Key Points
Question
Does a depression screening program for patients with breast cancer treated in community oncology practices using tailored implementation science–guided strategies result in a greater proportion of referrals to behavioral health compared with an education-only strategy?
Findings
In this cluster randomized trial involving 1436 patients with breast cancer at 6 medical centers, a higher proportion of patients at sites randomized to the tailored strategies compared with the educational-only strategy had appropriate referral to behavioral health following screening, 7.9% vs 0.1%, respectively, a difference that was statistically significant.
Meaning
An implementation-strategy guided depression screening program compared with an educational-only strategy resulted in higher proportion of referrals to behavioral health among patients with breast cancer treated in a community setting.
Abstract
Importance
Implementation of guideline-recommended depression screening in medical oncology remains challenging. Evidence suggests that multicomponent care pathways with algorithm-based referral and management are effective, yet implementation of sustainable programs remains limited and implementation-science guided approaches are understudied.
Objective
To evaluate the effectiveness of an implementation-strategy guided depression screening program for patients with breast cancer in a community setting.
Design, Setting, and Participants
A pragmatic cluster randomized clinical trial conducted within Kaiser Permanente Southern California (KPSC). The trial included 6 medical centers and 1436 patients diagnosed with new primary breast cancer who had a consultation with medical oncology between October 1, 2017, through September 30, 2018. Patients were followed up through study end date of May 31, 2019.
Interventions
Six medical centers in Southern California participated and were randomized 1:1 to tailored implementation strategies (intervention, 3 sites, n = 744 patients) or education-only (control, 3 sites, n = 692 patients) groups. The program consisted of screening with the 9-item Patient Health Questionnaire (PHQ-9) and algorithm-based scoring and referral to behavioral health services based on low, moderate, or high score. Clinical teams at tailored intervention sites received program education, audit, and feedback of performance data and implementation facilitation, and clinical workflows were adapted to suit local context. Education-only controls sites received program education.
Main Outcomes and Measures
The primary outcome was percent of eligible patients screened and referred (based on PHQ-9 score) at intervention vs control groups measured at the patient level. Secondary outcomes included outpatient health care utilization for behavioral health, primary care, oncology, urgent care, and emergency department.
Results
All 1436 eligible patients were randomized at the center level (mean age, 61.5 years; 99% women; 18% Asian, 17% Black, 26% Hispanic, and 37% White) and were followed up to the end of the study, insurance disenrollment, or death. Groups were similar in demographic and tumor characteristics. For the primary outcome, 7.9% (59 of 744) of patients at tailored sites were referred compared with 0.1% (1 of 692) at education-only sites (difference, 7.8%; 95% CI, 5.8%-9.8%). Referrals to a behavioral health clinician were completed by 44 of 59 patients treated at the intervention sites (75%) intervention sites vs 1 of 1 patient at the education-only sites (100%). In adjusted models patients at tailored sites had significantly fewer outpatient visits in medical oncology (rate ratio, 0.86; 95% CI, 0.86-0.89; P = .001), and no significant difference in utilization of primary care, urgent care, and emergency department visits.
Conclusions and Relevance
Among patients with breast cancer treated in community-based oncology practices, tailored strategies for implementation of routine depression screening compared with an education-only control group resulted in a greater proportion of referrals to behavioral care. Further research is needed to understand the clinical benefit and cost-effectiveness of this program.
Trial Registration
ClinicalTrials.gov Identifier: NCT02941614
This randomized clinical trial evaluates whether a tailored depression screening program yielded more referrals among patients with breast cancer treated at community medical oncology practices to behavioral health or mental health services than did a clinician education-only strategy.
