OBJECTIVES : To use Excel spreadsheet software to fit parametric survival distributions onto non-parametric Kaplan-Meier curve. The difference between Individual Patient Data (IPD) and survival data reconstructed in Excel and SAS will be reviewed and the impact of those differences on economic evaluation results will be assessed.

METHODS

: Three sets of patient data on Overall Survival (OS) were compared using different elicitation methods: “Original” IPD, “Reconstructed SAS” and “Reconstructed Excel”. Best-fit distribution was selected using visual observation, supported by graphical tests of linear plots of predicted probabilities, goodness-of-fit coefficients (R-squared) and the sum of squared errors of prediction (SSE). Outcomes in Incremental Cost-Effectiveness Ratio (ICER), Incremental Net Benefit (INB), incremental cost and life-years gained over short- and life-time horizons were compared for different data sets.

RESULTS

: Log-Normal, Log-Logistic and Weibull distributions applied in Excel showed best fit when visual test, R-squared and SSE were applied. Weibull and Exponential distributions showed significant differences with IPD data. Data on short-term (5 years) visual comparison and graphical test of parametric distributions generated by different data re-creation methods showed a close resemblance with data reconstructed from SAS to the Original IPD. Results of ICER and INB were dependent on the time horizon and selected parametric distribution from the model.

CONCLUSIONS

: Different approaches used in fitting parametric survival distribution yielded predicted probabilities that substantially differed to those using original IPD. Guidelines to economic evaluations would benefit from the inclusion of parametric survival analysis techniques with guidance in selecting the best fitting parametric survival distribution.