Chapter 6. Length of Stay among Female Clients in Substance Abuse Treatment

Empirical Model

Anderson and colleagues described a model of access and utilization of health care services that incorporated domains of "individual determinants" of health care utilization (Aday et al., 1999). Predisposing characteristics occur before onset and include demographic characteristics (e.g., age, race/ethnicity, education). Enabling characteristics include such factors as referral source for treatment and insurance status. The model suggests that organizational characteristics (e.g., provision of child care services and women-only facilities) also influence access and utilization. The distinction among predisposing, enabling, and organizational characteristics is used in our study to examine factors accounting for variation in retention among women.¹

Predisposing Characteristics: Age, Race, and Education

Age has been shown to be an important determinant of LOS in treatment, as some reports show that older adults remain in treatment longer (Wickizer et al., 1994). For example, analyses of women in residential drug treatment programs in the Drug Abuse Treatment Outcome Study (DATOS) found that age had a significant positive effect on retention (Grella, Joshi, & Hser, 2000). In addition, race/ethnicity has been found to be associated with LOS (McCaul et al., 2001). For example, in a study based in a health maintenance organization (HMO) outpatient alcohol and drug treatment program (Mertens & Weisner, 2000), characteristics predictive of retention among women included belonging to racial/ethnic groups other than African American. Education also has been related to LOS. In general, higher educational attainment is associated with longer LOS (Kelly, Blacksin, & Mason, 2001; Knight, Logan, & Simpson, 2001).

Other predisposing characteristics affecting LOS have been identified, including marital status, presenting substance abuse problem at admission, severity of substance use, age at first use, and mental disorder symptom severity (Fishman, Reynolds, & Riedel, 1999; Joe, Simpson, & Broome, 1999; Lang & Belenko, 2000; Maglione, Chao, & Anglin, 2000; Mertens & Weisner, 2000; Rowan-Szal, Joe, & Simpson, 2000; Veach, Remley, Kippers, & Sorg, 2000; Williams & Roberts, 1991). Although these characteristics are not major variables of interest in the present study, our analyses controlled for many of these potentially confounding variables.

Enabling Characteristics: Referral Source and Payment Source

Prior research has associated LOS in treatment with referral source (Wickizer et al., 1994). Women often are referred to substance abuse treatment through child protective services or the criminal justice system. One federally sponsored project supporting 35 residential treatment programs for substance-abusing pregnant or postpartum women found that child protective services or the criminal justice system referred 37 percent of female clients (Clark, 2001). Involvement with the criminal justice system has been associated with longer stays in substance abuse treatment, even if the criminal justice system was not the primary source of referral for treatment (Green et al., 2002; Grella, Anglin, Wugalter, Rawson, & Hasson, 1994; Nishimoto & Roberts, 2001).

Payment source may be a factor in predicting LOS in substance abuse treatment. Many insurance plans (including Medicare, Medicaid, and private plans) limit outpatient visits and residential stays for substance abuse treatment (Sharfstein, Stoline, & Goldman, 1993; Zarkin & Dunlap, 1999). In a study of outpatient admissions for alcohol treatment from New Jersey, type of insurance coverage was associated with the likelihood of dropping out of treatment (Mammo & Weinbaum, 1993).

Organizational Characteristics: Substance Abuse Treatment Programming for Women

Because the tendency of women to leave treatment earlier than men might be attributed to the inability of treatment programs to meet the complex medical, social, and emotional needs of women, research suggests that treatment programming factors need to be included in studies of retention (Roberts & Nishimoto, 1996). Studies have begun to examine the efficacy of substance abuse treatment programming for women and the "active ingredients" in programs that might be associated with improved outcomes among women, including retention (Copeland, Hall, Didcott, & Biggs, 1993; Stark, 1992).

One programmatic characteristic of substance abuse treatment intended to increase access and LOS for women is the provision of ancillary social services, such as child care. Limited access to child care services is one of the most important and frequently cited barriers to treatment among women who seek help (Allen, 1995; Coletti, 1998; Copeland, 1997; Grella, 1997; Wechsberg, Craddock, & Hubbard, 1998).

Studies also have found that programmatic changes to enable women to bring their children into residential treatment were associated with increased LOS (Hughes et al., 1995; Stevens, Arbiter, & Glider, 1989; Wobie et al., 1997). One study suggested that the earlier a mother's infant resides with her in the treatment setting, the longer her stay in treatment will be (Wobie et al., 1997). However, these studies were conducted with small samples of women in residential treatment.

Another characteristic of substance abuse programming associated with increased LOS for women is the provision of women-only treatment (Grella, 1999; Roberts & Nishimoto, 1996), which may foster longer stays in treatment by creating an environment more focused on women's issues. Women-only facilities are more likely than mixed-gender facilities to retain particularly vulnerable women, such as lesbian women, women with a history of physical or sexual abuse, or women who have worked in the sex trade (Copeland & Hall, 1992; Fullilove, Lown, & Fullilove, 1992; Grella, 1997; Pottieger, Inciardi, & Tressell, 1996).

Although these studies highlight factors important to retention for women, more nationally representative data are needed to examine client and organizational characteristics among subgroups of women in multiple types of treatment facilities. The current study increases our understanding of women and substance abuse treatment by examining LOS and experiences specific to females with substance use disorders. It also analyzes data from a large, nationally representative sample of clients in treatment facilities, making the findings more generalizable than those from previous studies, which have been limited to small convenience samples based at single treatment sites. In addition, no prior studies have analyzed a national sample of substance abuse treatment facilities to examine the role of organizational characteristics, such as on-site child care and women-only facilities, in enhancing the LOS of females with substance use disorders.

Methods

This chapter uses data from the Alcohol and Drug Services Study (ADSS), a nationally representative study designed to collect data about substance abuse treatment facilities, clients in treatment, and posttreatment status. The ADSS data, collected from 1996 to 1999, provide an important research opportunity to explore the experiences of women in substance abuse treatment. ADSS examined the characteristics of public and private substance abuse treatment facilities with active programs in all 50 States and the District of Columbia. More than 18,000 facilities were identified from the Substance Abuse and Mental Health Services Administration (SAMHSA) National Facility Register (NFR) and other supplementary independent sources or organizations. For additional information about the methods for ADSS, see Office of Applied Studies (OAS, 2003).

The sampling scheme varied by type of care in order to produce the best estimates of all treatment programs. To ensure reasonably precise and stable estimates of reported variables, hospital inpatient, residential, outpatient methadone, and combined facilities were oversampled relative to their proportional allocation.

For the Phase I survey of treatment programs, a screening telephone interview was administered between October 1996 and April 1997. The screening garnered responses from 2,395 treatment facilities representing 12,387 treatment facilities nationwide. The Phase I questionnaire administered by phone with facility directors collected information concerning each facility's most recent 12–month data on admissions, discharges, and treatment services offered.

At Phase II, the sample of 2,395 Phase I facility respondents was further reduced by eliminating hospital inpatient facilities, facilities in which 100 percent of clients were treated for alcohol use disorders, and facilities outside primary sampling units (PSUs). Of 1,052 remaining Phase I facilities, a target Phase II facility sample was set of 306 facilities, and 280 cooperated with all steps of the Phase II study protocol.² Substance abuse treatment discharges during the most recent 6 months were listed for these 280 facilities, and client discharge records were randomly selected from the client lists. To be included on the list, clients must have spent at least 1 day in treatment or have had at least one outpatient visit. Client-level discharge data were abstracted from facility records for 5,005 clients. A client record abstract form was used during site visits conducted with facility directors or administrators between August 1997 and April 1999. Facilities participating in Phase II data collection were assured that the information they shared was confidential.³

Upon completion of Phase II data collection, no imputation was performed to fill in missing values for key variables. Full sample weights were created. The sum of the full sample weights for each of the client records in the ADSS Phase II client abstract file provides the estimate of 2,229,060 total discharged clients in the population in a 12–month period. For the purpose of the present study, the Phase I and Phase II files were merged.

