Lost-to-follow-up study in systemic lupus erythematosus (SLE)

To determine the extent of and reasons for lost-to-follow-up, as well as its impact on outcome studies in a cohort of lupus patients.

As of September 1991, 247 patients, in a cohort of 621 patients with SLE, being followed in a long-term prognosis study, had not been seen since 1 March, 1990 and were considered lost-to-follow-up. Patients were contacted and encouraged to return for an evaluation or to answer a questionnaire by telephone. Descriptive statistics were used to compare the lost-to-follow-up and non-lost-to-follow-up patients and the survival experience during the lost-to-follow-up period was compared with that when patients were not considered lost-to-follow-up. Estimated survival curves with and without the information gained through contacts with lost-to-follow-up patients were compared.

Of the 247 patients, 29 have died, 66 returned for a full assessment, 84 completed a questionnaire and 68 (11%) were true lost-to-follow-up. The lost-to-follow-up patient group had 10% more Caucasians and 6% more males than the patients under regular follow-up. The estimated survival curves of the entire cohort with and without the new lost-to-follow-up data, calculated as of July 1992, were very similar. There was no evidence of a differential mortality rate during the period in which patients were lost-to-follow-up. Some suggestive evidence that the relative mortality rate comparing the rate during a period of lost-to-follow-up and during a period of active follow-up may depend on disease duration at the time of lost-to-follow-up was found.

While it would be prudent to limit lost-to-follow-up as much as possible, especially for outcomes such as mortality which do not necessarily require a clinic visit, it does not appear that significant bias will be present in prospective studies based on our single clinic database. Since the retrieved 179 lost-to-follow-up patients did not affect survival studies it is likely that the 68 true lost-to-follow-up patients will also not have an impact on prognostic studies. Lupus (2000) 9, 363-367