The American Journal on Addictions, 26: 572–580, 2017

Published 2017. This article is a U.S. Government work and is in the public domain in the USA
ISSN: 1055-0496 print / 1521-0391 online
DOI: 10.1111/ajad.12553

Three-Year Retention in Buprenorphine Treatment for

Opioid Use Disorder Nationally in the Veterans Health
Administration

Ajay Manhapra, MD ,1,2,3,4 Ismene Petrakis, MD,1,2 Robert Rosenheck, MD1,2

1
VA New England Mental Illness Research and Education Center, West Haven, Connecticut
2
Department of Psychiatry, Yale School of Medicine, New Haven, Connecticut
3
Department of Internal Medicine, Yale Medical School, New Haven, Connecticut
4
VA Hampton Medical Center, Hampton, Virginia

Background: Buprenorphine has become the major treatment for INTRODUCTION

opioid use disorder (OUD) but data on long treatment term retention
and its correlates are sparse.
Methods: All veterans with OUD treated in Veterans Health
Administration (VHA) facilities nationally in fiscal year (FY) 2012,
and who began treatment with buprenorphine as indicated by a first
prescription after the first 60 days of the year were identified with the
date of and their last prescription from FY 2012–2015. Veterans
were classified into four groups based on time from first to last
prescription: (0–30 days, 31–365 days; 1–3 years; and more than 3
years). These groups were compared on socio-demographic,
diagnoses and service, and psychotropic drug use. Kaplan-Meier
curves and Cox proportional hazards models were used to identify
variables independently associated with retention in buprenorphine
treatment.
Results: Veterans newly started on buprenorphine (n ¼ 3,151) were
retained in treatment for a mean duration of 1.68 years (standard
deviation [SD] 1.23), with 61.60% (n ¼ 1,941) retained for more
than a year and 31.83% (n ¼ 1,003) for more than 3 years. Cox
proportion hazards model showed that only black race (Hazards ratio
[HR] 1.26; standard error [SE] .06; p.0003), the Charlson index (HR
1.03; SE .01; p.0132) and emergency room visits during FY 2012
(HR 1.03; SE .01; p < .0001) were the only available variables
independently associated higher odds of buprenorphine
discontinuation.
Conclusions: Buprenorphine retention was substantial among
veterans treated in VHA, but few individual characteristics correlated
with retention.
Scientific Significance: Future research focused on identifying
further correlates of treatment retention is required to help devise
interventions to improve treatment continuation. (Am J Addict
2017;26:572–580)
Received October 6, 2016; revised February 7, 2017; accepted
March 27, 2017.
Address correspondence to Ajay Manhapra, Hampton VA
Medical Center, PRIME 5100 Emancipation Dr.Hampton, VA
23667.E-mail: ajay.manhapra@yale.ed
In 2012, over 2 million US adults were estimated to have
opioid use disorder (OUD), imposing a significant burden on
themselves, their families, and society.1 Substance use
disorders (SUD) including OUD are chronic relapsing
illnesses, not unlike diabetes or hypertension, that often
require long term treatment and have no reliable permanent
cure.2 However, patients who sustain participation in
treatment have more favorable outcomes with decreased
mortality, morbidity, and adverse psychosocial
consequences.3–5 Opiate Agonist Treatment (OAT) with
either a methadone or buprenorphine maintenance regimen
is considered to be the standard of care for eligible patients
with OUD.3,6 However, only a small proportion of eligible
patients with OUD receive such treatment.7 Dramatic
expansion of access to buprenorphine over the past decade
has increased the overall access to OAT while lowering the
restrictive requirement for treatment participation, resulting in
relative stagnation of older methadone maintenance.1,5,8–12
Despite the benefits with OAT for OUD, patient retention in
treatment can be challenging.2 This could especially be the
case with the newer OAT option of buprenorphine that impose
less stringent demands on patients and physicians, and
associated with higher dropout rates compared to the older
option of methadone as per reports from early clinical trials
and observational studies.5,8,13–15 But, real life large observa-
tional studies have also reported that despite the high early
drop out rates with buprenorphine, treatment re-engagement,
and retention in OAT is substantial on longer follow up,8 and
treatment attrition rate actually decreased during a decade,
where buprenorphine increasingly replaced methadone as the
preferred OAT choice.9 Buprenorphine has thus evolved as the
favored choice for OAT among both patients and physicians,
but there is sparse data regarding the long term retention in
buprenorphine treatment for OUD.16 A recent literature

