Wide variation in the percentage of donation after circulatory death donors across donor service areas – a potential target for improvement

Abstract

Background:

Substantial differences exist in the clinical characteristics of donors across the 58 donor services areas (DSAs). Organ procurement organization (OPO) performance metrics incorporate organs donated after circulatory determination of death (DCDD) donors, but do not measure potential DCDD donors.

Methods:

Using 2011–2016 UNOS data, we examined the variability in DCDD donors/all deceased donors (%DCDD) across DSAs. We supplemented UNOS data with CDC death records and OPO statistics to characterize underlying process and system factors that may correlate with donors and utilization.

Results:

Among 52,184 deceased donors, the %DCDD varied widely across DSAs, with a median of 15.1% (IQR [9.3%, 20.9%]; range 0.0–32.0%). The %DCDD had a modest positive correlation with 4 DSA factors: median match MELD, proportion of white deaths out of total deaths, kidney center competition, and %DCDD livers by a local transplant center (all Spearman coefficients 0.289–0.464), and negative correlation with 1 factor: mean kidney waiting time (Spearman coefficient −0.388). Adjusting for correlated variables in linear regression explained 46.3% of the variability in %DCDD.

Conclusions:

Donor pool demographics, waitlist metrics, center competition and DCDD utilization explain only a portion of the variability of DCDD donors. This requires further studies and policy changes to encourage consideration of all possible organ donors.

Introduction

The ongoing organ shortage necessitates efforts to expand the donor pool. One option to increase transplantable organs is to use both donation after neurologic determination of death (DNDD) and donation after circulatory determination of death (DCDD) donors. Organ procurement organizations (OPOs) play a vital role facilitating organ recovery and transplant within a given geographic boundary, or donor service area (DSA). Yet there are recognized differences in the number and characteristics of donors across DSAs. For example, noneligible liver grafts (defined as either older DNDD donors or DCDD donors) vary from 0% to nearly 20% of livers transplanted in a DSA.¹ Additionally, about a quarter of DCDD kidney grafts come from just 4 DSAs.² Several authors have suggested that elevating the performance of the lowest performing OPOs could substantially increase the number of organs available for transplant.^1,3,4

OPO performance is evaluated by Centers for Medicare and Medicaid Services (CMS) using two metrics, one that aims to measure the donation rate and another the organ yield. The donation rate metric has been criticized for not adequately capturing potential DCDD donors.^3–6 The donation metric, also known as the donor conversion rate, is calculated as actual organ donors/ “eligible deaths.” An actual donor is an individual whose organs are recovered with the intent to transplant. An “eligible death” is a hospitalized, brain-dead individual ≤75 years of age without contraindications to donation.⁷ A subsequent provision allowed an OPO to add a successful older or DCDD donor to both the numerator and denominator of the donor conversion rate, but potential DCDD donors are otherwise not tracked.⁷

Since the Institute of Medicine identified DCDD donors as one of the greatest potential avenues to augment the donor pool over a decade ago, utilization of DCDD donors has grown and clinical outcomes of DCDD organs have improved.^8,9 DCDD donors comprised 6.7% and 15% of deceased donors in 2006 and 2016, respectfully.¹⁰ Yet, reports continue to identify DCDD donors as one of the largest opportunities to expand the donor pool, with one study estimating 3,825 additional transplantable organs every year if all potential DCDD donors were referred.^5,11

The pathway to donation is complex and involves the alignment and cooperation of donor hospitals, donor characteristics, donor families, OPOs, and transplant centers (Figure S1). This is particularly true for DCDD donors, where the protocols vary by DSA and donor hospital.¹² Our study aimed to investigate the variability in the percent DCDD donors (%DCDD) across DSAs and whether the %DCDD is correlated to underlying process and system factors, such as potential donor referral base, donor characteristics, OPO characteristics, transplant center characteristics, waitlist metrics and procurement characteristics.

