Comparison of Demographic and Clinical Characteristics of Opioid-Dependent Admissions Between Community and Research-Based Treatment Settings

Abstract

Despite significant developments in pharmacotherapy and behavioral treatments for addiction, dissemination of new treatment methods into the community has been slow. It has been pointed out that treatments developed in research settings may be impractical in community treatment settings, which might help explain this lag. Participant recruitment and screening for research studies might partially explain this, as there is evidence that substance abusers who participate in clinical research are different on a number of measures from treatment-seekers. However, no study has directly compared treatment seekers with research participants drawn from similar populations using prospective methods. The present study compares the demographic, drug use and psychosocial problem severities, and personality traits of opioid-dependent people seeking help in a community (n=502) versus primarily research-based (n=459) drug abuse treatment setting; both settings offered a similar set of treatment services (opioid agonist medication and counseling). While the overall findings reveal numerous similarities between the samples, differences were observed. Most notably, there were significantly fewer women in the research sample than in the community-based treatment setting. Other differences included a modest but statistically significant increase in psychosocial problem severity in the community-based setting, and higher drug use problem severity in the research setting. Interestingly, many of these differences were strongest in women versus men.

1.0 Introduction

Many novel pharmacological and behavioral treatments for drug addiction have been tested in multiple research programs over the past few decades (Carroll, Sinha, Nich, Babuscio, & Rounsaville, 2002; Chutuape, Silverman, & Stitzer, 1999; Eissenberg et al., 1997; Foster, Brewer, & Steele, 2003; Griffith, Rowan-Szal, Roark, & Simpson, 2000; Jones et al., 1998; Kakko, Svanborg, Kreek, & Heilig, 2003; Pani, Maremmani, Pirastu, Tagliamonte, & Gessa, 2000), yet broad dissemination and implementation of these advances in community-based settings lags behind research findings (Institute of Medicine, 1998). A number of interrelated issues contribute to this ongoing challenge (Institute of Medicine, 1998; Russell & Orlinsky, 1996; Seligman, 1995; Strain & Stitzer, 1999). Among them is the concern that participants in research settings may differ in important ways from those seeking care in community-based sites, and that these differences may confound the effects of experimental interventions (McKay et al., 1998) and limit generalizability of results to non-research facilities.

Differences between research and publicly-funded community-based treatment programs in recruiting practices and eligibility criteria may affect the demographic and psychiatric characteristics of patients that choose to enroll at these facilities. For instance, research-based settings often attract participants through mixed media advertisements (Krupnick, Shea, & Elkin, 1986) that describe general inclusion criteria and offer financial compensation, while the advertising practices of publicly-supported community-based clinics are usually less extensive. In addition, research-based programs often use screening procedures that are narrowly defined by specific study protocols, whereas community-based programs employ more inclusive eligibility criteria.

No studies have prospectively compared the characteristics of substance users enrolled in either a research or community-based setting. Studies that have employed alternative methodologies provide some evidence that people who enroll in research differ from clinical populations. Humphreys and Weisner (Humphreys & Weisner, 2000), for instance, used exclusion criteria common to studies on alcohol problems to categorize people seeking community-based treatment for an alcohol use disorder as eligible or ineligible for research participation. These criteria produced a predominately Caucasian research-eligible sample with comparatively greater financial resources and less psychiatric, medical, and other psychosocial severity. Carroll et al. (Carroll, Nich, McLellan, McKay, & Rounsaville, 1999) compared participants enrolled in two clinical trials for cocaine dependence (Carroll et al., 1994; Carroll, Nich, Ball, McCance, & Rounsavile, 1998) with those in community-based cocaine and alcohol abuse treatment programs (McLellan et al., 1994). The results dovetail with those observed by Humphreys and Weisner (Humphreys & Weisner, 2000) in that research participants were more likely to be Caucasian and relied less on public assistance. Research participants also were more educated and reported more use of cocaine and other drugs.

The methods used by these studies possess both strengths and limitations. Each was retrospective and used large samples, enhancing the likelihood of detecting between-group differences. Yet the Humphreys and Weisner (Humphreys & Weisner, 2000) study relied on an analogue design that may not represent the combined and interactive effects of recruitment and screening, while the Carroll et al. (Carroll et al., 1999) study may have been confounded by the use of comparison groups that were recruited and examined at different locations and times.

