The Life History of a Cohort Study: Attrition and Its Effects on Analysis of Health Outcomes in the Cebu Longitudinal Health and Nutrition Survey

Adair, Linda S.; Perez, Tita Lorna L.; & Borja, Judith B. (Forthcoming). The Life History of a Cohort Study: Attrition and Its Effects on Analysis of Health Outcomes in the Cebu Longitudinal Health and Nutrition Survey. Philippine Population Review. NIHMSID: NIHMS994660

Adair, Linda S.; Perez, Tita Lorna L.; & Borja, Judith B. (Forthcoming). The Life History of a Cohort Study: Attrition and Its Effects on Analysis of Health Outcomes in the Cebu Longitudinal Health and Nutrition Survey. Philippine Population Review. NIHMSID: NIHMS994660

Octet Stream icon 11652.ris — Octet Stream, 1 kB (1955 bytes)

We document patterns of participation among adult women in the Cebu Longitudinal Health and Nutrition Survey (CLHNS), an ongoing community-based cohort study started in Metro Cebu, Philippines in 1983. The CLHNS recruited pregnant women from 33 randomly selected barangays in urban and rural areas, followed them intensely for 2 years, and subsequently in follow-up surveys in 1991, 1994, 1998, 2002, 2005, 2007, 2012, and 2015. For each survey, we categorized women according to causes of attrition as not located or left the Metro Cebu area, died, refused, or ineligible. We assessed how baseline maternal and household sociodemographic, economic, and environmental characteristics were related to each category of attrition vs. participation using multinomial logistic regression. We then assessed whether attrition influenced estimates of the association of maternal characteristics and behaviors with two health outcomes: systolic blood pressure (SBP), and depressive symptoms. We compared uncorrected models with those weighted by the inverse probability of participation. Decreased likelihood of failure to locate or migration (the most frequent reason for attrition) was associated with baseline rural residence, higher value of household assets, higher maternal schooling, and living with a spouse. Death rates increased across the years of the study and were significantly predicted only by age and urban residence. Refusal rates were low and were related to urban residence and higher SES. Despite substantial attrition, we found little evidence of selection bias in the SBP and depression score models.





JOUR



Adair, Linda S.
Perez, Tita Lorna L.
Borja, Judith B.



Forthcoming


Philippine Population Review












NIHMS994660

11652

Wink Plone Theme by Quintagroup © 2013.

Personal tools
This is themeComment for Wink theme