Methods and Materials

Data Sources

We analyzed national registry data from the Organ Procurement and Transplantation Network (OPTN) between 2011–2016. The Health Resources and Services Administration provides oversight to the activities of the OPTN contractor. The study was reviewed and exempted by the University of Pennsylvania Institutional Review Board (protocol #828646). We obtained supplemental data on US mortality from the CDC WONDER, an online database developed by the Centers for Disease Control and Prevention, and OPO performance metrics from the Scientific Registry of Transplant Recipients.¹³

Defining “Donor” and %DCDD

Since not all potential donors ultimately have their organs transplanted, precise definitions of donors along the pathway to organ donation have been described.¹⁴ An actual donor is an individual from whom ≥1 organ was recovered for the purpose of transplantation, whereas a utilized donor is an individual from whom ≥1 organ was transplanted. As a primary approach, we calculated the %DCDD in a DSA using actual donors (total actual DCDD donors/total actual deceased donors). As secondary analysis, the %DCDD was calculated using utilized donors. We examined the correlation between the approaches using Spearman’s rank correlation coefficient (r_s) to verify that %DCDD in as DSA was consistent with different definitions of donor (Table S1). Given that percentages may fluctuate rapidly when the denominator is small, a sensitivity analysis was performed using the DSAs in the top tercile of donor volume. Additionally, we examined the %DCDD donors by organ (kidney, liver, lung) in each DSA as well as the %DCDD donors that were kidney-only donors (Figure S2). Lastly, we examined the number of OPOs that met the Donation and Transplantation Community Practice goal of 10% DCDD donors by their goal-year (2013) through 2018.¹⁵

Donor Characteristics

DCDD donor characteristics were compared between the 6 DSAs in the bottom 10^th percentile of %DCDD (“Low %DCDD”) and the 6 DSAs from the top 90^th percentile of %DCDD (“High %DCDD”), similar to prior methodologies.² Donor factors included: age, sex, ethnicity, body mass index, kidney donor profile index, cause of death, hypertension, diabetes, hepatitis C (HCV), terminal creatinine, public health service (PHS) high-risk donor, cold ischemia time and donor warm ischemia time.

DSA factors and %DCDD

The %DCDD was compared to several DSA-specific variables that may be an indicator of potential donor referral base, donor characteristics, OPO characteristics, transplant center characteristics, waitlist metrics and procurement characteristics using a Spearman’s rank correlation coefficient. Variables related to the donor referral base included: the DSA population, the proportion of white deaths out of total deaths and donors-per-1000-deaths. Donor factors included: the mean organs used from a DCDD donor. OPO factors included: the mean donor conversion rate from 2011–2016, OPO full-time employees-per-10,000-individuals-in-a-DSA, DCDD kidney import ratio (total DCDD kidneys imported/total DCDD kidneys recovered), and DCDD kidney export ratio (total DCDD kidneys exported/total DCDD kidneys recovered). Transplant center competition was calculated using the inverted Herfindahl-Hirschman Index (HHI). An inverted HHI of 0 indicates no competition and a value closer to 1 indicates more competition.¹⁶ Theorizing that existence of a transplant center that performs more DCDD liver transplants might signal to the OPO a willingness to use DCDD organs, we examined the correlation between %DCDD and the highest %DCDD liver grafts used by a center within the DSA. Waitlist metrics included values used in organ-specific allocation (mean waitlist time for kidney recipients and median match model for end-stage liver disease [MELD] for liver recipients), time-to-kidney-transplant for 25th percentile of waiting list candidates and waitlist removal rates. Procurement factors included: mean donor warm ischemia time and mean distance from donor hospital to transplant center. See Appendix 1 for details on data sources and definitions of variables.

Multivariable Model

To determine amount of variability in the %DCDD explained by underlying process and system factors, all variables with at least weak univariate correlation (r_s>0.25 or r_s<−0.25) were included in a linear regression model. The normality of the residuals was examined using the Shapiro-Wilk test.

Estimating DCDD Donor Potential and Recent Trends

To simulate the potential increase in the number of organ donors, we calculated total number of DCDD donors if all lower-performing DSAs performed at the level of the median and 75^th percentile DSA. OPTN data from 2017–2018 was used to describe the more recent trends in %DCDD.

Statistical Analysis

All analyses were performed using Stata 15.0 (Statacorp LP, College Station, TX) using 2-sided hypothesis testing and P value of <0.05 as the criterion for statistical significance. Nonparametric continuous variables were compared between groups using Wilcoxon rank-sum tests. Categorical and binary variables were compared using the chi-square tests.