Sample

Of the 5,005 clients included in ADSS Phase II, 1,334 records were for female substance abuse treatment clients; 95 records were excluded because they indicated that the client was younger than age 18. The analyses used data from the remaining 1,239 records of adult female substance abuse treatment clients aged 18 or older (representing 504,177 clients nationwide). The final sample included females discharged from nonhospital residential treatment, outpatient methadone treatment, outpatient nonmethadone treatment, and combinations of these types of care.⁴

Measures

Measures of LOS and client characteristics were based on information abstracted from client treatment records during Phase II of ADSS. LOS was indicated by the number of days between the client's date of admission and the date of discharge from a substance abuse treatment facility. Exact age at admission was recorded. For descriptive analyses, clients were grouped into six age categories: 18 to 24 years, 25 to 34 years, 35 to 44 years, 45 to 54 years, 55 to 64 years, and 65 years or older. For multivariate analyses, age was treated as a continuous variable, and clients were grouped into three racial groups (white, unknown/not mentioned, and all other races),⁵ three educational groups (less than high school graduate, high school graduate/general equivalence diploma [GED] or higher, and unknown/not mentioned), two referral groups (criminal justice system vs. all other referral sources), and five payment groups (no payment or client self-payment, private health insurance, criminal justice system, Medicaid or Medicare, and other funding). Organizational characteristics were measured using data from Phase I of ADSS. A dichotomous variable was created to distinguish client records at facilities offering child care services from those at facilities not offering such services, as well as facilities serving women only.

Analytic Approach

Based on previous research, we hypothesized that predisposing characteristics of older age, white race, and higher educational attainment, as well as the enabling characteristic of criminal justice referral to treatment, would be associated with longer stay in treatment. We hypothesized that private health insurance, Medicaid, or Medicare payment for treatment would be associated with decreased LOS. We further hypothesized that organizational characteristics of receiving treatment at facilities offering child care services or at women-only facilities would be associated with longer stay in treatment.

One of the major issues confronting studies of treatment retention is how to address differences in LOS among types of care (Roberts & Nishimoto, 1996; Smith, North, & Fox, 1995; Strantz &Welch, 1995). Because many prior studies examining retention have analyzed clients separately by type of care received, descriptive analyses in the present study were stratified by type of care. Multivariate analyses were conducted on the full sample of adult female clients to preserve the large sample. Accordingly, three dummy variables were included in the model indicating treatment at residential facilities, outpatient nonmethadone facilities, or facilities offering a combination of types of care. Inasmuch as outpatient methadone facilities involve much longer stays in treatment than other types of care, and because ADSS collected only a subset of client information from records at outpatient methadone facilities, client records from outpatient methadone treatment facilities were excluded from the multivariate analysis. In addition, because LOS is associated with marital status, having children, presenting substance use disorder at admission, and receiving treatment at facilities offering transportation services (Broome et al., 1999; Fishman & Reynolds, 1999; Friedmann, Lemon, & Stein, 2001; Mammo & Weinbaum, 1993; Mertens & Weisner, 2000; Rowan-Szal et al., 2000; Veach et al., 2000; Williams & Roberts, 1991), the multivariate analysis controlled for these client and organizational characteristics.⁶ Although LOS has not been linked in the research literature with receiving treatment at facilities offering prenatal care or combined substance abuse treatment and mental health services, the multivariate analysis controlled for these organizational characteristics to avoid possible confounding.

Statistical procedures. Descriptive analyses used F tests to compare the distributions of mean LOS between (1) female clients with differing predisposing and enabling characteristics, (2) female clients at facilities offering child care services and female clients at facilities not offering child care services, and (3) clients at women-only facilities and female clients at mixed-gender facilities. Statistically significant differences in LOS were examined using two-tailed Student's t tests. Descriptive analyses of LOS were conducted using sample weights developed by Westat and using WesVar (Westat, Inc., 2000) and jackknife variance estimation.

Correlations between organizational characteristics were calculated to test for multicollinearity of variables prior to modeling and to test assumptions about relationships between these characteristics. Although many organizational characteristics were significantly correlated, Pearson product-moment correlation coefficients were relatively low (range = -0.09 to 0.40), indicating that multicollinearity was not an issue in the analysis. Therefore, all organizational characteristics were included in one model. Survival analysis, specifically Cox's proportional hazard regression (Hosmer & Lemeshow, 1999; Parmar & Machin, 1995), was used to model LOS. Survival analysis considers situations in which (1) a dependent variable represents a time to a terminal event, and (2) the duration of the study is limited in time. In this analysis, the terminal event was discharge from treatment. An observation is considered censored by the end of the study period if the terminal event had not yet occurred or if the observation was followed up to a certain time point, after which there was no further information on the client. For the purpose of this analysis, the time of study was determined as 3,000 days, and observations with a longer stay in treatment than 3,000 days were considered to be censored.

Hazard ratios (HRs) and 95 percent confidence intervals (CIs) were calculated. HR is a comparative measure of the association between exposure (measured in terms of one or more independent variables) and an outcome (e.g., discharge from treatment) over time for a defined population. In Cox's hazard regression analysis, it is assumed that (1) a hazard function is constant over time, and (2) all deviations are random. Under the assumption that a hazard function is constant, it is possible to say that a higher HR suggests that clients in the "exposed" group stay in treatment a shorter time than the comparison group. Furthermore, an HR of less than one indicates a variable that may be protective of leaving treatment early.

The comprehensive model analyzed LOS as a function of predisposing, enabling, and organizational characteristics of interest, controlling for facility type of care; marital status at admission; having children at admission; type of presenting substance use disorder at admission; and treatment at facilities offering prenatal care, transportation, or combined substance abuse treatment and mental health services. Survival analyses were conducted using SUDAAN (Shah, Barnwell, Hunt, & LaVange, 1994) with Taylor series variance estimation. All estimates are weighted.

Results

The characteristics of the sample used in the analyses are presented in Table 6.1. Of the 1,239 adult female substance abuse treatment clients in the ADSS Phase II sample, 12 percent were aged 18 to 24 years, 41 percent were aged 25 to 34 years, and 35 percent were aged 35 to 44 years. Almost 56 percent were white, 31 percent were black, 6 percent were from other racial groups, and 8 percent were of unknown race. At admission, a majority of the sample had only a high school education or less. At admission, most women in the sample were not married, 63 percent had a child/children, 55 percent were unemployed, 28 percent worked full or part time, and 23 percent lived with a spouse/partner. Alcohol was the most common substance of choice specified at admission (33 percent).

