Advancing Patient-Centered Metrics for Heart Transplantation: The Role of Days Alive and Outside the Hospital

Abstract

Background:

Heart transplantation (HT) survival and waitlist times are established outcome metrics. Patient-centered HT outcomes are insufficiently characterized. This study evaluates the role of days alive and outside the hospital (DAOH) as a candidate patient-centered HT performance measure.

Methods:

The study cohort included Medicare beneficiaries undergoing HT (July 2008-December 2017). The percent of days outside of hospital (%DOH) six months before (%DOH-BF) and percent of days alive outside of hospital 12 months after HT (%DAOH-AF) were evaluated along with adverse events (AEs, early: ≤3 months; late: 4-12 months). Patients were stratified by patient %DAOH-AF terciles. Risk-adjusted %DAOH was evaluated across hospitals.

Results:

A total of 5,104 beneficiaries underwent HT across 108 hospitals. Median [IQR] age was 62 [53-67] years, 23.9% were female, and 21.4% were African-American. The overall median %DOAH-AF was 92.9% [83.8%, 95.9%], varying by tercile: low: 71.8% [4.9%, 83.6%], intermediate; 92.9% [91%, 94%]; high 96.4% [95.9%, 97.3%]. The lowest (versus highest) tercile %DAOH-AF had a lower median %DOH-BF (88% [73%-97%] versus 92% [81%-98%]) and longer post-HT inpatient stay (54 [36-81] versus 13 [10-15] days). After HT, the lowest versus highest tercile had greater AEs burden in the early [allograft failure (16.1% versus 1.6%), stroke (12.1% versus 2.3%)], and late [stroke (5.1% versus 1.9%), sternal wound infection (5.0% versus 0.8%)] phases post-HT. Mean hospital %DAOH_adj was 80.5% (min:max 57.7%-96.7%).

Conclusions:

Post-HT %DAOH varies across beneficiaries and hospitals and is associated with AEs. Further research is warranted to assess the role and validity of %DAOH as a HT quality metric.

Central Figure:

Advancing Patient-Centered Metrics for Heart Transplantation: The Role of Days Alive and Outside the Hospital.

Introduction

Heart transplantation (HT) is recognized as the gold standard treatment for advanced heart failure refractory to guideline-directed medical therapy and improves patient survival, quality of life, and functional capacity^1,2. A limited set of institutional process measures (e.g., waitlist duration) and outcomes (e.g., mortality during waitlist and post-transplantation) are reported by the United Network for Organ Sharing (UNOS) for HT hospitals. Despite the elevated acuity for patients undergoing HT and the likelihood of post-discharge healthcare utilization, limited patient-centered hospital performance measures are reported by UNOS and the Scientific Registry of Transplant Recipients.

Days alive and outside the hospital (DAOH), an emerging patient-centered hospital performance measure, has been applied within medical and surgical cohorts.^3–7 Hospital performance evaluation is enhanced using DAOH, which accounts for a patient’s survival status and location (e.g., home, hospital, nursing home) before and following discharge.³ Prior work informs the relationship between increasing DAOH and improved quality of life.^8,9 A recent national evaluation of patients undergoing durable left ventricular assist device implantation found: (1) large-scale variability in the percentage of DAOH (%DAOH) over a 12-month horizon and (2) an independent relationship between %DAOH and patient health-related quality of life.¹⁰ Despite extensive public and policy interest focusing on hospital HT performance, DAOH evaluations in this context have been limited.^11,12

This study leverages national Medicare claims to evaluate DAOH as a patient-centered hospital HT performance metric. Specifically, this study: (1) tests the hypothesis that there will be variability between beneficiaries in percent days outside of hospital (%DOH) six months before and %DAOH twelve months following HT; (2) tests the hypothesis that there will be interhospital variability in risk-adjusted %DAOH twelve months following HT; and (3) associates %DOH and %DAOH with adverse events and healthcare utilization.

Methods

Study Population

This study evaluated adult Medicare beneficiaries undergoing HT between July 1, 2008, and December 31, 2017. The cohort was constrained to include beneficiaries with (1) complete 6-month pre- and 12-month post-HT follow-up or until death and (2) continuous Medicare fee-for-service Parts A and B enrollment during the 6-months preoperative through 12-months of follow-up (Figure S1). Patients with concomitant multi-organ (liver, lung, and/or kidney) transplantation were excluded. The American Hospital Association annual survey file was used to identify each hospital’s geographic location, teaching status, and number of intensive care unit beds. Each hospital’s total HT volume among adult patients between 2008 and 2016 was collected using data from the Organ Procurement and Transplantation Network. International Classification of Diseases code, 9^th and 10^th revisions codes were used to identify waitlist status within six months before HT (Table S1). Each beneficiary’s five-digit zip code was linked to a Distressed Community Index (DCI) value ranging from 0-100 provided by the Economic Innovation Group (2015-2019) to reach a relative economic well-being ranking.¹³

The study was limited to evaluating HT among Medicare beneficiaries due to Medicare’s standardized billing and reimbursement practices that facilitate rigorous research activities. Unlike commercial payers, the Centers for Medicare and Medicaid Services longitudinally tracks each beneficiary’s vital survival status and the location of reimbursed services (e.g., inpatient, outpatient, hospice).

