Interrupted Time Series Analysis of Donor Heart Use Before and After the 2018 UNOS Heart Allocation Policy Change

Abstract

Background:

Donor heart scarcity remains the fundamental barrier to increased transplant access. We examined whether 2018 United Network for Organ Sharing (UNOS) policy changes have had an impact on donor heart acceptance rates.

Methods and Results:

We performed an interrupted time series analysis in UNOS to evaluate for abrupt changes in donor heart-acceptance rates associated with the new policy. All adult donor offers were evaluated between 2015 and 2021 (n = 66,654 donors). Donor volumes and transplants increased during this period, but the donor acceptance rate declined significantly from 31% in quarter 3 of 2018 to 26% acceptance in quarter 3 of 2021 (slope change −0.4% per quarter; P < 0.001). We identified 2 trends associated with this decline: (1) a growing number of donors with high-risk features, and (2) decreased acceptance of donors with certain high-risk features in the new allocation system.

Conclusions:

Heart transplant volumes have increased in recent years as a result of increased donor volumes, but donor heart acceptance rates began decreasing under the current allocation system. Changes in the donor pool and acceptance patterns for certain donor-risk features may explain this shift and warrant further evaluation to maximize donor heart use.

Heart transplant remains the gold standard treatment for end-stage heart failure, providing patients with increased survival time and improved quality of life. However, organ scarcity remains the fundamental barrier to transplant access. The number of candidates awaiting heart transplant has grown rapidly in recent years, increasing by 42.6% between 2008 and 2019, and more than 7000 are now actively awaiting transplant.¹ In contrast, although the number of donor hearts has also grown over this period, and the number of annual transplants has increased, only ~ 3500 transplants were performed in 2020.¹ This profound supply-and-demand mismatch leads to prolonged wait times that are associated with increased chances of mortality, clinical decompensation and other adverse events.²

Previous work by Khush et al. demonstrated that, despite organ shortage and increasing demand for transplant, the percentage of hearts accepted for transplant declined from 2000 to 2010; only 1 in 3 hearts were accepted.³ There are many reasons for an organ to be declined for transplant, including donor demographics, such as age, donor-recipient size mismatch or medical comorbidities of the donor, and decisions about donor selection are complicated by the uncertain impact of multiple donor factors on post-transplant outcomes.

In 2018, the Organ Procurement and Transplant Network changed the criteria for heart allocation to improve stratification of adult patients awaiting transplant and decrease waitlist mortality. Thus far, these changes have achieved the desired impact on waitlist mortality,^4,5 but there have been numerous changes in waitlist management practices as a result.^1,2,5–7 We sought to understand whether practice changes associated with the new allocation system have also impacted rates of donor-heart acceptance.

Methods

Study Population and Data Source

We conducted a retrospective analysis using deceased-donor data from adult donors in the United Network for Organ Sharing (UNOS) registry. We included the 3 years before and after the October 2018 policy change (2015–September 2021) in our analysis. This study was deemed exempt by the University of Utah Institutional Review Board.

Outcome Definition and Comorbidity Selection

UNOS categorizes organs into 1 of 6 dispositions: authorization not requested, authorization not obtained, organ not recovered, recovered not for transplant, recovered for transplant but not transplanted, and transplanted. Our primary outcome was the percentage of organs designated as transplanted as a percentage of authorized donors during each quarter. We were also interested in whether acceptance patterns differed on the basis of perceived high-risk donor features. For this, we selected several variables a priori from the available data, including advanced donor age (> 40 years), history of intravenous drug use, history of cocaine use, history of tobacco use, history of diabetes, history of heavy alcohol use, terminal creatinine level, high Public Health Service risk for blood-borne disease, cardiac arrest, class II obesity (body mass index > 35 kg/m²), left ventricular ejection fraction (EF) < 50%, and donor natural circumstance of death (vs suicide, homicide, etc.). We excluded hepatitis C due to concurrent changing policies affecting hepatitis C-positive heart transplantation, which would violate necessary assumptions for time series analysis.