**Table 6.1 Characteristics of Female Substance Abuse Treatment Clients Aged 18 or Older at Admission: 1997–1999**
Characteristic	n	%
Age at Admission (years)
18 to 24	146	11.8
25 to 34	506	40.8
35 to 44	437	35.3
45 to 54	116	9.4
55 or older	34	2.7
Race
White	687	55.5
Black	381	30.8
American Indian or Alaskan Native	16	1.3
Asian or Pacific Islander	14	1.1
Other	38	3.1
Unknown/not mentioned	103	8.3
Education at Admission
Less than 8 years	20	1.6
8–11 years	301	24.3
Less than high school graduate, not otherwise specified	40	3.2
High school graduate/GED	358	28.9
Some college	210	17.0
College graduate	41	3.3
Postgraduate	13	1.1
Unknown/not mentioned	60	4.8
Methadone clients (data not collected)	196	15.8
Primary Source of Referral to Treatment
Self-referred/voluntary	367	29.6
Criminal justice system	336	27.1
Welfare office or other social service agencies	139	11.2
Other treatment facility	128	10.3
Health care or mental health providers	123	9.9
Family or friend	66	5.4
Other	18	1.4
Unknown/not mentioned	62	5.0
Primary Source of Payment for Treatment
Medicaid	323	26.1
Client self-payment	293	23.6
Private health insurance	147	11.9
No payment	67	5.4
Criminal justice system	51	4.1
Medicare	19	1.5
Other	192	15.5
Unknown/not mentioned	147	11.9
Marital Status at Admission
Separated/divorced	319	25.8
Never married	295	23.8
Married/common law	218	17.6
Single	143	11.5
Other	30	2.5
Unknown/not mentioned	38	3.1
Methadone clients (data not collected)	196	15.8
Have Child/Children at Admission
Yes	780	63.0
No	178	14.4
Unknown/not mentioned	85	6.9
Methadone clients (data not collected)	196	15.8
Employment at Admission
Unemployed	683	55.1
Full-time (35 hr/wk or more)	194	15.7
Part-time (less than 35 hr/wk)	100	8.1
Employed, not otherwise specified	51	4.1
Keeping house, not otherwise employed	74	6.0
Disabled	51	4.1
Other	22	1.8
Unknown/not mentioned	64	5.2
Living Arrangement at Admission
With spouse/partner	280	22.6
With parent(s)	139	11.2
With other family	120	9.7
Alone	120	9.7
With no other adult(s)/children only	90	7.3
No stable arrangement (include homeless, shelters)	82	6.6
With friends	52	4.2
Correctional facility	9	0.7
Other institution/closed facility	25	2.0
Unknown/not mentioned	126	10.2
Methadone clients (data not collected)	196	15.8
Substance of Choice Specified at Admission
Alcohol	411	33.2
Crack (if unable to separate, combined with cocaine)	175	14.1
Cocaine	111	9.0
Marijuana, hashish, THC	61	4.9
Heroin	48	3.9
Amphetamines (speed, methamphetamine, meth, stimulants, crank)	37	3.0
Other or multiple	44	3.6
No substance of choice	24	1.9
Substance(s) used not specified in record	132	10.7
Methadone clients (data not collected)	196	15.8
Presenting Substance Use Disorder at Admission
Alcohol and drug use	469	37.9
Drug use only (excluding alcohol)	260	21.0
Alcohol use only	259	20.9
Other	6	0.5
Unknown/not mentioned	49	4.0
Methadone clients (data not collected)	196	15.8
GED = general equivalence diploma; HMO = health maintenance organization; PPO = preferred provider organization; THC = tetrahydrocannabinol. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase II client data (1997–1999).

An estimated 14 percent of adult female substance abuse treatment clients received treatment at nonhospital residential facilities, 7 percent received treatment at outpatient methadone facilities, 49 percent received treatment at outpatient nonmethadone facilities, and 29 percent received treatment at combination facilities (Figure 6.1). The average LOS among adult female clients across types of care was 162 days. Among this population, the average LOS was 34 days at nonhospital residential facilities, 531 days at outpatient methadone facilities, 154 days at outpatient nonmethadone facilities, and 147 days at combination facilities (Figure 6.2).

Figure 6.1 Weighted Percentages of Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Facility Type of Care

Figure 6.1 D

Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Figure 6.2 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Facility Type of Care

Figure 6.2 D

Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Retention, by Predisposing Characteristics

Age at admission. Among adult female clients at outpatient nonmethadone facilities, the average LOS differed by age group (F = 10.76, df = 64, p < 0.001) (Table 6.2). Student's t test showed that among adult female clients at outpatient nonmethadone facilities, LOS was shorter among those aged 55 to 64 (mean = 50 days) than among those aged 18 to 54 (mean = 139 to 177 days). Analyses of age groups of adult female clients at nonhospital residential or outpatient methadone facilities could not be conducted because of small sample size in one or more age groups, weights that were zero for all records, or variables that were zero or a linear combination of other variables.

**Table 6.2 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Age at Admission and Facility Type of Care**
Facility Type of Care/Age at Admission	Weighted n	Average LOS	F	p
Total	504,177	161.7
Nonhospital Residential
18 to 24 years	9,360	32.6
25 to 44 years	35,315	37.0
35 to 44 years	22,981	28.2
45 to 54 years	3,042	39.3
55 to 64 years	*	*
65 years or older	*	*
Subtotal	72,325	34.0	----
Outpatient Methadone
18 to 24 years	2,257	733.0
25 to 44 years	14,713	702.6
35 to 44 years	14,439	395.5
45 to 54 years	4,759	296.9
55 to 64 years	*	*
65 years or older	0	n/a
Subtotal	36,269	531.4	----
Outpatient Nonmethadone
18 to 24 years	36,397	162.0
25 to 44 years	100,032	164.3
35 to 44 years	69,666	139.0
45 to 54 years	30,167	177.0
55 to 64 years	11,135	49.5
65 years or older	*	*
Subtotal	247,607	153.8	10.8	<0.001
Combination Facilities
18 to 24 years	16,782	52.6
25 to 44 years	47,022	284.0
35 to 44 years	54,567	120.3
45 to 54 years	25,526	31.7
55 to 64 years	*	*
65 years or older	0	n/a
Subtotal	147,976	146.7	2.03	0.10
Note: Of the total 1,239 female client records among clients aged 18 or older, 1,238 (99.9 percent) contained information about LOS and age at admission. Degrees of freedom: 64. * Low precision; no estimate reported. --- F value cannot be calculated due to small sample size in one or more age groups, weights that are zero for all records, or variables that are zero or a linear combination of other variables. n/a = LOS cannot be calculated due to zero sample size in this age group. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Race. Among adult female clients at outpatient methadone facilities, the average LOS differed by race (F = 4.85, df = 64, p = 0.004) (Table 6.3). However, small sample sizes for American Indian or Alaskan Native adult female clients at outpatient methadone facilities prevented further comparisons between racial groups. The average LOS was similar among racial groups of adult female clients at nonhospital residential (F = 2.06, df = 64, p = 0.11) or outpatient nonmethadone facilities (F = 0.71, df = 64, p = 0.59). Analyses of LOS by race among adult female clients at combination facilities could not be conducted.

