Assessment of adverse childhood experiences in women: a study of reliability and predictive validity of self-reporting versus medical records abstraction in an American population.

OBJECTIVE

To study the reliability and validity of adverse childhood experiences (ACE) scores measured using a questionnaire versus abstraction of medical records, and to test whether the scores vary by history of bilateral oophorectomy, or by age and presence of anxiety or depressive symptoms at the time of questionnaire administration.

DESIGN

The study involved a reliability component and a predictive validity component.

SETTING

A population-based sample in Olmsted County, Minnesota, was derived from the Mayo Clinic Cohort Study of Oophorectomy and Aging-2.

PARTICIPANTS

We included 198 women who underwent premenopausal bilateral oophorectomy for a non-malignant indication between 1988 and 2007 and 174 referent women of the same age randomly sampled from the general population (total of 372 women). At a later time (median of 22.7 years later), the women were contacted and invited to self-administer the ACE questionnaire during an inperson visit. Independent of the visit, their medical records were abstracted for ACE by a physician.

INTERVENTIONS: Questionnaire and abstraction-based ACE scores.

MAIN OUTCOME MEASURES: Agreement between the two ACE scores (reliability; weighted kappa statistics) and comparison of incidence of multimorbidity in women with ACE scores ≥1 vs 0 (predictive validity; survival analyses). Data were analysed in March and April 2024.

RESULTS

The 372 women in the study had a median age of 65 years at the time of ACE questionnaire self-administration (IQR, 62-69). Questionnaire-based ACE scores showed moderate agreement with abstracted ACE scores (weighted kappa 0.44 (95% CI 0.34 to 0.54)). The cut-off score of ACE ≥1 showed an overall fair agreement between the two scores (kappa 0.33 (95% CI 0.24 to 0.43)). A comparison of the cumulative incidence of multimorbidity in women with an ACE score ≥1 versus women with ACE score of 0 yielded a HR of 1.13 (95% CI 1.00 to 1.27) for abstracted ACE and 1.13 (95% CI 1.01 to 1.27) for questionnaire-based ACE. The best predictive validity was for ACE measured by both methods combined (either or) with a HR of 1.29 (95% CI 1.13 to 1.46).

CONCLUSIONS AND RELEVANCE: Questionnaire-based ACE scores have moderate agreement with medical records abstracted ACE scores. However, both sets of ACE scores are predictive of the accumulation of multimorbidity at older age and should be considered complementary.