Clinical Outcome

The study’s primary outcomes were %DOH and %DAOH, which were calculated by dividing each beneficiary’s number of DOH (numerator) by their pre-HT time exposure (for DOH) and post-HT time exposure (for DAOH).^3,10,14,15 The numerator was calculated as the total number of days spent in a facility subtracted from each patient’s pre- and post-HT follow-up period. For both periods, the number of days admitted within a hospital was computed as the cumulative duration a patient spent in any of the following facilities: (1) index or non-index hospital; (2) skilled nursing facility(s); (3) rehabilitation facility(s) or (4) hospice(s) (Supplemental Methods).

The use of Medicare claims data was approved by Michigan Medicine’s Institutional Review Board (HUM00089307, approved 9/29/2014) and through a Data Use Agreement (2014-27462). Data use agreements restrict the distribution of raw study-related data files, and data suppression rules prevent reporting of values with less than 11 observations. Requests for summary statistics will be shared upon review and approval by the study team. This study was compliant with ISHLT ethics statement.

The denominator was determined as the number of pre- (−6 months) and post-HT (+12) months. As needed, the number of days from a beneficiary’s death to the end of the twelve-month time horizon was assigned as days dead.¹⁰ The pre-HT %DOH (%DOH-BF) was calculated as DOH/180 days, while the post-HT %DAOH-AF was calculated (DAOH/365 days) (Figure S2).

Secondary outcomes included adverse events, readmission, statin utilization post-discharge (at one month), freedom from dialysis, cardiac rehabilitation, and death. Adverse events identified through billing codes (Table S1) included acute organ rejection, cardiac arrest, cytomegalovirus, endocarditis, gastrointestinal infections (GI), pneumonia, sternal wound infection, urinary tract infection, severe allograft failure (defined as the use of temporary mechanical circulatory support), dialysis, stroke, and the presence of a feeding tube. Several post-HT outcomes were assessed across an early (within three months of surgery) and late (four to twelve months) period.

Statistical Analyses

While %DAOH was evaluated as quartiles (Table S2) at the patient level, terciles were preferred given sample size constraints. A Kolmogorov-Smirnov test was performed to test the normality of continuous variables. Continuous variables were presented as mean (CI_95%) and compared using analysis of variance when normally distributed or as medians [IQR] and compared using Kruskal-Wallis tests when non-normally distributed. Categorical variables were presented as frequency and percentage and compared using χ² test or Fisher Exact test if any cell with frequency <5. To assess interhospital variation, first a generalized linear mixed model was used to calculate the hospital expected value of %DAOH adjusting for age, sex race, comorbidities, dual eligibility, hospital region location, hospital HT volume, hospital bed size, cardiac ICU hospital bed, hospital teaching status, and DCI. Next, the risk-adjusted %DAOH was calculated by multiplying the ratio of observed %DAOH and expected %DAOH (O:E ratio) in each hospital by the overall %DAOH mean. The hospital-specific CI_95%of risk-adjusted %DAOH was calculated by bootstrapping with replacement at the hospital level and without replacement at the patient level.¹⁶ The intraclass correlation coefficient (ICC) was estimated to quantify the amount of variation in %DAOH use attributed to the hospital.

The robustness of the primary findings was assessed through pre-specific subgroup analyses. Specifically, %DOH-BF and %DOAH-AF were compared across demographic subgroups: age (<60 years, 60-69 years, ≥70 years), sex, race (African-American, White, or Other), and comorbidities. Sankey diagrams displayed patient-specific trajectories between 6 months before and 12 months following HT.

Statistical tests were 2-sided (α-level of 0.05). Statistical analyses were performed with SAS version 9.4 (SAS Institute), with figures generated using Excel, Prism, and R version 4.3.1 (R Project).

Results

A total of 8,088 Medicare beneficiaries underwent HT (Figure S1). A total of 2,742 beneficiaries were excluded due to a lack of enrollment in Parts A & B during the study time window as well as 242 beneficiaries who underwent concomitant multi-organ transplantation. The study cohort included 5,104 Medicare beneficiaries undergoing HT across 108 hospitals. The overall median [IQR] hospital HT procedural volume for the study period was 38 [27 - 58]. Pre-HT characteristics are presented in aggregate and across terciles of %DAOH-AF (Table 1). The median [IQR] age of the cohort was 62 [53-67] years, 23.9% were female, 69.9% were White, and 21.4% were African-American. The mean risk-adjusted %DAOH was 80.5% [CI_95% 79.3%-81.7%], ranging from 57.7% to 96.7% across hospitals, with an ICC of 2.7% (Figure S3).

Table 1.

