Do Psychologists Help? Assessment and Evaluation of an Integrative Care Approach to Cancer Treatment

Psychosocial care has been shown to improve psychological and physiological functioning in cancer patients. However, as few as five percent of cancer patients engage in psychosocial care. Therefore, Study 1 of this dissertation developed measures of core TTM constructs (Stage of Change, Decisional Balance, SelfEfficacy) relevant for increasing engagement in psychosocial care among individuals diagnosed with cancer. Measure development entailed qualitative methods for item development and refinement followed by a series of quantitative analyses. The Stage of Change measure was validated against external constructs such as subjective present and future wellbeing. As expected, a chi-square test indicated that individuals in Action and Maintenance were significantly more likely to be in treatment than those in the preAction stages. Measures for Decisional Balance and Self-Efficacy were developed using splithalf, cross-validation procedures. In these, a series of Principal Component Analyses (PCAs) were conducted with half of the sample to narrow the item set and explore factor structure, and Confirmatory Factor Analyses (CFA) were conducted on the second half of the sample to confirm the factor structure and item loadings. For Decisional Balance, PCA supported two, 8-item factors, and CFA indicated a twofactor correlated model was the best fit to the data. For Self-efficacy, PCA supported two, 3-item factors, and CFA further supported this structure. Multivariate analyses indicated significant stage-construct relationships. Overall results supported the validity of the measures developed and laid the foundation for applying the TTM to psychosocial treatment acceptance among cancer populations. Implications for application of the TTM to cancer populations are discussed. Given that cancer patients frequently experience considerable distress during diagnosis and treatment, Study 2 described the development and utilization of a behavioral health program for cancer patients, at a small community hospital, as well as provided preliminary results on program efficacy. This program was co-developed by individuals from a university-based clinical psychology doctoral program and a community hospital. The behavioral health program was comprised of a licensed, PhD-level clinical psychologist and seven clinical psychology doctoral students, who met with patients in order to accrue clinical hours. Patients were typically referred by their oncologists or nurses. Distress, depression, and anxiety were evaluated for a small subsample of participants. From the time the program was initiated, 238 patients between ages 18 and 95 (M = 66.4) were evaluated over a three-year period. The majority of patients (77.8%) were offered psychosocial care. Although 49.8% declined treatment, 23.6% engaged in one session and 26.6% engaged in two or more. Patients who were referred through the STAR Program® were more likely to engage in psychosocial care than those who found out about behavioral health through other means. First, distress tracking may be improved if nurses, oncologists, and behavioral health providers administer measures. Second, partnerships between clinical psychology doctoral programs and hospitals may be mutually beneficial. Third, hospitals offering cancer treatment may benefit from obtaining STAR® certification, in order to generate referrals for comprehensive cancer care. These efforts can serve as a model for other hospitals seeking to integrate behavioral health into routine cancer treatment. Together, these two studies address the scarcity of studies on the intersection of cancer and mental health. As such, this work aimed to bridge the gap between the two disciplines, in order to prevent and treat mental health problems in cancer patients. Results of Study 1 may be used to guide researchers and clinicians in designing and implementing interventions. Study 2 methods and findings may be used to develop other behavioral health programs and to benchmark other integration efforts.