**Table 6.3 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Race and Facility Type of Care**
Facility Type of Care/Race	Weighted n	Average LOS	F	p
Total	456,276	159.7
Nonhospital Residential
White	39,110	38.1
Black	22,877	32.9
American Indian or Alaskan Native	0	n/a
Asian or Pacific Islander	*	*
Other	2,909	35.1
Subtotal	65,887	35.7	2.06	0.11
Outpatient Methadone
White	17,772	396.4
Black	9,042	941.7
American Indian or Alaskan Native	*	*
Asian or Pacific Islander	0	n/a
Other	4,356	552.5
Subtotal	31,981	563.0	4.85	0.004
Outpatient Nonmethadone
White	160,620	157.7
Black	54,995	145.1
American Indian or Alaskan Native	3,581	135.1
Asian or Pacific Islander	1,360	114.7
Other	2,646	112.7
Subtotal	223,201	153.5	0.71	0.59
Combination Facilities
White	93,533	68.1
Black	43,537	328.5
American Indian or Alaskan Native	*	*
Asian or Pacific Islander	*	*
Other	0	n/a
Subtotal	143,998	146.1	----
Note: Of the total 1,239 female client records among clients aged 18 or older, 1,135 (91.6 percent) contained information about LOS and race. Degrees of freedom: 64. * Low precision; no estimate reported. --- F value cannot be calculated due to small sample size in one or more racial groups, weights that are zero for all records, or variables that are zero or a linear combination of other variables. n/a = LOS cannot be calculated due to zero sample size in this racial group. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Education at admission. Among adult female clients at outpatient nonmethadone (F = 3.73, df = 56, p = 0.003) or combination facilities (F = 2.67, df = 56, p = 0.02), the average LOS differed by education at admission (Table 6.4). Among adult female clients at outpatient nonmethadone facilities, the LOS was shorter among those with fewer than 8 years of education (mean = 68 days) than among those with higher levels of education (mean = 28 to 173 days). The LOS also was shorter among those with postgraduate education (mean = 85 days) than among those with between 8 and 11 years of education (mean = 154 days) or high school graduates (mean = 173 days). Among adult female clients at combination facilities, the LOS was shorter among college graduates than among those with 8 to 11 years of education or high school graduates (college graduates: 17 days vs. 8 to 11 years: 56 days, t = 5.03, df = 56, p < 0.001; college graduates: 17 days vs. high school graduates: 79 days, t = 4.49, df = 56, p < 0.001).

**Table 6.4 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Education at Admission and Facility Type of Care**
Facility Type of Care/Education at Admission	Weighted n	Average LOS	F	p
Total	430,548	110.7
Nonhospital Residential
Less than 8 years	*	*
8–11 years	23,659	35.4
Less than high school graduate not otherwise specified	0	n/a
High school graduate/GED	29,610	35.7
Some college	15,234	28.3
College graduate	0	n/a
Postgraduate	0	n/a
Other
Subtotal	68,751	34.1	0.12	0.95
Outpatient Nonmethadone
Less than 8 years	2,819	67.5
8–11 years	74,947	154.1
Less than high school graduate not otherwise specified	8,250	132.2
High school graduate/GED	89,535	173.0
Some college	44,701	139.4
College graduate	12,626	127.8
Postgraduate	2,218	84.8
Other	0	n/a
Subtotal	235,096	154.6	3.73	0.003
Combination Facilities
Less than 8 years	*	*
8–11 years	35,168	56.1
Less than high school graduate not otherwise specified	5,421	35.2
High school graduate/GED	37,108	79.4
Some college	33,948	106.3
College graduate	4,455	16.7
Postgraduate	*	*
Other	0	n/a
Subtotal	126,701	70.6	2.67	0.02
Note: Of the total 1,239 female client records among clients aged 18 or older, 196 female methadone client records were excluded; of the remaining 1,043 records, 982 (94.2 percent) contained information about LOS and education at admission. Degrees of freedom: 56. * Low precision; no estimate reported. GED = general equivalence diploma; n/a = LOS cannot be calculated due to zero sample size in this education category. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Retention, by Enabling Characteristics

Primary source of referral. Among adult female clients at nonhospital residential facilities, the average LOS differed by primary source of referral to treatment (F = 2.53, df = 64, p = 0.047) (Table 6.5). Among adult female clients at nonhospital residential facilities, the LOS was longer among women with a criminal justice system referral (mean = 75 days) than among those referred by other sources (mean = 23 to 31 days). The LOS was similar among adult female clients with different referral sources at outpatient nonmethadone (F = 0.74, df = 64, p = 0.57) or combination facilities (F = 0.91, df = 64, p = 0.46). Analyses of the LOS by referral source among adult female clients at outpatient methadone facilities could not be conducted.

**Table 6.5 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Primary Source of Referral and Facility Type of Care**
Facility Type of Care/Primary Source of Referral	Weighted n	Average LOS	F	p
Total	474,299	163.7
Nonhospital Residential
Other treatment facility	13,759	22.7
Criminal justice system	9,742	74.5
Self-referred/voluntary, welfare office or other social service agencies, family, friend, or employer	36,971	30.9
Health care or mental health providers	9,022	24.2
Other	0	n/a
Subtotal	69,495	34.5	2.53	0.047
Outpatient Methadone
Other treatment facility	4,259	628.6
Criminal justice system	*	*
Self-referred/voluntary, welfare office or other social service agencies, family, friend, or employer	27,790	535.8
Health care or mental health providers	*	*
Other	*	*
Subtotal	33,505	539.4	---
Outpatient Nonmethadone
Other treatment facility	13,381	194.6
Criminal justice system	106,740	148.8
Self-referred/voluntary, welfare office or other social service agencies, family, friend, or employer	89,521	150.6
Health care or mental health providers	23,913	193.9
Other	*	*
Subtotal	236,632	157.2	0.74	0.57
Combination Facilities
Other treatment facility	23,562	71.0
Criminal justice system	24,309	99.6
Self-referred/voluntary, welfare office or other social service agencies, family, friend, or employer	79,243	195.1
Health care or mental health providers	6,746	39.7
Other	*	*
Subtotal	134,666	148.1	0.91	0.46
Note: Of the total 1,239 female client records among clients aged 18 or older, 1,176 (94.9 percent) contained information about LOS and referral source. Degrees of freedom: 64. * Low precision; no estimate reported. --- F value cannot be calculated due to small sample size in one or more categories, weights that are zero for all records, or variables that are zero or a linear combination of other variables. n/a = LOS cannot be calculated due to zero sample size in this category. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Primary source of payment. Among adult female clients at outpatient nonmethadone (F = 7.68, df = 64, p < 0.001) or combination facilities (F = 3.35, df = 64, p = 0.01), the LOS differed by primary source of payment for treatment (Table 6.6). Among adult female clients at outpatient nonmethadone facilities, the LOS was longer among those whose primary source of payment was no payment or client self-payment (mean = 139 days) or Medicare or Medicaid (mean = 227 days) than among those whose payment source was private health insurance (mean = 90 days). Among adult female clients at combination facilities, the LOS was longer among those whose payment source was Medicare or Medicaid (mean = 159 days) than among those whose payment source was private health insurance (mean = 27 days).

