The effect of organ–recipient gender match on pediatric heart transplant mortality is unknown. We analyzed the effects of gender and donor–recipient gender matching. Based on Organ Procurement and Transplant Network data, we performed a historical cohort study in a population of 3630 heart transplant recipients less than 18 years old. We compared unadjusted and adjusted mortality by recipient gender, donor gender and between gender-matched and gender-mismatched recipients. Female recipients had decreased survival compared to male recipients (unadjusted hazard ratio [HR] 1.16, confidence interval [CI] 1.02–1.31; p = 0.020). Organ–recipient gender mismatch did not affect mortality for either male or female recipients, though gender-mismatched females had the worst survival compared to gender-matched males, who had the best survival (unadjusted HR 1.26, CI 1.07–1.49; p = 0.005). After adjustment for other risk factors affecting transplant mortality, female recipients had decreased survival compared to male recipients (HR 1.27, CI 1.12–1.44; p = 0.020) and gender matching had no effect. In conclusion, gender mismatch alone did not increase long-term mortality for pediatric heart transplant recipients. However, there may be additive effects of gender and gender matching affecting survival. There are insufficient data at this time to support that recipient and donor gender should affect heart allocation in children.
body surface area
extracorporeal membrane oxygenation
Organ and Procurement Transplant Network
panel reactive antibodies
Standard Transplant Analysis Report
United Network for Organ Sharing
Multiple adult heart transplantation studies suggest that organ–recipient gender match improves late mortality [1-9]. The factors impacting this survival advantage are unknown but may include differences in heart mass and ratio of body surface area (BSA) [5, 7, 10]. Although well-designed adult heart transplant studies have not found an isolated effect of organ gender on mortality, these data have led to a recommendation that men should receive hearts from male donors . Whether these survival statistics are similar in pediatric patients is unknown. Multi-decade survival after pediatric heart transplantation has been difficult to achieve due to an inability to reduce late mortality despite steady increases in early pediatric heart transplant survival [11-16]. If gender matching improves survival in pediatric patients, emphasis on promoting organ–recipient gender match could improve late mortality.
Determination of the separate effects of recipient gender and organ gender as well as their interaction is necessary to analyze of the effect of organ–recipient gender match in pediatric heart transplantation. Interestingly, female organ gender and female recipient gender have been reported as risk factors for mortality in the 11th Official Pediatric Heart Transplantation Report . However, the effect of organ–recipient gender matching in pediatric heart transplantation has not been determined and may confound the effect of gender on mortality.
We conducted a historical cohort study to elucidate the effect of gender including recipient gender, donor gender and organ–recipient gender match on pediatric heart transplantation mortality in a large population of pediatric heart transplant recipients. To detect an effect on late mortality, we utilized a statistical analysis controlling for covariates known to influence long-term survival. We hypothesized that individuals with organ–recipient gender mismatch and female recipients would have increased mortality while organ gender would not affect mortality.
Materials and Methods
This study utilized a United Network for Organ Sharing (UNOS) Standard Transplant Analysis Report (STAR) File containing Organ and Procurement Transplant Network (OPTN) data as of May 2008. The STAR File included all the available data for first-time heart transplant recipients younger than 18 years of age at the time of transplant who were transplanted in the United States between January 1987 and May 2008. This study is exempt from IRB approval because the OPTN data used contains no patient identifiers.
All patients in the data set who were transplanted in the time period of 1994–2008 were included. Exclusion of recipients transplanted prior to 1994 ensured that the study cohort data included most variables shown to affect long-term survival in pediatric heart transplantation (all variables were not recorded prior to 1994). Other exclusion criteria consisted of repeat heart transplant recipients, multi-organ transplant recipients and omission of gender, death or follow-up time.
Patients were grouped into four categories according to their gender and organ–recipient gender match. Recipient and donor age and transplant era were categorized into groups based on clinical definitions for the purpose of the regression analysis. We examined the relationship between each potential risk factor one at a time versus mortality using univariate Cox regression. The chi-square/Fisher test was used to compare categorical baseline characteristics between groups. Continuous baseline characteristics were compared via the Kruskal–Wallis test. Survival curves were constructed using Kaplan–Meier analysis and compared between male and female recipients, between recipients who received hearts from male donors and recipients who received hearts from female donors, and between gender-matched and gender-mismatched recipients using the log rank test.