LIST OF TABLES
Organization [2] predicted a 70% increase over the next two decades, worldwide.
Although cancer incidence is expected to increase, the cancer death rate in the United States decreased by 23% between 1990 and 2012 [1]. Given the rise in cancer cases and the growing survivor population, there will also be increasing physical and emotional concerns associated with the disease and its treatment [3,4]. The multidimensional burden (i.e., vocational, financial, physical, interpersonal) of cancer undeniably makes it one of the most emotionally debilitating conditions [5].
The relationship between the physical and emotional burden of cancer is evidently strong [5][6][7][8]. For instance, depression has been shown to increase the length of hospitalization in lung cancer patients undergoing thoracic surgery [9]. Further, in a sample head and neck cancer patients, quality of life and negative coping styles were related to higher levels of anxiety and depression, as well as lower levels of optimism [10]. Remarkably, a 10-year study of 3,080 cancer survivors revealed that those with depression had double the risk of all-cause mortality, compared to those without depression [11]. Collectively, these findings reveal the pervasiveness of mental health issues among cancer patients, their staggering impact on physiological outcomes [9], survival rates [12], and accordingly, the need to address psychological concerns [10].
Psychosocial interventions, particularly in the form of evidence-based treatments and support groups, have been used to address a variety of cancer-related concerns, including quality of life [13,14], fatigue [15], pain [16], depression [17][18][19], and anxiety [20]. Further, psychosocial care may be used for increasing resilience and confidence, as well as for addressing fear of tumor progression [21]. Overall, this growing body of research has demonstrated effects largely in favor of psychosocial care.
Physiological outcome data further strengthens the case for psychosocial care among cancer patients. For instance, a seminal study revealed the effect of psychosocial treatment on survival of metastatic breast cancer patients, such that those who had participated in a support group were more likely to be survivors eight months after the intervention [22]. Notably, a more recent study revealed that pre-operative stress management improved immune functioning in men with prostate cancer undergoing radical prostatectomy [23]. Further, several studies found that psychosocial interventions were helpful in slowing disease progression [24][25][26] and increasing survival in cancer patients [27]. Collectively, studies support the potential for psychosocial interventions to improve cancer patients' physiological profile.
Numerous studies have explored the mechanisms and processes underlying the impact of psychosocial care on cancer patients' physiological functioning. For instance, a recent review of 16 randomized controlled trials (RCTs) that examined specific therapeutic components of treatments tailored for cancer patients revealed that alterations in cognitions, self-efficacy, mood disturbance, pain, and self-esteem were most important [28]. An earlier study established correlations between verbal or 5 written expressions of emotions and levels of tumor-infiltrating cancer lymphocytes in melanoma, suggesting that psychosocial interventions can enhance emotional expression to positively affect disease course and overall physiological functioning [26]. Similarly, Jensen discovered that repression of negative emotions was associated with an aggravated course of breast cancer and that psychological variables were twice as effective at predicting clinical outcomes as were biological indicators [29].
Psychosocial interventions can also improve adherence to various difficult cancer treatments, which can increase survival. Altogether, compelling evidence exists for the role that psychosocial interventions may play in cancer patients' physiological profiles.
Despite overwhelmingly strong evidence that psychosocial interventions improve psychological and physiological well-being in cancer patients, reluctance to accept psychosocial treatment prevails. For example, a study of 132 cancer patients revealed that only 28% participated in psychosocial support, with 88% of respondents being women with a history of breast cancer (72%). However, those who utilized support had positive attitudes towards therapeutic interventions and a desire to cope more effectively with their illness [30]. A recent study of 1,777 cancer survivors revealed that only 4.4% used psychosocial care and alarmingly, the majority (55.1%) never even discussed the possibility with their oncologists. Interestingly, the 4.4% that used psychosocial services reported high satisfaction with how their needs were addressed [31]. In light of these findings, treatment engagement strategies are needed, particularly by way of assessing readiness to change and developing interventions.
The Transtheoretical Model (TTM) has been found effective in assessing 6 readiness to change and in guiding interventions [32]. The TTM is an integrative and comprehensive model of intentional behavior change that incorporates processoriented variables to explain and predict how and when individuals change. TTMguided interventions have modified many health risk behavior changes, including adherence to medical protocols and treatments, such as mammography screening [33], medications [34,35], blood glucose monitoring [37], and blood donation [36].
Therefore, it offers a promising theoretical framework for assessing readiness to accept psychosocial care for cancer populations.

Design
A sequential process of measure development was used to develop measures of Stages of Change, Decisional Balance, and Self-efficacy [42]. A series of semistructured expert and research participant interviews were conducted. Item development was followed by exploratory, confirmatory, and external validation analyses.

Item Development
Initial item development was based on a comprehensive review of TTM measures for other behaviors (e.g., physical activity, high-fat diet, cigarette smoking).

8
Items were further developed from the literature on psychooncology and psychotherapy.

Expert Interviews
Following initial development, items were refined using feedback from experts in behavioral health, oncology, and the TTM. First, one licensed psychologist and PhD-level expert in working with cancer patients in a team-based oncology setting participated in a semi-structured interview on issues surrounding patient engagement in psychosocial care and provided feedback on the proposed set of items. Next, two oncologists provided feedback on issues that cancer patients commonly face with regard to diagnosis and treatment, as well as barriers to engaging in psychosocial care.
Finally, two PhD-level experts in the TTM reviewed the proposed set of items for clarity and face validity.

Qualitative Participant Interviews
After expert feedback was incorporated, 12 semi-structured qualitative interviews were conducted with cancer patients actively recruited from a community hospital. The goal of these interviews was to elicit feedback on item clarity, acceptability, and face validity. Participants had to be over the age of 18 and had to have a cancer diagnosis. All interviews were conducted in private patient rooms, while individuals were receiving chemotherapy. Participants reviewed and signed informed consent forms first. No participants withdrew from the study after reviewing informed consent. Participants then reviewed and completed the initially developed items and provided oral feedback. Participant feedback was discussed with the TTM experts and was incorporated to generate the final version of the survey.

Survey Administration
The survey was administered using SurveyMonkey™ online survey software.
Participants accessed the survey via an online link provided by Cint™, a targeted survey population and panel recruitment company. Individuals were asked to check a box indicating that they read the informed consent form and agreed to participate.
They were then routed to questions on eligibility criteria (same as those for qualitative interviews). Eligible individuals were then linked to the full survey. Data were extracted from SurveyMonkey™ into SPSS for exploratory analyses and to EQS for confirmatory analyses.