**Table 6.6 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Primary Source of Payment for Treatment and Facility Type of Care**
Facility Type of Care/Primary Source of Payment	Weighted n	Average LOS	F	p
Total	415,513	169.0
Nonhospital Residential
No payment or client self-payment	18,797	34.9
Private health insurance, fee-for-service, or HMO/PPO/managed care	*	*
Criminal justice system	0	n/a
Medicare or Medicaid	11,184	23.1
Other	32,738	37.1
Subtotal	65,070	33.1	1.62	0.19
Outpatient Methadone
No payment or client self-payment	9,912	921.2
Private health insurance, fee-for-service, or HMO/PPO/managed care	2,339	315.6
Criminal justice system	0	n/a
Medicare or Medicaid	12,631	411.2
Other	5,943	415.7
Subtotal	30,826	568.8	0.57	0.64
Outpatient Nonmethadone
No payment or client self-payment	82,417	138.6
Private health insurance, fee-for-service, or HMO/PPO/managed care	35,212	90.2
Criminal justice system	26,647	118.8
Medicare or Medicaid	41,234	227.0
Other	18,924	267.3
Subtotal	204,435	157.4	7.68	< 0.001
Combination Facilities
No payment or client self-payment	29,657	94.8
Private health insurance, fee-for-service, or HMO/PPO/managed care	34,619	27.0
Criminal justice system	*	*
Medicare or Medicaid	26,712	159.0
Other	19,819	510.3
Subtotal	115,183	159.3	3.35	0.01
Note: Of the total 1,239 female client records among clients aged 18 or older, 1,091 (88.1 percent) contained information about LOS and primary source of payment for treatment. Degrees of freedom: 64. * Low precision; no estimate reported. HMO = health maintenance organization; PPO = preferred provider organization; n/a = LOS cannot be calculated due to zero sample size in this category. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Retention, by Organizational Characteristics

Facilities offering child care services. Among adult female clients at nonhospital residential facilities, the LOS was longer among those at facilities offering child care services than among those at facilities without such services (child care: mean = 97 days vs. no child care: mean = 33 days, t = 4.77, df = 64, p < 0.01) (Table 6.7).

**Table 6.7 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Availability of Child Care Services and Facility Type of Care**
Facility Type of Care	Weighted n		Average LOS		t	p	df
Facility Type of Care	Child Care	No Child Care	Child Care	No Child Care	t	p	df
Nonhospital Residential	1,719	70,606	96.7	32.5	4.77	<0.01	64
Outpatient Methadone	3,731	32,538	386.8	548.0	-1.06	0.29	64
Outpatient Nonmethadone	76,348	171,259	168.4	147.3	0.55	0.58	64
Combination	41,069	106,906	353.0	67.5	1.29	0.20	64
Note: Of the 1,239 female client records among clients aged 18 or older, 1,238 (99.9 percent) contained information about LOS and availability of child care services and facility type. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Women-only facilities. Among adult female clients at nonhospital residential facilities, the LOS was longer among those at women-only facilities than among females at mixed-gender facilities (women-only: mean = 83 days vs. mixed-gender: mean = 22 days, t = 3.90, df = 64, p < 0.01) (Table 6.8). Among adult female clients at combination facilities, the LOS was shorter among those at women-only facilities than among those at mixed-gender facilities (women-only: mean = 52 days vs. mixed-gender: mean = 148 days, t = -2.01, df = 64, p = 0.048).

**Table 6.8 Weighted Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission, by Client Composition of Facility and Facility Type of Care**
Facility Type of Care	Weighted n		Average LOS		t	p	df
Facility Type of Care	Women-Only	Mixed-Gender	Women-Only	Mixed-Gender	t	p	df
Nonhospital Residential	13,956	58,369	83.1	22.3	3.90	<0.01	64
Outpatient Nonmethadone	2,484	245,123	295.5	152.4	0.90	0.37	64
Combination	1,454	146,522	51.9	147.7	-2.01	0.048	64
Note: Of the 1,239 female client records among clients aged 18 or older, 1,238 (99.9 percent) contained information about LOS and facility clientele composition and facility type. Of these, 153 client records from outpatient methadone facilities were excluded. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

Multivariate Analysis

A multivariate analysis showed that several predisposing, enabling, and organizational characteristics were associated with LOS among adult female clients (Table 6.9). Adult females who did not complete high school (HR = 1.32; 95 percent CI = 1.07, 1.63; p = 0.01) or whose primary source of referral was not the criminal justice system (HR = 1.32; 95 percent CI = 1.02, 1.70; p = 0.04) were more likely to leave treatment earlier than high school graduates or those referred by the criminal justice system. Adult female clients whose primary source of payment was other than private health insurance stayed in treatment longer than those whose primary source of payment was private health insurance. Adult females receiving treatment at facilities offering child care services (HR = 0.51; 95 percent CI = 0.36, 0.73; p = 0.0003) or at women-only facilities (HR = 0.34; 95 percent CI = 0.13, 0.89; p = 0.03) stayed in treatment longer than those at facilities without child care services or at mixed-gender facilities. These multivariate analysis results are similar to descriptive analysis findings, although multivariate analyses were conducted using a reduced sample (n = 962) because of missing data for one or more covariates. Age at admission and race were not associated with LOS among adult female clients in multivariate analyses. Some control variables also were significantly associated with LOS among women discharged from nonhospital residential, outpatient nonmethadone, or combination facilities. Women at nonhospital residential or combination facilities were more likely to leave treatment earlier than those at outpatient nonmethadone facilities. Adult female clients at facilities offering combined substance abuse treatment and mental health services stayed in treatment longer than those at facilities not offering these services. Contrary to expectations, receiving treatment at facilities offering prenatal care or transportation services was associated with leaving treatment earlier among women.

Discussion

This study examined the association between predisposing characteristics (age, race, and education), enabling characteristics (referral source and payment source), and organizational characteristics (availability of child care services and women-only facilities) and LOS among adult female substance abuse treatment clients. Consistent with findings from another study (Knight et al., 2001), adult females who did not complete high school were more likely to leave treatment earlier than high school graduates. Adult females referred by sources other than the criminal justice system were more likely to leave treatment earlier than those referred by the criminal justice system, which is similar to findings by Green et al. (2002). Adult female clients whose primary source of payment was private health insurance left treatment earlier than those with other payment sources. As noted in other studies (Hughes et al., 1995; Stevens et al., 1989; Wobie et al., 1997), women at facilities offering child care services stayed in treatment longer than women at facilities without such services. Adult females receiving treatment at women-only facilities were retained in treatment longer than those at mixed-gender facilities, which is consistent with the findings of other studies (Grella, 1999; Roberts & Nishimoto, 1996). In an unanticipated finding, age at admission and race were not associated with LOS among adult female clients, despite previous research that has linked these two client characteristics to LOS (Grella et al., 2000; McCaul et al., 2001; Mertens & Weisner, 2000; Strantz & Welch, 1995; Wickizer et al., 1994).

The results of the multivariate analysis support the positive effect of on-site child care and women-only treatment on LOS of women in substance abuse treatment. However, the results of the descriptive analysis suggest that this finding may only be relevant to women in nonhospital residential treatment. It is important to note that certain factors make it difficult to estimate the association between child care and LOS. For example, child care may be bundled with other services, such as relapse prevention groups and positive contingency awards for abstinence (Carroll, Chang, Behr, Clinton, & Kosten, 1995), or it may be associated with other confounders, such as the availability of social workers, physicians, and nurse practitioners to clients. In addition, facilities offering child care services or women-only facilities may be more likely than other facilities to offer parenting and women's issues groups or to serve clients with different characteristics, and these factors may result in longer LOS. Furthermore, although receiving treatment in women-only facilities was associated with LOS in the multivariate analysis and among women in nonhospital residential facilities in the descriptive analysis, among women in combination facilities, LOS was shorter in women-only facilities compared with mixed-gender facilities. The cross-sectional design of this study does not allow for conclusions about the direct causal effects on LOS of such factors as availability of child care or women-only facilities because of "confounding by indication" (Walker, 1996); that is, the clients were not randomly assigned to facilities and therefore may have made their choice of facilities based on their different attitudes toward treatment or child care, different personality traits, or other factors.

