Patient and Program Costs, and Outcomes, of Including Gender-Sensitive Services in Intensive Inpatient Programs for Substance Use

Abstract

Gender-sensitive services (GSS) attempt to make substance use treatment better for women, but at what cost and with what results? We sought answers to these questions in a federally-funded study by measuring separately the patient and provider costs of adding GSS, outcomes, and cost-outcome relationships for 12 mixed-gender intensive inpatient programs (IIP) that varied in amounts and types of GSS. GSS costs to female inpatients included time devoted to GSS and expenses for care of dependents while in the IIP. GSS costs to providers included time spent with patients, indirect services, treatment facilities, equipment, and materials. Offering more GSS was expected to consume more patient and provider resources. Offering more GSS also was expected to enhance outcomes and cost-outcome relationships. We found that average GSS costs to patients at the IIPs were $585 ($515 to $656) per patient. Average GSS costs to providers at the IIPs were $344 ($42 to $544) per patient. GSS costs to patients significantly exceeded GSS costs to providers. Contrary to previous research, offering more GSS services to patients did not result in significantly higher costs to patients or providers. IIPs offering more GSS may have delivered fewer traditional services, but this did not significantly affect outcomes, i.e., days until returning to another substance use treatment. In fact, median cost-outcome for these IIPs was a promising 35 treatment-free days, i.e., over a month, per $100 of GSS resources used by patients and providers.

Substance use disorders (SUD) can be especially detrimental to women because, relative to men, women with SUD develop more health problems, more psychiatric disorders, and exhibit a faster course of addiction and higher risk of death (Hersen, Turner, & Beidel, 2007; Navajits, Rosier, Nolan, & Freeman, 2007; Westermeyer & Boedicker, 2000; Wetherington, 2007). Women with SUD are less likely than men with SUD to enter treatment (Greenfield et al., 2007). Women also may delay entry longer after initial drug use, possibly reducing treatment effectiveness (Copeland& Hall, 1992; Hersen et al., 2007). Contributing to this delay may be a greater sense of responsibility for child-rearing (Brady & Ashley, 2005) and greater concern about losing custody of children (Grella, Ponlinsky, Hser, & Perry, 1999). Women with SUD also can be more challenging to treat than men with SUD and may not show as much improvement (French et al., 2002). These and other factors can reduce the effectiveness of women’s SUD treatments, which typically were developed for men (White, 1998).

For women, substance use not only may receive later treatment but also may be initiated and maintained by different psychological processes. Ashley, Marsden, and Brady (2003), Bepko (1991), and Brady and Ashley (2005) hypothesized that the greater powerlessness of women in male-dominated societies could motivate women to use drugs to gain illusionary control over peo- ple and experiences that were denied them in reality, offering temporary escape from memories of emotional, physical, and sexual abuse. These memories may be extremely common for women entering SUD treatment: 96% have experienced emotional abuse, 79% have been abused physically, and 51% have been abused sexually (Grupp, 2006).

Gender-Sensitive Services

Recognizing the potentially unique needs of female substance users, in 1984 the U.S. gov- ernment required that at least 5% of funding for new alcohol and drug use services be devoted to treatment of women. Agencies such as the National Institute of Drug Abuse (NIDA) have supported model programs to provide gender-sensitive services (GSS) (cf. Grella et al., 1999). Researchers and clinicians have collaborated to design GSS-focused treatments for SUD (e.g., Ettore, 2004; French, McCollister, Cacciola, Durell, & Stephens, 2002; Greenfield et al., 2007; Sun, 2006), often building on gender-sensitive services developed for depression, post-traumatic stress disorder, attention deficit hyperactivity disorder, general medical care, and rehabilitation within the criminal justice system (Bloom, Owen, & Covington, 2004; Cochran & Rabinowitz, 2003; Covington, 2007; Quinn, 2005; Salgado, Vogt, King, & King, 2002; Vlassoff & Moreno, 2002). These GSS for SUD have included:

• coping skills training,

• support of women’s self-care and self-esteem (including separate, private, safe bathrooms),

• addressing reoccurring psychiatric problems in treatment,

• self-efficacy education,

• gynecological and other health services,

• on-site childcare,

• supportive and other empowerment methods,

• women-only groups,

• family planning,

• parent training,

• vocational training, and

• trauma-focused services.

Outcome of GSS for Substance Abusing Women

As with most human services, research on GSS has focused more on outcomes than costs (cf. Yates, 1994). For example, Copeland, Hall, Didicott, and Biggs (1993) compared the outcomes of a “specialist women’s service” and “traditional mixed-sex service” for substance use. The “specialist women’s service” added childcare and a female-only staff to traditional treatment. These additions did not significantly affect drug use, social support, severity of depression, or self-esteem six months following treatment. Copeland and colleagues did, however, find that women with dependent children, with same-sex partners, or with a history of childhood sexual assault were significantly more likely to stay in programs offering GSS (Copeland & Hall, 1992; see also Claus et al., 2007).

Gender balance of patients in a treatment program also can affect outcomes of substance use treatment. For example, Niv and Hser (2006) compared outcomes for females in women-only versus mixed-gender programs, both of which included GSS such as child and family services. Women receiving GSS in the women-only program used more services while in treatment but, for the 9 months following treatment, reported significantly less drug use and were less likely to be arrested. Also, Greenfield, Trucco, McHugh, Lincoln, and Gallop (2007) found that women receiving GSS in women-only groups reported significantly (a) higher satisfaction with treatment,(b) greater decreases in psychiatric symptoms during and after treatment, and (c) less drug use 6 months after treatment, relative to women receiving GSS in mixed-gender groups.

