Telephone: 919-966-2516; FAX: 919-966-1700;
Article first published online: 22 APR 2012
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 18, Issue 5, pages 558–565, May 2012
How to Cite
Orman, E. S., Hayashi, P. H., Dellon, E. S., Gerber, D. A. and Barritt, A. S. (2012), Impact of nighttime and weekend liver transplants on graft and patient outcomes. Liver Transpl, 18: 558–565. doi: 10.1002/lt.23395
Eric S. Orman contributed to the study concept and design, the acquisition of data, the drafting of the manuscript, and the critical revision of the manuscript. Paul H. Hayashi contributed to the interpretation and analysis of data and the critical revision of the manuscript. Evan S. Dellon contributed to the study concept and design, the statistical analysis, the interpretation and analysis of data, and the critical revision of the manuscript. David A. Gerber contributed to the interpretation and analysis of data and the critical revision of the manuscript. A. Sidney Barritt IV contributed to the study concept and design, the statistical analysis, the interpretation and analysis of data, the drafting of the manuscript, and the critical revision of the manuscript.
This work was supported in part by the National Institutes of Health (grant 1KL2-RR025746-03 to A. Sidney Barritt IV and grant T32 DK07634 to Eric S. Orman).
This work was supported in part by the Health Resources and Services Administration (contract 231-00-0115). The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.
Telephone: 919-966-2516; FAX: 919-966-1700;
- Issue published online: 22 APR 2012
- Article first published online: 22 APR 2012
- Accepted manuscript online: 23 JAN 2012 07:32AM EST
- Manuscript Accepted: 15 JAN 2012
- Manuscript Received: 27 AUG 2011
Safety concerns have been raised about nighttime and weekend patient care, but it is unknown whether these issues affect liver transplantation. We sought to identify the impact of nighttime and weekend liver transplants on graft and patient survival. We used the United Network for Organ Sharing database to review adult liver transplants from 1987 to 2010. Comparisons were made between nighttime and daytime operations and between weekday and weekend operations. Cox proportional hazard ratios (HRs) were determined 30, 90, and 365 days after transplantation after we controlled for relevant factors; 94,768 transplants were included in the analysis. The patient survival rates at 30, 90, and 365 days for nighttime operations were 96%, 93%, and 86%, respectively. The patient survival rates at 30, 90, and 365 days for weekend operations were 95%, 92%, and 86%, respectively. These rates did not significantly differ from those for daytime and weekday operations, respectively. The graft failure rate was unchanged at 30 and 90 days for weekend transplants but was modestly increased at 365 days [HR = 1.05 (95% confidence interval = 1.01-1.11)]. Graft survival was unaffected by nighttime transplantation. Nighttime and weekend operations for liver transplantation do not affect patient or graft survival, and this testifies to the patient safety measures in place. Liver Transpl, 2012. © 2012 AASLD.
In 1999, the Institute of Medicine issued a report highlighting the substantial burden of medical errors, which were estimated to result in up to 98,000 deaths annually with an overall cost of $29 billion.1 These figures sparked considerable interest in systems-based initiatives to reduce medical errors and promote patient safety in various aspects of health care. Among the possible contributors to medical errors, one that has garnered particular attention is the provision of medical care outside regular daytime and weekday hours, which may be associated with an increased risk of medical errors because of provider fatigue and a lack of full available resources.
The associations between nighttime care, weekend care, and patient outcomes have been examined in multiple health care settings with conflicting results.2-9 Notably, organ transplantation has been the subject of multiple investigations because these procedures are commonly performed after hours on account of the timing of organ availability. The deleterious effect of nighttime kidney transplantation on patient outcomes has been seen inconsistently,10, 11 whereas the single study examining this relationship in the context of thoracic organ transplantation showed no difference.12 The only study evaluating the effect of the time of day on liver transplant outcomes showed no differences in various postoperative complications, but it did demonstrate an increased risk of early death in the group receiving nighttime transplants.13 Although these studies are helpful, many are single-center reviews and are limited by relatively small sample sizes and short study durations.
