Article Text

Original research article
Routine diversion of patients with STEMI to high-volume PCI centres: modelling the financial impact on referral hospitals
1. Elizabeth Barnett Pathak1,
2. Meg M Comins2,
3. Colin J Forsyth3 and
4. Joel A Strom4
1. 1Department of Internal Medicine, University of South Florida, Tampa, Florida, USA
2. 2Department of Health Policy and Management, University of South Florida, Tampa, Florida, USA
3. 3Department of Anthropology, University of South Florida, Tampa, Florida, USA
4. 4Department of Medicine, University of Florida College of Medicine, Jacksonville, Florida, USA
1. Correspondence to Dr Elizabeth Barnett Pathak; dr.elizabeth.pathak{at}gmail.com

## Abstract

Objective To quantify possible revenue losses from proposed ST-elevation myocardial infarction (STEMI) patient diversion policies for small hospitals that lack high-volume percutaneous coronary intervention (PCI) capability status (ie, ‘STEMI referral hospitals’).

Background Negative financial impacts on STEMI referral hospitals have been discussed as an important barrier to implementing regional STEMI bypass/transfer protocols. However, there is little empirical data available that directly quantifies this potential financial impact.

Methods Using detailed financial charges from Florida hospital discharge data, we examined the potential negative financial impact on 112 STEMI referral hospitals from losing all inpatient STEMI revenue. The main outcome was projected revenue loss (PRL), defined as total annual patient with STEMI charges as a proportion of total annual charges for all patients. We hypothesised that for most community hospitals (>90%), STEMI revenue represented only a small fraction of total revenue (<1%). We further examined the financial impact of the ‘worst case’ scenario of loss of all acute coronary syndrome (ACS) (ie, chest pain) patients.