Introduction
Implementation of guideline-recommended distress screening in medical oncology remains challenging.1 In oncology, distress is a multidimensional construct encompassing depression, anxiety, and other experiences affecting the ability to cope with cancer.1 There is a rich literature documenting the associations of depression with negative outcomes in patients with cancer, particularly breast cancer, including associations with decreased physical and social functioning, increased symptom burden, and poor quality of life,2,3,4 and the global prevalence of clinical depression in breast cancer was estimated to be approximately 30% in data collected from January 1, 2000, through March 20, 2019.5 Recognizing this, screening for depressive and other symptoms is recommended by the American Society of Clinical Oncology and others.6,7 Screening programs for distress are mandated for cancer center accreditation by the American College of Surgeons Commission on Cancer.8 However, depression and depressive symptoms remain underdetected and undertreated in patients with breast cancer.2,9
Although the need for large-scale screening seems intuitive, screening programs incur costs, and there is inadequate knowledge regarding key outcomes.10, While efficacy has been demonstrated in randomized trials at academic centers, typically showing increased number of referrals to psychosocial services,11 there is a paucity of evidence supporting the effectiveness of depression screening programs under routine-practice conditions.12 Oncology clinicians have expressed concerns regarding program acceptability, usefulness, and sustainability, and pilot programs have not been uniformly successful.10 The reduction in benefit in less tightly controlled settings may be due to lack of thorough consideration of local context and resources relevant to program implementation.13 Implementation science–guided studies have been largely overlooked.14,15 Implementation strategies that are feasible and responsive to local context may be critical elements of program adoption and sustainment.3,13,16 The purpose of this study was to evaluate whether a depression screening program for patients with breast cancer in community medical oncology practices using tailored implementation science–guided strategies resulted in a greater proportion of appropriate referrals to behavioral health compared with an education-only strategy.
Methods
Setting and Participants
We conducted the trial at 6 medical centers within Kaiser Permanente Southern California (KPSC), an integrated health care system providing comprehensive care to more than 4.5 million members. Membership in KPSC broadly reflects the socioeconomic and racial and ethnic diversity of Southern California.17 This study received approval from the KPSC Institutional Review Board. All patients with a new diagnosis of breast cancer and a consultation in medical oncology from October 1, 2017, through September 30, 2018, were included, with no exclusions by stage of disease, histology, sex, race and ethnicity, comorbidities, or other clinical or demographic characteristics. The study received a waiver for individual patient consent (passive enrollment). We followed the Consolidated Standards of Reporting Trials reporting guideline for cluster randomized trials (Figure 1). The trial protocol and statistical analysis plan are available in Supplement 1.
Figure 1. Flow of Participants Through the Trial.
Randomization
The principal statistician used SAS version 9.3 (SAS Institute Inc) to generate the randomization scheme for the 6 sites. Clusters were at the medical center level. Although cluster randomization may be less statistically efficient than individual randomization, it allowed evaluating the effectiveness of the program while avoiding contamination. Additionally, it balanced medical center–level factors that were otherwise unable to be reliably measured. Outcomes were analyzed at the individual patient level.
Study Design
We used an effectiveness-implementation hybrid study design.18 Hybrid designs have a priori focus on simultaneously assessing outcomes relevant to clinical effectiveness and implementation. The Consolidated Framework for Implementation Research (CFIR), which consists of 5 domains (intervention characteristics, inner setting, outer setting, individual characteristics, and implementation processes) was used to guide critical elements of the design. This included selection of the screening instrument and workflow adaptability (intervention characteristics), engagement of key clinical and administrative stakeholders during study development and planning (inner setting, process), and building on clinician self-efficacy and knowledge regarding the program (individual characteristics).19,20 The CFIR was also used for planning, coding, and analysis of qualitative data collected on implementation, not reported in this article. We used the Expert Recommendations for Implementing Change (ERIC) strategies21 to identify and select implementation strategies during study planning. With input from clinical and administrative staff, we selected ERIC strategies considered appropriate for the scientific question, feasible to use and familiar to staff, and replicable at scale using health system resources beyond the study time frame.
This trial used a pragmatic approach, guided by the Pragmatic-Explanatory Continuum Indicator Summary 2 (PRECIS-2). Our research question, design, and methods aligned with pragmatic trial methodological standards.22,23 The trial had several design elements to maximize both the utility of our findings and generalizability as described below.