572
review showed that few studies have reported buprenorphine
retention beyond 6 months, and the scant studies with longer
term information include data for only up to 1 year.17 Two
recent insurance claims-based studies focusing on healthcare
costs presented data on treatment retention, but lacked clinical
information on patient characteristics that are associated with
retention.14,18
The Veterans Health Administration (VHA) is the largest
integrated healthcare provider in United States serving over
five million US veterans annually. Veterans appear to have
similar rates of SUD as the general population, but are more
likely to receive treatment, although the overall treatment rates
remain low.19 Although overall utilization of buprenorphine
has increased substantially in VHA, since its introduction in
2003, paralleling non-VA national trends,1,20–22 the long-term
retention in buprenorphine treatment among veterans treated
at VHA facilities has not been studied. In this observational
study of veterans with OUD initiated on Buprenorphine OAT
during fiscal year (FY) 2012, we examined the duration of
buprenorphine service use up to and beyond 3 years along with
socio-demographic, diagnostic and health service, and
psychotropic medication use correlates of prolonged engage-
ment. These data will increase our understanding of correlates
of long-term retention in buprenorphine maintenance therapy,
potentially facilitating the development of interventions to
increase such retention.
METHODS
Sample and Data Source
National VHA administrative records from FY 2012
(October 1, 2011–September 30, 2012) were used to identify
all veterans with the diagnosis of OUD (ICD-9-CM codes
304.0x, 305.5x, and 304.7x—either opioid abuse or opioid
dependence). We then identified patients, who filled at least
one prescription for buprenorphine in VHA nationally during
the FY 2012. Patients receiving buprenorphine treatment were
identified by prescription fill data that included all patients,
who received at least one prescription for buprenorphine or
buprenorphine/naloxone tablets, and excluded patients receiv-
ing buprenorphine transdermal patch as it is designated for
pain management rather than OAT.
Measures
Duration of Buprenorphine Treatment
Using the VHA pharmacy benefits data, we identified the
first and last day each patient received buprenorphine
prescription, from October 1, 2011 (first day of FY 2012) to
September 30, 2015 (last day of FY 2015), a maximum of
4 years. We then calculated, the duration of treatment from the
first day of buprenorphine fill in FY 2012 and the last date
buprenorphine was filled from FY 2012–2015. We excluded
veterans who received any buprenorphine in the first 60 days of
FY 12 to identify a sample, who newly started a new episode of
treatment in FY 2012. We then divided them into four
Manhapra, Petrakis, and Rosenheck
groups based on the duration of retention in treatment: 0–30
days, 31–365 days; 1–3 years; and more that 3 years. We then
compared the four retention groups on sociodemographic,
diagnostic, and service and psychotropic medication use
characteristics from FY 2012 administrative data.
Diagnostic Characteristics and Service, and
Psychotropic Medication Use
Sociodemographic characteristics included age, gender,
income, receipt of Veterans Affairs (VA) disability compensa-
tion, or pension payments, service in the Iraq or Afghanistan
theater of war (OIF/OEF), and a designation of urban versus
rural residence based on zip codes and Rural–Urban Commut-
ing Area (RUCA) codes (depts.washington.edu/uwruca).
Recent homelessness was identified by participation in a
VHA specialized homeless service program or a V-60 code
indicating housing problems. Information on ethnicity was
obtained and included black, white, Hispanic, and mixed races.
Clinical diagnostic data included psychiatric and medical
diagnoses based on International Classification of Diseases,
9th edition [ICD-9] codes that were assigned to each patient at
least once during the study year. The Charlson Index,23 an
aggregate measure of medical co-morbidity diagnoses, has
been shown to predict the 10-year mortality for patients based
on the presence of a broad range of co-morbid medical
conditions and was used to assess medical co-morbidity (see
reference24 for details). Individual diagnoses that compose the
Charlson index were also examined. We included eight pain
diagnoses, based on ICD-9 diagnostic codes: any pain
diagnosis, herpetic pain (053.12 or 729.2), fibromyalgia
pain (729.1), musculoskeletal pain (338.xx, 719.4, 780.96),
skeletal-spasm pain (728.85, 781.0), pain from diabetes
(250.6, 357.2, 337.1), migraine and headache (346.x,
784.0), and pain and neuropathy (250.6, 357.2, 337.1, 338.
x, 719.4, 780.96, 729.1, 728.85, 781.0, 053.12, 729.2, 352.1,
350.1).
Data on co-morbid psychiatric diagnoses included all ICD-9
codes 290.00 through 319.99 (coded into 11 classes; available
on request or see reference24). VHA inpatient and outpatient
service utilization was documented including medical and
surgical outpatient visits, and emergency room visits.
Outpatient mental health specialty care, substance abuse
clinic visits, medical-surgical visits, and emergency depart-
ment visits were identified by standard VHA clinic stop codes
as were patients who experienced a psychiatric hospitalization
during the year.
Pharmacy benefit records documenting all VHA prescrip-
tions filled were used to identify the total number of
prescriptions filled by these patients in five psychiatric
medication classes: antidepressants, antipsychotics, sedative/
hypnotics/anxiolytics, mood stabilizers (anti-epileptics), and
lithium (the individual medications in each class available on
request or see reference24). The total number of psychotropic
prescriptions filled by each veteran during the year was
determined by summing the numbers of prescriptions in each
of these classes as well as the total number of classes from
September 2017 573
which a prescription was filled. Medications in those classes
that were not on the VHA formulary were not included in the
analysis.
Analysis
We compared the baseline characteristics between 0 and
30 day retention group, 31–365 day retention group, 1–3 year
retention group, and more that 3 years retention group using
bivariate analyses of sociodemographic and diagnostic
characteristics, health service use, and psychotropic medica-
tion fills. Veterans who remained on medication for 30 days or
less were the reference group. Given the large sample size and
the fact that it represents the entire population of interest,
effect sizes were used rather than p-values to identify
meaningful differences (ie, risk ratios for dichotomous
measures and Cohen’s d for continuous measures). Risk
ratios greater than 1.5 or less than .67 were considered to
represent substantial differences on dichotomous variables.
The difference between means divided by the pooled standard
deviation was used to calculate Cohen’s d for continuous
variables. Values greater than .20 were considered to represent
more than small differences.25 Variables were included only if
the base rate was more than 5%.
Kaplan–Meyer curves and Cox proportional hazards
models were then used to identify variables independently
associated with retention in buprenorphine treatment. Varia-
bles found to demonstrate substantial differences between
groups on bi-variate analyses were included in these
multivariate analyses.
RESULTS
Altogether 8,164 veterans received a diagnosis of OUD
during FY 2012 and filled at least one buprenorphine
prescription during FY 2012. After excluding 5,013 veterans,

