Cardiac Rehabilitation Use After Percutaneous Coronary Intervention

Among patients with coronary artery disease, participation in outpatient cardiac rehabilitation (CR) is associated with improved quality of life and reduced rates of readmission and cardiovascular mortality (1). For these reasons, CR is recommended in international guidelines (2,3). Despite its proven benefit, the use of CR remains poor with historical enrollment rates between 20–40% (4–6).

There are many reasons for this gap in CR utilization, including low rates of referral. Therefore, over the past decade there has been increased emphasis on improving rates of CR referral. However, there is a dearth of research examining the link between CR referral and utilization, and the patient factors associated with downstream utilization. Identification of such factors may help inform healthcare policies and hospital initiatives focused on improving CR use, not just referral.

In this context, we used a multi-center registry of PCIs performed in the state of Michigan combined with administrative claims to identify demographic, procedural, geographic, and socioeconomic factors associated with CR use after PCI.

Methods

We linked 2 data sources for this research. The first was the Blue Cross Blue Shield of Michigan Cardiovascular Consortium (BMC2) clinical PCI registry. BMC2 is a prospective, multicenter, statewide registry of all patients who undergo PCI at all non-federal hospitals in Michigan. The second data source was the Michigan Value Collaborative (MVC), a collaborative focused on improving the value of care in the state of Michigan. MVC developed and maintains a validated claims-based registry with 90-day price-standardized episodes of care from Medicare fee-for-service (FFS) and Blue Cross Blue Shield of Michigan preferred provider organization (BCBSM PPO) administrative claims (7). All clinically-related claims within 90 days after discharge from the index hospitalization or procedure were included in the episode.

We linked BMC2 PCI records to MVC’s 90-day episodes of care where PCI occurred through indirect matching of the index PCI procedure using multiple variables including hospital and operator National Provider Identifier numbers; admission, discharge, and procedure dates for the index hospitalization; and patient gender and date of birth.

Using the linked dataset, we evaluated consecutive patients who underwent PCI, including both inpatient and outpatient PCI, between January 2012 and October 2016, at 47 PCI-capable hospitals in Michigan and were discharged home. Referral to CR prior to PCI discharge was obtained from the clinical registry (8). We excluded patients who were deemed ineligible for CR referral as defined by the National Cardiovascular Data Registry CathPCI data dictionary v4.4 (8). We also excluded patients with a missing residential ZIP code.

CR use within the 90 days following discharge was obtained from administrative claims, based on the following coding: Current Procedural Terminology codes (93797 and 93798), Healthcare Common Procedure Coding System codes (G0422 and G0423) and revenue center code 943.

Among patients referred for CR, we assessed the association between CR attendance and patient factors including demographics, clinical characteristics, comorbidities, insurance status, and travel distances (Figure). Travel distances from the site of PCI to the nearest CR facility (>2 miles vs. ≤2 miles), and distance from the centroid of the patient’s ZIP code to the nearest CR facility were included in the models. Geographic distances were determined using the Google Maps application program interface through the R package ggmap. A list of CR facilities was obtained from our institution’s CR scheduling center.

Finally, using the patient’s residential ZIP code, we assessed whether publicly-available measures of ZIP code-level socioeconomic status were individually associated with CR attendance after adjusting for patient characteristics. Socioeconomic measures included the Area Deprivation Index (9) and ZIP code-level educational and poverty status obtained from the American Community Survey. Of note, patient ZIP codes were mapped to ZIP code tabulation areas, the geographic unit used by the American Community Survey. The Area Deprivation Index is a composite measure of neighborhood disadvantage, with a higher number indicating greater neighborhood disadvantage (9).

Statistical Analysis

We developed logistic regression models using robust standard errors accounting for clustering at hospitals to evaluate the association between the previously described covariates and the use of CR among referred patients. The association between covariates and CR use was expressed as odds ratios and 95% confidence intervals adjusted for all other patient factors included in the model. All analyses were performed using R version 3.2.1.

Results

Of 42,334 PCI episodes between January 2012 and October 2016, 30,075 (73.1%) were discharged alive to their home with a referral for CR. Of these, 26,168 (87.0%) had a valid residential ZIP code available and formed the study cohort. A total of 8,246 (31.5%) patients attended at least one CR session within 90 days after discharge.

Patients were more likely to attend CR with increasingly acute presentations for PCI such as STEMI and NSTEMI (Figure 1). The presence of comorbidities was generally associated with decreased odds of attending CR. Compared with patients insured by BCBSM PPO insurance, those with Medicare FFS insurance were less likely to attend CR after a referral was made (aOR: 0.81; 95% CI 0.72–0.90; p<0.001; Figure). Among patients with Medicare FFS, patients who are eligible for both Medicare and Medicaid were significantly less likely to attend CR after a referral was made as compared with those not eligible for either (aOR: 0.44; 95% CI 0.38–0.51; p<0.001).