Recruitment
Participants for qualitative interviews were recruited in person, by the primary

Qualitative Interview Sample
Twelve, one-on-one, qualitative interviews were conducted by a clinical psychology doctoral student. The average age of the participants was 65.5 (SD = 10.9) and all participants had a present cancer diagnosis. Seven of the participants identified as female and five identified as male. All 12 participants identified as White.

Measures Used
Cantril Self-Anchoring Striving Scale. The Cantril Self-Anchoring Striving Scale [43] was used to determine evaluative well-being outcomes. Individuals were asked to rate their current and future lives on a ladder scale from 0 to 10, in which 0 represented the worst possible life and 10 represented the best possible life. The first question aims to capture present subjective well-being: "Please imagine a ladder with steps numbered from 0 at the bottom, to 10 at the top. The top of the ladder represents the best possible life for you, and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time?" The second question aims to capture future subjective well-being: "On which step do you think you will stand about five years from now?" Individuals who rated their present lives a 7 or higher and their future lives an 8 or higher were classified as "thriving". Individuals who rated their current lives a 4 or lower and their future lives a 4 or lower were classified as "suffering". Individuals who met neither of these 11 criteria were classified as "struggling" (e.g., rated their current lives as 5 and future lives as 6).

Measures Developed
Stage of Change for Psychosocial Care. Participants were assigned to a Stage of Change based on their answers to a short series of questions. They were assigned to the Precontemplation stage if they indicated that they were not considering psychosocial care and did not intend to engage in it for the next 6 months; to Contemplation if they intended to seek psychosocial care within the next 6 months; and to Preparation if they intended to seek psychosocial care within the next 30 days.
Participants were assigned to the Action stage if they were receiving psychosocial care and had been for less than 6 months, and Maintenance if they had been receiving psychosocial care for 6 months or more.
Decisional Balance for Psychosocial Care. Thirteen items represented the Cons and eight items reflected the Pros. Respondents indicated how important each item was in their decisions of whether to accept psychosocial care, on a 5-point Likert scale, ranging from 1 = 'Not Important At All' to 5 = 'Extremely Important'.
Self-efficacy. Nine items assessed Self-efficacy. Items evaluated participants' confidence in their ability to engage in psychosocial care across a variety of challenging situations (e.g., feeling fatigued after chemotherapy). Participants indicated their confidence levels on a 5-point Likert scale, ranging from 1 = "Not At All Confident" to 5 = "Extremely Confident".

Data Analysis
Data were examined for violations of normality before exploratory and confirmatory analysis. A random half of the sample was used for the exploratory phase using principal components analysis (PCA) with varimax rotation on item correlation matrixes. PCAs determined the number of components and reduced scales to a smaller set of items. The number of components retained was based on the minimum average partial procedure (MAP) and parallel analysis [44,45]. Item selection was an iterative process that involved removing items for quantitative reasons (loadings <.40, or > .90 and correlations >.70 with other items, or high loadings [>.40] on multiple factors) and qualitative breadth of construct (to avoid redundancy and maintain conceptual breadth). The overall Cronbach alpha was examined to determine scale internal consistency.
The second half of the sample (n = 238) was used for confirmatory factor analysis (CFA). CFAs were used to evaluate the degree to which an independent portion of the data fit the model created by iterative PCAs. Model fit and factor loadings were evaluated. Final item selection was determined on the basis of item clarity, lack of redundancy, and conceptual breadth. Finally, Cronbach alphas and rho coefficients were examined to determine scale internal consistency. In the final phase, external validation analyses were conducted with the full sample (N = 475). First, the relationship between TTM constructs and Stages of Change was evaluated and compared to patterns seen in other areas of behavior change (Figures 1-3). Raw TTM construct scores (see Table 4) were translated to T-scores and weighted by group size to eliminate bias created by uneven Stage groups.
A chi-square test evaluated the association between participants' mental health 13 treatment status (in treatment versus not in treatment) and Stage of Change for Psychosocial Care. ANOVA also evaluated the relationship between Self-Efficacy and Stage of Change. Next, MANOVA evaluated relationships between Decisional Balance and Stage of Change. ANOVA determined whether individuals in the Action/Maintenance stages of change showed different levels of well-being than those in pre-Action stages. Then, regression analyses evaluated relationships between TTM constructs and subjective well-being. Finally, relationships between constructs were evaluated for consistency with patterns seen for other behaviors (e.g., physical activity, cigarette smoking).
Further, 23.4% of the sample reported multiple cancer diagnoses, as a result of metastasis. Given that 57 different cancer diagnoses were reported, variables were recoded such that diagnoses were organized according to organ system/site (e.g., gastrointestinal, gynecologic, skin), as presented by the National Cancer Institute (2016). Additional information regarding the sample's cancer diagnoses and treatment may be found in Table 2.