TABLE 1. Comparison of baseline characteristics between buprenorphine retention groups

Effect size comparison

with 0–30 days
category (Risk
Buprenorphine retention categories Ratios/Cohens d )
31–365 1–3 >3
0–30 days 31–365 days 1–3 years >3 years days years years
Total N ¼ 3151 N ¼ 435 N ¼ 775 N ¼ 938 N ¼ 1003
Percent of total 13.81% 24.60% 29.77% 31.83%
Mean age (SD) 44.652 (14.06) 42.770 (14.0) 41.988 (13.7) 42.695 (13.2) .14 .20 .14
Males 93.79 94.58 94.70 93.32 1.01 1.01 .99
Urban area residents 76.82 76.39 078.10 75.54 .99 1.02 .98
Large rural area 10.77 9.95 9.48 11.46 .92 .88 1.06
residents
Small rural area 7.73 8.62 6.21 6.76 1.12 .80 .87
residents
Isolated rural area 4.68 5.03 6.20 6.24 1.08 1.33 1.33
residents
OIF/OEF Era 24.83 29.68 31.77 32.20 1.20 1.28 1.30
Veterans
Homeless 31.49 35.48 33.90 30.41 1.13 1.08 .97
Service connected 29.66 23.23 25.05 26.12 .78 .84 .88
50%
VA pension 4.14 6.19 4.16 3.59 1.50 1.00 .87
Annual income in 15085.2 (16844.0) 16808.6 (24540.2) 16436.7 (17772.6) 15831.5 (17000.3) .09 .07 .04
dollars (SD)
Black 16.99 15.65 13.89 9.80 .92 0.82 .58
White 80.14 80.89 85.50 87.99 1.01 1.04 1.10
Hispanic 9.89 8.80 11.17 8.86 .89 1.13 .90
Other race .47 .69 .34 .63 1.45 .71 1.32
Unknown race 3.91 6.84 5.65 5.38 1.75 1.45 1.38
Mixed race 1.68 1.66 1.81 1.27 .99 1.08 .76
SD, standard deviation.

Indicates cohens d as the effect size.