We compared mortality between the above-mentioned groups before and after adjusting for covariates using Cox proportional hazard regression using two methods of multivariate adjustment. First, a multivariate Cox model was constructed adjusted for known risk factors for pediatric heart transplantation mortality and model-based hazard ratios (HR) were generated. The initial multivariate model included all potential risk factors, and the final model was selected using a backward stepwise procedure with a liberal p < 0.25 level of significance as the retention criterion. The records of 556 patients did not report serum bilirubin, so missing values were multiply imputed for the purpose of the multivariate analysis using chained equations. The final HR estimates were obtained by pooling the results across five imputations. Second, we used a propensity score adjusted Cox model. The propensity score was defined as the conditional probability of obtaining a gender-mismatched heart given the covariates, calculated using multiple logistic regression. Propensity score adjustments were made by including the propensity score by quartile in the Cox model. Statistical significance was set at p < 0.05.
The analysis yielded a cohort of 3630 first-time heart transplant recipients after removing 38 recipients for critical missing data (Figure 1). Comparison of baseline characteristics revealed that the groups differed significantly with regard to factors known to effect heart transplant mortality (Table 1). Male subjects receiving hearts from male donors tended to be older than other subjects, and the ratio of donor to recipient BSA was closer to one. Male recipients tended to have more comorbidities including congenital heart disease (52.4% of mismatched and 46.8% of gender-matched males). Gender-mismatched males also were on dialysis, required mechanical ventilation and were supported by extracorporeal membrane oxygenation (ECMO) most often.
|Female to female||Female to male||Male to female||Male to male||p-Value1|
|Recipient age (mean ± SD, years)||6.06 ± 6.05 (n = 677)||6.08 ± 6.19 (n = 869)||5.86 ± 5.82 (n = 906)||7.15 ± 6.50 (n = 1170)||0.00002|
|Congenital heart disease||262/680 (38.5%)||456/870 (52.4%)||365/908 (40.2%)||549/1172 (46.8%)||<0.00001|
|Serum creatinine (mean ± SD, mg/dL)||0.76 ± 1.58 (n = 641)||0.70 ± 1.36 (n = 825)||0.78 ± 1.80 (n = 861)||0.81 ± 1.96 (n = 1109)||0.0003|
|Dialysis prior to transplant||15/647 (2.3%)||23/840 (2.7%)||21/876 (2.4%)||16/1136 (1.4%)||0.19|
|Total bilirubin (mean ± SD, mg/mL)||1.93 ± 5.89 (n = 575)||2.13 ± 4.07 (n = 728)||1.64 ± 3.67 (n = 765)||1.77 ± 3.20 (n = 1006)||0.035|
|Hospitalized awaiting transplant||462/679 (68.0%)||595/864 (68.9%)||611/906 (67.4%)||773/1169 (66.1%)||0.608|
|Mechanical support prior to transplant||n = 680||n = 870||n = 908||n = 1172|
|ECMO||39 (5.7%)||69 (7.9%)||52 (5.7%)||62 (5.3%)||0.079|
|VAD||51 (7.5%)||60 (6.9%)||56 (6.2%)||109 (9.3%)||0.044|
|Intra-aortic balloon pump||6 (0.9%)||9 (1.0%)||1 (0.1%)||8 (0.7%)||0.087|
|Mechanical ventilation||111 (16.3%)||182 (20.9%)||177 (19.6%)||200 (17.1%)||0.052|
|Donor age (mean ± SD, years)||9.49 ± 11.56 (n = 677)||9.96 ± 12.05 (n = 866)||7.45 ± 8.32 (n = 907)||9.72 ± 10.22 (n = 1171)||0.002|
|Graft ischemic time (mean ± SD, hours)||3.56 ± 1.31 (n = 624)||3.68 ± 1.32 (n = 797)||3.67 ± 1.37 (n = 824)||3.56 ± 1.29 (n = 1084)||0.164|
|African-American donor to African-American recipient||33/680 (4.9%)||25 (2.9%)||26 (2.9%)||52 (4.4%)||0.054|
|Ratio donor to recipient body surface area||1.23 ± 0.34 (n = 619)||1.26 ± 0.41 (n = 796)||1.25 ± 0.35 (n = 826)||1.21 ± 0.33 (n = 1076)||0.013|
|Era of transplant||n = 680||n = 870||n = 908||n = 1172|
|1994–1996||124 (18.2%)||184 (21.1%)||161 (17.7%)||236 (20.1%)||0.230|
|1997–1999||134 (19.7%)||182 (20.9%)||178 (19.6%)||253 (21.6%)||0.651|
|2000–2003||191 (28.1%)||243 (27.9%)||277 (30.5%)||326 (27.8%)||0.523|
|2004–2008||231 (34.0%)||261 (30.0%)||292 (32.2%)||357 (30.5%)||0.309|
The final Cox model included all predictors that satisfied p < 0.25 retention criteria using backward stepwise search. The sample size was sufficiently large to allow the inclusion of all 15 risk factors in the model. Most have been reported in previous studies [17, 18]. Results are described in Table 2. Predictors that were not significant included BSA, head trauma as a cause of death and panel reactive antibodies (PRA), and were excluded from the final model.