Table 6.9 Weighted Cox's Proportional Hazard Regression Analysis of Length of Stay (LOS) among Female Substance Abuse Treatment Clients Aged 18 or Older at Admission Discharged from Nonhospital Residential Facilities, Outpatient Nonmethadone Facilities, or Combination Facilities
Independent Variables	HR (95% CI)	df	Adj df	p
Age at Admission	1.00 (0.99, 1.01)	1	1.00	0.64
Race (compared with white)		2	1.68
All other races	0.97 (0.71, 1.33)			0.86
Unknown/not mentioned	1.26 (0.89, 1.77)			0.19
Education at Admission (compared with high school graduate/GED or more)		2	1.81
Less than high school graduate	1.32 (1.07, 1.63)			0.01
Unknown/not mentioned	0.95 (0.65, 1.39)			0.78
Primary Source of Referral for Treatment (compared with criminal justice system)		1	1.00
Other sources	1.32 (1.02, 1.70)			0.04
Primary Source of Payment for Treatment (compared with private health insurance, fee-for-service or HMO/PPO/managed care)		4	3.78
No payment or client self-payment	0.67 (0.46, 0.97)			0.03
Medicare/Medicaid	0.54 (0.38, 0.78)			0.001
Criminal justice system	0.51 (0.31, 0.83)			0.007
Other funding	0.74 (0.47, 1.17)			0.20
Child Care Services (compared with services not offered)		1	1.00
Offered	0.51 (0.36, 0.73)			0.0003
Facility Clientele Composition (compared with mixed-gender)		1	1.00
Women-only	0.34 (0.13, 0.89)			0.03
Marital Status at Admission (compared with never married, widowed, separated/divorced, or single)		1	1.00
Married/common law	1.11 (0.87, 1.42)			0.40
Have Child/Children at Admission (compared with no child/children)		2	1.94
Unknown/not mentioned	1.15 (0.74, 1.78)			0.52
Have child/children	0.83 (0.63, 1.09)			0.18
Presenting Substance Use Disorder at Admission (compared with alcohol abuse only)		3	2.03
Drug abuse only	0.83 (0.65, 1.04)			0.10
Alcohol and drug abuse	0.77 (0.60, 1.00)			0.05
Substance not specified	0.51 (0.26, 0.98)			0.04
Facility Type of Care (compared with outpatient nonmethadone)		2	2.00
Nonhospital residential	4.39 (2.62, 7.35)			<0.0001
Combination	2.65 (1.84, 3.82)			<0.0001
Prenatal Care Services (compared with services not offered)		1	1.00
Offered	1.48 (1.09, 2.02)			0.01
Transportation Services (compared with services not offered)		1	1.00
Offered	1.61 (1.11, 2.34)			0.01
Combined Substance Abuse Treatment and Mental Health Services (compared with services not offered)		1	1.00
Offered	0.61 (0.47, 0.80)			0.0005
Number of observations used in the analysis: 962. Weighted count: 417,592. GED = general equivalence diploma; HMO = health maintenance organization; HR = hazard ratio; PPO = preferred provider organization. Source: SAMHSA, Office of Applied Studies, Alcohol and Drug Services Study (ADSS), Phase I data (1996–1997) and Phase II data (1997–1999).

The extent to which data were missing data was a major limitation of this study. The ADSS Phase II data were abstracted retroactively from client records, and limited data were collected about methadone clients. Missing data imposed limitations on conducting descriptive, inferential, and multivariate analyses. The survival analysis procedure eliminated all records in which any variable in the model included a missing value. The number of observations eliminated due to missing values progressively increased as the number of variables included in the model increased. It is possible that multicollinearity among some covariates may have affected our results. Although we tested for multicollinearity among organizational characteristics, we did not evaluate all covariates. These limitations may have reduced the statistical significance of some variables as correlates of LOS. In addition, many relevant variables could not be included in our multivariate analysis because of missing data, and not controlling for these factors may have caused some covariates to appear statistically significant that should not have.

In descriptive and inferential analyses, the lack of a statistically significant difference between two means may be misleading if extreme values in a small number of records skewed the means, resulting in larger standard errors. In addition, the potentially nonlinear relationship between age and LOS suggested by some descriptive analyses may explain why age was not correlated with LOS in the multivariate analysis.

This study explored the association between a set of single factors on retention. The multivariate analysis did not explicitly examine interactions between variables. However, addressing single separate factors was a necessary first step in exploring the relationships between study variables and LOS using a nationally representative sample of adult female substance abuse treatment clients.

The finding that adult female clients whose primary source of payment was private health insurance remained in treatment for shorter LOS than those with other payment sources suggests that insurance coverage for substance abuse treatment, which likely includes caps on coverage, has important implications for LOS. It is possible that women with insurance coverage do not leave treatment early of their own accord, but rather due to funding restrictions. However, we did not adjust for severity or "need" in our analyses because of data limitations. Also, it is possible that those who have private coverage are less severely impaired (possibly because they have some attachment to the labor force) and did not require as long a stay as the average client. This issue requires further examination of the role that private health insurance plays in women's reasons for leaving treatment and the effects that leaving treatment has on outcomes.

Providing on-site child care appears to be beneficial to women in substance abuse treatment. This finding may have implications for treatment providers, as well as for child welfare policy and the courts making custody and permanency decisions. Child welfare agencies and juvenile dependency courts are challenged to address clients' substance use disorders in order to achieve positive outcomes for children and families. Prior research has shown that unless the parent with a substance use disorder has been engaged in a treatment program or is otherwise moving into recovery, the child's prognosis for long-term emotional, social, and physical well-being is poor. A mother seeking substance abuse treatment may need to utilize the child care services at a substance abuse treatment facility. This may make the difference as to whether she seeks treatment, stays in treatment, and completes treatment. Child care has specifically been identified as a service that is needed to support child welfare clients' participation in substance abuse treatment (Semidei, Radel, & Nolan, 2001). Furthermore, the Adoption and Safe Families Act (ASFA) of 1997 has shortened the timeline for parents with substance use disorders whose children have been removed for abuse or neglect to seek treatment and stabilize their lives before termination of parental rights occurs. Therefore, offering child care services at treatment facilities, because of its association with increased LOS and improved outcomes, may be essential if families are to be afforded real opportunities for recovery within ASFA timelines.

Future research using targeted data collection approaches and large samples may be able to shed more light on the role of gender in LOS. These studies could help to identify a profile of women with children who would benefit most from treatment in the facilities providing on-site child care, as well as factors influencing the LOS of these women in treatment. More studies are needed that focus on facility characteristics. Full consideration will need to be given to the specifics of the therapeutic programs and their objectives that foster longer stays in treatment. In addition, more detailed service utilization measures are needed, such as the number of individual therapy visits, group therapy visits, and family counseling visits. Overall, the findings of the current study suggest that addressing women's specialized substance abuse treatment needs through offering of child care services or providing women-only treatment may lead to longer LOS and better substance abuse treatment outcomes.

References

Aday, L. A., Begley, C. E., Lairson, D. R., Slater, C. H., Richard, A. J., & Montoya, I. D. (1999). A framework for assessing the effectiveness, efficiency, and equity of behavioral healthcare. American Journal of Managed Care, 5 Special Number, SP25–SP44.

Adoption and Safe Families Act, Pub. L. No. 105–89 (1997).

Allen, K. (1995). Barriers to treatment for addicted African-American women. Journal of the National Medical Association, 87, 751–756.

Arfken, C. L., Klein, C., di Menza, S., & Schuster, C. R. (2001). Gender differences in problem severity at assessment and treatment retention. Journal of Substance Abuse Treatment, 20, 53–57.

