Transcript Open poster - CTN Dissemination Library

Predictors of study retention in addiction treatment trials
KORTE JE1, MAGRUDER KM1,2, KILLEEN TK1, SONNE SC1, SAMPSON RR1 and BRADY KT1,2
1. Medical University of South Carolina, Charleston, South Carolina, USA
2. Ralph H. Johnson VA Medical Center, Charleston, South Carolina, USA
INTRODUCTION
Differential attrition in research studies is a threat to study validity. Attrition during the
treatment phase of a study should be conceptualized and studied separately from
attrition during follow-up. Unbiased assessment of intervention effectiveness requires
good follow-up rates, and roughly equivalent follow-up rates between groups. By
focusing on retention during follow-up, we can improve our ability to maximize internal
and external validity of study results.
The current study is focused on attrition during follow-up, which is primarily relevant for
valid ascertainment of research outcomes (as opposed to retention during treatment,
which is relevant for both research outcomes and treatment planning).
We examined predictors of follow-up completion in the first 15 protocols to have locked
datasets in the NIDA Clinical Trials Network (CTN). We focused on gender and ethnicity
as the main predictors of interest.
RESULTS
Table 1. Protocol characteristics
The objective of the current study is to improve our understanding of why certain types of
clients are lost to follow-up, in order to design better study procedures to maximize
retention for all participants.
METHODS
 Design
• Secondary data analysis
• Pooled analysis of randomized controlled trials for drug addiction treatment
• First fifteen protocols in NIDA Clinical Trials Network (CTN) with datasets locked and
appropriate for analyses of client retention during follow-up
• CTN 01, 02, 03, 04, 05, 06, 07, 09, 10, 11, 13, 15, 18, 19, 21
 Retention definition
• Study-specific definition of retention
• Based on whether each client completed the final study assessment at which the
major outcome variable was collected (as defined by the respective study protocols)
 Study Population
• Each protocol included individuals meeting specific criteria for drug use (criteria varied
by protocol)
• Participants were included in the retention analyses if they met study criteria and were
randomized.
 Follow-up
• Scheduled study visits varied by protocol (see Table 1)
 Analysis
• Preliminary analyses: chi square for bivariate associations
• Logistic regression modeling to predict the odds of retention during follow-up
• Main predictors of interest: gender and ethnicity
• Stratified model by whether clients had taken prescribed methadone in 30 days
before study baseline
• Other covariates: age, length of study follow-up
NIDA Blending Conference, Albuquerque NM • April 22-23, 2010
Some of these differences are attributable to structural protocol characteristics (e.g. length of follow-up), or attributable to differences in the proportion of clients in
methadone maintenance programs (expected to increase retention).
CTN
protocol
number
Type of
intervention
Visit schedule
1
Medication trial
for inpatient
opiate detox
13 days detox;
13-day end of
then follow-up at treatment
1, 3, and 6
months
In multivariate models of retention, we found that use of prescribed methadone during the last 30 days, as expected, was a highly significant predictor of retention
during follow-up, with clients on methadone less than half as likely to be lost to follow-up (p<0.001).
2
Medication trial
for outpatient
opiate detox
13 days detox;
13-day end of
then follow-up at treatment
1, 3, and 6
months
3
Suboxone taper: 4 weeks of
3-month followcomparison of
stabilization;
up
two schedules
then 7 or 28
days of taper;
then follow-up at
1 and 3 months
post-taper
Stratifying by methadone status in logistic regression models (Table 4), we found that among methadone clients, no retention differences were seen by age, race,
gender, or length of study follow-up. In contrast, among clients not using prescribed methadone, we found highly significant differences. In comparison to African
Americans, white clients were 50% more likely to be lost to follow-up, whereas Hispanic/other clients were 30% less likely. Longer length of study follow-up was
associated with better retention; this finding bears further investigation and may be due to other study/client characteristics. Finally, in both methadone and nonmethadone clients, we found that women were 15% less likely to be lost to follow-up; this finding was significant in overall models and among non-methadone clients,
but not among models limited to methadone clients.