Results PRLs were $0.33 for every$100 of patient revenue statewide for STEMI and $1.73 for ACS. At the individual hospital level, the 90th centile PRL was$0.74 for STEMI and $2.77 for ACS. PRLs for STEMI were not greater in rural areas compared with major metropolitan areas. Hospital revenue centres that would be most impacted by loss of patients with STEMI were cardiology procedures and intensive care units. Conclusions Loss of patient with STEMI revenues would result in only a small financial impact on STEMI referral hospitals in Florida under proposed STEMI diversion/rapid transfer protocols. However, spillover loss of patients with ACS would increase revenue loss for many hospitals. • QUALITY OF CARE AND OUTCOMES This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ ## Statistics from Altmetric.com ### Key messages #### What is already known about this subject? • To our knowledge, this is the first study to directly study the financial impact of transfer of ST-elevation myocardial infarction (STEMI) patients on the small non-percutaneous coronary intervention (PCI)-providing hospitals (ie, ‘STEMI referral centres’). #### What does this study add? • Our results show that the vast majority of STEMI referral centres will experience only a very small financial impact from the loss of patients with STEMI through evidence-based transfer to a high-volume PCI center. #### How might this impact on clinical practice? • Cooperation between high-volume PCI-providing hospitals and STEMI referral centers is essential to ensure optimal clinical care. Numerous barriers to timely transfer of patients with STEMI may exist in these smaller community hospitals. Knowledge that hospital finances will not be adversely impacted by ‘giving away’ patients with STEMI can empower hospital physicians. ## Introduction Coronary heart disease continues to be the leading cause of death in the USA, and an estimated 525 000 new and 210 000 recurrent myocardial infarctions (MIs) will occur in 2015.1 Rapid access to primary percutaneous coronary intervention (PCI) for ST-segment elevated myocardial infarction (STEMI) patients results in earlier myocardial reperfusion that improves short and long-term outcomes.2–9 However, only about 20% of acute care hospitals in the USA have cardiac catheterisation laboratories,10 and only a subset of these hospitals offer primary PCI on a full-time basis.10 Additionally, PCI hospitals are more likely to be located in large urban centres, leaving many rural areas without access to PCI.11 Consequently, the American Heart Association, as part of its own initiative, sponsored a conference in early 2006 on development of systems of care for patients with STEMI.12 Barriers to implementing STEMI systems of care were examined from the perspectives of non-PCI hospitals,13 PCI-capable hospitals,14 emergency services,15 physicians,16 patients17 and payers.18 In an ideal system, emergency medical services (EMS) personnel would bypass non-PCI capable hospitals and transport patients directly to a high-volume PCI centre. In some cases, non-PCI-capable hospital bypass would be not feasible or patients self-transport to non-PCI capable hospitals, but hospital transfer could still result in PCI within 120 min of first medical contact.19 In these cases, ECG diagnosis of STEMI would be performed by EMS personnel prior to emergency department (ED) arrival, or by ED personnel within 10 min of arrival; patients would then be emergently transported to a high-volume PCI centre within 30 min.13 Clinical trials in Europe8 ,9 ,20–22 and a recent trial in the USA23 have shown that the routine transfer of patients with STEMI to PCI centres is safe and feasible, and can be accomplished rapidly within recommended time-to-reperfusion guidelines. Despite this persuasive evidence, controversy exists in the USA about the potential for implementing routine regional STEMI bypass/transfer protocols. A major concern for both non-PCI capable and low-volume PCI hospitals is the possible negative financial impact from losing patients to high-volume PCI centres.12 ,13 ,15 Recent policy statements have made dramatic statements about the detriment to these hospitals, for example: “…. their very survival may be threatened.”13 However, there is little empirical data available that directly quantifies this potential financial impact. In this study, we analysed empirical data for all patients with STEMI admitted to acute-care general hospitals in Florida and statistically modelled the potential negative financial impact on non-PCI capable hospitals and low-volume PCI hospitals of losing all inpatient patient with STEMI revenue as result of the proposed regionalisation plans that would either bypass those hospitals or require rapid transfer protocols. Contrary to some of the prevailing opinion, we hypothesised that for most community hospitals (>90%), STEMI revenue represented only a small fraction of total revenue (<1%) and consequently loss of those patients would not threaten hospital viability. We further examined whether the proportion of revenue from patients with STEMI varied by rurality, hospital characteristics (PCI volume and STEMI volume), hospital revenue centre (eg, intensive care unit) or patient insurance type. Finally, a second set of analyses modelled the financial impact of loss of all acute coronary syndrome (ACS) patient revenue, given arguments that STEMI referrals would lead to a wider loss of chest pain patients to high-volume PCI centres. ## Methods ### Study population and data sources Our study population consisted of adults aged 18 years and above who were discharged from Florida acute care general hospitals with a primary diagnosis of STEMI during 2006. Our primary data source was the Florida Agency for Health Care Administration (AHCA) Hospital Discharge Database. All hospitals in Florida (except VA/Federal hospitals) are required by state law to submit a mandatory dataset each year to AHCA. These data include all hospital discharge records, and therefore do not represent a sample, but rather surveillance with 100% coverage. AHCA makes available public-use data sets that have been de-identified for patients (but not for hospitals). Each discharge record contained patient-level data on source and day of admission, demographics, insurance status/payer, primary and secondary diagnoses (up to 30), primary and secondary procedures (up to 30), length of stay, discharge destination (including vital status at discharge) and financial charges for 22 specific revenue groups. Diagnoses and procedures were recorded using detailed codes from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Hospitals were identified by a unique numeric identification number. Additional data on hospital and community characteristics were obtained from the American Hospital Association Guide24 and from the US Census of Population and Housing.25 In this paper, we use the term ‘STEMI referral hospitals’ to refer to hospitals that were non-PCI capable (no PCI procedures performed on inpatients) or were low-volume PCI centres (fewer than 200 PCI procedures performed annually). The American College of Cardiology/American Heart Association (ACC/AHA) practice guidelines for PCI designate ‘high-volume’ centres as those hospitals that perform 400+ PCIs on an annual basis.26 We calculated hospital PCI volumes by summing the total number of procedures at each hospital, regardless of principal diagnosis. Patients with STEMI were identified by the principal diagnosis field on the discharge record, based on ICD-9-CM codes (410.0–410.6 and 410.8). Patients with ACS were identified by principal diagnosis codes 410 (STEMI, NSTEMI, and unspecified acute myocardial infarction (AMI)) and 411 (unstable angina). ### Financial analyses The main outcome in this study was projected revenue loss (PRL), defined as total annual patient with STEMI charges as a proportion of total annual charges for all patients. For each individual hospital, we multiplied the raw charge data by actual cost-to-charge ratios to obtain cost estimates, which were then used to provide the best estimates of PRL. This was a necessary step because in the USA, hospital charges are often inflated values. The ratio of actual costs to reported financial charges varies within hospitals (by department) and also between hospitals. Medicare Healthcare Cost Report Information System data were used to calculate cost-to-charge ratios at the department level. Hospital systems often report many hospitals under one Medicare Provider Number (MPN). In these instances, the cost ratios reported under one MPN were applied to each hospital within that system. For the analyses of specific hospital revenue centres, we chose cardiology, intensive care unit, radiology and pharmacy due to their relevance to patients with STEMI and ACS. Under a ‘worst case’ hypothetical scenario, we assumed that all STEMI inpatient revenue would be lost, and calculated the per cent reduction in total revenue that would result. We first calculated this measure for all STEMI referral hospitals across the state as a whole, and then for several statewide strata: community rurality index, hospital type, hospital revenue centre and payer. We then calculated the STEMI PRL for each individual hospital, and examined the distribution of PRL estimates in accordance with our primary hypothesis. 95% CIs were calculated for all PRLs. Hospital-level cost-to-charge ratios were used to adjust the PRL estimates for all strata except for revenue centres, for example, cardiology procedures, where for an accurate PRL calculation the department-level cost-to-charge ratios were used.27 To calculate the PRL, adjusted charges for all hospital inpatients regardless of diagnosis were summed to form the denominator. The sum of all STEMI patient-adjusted charges formed the numerator. For the hospital revenue centre analyses, we grouped financial charges data into four categories (from 22 categories provided on the discharge record). These categories were: (1) cardiology procedures; (2) intensive or coronary care unit; (3) radiology; (4) pharmacy. Not all hospitals had charges in each revenue centre. We also examined six major payer types: (1) Medicare, including HMO; (2) Medicaid, including HMO; (3) commercial; (4) self-pay/underinsured; (5) charity; (6) other (eg, Tricare military insurance, worker's compensation). There were$0 charges for charity patients at 33 hospitals (meaning these hospitals did not report charges classified as charity; they may have been reported with self-pay).