Patient Characteristics According to Terciles of Percent Days Alive and Outside the Hospital 12 Months after Heart Transplantation (%DAOH)

		Home time (%DAOH-AF terciles)
Variables	Overall	Low [≤88.4%]	Intermediate [88.7-94.7%]	High [≥95%]	P value
	[n=5,104]	[n=1,697]	[n=1,680]	[n=1,727]
%DOH-BF	90.0% [76.7%, 97.8%]	87.8% [73.3%, 96.7%]	89.4% [75.6%, 97.2%]	92.2% [80.6%, 98.3%]	< .0001
%DAOH-AF	92.9% [83.8%, 95.9%]	71.8% [4.9%, 83.6%]	92.9% [91.0%, 94.0%]	96.4% [95.9%, 97.3%]	< .0001
Patient Demographics
Age, y	62 (53,67)	63 (54, 67)	61 (52,67)	62 (53,67)	0.0044
< 60 years	2096 (41.1%)	667 (39.3%)	732 (43.6%)	697 (40.4%)	0.0511
60-69 years	2469 (48.4%)	847 (49.9%)	791 (47.1%)	831 (48.1%)
≥70 years	539 (10.6%)	183 (10.8%)	157 (9.4%)	199 (11.5%)
Sex
Male, n [%]	3882 (76.1%)	1258 (74.1%)	1274 (75.8%)	1350 (78.2%)	0.0209
Female, n [%]	1222 (23.9%)	439 (25.9%)	406 (24.2%)	377 (21.8%)
Race
White	3569 (69.9%)	1168 (68.8%)	1183 (70.4%)	1218 (70.5%)	0.6179
African-American	1091 (21.4%)	385 (22.7%)	350 (20.8%)	356 (20.6%)
Other	444 (8.7%)	144 (8.5%)	147 (8.8%)	153 (8.9%)
Distressed Community Index, Highest Quintile	823 (16.1%)	299 (17.6%)	262 (15.6%)	262 (15.2%)	0.6103
Distressed Community Index Score	45.4 [20.8, 70.2]	45.2 [20.9, 71.1]	45.1 [20.6, 69.9]	45.8 [20.8, 69.7]	0.7855
Dual Eligible Medicare/Medicaid	1627 (31.9%)	564 (33.2%)	554 (33.0%)	509 (29.5%)	0.0306
Comorbidities
Myocardial Infarction	1464 (28.7%)	465 (27.4%)	492 (29.3%)	507 (29.4%)	0.3597
Peripheral Vascular Disease	1487 (29.1%)	518 (30.5%)	524 (31.2%)	445 (25.8%)	0.0007
Cerebrovascular Disease	456 (8.9%)	210 (12.4%)	136 (8.1%)	110 (6.4%)	< .0001
Ischemic cardiomyopathy	2573 (50.4%)	917 (54.0%)	793 (47.2%)	863 (50.0%)	0.0003
Cardiogenic Shock	1608 (31.5%)	597 (35.2%)	512 (30.5%)	499 (28.9%)	0.0002
Dementia
Chronic Pulmonary Disease	2241 (43.9%)	739 (43.5%)	728 (43.3)	774 (44.8%)	0.6392
Respiratory Failure	1339 (26.2%)	485 (28.6%)	438 (26.1%)	416 (24.1%)	0.0113
Connective Tissue Disease-Rheumatic Disease	90 (1.8%)	37 (2.2%)	29 (1.7%)	24 (1.4%)	0.2113
Peptic Ulcer Disease	142 (2.8%)	64 (3.8%)	49 (2.9%)	29 (1.7%)	0.0009
Mild Liver Disease	543 (10.6%)	253 (14.9%)	147 (8.8%)	143 (8.3%)	< .0001
Moderate or Severe Liver Disease	84 (1.7%)	46 (2.7%)	18 (1.1%)	20 (1.2%)	0.0001
Diabetes without complications	1861 (36.5%)	647 (38.1%)	620 (36.9%)	594 (34.4%)	0.0687
Diabetes with complications	553 (10.8%)	199 (11.7%)	222 (13.2%)	132 (7.6%)	< .0001
Paraplegia and Hemiplegia	107 (2.1%)	62 (3.7%)	29 (1.7%)	16 (0.9%)	< .0001
Renal Disease	2615 (51.2%)	941 (55.5%)	862 (51.3%)	812 (47.0%)	< .0001
Dialysis	104 (2.0%)	51 (3.0%)	43 (2.6%)	10 (0.6%)	< .0001
Kidney transplant status	102 (2.0%)	40 (2.4%)	44 (2.6%)	18 (1.0%)	0.0019
Cancer	93 (1.8%)	32 (1.9%)	28 (1.7%)	33 (1.9%)	0.8431

Note: In compliance with the CMS suppression policy, all data involving less than 11 patients are not shown.

Median [IQR] or n(%) are shown.