|<5 years||1.00 (0.78–1.28)||0.998|
|>12 years||1.37 (1.12–1.67)||0.002|
|Congenital heart disease||1.51 (1.32–1.72)||<0.001|
|Elevated serum creatinine||1.18 (1.00–1.38)||0.049|
|Dialysis prior to transplant||2.01 (1.40–2.87)||<0.001|
|Elevated total bilirubin||1.20 (1.05–1.37)||0.007|
|Mechanical support prior to transplant|
|Mechanical ventilation||1.19 (1.00–1.41)||0.049|
|Donor age (reference)|
|<7 years||1.26 (0.98–1.62)||0.075|
|>25 years||1.39 (1.12–1.74)||0.003|
|African-American donor to African-American recipient||1.59 (1.18–2.15)||0.002|
|Era of transplant|
We examined the effects of gender and gender matching before and after adjusting for covariates. The univariate Cox regression analyses revealed that female recipients had increased mortality risk compared to male recipients (HR 1.16, confidence interval [CI] 1.02–1.31; p = 0.02). Gender-mismatched females as a group had the worst survival compared to gender-matched males with the best survival (HR 1.26, CI 1.07–1.49; p = 0.005). There was not a significant difference in mortality between gender-matched males and gender-matched females (HR 1.17, CI 0.97–1.40; p = 0.098). There was also not a significant difference in mortality between gender-mismatched males and gender-matched males (HR 1.13, CI 0.96–1.34; p = 0.15), or between gender-mismatched females and gender-matched females (HR 1.08, CI 0.90–1.31; p = 0.39; Table 3). Kaplan–Meier curves were generated to examine unadjusted survival by gender match. Survival was different across groups (p = 0.0431; log-rank test; Figure 2).
|Group comparison||HR||p-Value||Adjusted HR||p-Value||HR adjusted for propensity score1||p-Value|
|Female recipients vs. male recipients||1.16 (1.02–1.31)||0.020||1.27 (1.12–1.44)||0.020|
|Female hearts vs. male hearts||1.03 (0.91–1.17)||0.605||1.00 (0.88–1.13)||0.948|
|Gender-mismatched males vs. gender-matched males||1.13 (0.96–1.34)||0.148||1.09 (0.92–1.29)||0.330||1.13 (0.96–1.34)||0.149|
|Gender-matched females vs. gender-matched males||1.17 (0.97–1.40)||0.098||1.23 (1.02–1.48)||0.029||1.19 (0.99–1.43)||0.062|
|Gender-mismatched females vs. gender-matched males||1.26 (1.07–1.49)||0.005||1.38 (1.17–1.63)||0.000||1.30 (1.10–1.54)||0.002|
|Gender-mismatched females vs. gender-matched females||1.08 (0.90–1.31)||0.394||1.12 (0.93–1.36)||0.330||1.09 (0.91–1.32)||0.341|
We compared Cox model adjusted mortality across gender and gender-matched groups after adjusting for variables described in Table 2 (Table 3). After adjusting for the other variables, female recipients had decreased survival compared to male recipients (HR 1.27, CI 1.12–1.44; p = 0.02). Gender-matched males had the best survival compared to gender-mismatched females who had the worst survival (HR 1.38, CI 1.17–1.63; p < 0.001). Gender-matched females had increased mortality compared to gender-matched males (HR 1.23, CI 1.02–1.48; p = 0.029). After adjustment for propensity score, results were qualitatively similar to the covariate adjusted Cox model, although this difference did not reach p < 0.05 (HR 1.19, CI 0.99–1.43; p = 0.062). There was not a significant difference in mortality between gender-mismatched males and gender-matched males (HR 1.09, CI 0.92–1.29; p = 0.33), or between gender-mismatched females and gender-matched females (HR 1.12, CI 0.03–1.36; p = 0.33). Both multivariate methods provided similar results, confirming that our findings were robust, regardless of model assumptions.