4
Motivational
enhancement
28 day
treatment; then
follow-up at 1
and 3 months
3-month followup
5
Motivational
interviewing
Single 2-hour
session; then
follow-up at 1
and 3 months
3-month followup
6
Motivational
incentives (drug
free clinics)
12-week study,
with follow-up
visits at 1, 3,
and 6 months
after enrollment
6-month followup
Motivational
incentives
(methadone
clinics)
12-week study,
with follow-up
visits at 1, 3,
and 6 months
after enrollment
6-month followup
Smoking
cessation
treatment
9-week study,
with follow-up
visits at 9, 13,
and 26 weeks
after target
smoking quit
date
13-week followup
12-week study,
with follow-up
visits at 24, 36,
and 52 weeks
after enrollment
12-week end of
treatment
Telephone call
12-week study,
support to
with follow-up
increase postvisit at week 13
discharge
engagement
and decrease
drug/alcohol use
13-week followup
Motivational
enhancement in
pregnant
substance users
4-week study,
with follow-up
visits at 4 and
12 weeks after
study
12-week followup
“Seeking Safety”
treatment for
trauma for
women with
PTSD
6-week study,
with follow-up
visits at 1 week,
3, 6, and 12
months after
treatment
12-month followup
HIV/STD safer
sex for men
1 or 5 session
intervention,
with follow-up
visits
immediately,
and 3 and 6
months post
treatment
3-month followup
1 or 5 session
intervention,
with follow-up
visits
immediately,
and 3 and 6
months post
treatment
3-month followup
3 session
intervention,
with follow-up
visits at 1 and 3
months
3-month followup
OBJECTIVE
7
9
10
11
13
15
18
19
21
Medication trial
for opioiddependent
adolescents
HIV/STD safer
sex for women
Motivational
enhancement
for Spanishspeaking clients
Retention
definition
In preliminary analyses, we observed large differences in retention between different CTN protocols (Table 2).
We observed a significant interaction between ethnicity and gender in relation to study retention: this interaction was largely driven by Hispanic/other women not in
methadone treatment, who were approximately 40% less likely to be lost to follow-up than their African American counterparts. In contrast, white women (not in
methadone treatment) were approximately 20% more likely to be lost to follow-up than their African American counterparts.
Table 2. Client characteristics by protocol
1
2
3
4
5
Table 3. Proportion of clients completing study endpoint visit
6
7
9
10
11
13
15
18
19
21
%
Ethnicity
Race/ethnicity
Caucasian 59%
41%
71%
42%
84%
38%
28%
36%
70%
64%
37%
49%
48%
58%
0%
African American 21%
38%
11%
42%
12%
45%
54%
24%
0%
35%
33%
35%
25%
24%
0%
Hispanic 20%
22%
18%
16%
4%
16%
18%
40%
30%
1%
30%
17%
27%
18%
100%
Chi-square p-value
<0.0001
Black 70%
White 62%
Hispanic/other 74%
Gender
0.11
Male 67%
Gender
Female 59%
28%
33%
71%
58%
55%
43%
48%
42%
37%
100% 100% 0%
100% 11%
36.0
38.3
35.9
35.3
33.1
35.2
41.5
42.1
19.7
36.2
26.2
38.8
Age (mean)
39.2
39.5
32.7
Female 69%
Age
<0.0001
15-19 65%
20-29 64%
Retained to final followup visit
83%
80%
51%
68%
75%
68%
83%
80%
Using prescribed
methadone in last 30
days
1%
1%
1%
4%
0.4%
1%
100% 72%
60%
72%
74%
77%
68%
65%
78%
30-39 65%
3%
3%
12%
13%
51%
59%
2%
50-59 74%
40-49 73%
60+ 69%
Prescription methadone in last
30 days
Table 4. Odds ratios from multivariate logistic regression models,
modeling the odds of failing to complete the study endpoint visit
<0.0001
No 64%
Yes 80%
Model 1: Clients on methadone (last 30 days)
Age (1-year interval)
OR (95% CI)
p-value
0.99 (0.97, 1.004)
0.13
Ethnicity
SUMMARY AND CONCLUSIONS
Black -referenceWhite 0.95 (0.59, 1.5)
Hispanic/other 0.82 (0.49, 1.4)
0.77
Sex (female)
0.83 (0.58, 1.2)
0.31
Length of follow-up (days)
1.001 (0.998, 1.004)
0.37
0.37
Model 2: Clients not on methadone
Age (1-year interval)
OR (95% CI)
p-value
0.994 (0.986, 1.002)
0.12
Ethnicity
Black -referenceWhite 1.5 (1.2, 1.9)
Hispanic/other 0.70 (0.55, 0.89)
<0.0001
<0.0001
Sex (female)
0.84 (0.71, 1.004
0.06
Length of follow-up (days)
0.999 (0.998, 1.0)
0.03
 Clients on methadone were seen to have better retention than other clients;
further, among clients on methadone, other potential predictors were not
strongly related to retention in study follow-up.
 Among clients not on methadone, Caucasians were most likely to be lost to
follow-up, and Hispanic/other clients were least likely. Women were 15%
less likely to be lost to follow-up in comparison to men. In particular,
Hispanic/other women had better retention than any other group.
 By improving our ability to identify clients at increased risk of loss to follow-up
during research studies, we can devise strategies to maximize retention for
all participants. Future analyses will take into consideration other client and
protocol characteristics to further refine our assessment of factors related to
retention during follow-up.