The monthly percentage of beneficiaries in our study cohort on the HT waitlist increased from 6.4% at six months to 35.5% at one month before HT (Figure S4). The median %DOH-BF was 90% [76.7%, 97.8%], equating to 162 [138, 176] total days at home prior to HT (Table 2). Overall, the median days before HT spent in hospice and rehabilitation were 28 days [23, 60] and 15 days [8, 29], respectively (Table S3). The median %DOH-BF significantly increased across %DAOH-AF terciles: Low (87.8% [73.3%, 96.7%]) versus High (92.2% [80.6%, 98.3%]), p<0.0001. Overall, 92.9% [83.8%, 95.9%] of the 12 months following HT was spent alive and outside the hospital, equating to 339 [306, 350] total days at home (Table 3). Overall, the median days after HT spent in hospice was 5.5 days [1, 13] while 20 days [13, 31] in rehabilitation. Over one-half of patients spent over 90% of their time at home after HT (Figure 1A, Table S4). In contrast, patients in the highest %DAOH-AF HT tercile spent 90 more days home in the 12 months following HT compared to lowest tercile. By 12 months, 11.6% (590/5104) of the cohort died. Over 90% of patients who died had 0% DAOH-AF, compared to a %DAOH-AF greater than 90% for over 80% of patients alive 12 months after (Figure 1B–C).

Table 2.

Outcomes and Complications According to the Terciles of Percent Days Outside the Hospital 6 months before Heart Transplantation (%DOH-BF)

		Home Time (percentage of DOH-BF terciles)
Variables	Overall	Low ≤81.7%	Intermediate 82.2 to 95.6%	High ≥96.1	P value
Number of patients	(n=5104)	(n=1711)	(n=1767)	(n=1626)
% DOH-BF	90% (76.7%, 97.8%)	69.4% (57.2%, 76.7%)	90.6% (86.7%, 93.3%)	99.4% (97.8%,100%)	< .0001
Total days at home	162 (138,176)	125 (103,138)	163 (156,168)	179 (176,180)	< .0001
Total days in hospital over 6 months before HT	18 (4,41)	54 (41,76)	17 (12,24)	1 (0,4)	< .0001
Total days in hospital leading to HT	1 (0,13)	23 (1,48)	1 (0,7)	0 (0,1)	< .0001
Elective admission for HT	3325 (65.1%)	577 (33.7%)	1191 (67.4%)	1557 (95.8%)	< .0001
Patients in SNF before HT	NA	54 (3.2%)	<11	<11	< .0001
Total days in SNF	0 (0,0)	0 (0,0)	0 (0,0)	0 (0,0)	< .0001
Total days in SNF among patients in SNF	15 (7,25)	17 (9,28)	8 (6,11)	2 (2,2)	0.0186
Patients in Hospice before HT admission	NA	<11	0	0	0.0075
Total days in hospice	0 (0,0)	0 (0,0)	0 (0,0)	0 (0,0)	0.007
Total days in hospice among patients in hospice	28 (23,60)	28 (23,60)	No patients in denominator	No patients in denominator	N/A
Patients in Rehabilitation before HT admission	NA	51 (3.0%)	<11	0	< .0001
Total days in rehabilitation	0 (0,0)	0 (0,0)	0 (0,0)	0 (0,0)	< .0001
Total days in rehabilitation among patients in rehabilitation	15 (8,29)	18 (11,31)	9 (6,13)	No patients in denominator	0.0071

Note: In compliance with the CMS suppression policy, all data involving less than 11 patients are not shown.

Median [IQR] or n(%) are shown.

Abbreviation: Heart Transplantation- HT; %DOH-BF- Percent Days Outside the Hospital before Heart Transplantation; N/A: Not reportable due to Medicare data suppression rules; SNF - Skilled Nursing Facility

Table 3.

Outcomes and Complications According to the Terciles of Percent of Days Alive and Outside the Hospital (%DAOH-AF)