Costs of Gender-Sensitive Services

Extending previous research on GSS, the current study reports costs, outcomes, and cost-outcome relationships of adding GSS to mixed-gender treatment in intensive inpatient (IIP) settings. In one of the few prior studies of GSS costs, French et al. (2002) found that residential gender-sensitive treatment for pregnant and parenting women was substantially more costly than standard residential treatment, with an average total treatment cost of $8,035 versus $1,467 per patient. More recently, Yeom and Shepard (2007) found higher costs for outpatient services provided to women relative to men in an outpatient program, due in part to greater severity of substance use before treatment. We expected similar findings for GSS offered at the IIPs we studied.

Treating women should cost more for several reasons. For example, inpatient programs may find it difficult to admit and retain a sufficient number of women to offer women-only groups of sizes similar to those offered to men or to both genders. Smaller women-only groups would, then, be expected to result in each female patient consuming a higher proportion of the group leader’s time and of the group meeting space. This would raise the cost per member relative to male-only or mixed-gender groups. Costs of GSS also should be higher in inpatient settings if costs of separate women-only bath- and bedrooms could not be distributed over enough women to keep facilities costs per woman similar to facilities costs per man. We expected that these and other factors would result in relatively more provider and facilities resources being devoted to treatment of women than to treatment of men in mixed-gender inpatient programs.

We also measured the types, amounts, and monetary values of resources devoted to adding GSS to treatment by patients as well as providers in our cost study, in recognition of contributions of patients to treatment and as recommended by the NIDA manual on evaluating and improving cost, cost-effectiveness, and cost-benefit of SUD treatments (Yates, 1999). This is not intended to provide a comprehensive societal perspective on total costs of treatments that include GSS. Instead we thought it potentially useful to capture patient resources required by GSS separately from provider resources required by GSS. Patient resources can include time spent receiving services, time spent traveling to and from treatment sites, transportation to and from those sites, and dependent care required while patient caregivers participate in treatment (cf. Yates, 1980a, 1996). The amount of these patient resources may be higher when GSS is enhanced in treatment and may represent increased barriers to GSS for patients: we sought to measure them to find out. Assigning monetary value to these patient resources allows the value of those resources to be compared to already- monetized provider resources consumed by GSS. This comparison could reveal whether patients or providers devote more resources when GSS are increased, or whether there is possibly an inverse relationship between patient and provider resources consumed when GSS is increased.

Cost-Outcome Relationship After Adding GSS to SUD Treatment

Inclusion of costs and outcomes in the same investigation of SUD treatment has been especially rare, with the notable exceptions of Barnett and Swindle (1997), French and colleagues (e.g., French, Dunlap, Zarkin, McGeary, & McLellan, 1997), and Mannix (2010). French et al. (2002) found a benefit/cost ratio of 3:1 for gender-sensitive treatment for pregnant and parenting women by comparing self-reported use of health and other services after versus before treatment. In addition to examining costs, Yeom and Shepard (2007) found that an outpatient substance use treatment expended more resources for its female patients while achieving no greater effectiveness, defined as no self-reported drug use during the 6 months following treatment. Of course, that was only one program and it was outpatient: GSS offered in other programs could lead to better out-comes for female patients — but, it can be asked, at what additional cost, especially if the treatment program has to accommodate residential needs of women as well as men?

Hypotheses

The present study examined GSS in IIPs because of the individually-focused, labor-inten- sive nature of GSS in residential settings. Attempting to augment findings of Tang, Claus, Orwin, Kissin, and Ariera (2012), we examined GSS costs, outcomes, outcome/cost, and cost/outcome ratios in 12 mixed-gender IIPs that differed in GSS offered. We expected that offering more GSS in mixed-gender IIPs would:

1. increase costs for female patients

2. increase costs for providers,

3. improve outcomes for female patients, and

4. improve cost-outcome relationships for female patients,

in that more gender-sensitive treatment should result in lower costs of GSS added per day before returning to an SUD treatment.

Given that Tang et al. (2014) examined treatment provided in naturalistic settings, without random assignment to treatment or waiting list conditions or to different levels of gender-sensitivity, the present study can be considered an investigation of costs of GSS to patients and providers, and of relationships between costs and outcomes of GSS.

Method

Participants

Programs

Data on IIP gender-sensitivity and costs were collected during visits to IIPs in a state in the western United States. The present research on costs and cost-outcome relationships was part of a NIDA-funded study of how GSS might affect treatment. Sixteen IIPs met initial criteria of providing treatment in residential mixed-gender settings with at least 1 female to every 10 males. This was reduced to 13 IIPs by closure of one program and by discovering during site visits that two other programs actually provided de facto women-only treatment due to complete gender segregation. Data needed to assess patient as well as program costs were collected via structured group and individual interviews with patients, program directors, clinical directors, and two counselors at each IIP, as detailed below. Each IIP received $750. Institutional Review Boards at each organization providing or analyzing data approved this study.

At one of the 13 IIPs, female patients announced after the interview that they did not want the research to include data on their post-IIP use of substance use services. This allowed us to measure costs for the 13 IIPs, but constrained us to analyze and report outcomes and cost-outcome relationships for 12 IIPs.

