Deborah B Nelson1, Mary D Sammel, Ellen W Freeman, Li Liu, Elizabeth Langan, Clarisa R Gracia. 1. Department of Biostatistics and Epidemiology, Center for Clinical Epidemiology and Biostatistics, Hospital of the University of Pennsylvania, University of Pennsylvania School of Medicine, Philadelphia, USA. dnelson@cceb.med.upenn.edu
Abstract
OBJECTIVE: Identifying clinical markers and characteristics of the transition to menopause is an important woman's health issue, and recent long-term, prospective, cohort studies are just beginning to offer insight into methods of predicting the transition to menopause. One of the major challenges of conducting prospective cohort studies is the problem of attrition-both the retention of study participants and the influence of dropouts on the final study results. We conducted this systematic analysis to: 1) identify baseline predictors of subsequent long-term participation, and 2) determine the demographic, psychosocial, and hormonal differences between participants and dropouts among a group of premenopausal women enrolled in a longitudinal study of ovarian aging. DESIGN: Using data from the Penn Ovarian Aging study, premenopausal women aged 35 to 47 enrolled in the study were classified as either Active (full participants), Skipped, or Dropped participants based on their visit pattern during a 4-year study interval. RESULTS: Most of the demographic and psychosocial variables did not significantly differ between the Active, Skipped, or Dropped groups. There was no racial difference in study participation. The Dropouts were more likely to have a high school education and were less likely to report menopause symptoms compared with the Actives (P < 0.01). The Skipped group reported more anxiety (P < 0.05), and members were more likely to have less than a high school education compared with the Actives (P < 0.03). Hormone levels (estradiol, follicle-stimulating hormone, luteinizing hormone, dehydroepiandrosterone, testosterone) at enrollment were within the premenopausal range and did not significantly differ among the three study groups. These findings remained after adjustment for covariates and hormone levels in multivariate analyses. CONCLUSIONS: Education, anxiety levels and menopause symptoms at baseline differed marginally between the women participating fully and those who dropped out or skipped multiple assessments. These findings are important and indicate that long-term study participation rates do not differ substantially by racial group or any of the other demographic or hormonal characteristics examined.
OBJECTIVE: Identifying clinical markers and characteristics of the transition to menopause is an important woman's health issue, and recent long-term, prospective, cohort studies are just beginning to offer insight into methods of predicting the transition to menopause. One of the major challenges of conducting prospective cohort studies is the problem of attrition-both the retention of study participants and the influence of dropouts on the final study results. We conducted this systematic analysis to: 1) identify baseline predictors of subsequent long-term participation, and 2) determine the demographic, psychosocial, and hormonal differences between participants and dropouts among a group of premenopausal women enrolled in a longitudinal study of ovarian aging. DESIGN: Using data from the Penn Ovarian Aging study, premenopausal women aged 35 to 47 enrolled in the study were classified as either Active (full participants), Skipped, or Dropped participants based on their visit pattern during a 4-year study interval. RESULTS: Most of the demographic and psychosocial variables did not significantly differ between the Active, Skipped, or Dropped groups. There was no racial difference in study participation. The Dropouts were more likely to have a high school education and were less likely to report menopause symptoms compared with the Actives (P < 0.01). The Skipped group reported more anxiety (P < 0.05), and members were more likely to have less than a high school education compared with the Actives (P < 0.03). Hormone levels (estradiol, follicle-stimulating hormone, luteinizing hormone, dehydroepiandrosterone, testosterone) at enrollment were within the premenopausal range and did not significantly differ among the three study groups. These findings remained after adjustment for covariates and hormone levels in multivariate analyses. CONCLUSIONS: Education, anxiety levels and menopause symptoms at baseline differed marginally between the women participating fully and those who dropped out or skipped multiple assessments. These findings are important and indicate that long-term study participation rates do not differ substantially by racial group or any of the other demographic or hormonal characteristics examined.