Results

There were 52,184 deceased donors: 44,343 DNDD donors and 7,841 DCDD donors. DCDD donors yielded fewer organs per donor than DNDD donors (mean 1.9 organs vs 3.3 organs), were more often kidney donors (84.3% donated ≥ 1 kidney vs 76.4% in DNDD) and more frequently had no organs transplanted (13.0% of donors vs 4.6% of donors) (all p<0.001, Table S2).

The %DCDD donors by DSA ranged from 0.0% to 32.0% (Figures 1A & 1B, Table S3). DSAs were ranked by %DCDDs: the 90^th percentile had DSAs > 25.6% DCDD donors, the 75^th percentile had DSAs with 20.9–25.5% DCDD donors, the 50^th percentile had DSAs with 15.1–20.8% DCDD donors, the 25^th percentile had DSAs with 9.3–15.0% DCDD donors, and the <25^th percentile had DSAs < 9.3% DCDD donors (Figure 1C). In 2013, 18 OPOs failed to meet the Donation and Transplantation Community Practice goal of 10% DCDD donors; by 2018, 5 OPOs did not meet the goal (Figure S3).

Figure 1.

A. Total DCDD and DNDD donors for each DSA with DSA ranked from lowest percent DCDD donors to highest percent DCDD donors. B. Percent DCDD donors out of total deceased donors with DSA ranked from lowest percent DCDD donors to highest percent DCDD donors. C. Heat map showing the percentage of DCDD donors in each DSA by percentile.

Comparing the high %DCDD and low %DCDD DSAs, we found that DCDD donors from high %DCDD DSAs were older (43 vs 36, p<0.001) and more often white (89.5% vs 74.8%, p<0.001) (Table 1). Comorbidities such as hypertension, diabetes, HCV or PHS high-risk did not vary among the high and low DSAs.

Table 1.

Comparison of Donor Characteristics Between Low and High %DCDD DSA. Low and High % DCDD groups include the bottom and top 6 DSAs when ranked by %DCDD

	DCDD Donors				DNDD Donors
	Low %DCDD (6 OPOS, donors=163)	High %DCDD (6 OPOS, donors=1253)	p-value		Low %DCDD (6 OPOS, donors=4827)	High %DCDD (6 OPOs, donors=3323)	p-value
Age in Years, median (IQR)	36 (23, 48)	43 (27, 53)	<0.001		41 (25, 53)	40 (25, 54)	0.97
Male, n (%)	99 (60.7%)	826 (65.9%)	0.19		2782 (57.6%)	1980 (59.6%)	0.079
Ethnicity, n (%)
White	122 (74.8%)	1122 (89.5%)	<0.001		2594 (53.7%)	2744 (82.6%)	<0.001
Black	19 (11.7%)	31 (2.5%)			1125 (23.3%)	194 (5.8%)
Hispanic	20 (12.3%)	62 (4.9%)			628 (13.0%)	243 (7.3%)
Asian	2 (1.2%)	20 (1.6%)			44 (0.9%)	67 (2.0%)
Other	0 (0.0%)	18 (1.4%)			436 (9.0%)	75 (2.3%)
BMI(kg/m2), median (IQR)	26.7 (22.6, 32.4)	26.8 (23.1, 31.7)			26.4 (22.7, 30.9)	26.5 (22.8, 31.1)	0.30
KDPI, median (IQR)	50 (28, 66)	54 (33, 75)	0.002		55 (27, 81)	47 (22, 72)	<0.001
Cause of Death, n(%)
Anoxia	71 (43.6%)	629 (50.2%)	0.007		1127 (23.3%)	1262 (38.0%)	<0.001
CVA	24 (14.7%)	219 (17.5%)			1879 (38.9%)	954 (28.7%)
Trauma	59 (36.2%)	337 (26.9%)			1713 (35.5%)	1021 (30.7%)
CNS Tumor	1 (0.6%)	0 (0.0%)			27 (0.6%)	6 (0.2%)
Other	8 (4.9%)	68 (5.4%)			81 (1.7%)	80 (2.4%)
Hypertension, n(%)	40 (24.7%)	370 (29.6%)	0.19		1867 (38.8%)	958 (29.0%)	<0.001
Diabetes Mellitus, n(%)	9 (5.5%)	88 (7.0%)	0.47		622 (12.9%)	306 (9.3%)	<0.001
HCV, n(%)	2 (1.2%)	29 (2.3%)	0.37		203 (4.2%)	226 (6.8%)	<0.001
Terminal Creatinine (mg/dL), median (IQR)	0.8 (0.5, 1.0)	0.8 (0.6, 1.1)	0.044		1.1 (0.8, 1.6)	0.9 (0.7, 1.3)	<0.001
PHS High-Risk Donor, n(%)	21 (12.9%)	211 (16.9%)	0.20		720 (14.9%)	733 (22.1%)	<0.001
CIT for liver grafts, in hours, median (IQR)	4.9 (4.0, 5.6)	6.0 (5.1, 7.3)	<0.001		5.8 (4.5, 7.3)	6.5 (5.1, 8.3)	<0.001
CIT for kidney grafts in hours, median (IQR)	20.6 (15.5, 25)	16.8 (11.9, 21.5)	<0.001		16.3 (11.0, 22.5)	14.0 (9.8, 19.6)	<0.001
Donor WIT in minutes*, median (IQR)	21 (17, 28)	13 (10, 20)	<0.001

