# 6  Description of statistical methods and data presentation

## 6.1 General Considerations

The analysis of this study was conducted with an exploratory intent.

All analysis variables were tabulated with summary statistics, and graphical representations were used as appropriate. Statistical tests were used to highlight interesting aspects of the data, such as differences between the groups in the analyzed endpoints. The tests were conducted with a two-sided alternative and the p-values will be reported. Statistical significance is declared for p-values below 5%. If appropriate, 95% confidence intervals for point estimates based on suitable distributions were additionally provided. No correction of the significance level for multiple comparisons was performed.

In general the analysis variables were compared by means of statistical tests between the EPS group and the two Control groups taken separately.

All data analyses were performed using the statistical package R 2.0.1 (The R Foundation for Statistical Computing), for Windows XP. Statistical support was provided by Corrado Bernasconi, M.D. Ph.D.

## 6.2 Analysis population, completeness of data, patients per country

Based on the definition of the ESP and the 2 Control groups (see 2.1 and 2.2) Eurotransplant provided data on a total of 3456 patients transplanted between 4 January 1999 and 4 January 2004. 7 patients from non-heart beating donors and 18 ESP patients for whom either the donor or the recipient was less than 65 years of age at time of transplant were excluded from all analyses, leaving a total number of 3431 patients for the analysis. Due to the definition of the two Control groups there is an overlap of 109 patients between Control 1 and 2(Table 5; see also 3.2.).

 N % Total number of patients (excluded patients: see below) 3431 100.00 ESP 1406 40.98 Control 1 (ETKAS, donor ≥ 65 y) 446 13.00 Control 2 (ETKAS, recipient ≥ 60 and 1687 49.17 Overlap between Control 1 and 2 109 3.18

Figure 7 shows the number of patients in each of the groups by year of transplantation. The number of patients transplanted in the ESP increased from 227 in 1999 to 382 in 2003.

“Updated patient group”: Out of 64 centres that were contacted and asked to provide missing information, 50 centres (79%) completed and returned data on a total of 2903 patients (85% of all patients, ESP Group n=1294; Control 1 n=355, and Control 2 n=1346 – considering also the overlaps)

The analysis for all evaluations concerning rejections was performed with the “updated patient population” only. In the data collection tool "no information” concerning rejection events and "no rejection" were not distinguished. Since detailed rejection data are available only for the "updated patient group", it was decided to restrict the analysis to this population.

The time of documented follow up was comparable in all groups indicating that there is no systematic error in the data capture (data not shown).

The majority of patients in all groups were transplanted in Germany (71.8%), followed by Austria (10.3%), Holland (9.1%) and Belgium (8%)(Figure 8). 83.8% of all ESP patients were transplanted in Germany.

## 6.3 Missing data/outliers

No imputation of missing data was done. This implies that some of the analyses could in effect be performed only on a subset of the entire analysis population.

In particular for the analysis of patient and graft survival, several patients had to be excluded due to the fact that an event or censoring time could not be determined. Reasons were that such times were missing, negative (death, graft loss or loss to follow up reported as having occurred before transplantation) or events occurred after database closure (the maximum accepted event date was 06 June 2005). A maximum of 15.5% of the patients was excluded from such analyses.

As far as the analysis of rejection is concerned, it should be mentioned that the outcome “no rejection” could not be distinguished form the absence of rejection information. For analysis purposes, both cases were considered as no rejection, but the analysis population was restricted to the 2877 patients for which rejection information was collected.

With regard to SCr values it was agreed to exclude outliers (value < 10 or > 1000 μmol/l). Waiting times < 4 weeks and > 15 years were excluded from the waiting time analysis. However, values incorrectly expressed in mg/dl were kept if the value with supposedly correct unit fell between the 100-300 μmol/l limits.

## 6.4 Definition of main analysis and derived variables

### 6.4.1 Calculation of Creatinine Clearance

Creatinine Clearance was calculated using the Cockroft-Gault Formula (Cockcroft and Gault, 1976). Baseline demographic data for weight, height and gender at time of transplantation were used.

### 6.4.2 Acute rejection

Acute rejection episodes reported with a normal biopsy as well as cases with no biopsy and cases that were not treated were not considered acute rejections in the analysis.

### 6.4.3 Waiting time

The time on the waiting list for transplantation is defined as the time between first dialysis and transplantation.

### 6.4.4 Immunosuppressive therapy

When entering the immunosuppression at 6 and 12 Months the following time windows applied

 • month 6 +/- 14 days • month 12 +/- 1 months

### 6.4.5 Graft function

Immediate graft function: No dialysis required within the first 7 days post transplant

Delayed graft function: One or more dialysis required within the first 7 days post transplant.

When entering serum creatinine values the following time windows applied:

 • week 2 +/- 2 days • month 1 +/- 3 days • month 3 und 6 +/- 14 days • year 1-5 +/- 1 month

“Yes” was recorded if an event occurred at any time during the observation period. The number of adverse events per patient was not recorded. Only selected adverse events were recorded: opportunistic infections, malignancies and cardiovascular events (see 5.2.3 for a definition)

### 6.4.7 Follow up on patient’s last visit

 Date last seen Date of patient’s most recent visit to the transplant centre and Clinical condition Clinical condition as judged by the treating physician (poor, good, excellent)

### 6.4.8 Hospital stay

Readmissions to hospital (number) number of readmissions to hospital (= same location as transplant centre) for any reason during the observation period. Completion of this field seemed to cause some difficulties and inconsistencies might impact on analysis.

In hospital days during readmissions cumulative number of days for all readmissions to hospital (= location as transplant centre) during the observation period. Completion of this field seems to cause some difficulties and inconsistencies might impact on analysis.

## 6.5 Demographic and Baseline Data

Demographic and background data are summarized for the ESP and the 2 Control groups.

The subdivision into categories of certain variables is only used in the presentation of summary statistics to but not in the statistical models except for the grouping of the reason for end-stage renal disease (ESRD), cause of graft loss and death, and the preservation solution.

## 6.6 Outline of statistical methods

Descriptive statistics were provided according to the nature of variables: number of observations, mean, standard deviation, minimum and maximum and quartiles for continuous variables, and frequency counts and percentages for categorical variables. Summary statistics were presented by group and in addition in selected cases broken down by other categorical variables, such as the transplantation year.

Continuous variables were compared by means of the Wilcoxon rank-sum test. Categorical variables, including proportions were analyzed with the chi-square test or in selected cases with Fisher’s exact test. Survival times were analyzed using the Kaplan-Meier method. A plot of the estimated probabilities of survival was created and log-rank tests of the difference between groups in survival probabilities were carried out.

Cox regression analysis was used to additionally evaluate the impact of baseline and treatment characteristics (including HLA matches and selected IS regimens) on:

• Patient survival
• Graft survival
• Time to AR

Cox regression models were also used to evaluate early graft function and the occurrence of AR as predictors for patient and graft survival.

 © Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme. DiML DTD Version 4.0 Zertifizierter Dokumentenserver der Humboldt-Universität zu Berlin HTML-Version erstellt am: 05.07.2006