Therefore, using the United Network for Organ Sharing (UNOS) Standard Transplant Analysis and Research files, which contain information on all liver transplants performed in the United States since 1987, we undertook this study to determine whether the operative time of day and weekend operations are associated with adverse patient outcomes, such as reduced patient and graft survival.
PATIENTS AND METHODS
We used the UNOS Standard Transplant Analysis and Research data set files for liver transplants as of March 4, 2011. These files contained extensive patient-level data on all deceased and living donor transplants in the United States reported to the Organ Procurement and Transplantation Network between October 1, 1987, and December 31, 2010. Nighttime operations were defined as those started after 7 PM or before 7 AM. Operative start times were determined by the addition of the cold ischemia time to the cross-clamp time at procurement. Weekend operations were defined as those started between 5 PM on Friday and 8 AM on Monday. The Model for End-Stage Liver Disease (MELD) era was defined as the time period after February 27, 2002 (when the MELD scoring system was implemented as the method of allocation for liver transplants). All statistical analyses were performed with Stata 11 (StataCorp, College Station, TX). Approval for the study was granted by the University of North Carolina institutional review board.
Bivariate analyses were used to compare daytime and nighttime operations as well as weekday and weekend operations. Chi-square tests were used for categorical variables, the Student t test was used for normally distributed continuous variables, and the Kruskal-Wallis test was used for nonnormal continuous variables when it was appropriate. Kaplan-Meier analysis was used to generate survival tables.
Multivariate analysis with Cox proportional hazards modeling was used to calculate hazard ratios (HRs), which were adjusted for potential confounding and relevant recipient, donor, and operative factors. The main outcomes were all-cause mortality 30, 90, and 365 days after transplantation and graft failure at the same time points. The results of the bivariate analyses were used to select candidate variables for the models. These included the following: recipient race, sex, age, diabetes, body mass index (BMI), vasopressor support, dialysis at transplant, previous upper abdominal surgery, and portal vein thrombosis; pre-MELD era versus MELD era; warm and cold ischemia times; donor age; split liver donation; donation after cardiac death (DCD); donor cause of death, sex, and location (local, regional, national, or foreign); and retransplantation. Before modeling, potential interactions were assessed, and no significant effect modifiers were identified. Variables for which more than 33% of the data were missing were also excluded from multivariate analyses. The models were reduced with a backwards elimination strategy; we retained a priori variables that were significant at a P < 0.10 level or those thought to be clinically important. Post hoc sensitivity analyses were performed with different definitions of day, night, and weekend and for the MELD era so that MELD score could be included as a measure of pretransplant disease severity.
Study Population Characteristics
After the exclusion of patients who underwent transplantation before the age of 18 years, data from 94,768 liver transplants performed at 150 unique centers between 1987 and 2010 were analyzed (Table 1); 51,717 of these transplants (55%) occurred during the MELD era. The mean age at transplant was 51 ± 10.9 years, 63% of the patients were male, and 76% were Caucasian. Viral hepatitis (including hepatitis B and hepatitis C) was the indication for 34% of the transplants, and this was followed by alcoholic cirrhosis (13%) and cryptogenic cirrhosis/nonalcoholic steatohepatitis (11%). Fulminant hepatic failure was the cause for 6% of the population, whereas the other category, which included diagnoses such as autoimmune hepatitis, primary biliary cirrhosis, and primary sclerosing cholangitis, accounted for 36% of the population. The mean MELD score at transplant was 21 ± 9.4 for those patients undergoing transplantation in the MELD era. Twenty-one percent of the population had diabetes, and the median BMI was 27 kg/m2 (interquartile range = 24-31 kg/m2). A majority of the donors (60%) were male; the mean donor age was 38 ± 17 years. The median cold ischemia time was 7.6 hours (interquartile range = 5.6-10 hours). The overall posttransplant survival rates were 96% at 30 days, 92% at 90 days, and 86% at 1 year.