Patients

Group administrations of structured self-report forms were conducted with 76 female patients who had been at the IIPs for at least a week, to assure they were familiar with GSS activities and resources used in those activities. Each woman received a $25 gift card just before the forms were distributed. Three women treated at the 12 IIPs for which outcomes could be examined had died by the end of the two-year followup period. Two of those three had reentered a treatment before dying, however, and were retained in analyses of effectiveness and costs. Time to treatment reentry also could have been affected by incarceration, but none of the women inter- viewed spent more than three days incarcerated post-IIP. Table 1 provides detailed demographic information for female participants at the 12 IIPs. The largest percentage of participants (83%) reported being polysubstance users, which included opiates and cocaine, as well as alcohol. The second most commonly used drug reported by participants was marijuana.

Table 1.

Participant Characteristics

Measure	n	Mean	Median	SD
Age	76	38.51	40	11.03
Years of education	76	11.70	12	2.10
Number of children	64	1.30	1	1.58
	n	%
Race
Caucasian	48	63%
African American	9	12%
Multiracial	9	12%
tive American	6	8%
Asian American	1	1%
other	2	3%
Employment Status at Intake
Unemployed	56	74%
Not working, disabled	14	18%
Employed full time	2	3%
Employed part time	2	3%
Homemaker	1	1%
Institutionalized	1	1%
Polysubstance or Single Drug Use
Polysubstance use	63	83%
Single drug use	13	17%
Primary Drug Used
Alcohol	34	45%
Methamphetamine	16	21%
Cocaine	11	14%
Heroin	6	8%
Oxycodone, Hydrocodone	6	8%
Other opiates or synthetics	2	3%
Marijuana	1	1%
Secondary Drug Used
Marijuana	20	26%
Alcohol	12	16%
Cocaine	10	13%
Methamphetamine	7	9%
Tobacco	6	8%
Oxycodone, Hydrocodone	3	4%
Amphetamines	2	3%
Heroin	1	1%
Opiate substitute	1	1%
Other opiates or synthetics	1	1%
Mental Health Services
Receives mental health services	14	18%
Does not receive mental health services	44	58%
Needs mental health services	18	24%
Psychoactive Medication
Takes	22	29%
Does not take	53	70%
Needs psychoactive medication	1	1%

Data Collection for Gender Sensitivity and Effectiveness

Observation protocol for rating gender sensitivity of IIPs.

Both interviewers independently rated each IIP as (a) institutional versus home-like, (b) allowing privacy in washrooms, bedrooms, and (c) overall gender sensitivity for 33 characteristics of the program on 5-point scales (1 = low, 5 = high) as detailed in Tang et al. (2012). The interviewers discussed any disparate ratings to reach consensus. These ratings were analyzed to determine the gender-sensitivity of IIPs by Tang et al. (2012). Tang et al. used Rasch modeling (e.g., Conrad & Smith, 2004) combined with literature detailing gender-sensitive services (e.g., Grella, 2008) to assess the gender-sensitivity of treatment orientation, administration and staff, organization characteristics, women’s services, children’s services, general services, and physical design of the treatment facility. This resulted in four ordinal groups of IIPs with respect to the gender-sensitivity of services offered (cf. Tang et al., 2012).

Demographics data.

State Medicaid databases provided individual-level data on patient demographics, presenting problems at the time of IIP treatment, and days receiving IIP treatment.

Effectiveness data

Just as Barnett and Swindle (1997), Mannix (2010), and Tang et al. (2012) measured the outcomes of gender-sensitive and other SUD treatments according to amount of time elapsed between completion of the index IIP treatment and the next treatment, so did we. Substance use treatments received during the 24 months following initiation of IIP treatment were recorded in state databases. We viewed these data as potentially more valid that self-report of next treatment with briefer follow-up periods used in prior research. No direct measure of substance abuse following treatment was available.

Procedure

The order of site visits to collect cost data at different IIPs was determined by constraints of IIP staff availability and proximity of IIPs already scheduled for interviews. Written informed consent was obtained from each participant before each interview. At each IIP the program director, clinical director, and two counselors were questioned individually and privately using protocols described above. At each IIP 4 to 10 female patients reported their personal, family, and other resources expended on treatment activities by sitting at tables and completing questionnaires. Immediately following each site visit, data collectors completed the protocol described above to assess IIP gender sensitivity.

Group form administrators and interviewers

The female interviewer held a Masters of Social Work degree and was an experienced qualitative researcher. The male interviewer held a Ph.D. in Counseling Psychology and had 24 years of experience as a substance use researcher and clinician.

Calculating Costs of GSS per Patient for Each IIP: Cost model

We based our cost assessment on the Cost-Procedure-Process-Outcome Analysis (CPPOA) model (Yates, 1980a, 1996, 1999). CPPOA conceptualizes costs as the values of the types and amounts of resources used for specific activities within a human service program, so those specific resource amounts then can be associated with changes in targeted biopsychosocial processes and then to proximal nonmonetary and distal monetary outcomes. Focusing on costs of GSS offered in the IIPs, our methods incorporated Yates’ (1980a) additional concept that the value and amount of the same resource can differ according to the assessor’s perspective. For example, different interest groups involved in the receipt, provision, funding, or evaluation of treatment might value participants’ time as worth nothing, or as worth what they could have earned had they not been receiving treatment, plus what was paid to others for the childcare the participant otherwise would have provided (cf. Yates et al., 1979).