574 3-Year Buprenorphine Retention in OUD September 2017

who filled a buprenorphine prescription in the first 60 days of
the FY, we were left with 3,151 patients, who were assumed to
have begun a new episode of treatment on buprenorphine in
FY 2012.
Those started newly on buprenorphine were retained in
treatment for a mean duration of 612.77 days [1.68 years],
(standard deviation [SD] 448.22 days [1.23 years]), with
13.80% (n ¼ 435) engaged for 30 days or less, 24.60%
(n ¼ 775) from 31 days to a year, 29.77% (n ¼ 938) between 1
and 3 years, and 31.83% (n ¼ 1,003) more than 3 years. The
mean number of days of retention was 6.58 days (SD 8.94) in
the 30 days or less group, 157 days (SD 96 days) in the 31–365
days group, 758 days (2.1 years, SD 238 days) in the 1–3 year
group, and 1,222 days (3.5 years; SD 75 days) in the >3 year
group.
Comparison of FY 2012 Characteristics Between
Different Retention Groups
There were no substantial differences between groups in
any demographic characteristics, except for black race (see
Table 1). The proportion of the sample, who self identified as
black were substantially less likely to be among those retained
for more than 3 years compared to the proportion in the 30 days
or less group.
There were no substantial differences between the retention
groups in baseline medical comorbidities, except for overall
Charlson comorbidity index which was significantly lower in
the >3 year retention group compared to 30 days or less group
(Table 2). There was no difference in the high proportion of
patients with a pain diagnosis across the groups. Although
large proportions of patients had other comorbid substance use
disorder diagnoses than OUD (averaging at least one
additional SUD diagnosis per patient) and psychiatric use
disorder diagnoses (averaging two or more diagnoses per
patient), there was no substantial difference across different
retention categories (see Table 2).
Emergency room visits during FY 2012 were lower among
>the 3 year retention group compared to the 30 days or less
retention group (Table 3). As expected, use of all outpatient
visits and any psychiatric or substance abuse outpatient visits
were substantially higher among all the rest of the retention
groups compared to 30 days or less group. Over four-fifths of
the sample received some other psychotropic medications,
with no significant differences across groups (see Table 4).
Survival Analysis and Multivariate Cox Proportional
Hazard Analysis
Cox proportion hazards model showed black race (Hazard
ratio [HR] 1.26; standard error [SE] .06; p.0003), a higher
Charlson index (HR 1.03; SE .01; p.0132) and greater numbers
of emergency room visits during FY 2012 (HR 1.03; SE .01;
p < .0001) were independently associated with a greater
likelihood of discontinuing buprenorphine treatment. Lower
retention in buprenorphine treatment among those self
identified as black compared to others is illustrated in
Kaplan–Meir survival curve in Fig. 1.
Manhapra, Petrakis, and Rosenheck
Facility Based Variations in Buprenorphine
Engagement
Although we observed only minimal variations in
buprenorphine retention duration based on individual charac-
teristics, there was somewhat greater variation based on the
facility, where patients received treatment. Of the 127 VHA
facilities across the nation, where OUD patients were newly
initiated on buprenorphine in FY2012, 91 facilities had 10 or
more new patients and thus were considered for comparative
analysis (Fig. 2). Although buprenorphine retention rates
varied substantially at the extreme ends, the coefficients of
variation reflected only modest variability across facilities for
both more than 1 year retention (average 61.0%, SD 12.6%:
coefficient of variation ¼ .21) and more than 3 year retention
(average 30.9%, SD ¼ 10.58%: coefficient of variation ¼ .34).
DISCUSSION
This study of retention of patients initiating buprenorphine
OAT nationally in the VHA found almost two-thirds of
patients with OUD continued to be engaged in treatment after
1 year and almost one-third for 3 years. It was especially
notable that few individual patient characteristics were
associated with substantially more prolonged OAT retention
with the exception of black race, overall medical severity
indicated by Charlson comorbidity index, and frequent
emergency room visits during the first year of treatment, all
of which were independently associated with a shorter
duration of participation. Modest variation was observed
between VHA facilities from which patients received their
care based on the coefficient of variation.
Retention rates observed in the real-world VHA data are
substantially higher than those observed in multiple other
studies. In prior reports from clinical trials data, buprenorphine
treatment retention varied from 20% at 52 weeks to 53% at
1 year and 25% at 2 years in more recent studies.26,27 One-year
retention rates reported from naturalistic studies are generally
lower than in the VHA sample reported here, with 32–57% in
small practice settings,28–31 just above 20% in a national
program in Australia,8 and 42–45% in insurance claims data
based studies from United States (about 20% retention after 3
years).14,18 Comprehensive Statewide implementation of
office based opioid treatment in Massachusetts in 2012
showed a 1-year retention rate in buprenorphine OAT of 65%,
the only report that approached retention rates reported here.32
It is notable that retention rates were highest in our national
VHA study and in Massachusetts both of which are large
public sector programs that serve patients, who are most likely
unable to seek treatment from private office practices. VHA
facilities tend to provide more comprehensive care than non-
VHA substance abuse treatment centers and this may also
facilitate retention.33
Although not predictive of retention in buprenorphine
OAT, this study provides far more extensive information than
previously available on the characteristics of patients initiating
September 2017 575
TABLE 2. Comparison of baseline medical, substance use disorder, and psychiaytric diagnoses between buprenorphine retention groups