Exploratory Analyses
Exploratory procedures included PCA with varimax rotations. Sample size (n = 237) was adequate based on existing literature [46]. Decisions regarding retention of components were based on parallel analysis and minimum average partial procedures (MAP), both of which have been found to be accurate methods. Exploratory analyses were used to determine the number of components, the correlation between components, and the loadings of items on these components. Items with poor (<.40) and/or complex loadings (>.40) on more than one factor were removed. In later steps, items with content overlap were removed.

Decisional Balance
Twenty-one decisional balance items were included in the initial exploratory factor analysis. PCA with varimax rotation on the 21 x 21 matrix of item intercorrelations was conducted to determine the factor structure of the decisional balance measure. A total of 6 iterative PCAs were conducted, which reduced the original pool of 21 items to 16, with 8 items reflecting Pros and 8 items reflecting Cons. Parallel analysis indicated a two-factor solution. Examination of the item content revealed that one factor (8 items) clearly reflected the pros of utilizing psychosocial services and one factor (8 items) clearly reflected the cons of utilizing psychosocial services. All item loadings were above 0.522. Internal consistency was excellent for the Pros scale (α = 0.933) and good for the Cons scale (α = 0.809).
Together, the two factors accounted for 56.41% of the total variance (35.66% for Pros and 20.74% for Cons). The retained items can be viewed in Figure 4.

Decisional Balance -Short Form (DB-SF)
For development of the DB-SF, the 16 decisional balance items from the full measure were included in the initial exploratory factor analysis. PCA with varimax rotation on the 16 x 16 matrix of item intercorrelations was conducted to determine the factor structure of the measure. A total of 3 iterative PCAs was conducted, which reduced the original pool of 16 items to 8, with 4 items reflecting Pros and 4 items reflecting Cons. Parallel analysis indicated a two-factor solution, which was retained.
Examination of item content revealed that one factor (4 items) clearly reflected the pros of utilizing psychosocial services and one factor (4 items) reflected the cons of utilizing psychosocial services. All item loadings were above .641. Internal consistency was good for the Pros scale (α = .874) and acceptable for the Cons scale (α = .716). Together, the two factors accounted for 61.94% of the total variance (38.29% for Pros and 23.66% for Cons). The final set of retained items can be viewed in Figure 5.

Self-efficacy
All nine Self-efficacy items were included in the initial exploratory factor analysis. PCA with varimax rotation on the 9 x 9 matrix of items intercorrelations was conducted to determine the factor structure of the measure. Four PCAs were conducted, which reduced the initial pool of nine items to six. MAP and parallel analysis supported a single component solution. However, PCA supported a twocomponent solution. Therefore, the two-factor solution was retained. Examination of the item content revealed that one factor (3 items) clearly reflected the physical challenges to utilizing psychosocial services (α = .904) and one factor (3 items) clearly reflected the social and emotional challenges utilizing psychosocial services (α = .757). Item loadings ranged from .667 to .919. The resulting scale had good internal consistency (α = .826) and accounted for 75.46% of the total variance. The final set of retained items can be found in Figure 6.

Confirmatory Analyses
Confirmatory factor analyses were conducted with the structural equation

Decisional Balance
The following measurement models were compared for the 16-item Decisional Balance measure: (1) a null model that supported 16 independent variables and no latent factors; (2) a single-factor model; (3) a two-factor uncorrelated model; and (4) a two-factor correlated model. Fit indices for each model are summarized in Table 3.
The two-factor correlated model showed the best fit to the data. Factor loadings ranged from .464 to .878. Fit indices suggested good model fit, χ 2 (103) = 349.563, p < .001, CFI = .928, RMSEA = .075. The correlation between the two scales was r = .147 and rho coefficients were excellent for Pros (ρ = .932) and good for Cons (ρ = .816).
The final items and their loadings in the confirmatory subsample are presented in Figure 5.

Decisional Balance -Short Form (DB-SF)
The following measurement models were compared for the 8-item Decisional Balance (SF) measure: (1) a null model that supported 8 independent variables and no latent factors; (2) a one-factor model; (3) a two-factor uncorrelated model; and (4) a two-factor correlated model. Fit indices for each model are summarized in Table 3.
The two-factor correlated model showed the best fit to the data. Factor loadings ranged from .641 to .893. Fit indices suggested good model fit, χ 2 (19) = 68.56, p < .001, CFI = .962, RMSEA = .078. The correlation between the two scales was r = .14 and rho coefficients were good for Pros (ρ = .872) and acceptable for Cons (ρ = .755). The final items and their loadings in the confirmatory subsample are presented in Figure 5.

Self-Efficacy
The following measurement models were compared for the Self-efficacy scale: (1) a null model that supported six independent variables and no latent factors; (2) Table 3.

