Maximizing Retention in an Urban Prospective Cohort Study Elaina Murray, BS 1 ; Kate Beatty, MPH 1 ; Louise F. Flick, DrPH 1 ; Michael Elliott, PhD 1 ; Lisa V. John, PhD 2 ; Vetta Sanders Thompson, PhD 3 ; Allison King, MD, MPH 3 ; Laura Bernaix, PhD, RN 4 ; Candi LeDuc, RN 4 ; Elizabeth Lacy, RN 4 ; Kristi Helmkamp, RN 2 ; Amanda S. Harrod, MPH 1 ; Nikki Weinstein, MSW 2 1 Saint Louis University, 2 Battelle Memorial Institute, 3 Washington University in Saint Louis, and 4 Southern Illinois University-Edwardsville
14
Embed
Maximizing Retention in an Urban Prospective Cohort Study Elaina Murray, BS 1 ; Kate Beatty, MPH 1 ; Louise F. Flick, DrPH 1 ; Michael Elliott, PhD 1 ;
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Maximizing Retention in an Urban Prospective Cohort Study
Elaina Murray, BS1; Kate Beatty, MPH1; Louise F. Flick, DrPH1; Michael Elliott, PhD1; Lisa V. John, PhD2; Vetta Sanders Thompson, PhD3; Allison King, MD, MPH3; Laura Bernaix, PhD, RN4; Candi LeDuc, RN4; Elizabeth Lacy, RN4; Kristi Helmkamp, RN2; Amanda S. Harrod, MPH1; Nikki Weinstein, MSW2
1Saint Louis University, 2Battelle Memorial Institute, 3Washington University in Saint Louis, and 4Southern Illinois University-Edwardsville
Presenter Disclosures
The following personal financial relationships with commercial interests relevant to this presentation
existed during the past 12 months:
Elaina Murray
• No relationships to disclose
National Children’s Study (NCS)St. Louis City LocationBackground
• NCS will examine the effects of the environment on the growth, development, and health of children across the US, following them from before birth until age 21
• St. Louis City: Recruiting door-to-door in randomly selected segments (neighborhoods)• Segments, groups of 8 to 25 city blocks distributed around city• Eligible women were pregnant or trying to conceive
• Retaining participants in multi-year prospective cohort studies presents challenges, especially in urban settings• Early identification of participants at risk for attrition may enhance
retention
3
Research Questions
4
• Can we predict missed appointments from SES characteristics?
• Is the use of a subjective risk ranking better at predicting risk than SES characteristics alone?
Methods
5
Subjective Risk Ranking of participants• Data collectors assigned participants to
either low, medium, or high risk of loss to follow-up
• High risk characteristics: •Chaotic family life•Very busy•Disengaged from the Study•Financial or housing crisis
Methods
Data Set• SES variables pulled from larger NCS database (N =97)
• Race- White- Black
• Age- 18-24- 25-34- 35-44
• Education- Less than HS or HS grad- Some college or more
• Risk Ranking• Low/Medium• High
• # missed appointments• None• One or more
6
Statistical Analysis
7
• Used IBM SPSS Statistics 20• Preliminarily used Chi-square analysis to look for
associations• Binary logistic regression to develop propensity scores
to summarize SES variables (DV=risk ranking, IV = race, age, & education)
• Binary logistic regression to develop 4 models (DV = missed appointments):• Model 1: IV = Race, age, and education• Model 2: IV = Risk only• Model 3: IV = Race, age, education, and risk• Model 4: IV = Propensity score and risk
Statistical Analysis
• Propensity score•Typically used to match cases/controls•Also used to control for demographic variables by creating one score to account for all the SES variables-Used when not interested in the effects of each SES variable alone-Good for small sample sizes, increases power
8
Statistical Analysis
9
• Propensity score (cont.)•Ran logistic regression using risk ranking as outcome-Included all SES/demographic variables as predictors of risk ranking
•Saved predicted probabilities to use as propensity score for each participant
Results
Total (%) Missed Appointments χ2
None (%) One or more (%)
Race, N=67
White 50.7 40.8 33.3 χ2=4.307df=1*
Black 49.3 59.2 66.7
Age, N=67 χ2=.572df=2
18-24 34.5 32.2 40.0
25-34 50.0 52.5 44.0
35-44 15.5 15.3 16.0
Education, N=67 χ2 =.116df=1
<HS or HS Graduate 44.6 45.8 41.7
Some college or more
55.4 54.2 58.3
Risk Ranking, N=72 χ2=8.698df=1**
Low/Medium 75.9 84.5 52.4
High 24.1 15.5 47.6
Table 1. χ2 Analysis of Demographics & Risk with Number of Missed Appointments