .001). Finally, referral to a rehabilitation program at hospital
discharge is a strong predictor of later participation (54.4% versus 20.3%, P .001). Table 2 shows the results of a multivariate logistic regression predicting the odds of referral to a cardiac rehabilitation program at the time of hospital discharge. Recall that 160 of the 1475 were referred. (Only 10 cases had missing information on some of the variables in the model.) Among significant predictors the strongest included: catheterization of the MI patient, the type of provider with whom the patient has an appointment, bypass surgery, and age of the patient. In particular, the odds of referral are 2.6 times larger among catheterized patients (P .001), and almost 1.7 times higher among patients who had bypass surgery (P .01). Patients scheduled for an appointment with a cardiologist or cardiac surgeon had more than twice the odds of being referred to cardiac rehabilitation (1/.43 2.33, P .02). Patients aged 45–59 had the greatest odds of referral, two times greater than the youngest MI patients (P .05). Table 3 examines the odds of participation in a cardiac rehabilitation program. Given the differences in the time interval for the follow-up interviews, it comes as no surprise that the odds of participation are only half as large (.52, P .02) in the Phase II study group. In addition, bypass surgery remains a powerful predictor of participation: patients with Table 2 Predictors of referral at hospital discharge Predictor Phase 0 Phase I, 1 Phase II Sex 0 male, 1 female Race 0 non-Hispanic White 1 minority Age (Reference: 26–44) 45–59 60–74 75–93 Insurance: (Reference: private) Medicare Medicaid No ins. Missing Charlson Index (Reference group: MI only) Comorbidities 1–2 Comorbidites: 2 Prior MI 0 no, 1 yes Catheterization 0 no, 1 yes Angioplasty 0 no, 1 yes Bypass surgery 0 no, 1 yes Provider: (Reference group: cardiologist) Family phys. Other provider
Adjusted odds-ratio
95% CI for odds-ratio
P-value
.75
.50–1.14
.18
1.18
.82–1.71
.37
1.03
.63–1.70
.90
2.02 1.18 1.60
.99–4.13 .54–2.58 .66–3.88
.05 .68 .30
.98 1.01 .65 .18
.59–1.64 .41–2.54 .22–1.90 .02–1.37
.95 .98 .43 .10
1.29 1.34
.87–1.90 .75–2.38
.20 .32
1.13
.70–1.85
.61
2.60
1.34–5.07
.00
.80
.52–1.22
.29
1.67
1.11–2.50
.01
.43 .62
.22–.86 .34–1.11
.02 .11
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030 Table 4 Predictors of participation post discharge: stepwise results
Table 3 Predictors of participation 6 or 12 weeks post-discharge Predictor Phase 0 Phase I, 1 Phase II Sex 0 male, 1 female Race 0 non-Hispanic White 1 minority Age (Reference: 26–44) 45–59 60–74 75–93 Insurance: (Reference: private) Medicare Medicaid No insurance Missing Charlson Index (Reference: MI only) 1–2 Comorbidities 2 Comoridities Prior MI 0 no, 1 yes Catheterization 0 no, 1 yes Angioplasty 0 no, 1 yes Bypass surgery 0 no, 1 yes Provider: (Reference: cardiologist) Family physicians Other provider Discharge referral to rehab 0 no, 1 yes
1029
Adjusted 95% CI odds-ratio for odds-ratio P-value
Predictor Phase 0 3 months interview 1 6 weeks interview Marital status 0 not married 1 married Bypass surgery 0 no, 1 yes Discharge referral to cardiac rehabilitation 0 no, 1 yes
Adjusted odds-ratio
95% CI for odds-ratio
P-value
.52
.29–.92
.02
.57 1.15
.30–1.10 .42–3.16
.09 .78
1.83 1.01 1.42
.65–5.21 .32–3.15 .36–5.66
.25 .99 .62
1.00 1.22 .52 .22
.44–2.24 .21–7.06 .10–2.86 .02–2.96
.99 .83 .46 .25
.77 .41
.43–1.36 .13–1.31
.37 .13
.59
.30–1.19
.14
.81
.27–2.46
.71
1.36
.66–2.78
.40
4. Discussion
3.19
1.52–6.73
.00
.69 .31
.22–2.21 .09–1.09
.53 .07
4.07
2.07–8.04
.00
Despite the available evidence and clinical guidelines for MI patients published by such groups as the American Association of the Cardiovascular and Pulmonary Rehabilitation [3], overall participation in exercise-based cardiac rehabilitation programs remains low. Data from our study show only 16% of the MI patients were referred to a cardiac rehabilitation program at discharge, and only 26% of the patients later interviewed in the community actually participated in such a program. However, 54% of the patients who were referred at discharge did participate at the time of the interview and, in the multivariate analysis, referral emerged as the strongest predictor of participation. This finding supports those of Ades et al. [2] who showed that physicians’ recommendation