Self-efficacy by Stages of Change
Multivariate analysis of variance (MANOVA) indicated that Self-Efficacy was significantly different across the Stages of Change, (F(8,906) = 6.18, p < .001, Wilks' l = .899; η 2 = .05). Follow-up ANOVAs indicated significant between-stage differences on the Physical (F(4,456) = 4.31, p < .01, η 2 = .04) and Social/Emotional (F(4,459) = 10.49, p < .001, η 2 = .08) factors. Follow-up comparisons showed that Self-efficacy of individuals in the Precontemplation and Preparation stages was substantially lower than that of those in the Action and Maintenance stages. Weighted T-scores of Self-efficacy at each Stage of Change are presented in Figure 4.

Self-Efficacy
This research supported a two-factor correlated model for Self-Efficacy. This finding diverged from some previous research on Self-Efficacy across other health 24 behavior change, in which a single-factor scale was supported. Nonetheless, Self- Validation analyses further supported the developed Self-Efficacy measure.
Self-Efficacy was significantly different across Stages of Change, demonstrating its utility in facilitating readiness to change. More importantly, Self-Efficacy was significantly lower in the pre-Action stages, compared with the Action stages, validating its role in acceptance of psychosocial care. The relationship between Self-Efficacy and present well-being provided further external validation for this measure, as those with greater confidence for engaging in psychosocial care across a range of challenging situations had greater subjective present well-being. These results indicate 25 that Self-Efficacy may be an essential component for feedback in an intervention or feedback session aimed at reducing reluctance or ambivalence to meet with a mental health provide or to attend a support group.     Note. N = 238; χ 2 = chi square; df = degrees of freedom; CFI = comparative fit index; RMSEA = root mean square error of approximation; CI = confidence interval; *p < .001; **p < .01. Note. Mean = average sum score; higher scores indicate more importance for Pros and Cons and more confidence for Self-Efficacy; SD = standard deviation; scores in parentheses indicate those from the Decisional Balance Short -Form (SF) measure 39     (Woltmann et al., 2012). Similarly, a review of psychosocial interventions for cancer revealed that the biomedical model of disease does not take into account all of the complex factors involved in cancer, underscoring the need for a broader, more integrative framework for cancer care that integrates psychosocial factors (Shapiro et al., 2001). Notably, recent models have converged on the use of multimodal, multidisciplinary interventions to decrease cancer-related morbidity, increase survival rates, improve physical and psychological health outcomes, decrease hospital readmissions, and reduce healthcare costs (e.g., Mehnert & Koch, 2008;Purushotham, et al., 2013;. Findings from a recent study of 1,083 women with breast cancer generated recommendations for patient education, screening for psychosocial distress, and tailoring psychosocial interventions for older women (Mehnert & Koch, 2008). Unfortunately, despite evidence that up to 35% of patients with cancer experience significant distress, only five percent obtain psychological help (National Comprehensive Cancer Network, 2016). Access to effective psychosocial care is often limited by lack of systematic approaches to assessment, scarcity of psychosocial services, and patient reluctance to accept treatment, mainly due to perceived stigma (Zabora et al., 2001). Nonetheless, the literature has evolved to encourage broader and better integrated models of care, rather than treating cancer from a solely biomedical model.

Limitations and Future Directions
Given the limitations of the biomedical model, research supports that multidisciplinary collaborative care teams are more likely to deliver favorable cancer 51 treatment outcomes. Notably, a randomized trial of psychosocial support groups revealed that the use of multidisciplinary collaborations enhanced enrollment rates in psychosocial interventions (Goodwin et al., 2000). Such improvements have often been attributed to having cancer treatment providers (e.g., oncologists, nurses) introduce and recommend behavioral health treatment, thereby increasing engagement. However, due to insufficient behavioral health providers in oncology settings, nurses and oncologists are often expected to screen for patient distress and to provide therapeutic services (Fallowfield, Ratcliffe, Jenkins, & Saul, 2001).
Problematically, Sollner et al. (2001) found that oncologist recommendations for counseling did not correlate with patient distress, implying that oncologists' ability to identify patients in distress is generally insufficient. Additionally, a recent study of 448 oncologists revealed that 38% of their patients experienced psychological distress requiring intervention, but only half of those oncologists had any mental health services affiliated with their practices. Additionally, only 47% made a referral for psychosocial services (Muriel et al., 2015). These data suggests that multidisciplinary teams, representing professionals with different areas of expertise (e.g., nursing, oncology, psychology) may be more likely to deliver effective care and to enhance treatment outcomes. However, providing psychosocial care to cancer patients comes with numerous barriers, including the need for systematic approach to identifying patients with unmet psychosocial needs, as well as provider, patient, financial, and organizational challenges (Fann & Sharpe, 2012).

52
STAR® is a free access, evidence-based program that provides nutrition counseling, physical rehabilitation, caregiver support, monitoring tools, and behavioral health services for cancer patients (Kirschner et al., 2013;Silver, Baima, Mayer, 2013;Silver & Gilchrist, 2011;Silver & Mayer, 2007;Silver, 2007;2010;2011;2013;2014;2014;Silver et al., 2015a,b). The which often generate a self-sustaining service and decrease the economic impact of cancer on patients, caregivers, and the healthcare system; and 5) enhanced community education by supporting local and regional awareness initiatives (STAR®, 2015).