^*For donors prior to 3/31/2015, after which the variable was no long collected.

Abbreviations: BMI, Body Mass Index; CIT, Cold Ischemia Time; CNS, Central Nervous System; CVA, Cerebrovascular Accident; DCDD-donation after circulatory determination of death; DNDD-donation after circulatory determination of death; HCV, Hepatitis C Virus; IQR, interquartile range; KDPI, Kidney Donor Profile Index; PHS, Public Health Service; WIT, Warm Ischemia Time

On univariable analysis, the %DCDD was modestly correlated with five of the DSA-specific underlying process and system variables (Table 2, Figure S4); the %DCDD was positively correlated to the proportion of white deaths out of total deaths (r_s=0.410), the median MELD for liver recipients (r_s=0.464), kidney transplant center competition (r_s=0.289), and highest %DCDD livers by a local transplant center (r_s=0.333). The %DCDD has a modest negative correlation with the post kidney allocation system (KAS) mean waitlist time for kidney recipients (r_s = −0.388). The five variables explained 46.3% of the variability in the %DCDD (Table S4).

Table 2.

Univariate Correlation of % DCDD with DSA Variable

	Spearman coefficient, rs	p-value
Donor Referral Base
2015 DSA population	−0.024	0.857
Proportion of white deaths out of total deaths in DSA*	0.410	0.002
Donors-per-1000-Deaths in DSA*	0.206	0.124
Donor Characteristics
Mean DCDD Organs Used per DCDD donor in DSA**	−0.241	0.072
OPO Characteristics
Mean standardized donor conversion rate (2011–2016)	−0.124	0.355
OPO FTEs per 10,000 population served⁜	−.051	0.741
DCDD kidney export ratio **    (total DCDD kidneys exported/total DCDD kidneys recovered)	−0.226	0.091
DCDD kidney import ratio**    (total DCDD kidneys imported/total DCDD kidneys recovered)	−0.237	0.076
Transplant Center Characteristics
Kidney Transplant Center Competition (Inverted HHI, where a larger number indicates more competition)	0.289	0.028
Liver Transplant Center Competition (Inverted HHI, where a larger number indicates more competition)†	0.112	0.428
Highest %DCDD liver transplants of any center within the DSA	0.333	0.015
Waitlist Metrics
Pre-KAS mean waitlist time for kidney recipients in DSA	0.012	0.931
Post-KAS mean waitlist time for kidney recipients in DSA	−0.388	0.003
Time-to-kidney-transplant for 25th percentile of waitlist candidates	−0.096	0.472
Kidney waitlist recipient removal rate in DSA	−0.087	0.514
Median match MELD for liver recipients in DSA†	0.464	0.001
Liver waitlist recipient removal rate in DSA†	0.202	0.152
Procurement Characteristics
Mean Donor Warm Ischemia Time**	0.019	0.889
Mean distance from donor hospital to kidney recipient transplant center**	0.056	0.678
Mean distance from donor hospital to liver recipient transplant center††	0.204	0.164