The present study addresses these concerns by using a prospective design to evaluate admissions to a primarily research-oriented drug abuse treatment clinic (Behavioral Pharmacology Research Unit-BPRU) versus a primarily community-based treatment clinic (Addiction Treatment Services-ATS). These two programs are located on the Johns Hopkins Bayview Medical Center campus, draw from similar communities, and employ similar treatment approaches that include opioid agonist medication and counseling services. Participants from each program were compared on demographics and multidimensional measures of problem severity and psychological traits.

2.0 Methods

2.1 Participants

Participants were 961 admissions to opioid agonist treatment from April 1984 to February 1998. All participants met DSM-III-R criteria for opioid dependence and Food and Drug Administration requirements for treatment with opioid agonist medications. Participants were paid between $30.00 or $40.00 for completing the intake assessment measures (higher reimbursement rates were used in later years of the study). All data were collected as part of a long-term study of the demographic, drug use, personality, and psychosocial characteristics of treatment-seeking opioid-dependent people. The study was reviewed and approved by the Johns Hopkins Institutional Review Board for research with human subjects.

2.2 Treatment Settings, Recruitment, and Screening Procedure

Participants represented consecutive admissions to a series of outpatient treatment studies that were conducted between 1984 and 1998; participants in these primary studies also provided informed written consent to participate in the present study. Participants sought admission to either ATS or the BPRU.

2.2.1 BPRU Program

BRPU is a research-based program that maintains an outpatient clinic supporting both behavioral and pharmacology studies of opioid and other drug dependence. BPRU actively recruits research participants through advertisements in local papers and other media. The BPRU admission process used a pre-admission assessment and screening, followed by a final assessment (including a medical assessment and physical examination) at the time of admission. Major exclusion criteria were: 1) significant psychiatric illness (e.g., schizophrenia), 2) significant medical problem (e.g., diabetes), 3) cognitive impairment that would impede ability to consent or participate, or 4) pregnancy. Medical exclusions were based on the judgment of the medical staff in the program based on a history, physical examination, and routine laboratory work. Admission and opioid agonist medication typically commenced within two weeks of the pre-admission screening.

2.2.2 ATS Program

ATS is an outpatient hospital-based community treatment program that did not actively recruit patients over the course of the study; admission was based primarily on the presence of opioid dependence disorder and treatment slot availability. The most frequent referral source to the program was the Baltimore Substance Abuse Systems, the primary entity responsible for oversight and coordination of publicly supported substance abuse treatment programs in Baltimore, Maryland. The ATS admission process consisted of an intake evaluation of psychiatric, medical, and substance use history. Participants were admitted on the day of their initial assessment.

2.3 Charges for Services

There were typically no charges for treatment services provided to participants in the BPRU studies. All admissions to ATS were charged the same fee although this changed over the course of the years, ranging from about $65.00 per week to $87.50 (current). The standard charge was adjusted by a sliding fee schedule issued annually by the Maryland State Department of Health and Human Services. The sliding fee schedule reduces the standard charge for uninsured patients based on family income and number of dependents. Over the years included in this report, the adjusted fee ranged from as low as $5.00 per week to about $70.00 per week; the average reduced charge over study years was $31.00 per week.

2.4 Study Measures

All participants were administered a standard demographic assessment common to both settings, the Addiction Severity Index (ASI; McLellan et al., 1992), and the NEO-Personality Inventory (NEO-PI and NEO-PI-R; McCrae & Costa, 1987). The demographic questionnaire and the ASI were administered by trained interviewers, the NEO-PI was computer administered. Participants who had difficulty reading or understanding NEO-PI questions could activate a computerized verbal recording of the questions. Research assessment staff was present during the completion of the NEO-PI and assisted all participants who required additional help.

