My SciELO
Services on Demand
Journal
Article
Indicators
- Cited by SciELO
- Access statistics
Related links
- Cited by Google
- Similars in SciELO
- Similars in Google
Share
Gaceta Sanitaria
Print version ISSN 0213-9111
Abstract
ALVES, Elisabete et al. Medical record review to recover missing data in a Portuguese birth cohort: agreement with self-reported data collected by questionnaire and inter-rater variability. Gac Sanit [online]. 2011, vol.25, n.3, pp.211-219. ISSN 0213-9111.
Objectives: To assess the yield of medical record review to recover missing data originally collected by questionnaire, to analyze the agreement between these two data sources and to determine interobserver variability in clinical record review. Methods: We analyzed data from a birth cohort of 8,127 women who were consecutively recruited after giving birth from 2005-2006. Recruitment was conducted at all public maternity units of Porto, Portugal. We reviewed the medical records of 3,657 women with missing data in the baseline questionnaire and assessed agreement between these two sources by using information from participants with data from both sources. Interobserver variability was assessed by using 400 randomly selected clinical records. Results: Data on pregnancy complications and maternal anthropometric parameters were successfully recovered. Agreement between the questionnaire and records in family history data was fair, particularly for cardiovascular disease [k=0.27; 95% confidence interval (95%CI): 0.23-0.32]. The highest agreement was observed for personal history of diabetes (k=0.82; 95%CI 0.70-0.93), while agreement for hypertension was moderate (k=0.60; 95%CI 0.50-0.69). Discrepancies in prepregnancy body mass index classes were observed in 10.3% women. Data were highly consistent between the two reviewers, with the highest agreement found for gestational diabetes (k=1.00) and birth weight (99.5% concordance). Conclusion: Data from the medical records and questionnaire were concordant with regard to pregnancy and well-known risk factors. The low interobserver variability did not threaten the precision of our data.
Keywords : Epidemiologic methods; Medical records; Interobserver variability; Anthropometry; Cohort studies.