are the most powerful predictors of patient participation in a cardiac rehabilitation program. We observed a “provider effect” in the referral of postMI patients. Merely scheduling an appointment with a community-based cardiologist or cardiac surgeon increased the likelihood of a patient being referred to rehabilitation at the time of discharge. While this finding attests to the fact that cardiac specialists have a greater awareness of the importance of cardiac rehabilitation programs, it is also troubling. The implication is that the less severely ill post-MI patients, treated by generalist physicians who appear to be less likely to refer for cardiac rehabilitation programs, are missing opportunities for cardio-health improvements. Our finding that married patients are more likely to participate in cardiac rehabilitation programs again underscores the importance of social support and encouragement [4,5]. A gender difference was suggested but not significant in the multivariate analysis. Other studies have shown evi-
bypass surgery had more than three times greater odds (3.19, P .001) of attending compared to patients without bypass surgery. However, while catheterization predicted referral to rehabilitation at the time of hospital discharge, it did not predict actual participation in such programs. Similarly, the “provider effect” (whether or not a patient goes to a cardiologist or primary care practitioner in the community) was absorbed once discharge referral was added as a predictor of actual participation. In fact (16.32 Wald statistic), referral at discharge had the strongest effect, with referred patients having more than 4 times greater odds of participation (P .001). The patterns did not differ when general internists and family practitioners were separated. Neither health insurance status nor comorbidity affected a person’s odds of participating in cardiac rehabilitation. In this sample, women had lower odds of participation (.57) than men. While this effect is not statistically significant (P .09), the results warrant further study. Because information from medical records was limited, we introduced additional measures from the in-hospital interview. There were 314 cases that yielded data from both their hospital and the follow-up interviews. The additional predictors included were patient marital status, household income, employment status and depression (Center for Epidemiologic Studies-
.61
.35–1.06
.08
2.38
1.15–4.94
.02
2.81
1.55–5.08
.00
4.03
2.03–7.75
.00
Depression scores obtained in the hospital interview). Given the number of variables and limited sample size, we opted for a step-wise approach to logistic modeling. Table 4 displays the final results. Four predictor variables were retained. Only marital status appears to have an additional effect on participation in cardiac rehabilitation: married patients had more than twice greater odds of participation (2.38, P .02).
1030
K. Barber et al. / Journal of Clinical Epidemiology 54 (2001) 1025–1030
dence that women are less likely to be referred and to participate in cardiac rehabilitation programs [17]. Our data contain suggestions that this gender effect may, in part, be a case of under-referral of the elderly. We observed that patients in the younger age group of 45–59 were most likely to be referred, yet women generally have their first MI 10 years later than men. These data confirm the correlation between gender and age, with older MI patients much more likely to be female. As a result, the gender effect is no longer significant in the multivariate analysis, after the effects of age have been taken into account. Overall our evidence suggests the strong role that providers and their referral activity play in the actual participation of post-MI patients in cardiac rehabilitation programs. Patients who were scheduled to visit a cardiologist or cardiac surgeon post-discharge were more likely to have had catheterization and cardiac bypass surgery. They were more likely to be male, Caucasian, under the age of 75 years, and to have private insurance. Physician influence is evident in our data. While physician attitude may not be enough to overcome patient receptiveness, motivation, or compliance behavior, it is clear that post-MI patients followed up in the community by a cardiologist or cardiac surgeon have a much greater probability of getting involved