To further explore the gender-based differences, we examined short-term (30 day, 90 day and 1 year) and conditional long-term (5 year) survival. At 90 days, unadjusted survival between groups differed (p = 0.028); however, it was not significantly different at 30 days or at 1 year after transplant (Table 4). Long-term survival at 5 years was significantly different conditional on survival for 90 days, but was not significantly different conditional on survival for 30 days or 1 year (Table 5).
|Survival time||Female to female||Female to male||Male to female||Male to male||p-Value1|
|n = 680||n = 870||n = 908||n = 1172|
|30 days||93.1 (91.2–95.0)||91.8 (90.0–93.7)||92.4 (90.7–94.2)||94.4 (93.0–95.7)||0.133|
|90 days||90.1 (87.8–92.4)||87.4 (85.1–89.6)2||89.9 (87.9–91.9)||91.5 (89.9–93.2)||0.028|
|1 year||83.7 (80.9–86.6)||83.2 (80.7–85.8)||84.0 (81.6–86.5)||87.0 (85.0–88.9)||0.081|
|Condition||Female to female||Female to male||Male to female||Male to male||p-Value1|
|n = 680||n = 870||n = 908||n = 1172|
|Survived 30 days||75.3 (71.3–79.3)||77.3 (73.9–80.6)||74.5 (70.9–78.1)||78.9 (76.0–81.7)||0.107|
|Survived 90 days||77.8 (73.8–84.3)||80.9 (77.6–84.3)2||75.5 (71.8–79.2)||81.3 (78.5–84.1)2||0.015|
|Survived 1 year||82.0 (78.0–86.1)||83.5 (80.1–86.9)||78.3 (74.4–82.2)||83.9 (81.1–86.6)||0.060|
We calculated the smallest effect size that could be detected from the data available. Given the size of the pediatric cohort we investigated, the maximal effect that could have been present with 95% confidence would be a 7.74% decrease in adjusted survival at 5 years for female recipients receiving gender-mismatched hearts versus gender-matched and a 5.79% decrease in adjusted survival among male recipients who are mismatched versus matched (Table 6). The observed survival benefit is 1.52% for males and 1.67% for females calculated for this cohort (Table 6).
|Time||Adjusted survival||Difference in survival|
|%surviving (±SE)||% (CI range)|
|Year 1||87.43 (0.80)||86.35 (0.93)||−1.08 (−3.47–1.32)|
|Year 3||81.34 (1.08)||79.81 (1.26)||−1.53 (−4.79–1.72)|
|Year5||76.50 (1.30)||74.63 (1.52)||−1.87 (−5.79–2.06)|
|Year 1||84.73 (1.12)||83.00 (1.06)||−1.73 (−4.75–1.29)|
|Year 3||77.51 (1.52)||75.10 (1.41)||−2.41 (−6.47–1.64)|
|Year 5||71.86 (1.82)||68.96 (1.67)||−2.89 (−7.74–1.96)|
This study represents the first large population-based study designed to assess the effect of both gender and gender matching in long-term pediatric heart transplantation mortality. In a population of 3630 patients, we find that male recipients have improved survival compared to female recipients. We also demonstrated, by both our univariate and multivariate analyses, that gender-matched males have the best survival outcome and gender-mismatched females have the worst survival. Based on the differences noted in the baseline characteristics of these groups, size differences and difficulties in size matching for female recipients may contribute. However, gender matching did not improve survival among either male or female recipients.
The 10th Official Pediatric Heart Transplantation Report found that recipients of female donor hearts have increased 5-year mortality . However, that analysis of the effect of donor organ gender on mortality was likely confounded by the effect of recipient gender we found in our study. By isolating the effect of gender matching, we found that donor organ gender did not significantly affect survival for male or female recipients.
Analysis of the gender effect on mortality has been studied in nontransplant pediatric cardiac surgical patients as well. In large populations of pediatric congenital heart surgical patients, previous studies reported increased in-hospital and 30-day postdischarge mortality (OR 1.51 for in-hospital and OR 1.18 for in-hospital plus postdischarge) for female patients [19-21]. The etiology for the increased mortality of females is unclear and likely multifactorial in origin, but may include hormonal or genetic differences.