|Total transplants (n)||94,768|
|MELD era [n (%)]||51,717 (55)|
|Mean age at transplant (years)*||51 ± 10.9|
|Male sex [n (%)]||60,176 (63)|
|Caucasian [n (%)]||71,739 (76)|
|Mean MELD score at transplant*†||21 ± 9.4|
|Indication for transplantation [n (%)]|
|Acute liver failure||5945 (6)|
|Viral hepatitis||31,763 (34)|
|Retransplantation [n (%)]||9491 (10)|
|ABO blood group [n/N (%)]|
|Diabetes [n/N (%)]||15,975/77,485 (21)|
|Dialysis [n/N (%)]||6716/79,384 (8)|
|BMI (kg/m2)‡||26.9 (23.6-31.0)|
|Male sex [n/N (%)]||57,124/94,755 (60)|
|Age (years)*||38 ±17|
|Type [n/N (%)]|
|Deceased donor||75,493/77,735 (97)|
|Split liver [n (%)]||3932 (4)|
|Cause of death [n/N (%)]|
|CNS tumor||661/91,694 (1)|
|Donor location [n (%)]|
|Nighttime procurement [n (%)]||56,665 (60)|
|Nighttime transplantation [n/N (%)]||31,143/82,686 (38)|
|Weekend operation [n/N (%)]||32,079/91,659 (35)|
|Warm ischemia time (minutes)‡||45 (35-60)|
|Cold ischemia time (hours)‡||7.6 (5.6-10)|
|Previous abdominal surgery [n/N (%)]||31,928/76,376 (42)|
|Portal vein thrombosis [n/N (%)]||3267/60,361 (5)|
|Multiple vasopressors [n/N (%)]||1893/91,598 (2)|
|Overall graft survival (%)|
|Overall patient survival (%)|
Nighttime Operation and Effect on Survival
A bivariate analysis of the nighttime and daytime populations is shown in Table 2. A higher proportion of operations were performed at night during the MELD era (60% versus 50% P < 0.01). For patients undergoing transplantation in the MELD era, there was no difference in the MELD score at the time of transplantation (21.2 for daytime procedures versus 21.3 for nighttime procedures, P = 0.58). Although the mean age of patients undergoing transplantation at night was statistically significant, it was not clinically different from the mean age of the daytime group (51.1 years for the nighttime group versus 50.7 years for the daytime group, P < 0.01). This clinically insignificant but statistically significant difference also held for diabetes (21% versus 20%, P < 0.01), BMI (27.1 versus 26.9 kg/m2, P < 0.01), and the warm ischemia time (45 versus 46 minutes, P < 0.01). The cold ischemia time was shorter for nighttime operations versus daytime operations [7 hours (interquartile range = 5.2-9.4 hours) versus 8 hours (interquartile range = 6-10.5 hours), P < 0.01]. Daytime transplants were more likely to involve nighttime organ procurement (70% for daytime transplants versus 40% for nighttime transplants, P < 0.01). There was a statistically significant difference in the number of nighttime liver transplants versus daytime liver transplants during the weekend (42% versus 30%, P < 0.01) due to the definition of weekend. The patient and graft survival rates did not differ for daytime and nighttime transplants at 30 days (96% versus 96% for patients and 92% versus 93% for grafts), 90 days (92% versus 93% for patients and 88% versus 89% for grafts), and 365 days (86% versus 86% for patients and 81% versus 81% for grafts). The timing of transplants was also compared on an hourly basis, and no differences in patient survival were seen at 30, 90, or 365 days (Fig. 1).