Significant between-program differences were noted for the times of administration of both the ASI and the NEO-PI. ATS participants completed the ASI 3-4 weeks after admission (M = 24.0 days, SD = 58.7 days), while BPRU admissions completed this assessment on or close to the day of admission (M = 0 days, SD = 0.3 days; t = -8.8; df = 959, p < .001). Similarly, ATS patients completed the NEO an average of 34.2 days (SD = 32.9 days) post-admission, while BRPU participants completed the NEO-PI on the first day of screening, which was an average of 9.1 days (SD = 8.0 days) prior to admission (t = 27.4, df = 959, p < .001).

2.4.1 Addiction Severity Index (ASI- 5^th edition; McLellan, Luborsky, Woody, & O’Brien, 1980; McLellan et al., 1992)

The ASI is a semi-structured interview that assesses problem severity over the previous 30 days in seven areas commonly affected by substance use: drug and alcohol use, psychiatric, medical, family/social, legal, and employment functioning. ASI composite scores have been shown to have good reliability and validity (McLellan et al., 1992).

2.4.2 NEO-PI and NEO-PI-Revised (NEO-PI-R; Piedmont, 1998)

The original NEO-PI was administered to participants admitted before April, 1993; after this date, the NEO-PI-R was administered. A similar proportion of patients at each site received the NEO-PI-R. The NEO-PI and its successor are self-report instruments that measure personality traits defined by the Five-Factor Model of Personality (Carter et al., 2001; P. T. Costa Jr., Busch, Zonderman, & McCrae, 1986; P. T. Costa Jr. & McCrae, 1987; P. T. Costa Jr. & McCrae, 1988; McCrae, Costa, & Arenberg, 1980; McCrae & Costa, 1985; McCrae & Costa, 1987; McCrae & Costa, 1989a; McCrae & Costa, 1989b; Piedmont, 1998). The five traits measured are Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Previous studies of these instruments in opioid dependent outpatients have demonstrated good stability that is only minimally affected by drug use (Carter et al., 2001).

2.5 Interviewer Training

The ASI and NEO-PI were administered in both facilities by the same assessment staff, directed by Kori Kindbom, M.A. and Robert K. Brooner, Ph.D. Study interviewers had either a Bachelor’s or a Master’s degree in the behavioral sciences and completed a structured training protocol for all of the measures employed in the study (e.g. Brooner, King, Kidorf, Schmidt, & Bigelow, 1997). The training protocol included both didactic review and practice interviewing, along with co-rating interviews done by an expert.

2.6 Data Analysis

Because the two treatment samples had substantially different gender compositions (see section 3.1, below), analyses were stratified by gender. Age and amount of treatment exposure prior to completing the ASI and NEO also differed across the two programs and were used as covariates in general linear models (GLMs). Lifetime months of drug use were extracted from individual items in the ASI as another measure of baseline drug use severity. These were compared using similar methods, controlled for gender, age, education, and race (all significant predictors in univariate analyses). All statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) version 11 for Mac OS X.

3.0 Results

3.1 Comparison of ATS and BPRU samples on Demographic Characteristics

The demographic composition of the two treatment settings differed in gender and age. The BPRU sample had almost twice the percentage of males (64.7% vs. 32.7%; p < 0.001) and was slightly older than their ATS counterparts (37.4 vs. 35.3 years; p<0.001). Racial composition was nearly identical (proportion Caucasian: 49.9% ATS, 45.75% BPRU; p = 0.570), as was educational level (ATS 11.06 years, BPRU 11.21 years; p = 0.200).

3.2 Comparison of ATS and BPRU samples on ASI scores

Table 1 shows that BPRU men and women had higher Drug Use composite scores than their ATS counterparts, although the magnitude of difference was greater for women (Beta = 0.45) than men (Beta = 0.15). In addition, men and women in BPRU had lower Medical and Family/Social composite scores compared to participants in ATS, and women in BPRU had higher Legal composite scores, while men in BPRU lower Psychiatric composite scores.

Table 1.