Effect size comparison with 0–30 days

Buprenorphine retention categories category (Risk Ratio/Cohens d )
0–30 31–365 1–3 31–365 365 days–3 >3
days days years >3 years days years years
Total N ¼ 3151 N ¼ 435 N ¼ 775 N ¼ 938 N ¼ 1003
Percent of total 13.81% 24.60% 29.77% 31.83%
Medical diagnoses
Seizures 1.38 1.68 2.03 1.10 1.22 1.47 .80
Insomnia 12.87 12.26 13.01 10.67 .95 1.01 .83
Myocardial infarction .92 .39 .75 .30 .42 .81 .33
Congestive Heart failure 47.13 48.65 44.67 44.57 1.03 .95 .95
Peripheral vascular disease 2.30 1.94 2.67 1.50 .84 1.16 .65
Cerebrovascular accident 3.45 1.55 1.60 1.10 .45 .46 .32
Chronic obstructive airway 14.02 12.65 11.30 13.36 .90 .81 .95
disease
Hepatic disease 13.10 12.26 11.30 11.17 .94 .86 .85
Diabetes mellitus 10.12 9.29 8.32 9.17 .92 .82 .91
Renal disease 3.22 1.68 1.39 1.30 .52 .43 .40
Cancer 3.00 3.61 2.13 2.30 1.21 .71 .77
Charlson full index 1.88 1.75 1.50 1.46 .07 .19 .22
Any pain 74.02 72.39 72.50 70.49 .98 .98 .95
Musculoskeletal pain 40.46 39.10 39.45 37.79 .97 .97 .93
Substance use disorders
Alcohol 46.67 52.00 44.99 42.17 1.11 .96 .90
Cocaine 27.36 30.97 27.93 25.42 1.13 1.02 .93
Cannabis 20.46 26.97 20.26 19.54 1.32 .99 .96
Sedatives 12.64 11.74 12.69 10.86 0.93 1.00 .86
Amphetamine 5.06 7.61 6.61 5.58 1.51 1.31 1.10
Hallucinogen .23 .26 1.49 1.20 1.12 6.49 5.20
Psychiatric diagnoses
Bipolar 13.10 13.03 11.73 11.67 .99 .89 .89
Major depression 23.21 23.61 23.45 22.13 1.02 1.01 .95
Other depression 53.79 55.10 54.37 54.84 1.02 1.01 1.02
PTSD 43.22 44.90 42.21 41.48 1.04 .98 .96
Anxiety disorder 37.70 39.87 38.17 39.28 1.06 1.01 1.04
Adjustment disorder 12.64 13.03 12.90 12.56 1.03 1.02 .99
Personality disorder 8.97 12.90 8.42 7.78 1.44 .94 .87
Schizophrenia 2.76 3.10 3.73 1.79 1.12 1.35 .65
Other Psych diagnoses 34.94 34.71 32.41 32.70 .99 .93 .94

Indicates when cohens d was used as the effect size value and the rest have risk ratios.

buprenorphine OAT. In this VA sample over 90% were male,
over 30% with current or recent homelessness, 74% had pain
diagnoses, frequent diagnoses of other addictive disorders
(half with alcohol use disorder, one-third with cocaine or
amphetamine, one quarter marijuana) with high rates of
psychiatric comorbidity (with an average of two diagnoses per
patient), and over 80% filling psychotropic medication
prescriptions during the year and almost one third with a
psychiatric hospitalization. While these high levels of co-
morbidity and psychotropic medication use suggest high
vulnerability, they may also indicate high levels of distress and
motivation for treatment. This study did not compare veterans
initiating buprenorphine OAT with others with OUD, but such
a comparison may further illuminate the distinctive character-
istics of this sample.
As comprehensively summarized by Dreifuss et al,34
prior small studies have reported that OAT retention for
OUD could be influenced by multiple individual patient
level characteristic including younger age, black race, male
gender, being unmarried/not living with a stable partner,
cocaine use, problematic and/or frequent alcohol use,
comorbid SUD, more severe psychiatric problems, and

576 3-Year Buprenorphine Retention in OUD September 2017

TABLE 3. Comparison of baseline health service and psychotropic medication use between buprenorphine retention groups

Effect size comparison with

0–30 days category (Risk
Buprenorphine retention categories Ratio/Cohens d )
31–365 1–3 >3
0–30 days 31–365 days 1–3 years >3 years days years years
Total N ¼ 3151 N ¼ 435 N ¼ 775 N ¼ 938 N ¼ 1003
Percent of total 13.81% 24.60% 29.77% 31.83%
Healthcare utilization
Any mental health inpatient 35.17 32.39 31.56 25.62 .92 .90 .73
treatment
Emergency room (SD) 2.76 (3.90) 2.38 (4.19) 2.40 (3.61) 1.82 (3.01) .10 .10 .26
Medical/surgical visits (SD) 10.39 (11.70) 10.21 (11.56) 9.78 (10.99) 9.28 (10.55) .02 .06 .10
All outpatient visits (SD) 46.39 (58.60) 61.22 (56.42) 58.33 (60.87) 59.63 (55.08) .26 .21 .23
Any Psych/ substance abuse 36.00 (55.70) 51.02 (53.73) 48.55 (57.92) 50.35 (52.36) .27 .23 .26
visits (SD)
Mean number of prescriptions (SD)
Antidepressants 10.93 (22.00) 11.47 (27.57) 11.06 (18.27) 9.70 (15.26) .03 .01 .06
Antipsychotics 4.76 (11.74) 4.36 (11.73) 4.81 (12.97) 3.23 (10.23) .03 .00 .13
Anxiolytics/sedatives/hypnotics 4.25 (9.17) 3.79 (9.10) 3.84 (7.97) 3.83 (9.39) .03 .02 .08
Stimulants .23 (1.49) .16 (1.15) .18 (1.22) .42 (3.89) .05 .05 .05
Anticonvulsants or mood 5.06 (15.04) 4.51 (13.41) 3.84 (10.88) 3.10 (8.56) .05 .11 .17
stabilizers
Lithium .30 (1.99) .51 (4.00) .57 (5.92) .25 (2.10) .05 .07 .01
All psychotropics 25.52 (45.10) 24.79 (45.59) 24.30 (39.27) 20.52 (33.13) .02 .03 .12
Percent getting prescription
Antidepressants 70.57 76.65 76.86 75.18 1.09 1.09 1.07
Antipsychotics 37.47 32.90 33.69 28.31 .88 .90 .76
Anxiolytics/sedatives/ghypnotics 50.35 45.42 44.67 43.67 .90 .89 .87
Stimulants 3.68 2.97 3.63 3.59 .81 .99 .98
Anticonvulsants or mood 40.92 40.13 34.65 32.70 .98 .85 .80
stabilizers
Lithium 3.68 3.48 3.20 2.69 .95 .87 .73
All psychotropics 83.44 87.74 84.86 85.94 1.05 1.02 1.03
SD:, Standard deviation.