Purpose of Current Study
As the literature supporting the efficacy of biobehavioral cancer care continues to grow (e.g., Lutgendorf & Anderson, 2015) a dearth of reporting on psychosocial cancer care programs prevails. Notably, a recent study revealed the scarcity of studies on the intersection of cancer and mental health and suggested the need to bridge the gap between these two disciplines, in order to prevent and treat mental health problems in cancer patients (Purushotham et al., 2013). Collectively, research has identified: 1) a gap between the need for and delivery of services; 2) that dual screening for psychological distress and physical impairment is critical for optimal outcomes; and 3) integrated rehabilitative services are cost-effective.
Accordingly, the current study describes the development, preliminary evaluation, and utilization of a behavioral health program integrated into routine cancer care at a STAR Program® -certified hospital.

Setting and Program Description
The study took place at an independent, non-profit acute care hospital serving [area masked for anonymous review]. Prior to the beginning of data collection, the program was collaboratively designed by hospital administrators, clinicians, and staff, as well as faculty and two graduate students from a clinical psychology doctoral program at a northeastern, public research university. The goals of this program were to: 1) increase access to behavioral health services to cancer patients; and 2) provide students with a one-year, formally supervised clinical training experience.
In behavioral health staff. The purpose of NCCN administration was to: 1) screen for distress; 2) provide preliminary data for behavioral health staff; and 3) prioritize patient assignments (i.e., in the event of understaffing, patients with a higher distress scores would be seen first).

Measures
Demographics. Patients who accepted psychosocial treatment provided their gender, age, race/ethnicity, marital status, and employment status during the intake. A retrospective chart review was conducted to obtain this information from individuals who only attended support groups or those who were referred, but declined psychosocial care. For all patients, medical information was obtained, including cancer site, cancer stage, and cancer treatment type.
Cancer Staging. The TNM system is one of the most widely used cancer staging systems. It is based on the size and extent of the primary tumor (T), the degree of spreading to nearby lymph nodes (N), and the presence or absence of metastasis (M).
A number is added to each letter to indicate the size or extent of primary tumor and degree of cancer proliferation. Primary Tumor (T) can be noted as TX (primary tumor cannot be evaluated), T0 (no evidence of primary tumor), Tis: Carcinoma in situ (CIS; abnormal cells are present, but have not spread to neighboring tissue and may become 56 cancerous), T1, T2, T3, T4 (size and/or extent of the primary tumor). Regional Lymph Nodes (N) may be noted as NX: Regional lymph nodes cannot be evaluated, N0: No regional lymph node involvement; N1, N2, N3: Degree of regional lymph node involvement. Metastasis (M) may be noted as MX: distant metastasis cannot be evaluated; M0: no distant metastasis; M1: distant metastasis is present (Greene & Sobin, 2002).
Depression. The Beck Depression Inventory (BDI-II) is a 21-item, self-report rating inventory that measures characteristic attitudes and symptoms of depression (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961), such as hopelessness and loss of interest.
Anxiety. The Beck Anxiety Inventory (BAI) is a 21-item self-report inventory for measuring the severity of anxiety, with higher scores reflecting higher anxious symptomatology. It has a high internal consistency (α-.92) and test-retest reliability over 1 week r(81) = .75. The BAI discriminates anxious diagnostic groups (panic disorder, generalized anxiety disorder, etc.) from non-anxious diagnostic groups (major depression, dysthymic disorder, etc.). The BAI has been used in various medical and health settings, including bone marrow transplantation (Manne et al., 2001), coronary artery bypass surgery (Hartford, Wong, & Zakaria, 2002), and chronic pulmonary disease (Cully et al., 2006).

Distress. The National Comprehensive Cancer Network (NCCN) Distress
Thermometer for Cancer Patients is a self-report measure that has been used to identify patients with elevated psychological distress, in various domains, including financial, emotional, and physical, with the patients noting subjective levels of distress, ranging from 0 (no distress) to 10 (extreme distress) (Goebel & Mehdorn, 2011;Holland & Bultz, 2007;Jacobsen et al., 2005;Patel, Sharpe, Thewes, Bell, & Clarke, 2011).

Analytic Plan
All data analyses were conducted using the Statistical Package for the Social Sciences, Version 24.0 (SPSS 24.0). First, a series of chi-square tests of independence were conducted to determine associations between demographic variables and medical characteristics. Second, chi-square tests were run to determine relationships between demographic and medical characteristics, and behavioral health program utilization, respectively. Finally, repeated-measures within-subjects ANOVAs, with Bonferroni corrections, were run to examine changes in depression and/or anxiety between the first and second, as well as first and final sessions with behavioral health providers.
Alpha was established at the .05 level. Additional information regarding patient demographics may be found in Table 1.