		Home time (%DAOH-AF terciles)
Variables	Overall	Low (≤88.4%)	Intermediate (88.7-94.7%)	High (≥95%)	P value
	(n=5,104)	(n=1,697)	(n=1,680)	(n=1,727)
%DAOH-AF	92.9% [83.8%, 95.9%]	71.8% [4.9%, 83.6%]	92.9% [91.0%, 94.0%]	96.4% [95.9%, 97.3%]	< .0001
Total days at home after HT	339 [306, 350]	262 [18, 305]	339 [332, 343]	352 [350,355]	< .0001
Total days in hospital	23 [14, 40]	54 [36,81]	26 [22,32]	13 [10,15]	< .0001
Days before HT	1 [0,13]	1 [0,16]	1 [0,14]	1 [0,9]	0.0046
Days after HT	15 [11, 23]	25 [14, 42]	17 [13, 23]	11 [9, 14]	< .0001
Elective admission for HT	3325 (65.1%)	1072 (63.2%)	1077 (64.1%)	1176 (68.1%)	0.0057
Patient in SNF	224 (4.4%)	211 (12.4%)	13 (0.8%)	0 (0%)	< .0001
Total days in SNF	0 [0,0]	0 [0,0]	0 [0,0]	0 [0,0]	< .0001
Total days in SNF among patients in SNF	33 [15, 62]	35 [19, 62]	9 [7, 12]	No patients in denominator	< .0001
Patients in hospice	18 (0.4%)	18 (1.1%)	0	0	< .0001
Total days in hospice	0 [0,0]	0 [0,0]	0 [0,0]	0 [0,0]	< .0001
Total days in hospice among patients in hospice	5.5 [1,13]	5.5 [1,13]	No patients in denominator	No patients in denominator	N/A
Patients in rehabilitation	NA	247 (14.6%)	48 (2.9%)	<11	< .0001
Total days in rehabilitation	0 [0,0]	0 [0,0]	0 [0,0]	0 [0,0]	< .0001
Total days in rehabilitation among patients in rehabilitation	20 [13, 31]	22 [14, 35]	12 [9,16]	7 [3,8]	< .0001
Readmission (12 months post discharge)
Readmission at the index hospital
Early (≤ 3 months)	2095(41.0%)	965 (56.9%)	867 (51.6%)	263 (15.2%)	< .0001
Late (4-12 months)	832 (16.3%)	215 (12.7%)	384 (22.9%)	233 (13.5%)	< .0001
Readmission at another hospital
Early (≤ 3 months)	336 (6.6%)	235 (13.9%)	77 (4.6%)	24 (1.4%)	< .0001
Late (4-12 months)	428 (8.4%)	229 (13.5%)	157 (9.4%)	42 (2.4%)	< .0001
Adverse Events
Acute organ rejection after transplantation
Early (≤ 3 months)	3417 (66.9%)	1106 (65.2%)	1161 (69.1%)	1150 (66.6%)	0.0485
Late (4-12 months)	376 (7.4%)	108 (6.4%)	132 (7.9%)	136 (7.9%)	0.1537
Cardiac Arrest
Early (≤ 3 months)	263 (5.2%)	165 (9.7%)	59 (3.5%)	39 (2.3%)	< .0001
Late (4-12 months)	69 (1.4%)	58 (3.4%)	<11	<11	< .0001
Cytomegalovirus
Early (≤ 3 months)	628 (12.3%)	230 (13.6%)	213 (12.7%)	185 (10.7%)	0.0485
Late (4-12 months)	533 (10.4%)	208 (12.3%)	197 (11.7%)	128 (7.4%)	0.1537
Endocarditis
Early (≤ 3 months)	234 (4.6%)	111 (6.5%)	57 (3.4%)	66 (3.8%)	<.0001
Late (4-12 months)	121 (2.4%)	62 (3.7%)	41 (2.4%)	18 (1.0%)	<.0001
Gastrointestinal infection
Early (≤ 3 months)	195 (3.8%)	137 (8.1%)	45 (2.7%)	13 (0.8%)	<.0001
Late (4-12 months)	182 (3.6%)	105 (6.2%)	52 (3.1%)	25 (1.5%)	<.0001
Pneumonia
Early (≤ 3 months)	881 (17.3%)	540 (31.8%)	236 (14.1%)	105 (6.1%)	<.0001
Late (4-12 months)	425 (8.3%)	215 (12.7%)	155 (9.2%)	55 (3.2%)	<.0001
Sternal Wound Infection
Early (≤ 3 months)	425 (8.3%)	281 (16.6%)	109 (6.5%)	35 (2.0%)	<.0001
Late (4-12 months)	132 (2.6%)	85 (5.0%)	33 (2.0%)	14 (0.8%)	<.0001
Urinary Tract Infection
Early (≤ 3 months)	319 (6.3%)	174 (10.3%)	107 (6.4%)	38 (2.2%)	<.0001
Late (4-12 months)	358 (7.0%)	176 (10.4%)	106 (6.3%)	76 (4.4%)	<.0001
Allograft Failure
Early (≤ 3 months)	369 (7.2%)	273 (16.1%)	68 (4.1%)	28 (1.6%)	< .0001
Late (4-12 months)	NA	20 (1.2%)	<11	no patients	< .0001
Dialysis
Early (≤ 3 months)	NA	144 (8.5%)	20 (1.2%)	<11	< .0001
Late (4-12 months)	76 (1.5%)	64 (3.8%)	<11	<11	< .0001
Stroke
Early (≤ 3 months)	316 (6.2%)	205 (12.1%)	72 (4.3%)	39 (2.3%)	< .0001
Late (4-12 months)	186 (3.6%)	86 (5.1%)	68 (4.1%)	32 (1.9%)	< .0001
Feeding tube
Early (≤ 3 months)	NA	30 (1.8%)	<11	no patients	< .0001
Late (4-12 months)	22 (0.4%)	22 (1.3%)	no patients	no patients	< .0001
Statin use at 1 month following discharge	1410 (27.6%)	405 (23.9%)	501 (29.8%)	504 (29.2%)	0.0001
Freedom from Dialysis
Early (≤ 3 months)	4934 (96.7%)	1553 (91.5%)	1660 (98.8%)	1721 (99.7%)	< .0001
Late (4-12 months)	4858 (95.2%)	1489 (87.7%)	1651 (98.3%)	1718 (99.5%)	< .0001
Cardiac Rehabilitation following discharge
Early (≤ 3 months)	1850 (36.2%)	304 (17.9%)	671 (39.9%)	875 (50.7%)	< .0001
Late (4-12 months)	742 (14.5%)	247 (14.6%)	291(17.3%)	204 (11.8%)	< .0001
Death
Early (≤ 3 months)	390 (7.6%)	390 (7.6%)	no patients	no patients	< .0001
Late (4-12 months)	NA	199 (11.7%)	<11	no patients	< .0001