Ashley, O. S., Marsden, M. E., & Brady, T. M. (2003). Effectiveness of substance abuse treatment programming for women: A review. American Journal of Drug and Alcohol Abuse, 29, 19–53.

Baekeland, F., & Lundwall, L. (1975). Dropping out of treatment: A critical review. Psychological Bulletin, 82, 738–783.

Bartholomew, N. G., Rowan-Szal, G. A., Chatham, L. R., & Simpson, D. D. (1994). Effectiveness of a specialized intervention for women in a methadone program. Journal of Psychoactive Drugs, 26, 249–255.

Broome, K. M., Flynn, P. M., & Simpson, D. D. (1999). Psychiatric comorbidity measures as predictors of retention in drug abuse treatment programs. Health Services Research, 34, 791–806.

Carroll, K. M., Chang, G., Behr, H., Clinton, B., & Kosten, T. R. (1995). Improving treatment outcome in pregnant, methadone-maintained women: Results from a randomized clinical trial. American Journal on Addictions, 4(1), 56–59.

Clark, H. W. (2001). Residential substance abuse treatment for pregnant and postpartum women and their children: Treatment and policy implications. Child Welfare, 80(2), 179–198.

Coletti, S. D. (1998). Service providers and treatment access issues. In C. L. Wetherington & A. B. Roman ( Eds.), Drug addiction research and the health of women (pp. 236–244, NIH Publication No. 98–4290). Rockville, MD: National Institute on Drug Abuse.

Comfort, M., Loverro, J., & Kaltenbach, K. (2000). A search for strategies to engage women in substance abuse treatment. Social Work in Health Care, 31(4), 59–70.

Condelli, W. S. (1994). Domains of variables for understanding and improving retention in therapeutic communities. International Journal of the Addictions, 29, 593–607.

Copeland, J. (1997). A qualitative study of barriers to formal treatment among women who self-managed change in addictive behaviours. Journal of Substance Abuse Treatment, 14, 183–190.

Copeland, J., & Hall, W. (1992). A comparison of women seeking drug and alcohol treatment in a specialist women's and two traditional mixed-sex treatment services. British Journal of Addiction, 87, 1293–1302.

Copeland, J., Hall, W., Didcott, P., & Biggs, V. (1993). A comparison of a specialist women's alcohol and other drug treatment service with two traditional mixed-sex services: Client characteristics and treatment outcome. Drug and Alcohol Dependence, 32, 81–92.

De Leon, G., Melnick, G., & Kressel, D. (1997). Motivation and readiness for therapeutic community treatment among cocaine and other drug abusers. American Journal of Drug and Alcohol Abuse, 23, 169–189.

Egelko, S., Galanter, M., Dermatis, H., & DeMaio, C. (1998). Evaluation of a multisystems model for treating perinatal cocaine addiction. Journal of Substance Abuse Treatment, 15, 251–259.

Fishman, J., Reynolds, T., & Riedel, E. (1999). A retrospective investigation of an intensive outpatient substance abuse treatment program. American Journal of Drug and Alcohol Abuse, 25, 185–196.

Friedmann, P. D., Lemon, S. C., & Stein, M. D. (2001). Transportation and retention in outpatient drug abuse treatment programs. Journal of Substance Abuse Treatment, 21, 97–103.

Fullilove, M. T., Lown, E. A., & Fullilove, R. E. (1992). Crack 'hos and skeezers: Traumatic experiences of women crack users. Journal of Sex Research, 29, 275–287.

Gottheil, E., McLellan, A. T., & Druley, K. A. (1992). Length of stay, patient severity and treatment outcome: Sample data from the field of alcoholism. Journal of Studies on Alcohol, 53, 69–75.

Green, C. A., Polen, M. R., Dickinson, D. M., Lynch, F. L., & Bennett, M. D. (2002). Gender differences in predictors of initiation, retention, and completion in an HMO-based substance abuse treatment program. Journal of Substance Abuse Treatment, 23, 285–295.

Grella, C. E. (1997). Services for perinatal women with substance abuse and mental health disorders: The unmet need. Journal of Psychoactive Drugs, 29, 67–78.

Grella, C. E. (1999). Women in residential drug treatment: Differences by program type and pregnancy. Journal of Health Care for the Poor and Underserved, 10, 216–229.

Grella, C. E., Anglin, M. D., Wugalter, S. E., Rawson, R., & Hasson, A. (1994). Reasons for discharge from methadone maintenance for addicts at high risk of HIV infection or transmission. Journal of Psychoactive Drugs, 26, 223–232.

Grella, C. E., Joshi, V., & Hser, Y. I. (2000). Program variation in treatment outcomes among women in residential drug treatment. Evaluation Review, 24, 364–383.

Haller, D. L., Miles, D. R., & Dawson, K. S. (2002). Psychopathology influences treatment retention among drug-dependent women. Journal of Substance Abuse Treatment, 23, 431–436.

Hosmer, D. Jr., & Lemeshow, S. (1999). Applied survival analysis: Regression modeling of time to event data. New York: Wiley.

Hser, Y. I., Polinsky, M. L., Maglione, M., & Anglin, M. D. (1999). Matching clients' needs with drug treatment services. Journal of Substance Abuse Treatment, 16, 299–305.

Hubbard, R. L., Craddock, S. G., Flynn, P. M., Anderson, J., & Etheridge, R. M. (1997). Overview of 1–year follow-up outcomes in the Drug Abuse Treatment Outcome Study (DATOS). Psychology of Addictive Behaviors, 11, 261–278.

Hughes, P. H., Coletti, S. D., Neri, R. L., Urmann, C. F., Stahl, S., Sicilian, D. M., & Anthony, J. C. (1995). Retaining cocaine-abusing women in a therapeutic community: The effect of a child live-in program. American Journal of Public Health, 85, 1149–1152.

Joe, G. W., Simpson, D. D., & Broome, K. M. (1999). Retention and patient engagement models for different treatment modalities in DATOS. Drug and Alcohol Dependence, 57, 113–125.

Kelly, P. J., Blacksin, B., & Mason, E. (2001). Factors affecting substance abuse treatment completion for women. Issues in Mental Health Nursing, 22, 287–304.

Kingree, J. B. (1995). Understanding gender differences in psychosocial functioning and treatment retention. American Journal of Drug and Alcohol Abuse, 21, 267–281.

Knight, D. K., Logan, S. M., & Simpson, D. D. (2001). Predictors of program completion for women in residential substance abuse treatment. American Journal of Drug and Alcohol Abuse, 27, 1–18.

Lang, M. A., & Belenko, S. (2000). Predicting retention in a residential drug treatment alternative to prison program. Journal of Substance Abuse Treatment, 19, 145–160.

Luchansky, B., He, L., Krupski, A., & Stark, K. D. (2000). Predicting readmission to substance abuse treatment using state information systems: The impact of client and treatment characteristics. Journal of Substance Abuse, 12, 255–270.

Maglione, M., Chao, B., & Anglin, D. (2000). Residential treatment of methamphetamine users: Correlates of drop-out from the California Alcohol and Drug Data System (CADDS), 1994–1997. Addiction Research, 8(1), 65–79.

Mammo, A., & Weinbaum, D. F. (1993). Some factors that influence dropping out from outpatient alcoholism treatment facilities. Journal of Studies on Alcohol, 54, 92–101.