|Daytime Procedures||Nighttime Procedures||P Value|
|Total transplants [n (%)]||51,543 (62)||31,143 (38)|
|MELD era [n (%)]||25,717 (50)||18,549 (60)||<0.01|
|Mean age at transplant (years)*||50.7 ± 11||51.1 ± 11||<0.01|
|Male sex [n (%)]||32,773 (64)||19,864 (64)||0.56|
|Caucasian [n (%)]||38,987 (76)||23,706 (76)||0.02|
|Mean MELD score at transplant*†||21.2 ± 10||21.3 ± 9||0.58|
|Indications for transplantation [n (%)]||<0.01|
|Acute liver failure||3057 (6)||2072 (7)|
|Viral hepatitis||17,399 (34)||10,250 (33)|
|Cryptogenic/NASH||5616 (11)||3484 (11)|
|Alcohol||7041 (14)||4048 (13)|
|Other||18,430 (36)||11,289 (36)|
|Retransplantation [n (%)]||5181 (10)||3244 (10)||0.09|
|Diabetes [n/N (%)]||8298/40,814 (20)||5534/26,009 (21)||<0.01|
|Dialysis [n/N (%)]||3546/41,978 (8)||2355/26,620 (9)||0.07|
|Median BMI (kg/m2)||26.9||27.1||<0.01|
|Male sex [n/N (%)]||31,118/51,535 (60)||18,911/31,139 (61)||0.32|
|Age (years)*||38.3 ± 17||38.1 ± 17||0.048|
|Type [n/N (%)]||<0.01|
|Deceased donor||41,581/42,552 (98)||25,869/26,902 (96)|
|DCD||971/42,552 (2)||1033/26,902 (4)|
|Split liver||514 (1)||366 (1)||0.42|
|Cause of death [n/N (%)]||<0.01|
|Anoxia||5564/40,814 (11)||3888/26,009 (12)|
|CVA||21,410/51,528 (42)||12,380/31,137 (40)|
|Trauma||20,468/40,814 (40)||12,684/26,009 (41)|
|CNS tumor||340/40,814 (1)||237/26,009 (1)|
|Other||3746/40,814 (7)||1948/26,009 (6)|
|Location [n (%)]||<0.01|
|Local||34,605 (67)||20,555 (66)|
|Regional||12,314 (24)||7141 (23)|
|National||4548 (9)||3413 (11)|
|Foreign||76 (0)||34 (0)|
|Nighttime procurement [n (%)]||35,859 (70)||12,364 (40)||<0.01|
|Weekend operation [n (%)]||15,279 (30)||13,084 (42)||<0.01|
|Warm ischemia time (minutes)‡||46 (36-60)||45 (35-60)||<0.01|
|Cold ischemia time (hours)‡||8 (6-10.5)||7 (5.2-9.4)||<0.01|
|Previous abdominal surgery [n (%)]||16,868/40,751 (41)||11,088/25,765 (43)||<0.01|
|Portal vein thrombosis [n (%)]||1628/30,437 (5)||1197/20,993 (6)||0.08|
|Multiple vasopressors [n (%)]||977/51,445 (2)||676/31,081 (2)||<0.01|
|Overall graft survival (%)|
|Overall patient survival (%)|
Unadjusted Cox proportional hazards modeling showed that in comparison with daytime liver transplant patients, nighttime liver transplant patients had a 0.96 hazard of death [95% confidence interval (CI) = 0.89-1.03] at 30 days, a 0.97 hazard of death (95% CI = 0.92-1.02) at 90 days, and a 0.99 hazard of death (95% CI = 0.95-1.03) at 365 days. A fully adjusted model was created that controlled for age, race, sex, diabetes, BMI, vasopressor support, dialysis, previous upper abdominal surgery, portal vein thrombosis, retransplantation, donor location, pre-MELD era versus MELD era, diagnosis, donor age, cold and warm ischemia times, nighttime organ procurement, donor cause of death, and weekday operations versus weekend operations. This model was then reduced by backwards elimination. Other models were created through alterations in the daytime and nighttime definitions (eg, 3 AM to 3 PM per Lonze et al.13) and through the substitution of variables that were not included by the original a priori methodology. Additionally, a separate analysis was performed for the MELD era so that MELD score could be included as a measure of disease severity. The results for each of these models were unchanged from those for the unadjusted model (data not shown). The final model, which controlled for weekend operations, nighttime procurement, cold ischemia time, donor age, split livers, DCD, diagnosis, recipient age and sex, dialysis, vasopressor use, previous upper abdominal surgery, portal vein thrombosis, retransplantation, donor location, and pre-MELD era versus MELD era, also showed no differences in graft or patient survival at 30, 90, or 365 days (Table 3).
