Life Tables

Life tables are a way to estimate risks when the occurrence of disease is a function of time and we would otherwise estimate incidence. It does this by stratifying the data by the time individuals are followed looking for the disease to occur. Within each study-relative time period, the number of events (e.g., cases of disease) and the number of withdrawals (persons for whom follow-up ended) are recorded. From this information, the probability of making it through the time period without having the event is calculated. There are two ways this is done. The most popular way is to assume that withdrawals occur at the end of the time period. This is the Kaplan–Meier method. The other way is to assume that withdrawals occur uniformly throughout the time period and thus, on the average, in the middle of the period. This is the Cutler-Ederer or actuarial method. Risks for a time period are calculated by taking the complement of the cumulative probability of avoiding the event. Life tables are compared by estimating risk ratios or risk differences. The null hypotheses that the risk ratio is equal to one and the risk difference is equal to zero in the population are tested using the log-rank test. The log-rank test compares observed and expected valuesExpected values for 2 × 2 tables created for each time period. It can be performed in RR. Life tables can be summarized and compared graphically using survival plots. A survival plotPlot has time on the abscissa and the cumulative probability of avoiding the event on the ordinate. Points are connected with stepwise lines.