Registry reportThe registry of the international society for heart and lung transplantation: twentieth official adult heart transplant report—2003☆
Section snippets
Statistical methods
Survival rates were calculated using the Kaplan–Meier method1 and compared using the log-rank test. Multivariate analyses were performed using logistic regression analysis.2 Weights were used to account for incomplete follow-up. Patients with known status (e.g., alive or dead) at the timepoint of interest were assigned a weight of 1; patients with incomplete follow-up were assigned a weight proportional to the length of the interval for which their status was known. For example, in the analysis
Heart transplant demographics
The number of heart transplant procedures reported to the Registry each year continues to decrease. This appears to be due to decreased reporting from centers outside the USA. Figure 1 depicts the average annual heart transplant center volume from January 1, 1998 through June 30, 2002. Note that 25% of actively reporting heart transplant centers averaged <5 procedures per year, 45% averaged <10 per year, and 77% averaged <20 procedures per year. Only 6% of centers reporting averaged >40
Post-operative immunosuppression
Figure 2 demonstrates the trends in induction immunosuppression during the last 2.5 years. Note that peri-operative anti-lymphocyte antibody use continues to decrease; only 42% of patients transplanted in the first 6 months of 2002 received any form of prophylactic antibody therapy. FIGURE 3, FIGURE 4 portray the maintenance immunosuppressive protocols in place at the 1- and 5-year follow-up reports. To provide the most up-to-date analysis possible, only patient data obtained between January
Survival
Heart transplant actuarial survival curves and graft half-time calculations are shown in FIGURE 5, FIGURE 6. Note that the survival curve for the entire cohort (Figure 5) continues to demonstrate that, after the steep fall in survival during the first 6 months, survival decreases at a very linear rate, even beyond 15 years post-transplant. In addition, there does not appear to be a point beyond which the slope of the survival curve decreases and approaches that seen for the general
Risk factors for mortality
Ideally, identification of risk factors should lead to changes in medical practice, which may then alter the impact of these factors on outcome. It follows that to keep up with the evolution of immunosuppressive therapy and post-transplant management, the cohorts analyzed must be as current as possible to provide clinically useful data. Accordingly, we focus this year’s analysis on the most recent cohort transplanted from January 1999 through June 2001, and compare these patients with those
Causes of death
Adjudication of primary cause of death is particularly problematic in multicenter registries. This is due not only to actual difficulties in assigning priority to the multiple competing events, but also to the difficulties inherent in a registry reporting format utilizing non-uniform definitions. These limitations aside, review of the stated causes of death after heart transplantation can provide useful information. Included in this year’s Registry slide set (www.ishlt.org/registries/) is a
Post-transplant morbidity
This year’s analysis adds those transplants performed from December 2000 through June 2002 to the cohort examined in last year’s analysis. The cumulative incidence of hypertension, renal dysfunction, diabetes and CAV did not change significantly (www.ishlt.org/registries/). This year, we examined CAV differently than in prior years by determining risk factors for CAV occurring within 5 years post-transplant and for CAV occurring between 5 and 7 years (patients without CAV at 5 years). Table IV
Conclusions
For a registry or database to be clinically useful and directly impact patient care it must provide up-to-date and accurate information. At first glance, the discrepancies between the successive yearly reports suggest problems with data validity or analysis. However, after careful review it becomes clear that the changing risk factors and outcomes are more a reflection of changing medical management and patient selection—proof that we actually do “learn from our mistakes.” It is likely that, as
References (2)
- et al.
Nonparametric estimation from incomplete observations
J Am Stat Assoc
(1957) - et al.
Cited by (0)
- ☆
All figures and tables from this report and a more comprehensive set of Registry slides are available at www.ishlt.org/registries/.