|HR (95% CI)|
|30 days||0.93 (0.88-0.98)||1.00 (0.92-1.08)||0.96 (0.89-1.03)||0.94 (0.85-1.05)|
|90 days||0.94 (0.90-0.98)||1.00 (0.94-1.06)||0.97 (0.92-1.02)||0.98 (0.90-1.06)|
|1 year||0.96 (0.94-1.00)||1.01 (0.96-1.06)||0.99 (0.95-1.03)||0.99 (0.94-1.05)|
|30 days||1.03 (0.98-1.08)||1.04 (0.96-1.13)||0.99 (0.93-1.07)||0.93 (0.84-1.04)|
|90 days||1.03 (0.98-1.07)||1.04 (0.97-1.10)||1.01 (0.96-1.06)||0.98 (0.91-1.06)|
|1 year||1.02 (0.99-1.06)||1.05 (1.01-1.11)||1.02 (0.98-1.06)||1.02 (0.97-1.08)|
Weekend Operation and Effect on Survival
A bivariate analysis of the weekday and weekend populations is shown in Table 4. Although there were statistically significant differences, none of the differences were clinically meaningful. A higher proportion of weekend operations were performed in the MELD era (55% versus 54%, P = 0.01). For patients undergoing transplantation in the MELD era, the MELD scores at the time of transplantation were higher on the weekend (21.6 versus 21.2, P < 0.01). There were no differences in the median BMIs or the proportions of patients with diabetes when weekend and weekday transplants were compared. On weekends, organ procurement was less likely to occur at night (53% versus 61%, P < 0.01), but transplants were more likely to occur at night (46% versus 33%, P < 0.01) because of the definition of weekend. The warm ischemia times did not differ between weekends and weekdays. The cold ischemia times were shorter on weekends (7.5 versus 8.0 hours, P < 0.01). The patient and graft survival rates were not different at 30, 60, or 365 days for weekend transplants versus weekday transplants.
|Total transplants [n (%)]||59,580 (65)||32,079 (35)|
|MELD era [n (%)]||32,019 (54)||17,517 (55)||0.01|
|Mean age at transplant (years)*||50.9 ± 11||50.8 ± 11||0.8|
|Male sex [n (%)]||37,969 (64)||20,431 (64)||0.91|
|Caucasian [n (%)]||44,921 (75)||24,315 (76)||0.04|
|Mean MELD score at transplant*†||21.2 ±9||21.6 ± 10||<0.01|
|Indication for transplantation [n (%)]||0.08|
|Acute liver failure||3698 (6)||2141 (7)|
|Viral hepatitis||20,051 (34)||10,678 (33)|
|Cryptogenic/NASH||6520 (11)||3516 (11)|
|Alcohol||7922 (13)||4302 (13)|
|Other||21,389 (36)||11,442 (36)|
|Retransplantation [n (%)]||6079 (10)||3339 (10)||0.33|
|Diabetes [n/N (%)]||9966/48,291 (21)||5487/26,193 (21)||0.32|
|Dialysis [n/N (%)]||4233/49,578 (9)||2439/26,861 (9)||0.01|
|BMI (kg/m2)‡||27 (24-31)||27 (24-31)||0.99|
|Male sex [n (%)]||36,109/59,572 (61)||19377/32,075 (60)||0.71|
|Age (years)*||38.3 ± 17||37.9 ± 17||<0.01|
|Type [n/N (%)]||0.08|
|Deceased donor||48,883/50,375 (97)||26,608/27,358 (97)|
|DCD||1492/50,375 (3)||750/27,358 (3)|
|Split liver [n (%)]||762 (1)||336 (1)||<0.01|
|Cause of death [n/N (%)]||<0.01|
|Anoxia||6944/59,565 (12)||3509/32,073 (11)|
|CVA||24,514/59,565 (41)||12,981/32,073 (40)|
|Trauma||23,683/59,565 (40)||13,270/32,073 (41)|
|CNS tumor||406/59,565 (1)||254/32,073 (1)|
|Other||4018/59,565 (7)||2059/32,073 (6)|
|Location [n (%)]||0.01|
|Local||40,110 (67)||21,353 (67)|
|Regional||13,947 (23)||7546 (24)|
|National||5451 (9)||3136 (10)|
|Foreign||72 (0)||44 (0)|
|Nighttime procurement [n (%)]||36,346 (61)||17,210 (54)||<0.01|
|Nighttime transplantation [n (%)]||18,059/54,323 (33)||13,084/28,363 (46)||<0.01|
|Warm ischemia time (minutes)‡||45 (36-60)||46 (35-60)||0.09|
|Cold ischemia time (hours)‡||8.0 (6-10)||7.5 (5.6-10)||<0.01|
|Previous abdominal surgery [n/N (%)]||19,926/47,696 (42)||10,871/25,747 (42)||0.24|