Indicates when cohens d was used as the effect size value and the rest have risk ratios.

personality disorders. Surprisingly in our study, black race

was the only characteristic found to be associated with
earlier discontinuation, although all were common in our
sample. Of course, the low percentage of women in this
study and in the VA population in general makes it difficult
to comment on the role of gender. The higher Charlson
comorbidity index and higher frequency of ED visits,
characteristics associated with shorter treatment retention,
may reflect the presence of patients with more severe opioid
dependence and medical disease burden, who might find it
harder to stay engaged due to the demands for regular
attendance and adherence to regulations that even bupre-
norphine treatment requires. More specifically, the higher
FIGURE 1. Kaplan–Meir survival analysis on probability of
Charlson indices among patients who discontinued treatment
retention in buprenorphine treatment: Black versus other races.
may reflect discontinuation due to higher mortality.

Manhapra, Petrakis, and Rosenheck September 2017 577

 FIGURE 2. Facility based variation buprenorphine retention duration among opioid use disorder patients initiated on buprenorphine treatment in
FY 2012.
African Americans with OUD appear to be facing two
identifiable impediments regarding buprenorphine in that they
tend to be started on buprenorphine less frequently,35 and are
retained for a shorter duration compared to other racial groups.
Gryczynski et al36 in a study of buprenorphine retention
among African Americans have reported that despite a
majority wanting to stay in treatment for more than 6 months,
a substantial proportion left treatment early. Only a small
proportion left for other providers (14%), finished treatment
successfully (4%), or were dismissed due to positive urine
toxicologies (9%), while that majority left due to program
related issues like conflict with staff (24%), missing too many
scheduled visits (17%), program conflict with life obligations
(17%), or discomfort with provider strictness (4%). Such
programs identified structural factors and other organizational
factors that influenced buprenorphine adoption and retention
generally,37–40 and may have impacted African Americans
more strongly than others. At present, however, the lower
retention among black veterans lacks a clear explanation, but
deserves further study.
Since very few of the individual veterans characteristics
were associated with buprenorphine retention in this study, it
is possible that structural and organizational variability
between programs are more likely to be salient as reflected
in the variability observed in retention between different
facilities. A recent study from Medicaid program in the state of
Massachusetts that had a statewide office-based opioid
578 3-Year Buprenorphine Retention in OUD
treatment program implementation showed substantial varia-
tion in structure of treatment and adherence to guidelines
across the state.41 Substantial variations across individual
VHA facilities nationally regarding buprenorphine utilization
reported in an earlier study42 and buprenorphine treatment
retention in our study together points to a need for further
qualitative examination of organizational level practice
variations and possible policy interventions to increase
buprenorphine utilization and retention among veterans
treated in VHA.
Limitations
Several limitations deserve note. We lacked data on dose
strength of buprenorphine, duration, and severity of OUD,
type of opioid used or psychosocial factors, which can all
influence treatment retention.14,18,34 We also did not have
information on treatment interruptions, although we think
such interruptions do not diminish the importance of evidence
of sustained involvement. Prior studies have reported that
about 70% of the patients retained up to 1 year on
buprenorphine refilled it persistently,15 and more than 90%
of patients retained in buprenorphine treatment for 12 months
were no longer actively using substances,31 suggesting that a
substantial proportion of patients retained at 1 year are
probably continuously engaged in treatment for OUD
successfully. Data on other details on treatment adherence
September 2017
might also have been informative, but would not have changed
our major findings.
Another limitation of this study is that we lack data on
reasons for discontinuation. It is likely that some proportion of
patients leave buprenorphine treatment voluntarily, either to
return to opioid misuse, or because they no longer need
treatment. In one 6-month follow up study, patients expected
to stay in treatment for not longer than 6 months on average,
while average length of stay in the program was close to 4.5
months, pointing toward an expectable gap between patient
expectations and actual retention.36 Proportions of patients