Statistical Methods

We used interrupted time series analysis to measure changing patterns of donor heart acceptance before and after the allocation policy change. Time series analysis provides quasi-experimental inference about whether a policy change has had a causal impact on the outcome of interest, in this case, the change in slope of the donor acceptance rate per quarter in the postallocation period. No lag time was introduced in the model, and no autocorrelation was assumed. We also performed time series analysis for donors with each of the above-mentioned high-risk features by using a 2-sided Bonferroni-adjusted alpha level of 0.004 to account for multiple testing. Analyses were performed in Stata, version 17.0 (StataCorp, College Station, TX).

Results

Demographic Trends

Identified potential donors numbered 33,271 in the preallocation period, and 33,383 potential donors were identified in the postallocation period. Several donor factors were significantly different after the policy change (Table 1). Of these, potentially relevant changes included an increase in donor anoxic deaths (46% vs 40%) and a slight increase in tobacco history (25% vs 23%).

Table 1.

Demographics of Deceased Donors

Characteristic	Prior to policy change (2015–2018) n = 33,271	After policy change (2018–2021) n = 33,383	P value
Mean age of donor (SD)	43 (15)	44 (15)	<0.001
Male sex	20032 (60.2%)	20386 (61.1%)	0.023
Circumstance of death	4390 (13.2%)	3794 (11.4%)	<0.001
MVA	3290 (9.9%)	2862 (8.6%)
Suicide	1499 (4.5%)	1229 (3.7%)
Homicide	4676 (14.1%)	5879 (17.6%)
Non-MVA	14708 (44.2%)	15367 (46.0%)
Death from natural causes	4708 (14.2%)	4248 (12.7%)
Unknown or other	13301 (40.0%)	15359 (46.0%)	<0.001
Anoxia as cause of death	1826 (6.6%)	1688 (6.9%)	0.16
Cardiac arrest	7200 (85.1%)	7030 (86.3%)	0.032
Hypertension	4199 (12.7%)	4644 (14.1%)	<0.001
Diabetes	20032 (60.2%)	20386 (61.1%)	0.023
Mean BMI (SD)	29 (7)	29 (7)	<0.001
Terminal creatinine (SD)	1.6 (2)	1.7 (2)	<0.001
Tobacco history	7433 (22.8%)	7948 (24.5%)	<0.001
Heavy alcohol	6767 (20.3%)	7456 (22.3%)	<0.001
IV drug use	4416 (13.3%)	4726 (14.2%)	<0.001
Cocaine	7499 (22.5%)	7980 (24.4%)	<0.001
CMV	20862 (62.7%)	20945 (62.7%)	0.92
EBV	30206 (90.8%)	30211 (90.5%)	0.2
HCV	1722 (5.2%)	2277 (6.8%)	<0.001
PHS high risk for bloodborne disease	8933 (26.8%)	8928 (26.7%)	0.76

BMI, body mass index; CMV, cytomegalovirus; EBV, Epstein-Barr virus; HCV, hepatitis C virus; IV, intravenous; MVA, motor vehicle accident; PHS, public health service; SD, standard deviation.

Overall Acceptance Rates

From January 2015 to September 2021, the number of potential donors increased from 1872 to 2926 per quarter, whereas adult heart transplants increased less rapidly, from 549 to 775 per quarter, due to a decreased rate of acceptance/offer (Fig. 1). Prior to the allocation change, quarterly percentage of acceptance had been increasing from 29% in 2015 to 31% in quarter 3 of 2018. After the policy change, the percentage of acceptance began decreasing to 26% in quarter 3 of 2021 (slope change −0.4% per quarter; P < 0.001) (Fig. 1).

Fig. 1.

Heart acceptance rates before and after policy change. Visual take home graphic.

Acceptance Rates for Donors With High-Risk Features

The allocation policy was associated with significant decreases in the acceptance rates of donors with certain risk features. After the policy change, donor hearts were accepted at a decreased rate if donors had a history of intravenous drug use (slope change −1.5 per quarter; P < 0.001), anoxia as cause of death (slope change −0.7; P <.001), heavy alcohol use (slope change −0.7; P < 0.001), cocaine use (slope change −0.9; P < 0.001), elevated public health service infection risk (slope change −0.6; P = 0.002), or tobacco use (slope change −0.5; P < 0.001) (Fig. 2). There was no significant change in the acceptance rates of donors aged > 40, history of diabetes, terminal creatinine level > 2 mg/dL, class II obesity, history of cardiac arrest, LVEF < 50%, or donor natural circumstance of death.