Demographic and Medical Characteristics
In addition to sociodemographic information, medical information was collected. The most common cancer diagnoses were breast (18.1%) and lung (10.1%).
Further, 19.1% of the sample reported multiple cancer diagnoses, as a result of metastasis. Given that 29 different cancer diagnoses were prevalent in the sample, variables were recoded such that diagnoses were organized according to organ system/site (e.g., gastrointestinal, gynecologic, skin), as presented by the National Cancer Institute (2016). This revealed that the most common cancer sites were breast (20.4%) and gastrointestinal (16.6%). From the date of program initiation to June 30, 2016, 17 patients (7.14%) who were offered behavioral health treatment passed away.
Additional information regarding the sample's medical characteristics may be found in Table 2.

Treatment Providers
Three years of data revealed that of the 125 patients seen by the behavioral

Program Efficacy
Change in depression and anxiety scores between patients' first and last sessions was calculated (see Figure 2).

Clinical Characteristics and Program Utilization
A series of chi square tests of independence revealed relationships between clinical characteristics and behavioral health utilization. No relationship between cancer organ system site and treatment acceptance was observed, χ 2 (22, n = 237) = 28.96, p = .15. However, a relationship between cancer diagnosis and treatment acceptance was observed, χ 2 (33, n = 237) = 53.55, p = .01, such that being diagnosed with leukemia, lymphoma, multiple myeloma, lung, pancreatic, and gastric cancer was associated with accepting treatment, compared with those diagnosed with cancers with higher survival rates (e.g., breast; NIH, 2016). Further, there was an association between cancer diagnosis and engagement in at least two behavioral health encounters, χ 2 (66, n = 237) = 87.79, p = .04. The association between treatment acceptance and cancer stage was not significant, χ 2 (8, n = 237) = 13.87, p = .09. Chi square tests revealed between receiving chemotherapy and to accepting treatment, compared with those who were receiving multiple cancer treatments, χ 2 (10, n = 237) = 24.50, p < .01. The association between engaging in at least two sessions and receiving only one treatment was also significant, χ 2 (20, n = 237) = 37.77, p < .01.

DISCUSSION
Results supported the development, preliminary efficacy, and overall utilization of the behavioral health program. As such, the present study provides preliminary data and evidence for establishing and maintaining a partnership between a hospital and clinical psychology doctoral program. Results may be used to benchmark other behavioral health integration efforts.

Program Development and Implementation
The behavioral health program was collaboratively designed and implemented,  Williams et al., 2015), the present program involved training for all clinicians, in addition to formal supervision.
Specifically, all clinicians were STAR®-certified and trained in providing evidencebased treatments (EBTs) for cancer patients.

Evaluation
The present study reports only preliminary results on program efficacy.

Program Efficacy
Given challenges with BAI and BDI administration, results on changes in anxiety and depression were limited, by the small sample size over time. Further, the lack of a control or comparison group limited our ability to draw conclusions regarding program efficacy. Nonetheless, these preliminary findings are promising with regard to intervening on depression among cancer patients. Notably, the present study not only revealed significant changes in depression between first and last session (i.e., from mild to minimal levels of depression), but between the first and second session (e.g., from mild to minimal levels depression). This finding is consistent with the psychotherapy literature on large treatment gains that are often observed in the first few sessions (Cooper, 2008). However, the present study revealed statistically nonsignificant findings with regard to changes on anxiety. Given the unique challenges that cancer patients are often faced with, treating anxiety might be especially difficult. For instance, addressing concerns related to fear of tumor progression and beginning new cancer treatments may be particularly anxietyprovoking and difficult to address (e.g., Brix et al., 2008). Nonetheless, a clinically significant decrease in anxiety was observed (i.e., from mild to minimal levels of anxiety).

Patient Demographic and Medical Characteristics
Consistent with epidemiological data on cancer incidence and prevalence, the most common diagnosis in this study sample was breast cancer, followed by lung cancer (National Cancer Institute, 2016). Additionally, consistent with national data, the majority of individuals in this sample had multiple forms of treatment, often surgery as their primary treatment, followed by chemotherapy (National Cancer Institute, 2016).