Figure 1: Factors associated with cardiac rehabilitation use after percutaneous coronary intervention.

A) Patient factors associated with cardiac rehabilitation use. B) Individual ZIP code-level socioeconomic factors associated with cardiac rehabilitation use after adjusting for patient-level factors. The adjusted odds ratio for each ZIP code-level socioeconomic variable was obtained by individually adding each variable to the base model adjusting for patient-level characteristics. Dual eligibility refers to being eligible for both Medicare and Medicaid insurance. Abbreviations: CR=cardiac rehabilitation; NSTEMI=non-ST-elevation myocardial infarction; PCI=percutaneous coronary intervention; SA=stable angina; STEMI=ST-elevation myocardial infarction; USA=unstable angina.

The distance from a patient’s ZIP code centroid to the nearest CR was not significantly associated with increased odds of attending CR (aOR: 1.00; 95% CI 0.98–1.02; p=0.926). Patients who underwent PCI at sites where the nearest CR facility was >2 miles away were significantly less likely to attend CR compared with patients who underwent PCI at sites where the nearest CR facility was ≤2 miles away (aOR: 0.27; 95% CI 0.14–0.55; p<0.001). Patients living in ZIP codes with higher levels of educational attainment were significantly more likely to attend CR (Figure). Patients living in ZIP codes with a higher proportion of families below 125% of the federal poverty level or a higher Area Deprivation Index were associated with a trend towards lower odds of attending CR (Figure).

Discussion

Existing health care policies have succeeded in increasing CR referral rates following PCI. However, this success has not translated into high CR utilization, which remains below one-in-three patients following PCI. Additionally, CR utilization was highest among patients with private insurance and who had their PCI at a healthcare site with a closely located (and possibly affiliated) CR facility. However, the distance between a patient’s home ZIP code and the nearest CR facility was not associated with CR attendance, whereas the distance from the PCI facility to the nearest CR facility was, suggesting that the distance between the PCI facility and the nearest CR facility may be more important than the distance from the patient’s ZIP centroid to the nearest CR facility. Finally, multiple socioeconomic factors were associated with CR attendance. Taken together, these findings suggest that simply improving rates of referrals may only partially improve downstream CR utilization.

Only one-third of patients who received a referral for CR after PCI attended at least one session—an estimate that is consistent with prior research (5). This gap highlights the need for novel quality improvement initiatives focused on CR use even after prompt in-hospital referral. Similar to prior research, we found that patient insurance status and geographic socioeconomic factors were significantly associated with CR attendance (5). For instance, patients insured by BCBSM PPO, a private insurer, were more likely to attend CR compared with patients insured by Medicare FFS. We speculate that the reason for these differences is multifactorial. For instance, BCBSM PPO plans may have different out-of-pocket costs for patients. If a patient attends a typical 36-session CR program, even a $30 co-pay per session adds up to >$1,000 dollars. Furthermore, attending CR may interfere with employment, placing further financial stress on patients. Indeed, some area-level socioeconomic factors are strongly associated with CR attendance.

Insurers, including Medicare, have considered various ways of incentivizing CR use. In 2016, Medicare announced the Cardiac Rehabilitation Incentive Payment Model, where Medicare would pay hospitals for each session of CR that patients attended after treatment for acute myocardial infarction or coronary artery bypass graft surgery. This program was designed to incentivize hospitals to invest in initiatives aimed at improving CR use. However, this incentive program was cancelled in December 2017 (10). Novel payment models, such as bundled payments, may also incentivize CR use given that CR has been associated with decreased hospitalizations which are an important driver of payment variation in PCI bundles of care (1).

Limitations

Our findings should be considered in the context of some important limitations. First, our findings were limited to a single state with a long-standing quality improvement program, thus limiting the generalizability of our findings to other states. Second, we were only able to evaluate CR use in patients insured by Medicare FFS or BCBSM PPO. However, these represent two large insurers in the state of Michigan with diverse patient populations. Third, we only had access to area-level, rather than patient-level, socioeconomic factors and geographic distances which may have different associations with CR utilization and may mask the extent of patient-level variability, potentially subjecting these analyses to the ecological fallacy and misclassification bias.

Conclusions

These findings are instructive to hospitals, physicians, and other policymakers seeking to improve CR use after PCI. From a policy perspective, novel insurance designs that incentivize both CR referral and CR use may be beneficial given the association between insurance type and CR referral and use. Moreover, interventions aimed at reducing socioeconomic disparities, such as financially incentivizing patients to attend CR may prove beneficial. Further research is needed to understand the mechanisms by which socioeconomic factors influence CR attendance, so that policies are more precisely designed to improve CR attendance.

Abbreviations

PPO

preferred provider organization

CR

cardiac rehabilitation

FFS

fee-for-service

MVC

Michigan Value Collaborative

NCDR

National Cardiovascular Data Registry

PCI

percutaneous coronary intervention