1 INTRODUCTION

The six-tiered prioritization scheme implemented with the 2018 United Network for Organ Sharing (UNOS) policy change was designed to prioritize the sickest patients awaiting heart transplantation with the goal of decreasing waitlist mortality and improving donor organ availability. Several analyses of the 2018 policy change have examined the effect of the policy change on either waitlist outcomes and/or short-term post-transplant survival. The main findings of these analyses have demonstrated a decrease in waitlist death/deterioration, an increase in rates of heart transplantation, and either worse or unchanged short-term post-transplant survival.^1-8 Some analyses were limited in interpretation due to informative censoring and repeated analyses have begun to address this issue.^{2, 8-10}

An analysis of Status 1 patients bridged to heart transplantation with veno-arterial extracorporeal membrane oxygenation (VA-ECMO) identified reduced waitlist mortality, with shorter waitlist times, and improved post-transplant survival.³ Prior analyses demonstrated shorter waitlist times, reduced waitlist mortality and equivocal post-transplant survival in patients bridged to heart transplant with an intra-aortic balloon pump (IABP).^{5, 11, 12} One current limitation is that most analyses have examined outcomes for either all statuses or have focused on high-priority (Status 1 and Status 2) patients alone. One recent study by Fuery et al.¹³ provided an analysis of “low-priority” patients (Old 2, New 6) and contrasted these findings with “high-priority” patients (Old 1A, New 1 and 2). The authors found that low-priority patients had similar post-transplant survival before and after the UNOS policy change and longer waitlist times. Both low- and high-priority patients had shorter waitlist times after the policy change, however, lower priority patients received hearts from donors with greater comorbidities. While informative, this study did not assess for a potential interaction between the UNOS policy change and priority status and did not perform a competing risks analysis on waitlisted patients. As a result, it is currently unknown whether the observed benefits in waitlist and post-transplant outcomes in high-priority status patients negatively affect low-priority patients.

Thus, the purpose of this study was to evaluate the combined effect of priority status and policy change on waitlist and post-transplant outcomes for first-time adult heart transplant candidates and recipients and to determine if the gains seen in outcomes among high-priority patients came at a cost to low-priority patients.

2 MATERIALS AND METHODS

3 RESULTS

3.4 Waitlist outcomes

3.4.1 Death/deterioration

The cumulative incidence plots for the waitlist outcome of death/deterioration stratified by policy era and initial listing priority status are shown in Figure 1. The cumulative incidence of waitlist death/deterioration was lower in the POST era when compared to the PRE era for both low- and high-priority patients (p = .034 and p < .001, respectively). However, this difference was noticeably larger in the high-priority patients (Figure 1—right panel) than in the low-priority patients (Figure 1—left panel). The results of the Fine-Gray model, adjusted for additional confounders, are shown in Table 4. There was a non-significant trend toward a beneficial effect of PC on the cumulative incidence of waitlist death/deterioration for patients initially listed as low priority (adjusted sdHR: 0.86 [0.73–1.01]; p = .07 - see Table 4). Patients who were initially listed at high priority status have a significantly lower cumulative incidence of waitlist death/deterioration in the POST era than those who are in the PRE era (adjusted sdHR: 0.51 [0.41–0.63]; p-value < .001 - see Table 4). When looking at the interaction term (Status × Policy), we found a significant dependence between policy era and initial priority status on the cumulative incidence of waitlist death/deterioration (adjusted sdHR: 0.59 [0.45–0.78], p-value < 0.001 - see Table 4). In contrast to the subdistribution hazards results, the cause-specific hazard ratios (see Table 5) examining the effect of the policy era on death/deterioration were not significant for initial high-priority status patients (adjusted cause-specific hazard ratio [csHR]: 1.07 [0.87–1.31], p = .55). On the other hand, there was a trend toward a protective effect of policy era on initial low-priority status patients (adjusted csHR: 0.86 [0.73–1.01], p = .07), but did not meet significance. We did not observe a dependence (or interaction) between policy era and initial priority status (Status × Policy) on the cause-specific hazard (adjusted csHR: 1.23 [0.95–1.61], p = .12). Overall, there was a discrepancy in the sdHR and csHR for the waitlist outcome of death/deterioration.

TABLE 4. Multivariable competing risks analysis looking at waitlist outcomes of death/deterioration or heart transplantation using a Fine-Gray proportional sub-distribution hazards model

	Waitlist death/deteriorationa		Heart Transplantationb
	sdHR (95% CI)	p-value	sdHR (95% CI)	p-value
Post vs. Pre (initial Low-priority status)	0.86 (0.73, 1.01)	.07	1.08 (1.02, 1.14)	.007
Post vs. Pre (initial High-priority status)	0.51 (0.41, 0.63)	<.001	2.16 (1.96, 2.38)	<.001
Interaction Term (status × policy)	0.59 (0.45, 0.78)	<.001	2.00 (1.79, 2.23)	<.001

• Notes: Each main effect is modeled as a binary covariate: Policy = PRE/POST, Priority Status = LOW/HIGH as defined in the main text. While the interaction term, by itself, is difficult to interpret, the p-value assess whether or not the effect policy (PRE/POST) has on the outcome of interest is dependent on/modified by priority status (Low/High) and vice versa. Abbreviation: sdHR, sub-distribution hazards ratio.
• ^a Adjusted for age, gender, race, blood type, BMI, diabetes mellitus, mechanical ventilation at registration, and prior cardiac surgery after listing.
• ^b Adjusted for age, gender, race, blood type, BMI, diabetes mellitus, mechanical ventilation at registration, prior cardiac surgery after listing, ischemic cardiomyopathy, and dialysis after listing (12 observations were omitted due to missing values). The sdHR in the first two rows restricts calculation to either only initial low-priority status at listing patients or initial high-priority status at listing patients, respectively.

