Is patient drop out from a longitudinal study of lung function predictable and reversible?

Predictive factors for individuals dropping out from a longitudinal study of lung function have not been described, nor have methods for re-recruiting dropouts been developed. Our aims were to identify factors predictive of dropout and methods of re-recruiting dropouts from a longitudinal study of lung function. Details at enrollment into a longitudinal birth cohort study were compared between those who dropped out and those who were assessed 6 and 11 years later. Strategies for identifying individuals successfully contacted at 11 but not 6 years were described.

Of 253 infants enrolled prenatally, 123 (49%) were followed up at age 6 and 194 (77%) at age 11 years. At 6 years, independent risk factors for a child dropping out were: no other sibling enrolled (OR, 4.0; 95% Cl, 1.6, 10.1; P = 0.003); being first-born (OR, 2.8; 95% Cl, 1.5, 5.2; P = 0.001); mother completing education at age 15 years (OR, 2.5; 95% Cl, 1.3, 4.9; P = 0.007); and mother born overseas (OR, 2.3; 95% Cl; 2.8; P = 0.01). The following were predictive of a child dropping out at 11 years: mother born overseas (OR, 4.8; 95% Cl, 2.2, 10.2; P < 0.001); paternal smoker (OR, 3.1; 95% Cl, 1.5, 6.5; P = 0.003); no other sibling enrolled (OR, 4.1; 95% Cl, 1.1, 15.7; P = 0.04); and being first-born (OR, 2.1; 95% Cl, 1.0,4.3; P < 0.05). Of the 74 families successfully followed up at 11 but not 6 years, 35 (47%) were contacted by using data already held in study records, and 32 (43%) were contacted by databases available to the public.

In conclusion, factors present at enrollment and predictive of dropout from a longitudinal cohort study can be identified. Using a variety of methods, re-recruitment is possible for many individuals who previously dropped out. (C) 2003 Wiley-Liss, Inc.