McCaul, M. E., Svikis, D. S., & Moore, R. D. (2001). Predictors of outpatient treatment retention: Patient versus substance use characteristics. Drug and Alcohol Dependence, 62, 9–17.

McComish, J. F., Greenberg, R., Kent-Bryant, J., Chruscial, H. L., Ager, J., Hines, F., & Ransom, S. B. (1999). Evaluation of a grief group for women in residential substance abuse treatment. Substance Abuse: Official Publication of the Association for Medical Education and Research in Substance Abuse, 20(1), 45–58.

McCusker, J., Bigelow, C., Luippold, R., Zorn, M., & Lewis, B. F. (1995). Outcomes of a 21–day drug detoxification program: Retention, transfer to further treatment, and HIV risk reduction. American Journal of Drug and Alcohol Abuse, 21, 1–16.

McKay, J. R., Alterman, A. I., McLellan, A. T., & Snider, E. C. (1994). Treatment goals, continuity of care, and outcome in a day hospital substance abuse rehabilitation program. American Journal of Psychiatry, 151, 254–259.

Mertens, J. R., & Weisner, C. M. (2000). Predictors of substance abuse treatment retention among women and men in an HMO. Alcoholism: Clinical and Experimental Research, 24, 1525–1533.

Moos, R. H., Brennan, P. L., & Mertens, J. R. (1994a). Diagnostic subgroups and predictors of one-year re-admission among late- middle-aged and older substance abuse patients. Journal of Studies on Alcohol, 55, 173–183.

Moos, R. H., Mertens, J. R., & Brennan, P. L. (1994b). Rates and predictors of four-year readmission among late-middle-aged and older substance abuse patients. Journal of Studies on Alcohol, 55, 561–570.

Nishimoto, R. H., & Roberts, A. C. (2001). Coercion and drug treatment for postpartum women. American Journal of Drug and Alcohol Abuse, 27, 161–181.

Office of Applied Studies. (2003). Alcohol and Drug Services Study (ADSS): Methodology report: Phases I, II, and III. Rockville, MD: Substance Abuse and Mental Health Services Administration. [Available as a PDF at http://www.oas.samhsa.gov/adss.htm]

Parmar, M., & Machin, D. (1995). Survival analysis: A practical approach. New York: Wiley.

Pettinati, H. M., Belden, P. P., Evans, B. D., Ruetsch, C. R., Meyers, K., & Jensen, J. M. (1996). The natural history of outpatient alcohol and drug abuse treatment in a private healthcare setting. Alcoholism: Clinical and Experimental Research, 20, 847–852.

Pottieger, A. E., Inciardi, J. A., & Tressell, P. A. (1996, August). Barriers to treatment entry for women crack users. Paper presented at the 91^st Annual Meeting of the American Sociological Association, New York.

Public Health Service Act, 42 U.S.C. § 241 (d) (1999).

Roberts, A. C., & Nishimoto, R. H. (1996). Predicting treatment retention of women dependent on cocaine. American Journal of Drug and Alcohol Abuse, 22, 313–333.

Rowan-Szal, G. A., Joe, G. W., & Simpson, D. D. (2000). Treatment retention of crack and cocaine users in a national sample of long term residential clients. Addiction Research, 8(1), 51–64.

Sayre, S. L., Schmitz, J. M., Stotts, A. L., Averill, P. M., Rhoades, H. M., & Grabowski, J. J. (2002). Determining predictors of attrition in an outpatient substance abuse program. American Journal of Drug and Alcohol Abuse, 28, 55–72.

Semidei, J., Radel, L. F., & Nolan, C. (2001). Substance abuse and child welfare: Clear linkages and promising responses. Child Welfare, 80(2), 109–128.

Shah, B. V., Barnwell, B. G., Hunt, P. N., & LaVange, L. M. (1994). SUDAAN user's manual: Release 6.4. Research Triangle Park, NC: Research Triangle Institute.

Sharfstein, S. S., Stoline, A. M., & Goldman, H. H. (1993). Psychiatric care and health insurance reform. American Journal of Psychiatry, 150, 7–18.

Simpson, D. D., Joe, G. W., & Broome, K. M. (2002). A national 5–year follow-up of treatment outcomes for cocaine dependence. Archives of General Psychiatry, 59, 538–544.

Simpson, D. D., Joe, G. W., Broome, K. M., Hiller, M. L., Knight, K., & Rowan-Szal, G. A. (1997a). Program diversity and treatment retention rates in the Drug Abuse Treatment Outcome Study (DATOS). Psychology of Addictive Behaviors, 11, 279–293.

Simpson, D. D., Joe, G. W., Fletcher, B. W., Hubbard, R. L., & Anglin, M. D. (1999). A national evaluation of treatment outcomes for cocaine dependence. Archives of General Psychiatry, 56, 507–514.

Simpson, D. D., Joe, G. W., & Rowan-Szal, G. A. (1997b). Drug abuse treatment retention and process effects on follow-up outcomes. Drug and Alcohol Dependence, 47, 227–235.

Smith, E. M., North, C. S., & Fox, L. W. (1995). Eighteen-month follow-up data on a treatment program for homeless substance abusing mothers. Journal of Addictive Diseases, 14(4), 57–72.

Stark, M. J. (1992). Dropping out of substance abuse treatment: A clinically oriented review. Clinical Psychology Review, 12, 93–116.

Stevens, S., Arbiter, N., & Glider, P. (1989). Women residents: Expanding their role to increase treatment effectiveness in substance abuse programs. International Journal of the Addictions, 24, 425–434.

Strantz, I. H., & Welch, S. P. (1995). Postpartum women in outpatient drug abuse treatment: Correlates of retention/completion. Journal of Psychoactive Drugs, 27, 357–373.

Veach, L. J., Remley, T. P. Jr., Kippers, S. M., & Sorg, J. D. (2000). Retention predictors related to intensive outpatient programs for substance use disorders. American Journal of Drug and Alcohol Abuse, 26, 417–428.

Walker, A. M. (1996). Confounding by indication. Epidemiology, 7, 335–336. Comment in Epidemiology, 8, 110–111, 219–220, 335. Comment in Epidemiology, 7, 434–436.

Wechsberg, W. M., Craddock, S. G., & Hubbard, R. L. (1998). How are women who enter substance abuse treatment different than men? A gender comparison from the Drug Abuse Treatment Outcome Study (DATOS). Drugs & Society, 13(1/2), 97–115.

Westat, Inc. (2000). WesVarTM 4.0 user's guide. Rockville, MD: Author.

Westreich, L., Heitner, C., Cooper, M., Galanter, M., & Guedj, P. (1997). Perceived social support and treatment retention on an inpatient addiction treatment unit. American Journal on Addictions, 6, 144–149.

Wickizer, T., Maynard, C., Atherly, A., Frederick, M., Koepsell, T., Krupski, A., & Stark, K. (1994). Completion rates of clients discharged from drug and alcohol treatment programs in Washington State. American Journal of Public Health, 84, 215–221.

Williams, M. T., & Roberts, C. S. (1991). Predicting length of stay in long-term treatment for chemically dependent females. International Journal of the Addictions, 26, 605–613.

Wobie, K., Eyler, F. D., Conlon, M., Clarke, L., & Behnke, M. (1997). Women and children in residential treatment: Outcomes for mothers and their infants. Journal of Drug Issues, 27, 585–606.

Zarkin, G. A., & Dunlap, L. J. (1999). Implications of managed care for methadone treatment: Findings from five case studies in New York State. Journal of Substance Abuse Treatment, 17, 25–35.