TABLE 5. Multivariable competing risks analysis looking at waitlist outcomes of death/deterioration or heart transplantation using a cause-specific hazards model

	Waitlist death/deteriorationa		Heart transplantationb
	csHR (95% CI)	p-value	csHR (95% CI)	p-value
Post vs. Pre (initial Low-priority status at listing)	0.86 (0.73, 1.02)	.09	1.07 (1.00, 1.13)	.046
Post vs. Pre (initial High-priority status at listing)	1.07 (0.87, 1.31)	.55	2.20 (2.04, 2.39)	<.001
Interaction Term (status × policy)	1.23 (0.95, 1.61)	.12	2.07 (1.87, 2.29)	<.001

• Notes: Each main effect is modeled as a binary covariate: Policy = PRE/POST, Priority Status = LOW/HIGH as defined in the main text. While the interaction term, by itself, is difficult to interpret, the p-value assess whether or not the effect policy (PRE/POST) has on the outcome of interest is dependent on/modified by priority status (low/high) and vice versa. Abbreviation: csHR, cause-specific hazards ratio.
• ^a Adjusted for age, gender, race, blood type, BMI, diabetes mellitus, mechanical ventilation at registration, and prior cardiac surgery after listing.
• ^b Adjusted for age, gender, race, blood type, BMI, diabetes mellitus, mechanical ventilation at registration, prior cardiac surgery after listing, ischemic cardiomyopathy, and dialysis after listing (12 observations were omitted due to missing values). The csHR in the first two rows restricts calculation to either only initial low-priority status at listing patients or initial high-priority status at listing patients, respectively.

3.4.2 Heart transplantation

The cumulative incidence of heart transplantation (Figure 1) was significantly higher in the POST era than in the PRE era for those listed as either high (adjusted sdHR: 2.16 [1.96–2.38], p < .001) or low (adjusted sdHR: 1.08 [1.02–1.14], p = .007) priority. The interaction term (Status × Policy) demonstrated a significant dependence of policy era and initial priority status (adjusted sdHR: 2.00 [1.79–2.23], p < .001) on heart transplantation (Table 4). The cause-specific hazard ratios (csHR) for the waitlist outcome of heart transplantation were nearly identical to the corresponding sdHRs (see Table 4 and Table 5).

3.5 One-year post-transplant survival

We also examined the impact of policy era and final priority status at transplant on 1-year post-transplant survival. Kaplan–Meier curves of the four subgroups, stratified by policy status and priority status, are shown in Figure 2. We observed that 1-year post-transplant survival did not differ when stratified by policy change and final priority status (log-rank p-value = 0.22). The multivariable Cox model (Table 6) suggests there was no impact of policy change, irrespective of listing status, on 1-year post-transplant survival (HR: 1.03 [0.70–1.51], p-value = .90); see Table 6. Furthermore, there was no significant interaction between policy and listing status on 1-year post-transplant survival (HR: 0.85 [0.55–1.32], p = .5; see Table 6).

TABLE 6. Multivariable Cox's regression (39 observations removed due to missing values)

Variable	Hazard ratio (95% CI)	p-value
Policy change	1.03 (0.70, 1.51)	.90
Listing status at transplant	1.26 (0.93, 1.70)	.14
Interaction (Policy × Listing status)	0.85 (0.55, 1.32)	0.50

• Notes: Analysis was adjusted for age, gender, race/ethnicity, BMI at transplant, blood type, ischemic cardiomyopathy, prior cardiac surgery, ventilator, dialysis after listing, diabetes mellitus, total bilirubin, creatinine at transplant, and intravenous antibiotics 14 days prior to a heart transplant, ischemic time.

4 DISCUSSION

In this analysis of UNOS data pre- and post-policy change, we observed the following: (1) POST era patients had a significant reduction in the cumulative incidence of waitlist death/deterioration and increase in heart transplantation, (2) waitlist outcomes were either the same or improved for low-priority patients, and (3) there was no significant difference in 1-year post-transplant survival across all subgroups. Taken together, adult, first-time, high-priority, POST era patients received a targeted reduction in waitlist death/deterioration, and an increase in access to heart transplantation, without worsening waitlist or survival outcomes in low-priority patients. As such, the overall effects of the UNOS policy change appear to be meeting the intended goal of improving waitlist survival via improved access to donor hearts for the highest urgency patients and does so without harming low-priority patients.