Our secondary financial analyses focused on PRL from all ACS and followed the same methods outlined above for STEMI.

### Transfer time analyses

We used hospital street addresses and geographic information systems methods to calculate road network travel distances and transport time from each STEMI referral hospital to the nearest high-volume PCI centre, for the subset of hospitals whose annual PRL for STEMI was >1% (>$1 per$100 of total patient revenue).

### Exclusions

We excluded certain hospital types from our analyses: children's hospitals, psychiatric and behavioural health facilities, and long-term care facilities. Data were not available for Veteran's Administration facilities which may infrequently accept patients with AMI through their EDs. We also excluded eligible acute care community hospitals which did not discharge any patients with STEMI during the study period (n=6). Patients were excluded only on the basis of age (<18 years of age). We did not exclude either ‘transfers in’ or ‘transfers out’ to/from other acute care hospitals, in order to maximise the ‘worst case scenario’ approach to examining our primary hypothesis—in other words, to maximise the estimate of projected revenue lost. Three critical access hospitals28 with a total of 21 STEMI discharges did not report costs to Medicare; therefore cost-to-charge ratios for these hospitals could not be calculated and they were excluded from all analyses.

## Results

### PRL for STEMI

In this study, 2082 patients with STEMI were admitted to 112 STEMI referral hospitals with an annual PCI volume of <200 procedures/year (table 1). The typical patient was a white male admitted to the hospital through the emergency department. The most common age group was 65–84 years (44.4%), with 31.2% of patients <65 years and 19.4% of patients >85 years of age. Over 90% of patients had some type of health insurance, with 2.3% identified as charity patients and 7.2% as self-pay/underinsured. Almost 70% of these patients had a length of stay >2 days. Among the subset of patients who were eventually transferred to another (presumably larger) hospital, almost 50% were not transferred until 2+ days after admission.