In Weiss et al's  adult heart transplantation study of 18 240 patients, gender-matched males had the best survival outcome and gender-mismatched females had the worst survival (HR 1.25, CI 1.07–1.43; p = 0.003). Gender-mismatched males had increased mortality compared to gender-matched males (HR 1.15, CI 1.02–1.30; p = 0.02) and there was no difference between gender-matched and gender-mismatched females (HR 1.24, CI 0.92–1.35; p = 0.31) . The large size of the adult heart transplantation population allows for detection of a small but statistically significant difference in mortality based on gender match for male recipients. Because of the size of the cohort we investigated, we could not exclude the possibility of a very small effect of gender matching. However, whether the possibility of increasing survival by a maximum of 5.79% in males or 7.74% in females warrants selecting only gender-matched organs is unclear.
In addition, in the adult study, mean donor–recipient BSA ratio was 0.9 in both gender-matched and unmatched males . Donor–recipient BSA ratio in our pediatric sample was >1.2 in all gender groups. Effect of size mismatch is unclear, but pediatric subjects tended to receive adequate or oversized hearts. Weight ratio has not been demonstrated to be an important factor in pediatric recipients . In adults, weight ratio has not been shown to predict mortality except in the setting of high pulmonary vascular resistance . It is possible that donor gender may have more impact in adult males if donor size is not as favorable or if statistical adjustment for effect of size was not adequate.
Nonetheless, some of our findings in pediatric patients are consistent with those in adults that female recipients fare worse than male recipients and gender-matched males, as a group, have the best survival. However, gender matching in pediatric patients did not confer any statistically significant advantage and could improve survival by at most 5.8% in males and 7.7% in females with 95% confidence. In the pediatric cohort, era of transplantation (particularly after 2000 compared with prior to 2000) and morbidities (such as congenital heart disease or need for mechanical support or dialysis) had greater impact on survival than gender effects.
We believe the strength of our study is the inclusion of 15 variables in our Cox model. It should be noted that all of the covariates used in our model had already been established in previous reports  and we included all of the known risk factors available in the database to avoid reporting-biased survival data. In the unadjusted analysis, many of these factors, including era of transplant and comorbidities, had a significant effect on transplant survival. However, even after adjusting for these factors, recipient gender still played a role in long-term survival. The role for gender in transplant survival is unclear and may include genetic as well as psychosocial factors that could not be assessed in this study and factors such as puberty, which may only evolve after transplant. Additional studies to better characterize the effects of gender are needed.
Many limitations are presented by using the OPTN database for this study. There is little detail available describing the pretransplant heart disease. Serum bilirubin was missing for 556 patients in the early transplant era and rather than exclude these subjects, we imputed values. We also included only those transplants performed after 1994 so as to minimize missing data for other variables. We were unable to evaluate the effect of pretransplant disease and prior surgery (for congenital heart disease). Surgical factors that could not be examined may have a gender-specific impact on early survival after heart transplant just as in other cardiothoracic surgery in children [19-21]. Posttransplant immunologic treatment is not recorded in this registry. We have no knowledge of selection of immune suppression based on gender in common practice, but such effects cannot be excluded from these data. The OPTN registry is not an event-driven database; data used for this study are based on yearly report. However, the OPTN data are based on information collected for all transplants performed in the United States reflecting organ allocation within the United States during the period 1994–2008.
In summary, we performed the first large population-based study designed to assess the gender effect on long-term mortality in pediatric heart transplantation. Gender matching did not significantly improve long-term survival among male or female recipients, so, at this time, gender should not play a role in decisions regarding organ allocation. We report that male heart transplant recipients have improved survival compared to female recipients and gender-matched males have the best survival and gender-mismatched females have the worst survival. The increased mortality of gender-mismatched females compared to their male counterparts may be explained in part by the well-published finding that females have increased risk of mortality after pediatric cardiac surgery and there may possibly be an additive effect of gender mismatch. While we report no difference in mortality attributable to isolated effects of gender match in male or female recipients, we do not exclude the possibility that the small effect of mismatch is real, even though its clinical significance on survival based on our data is unclear. We hypothesize that gender, gender match and additional factors incrementally contribute to gender-mismatched females having the worst survival outcome after heart transplantation. Both biologic and psychosocial factors related to gender may be important in long-term survival after heart transplant and deserve further study.
This work was supported in part by Health Resources and Services Administration contract 234-2005-370011C. 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 U.S. Government. This work was presented in part at the American Heart Association Scientific Sessions, November 2011.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.