with successful completion of treatment and tapering off
buprenorphine have been reported to be in single digits in prior
studies suggesting most discontinuation is premature.31,36 A
review of methadone maintenance outcome studies reported
that only about 30% of those with OUD, who discontinue
methadone maintenance stay abstinent over 1–2 years of
follow up.43 Similar post-discontinuation data is not available
for this or other buprenorphine maintenance studies, but is
much needed. We also lacked mortality data or additional
outcome data that would illuminate the implications of
variations in treatment retention. As with any observational
study using administrative data, this study provides only
correlational perspectives on reasons for discontinuation.
CONCLUSIONS
In this long-term, national VHA study retention in
buprenorphine OAT was substantial but few individual patient
characteristics were associated with retention. It is encourag-
ing that this level of retention was achieved in a sample with
high levels of co-morbidity. Intervention research is needed to
identify methods for increasing buprenorphine retention
among all segments of this vulnerable population.
Ajay Manhapra was supported by the Research in Addiction
Medicine Scholars (RAMS) Program, R25DA033211 from the
National Institute on Drug Abuse, and VA/OAA Interprofes-
sional Advanced Fellowship in Addiction Treatment.
Declaration of Interest
The authors report no conflicts of interest. The authors
alone are responsible for the content and writing of this paper.
REFERENCES
1. Jones CM, Campopiano M, Baldwin G, et al. National and state treatment
need and capacity for opioid agonist medication-assisted treatment. Am J
Public Health. 2015; 105:e55–e63.
2. Nih CP. Effective medical treatment of opiate addiction. National
consensus development panel on effective medical treatment of opiate
addiction. JAMA. 1998;280:1936–1943.
3. Center for Substance Abuse T. SAMHSA/CSAT Treatment improvement
protocols. Clinical guidelines for the use of buprenorphine in the
Manhapra, Petrakis, and Rosenheck
treatment of opioid addiction. Rockville (MD): Substance abuse and
mental health services administration (US); 2004.
4. Saxon AJ, Commentary on Burns et al. (2015): Retention in
buprenorphine treatment. Addiction. 2015;110:656–657.
5. Bell J, Trinh L, Butler B, et al. Comparing retention in treatment and
mortality in people after initial entry to methadone and buprenorphine
treatment. Addiction. 2009;104:1193–1200.
6. Center for Substance Abuse T. SAMHSA/CSAT Treatment improvement
protocols. Medication-assisted treatment for opioid addiction in Opioid
Treatment Programs. Rockville (MD): Substance abuse and mental health
services administration (US); 2005.
7. SAMHSA. Results from the 2013 national survey on drug use and health:
Summary of national findings, NSDUH Series H-48, HHS Publication No.
(SMA) 14–4863. Rockville, MD 2014.
8. Burns L, Randall D, Hall WD, et al. Opioid agonist pharmacotherapy in
New South Wales from 1985 to 2006: Patient characteristics and patterns
and predictors of treatment retention. Addiction. 2009;104:1363–1372.
9. Riksheim M, Gossop M, Clausen T. From methadone to buprenorphine:
Changes during a 10 year period within a national opioid maintenance
treatment programme. J Subst Abuse Treat. 2014;46:291–294.
10. SAMHSA. Trends in the use of Methadone and Buprenorphine at
substance abuse treatment facilities: 2003 to 2011. Rockville, MD:
Substance abuse and mental health services administration, Center for
Behavioral Health Statistics and Quality; 2013.
11. Turner L, Kruszewski SP, Alexander GC. Trends in the use of
buprenorphine by office-based physicians in the United States, 2003–
2013. Am J Addict. 2014;24:24–29.
12. Nosyk B, Anglin MD, Brissette S, et al. A call for evidence-based medical
treatment of opioid dependence in the United States and Canada. Health
Aff. 2013;32:1462–1469.
13. Mattick R, Kimber J, Breen C. Buprenorphine maintenance versus
placebo or methadone maintenance for opioid dependence. Cochrane
Database of Systematic Reviews 2008, Issue 2. Art. No.: CD002207.
Cochrane Library. 2000.
14. Clay E, Khemiri A, Zah V, et al. Persistence and healthcare utilization
associated with the use of buprenorphine/naloxone film and tablet
formulation therapy in adults with opioid dependence. J Med Econ.
2014;17:626–636.
15. Lo-Ciganic WH, Gellad WF, Gordon AJ, et al. Associations between
trajectories of buprenorphine treatment and emergency department and
inpatient utilization. Addiction. 2016;111:892–902.
16. Bart G. Maintenance medication for opiate addiction: The foundation of
recovery. J Addict Dis. 2012;31:207–225.
17. Timko C, Schultz NR, Cucciare MA, et al. Retention in medication-
assisted treatment for opiate dependence: A systematic review. J Addict
Dis. 2016;35:22–35.