^*Includes 57 DSAs, excludes Puerto Rico (PRLL) for which information is not available from the CDC

^**Includes 57 DSAs, excludes DSA with 0 DCDD donors

^⁜Includes 45 DSAs, excludes 3 OPOs which function as additional services (bloodbank, bioengineering) and 10 OPOs with missing information

^†Includes 52 DSAs, excludes DSA without a liver transplant program

^††Includes 48 DSAs, excludes DSA for which no DCDD liver was used locally

Abbreviations: DCDD, donation after circulatory determination of death; DSA, donor service area; FTE, full-time employee; HHI, the Herfindahl–Hirschman Index; KAS, Kidney Allocation System; MELD, Model in End-Stage Live Disease; OPO, organ procurement organization

Increasing the %DCDD donor to the median and 75^th percentile could yield about 302 and 735 new donors per year, respectively, which would result in an estimated 605 and 1430 new transplants (Table S5). The past two years (2017–2018) has continued to demonstrate growth and variation in the %DCDD by DSA, ranging from 0.0% to 38.7% (Table S1).

Discussion

The %DCDD varied substantially by DSA, from 0% to 32.0% of all deceased donors, and remained consistent when a donor was defined an actual or utilized donor. While our first aim was to describe the variation in %DCDD donors across DSA, we sought to explore the underlying process and system factors that may correlate to DCDD donor use (Figure S1). We found five modest correlations, but also found that over half of the variation remained unexplained by our variables. This unexplained variation leads toward future, prospective studies focusing on elements not captured in our data – such as: OPO organizational priorities, donor referrals, how donation is discussed with families, and center behavior.

DSAs with more kidney transplant center competition or centers that use a greater proportion of DCDD livers were associated with a greater %DCDD donors. Our finding is congruent with prior research that more competitive kidney transplant markets are associated with an increased use of higher kidney donor risk index grafts.¹⁶ Although conjecture, OPOs may prioritize pursuing DCDD donors, if they have reason to believe that a center(s) will accept the organs. A study from a single OPO found that DCDD donors were an estimated 1.6 times costlier to the OPO than DNDD donors because some donors do not progress to donation and fewer organs are transplanted per donor.¹⁷ To what extent the financial implications of evaluating a DCDD donor, which inherently has a lower potential for organ utilization, may bear are unclear. Remuneration or other incentives may be necessary when assessing donors with lower organ yields.

Some waitlist metrics were associated with %DCDD but not in a consistent way. Higher MELDs were associated with higher %DCDD. Prior research has suggested that higher median MELDs may influence programs to pursue greater risk liver grafts,¹⁸ while a more recent study found no correlation between MELD and acceptance of older or DCDD donors.¹ Post-KAS mean waitlist time was negatively associated with %DCDD, meaning areas with a greater percentage of DCDD donors had shorter waiting time for their kidney recipients. Yet, time-to-kidney-transplant for the 25^th percentile of waitlist candidates had no association with %DCDD, which indicates that local organ scarcity may not directly influence the availability of DCDD organs.

Examining donor factors, we found that DCDD donors from high %DCDD DSAs were significantly older than those from low %DCDD DSAs, yet the age of the DNDD donors was similar. This observation may imply that high %DCDD DSAs have more liberal age criteria for DCDD donors. Areas where white individuals comprised a larger proportion of all deaths also correlated with a higher %DCDD. It is unknown whether this association is reflective of differences in referral to the OPO, OPO activity, family/patient wishes or other factors. Racial/ethnic minorities may be less likely than whites to consent to organ donation.¹⁹

We acknowledge that our covariates are not comprehensive of all variables affecting local transplant markets, such as recipient preferences, surgeon/center behavior, or day of the week and time of the organ offer.²⁰ Additionally, there may be interplay between factors, as acceptance practices of transplant centers may influence the OPO’s assessment of future donors. Yet the magnitude of variation highlights the need to address additional factors that may influence differences among-OPOs, which may include how donation is discussed with families, establishing organizational best-practices, and the professional relationships between OPOs, donor hospitals, and transplant centers.^21–23 Collaborative reviews between transplant centers and OPOs regarding donor utilization as well as stewardship within a center regarding outcomes of turned-down offers may be insightful towards changing practices.