Costs to patients

The patient survey format was based on methods detailed in Yates (1980a, 1996, 1999) and implemented by Yates, Haven, and Thoresen (1979) and Yates (1980b, 1987). Interviews described below were conducted between 2008 and early 2009. State databases provided data on treatment, patient demographics, and all other variables for patients receiving IIP treatment between 2005 and 2009, but the present study was constrained to only use data for female patients who consented to both interviews and use of data on when they first reentered treatment post-IIP.

Patient reports of types and amounts of resources used

Visits to each IIP allowed female patients and staff to report confidential data on the amounts and costs of time, transportation, and other resources devoted to specific GSS activities using a structured form [provided in Appendix A]. This information provided the basis for assessing patient treatment costs. In a group meeting room from which staff were excluded, interviewers asked female patients to estimate amounts of resources they provided for specific program activities. After completing practice items for “getting ready in the morning,” patients estimated mean, minimum, and maximum duration and frequency per day for their participation in:

1. individual counseling,

2. family or couples counseling,

3. mixed-gender group therapy or education,

4. women-only group therapy or education,

5. medication management,

6. other medical services,

7. other appointments (e.g., legal, child custody, housing), and

8. other services.

Each patient recorded her estimates on a separate form with columns for “Time (minutes per day)” and subcolumns for “Average,” “Least,” and “Most.” To the right of these were columns for “# time(s) per week” and “Transportation (round trip)” with subcolumns for “Yes/No: Time Taken?” and “Did you pay? How much?”. Patients then were asked to describe licit “wages lost or given up due to being in treatment” and any payments they made for dependent care during their stay at the IIP, again with subcolumns for “Average,” “Least,” “Most,” and blanks for recording “$ per month (or $ per day for days per week).”

Calculation of patient costs

Following methods to value the time of patients who were students, homemakers, or unemployed (Fleming et al., 2000), the monetary value of patient time spent in a GSS for which they had been interviewed was calculated by multiplying (a) the time reported by the patient for the GSS by the frequency reported for the GSS and then by (b) either the minimum hourly wage for the state in which the IIPs operated or, if higher, the hourly wage reported by the patient. Cost of dependent care purchased while a primary caregiver was being treated in the IIP was averaged for each program, and converted to an hourly value. Because patient transportation to and from IIPs at the start and end of the residential stay varied markedly according to the arbitrary location of the patient before treatment, and could be substantial given the distances sometimes involved, costs of that transportation were excluded from analyses.

Costs to providers and programs

To collection additional data on the types and amounts of resources used by treatment programs, individual program administrators and staff were queried in direct question-and-response format, with specific responses being recorded immediately on forms.

Program director reports of types and amounts of resources used

Structured question- and-response interviews of program directors were conducted jointly by the female and male in- terviewers described above, presenting 102 specific items querying (a) respondent background,(b) program structure and philosophy, (c) patient admission patterns, (d) children’s services of- fered, (e) staff competencies and training, (f) program challenges, and (g) program costs. [Appen- dix B provides the entire question-and-answer protocol and all items.] Directors also were given a table [Appendix C] requesting for each IIP staff member: (a) years of experience, (b) salary, (c) percent of time dedicated to the IIP, and (d) program benefits such as medical coverage.

Clinical director and counselor reports of types and amounts of resources used

During the same site visit at which the program director was questioned, the clinical director responded verbally to 170 questions [Appendix D] regarding: (a) respondent experience and training, (b) treatment philosophy, (c) patient population, (d) patient assessment and treatment implementation, (e) treatment planning, (f) specific services for patients, (g) services available to patients’ children and families, (h) discharge planning, (i) post-treatment housing services, (j) continuing care ser- vices, and (k) respondent demographics. Counselor interviews covered 263 items [Appendix E] focusing on patient access to counseling, children, continuing care, barriers to treatment, details about the IIP environment, and satisfaction with treatment.

Calculation of treatment provider and program costs.

The amount and monetary value of each type of resource used to provide specific GSS were estimated for each IIP separately. Services shown by Tang et al.’s (2012) Rasch analyses to be important to gender sensitivity were deemed costable if they consumed specific identifiable and quantifiable resources from providers, patients, or treatment facilities or any combination of these. Relationship counseling, for example, required discernible and finite amounts of patient time, counselor time, and office space according to interview data. Noncostable activities were those which consumed negligible or no identifiable specific resources, such as “mentioning ‘behavioral health’ in the program’s mission statement.” Table 2 indicates which IIP services were costed in the present study, the average overall gender sensitivity rating for the service, and which interview or interviews provided data for each service.

Table 2.