Note: In compliance with the CMS suppression policy, all data involving less than 11 patients are not shown. Median [IQR] or n(%) are shown.

Gastrointestinal infection includes Clostridium difficile colitis, Cholecystitis, and Diverticulitis

Abbreviations: Heart transplantation- HT; Not reportable due to Medicare data suppression rules- N/A; Percent Days Alive and Outside of Hospital after Heart Transplantation- %DAOH-AF; Skilled Nursing Facility- SNF

Figure 1a-c. Distribution of Percentage of Days Alive and Out of Hospital (DAOH) Following Heart Transplantation.

Distribution of percent DAOH (Pareto diagram) following heart transplantation (percentage of DAOH) in the entire cohort (A), among patients who were alive at 1 year (B), and among patients who died (C). The x-axis and y-axis are specific to each individual graph in Figure 1.

The median %DAOH-AF varied appreciably by %DAOH-AF tercile: Low (71.8% [4.9%, 83.6%]) versus High (96.4% [95.9%, 97.3%]), p<0.0001. Patients with the highest versus lowest %DAOH-AF tercile were more often male, had lower rates of Medicare and Medicaid dual eligibility, had lower frequencies of several preoperative comorbidities (e.g. cerebrovascular, renal, hepatic disease) while less likely to be in preoperative cardiogenic shock (Table 1). There were no appreciable differences in patient age, race, or DCI across terciles or median %DAOH-AF (Figure S5). There was no significant effect by year across %DAOH-AF terciles, p=.0765.

The median hospital stay after HT was 15 days [11,23] for the cohort, Table 3. The length of the initial hospitalization was 14 days (25 versus 11, p<.0001) shorter for those in the highest versus lowest tercile of %DAOH-AF. Patients within the highest versus lowest %DAOH-AF tercile were more likely to live in the South and less likely to live in the Midwest, Table 4. Patients in the lowest tercile %DAOH-AF were more likely to receive a HT at a very large (500+ beds) hospital.

Table 4.

Hospital Characteristics According to Terciles of Percent Days Alive and Outside the Hospital at 12 Months after Heart Transplantation (%DAOH-AF).

		Home time (%DAOH-AF terciles)
Variables	Overall	Low (≤88.4%)	Intermediate (88.7-94.7%)	High (≥95%)	P value
	(n=5,104)	(n=1,697)	(n=1,680)	(n=1,727)
Location
Midwest	1120 (21.9%)	359 (21.2%)	400 (23.8%)	361 (20.9%)	< .0001
Northeast	1006 (19.7%)	417 (24.6%)	361 (21.5%)	228 (13.2%)
South	2089 (40.9%)	688 (40.5%)	624 (37.1%)	777 (45.0%)
West	889(17.4)	233 (13.7%)	295 (17.6%)	361 (20.9%)
Hospital Transplantation Volume
Quartile 1 (smallest)	303 (5.9%)	110 (6.5%)	102 (6.1%)	91 (5.3%)	0.0657
Quartile 2	966 (18.9%)	352 (20.7%)	314 (18.7%)	300 (17.4%)
Quartile 3	1280 (25.1%)	417 (24.6%)	436 (26.0%)	427 (24.7%)
Quartile 4 (largest)	2555 (50.1%)	818 (48.2%)	828 (49.3%)	909 (52.6%)
Hospital Structure
Bed size					0.0026
100-299	144 (2.8%)	37 (2.2%)	59 (3.5%)	48 (2.8%)
300-499	632 (12.4%)	185 (10.9%)	198 (11.8%)	249 (14.4%)
500+	4328 (84.8%)	1475 (86.9%)	1423 (84.7%)	1430 (82.8%)
Cardiac intensive care unit hospital beds,	32 (22,44)	32 (23,44)	32 (24,44)	29 (20,40)	< .0001
Academic	4958 (97.1%)	1634 (96.3%)	1636 (97.4%)	1688 (97.7%)	0.0296

Note: In compliance with the CMS suppression policy, all data involving less than 11 patients are not shown. Median [IQR] or n (%) are shown.

Abbreviations: Percent days alive and outside the hospital- %DAOH; Skilled Nursing Facility- SNF

Figure 2 shows daily cross-sectional distribution of patient’s status before and following HT. Patient post-discharge status (e.g., visits, admissions) differed significantly among %DAOH-AF terciles. The lowest versus highest %DAOH tercile used more SNF (12.4% versus no patients, p<.0001) and hospice (1.1% versus no patients, p<.0001) (Table 3).

