Objective: To assess the agreement between postal and interviewer-administered versions of the Frenchay Activities Index (FAI) and to assess the criterion validity of the postal version, using interviewer administration as a gold standard.
Design: Comparison of responses to FAI administered by post and then by interview (median delay 10 days).
Subjects: Forty-eight Oxfordshire residents admitted to hospital with acute stroke between 1 August 1994 and 31 January 1995 and discharged alive within six months of their stroke.
Results: The limits of agreement of the total FAI score are from -5.4 to 7.2. The kappa statistic for each of the 15 individual items that make up the FAI ranged from 0.35 to 1. For nine items, agreement was moderate or fair, and for six items, agreement was good or very good. The mean difference between the overall scores was 0.9 (95% confidence interval: -0.1 to 1.9). The correlation between the overall scores was 0.94 (Spearman's rank correlation coefficient).
Conclusion: The postal version of the FAI is a satisfactory alternative to direct administration, but poor agreement in scores for individual patients emphasizes that the two approaches should not be used sequentially to monitor individual patients.