Comparisons of Treatment Settings on ASI Composite Scores

ASI Problem Area1	ATS (community) n=502	BPRU (research) n=459	F (1 d.f.)	p-value
Men	n=164	n=297
Medical	0.277	0.112	27.916	<0.001
Employment	0.732	0.695	1.780	0.183
Alcohol	0.057	0.043	1.181	0.278
Legal	0.149	0.168	0.752	0.386
Family/Social	0.143	0.080	12.776	<0.001
Psychiatric	0.091	0.035	15.273	<0.001
Drug	0.317	0.347	10.204	0.001
Women	n=338	n=162
Medical	0.210	0.120	3.883	<0.001
Employment	0.799	0.789	0.680	0.497
Alcohol	0.030	0.030	1.244	0.214
Legal	0.090	0.144	4.062	<0.001
Family/Social	0.164	0.102	4.290	<0.001
Psychiatric	0.080	0.060	1.633	0.102
Drug	0.232	0.352	122.70	<0.001

¹All means adjusted for age and time in treatment

3.3 Comparison of ATS and BPRU samples on Lifetime Drug Use

Although lifetime months using multiple drugs, alcohol, and heroin differed by setting (ATS vs. BPRU), these differences were not statistically significant when age, education, gender, and race were entered into the model (polysubstance F=1.184, p=0.277; heroin F=0.92, p=0.338; cocaine F=0.157, p=0.692; alcohol F=3.377, p=0.066).

3.4 Comparison of ATS and BPRU samples on NEO-PI scores

Table 2 shows that BPRU women had higher Neuroticism and Agreeableness factor scores than ATS women, and both men and women at BPRU had lower Conscientiousness factor scores than their ATS counterparts.

Table 2.

Comparison of Treatment Settings on Personality Traits

NEO-PI Factor	ATS (community) n=502	BPRU (research) n=459	F (1 d.f.)	p-value
Men	n=164	n=297
Neuroticism	57.292	60.446	2.987	0.085
Extraversion	46.529	46.418	0.005	0.943
Openness	46.372	44.799	1.061	0.303
Agreeableness	43.569	43.791	0.016	0.901
Conscientiousness	43.235	37.159	8.796	0.003
Women	n=338	n=162
Neuroticism	56.275	59.077	9.425	0.002
Extraversion	47.779	46.384	2.452	0.118
Openness	44.531	45.574	1.961	0.162
Agreeableness	37.084	40.590	8.578	0.004
Conscientiousness	41.860	36.612	21.975	<0.001

4.0 Discussion

This appears to be the first prospective report on possible selection biases in opioid users seeking treatment in a community-based versus primarily research-based drug abuse treatment setting. Some between-group differences were found but they were most notable in women, who were also underrepresented in the BPRU (research) sample. Overall findings revealed as many similarities as differences in participants seeking treatment in the community versus research-based setting. Consistent with this basic pattern and interpretation, many of the statistically significant differences between the groups seemed clinically unimpressive. A discussion of these findings and their implications is presented below.

4.1 Demographic Characteristics

The community setting enrolled a larger proportion of women than the research-based treatment setting, even though neither program explicitly used gender as an admission criterion. BPRU excluded individuals who screened positive for severe psychiatric or medical problems. This practice may have inadvertently reduced the number of female admissions. Prior research has shown that women are more likely than men to report psychological distress and problems (Brown, Alterman, Rutherford, Cacciola, & Zaballero, 1993; Dansky, Byrne, & Brady, 1999; Davis & DiNitto, 1996; Grella & Joshi, 1999; McLellan et al., 1992; Schneider, Kviz, Isola, & Filstead, 1995); this may have led to a higher rate of exclusion in women versus men during the screening process. Interestingly, the two samples were otherwise similar in demography, a finding that is inconsistent with earlier reports (Carroll et al., 1994; Carroll et al., 1999). Although speculative, the demographic similarities observed in the present report may be related to the fact that the research and community-based programs are located on the same campus, draw from the same geographic areas of the city, and represent a similar socioeconomic stratum.