Fig. 2.

Heart acceptance rates before and after policy change by risk factor. IV, intravenous; PHS, public health service.

Discussion

Multiple initiatives have been underway to increase the availability of donor heart organs in the United States, including increased use of hepatitis C-positive hearts and the use of hearts donated after cardiac death. These initiatives, in addition to population growth and an increased number of donations after opioid overdose, have led to a significant increase in the number of available donors and subsequent heart transplantations over the past decade.¹ Still, there remains a critical shortage of donor organs relative to the number of patients awaiting transplant, and there is an imperative to optimize the use of potential donors to maximize benefit.

Changes in the United States adult heart allocation policy in 2018 have resulted in reduced waitlist times and waitlist mortality.^4,8 Nevertheless, our current analysis suggests that the new policy has also been associated with a significant decline in the rate of donor-heart acceptance, down from 31% of donors in quarter 2 of 2018 to 26% of donors in quarter 3 of 2021 (Fig. 1).

One possible explanation for this is that the expanded pool of donors seen over the past decade increasingly includes donors with marginal or high-risk features who would have been declined under either allocation policy. For instance, we observed a significant increase in the percentage of donors who died from anoxic injury from 40% to 46%; they are more likely than the average donor to be declined out of concern for ischemic cardiac injury. This possibility is consistent with our findings, but our data also point to additional mechanisms related to changing risk-tolerance for several donor features.

We analyzed donor features potentially associated with higher risk and observed several where the rate of acceptance had been rising before the 2018 but began decreasing significantly immediately following the policy change (Fig. 2). These features include anoxic death (which would have a multiplicative effect with the increased prevalence of this among donors) in addition to intravenous drug use, heavy alcohol consumption, cocaine use, behaviors resulting in public health service elevated infectious risk, and a history of tobacco use.

The reason donor offers with these features are now more likely to be declined is unknown. However, multiple other waitlist practice changes have been observed under the revised allocation policy, including a large increase in the use of temporary mechanical circulatory support and a decrease in left ventricular assist device use as a bridge to transplant. It is possible that we are seeing a similar pattern with donor-heart acceptance rates, where the expected wait time for sicker patients has decreased so that providers see less urgency to accept potentially marginal donor hearts. These offers would then go to patients lower on the waiting list, but providers for lower risk ambulatory patients may be reluctant to accept higher risk hearts when patients are reasonably stable. It would be important to determine whether increasing acceptance rates for donors with these features to the level typical before the allocation change could result in further increases in transplant volumes while maintaining favorable post-transplant outcomes. Further research will be needed to ascertain the exact reasons why acceptance rates have decreased for these donors and which organs with higher-risk features can be safely transplanted.

Limitations

This study was limited by its retrospective and observational design. Time-series analyses are used to approximate causal inference, but bias may exist if other changes are occurring around the same period external to the allocation-policy change. Moreover, the donor features we observed that had declining acceptance may be confounded by other donor or recipient features. Importantly, the overlap of the study time frame with the COVID-19 pandemic cannot be ignored; however, we observed a change in acceptance rates that began prior to the onset of the pandemic.

Conclusions

Donor offers have increased in recent years, but the rate of donor heart acceptance decreased following the 2018 UNOS heart-allocation policy change. Organs with certain risk features are now more likely to be declined, suggesting additional practice changes associated with the policy change that have not been identified previously. Whether these changes could be addressed to increase donor organ use and decrease waiting list times without worsening post-transplant outcomes should be explored.

Lay Summary

• Rates of donor heart acceptance have declined following recent changes in the donor heart allocation system.

• Providers appear to be a bit more selective about which hearts are accepted for a patient.

• We need better understanding of these changes in order to make sure patients are not waiting longer than necessary to receive a heart transplant.

In 2018, new rules were put in place to govern the order in which patients receive a heart transplant based on how sick they are. This has led to many positive changes, but it also may have resulted in a decline in the percentage of potential donor hearts that are used. We need to better understand the reasons for this to ensure patients have access to a heart transplant as quickly as possible.