|Portal vein thrombosis [n/N (%)]||2089/37,327 (6)||1084/20,353 (5)||0.17|
|Multiple vasopressors [n/N (%)]||1195/59,449 (2)||696/32,006 (2)||0.1|
|Overall graft survival (%)||0.21|
|Overall patient survival (%)||0.26|
Unadjusted Cox proportional hazards modeling showed that in comparison with weekday liver transplant patients, weekend liver transplant patients had a 0.99 hazard of death (95% CI = 0.93-1.07) at 30 days, a 1.01 hazard of death (95% CI = 0.96-1.06) at 90 days, and a 1.02 hazard of death (95% CI = 0.98-1.06) at 365 days. A fully adjusted model was created that controlled for age, race, sex, diabetes, BMI, vasopressor support, dialysis, previous upper abdominal surgery, portal vein thrombosis, retransplantation, donor location, pre-MELD era versus MELD era, diagnosis, donor age, cold and warm ischemia times, nighttime organ procurement, donor cause of death, and daytime operations versus nighttime operations. Other models were created similarly to those previously mentioned. The results for each of these models were unchanged from the results for the unadjusted model (data not shown). The final model showed no differences in patient survival at 30, 90, or 365 days. The final model, which controlled for nighttime operations, nighttime procurement, cold ischemia time, donor age, split livers, DCD, diagnosis, recipient age and sex, dialysis, vasopressor use, previous upper abdominal surgery, portal vein thrombosis, retransplantation, donor location, and pre-MELD era versus MELD era, showed a small increase in the hazard of graft failure at 365 days only [1.05 (95% CI = 1.01-1.11); Table 3]. The timing of transplants was also compared on a daily basis, and no differences in patient or graft survival were seen at 30, 90, or 365 days.
This retrospective review of nearly 100,000 liver transplants performed over more than 2 decades shows no differences in graft or patient survival when operations were performed during daytime hours versus nighttime hours. For weekend transplants, there were no differences in patient survival at any time point or in graft survival after 30 and 90 days, although we did detect a slight decline in graft survival after 1 year. This study shows findings similar to those of the recent UNOS registry study by George et al.12 for thoracic transplants, but it presents findings contradicting smaller studies performed by Lonze et al.13 for liver transplants and by Fechner et al.10 for kidney transplants.
Our study provides reassurance about outcomes to the transplant community and patients awaiting transplantation. It highlights patient safety mechanisms, that is, the availability of appropriate numbers of surgeons and ancillary staff after hours and on weekends. Call schedules that are already in place allow patients undergoing liver transplantation after hours and on the weekend to benefit from the same survival outcomes enjoyed by patients undergoing these procedures during the standard workday. Because the timing of transplantation can rarely be predicted or scheduled, the lack of a decline in safety at night and on weekends is reassuring.
When a data set of this size is being analyzed, care must be taken to not misinterpret statistically significant results that are not clinically important. Many variables are trivially different, such as the age at transplant (51.1 years for nighttime operations versus 50.7 years for daytime operations, P < 0.01), yet because of the large sample size and tremendous power, they are highly statistically significant.