in cardiac rehabilitation. It appears that stronger encouragement needs to be provided to the broader range of MI patients who stand to benefit from cardiac rehabilitation. Acknowledgments The Blue Cross/Blue Shield of Michigan Foundation, Michigan State University Foundation-College of Human Medicine at Michigan State University, and all hospitals involved in the study provided funding for this project. Investigators Participating in the Michigan State University Inter-Institutional Collaborative Heart Study: Firas Akhrass MD, Chris Colenda PhD, Francesca Dwamena MD, Hassan El-Tamimi MD, Joseph Gardinei PhD, Ken Gaines MD, Margaret Holmes-Rovner PhD, Jill Kroll PhD, Joel Kupersmith, MD, Raj Nfitra MD, David Rovner MD, Peter Vasilenko PhD, and Ralph Watson MD, (MSU, East Lansing); Aryeh Stein PhD, Emory University, Atlanta. Kimberly Barber MS, Sue Davis RN, Del DeHart MD, (Saginaw Cooperative Hospitals, Inc., Saginaw); Saginaw Cooperative Hospitals, Inc.; Jack Ferlinz MD, Aleda E. Lutz (VAMC, Saginaw). Michelle Debemardi RN, Barbara McIntosh MD, D. Kay Taylor PhD, (Hurley Medical Center, Flint); Susan Smith MD, (McLaren Regional Medical Center, Flint). Acknowledgement of Support Abin Achrekar MPH, Fawzia Ahmed MS, Usman Alam, Elisa Arfianti, Amy Cairns, Qin Chen, East Lansing; Peter
Fattel, Syed Fazal-Ur-Rehman, V. Kandallu, Anton Koinev, Annette McLane BS, Danutta Podgorska, Gennaro Polverino, Manfred Stommel, PhD (MSU, East Lansing). Leslie Franke BS and O. David Tenny, (Saginaw Cooperative Hospitals, Inc., Saginaw). Jeanne Minor, (Rochester). Kathy Sands, (Flint).
References [1] Campbell NC, Grimshaw JM, Ritchie LD, Rawles JM. Outpatient cardiac rehabilitation: are the potential benefits being realized? J R Coll Physicians Lond 1996;30:514–9. [2] Ades PA, Waldmann ML, Polk DM, Coflesky JT. Referral patterns and exercise response in the rehabilitation of female coronary patients aged 62 years. Am J Cardiol 1992;69:1422–5. [3] American Association of Cardiovascular and Pulmonary Rehabilitation. Guidelines for cardiac rehabilitation programs 2nd ed. Champaign, IL: Kinetics Books, 1995. [4] Wallace JP, Raglin JS, Jastremski CA. Twelve months adherence of adults who joined a fitness program with a spouse vs without a spouse. J Sports Med Phys Fitness 1995;35:206–13. [5] Ginzel AR. Women’s compliance with cardiac rehabilitation programs. Prog Cardiovasc Nurs 1996;11:30–5. [6] Moore SM. Women’s views of cardiac rehabilitation programs. J Cardiopulm Rehab 1996;16:123–9. [7] Crowe JM, Runions J, Ebbesen LS, Oldridge NB, Streiner DL. Anxiety and depression after acute myocardial infarction. Heart Lung 1996;25:98–107. [8] Oldridge NB, Furlong W, Feeny D, Torrance G, Guyatt G, Crowe J, Jones N. Economic evaluation of cardiac rehabilitation soon after acute myocardial infarction. Am J Cardiol 1993;72:154–61. [9] Leitch JW, Newling RP, Basta M, Inder K, Dear K, Fletcher PJ. Randomized trial of a hospital-based exercise training program after acute myocardial infarction: cardiac autonomic effects. J Am Coll Cardiol 1997;29:1263–9. [10] Adachi H, Koike A, Obayashi T, Umezawa S, Aonuma K, Inada M, Korenaga M, Niwa A, Marumo F, Hiroe M. Does appropriate endurance exercise training improve cardiac function in patients with prior myocardial infarction? Eur Heart J 1996;17:1511–21. [11] Heldol M, Sire S. Effects of intensive exercise training on lipid levels in high risk post-MI patients. Eur Heart J 1994;15:1362–7. [12] Wenger NK. Rehabilitation of the cardiac patient. Advance Cardiovasc Med 1996;3:1–7. [13] 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;14:831–5. [14] O’Connor GT, Buring JE, Yusuf S, Goldhaber SZ, Olmstead EM. An overview of randomized trials of rehabilitation with exercise after myocardial infarction. Circulation 1989;80:234–44. [15] Lavie CJ, Milani RV. Effects of cardiac rehabilitation and exercise training programs in patients or 75 years of age. Am J Cardiol 1996;78:675–7. [16] Balady GJ, Jette D, Scheer J, Downing J. Changes in exercise capacity following cardiac rehabilitation in patients stratified according to age and gender. Results of the Massachusetts Association of Cardiovascular and Pulmonary Rehabilitation Multicenter Database. J Cardiopulmonary Rehab 1996;16:38–46. [17] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. [18] Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1993;47:1245–51.