Costable Gender-Sensitive Services

Mean Level of IIP Gen- der-Sensitive Services (GSS) 1–5 scale				Costable Gender-Sensitive Services in One or More IIPs in the Level of GSS	Source or sourcesa
I	II	III	IV
4.00	4.3	4.67	5.00	Women receive relationship counseling	P, CD, C
	3
4.33	4.6	5.00	5.00	Women receive sex education	PD, CD
	7
3.67	4.0	4.00	4.75	Women participate in spiritual or cultural activities	CD, C
	0
6.00	5.0	3.17	3.33	Fewer women share a bathroom in the IIP	C
	0
3.33	1.3	2.00	4.25	Women receive trauma counseling	PD, CD
	3
4.00	4.5	4.00	5.00	Women receive assertiveness and self-efficacy training	CD
	0
4.00	3.3	5.00	5.00	Women receive women’s health information	CD
	3
3.00	3.6	3.33	4.75	Women participate in physical activities	CD
	7
4.00	4.3	5.00	5.00	Women participate in social and recreation activities	CD
	3
1.33	1.3	2.33	3.00	Parenting skills are addressed during assessment	PD, CD
	3
2.00	4.6	5.00	5.00	Social support is addressed during assessment	CD
	7
2.67	4.0	4.33	4.75	Grief and loss are addressed during assessment	CD
	0
2.67	3.6	5.00	5.00	Domestic violence is addressed during assessment	CD
	7
1.67	2.3	4.67	5.00	Safety is addressed during assessment	CD, C
	3
2.33	3.3	4.33	4.50	Sexuality is addressed during assessment	PD, CD
	3
2.00	4.5	5.00	5.00	Stage of change is addressed during assessment	CD
	0
1.33	1.3	2.00	3.50	Life skills are addressed during assessment	CD
	3
1.67	1.3	3.00	4.00	Vocational needs are addressed during assessment	CD
	3
2.33	3.6	5.00	5.00	Housing needs are addressed during assessment	CD
	7
2.00	3.0	4.67	4.50	Spirituality/religion/culture are addressed during assessment	CD
	0
2.33	2.0	2.67	4.67	Women are connected to social supports by discharge	CD
	0
4.33	3.3	3.00	5.00	Women have concrete posttreatment housing plan	CD
	3
16.0	10.	10.0	30.0	Average hours each staff spends per year on training related to women’s recovery	PD, CD
	0
100	100	66.7	100	Percent of programs providing treatment to women with acute psychiatric condition	CD, C

^aSources of on this gender-sensitive services:

P: Patient

PD: Program Director

CD: Clinical Director

C: Counselor

Monetary values of each resource used for these GSS were derived from the best information available, typically from the program director interviews described earlier. Costs of resources, e.g., staff salary plus benefits, were multiplied by the number of units used for an activity, e.g., 2 hours of staff time per group meeting, and the number of times that activity was reported to occur during the patient’s treatment at the IIP, e.g., once per week for 4 weeks. If data on resource values were not available from program directors, information from interviews of clinic directors, counselors, and patients was used. This information was supplemented by state officials familiar with the IIPs. For example, an official reported that sex education in IIPs consumed 1.5 hours of counselor time, patient time, and office use. The monetary value of counselor time was obtained from the table described above that was completed by the program director for each IIP as part of the question-and-answer process [again, Appendix C].

The monetary value of office space (dollars per square foot per hour) was found by dividing total mortgage or rental cost reported by the program director for all buildings used in the IIP by (a) the number of days during the mortgage or rental payment period for which the facility was used for treatment, based on the modal length of stay reported for the program,¹ and then by (b) the number of hours the facility could be used, i.e., 18 hours, allowing time for patients to sleep and engage in grooming and recreation. That total IIP space cost per hour then was divided by the number of square feet used by the IIP according to the program director, resulting in the cost per square foot per hour for the IIP.

The monetary value of space used for a GSS was calculated by multiplying the hourly cost of office space used for that GSS by the mean number of hours the activity consumed during a patient stay at the IIP according to patient reports for that IIP or, if patient reports were unavailable, state officials’ estimates. Monetary values of additional resources, including administrative space, staff meeting rooms, hallways, staff restrooms, office equipment, and materials, were summed and then distributed across specific GSS in proportion to the previously calculated cost of space devoted to the different GSS, as recommended by Yates (1999).

Total costs of GSS.

Total costs of each GSS activity were calculated by summing, for each female patient, costs to the patient and provider (time, facilities, and other overhead) for the activ- ity. This cost then was adjusted to reflect the reality that, according to program directors, not all female patients participated in all GSS. Because use of the same resources by fewer female patients would increase the cost per patient for the activity, the cost totals for each GSS were divided by the proportion of patients reported to have participated in that activity to arrive at a total cost per patient for each GSS.² If the treatment activity occurred in groups, the cost of the activity per patient was calculated by dividing provider and overhead costs for the group by the number of female patients who typically attended groups.

Calculating Effectiveness and Cost-Outcome Relationship of GSS

Effectiveness calculations.

Days to treatment reentry following the IIP stay had been used in previous research on gender-sensitive and other SUD programs (Barnett & Swindle, 1997; Mannix, 2010; Tang et al., 2012). As noted earlier, this measure of effectiveness was available in state databases for the 24 months following IIP entry by each patient. It was the only effectiveness measure available. Treatment starting less than a month after discharge from the IIP was considered a continuation of IIP treatment. Any in- or outpatient substance use treatment occurring one month or more following discharge from the IIP was considered new treatment. Finally, a Kaplan-Meier survival analysis examined number of days until treatment reentry as a function of gender sensitivity.

Cost-outcome relationship calculations

Cost-outcome analysis began with a scatterplot of days to treatment reentry versus cost of providing gender-sensitive IIP services, following the descriptive data analysis procedures developed by Siegert and Yates (1980). These and all other cost-outcome analyses used patient + provider costs per patient. To quantify the cost-outcome relationship, the number of days until the patient reentered treatment was divided by that patient’s GSS cost. These individual-level outcome/cost ratios of treatment-free days per dollar spent on GSS were multiplied by 100 to improve usability.