There was a statistically significant decline (5%) in graft survival 365 days after transplantation when the surgery occurred on the weekend. One plausible explanation may be the surgical team's call schedule. Both patient survival and graft survival decline when the primary surgeon has performed a high volume of transplants within a short period of time.14 At centers where the primary surgical team is on call for a week at a time or for the whole weekend, there may be a resultant decrease in graft survival later in the weekend if the team has performed many operations in the preceding days. Presumably, such a decrease would be due to impaired technical performance or perioperative decision making, both of which could affect 1-year outcomes. Another explanation for a decline in graft survival at 1 year could involve the support staff for surgical services. During the weekend, these individuals may form a cross-disciplinary team involved in a heterogeneous set of operative cases, including trauma, orthopedics, and surgical emergencies. During weekdays, the surgical team is likely to be composed of dedicated transplant personnel. Procurement may be left to local surgical teams at non–liver transplant centers more often on weekends, whereas transplant teams may travel to the donors on weekdays and weeknights. Perioperative support, including the physical presence of multiple medical experts (eg, dedicated transplant pharmacists), may be less available on weekends. Therefore, there are plausible reasons for weekend transplants to have worse graft survival.
However, one might expect such effects to manifest earlier at 30 and 90 days and at 1 year. Also, our study examined multiple outcomes at various time points (12 adjusted HRs were measured in all). In these circumstances, it is not unusual to find a statistically significant result by chance (ie, a type I error) or an epiphenomenon.
Data on the reasons for graft failure are collected in the UNOS database. No differences in the etiologies of graft failure were seen in bivariate analyses of biliary complications, acute or chronic rejection, de novo hepatitis, primary disease recurrence, or infection when weekday transplants were compared to weekend transplants. However, these data must be interpreted with caution, and the analyses are not presented because of the large number of missing data.
Other studies have raised concerns about nighttime and weekend procedures as well as nighttime and weekend admissions for a variety of medical problems. Studies of gastrointestinal bleeding,5, 15, 16 intensive care unit admission,17 stroke admission,3 and renal failure4 have all found a deleterious impact from either night or weekend admissions. Most of these studies are from single centers and, in comparison with this study, are markedly smaller.
Database registry research has its limitations. The UNOS database lacks the granular patient information that many single-center studies are able to access. The inclusion of such factors may alter the results, and thus residual confounding is a concern. Additionally, coding or data entry errors cannot be resolved at the level of individual patients. If these errors occur in a nonrandom fashion, the results may be biased, but there is no strong reason to suspect this.
Our definition of the start time may be subject to scrutiny. UNOS does not record the actual time at which the recipient surgery is started, so we chose to add the cold ischemia time to the donor cross-clamp time as an approximation of the start time. In practice, this would reflect the time at which the graft is placed into the recipient and would, therefore, exclude the hepatectomy, which may be the most difficult part of the operation and can take several hours. The assignment of daytime operations versus nighttime operations is arbitrary as well. The decision to use 7 AM to 7 PM to define daytime was based on the study by George et al.12 and was an attempt to provide some consistency across our investigations. In an attempt to overcome this arbitrary day/night definition, other divisions were analyzed to facilitate comparisons with other relevant studies, such as the study by Lonze et al.13 The bivariate hour-by-hour analysis also showed no differences (Fig. 1), and this mitigated any variability that could be attributed to either the start-time definition or the day/night definition. The fact that there were no differences in the outcomes no matter how the data were analyzed gives some strength to our ultimate conclusions.
In summary, according to a large multicenter registry database containing nearly 100,000 liver transplants and spanning more than 20 years, neither nighttime liver transplantation nor weekend liver transplantation has an adverse impact on patient or graft survival. The lack of differences in patient and graft outcomes is a testament to current practices covering off-hour and weekend transplants. These data should provide some reassurance to patients and transplant centers that posttransplant outcomes are not significantly affected by the timing of surgery. With increasing public focus on patient safety, the collection of more precise data on the time of transplantation and available personnel may be warranted to provide better vigilance for off-hour transplants in the future.
- 1To Err Is Human: Building A Safer Health System. Washington, DC: National Academies Press; 2000., , .