18. Khemiri A, Kharitonova E, Zah V, et al. Analysis of buprenorphine/
naloxone dosing impact on treatment duration, resource use and costs in
the treatment of opioid-dependent adults: A retrospective study of US
public and private health care claims. Postgrad Med. 2014;126:113–120.
19. Golub A, Vazan P, Bennett AS, et al. Unmet need for treatment of substance
use disorders and serious psychological distress among veterans: A
nationwide analysis using the NSDUH. Mil Med. 2013;178:107–114.
20. Gordon AJ, Geppert CM, Saxon A, et al. Models for implementing
buprenorphine treatment in the VHA. Federal Pract. 2009;26:48–57.
21. Gordon AJ, Kavanagh G, Krumm M, et al. Facilitators and barriers in
implementing buprenorphine in the veterans health administration.
Psychol Addict Behav. 2011;25:215–224.
22. Stanton A. The SAMHSA evaluation of the impact of the DATA waiver
program. Summary report. Rockville, MD: WESTAT. 2006.
23. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying
prognostic comorbidity in longitudinal studies: Development and
validation. J Chronic Dis. 1987;40:373–383.
24. Wiechers IR, Kirwin PD, Rosenheck RA. Increased risk among older
veterans of prescribing psychotropic medication in the absence of
psychiatric diagnoses. Am J Geriatr Psychiatry. 2014;22:531–539.
September 2017 579
25. Ferguson CJ. An effect size primer: A guide for clinicians and researchers.
Professional Psychology: Research and Practice. 2009;40:532–538.
26. Ling W, Wesson DR, Charuvastra C, et al. A controlled trial comparing
buprenorphine and methadone maintenance in opioid dependence. Arch
Gen Psychiatry. 1996;53:401–407.
27. Fiellin DA, Moore BA, Sullivan LE, et al. Long-term treatment with
buprenorphine/naloxone in primary care: Results at 2–5 years. Am J
Addict. 2008;17:116–120.
28. Haddad MS, Zelenev A, Altice FL. Integrating buprenorphine maintenance
therapy into federally qualified health centers: Real-world substance abuse
treatment outcomes. Drug Alcohol Depend. 2013;131:127–135.
29. Parran TV, Adelman CA, Merkin B, et al. Long-term outcomes of office-
based buprenorphine/naloxone maintenance therapy. Drug Alcohol
Depend. 2010;106:56–60.
30. Soeffing JM, Martin LD, Fingerhood MI, et al. Buprenorphine
maintenance treatment in a primary care setting: Outcomes at 1 year.
J Subst Abuse Treat. 2009;37:426–430.
31. Alford DP, LaBelle CT, Kretsch N, et al. Collaborative care of opioid-
addicted patients in primary care using buprenorphine five-Year
experience. Arch Intern Med. 2011;171:425–431.
32. LaBelle CT, Han SC, Bergeron A, et al. Office-based opioid treatment
with buprenorphine (OBOT-B): Statewide implementation of the
massachusetts collaborative care model in community health centers.
J Subst Abuse Treat. 2016;60:6–13.
33. SAMHSA. The N-SSATS Report: Substance abuse treatment facilities
operated by the Department of Veterans Affairs (November 11, 2009).
Rockville, MD: Substance abuse and mental health services administra-
tion, Office of Applied Studies; 2009.
34. Dreifuss JA, Griffin ML, Frost K, et al. Patient characteristics associated
with buprenorphine/naloxone treatment outcome for prescription opioid
580 3-Year Buprenorphine Retention in OUD
dependence: Results from a multisite study. Drug Alcohol Depend.
2013;131:112–118.
35. Manhapra A, Quinones L, Rosenheck R. Characteristics of veterans
receiving buprenorphine vs methadone for opioid use disorder nationally
in the Veterans Health Administration. Drug Alcohol Depend.
2016;160:82–89.
36. Gryczynski J, Mitchell SG, Jaffe JH, et al. Leaving buprenorphine
treatment: Patients’ reasons for cessation of care. J Subs Abuse Treat.
2014;46:356–361.
37. Mitchell SG, Gryczynski J, Schwartz RP, et al. A randomized trial of intensive
outpatient (IOP) vs standard outpatient (OP) buprenorphine treatment for
African Americans. Drug Alcohol Depend. 2013;128:222–229.
38. Knudsen HK, Ducharme LJ, Roman PM. Early adoption of buprenor-
phine in substance abuse treatment centers: Data from the private and
public sectors. J Subst Abuse Treat. 2006;30:363–373.
39. Knudsen HK, Ducharme LJ, Roman PM. The adoption of medications
in substance abuse treatment: Associations with organizational
characteristics and technology clusters. Drug Alcohol Depend.
2007;87:164–174.
40. Knudsen HK, Ducharme LJ, Roman PM, et al. Buprenorphine diffusion:
The attitudes of substance abuse treatment counselors. J Subst Abuse
Treat. 2005;29:95–106.
41. Baxter JD, Clark RE, Samnaliev M, et al. Adherence to buprenorphine
treatment guidelines in a medicaid program. Subst Abuse.
2015;36:174–182.
42. Oliva EM, Harris AH, Trafton JA, et al. Receipt of opioid agonist
treatment in the Veterans Health Administration: Facility and patient
factors. Drug Alcohol Depend. 2012;122:241–246.
43. Kornor H, Waal H. From opioid maintenance to abstinence: A literature
review. Drug Alcohol Rev. 2005;24:267–274.
September 2017