Table 1

Characteristics of STEMI (n=2082) and ACS patients (n=12 228) at STEMI referral hospitals* in Florida

Characteristics of the STEMI referral hospitals are shown in table 2. The majority of these hospitals performed no PCIs (69.6%), and 93% admitted<51 patients with STEMI—equivalent to<1 patient with STEMI per week on average. According to self-reports in an American Hospital Association survey, 26.8% offered diagnostic cardiac catheterisation and 44.6% had a medical intensive care unit (MICU); however, 42.6% of hospitals did not participate in the survey. Almost half of all hospitals were for-profit (49.1%) and the majority were located in a medium or large metropolitan area (76.7%), that is, in an urban area with a population of at least 250 000. Only 10 STEMI referral hospitals were located in rural areas or small towns.

Table 2

Characteristics of STEMI referral hospitals* (n=112) in Florida

We first examined average PRL for STEMI for the state as a whole (table 3). Across all 112 STEMI referral hospitals which admitted at least 1 patient with STEMI annually, the annual PRL was $0.33 for every$100 of total patient revenue. We then divided the hospitals into groups based on specific hospital characteristics and calculated annual PRLs for each group. The PRL for rural areas ($0.24) was lower than the PRL for major metro areas ($0.33). The average PRL was greatest among hospitals with the highest PCI volume ($0.96) and highest patient with STEMI volume ($1.08). Average PRLs varied by hospital revenue centre, with the highest average PRL for cardiology services ($2.27). Average PRL was higher for Medicare patients ($0.36) than for commercial patients ($0.30). Table 3 PRLs resulting from proposed diversion of STEMI patients and ACS* patients away from referral hospitals with low or no PCI capability: summary results Annual STEMI PRLs for individual hospitals ranged from$0.01 to $3.82 (table 4). The PRL at the 90th centile was$0.74/$100 of total patient revenue. The 90th centile PRL was less than 1% ($1) in all categories of urbanicity/rurality. However, the 90th centile PRL exceeded 1% for the four hospitals with highest PCI volume ($3.82) and for the eight hospitals with highest STEMI volume ($3.82).

Table 4

PRLs resulting from proposed diversion of STEMI patients away from referral hospitals with low or no PCI capability: individual hospital results

### PRL for ACS

We examined PRL for ACS under assumption of a ‘worst case’ scenario in which STEMI diversion/rapid transfer protocols would result in the spillover loss of all acute chest pain patients from STEMI referral hospitals. There were 12 228 patients with ACS admitted to STEMI referral hospitals in Florida during the study period (table 1). Average annual PRL for ACS was over five times higher than the PRL for STEMI alone ($1.73 vs$0.32) but still <2% (table 3). Average annual PRLs exceeded 2% for hospitals with highest PCI volume ($2.53) and highest STEMI volume ($2.84), for Medicare patients ($2.24) and for the following specific revenue centres: cardiology ($8.93) and MICU/cardiac care unit (CCU) ($3.45). For individual hospitals, the 90th centile value of PRL for ACS was$2.77 (table 5). The median (50th centile) value was $1.49, indicating that > 50% of these hospitals would experience revenue loss of > 1% if all patients with ACS were lost. Median PRLs were highest for the four hospitals with highest PCI volume ($2.53) and for the eight hospitals with highest STEMI volume ($2.89). Interestingly, the 90th centile PRL for cardiology services was$14.42—indicating that non-ACS patients accounted for > 86% of cardiology revenue in the majority of these STEMI referral hospitals.

Table 5

PRLs resulting from hypothetical routine diversion of all ACS patients away from referral hospitals with low or no PCI capability: individual hospital results

### Hospitals with high STEMI PRLs

Of 112 referral hospitals which discharged one or more patients with STEMI, there were five hospitals with STEMI PRLs > 1% (table 6). All were located in metropolitan areas of >250 000 population, and all were less than a 31 min drive to the nearest high volume PCI centre. Hospitals C and E were unprofitable non-PCI capable centres with sizeable patient with STEMI volumes, located in very close proximity to high-volume PCI centres.