There are multiple reports of how a change in OPO leadership or targeted interventions have increased DCDD donation. For example, DCDD donors nearly doubled the year after the University of Wisconsin Hospital and Clinics OPO held focus groups to identify barriers to DCDD donation and used the information to design materials for hospital staff.²⁴ Other OPOs report substantial increases in DCDD donors after defining policies for donor referrals, establishing organizational clarity and/or setting operational policies.^25,26

A policy implication of this variation is the current characterization of OPO performance via the donor conversion rate and organ yield metric. Members of our group have previously proposed changes to the metrics to encourage the use of all donors.³ These new metrics are: 1) donation percentage (which uses a broader definition of a potential donor, including DCDD donors) and 2) organs transplanted per possible donor. Another policy implication might be the support of progressive recovery practices, such as OPO-designated procurement surgeons, having local surgeons procure and ship organs, or dedicated donor centers, as this might decrease the resources commitment for an accepting center and increase the likelihood of utilization, especially for kidney-only donors.^27,28 Such policy changes combined with learning from OPOs with experience explicitly targeting improvements in DCDD donation may increase the number of deceased donor organs available.

The aim of the study was to describe the variation in DCDD donors. A limitation of this study is that it reports a variation without providing data on what drives the variation, which is likely multifactorial and interrelated. Analyzing DCDD donors as a percentage of total deceased donors may introduce bias as OPOs with a large number of DNDD donors may be appear worse than those with fewer DNDD donors. We performed a sensitivity analysis using the top-tercile of DSAs by donor volume to mitigate this bias and found similar variation in the %DCDD (Table S1). Additionally, there is the theoretical concern that OPOs may divert some DNDD donors to DCDD donors to facilitate earlier procurement times. Yet, multiple reports have demonstrated that policies aimed at increasing DCDD donors leads to increases in all deceased donors.^29,30

There were several limitations due to the clinical granularity of our data and sample size. It was not possible to report the potential DCDD donors because patient-level clinical information that determine brain-death status and medical contraindications to donation for referrals to OPOs are not nationally tracked. Exploring whether there is variation in number and clinical characteristics of referrals, and reasons why referrals do not become donors requires individualized OPO study. Although, we examined the association of %DCDD to waitlist metrics, we are not able to determine whether all DCDD donors would have been an acceptable organ-specific donor. Lastly, given the small number of DSAs and only modest correlations, any conclusions of associations should be viewed with caution.

Another limitation is our inability to adequately adjust for regional variation. The population of each DSA varies by demographics and comorbidities which may both affect an individual’s potential as a DCDD donor and willingness to donate. There may be other regional trends at play. For example, as the tragedy of overdose deaths have risen with the opioid epidemic, there is an increase in donors with anoxic brain injury.³¹ There are no reports on how frequently opioid overdose leads to DNDD versus DCDD donation, yet opioid deaths may affect each DSA at different rates.³²

In conclusion, this study highlights the vast variation in the percent DCDD out of all deceased donors across the 58 DSAs – with DCDD donors comprising 0.0 to 32.0% of a DSAs total deceased donors. DCDD donors may be an under-utilized source of transplantable organs in certain areas. As a community seeking innovative approaches to attenuate the organ shortage crisis for our patients, we must underscore the need to incentivize and track all possible organ donors. Potential strategies to increase organs include changes to OPO performance metrics and more research into what drives the disparity in %DCDD.

Abbreviations:

CDC

Centers for Disease Control

CMS

Centers for Medicare and Medicaid Services

DCDD

donations after circulatory determination of death

DNDD

donations after neurologic determination of death

DSA

donor service area

HCV

hepatitis C virus

KAS

kidney allocation system

MELD

model for end-stage liver disease

PHS

public health service

OPO

organ procurement organization

OPTN

Organ Procurement and Transplantation Network

%DCDD

percent DCDD donors