Results

Which Gender-Sensitive Services Were Most Time-Intensive?

Mixed-gender groups were the most time-consuming and most frequent GSS reported by female patients, means (Ms) = 205.3 minutes per episode, occurring 6.7 times per week, i.e., almost once per day. Women-only groups were the next longest in duration per episode and next most frequent, Ms = 104.7 minutes, 5.1 times per week. Medical services and family or couples counseling typically consumed about 1.5 to 1.7 hours (Ms = 101.0 and 79.4 minutes respectively), between once and twice a week (Ms = 1.5 and 1.9). Individual counseling sessions lasted about a half hour (M = 26.8 minutes), occurring about twice a week (M = 1.9).

Costs of Gender-Sensitive Services

GSS costs per program

For GSS at each of the 13 original IIPs, patients typically devoted resources of significantly higher total monetary value than did providers (Ms = $585.41 versus $344.42 per patient per stay, medians (Mdns) = $594.92 versus $385.38, respectively), paired t(12)= 4.66, p = .001, Wilcoxon Rank-Sum test, W(13) = 89, p < .002 (Table 3). These and all other ps in our analyses were two-tailed. Analyses were nonparametric to avoid unrepresentative skewing by the occasional high costs commonly found in services research. Both separately and summed, patient and provider GSS costs were not significantly higher in the more gender-sensitive IIP groupings, according to Kruskal-Wallis one-way analyses of variance, ps > .33, as illustrated in Figure 1.

Table 3.

Program-Level Resources Invested by Patients and Providers in Gender-Sensitive Services, Av- eraged Per Patient Per Stay

Gender- sensitive service (GSS) level	Intensive inpatient programs (IIP)	Resources from the average pa- tient	Resources from providers	Total of resources from patients + providers
I (least)	Program 1	$635.21	$42.27	$677.68
		(76.41)	(5.09)	(81.54)
	Program 2	$566.87	$339.57	$906.45
		(144.85)	(86.80)	(231.65)
	Program 3	$608.59 (0.00) a	$451.71 (0.00) a	$1,060.40 (0.00) a
II	Program 4	$602.14	$217.20	$819.34
		(42.25)	(15.25)	(57.50)
	Program 5	$594.92	$397.89	$992.81
		(8.36)	(5.59)	(13.95)
	Program 6	$533.42	$487.97	$1,021.39
		(155.13)	(141.94)	(297.06)
III	Program 7	$523.40	$84.19	$607.58
		(81.47)	(13.12)	(94.59)
	Program 8	$581.92	$521.92	$1,103.83
		(137.16)	(123.00)	(260.16)
	Program 9	$636.11	$385.38	$1,021.49
		(40.59)	(24.61)	(65.20)
IV (most)	Program 10	$655.85	$158.55	$814.40
		(112.89)	(27.29)	(140.18)
	Program	$527.73	$378.58	$906.31
		(152.47)	(109.39)	(261.86)
	Program	$514.73	$544.37	$1,059.10
		(—) b	(—) b	(—) b
	Program 13	$629.49	$467.92	$1,097.41
		(167.04)	(124.11)	(291.15)
mean		$585.41	$344.42	$929.86
minimum		$514.73	$42.27	$607.58
maximum		$655.85	$544.37	$1,103.83

Note. Parenthesized values are standard deviations.

^aStandard deviations of zero (0) indicate that all women in the study at that IIP stayed the same number of days and thus had the same cost.

^bStandard deviations were not calculable for Program 12 because, as reported in text of this manuscript, individual-level cost data for women in this program had to be discarded because the women withdrew consent after completing the Patient Interview. Cost values were derived from Program Director, Clinical Director, and Counselor Interviews. The modal length of stay for fe- male patients in other IIPs in this GSS level, i.e., 21 days, was used to estimate costs for this IIP.

Figure 1. Patient and Provider Costs of 13 Intensive Inpatient Programs for Substance Use.

Stacked bar graphs showing program-level patient costs (darker, lower sections of bars) and provider costs (lighter, upper sections of bars) for the four levels of gender-sensitive services (I, II, III, and IV, ranged from offering the least to most gender-sensitive services [GSS]) in 13 intensive inpatient programs for substance use.

GSS costs for gender-sensitivity groupings

Summarizing costs for the four groupings of gender sensitivity found by Tang et al. (2012), Table 4 shows mean and median GSS costs per patient for each grouping of gender-sensitive IIPs, for patient and provider perspectives separately. These costs were calculated by multiplying the cost per day of providing gender-sensitive services within a particular program for each patient by the number of days that patient participated in IIP treatment.

Table 4.

Median Patient and Provider Costs of Gender-Sensitive Services per Patient for Programs in Different Levels of Gender Sensitivity

Gender- sensitive service (GSS) level	Female patients in IIPs at GSS level	Median cost of gender-sensitive services per pa- tient per day	Median Cost of Gender-Sensitive Ser- Vices per Patient for IIP Treatment
Patient Cost	Provider Cost
I (least)	20	$30.21	$630..46	$311.30
		(22.59, 37.87)	(423.40, 850.05)	(28.20, 509.40)
II	24	$33.09	$614.73	$411.06
		(29.26, 48.64)	(203.20, 688.00)	(185.92, 534.52)
III	9	$34.05	$657.20	$398.35
		(21.70, 52.56)	(355.11, 803.59)	(57.19, 720.65)
IV (most)	23	$40.97	$651.30	$218.95
		(37.84, 43.16)	(175.91, 905.67,)	(126.21, 483.90)
median	21.5	$34.05	$630.46	$350.92

Note. Parenthesized values are minima and maxima. Also, the relationship between gender sensitivity of treatment and GSS cost approached but did not achieve statistical significance, r_s(75) =.20, p = .08.

