Rates, predictors and variability of interhospital transfers: A national evaluation

Among transferred patients, the top 5 primary diagnoses at time of transfer included AMI (12.2%), congestive heart failure (CHF) (7.2%), sepsis (6.6%), arrhythmia (6.6%), and pneumonia (3.4%). Comorbid conditions most commonly present in transferred patients included CHF (52.6%), renal failure (51.8%), arrhythmia (49.8%), and chronic obstructive pulmonary disease (COPD; 37.0%). The most common day of transfer was day after admission (hospital day 2, 24.7%), with 75% of transferred patients transferred before hospital day 6 (Appendix B).

After adjusting for all other patient and hospital characteristics and clustering by hospital, the following variables were associated with greater odds of transfer: older age, male sex, nonblack race, non-Medicaid co-insurance, higher comorbidity (HCC score), lower DRG-weight, and fewer hospitalizations in the prior 12 months. Beneficiaries also had greater odds of transfer if initially hospitalized at smaller hospitals, nonteaching hospitals, public hospitals, at hospitals in the Northeast, those with fewer specialty services, and those with a low CMI (Table 2).

In examining the between-hospital variability in IHT, our unadjusted model estimated an average transfer rate of 1.79%, and showed a variance estimate of 1.33 (P=0.009), demonstrating that 95% of hospitals have transfer rates between 0.83% and 3.80%. The variance estimate increased by 19% to 1.58 (P=0.009) when adjusting for patient characteristics. After adjusting for hospital characteristics, variance decreased by 83% to 0.28 (P=0.01), showing 95% of hospitals have transfer rates between 1.26% and 2.54% (Figure 2).

DISCUSSION

In this nationally representative study of 6.6 million Medicare beneficiaries, we found that 1.5% of patients were transferred between acute care facilities and were most often transferred prior to hospital day 6. Older age, male sex, nonblack race, higher medical comorbidity, lower DRG weight, and fewer recent hospitalizations were associated with greater odds of transfer. Initial hospitalization at smaller, nonteaching, public hospitals, with fewer specialty services were associated with greater odds of transfer, while higher CMI was associated with a lower odds of transfer. The most common comorbid conditions among transferred patients included CHF, renal failure, arrhythmia, and COPD; particularly notable was the very high prevalence of these conditions among transferred as compared with nontransferred patients. Importantly, we found significant variation in IHT by region and a large variation in transfer practices by hospital, with significant variability in transfer rates even after accounting for known patient and hospital characteristics.

Among our examined population, we found that a sizable number of patients undergo IHT—more than 100,000 per year. Primary diagnoses at time of transfer consist of common inpatient conditions, including AMI, CHF, sepsis, arrhythmia, and pneumonia. Limited prior data support our findings, with up to 50% of AMI patients reportedly undergoing IHT,^3-5 and severe sepsis and respiratory illness reported as common diagnoses at transfer.¹¹ Although knowledge of these primary diagnoses does not directly confer an understanding of reason for transfer, one can speculate based on our findings. For example, research demonstrates the majority of AMI patients who undergo IHT had further intervention, including stress testing, cardiac catheterization, and/or coronary artery bypass graft surgery.^5,26 Thus, it is reasonable to presume that many of the beneficiaries

transferred with AMI were transferred to receive this more specialized cardiac care. We further found the majority of patients are transferred prior to hospital day 6 with the highest prevalence on day 2, supporting the hypothesis that these patients may be transferred for receipt of specialty services for their admission diagnosis. However, we cannot prove this presumption, and for other conditions, such as pneumonia, the plan after IHT is less obvious. There are numerous possible reasons for transfer,¹ including patient preference and prior affiliation with receiving hospital. Further research is required to more fully define these reasons in greater detail.

We additionally found that certain patient characteristics were associated with greater odds of transfer. Research suggests that transferred patients are “sicker” than nontransferred patients.^1,11 Although our findings in part confirm these data, we paradoxically found that higher DRG-weight and 4 or more hospitalizations in the past year were actually associated with lower odds of transfer. In addition, the oldest patients in our cohort (85 years or older) were actually less likely to be transferred than their slightly younger counterparts (75 to 84 years). These variables may reflect extreme illness or frailty,²⁷ and providers consciously (or subconsciously) may factor this in to their decision to transfer, considering a threshold past which transfer would confer more risk than benefit (eg, a patient may be “too sick” for transfer). Indeed, in a secondary analysis without hospital characteristics or comorbidities, and with fixed effects by hospital, we found the highest rates of IHT in patients in the middle 2 quartiles of DRG-weight, supporting this threshold hypothesis. It is also possible that patients with numerous hospitalizations may be less likely to be transferred because of familiarity and a strong sense of responsibility to continue to care for those patients (although we cannot confirm that those prior hospitalizations were all with the same index hospital).