Table 6

Florida referral hospitals with >1% PRL from proposed STEMI patient diversions/rapid transfers (n=5)

## Discussion

We found an average PRL of 0.33% for the entire state and for individual hospitals, a 90th centile PRL of 0.74%, confirming our hypothesis that STEMI charges represented <1% of total charges for more than 90% of Florida community hospitals. As expected, PRL was highest (0.96%) in low-volume PCI hospitals which still performed more than 52 PCIs in 2006. Nonetheless, this subset consisted of only four hospitals, or 3.5% of the total sample. The lowest PRL of 0.22% occurred in the 14 hospitals which performed between 1 and 11 PCIs in 2006, while the PRL averaged 0.28% in the 78 hospitals which did not perform any PCIs in 2006.

Some fear that the implementation of regional STEMI care systems incorporating bypass/transfer protocols for PCI would irrevocably damage the financial viability of community hospitals. Rathore et al29 argue that ‘cardiovascular care is a financially attractive service for hospitals’ that provides up to 35% of total revenue, revenue which helps subsidise other necessary but less lucrative services. The authors warn that establishment of regional STEMI centres would increase the gap ‘between the cardiac ‘haves’ and the ‘have nots.’ Another article concludes that non-PCI capable hospitals are the ‘most likely to suffer a significant financial impact with the development of such systems, and their very survival may be threatened.’13 However, our study found that, even under the very worst-case scenario where all ACS in-patient revenue would be lost, average PRL for STEMI referral hospitals would remain under 2%. Furthermore, our analyses of PRL for ACS reveal that cardiology revenue centres within these hospitals receive the overwhelming majority of their revenue from non-ACS patients.

Our findings are consistent with other studies that have examined the financial impact of regionalisation on community hospitals. For example, Chappel et al30 concluded that rural hospitals would not suffer significant losses in procedure volume or revenue if specialised surgical procedures, including angioplasty, were diverted to high-volume hospitals. Another study estimated that low-volume rural PCI hospitals in Iowa would lose a combined $10.1 million in revenue if patients were regionalised to high-volume hospitals.31 However, the Iowa study characterised ‘low volume’ hospitals as any hospital performing <400 PCIs annually, much higher than our cut-off point of 200. Additionally, in our study, the highest PRLs were observed in those referral hospitals in medium and major metro areas ($0.40 and $0.33, respectively), not from those in rural locations. Within hospitals, cardiology procedures and intensive or coronary care unit revenue centres experienced the highest PRLs. Not surprisingly, cardiology procedures units showed the highest statewide PRL of any group, reaching$8.93/$100 of revenue for all patients with ACS, but only$2.27 for patients with STEMI. The PRLs in intensive or coronary care units reached $3.45 per$100 of revenue for patients with ACS. However, this figure drops to $0.59/$100 for patients with STEMI.

Still, the prevailing characterisation of cardiac services as key profit-makers is based largely on historically favourable Medicare reimbursement rates, which in recent years have helped fuel a proliferation of specialty cardiac hospitals in many states.32 In our study, the majority of patients (62%) were covered by Medicare. Total PRL for this group per $100 of STEMI charges was only$0.36, although this was slightly higher than the PRL for commercial payers ($0.30) and Medicaid ($0.23). When focusing on all patients with ACS, the PRL for Medicare reached $2.24, more than twice as high as that for commercial payers ($1.06). However, Medicare reimbursement rates for cardiac services were reduced in 2009,33 suggesting future PRL for Medicare patients will be even lower than our prediction. Of note, a recent study found that hospital financial strain caused by reduced reimbursement rates did not negatively affect the quality or timeliness of care offered to patients with AMI, regardless of patient insurance status.34

For both STEMI and ACS as a whole, PRLs were greatest for charity and self-pay/underinsured patients. Hospitals are required by law to treat these patients when they present to the ED, regardless of insurance status. However, given that in reality hospitals may collect very little revenue for these patients, diversion or rapid transfer of these patients would likely result in a net financial benefit for the referral hospital.

Critics have argued that the success of regional systems of transfer for PCI in Europe, for example as in Romania,35 cannot be generalised to the USA, where transfer distances tend to be much longer. However, regional systems in North Carolina and Minnesota have achieved promising results, even with patients in rural areas who have had to be transported over considerable distances.36 ,37 The total costs and benefits of implementing such systems within the USA should be examined more closely to facilitate policy decisions.