Figure 2. Association Between Percentage of Days Alive and Out of Hospital (DAOH) and Adverse Events Following Heart Transplantation.

This figure represents the daily cross-sectional analysis of the distribution of patients after (365 days) heart transplantation. The colored area under each curve represents the proportion of patients in each setting. For each day, the sum of the proportion of patients is 100%. The proportion of patients in various locations (alive at home, dead, in a hospital, in a skilled nursing facility location, in a rehabilitation center, or hospice) is represented by each color. The graph represents 365 individual bar graphs showing the proportion of patients who were: (1) alive at home; (2) hospitalized in the transplanting hospital; (3) re-hospitalized in the index hospital; (4) hospitalized in another hospital; (5) in hospice; (6) in a rehabilitation center; (7) in a skilled nursing facility location; or were (8) dead on each day of follow-up in the entire cohort.

The overall readmission rate to index hospitals was 57.3%. Readmissions were greater in the early post-discharge period (≤ 90 days) in the lowest versus highest %DAOH-AF tercile (56.9% versus 15.2%, p<.0001). In the late post-discharge period (>90 days), the readmission rates at index hospitals were slightly reduced among the lowest versus highest %DAOH-AF terciles (12.7% versus 13.5% p<.0001). Figure S6 represents a Sankey diagram depicting the care trajectory following HT.

The rates of AEs before and after HT are displayed in Table S5. The rates of early, late, and cumulative AEs within 12 months of HT varied across %DAOH-AF terciles (Table 3; Figure S7). For instance, the differences in early AE rates across %DAOH-AF terciles were greatest for pneumonia (31.8% versus 6.1%) followed by sternal wound infections rates (16.6% versus 2.0%), both p<.0001. The differences in late AE across %DAOH-AF terciles were greatest for pneumonia (12.7% versus 3.2%) followed by urinary tract infections (10.4% versus 4.4%), both p<0.0001. Cardiac arrest, pneumonia, urinary tract infection, GI infection, endocarditis, and sternal wound infections were associated with an appreciable decrease in %DAOH across the entire cohort (Figure 3).

Figure 3. Association Between Percentage of Days Alive and Outside the Hospital (DAOH) and Adverse Events Following Heart Transplantation.

Percentage of DAOH following heart transplantation (percentage of DAOH) according to the presence or absence of the specified complications during follow-up. The line in bars represents the median and the boxes Q1 through Q3 for percentages of DAOH. All p-values were <.0001 except for acute organ rejection which is p=.0005

Discussion

This national evaluation of Medicare claims informs the use of %DAOH as a candidate patient-centered HT performance measure. Three findings were identified: (1) variability in risk-adjusted %DAOH existed across hospitals; (2) large variability exists in the %DOH 6 months prior to and first 12 months following HT; and (3) variability in %DAOH twelve months after HT was associated with early and late complications and markers of high hospital resource utilization.

Prior studies have evaluated the variability in %DAOH within the context of durable left ventricular assist device (dLVAD) implantation. Noly and colleagues used national Medicare claims linked to clinical registry data from The Society of Thoracic Surgeons Intermacs National Database (STS Intermacs) to evaluate the use of %DAOH as a potential dLVAD quality metric.¹⁰ The DAOH was calculated for this cohort of 3,387 patients within 180 days before (DOH-BF) and 1 year after dLVAD implantation (DAOH-AF). While the median [IQR] %DOH-BF dLVAD was 89% [82%-93%], a five percent absolute difference existed across the lowest (85%) versus highest (90%) terciles, p<.001.

The present study adds to the literature by evaluating risk-adjusted %DAOH across HT hospitals and variation of %DOH-BF across beneficiaries. The observed variability in %DOH-BF may be partly attributed by fewer elective admissions among the lowest versus highest %DOH-BF tercile (33.7% versus 95.8%, p<0.0001). High rates of elective admissions may be attributed to the study’s date range (i.e., prior to the 2018 UNOS Allocation policy change that prioritizes higher acuity patients).¹⁷ The study findings also suggest that %DOH-BF may be a potential indicator of post-HT AEs. Patients who experienced cardiac arrest or infection following HT had lower %DOH-BF and %DAOH-AF.