It is also notable that odds of transfer differed by race, with black patients 17% less likely to undergo transfer compared to whites, similar to findings in other IHT studies.¹¹ This finding, in combination with our demonstration that Medicaid patients also have lower odds of transfer, warrants further investigation to ensure the process of IHT does not bias against these populations, as with other well-documented health disparities.^28-30

The hospital predictors of transfer were largely expected. However, interestingly, when we controlled for all other patient and hospital characteristics, regional variation persisted, with highest odds of transfer with hospitalization in the Northeast, indicating variability by region not explained by other factors, and findings supported by other limited data.³¹ This variability was further elucidated in our examination of change in variance estimates accounting for patient, then hospital, characteristics. Although we expected and found marked variability in hospital transfer rates in our null model (without accounting for any patient or hospital characteristics), we interestingly found that variability increased upon adjusting for patient characteristics. This result is presumably due to the fact that patients who are more likely to be transferred (ie, “sick” patients) are more often already at hospitals less likely to transfer patients, supported by our findings that hospital CMI is inversely associated with odds of transfer (in other words, hospitals that care for a less sick patient population are more likely to transfer their patients, and hospitals that care for a sicker patient population [higher CMI] are less likely to transfer). Adjusting solely for patient characteristics effectively equalizes these patients across hospitals, which would lead to even increased variability in transfer rates. Conversely, when we then adjusted for hospital characteristics, variability in hospital transfer rates decreased by 83% (in other words, hospital characteristics, rather than patient characteristics, explained much of the variability in transfer rates), although significant unexplained variability remained. We should note that although the observed reduction in variability was explained by the patient and hospital characteristics included in the model, these characteristics do not necessarily justify the variability they accounted for; although patients’ race or hospitals’ location may explain some of the observed variability, this does not reasonably justify it.

This observed variability in transfer practices is not surprising given the absence of standardization and clear guidelines to direct clinical IHT practice.¹⁷ Selection of patients that may benefit from transfer is often ambiguous and subjective.⁶ The Emergency Medical Treatment and Active Labor Act laws dictate that hospitals transfer patients requiring a more specialized service, or when “medical benefits ... outweigh the increased risks to the individual...,” although in practice this provides little guidance to practitioners.¹ Thus, clearer guidelines may be necessary to achieve less variable practices.

Our study is subject to several limitations. First, although nationally representative, the Medicare population is not reflective of all hospitalized patients nationwide. Additionally, we excluded patients transferred from the emergency room. Thus, the total number of patients who undergo IHT nationally is expected to be much higher than reflected in our analysis. We also excluded patients who were transferred more than once during a given hospitalization. This enabled us to focus on the initial transfer decision but does not allow us to look at patients who are transferred to a referral center and then transferred back. Second, given the criteria we used to define transfer, it is possible that we included nontransferred patients within our transferred cohort if they were discharged from one hospital and admitted to a different hospital within 1 day. However, on quality assurance analyses where we limited our cohort to only those beneficiaries with corresponding “transfer in” and “transfer out” claims (87% of the total cohort), we found no marked differences in our results. Additionally, although we assume that patient transfer status was coded correctly within the Medicare dataset, we could not confirm by individually examining each patient we defined as “transferred.” However, on additional quality assurance analyses where we examined randomly selected excluded patients with greater than 1 transfer during hospitalization, we found differing provider numbers with each transfer, suggesting validity of the coding. Third, because there are likely many unmeasured patient confounders, we cannot be sure how much of the between-hospital variation is due to incomplete adjustment for patient characteristics. However, since adjusting for patient characteristics actually increased variability in hospital transfer rates, it is unlikely that residual patient confounders fully explain our observed results. Despite this, other variables that are not available within the CMS or AHA datasets may further elucidate hospital transfer practices, including variables reflective of the transfer process (eg, time of day of patient transfer, time delay between initiation of transfer and patient arrival at accepting hospital, accepting service on transfer, etc.); other markers of illness severity (eg, clinical service at the time of index admission, acute physiology score, utilization of critical care services on arrival at receiving hospital); and other hospital system variables (ie, membership in an accountable care organization and/or regional care network, the density of nearby tertiary referral centers (indicating possible supply-induced demand), other variables reflective of the “transfer culture” (such as the transfer rate at the hospital or region where the attending physician trained, etc.). Lastly, though our examination provides important foundational information regarding IHT nationally, this study did not examine patient outcomes in transferred and nontransferred patients, which may help to determine which patients benefit (or do not benefit) from transfer and why. Further investigation is needed to study these outcomes.