Program Utilization
Results revealed promise with regard to behavioral health integration in a In order to evaluate this program and its utilization more broadly, we compared its utilization data to that of the general psychotherapy literature. First, the mean number of sessions attended in this study was 2.77, ranging from zero to 96 sessions.
Although it is challenging to make recommendations regarding the number of therapy sessions needed to meet criteria for remission or "recovery", a dose-response relationship does exist (Cooper, 2008). However, it is important to note that sudden treatment gains on acute and symptomatic problems, as would be expected with a cancer patient population, would tend to happen more quickly than change on more longstanding problems (i.e., personality-based diagnoses) (e.g., Cooper, 2008;Kopta et al., 1994). Notably, although the average number of sessions that patients engaged in was few, research has established a 'law of diminishing returns', meaning that as patients have more sessions, the added benefit of each session actually begins to decrease (Cooper, 2008). To illustrate, research has revealed that the degree of improvement between session 53 and 104 is approximately the same as between sessions two and four (Cooper, 2008). The mean number of sessions attended, in the present study, was consistent with the general psychotherapy research, which has demonstrated that on average, patients drop out after just two sessions (Swift & Greenberg, 2012). However, of those who engaged in treatment (n = 119) in the present study, 42% attended three or more sessions, 10.9% attended two sessions, and 47.1% attended only one. The present research revealed that despite the unique challenges that cancer patients face, many committed to more than three sessions, a number greater than what has been observed in the general psychotherapy research.

Demographics and Program Utilization
Results revealed some associations between patient demographics and use of the behavioral health program. First, individuals who were not married were more likely to utilize the program and to engage in two or more visits. Patients who are married might be perceiving their spousal support as sufficient enough to decline psychosocial treatment. Second, patients who were not employed were more likely to accept treatment, suggesting that engaging in psychosocial care may be an additional and demanding time commitment, given investment in work and cancer treatment. For patients who are employed while receiving cancer treatment, providing "bedside" psychosocial care may be especially important, in order to eliminate or minimize the time commitment related to psychosocial care. Finally, patients who were in a younger age group were more likely to engage in psychosocial treatment. This finding may be interpreted in the context of mental health stigma and is consistent with previous research findings on mental health stigma among older age groups (e.g., Brenes et al., 2015;Conner et al., 2010;Sirey et al., 2001). Strategies to address this stigma may include psychoeducation and having oncologists or nurses introduce behavioral health services and its providers. Interestingly, analyses revealed no gender differences with regard to treatment utilization. This finding is contrary to previous findings in which men were less likely to seek or accept psychosocial care (Clement et al., 2015;Vogel et al., 2014).

Clinical Characteristics and Program Utilization
Results revealed some associations, with regard to patient clinical characteristics and use of the behavioral health program. First, a relationship between cancer diagnosis and treatment acceptance and engagement in two or more visits, was observed, such that those with multiple cancer diagnoses, due to metastasis, were more likely to decline treatment. Further, patients with leukemia, lung cancer, lymphoma, multiple myeloma, pancreatic, and gastric cancer were more likely to decline treatment. This finding may be due to decreased survival rates for the aforementioned cancers (CDC, 2016;NIH, 2016), compared with cancers with higher survival rates (e.g., breast), for which patients were more likely to accept psychosocial treatment.
Given that severe levels of psychological distress may interfere with coping and cancer treatment recommendations, it may be especially important for cancer treatment providers (oncologists, nurses) to encourage and support psychosocial care for these individuals. Second, results revealed that patients who were receiving chemotherapy were more likely to engage in psychosocial care than patients receiving multiple cancer treatments (e.g., chemotherapy, radiation). Providing "bedside" psychosocial care may address this barrier, as patients can have individual therapy sessions while receiving chemotherapy, thereby minimizing time commitment.

Limitations and Future Directions
This study has several limitations. First, only a small subset of the sample had baseline well-being data available. An even smaller subset of the sample had followup well-being data available. Future research should implement a systematic approach to progress monitoring, in order to maximize well-being assessment. Second, present findings are based on a sample that is mainly White and non-Hispanic. Given the differences in mental health stigma among non-White populations (Nadeem et al., 2007), additional research examining the utilization of psychosocial services in non-White populations is warranted. Third, no anxiety or depression data was available on patients who declined psychosocial care, thereby limiting our ability to compare groups. Fourth, although the behavioral health program is a segment of the STAR® program that sought to provide comprehensive care, this program was fundamentally not integrated. Specifically, oncologists were often not in contact with behavioral health providers past the initial referral, demonstrating separate, rather than integrated care (Eickmeyer et al., 2013). This, and other programs, should seek to provide an integrated approach, such that oncologists and behavioral health providers exchange clinical data regarding shared patients. Conducting weekly team rounds would be an excellent platform for exchanging crucial patient information that can inform and tailor treatment. Despite its limitations, the study has numerous strengths and may be used to guide future investigations and designs of behavioral health programs. First, this program represents an important step towards improved integration of patient care, using a multidisciplinary care approach to treatment, with enhanced access to psychosocial services and care. Given that psychosocial care of cancer patients has traditionally been viewed as separate from routine medical care, the present study assessed and evaluated the implementation of a more comprehensive approach to cancer. This study also implemented monitoring and maximization of treatment  Table 2.

Depression and Anxiety Scores across Sessions
Note. Total BDI and BAI scores across three time points. The solid line shows the depression scores, while the dashed line shows the anxiety scores. Change in depression scores was significant between the first and second session and between the first and final session. Change in anxiety scores was neither significant between the first and second session, nor between the first and final session.