Factors associated with attrition in a longitudinal study of health risk behaviours and conditions among adolescents in Ibadan, Nigeria.

BACKGROUND

Use of longitudinal design in research improves understanding of adolescent health. In this paper, we present factors associated with attrition in a pilot longitudinal study among adolescents in Ibadan, Nigeria.

METHODS

Adolescents were recruited from private and government-owned junior secondary schools using multi-stage sampling and interviewed over three data collection waves (2017, 2018 and 2019).

RESULTS

A total of 1067 (99.4%) of the 1073 adolescents recruited were willing to participate and were interviewed in wave one. Mean age at baseline was 11.9 ± 1.2 years and 34.9% owned a personal mobile phone. Of the 1067 adolescents, 192 (18.0%) were not willing to be followed up while 875 (82.0%) were willing to be followed up by home visit (70.2%), phone call (21.3%), text message (14.3%) or online chat-based message (4.8%). Overall attrition rate (proportion of adolescents lost to follow-up during waves two and/or three compared with the baseline sample) was 66.5% with 396 (37.1%) and 315 (46.9%) respondents lost to follow-up during waves two and three respectively. Common reasons for attrition were use of pseudonyms instead of real names, which many adolescents could not remember during subsequent data collection waves, relocation to a different school or neighborhood, school drop-out and closure of two schools. Adolescents in private versus government-owned schools (AOR = 3.35; CI =  2.39 - 4.69), those who did not have personal mobile phones (AOR = 1.43, CI =  1.03 - 1.98) and those engaging in remunerated work (AOR =  2.04, CI =  1.19 - 3.49) were more likely to be lost to follow-up.

CONCLUSIONS

Attrition was high despite high willingness to participate in the study. Whereas technology has made follow-up of study participants in high-income countries easier, multiple, and cost intensive methods to minimize attrition may be required in low-resource settings.