One limitation of our study is that we were unable to account for the impact of routine STEMI transfer on patient costs. A system of transfer could result in added financial burdens for patients if they end up receiving care from physicians who do not participate in their insurance network.18 Such a system would need to establish protocols to prevent undue financial obligations for patients. Furthermore, a system of routine transfer of patients with STEMI would need to ensure all participating hospitals are rewarded for achieving efficient transfer times. A recent study in one hospital discovered that while achieving reduced door-to-balloon time decreased costs, it also resulted in lower payments. Consequently, the financial benefits of achieving greater efficiency and quality of care accrued solely to payers in this instance.38 Fenter et al18 recommend establishing a system where a single payment is shared among ‘the referring, transporting and receiving providers’, so as to encourage coordination rather than competition between providers.

Our study population of patients with STEMI was defined on the basis of ICD-9-CM diagnosis codes, which may have resulted in a small degree of misclassification bias in our study.

Despite the advantages of early PCI for treating STEMIS, the costs of creating PCI capability in non-PCI capable hospitals can be prohibitive. Moreover, while expanding PCI capability to those hospitals that already have catheterisation laboratories has been shown to be feasible,39 this strategy will be likely to increase geographic access only minimally.40 Ultimately, a coordinated system of care for STEMIs will need to rely on both referral and receiving hospitals. A successful system can implement safeguards to protect the revenue of non-PCI capable hospitals. For example, the Boston EMS routinely transports patients with STEMI to PCI-capable hospitals, but its point of entry plan classifies these hospitals as ‘STEMI hospitals’ and PCI-referral hospitals as ‘cardiac’. This helps ensure non-PCI capable hospitals continue to receive non-STEMI cardiac patients.41

STEMI referral hospitals are likely to continue to play a key role in treating STEMIs even if routine protocols for transfer to STEMI-receiving hospitals are implemented. Some patients with STEMI will have contraindications for PCI, making transfer unnecessary. In situations where patients presenting to a non-PCI capable hospital cannot undergo PCI within 120 min of first medical contact, treatment with fibrinolytics remains the recommended therapy unless contraindicated.19 Finally, there may be therapeutic and economic advantages in using facilitated PCI, or using PCI following fibrinolytic therapy. Coleman et al42 assert that major adverse cardiac end points and hospital costs were reduced significantly in facilitated versus primary PCI for transferred patients. Importantly, though, randomised controlled trials have not conclusively demonstrated advantages to facilitated PCI.43 Regardless of the treatment option, the future success of a system of routine transfer of patients to PCI-capable facilities will depend on mutually beneficial collaboration between hospitals that de-emphasises aggressive competition for revenues.

In conclusion, in this study we focused solely on the financial detriments to implementing regional systems of STEMI care in the USA. There are many other types of barriers to widespread adoption of STEMI transfer protocols. Transfer travel times are one important potential barrier.39 In Florida, a recent study in Florida found that the majority of potential STEMI referral hospitals were capable of transferring patients within guideline-recommended time frames.44 However, in other regions of the USA, rapid transfer may not always be possible.

Finally, future research is needed to confirm the long-term cost-savings and patient benefits from STEMI systems of care from a system-wide or national perspective. In the UK, the National Health Service found that “PCI appears to be more cost effective than thrombolysis for people with AMI.”45 The National Infarct Angioplasty Project in the UK estimated a net benefit of 333.5 million pounds from the implementation of a national system with routine transfer for PCI.46

## Acknowledgments

The authors would like to acknowledge Jean Paul Tanner and Gabriella Anic for their assistance with preparation of the hospital transfer travel time database for Florida.

View Abstract

## Footnotes

• Contributors EBP and JAS conceived the study. All authors contributed to data analyses and report writing.

• Funding This study was funded by a Grant-in-Aid (085574E) from the American Heart Association, Southeast Affiliate (Dr. Pathak, PI).

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.

• Data sharing statement The data used in this study are publicly available from the Florida State Agency for Health Care Administration.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.