One noteworthy result to focus on is the discrepancy between the subdistribution and cause-specific hazard ratios for the waitlist outcome of death/deterioration. Specifically, there was no decrease in the cause-specific HR for (1) the interaction (Status × Policy) between the policy era and initial priority status or (2) the effect of the policy era on only the initial high-priority status patients. The sdHR on the other hand was reduced (protective) for these two categories (interaction and initial high-priority status patients). We interpret this discrepancy to mean that the decreased cumulative incidence of waitlist death/deterioration is driven primarily by the rapid rate of heart transplantation rather than by the policy change per se. In other words, the primary byproduct of the policy change appears to be more rapid heart transplantation in higher acuity patients, and this secondarily leads to a decreased cumulative incidence of death or deterioration. We base this interpretation^{18, 19} on the understanding that a cause-specific hazards model provides a causal (or etiologic) interpretation of the effect of policy change on outcomes, whereas the subdistribution hazards model is limited to predicting the outcome of the patient at some time after being listed for a heart transplant. This is consistent with the lack of an effect of policy change on waitlist death/deterioration in our cause-specific hazards analysis. Regardless of the interpretation, the policy change appears to be meeting its overall goal of rapidly providing donor hearts to the highest acuity patients on the waitlist. Moreover, there is no signal for harm to any patient subgroup in our analysis.

Our study adds to the literature by providing an analysis of the interaction, or dependence between, the UNOS policy change of 2018 and patient priority status. Our analysis sought to answer whether the gains seen in hospitalized and mechanically supported patients in the new allocation system came at the expense of worse outcomes in lower priority, ambulatory patients. Based on this analysis, lower priority patients had no observable adverse effects. As alluded to above, one potential mechanism for the observed targeted benefits to high priority, POST era patients is that status upgrades provide a means to “rescue” waitlisted ambulatory patients that suffer from an acute decompensation; we did observe an overall increase in heart transplantation in low-priority patients in the POST era. Similarly, patients with no prior history of cardiomyopathy who present in acute cardiogenic shock also benefit from the “rescue” effect of the policy change. A recent analysis highlighted that one drawback of the rapid time to heart transplantation under the new policy has been an association with a small, but significant reduction in observed myocardial recovery.²⁰ Additional tools to identify patients that have an increased likelihood of recovery are therefore needed.

Another important observation is that there was a similar proportion of patients on VA-ECMO at the time of listing for both the PRE and POST eras, however, there were approximately 4.9 times more patients transplanted from VA-ECMO in the POST era compared to the PRE era. This further highlights the importance of an allocation policy that prioritizes patients based on clinical acuity. As shown by Gonzalez et al,³ there was both a substantial waitlist and post-transplant survival benefit in VA-ECMO bridged patients during the POST era. Similarly, we previously observed¹² significant waitlist benefits for POST era IABP-supported (status 2) patients bridged to heart transplantation without any difference in post-transplant survival. Additionally, our analysis identified a nearly 3-fold increase in the number of patients undergoing heart transplantation from mechanical ventilation. This appears to support the idea that in the POST era transplant centers are more inclined to take risks on sicker patients than before; however, an alternative interpretation is that centers are instead prioritizing patients with “sicker therapies.” Whether this represents patient advocacy, “gaming” or simply treating patients according to agreed-upon consensus cardiogenic shock criteria is debatable. Nonetheless, future iterations of the organ allocation policy should consider instances where the cardiogenic shock criteria fall short in risk stratification and whether additional markers of end-organ malperfusion should be incorporated.²¹

We deliberately selected our study inclusion time periods to minimize the risk of informative censoring to our post-transplant survival analysis. With this in mind, repeated analysis with more complete follow-up will be important to ensure that reported findings to date regarding the effects of the allocation policy remain accurate.^{9, 10}

Another notable finding in our analysis is that fewer women donors appear to be selected in high-priority patients. In concert with this, there has been less utilization of female donors for male recipients, especially in high-priority patients. While this change was not found to be a predictor of post-transplant survival, higher status patients are likely seeing more organ offers and thus some programs may have less incentive to accept gender mismatched organs. This finding warrants a dedicated evaluation.

5 CONCLUSIONS

We observed that in the post-policy era that high-priority transplant candidates have a large and targeted reduction in waitlist death/deterioration, while waitlist outcomes for low-priority patients were either improved or no different. One-year post-transplant survival was not affected by policy change or priority status. Taken together, these findings support the conclusion that high-priority patients have benefited from the policy change and low-priority patients have not been harmed. Additional analyses of other patient subgroups are needed to evaluate for potential inequities that may remain despite the recent UNOS policy change.

AUTHOR CONTRIBUTIONS

Study design: AMW, ECD, ESK, ASV. Data analysis: AMW, ECD, ESK, ASV. Writing of the manuscript: AMW, ECD, ESK, SST, MWF, LZ, KP, ASV.

FUNDING INFORMATION

ESK was supported by the National Institute of Environmental Health Sciences (NIEHS) grant # T32ES013678.

DISCLOSURE

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.