Several single-center studies have evaluated the role of DAOH following HT.^11,12,18 M’Pembele and colleagues undertook a single-center study of 187 patients undergoing HT between 2010-2020.¹¹ While the DAOH calculation was similar in most respects to the methodology applied within this study, M’Pembele distinctively considered facility admissions as DAOH. The mean (SD) age of their cohort was 55 (11) years, and the median DAOH was 295 days [223-322]—44 days lower than the present series. M’Pembele identified several predictors of DAOH one year following HT: recipient diabetes, recipient eGFR, mechanical ventilation, and postoperative renal replacement therapy. Roth and colleagues leveraged a similar cohort also from the same university hospital to evaluate the relationship between predictors of HT and mortality after HT.¹² The investigators identified Intermacs Patient Profile, pre-existing chronic kidney disease, use of an RVAD, postoperative dialysis, and tracheotomy as significant variables associated with decreased DAOH, with only dialysis being predictive of mortality. Bruce and colleagues leveraged a cohort of 235 patients from the Mayo Clinic to evaluate whether mechanical circulatory support as a bridge to transplantation adversely impacted post-HT DAOH.¹⁸ The investigators evaluated predictors of DAOH, which, unlike the approach applied in this present series, did not account for long-term care, SNF or hospice. The median (IQR) DAOH was 348 [335-354], which was similar for patients with (347 [336-353]) versus those without MCS (348 [334-354]), p=0.43. Post-UNOS HT donor allocation amendment policy, pulmonary hypertension, and total ischemic time were independently associated with reduced DAOH.

Days alive and outside of the hospital presents a multidimensional performance outcome following HT, given the incorporation of both survival and location outside of the hospital, the latter serving as a surrogate of patient HRQoL. The present study builds upon the work of Noly and colleagues by validating DAOH as a candidate HT metric. Similar to Noly and colleagues, %DOH-BF did not appreciably differ across adverse event categories. The median %DAOH-AF was lower among patients developing adverse events, including post-procedural cardiac arrest, pneumonia, gastrointestinal infections, and sternal wound infections. Further, patients in the lowest %DAOH-AF tercile were more likely to experience higher associated rates of early and late adverse events following HT.

Together, these findings support the use of %DAOH-AF as a candidate quality measure for HT. Two distinct advantages exist for using DAOH as an adjunct to existing hospital HT quality measure reporting. First, while the current reporting of metrics such as post-HT survival, survival on the waitlist, and duration of waiting time to HT are critical, these metrics lack granularity involving the patient’s disposition status and use of hospital resources. The DAOH measure leverages several constructs to evaluate not only the patient’s vital status, but the number of days an individual is living outside of an institutionalized setting (hospital, nursing home, rehabilitation center). Importantly, DOAH is a patient-centered metric. The DOAH metric may additionally assess changes in hospital resource utilization during waitlist periods, exemplified by the change in UNOS heart allocation amendment in 2018 that increased days hospitalized while awaiting HT. A strong association with adverse events further supports the validity of the DOAH measure. Second, the national implementation of the DAOH measure is facilitated by our team’s sole use of Medicare claims absent the need for patient contact. A similar approach within the setting of commercial payer data may be considered, although may be limited by a lack of longitudinal follow-up for patients who change commercial payer plans.

Limitations

The study should be interpreted in the context of several limitations. First, while the use of Medicare claims data may limit the generalizability of this study’s findings, Medicare: (1) is the largest national healthcare payer; (2) has covered an increasing number of HT over time¹⁹; and (3) provides a longitudinal unbiased ascertainment of patient disposition (e.g., vital status, location). Second, using administrative data might not accurately reflect waitlist status and limited the inclusion of clinically pertinent variables to risk-adjust %DAOH (e.g., recipient status at transplant, UNOS zone). However, this study showed an increasing percentage of beneficiaries waitlisted each month. Additionally, the study identified variation in risk-adjusted %DAOH across hospitals suggesting potential as a hospital quality marker. Third, while unmeasured confounding is a potential limitation in any observational study, this study utilized sensitivity analyses to confirm the robustness of the relationship between %DOH and %DAOH and preoperative characteristics. Fourth, while there is a risk of imprecisely identifying adverse events, this study leveraged validated administrative coding algorithms. Fifth, while this study evaluates %DAOH in the pre-2018 allocation era, the findings of this study remain pertinent given the limited patient-centeredness in existing hospital HT performance measures. Last, while this study did not track all important patient-centered measures (e.g., health-related quality of life), patients describe increased time at home as a positive sign of recovery.²⁰

Conclusion

In conclusion, marked variability in DAOH within the first year after HT exists among Medicare beneficiaries and across hospitals. This study highlights the usefulness of %DAOH as a quality measure for HT, given its ease of calculation and association with early and late occurring adverse events. Future work should evaluate determinants of and interventions to address unwarranted variability in %DAOH within the context of physician-led collaborative learning organizations.^21,22 Further, additional research should focus on evaluating the role and validity of %DAOH in the post-2018 era, including among non-Medicare populations.

Abbreviations:

AEs

Adverse Events

DOH

Days Outside the Hospital

DAOH

Days Alive and Outside the Hospital

DCI

Distressed Community Index

GI

Gastrointestinal

HT

Heart Transplantation

IQR

Interquartile Range

LVAD

Left Ventricular Assist Device

%DOH-BF

Percent Days Outside Hospital before Heart Transplantation

%DAOH-AF

Percent Days Alive and Outside Hospital after Heart Transplantation

SD

Standard deviation

UNOS

United Network of Organ Sharing