4.2 Drug use, medical, and other psychosocial problem severity

Consistent with some past reports (Carroll et al., 1999), participants enrolled in a primarily research-based setting reported more drug use problems than those in a community setting, although this finding was much more pronounced in women (Drug Use composite scores: ATS: 0.23 vs. BPRU: 0.35). The potential importance of this finding is well documented by prior work showing that baseline drug use severity has prognostic significance (Compton, Cottler, Jacobs, Ben-Abdallah, & Spitznagel, 2003; Downey, Helmus, & Schuster, 2000; Magura, Rosenblum, Fong, Villano, & Richman, 2002; McKay et al., 1998). Interestingly, the difference in baseline Drug Use composite severity scores in the present study resulted from low scores in the ATS sample (especially women) rather than very high scores in BPRU participants. For example, Drug Use composite scores ranging from 0.3 to 0.4 are typical of treatment-seeking opioid-dependent samples (e.g. Brooner et al., In Press), which make the scores in the ATS sample notable. The reason for the comparatively low ATS scores is uncertain. It may be related to the group difference in the amount of treatment received prior to administration of the study measures (approximately three and one-half weeks). Studies have shown large reductions in self-reported drug use and other psychosocial problems within the first month of treatment (Strain, Stitzer, Liebson, & Bigelow, 1993; Strain, Stitzer, Liebson, & Bigelow, 1996; Strain, Bigelow, Liebson, & Stitzer, 1999). However, the difference in the amount of treatment provided in the present study was at least partially controlled by the covariate analyses. Lower drug use scores by the ATS group could also reflect unique aspects of the early treatment experience at this setting (the ATS clinic), distinctive features of the local community (the Baltimore area), or other characteristics of this population that are not captured by the current assessments.

The lower Medical and Psychiatric composite severity scores in BPRU versus ATS participants appear counterintuitive given their higher Drug Use composite score. These findings, however, may be related to the pre-admission screening process at BPRU, which excluded applicants reporting substantial medical or psychiatric problems and distress. The same cannot be said for the difference in the Family/Social composite severity scores. The pre-admission screening in BPRU did not try to exclude people with family and social problems, but may have inadvertently done so by excluding those reporting higher levels of psychiatric distress. It should also be noted that these scores were low for both groups, and that prior work suggests that the ASI may underestimate impairment in drug-dependent people with long-standing family and social problems (Brooner, Schmidt, & Herbst, 2002).

4.3 NEO-PI factor scores

The best overall summary of the personality traits of the two samples may be that the differences observed between the BPRU and ATS groups are less impressive than the differences between each of them compared to the general population (P. T. Costa & McCrae, 1992). The same overall pattern of findings was also noted in a study that compared the NEO-PI scores of drug-dependent patients with versus without distinct patterns of other psychiatric comorbidity (Brooner, Herbst, Schmidt, Bigelow, & Costa, 1993). While psychiatric comorbidity was associated with some differences in personality trait standings in that study, the subgroups were much more similar to each other than they were to the general population. It is also noteworthy that three of the four personality trait differences associated with treatment setting in the present study were found in women. The reasons for these gender specific findings are unclear and may be related to multiple factors, including both program (e.g., differences in pre-admission screening goals and procedures) and even subtle individual differences. Perhaps most importantly, while these differences were statistically significant they were largely unimpressive from a clinical perspective.

4.4 Strengths, Limitations, and Implications

The study has several important strengths and limitations. The primary strength is the prospective study design and comparatively large sample of men and women. Another strength of the study is that both treatment settings offered a similar set of clinical services and drew from the same urban community, enhancing the likelihood of detecting differences based on admission screening and possible self-selection biases. A major weakness of the study was the difference between groups in number of days of treatment prior to completing the study measures. This difference was controlled statistically but the analysis may not have completely eliminated the problem, particularly with the ASI that is shown to change over time in response to treatment. In contrast, the NEO-PI has been shown to have good retest reliability over months and years (McCrae et al., 1980), including in treatment-seeking drug-dependent patients tested on the day of admission and after 4 months of treatment (Carter et al., 2001). Lastly, while the study enjoyed a comparatively large sample, the results nonetheless require replication in other programs, samples, and cities.

The study has several implications that may prove important to the field. The first is that there may be more similarities than differences in opioid-dependent people seeking help from primarily research versus community-based treatment settings. If the present findings are replicated in other samples and programs, this would meaningfully extend the external validity of studies conducted in research programs to community-based programs. The under-representation of women in the BPRU setting also provides additional support for the ongoing efforts to include more women in drug abuse treatment research. The possibility that women enrolling in a primarily research-based setting might have greater drug use severity must also be considered when generalizing results to community-based treatment settings.