The apparent mildly positive association between level of gender-sensitive services and total GSS cost approached but did not achieve statistical significance, r_s(75) = .20, p = .08. We did, however, find that women spent significantly fewer days in IIPs that were more gender-sensitive, r_s(76) = −.45, p < .001.

Outcomes of Gender-Sensitive Services

IIP gender sensitivity and days until the patient returned to treatment (Table 5) were not significantly associated, according to a Kruskal-Wallis test, X²(3, n = 75) = 4.88, p = .18. A Cox proportional hazards regression of days to treatment reentry used the four levels of gender sensi- tivity as the stratifying variable with patient age and baseline alcohol problem days as covariates because these variables were found to be statistically significantly different across GS levels in exploratory analyses?. The overall model revealed no significant effects for gender sensitivity of the IIP. Neither age, p = .59, nor number of reported alcohol problem days prior to treatment, p =.72, were significant covariates. Although additional variables could have been included in this regression, the statistical power of detecting effects would have been diminished excessively. Also, nonparametric statistical analyses (chi-square tests) compared the four gender-sensitive groups on additional variables: Race (48 Caucasian versus 27 in all other categories), Employment Status at Intake (56 unemployed versus 20), Primary Drug of Use (34 alcohol, 42 all other), Mental Health Services Received (44 do not receive, 32 receive or need), and Psychoactive Medication (53 do not receive, 23 take or need). These analyses found a significant difference among gender-sensitivity groups in race, χ2(3, n = 75) = 30.50, p = .01, but no significant differences in these factors between treatment groups at intake: Employment Status: χ²(3, n = 76) = 3.86, p = .28, Primary Drug Use: χ²(3, n = 76) = 2.69, p = .44, Receipt of Mental Health Services: χ²(3, n = 76) = .28, p= .96, and Psychoactive Medication: χ²(3, n = 76) = 1.43, p = .70.

Table 5.

Median Days to Treatment Reentry for Patients in IIP Treatments Differing in Gender Sensitivity

Gender-sensitive service (GSS) level	Female patients in IIPs at GSS level	Median days to treatment reentry
I (least)	20	279.0
	(5, 9)	(37, 703)
II	24	701.5
	(5, 10)	(65, 724)
III	9	572.0
	(2, 4)	(85, 708)
IV (most)	23	377.0
	(6, 10)	(56, 730)
median	21.5	474.5

Note. Parenthesized values are minima and maxima. Gender sensitivity was not significantly associated with days until patients returned to treatment, χ²(3, n = 75) = 4.88, p = .18.

Before undertaking the survival analysis, several variables were explored as potential co-variates. Based on the literature and discussion with the research team, variables that were thought to potentially impact treatment re-entry were: days of alcohol use in the past month (preceding the index treatment), days of drug use in the past month, number of mental health problem days in the past month, number of treatment episodes in the past two years, number of substance abuse treatment days in the past two years, index IIP treatment completion or not, and age. None of these potential covariates was found to be significantly related to the outcome variable, days to treatment re-entry, after Pearson correlations were performed. The correlation coefficients from these exploratory analyses are shown in Table 6.

Table 6.

Pearson Correlation Coefficients, Potential Covariates with the Outcome Variable “Days to Any Treatment”

Potential outcome predictor variable	r	P
Alcohol problem days in last month	−0.02	0.85
Drug problems days in last month	−0.04	0.74
Mental health problem days in last month	−0.03	0.83
Number of previous treatment episodes	−0.07	0.54
Number of days of previous treatment	0.07	0.57
Treatment completion	0.14	0.22
Age	0.07	0.55

Also, survival analysis found no significant difference in the effectiveness of IIPs in the four strata of gender sensitivity after a Mantel-Cox Log Rank comparison, χ² (3, n = 75) = 3.64, p= .30. Even comparing the lowest level of gender-sensitive treatment to the combined three higher levels of gender-sensitive treatment still found that percentages of patients not returning to treat- ment (35% versus 43%) were not significantly different, z <.7.

Cost-Outcome Relationships for Treatments Varying in Gender-Sensitivity

Figure 2 displays the result of plotting days to treatment reentry as a possible function of cost of providing gender-sensitive IIP services: neither an obvious direct nor obvious inverse association is apparent. IIPs higher in gender-sensitivity had no better or worse outcome/cost ratios than other IIPs according to a Kruskal-Wallis ANOVA, χ²(3, n = 75) = 3.50, p = .32, as well as a Kaplan-Meier survival analysis, χ² (3, n = 75) = 3.35, p = .34. Nonparametric analyses and medians are reported here because distributions of costs were nonnormal. median dollars for patients + providers per treatment-free day were $2.62 per patient.

Figure 2. Outcomes of Substance Use Treatment in IIPs as a Possible Function of Cost of Gender-Sensitive Services.

Patient-level scatterplot of days until reentry to any substance use treatment versus patient + provider cost of gender-sensitive services. There is a concentration of points at the top of the graph because many patients did not reenter treatment during the 2-year follow-up.

Discussion

Despite efforts to improve outcomes, substance users who have received intensive inpatient treatment frequently return to treatment within a few months (Ashley, Marsden, & Brady, 2003). One way to enhance outcomes may be to adapt the activities of treatment to needs of specific subpopulations. We have reported and analyzed costs, outcomes, and cost-outcome indices for 12 intensive inpatient programs which varied in how much they adapted treatment to needs of female substance users. These gender-sensitive services (GSS) included relationship counseling, self-efficacy training, trauma counseling, assessment of parenting skills, women-only groups, and more (Greenfield et al., 2007; Grella et al., 1999; Tang et al., 2012). More gender sensitive services were not significantly associated with either higher costs or better outcomes.

The present study asked whether inclusion of more of these gender-sensitive activities in mixed-gender intensive inpatient programs (IIPs) changed costs, outcomes, or cost-outcome relationships of substance use treatment for women. Descriptive, graphic, and statistical analyses suggested that higher levels of gender-sensitive services were no more costly than lower levels of gender sensitivity in a dozen different programs – all intensive inpatient, all treating men as well as women – in a major state in the US. Separate and summed values of resources devoted by patients and providers to these gender-sensitive services did not increase significantly as program gender-sensitivity increased. Comparisons to other IIPs in other states were not possible, given the absence of comparable data on marginal GSS costs for IIPs in those states.

In the broader context of substance use treatment delivered in a wide range of settings, and contrary to French et al. (2002), we found that more gender-sensitive services could be offered with no statistically significant increase in treatment costs. Reimbursement for all programs involved in the present study was $90.18 per patient per day with no additional funding for programs offering more gender-sensitive services. Because programs usually must constrain expenditures to not surpass revenues, and because we found that gender-sensitive services did require substantial patient and provider resources, offering more gender-sensitive services might have been balanced in some programs by offering fewer traditional services. If this adjustment was made, it does not seem to have impacted treatment outcomes noticeably: a variety of statistical analyses found no significant differences in outcome among the four ranked levels of gender-sensitive treatment, even after covarying for patient age and pretreatment severity of substance (alcohol) use.

Of course, it is possible that offering more gender-sensitive services actually is more effective. Although it is a common outcome metric in other SUD treatment research, as noted earlier, time until treatment reentry may not have been the best measure on which to focus when evaluating effects of gender-sensitive services. It was the only outcome measure available to us. Returning sooner to treatment could be a desirable help-seeking behavior, possibly fostered by higher self-efficacy. Moreover, not all patients who received additional treatment may have already reinitiated drug use: some could have been acting to avoid relapse. Recognizing heightened relapse risk, and then reengaging with treatment, could be a positive rather than negative development in the evolution of patients’ self-management of substance use (Barnett & Swindle, 1997; Yates, 1985). Measures of treatment effectiveness that are themselves gender-sensitive may be needed to best detect the effects of more gender-sensitive regimens.

Of course, our study used natural variation in treatment gender-sensitivity, which can maximize some forms of validity but also could create several confounds. Random assignment to different levels of gender-sensitivity could reduce these confounds, although the ethics of such research might be challenged by some. Furthermore, although each program director was asked, “What percentage of female patients receive…?” a particular service, program directors could have been reporting these rates based on their impressions of the population’s need for particular gender-sensitive services, rather than the services’ actual availability. Finally, summing patient and provider costs could be viewed as an exaggeration of “real” costs if only costs to providers are of concern. We did, of course, report provider-only costs (in Table 3 as noted previously) for those who wish to focus exclusively on them.

In sum, we did not find that treatments which were similar in many respects (e.g., all intensive inpatient, all in the same state), but which differed in gender-sensitivity, were significantly more or less costly, had significantly better or worse outcomes, or had significantly better or worse cost-outcome relationships. Further research on gender-sensitivity of substance use treatment, and examination of whether health, criminal justice, and other human services change following GSS, could clarify whether the costs of providing specialized care to women are “worth it” in terms of savings in future services, i.e., whether adding more gender-sensitive services is cost-beneficial (Yates, 2009).

Highlights.

• addition of gender-sensitive services to traditional substance abuse treatment does require more resources, but extra cost is minor.

• No differences were found in cost-effectiveness among treatments of varying levels of gender sensitivity

• Investment of $100 in gender-sensitive services adds 26 to 30 treatment-free days per patient.

Biographies

Sarah Hornack vita

Sarah Hornack is a licensed clinical psychologist and national service provider at Children’s National Medical Center in Washington, DC. After receiving her BA from the University of North Carolina at Chapel Hill and her Ph.D in Clinical Psychology from American University in 2014, Dr. Hornack completed the Fellowship Program in Pediatric Psychology at Mt. Washington Pediatric Hospital in 2015.

Brian T. Yates vita

Brian Yates has published or has in-press 86 articles, book chapters, and online pieces, plus 5 books. Most of his publications apply cost-inclusive evaluation to the systematic evaluation and improvement of human services. In 1976 he received a Ph.D. in psychology from Stanford University and began working as a full-time tenure-track professor at American University in Washington, where he continues to teach, write, and conduct and supervise research in a clinical doctoral training program. Dr. Yates has presented workshops and research in Australia, Canada, Romania, the UK, and New Zealand as well as the US.