Intervention
The depression screening program followed guideline recommendations,24 offering screening with the 9-item Patient Health Questionnaire (PHQ-9) to all newly diagnosed patients with breast cancer, with repeated screening encouraged at follow-up visits. We used the PHQ-9 scoring rubric, for which 0 to 9 was mild; 10 to 19, moderate; and 20 or more, severe.25 Patients with mild scores received general information about KPSC and community behavioral health resources. Patients with moderate scores were referred to either the oncology licensed clinical social worker (LCSW), depression care management (staffed with LCSWs and nurse practitioners), or both. Patients with severe scores were directly referred to behavioral health (psychiatry or psychology), provided with an immediate telephone crisis consultation, or both, as appropriate. Program education consisted of up to 4 education sessions: 1 in-person site visit and 2 to 3 teleconference calls.
Tailored Implementation Strategies
We selected 3 implementation strategies for the intervention sites: tailored audit and feedback, facilitation, and adaptable workflow. Audit and feedback is often a necessary element for implementing practice change but may not be sufficient to sustain practice change alone.26 Sites received weekly emails with tailored anonymized audit and feedback reports of progress compared with the other intervention sites: proportion eligible, proportion screened, proportion appropriately referred graphed over time. Facilitation is a guided interactional process to aid implementation and sustainment of practice change.27 A nurse researcher led the facilitation activities, consisting of monthly teleconference check-ins and quarterly in-person site visits to address issues (eg, staff turnover, technical problems) with each tailored-intervention site individually. The nurse received training materials and mentoring in facilitation from the study principal investigator (Hahn). Clinical workflows at each site were adapted to address unique local context and resources. The critical functions of the program—offering screening and using the scoring rubric for appropriate referral—were mandated; the forms that the screening took were adaptable. For example, the screening could be given on paper or entered directly into the electronic medical record (EMR) by a nurse or the timing of the screening could be before or after measuring vital signs.
Screening Instrument
The screening instrument was the PHQ-9, a widely used instrument that has been validated for cancer distress screening28,29 and was available within the EMR. The PHQ-9 was in use in other KPSC departments (behavioral health, obstetrics, primary care).
Education-Only Control
Education-only controls sites were provided with general education about the screening program at study initiation; the PHQ-9 questionnaire and scoring-referral algorithm were available to control sites to use at their discretion. This is comparable with the approach often used for program implementation outside of research studies, which typically feature initial education for clinic staff for knowledge building but lack ongoing support.15
Outcomes and Data Collection
The primary outcome was the percentage of eligible patients screened and referred (based on PHQ score) at the tailored-intervention vs the education-only sites. Secondary outcomes included the proportion with complete referral, defined as receiving any type of visit (telephone, video, in-person) with a behavioral health clinician; and outpatient utilization for oncology, primary care, urgent care, and emergency department (ED). There is little known about the effect of a depression screening program on utilization for outpatient and ED clinical visits, and there was not an a priori hypothesis for the findings. It is possible that a program designed to identify and refer patients with symptoms of depression could decrease utilization for these services, but utilization could increase due to patients’ seeking mental health care from their oncologist or primary care physician or from the urgent care or ED setting. We measured utilization using rates of outpatient visits to medical oncology, primary care, and urgent and ED care, measured from the time of their initial medical oncology consultation through May 31, 2019, death, or disenrollment. Patient-reported outcomes were also measured but are not presented in this article.
Post Hoc Analysis
The prespecified primary outcome was limited to measuring referrals to and visits with behavioral health services only in patients screened with the PHQ-9, which may have biased the study toward a positive finding. To address this, a post hoc analysis was included to measure all referrals and visits to behavioral health in all participants regardless of screening status. Referring department (oncology, primary care, or other specialty care) was also examined.
Covariates
In accordance with the pragmatic nature of the trial, all covariates were based on data available within the EMR: patient age, sex, partner status, race and ethnicity, preferred language, insurance type, census-track education and income (linked geocoded data), cancer stage, and Charlson Comorbidity Index categorized into 3 groups: 0, low; 1-3, medium; and 4 or more, high. Race and ethnicity were prespecified to be included in the adjusted models for outpatient utilization to account for confounding due to sociodemographic characteristics and were self-reported by the patient in the EMR via fixed categories. Cancer stage was obtained through the KPSC pathology database and verified via chart review as needed.
Sample Size Calculation
The sample size for the primary outcome was calculated based on estimates of annual incident breast cancer cases within participating KPSC centers. Because this was an effectiveness study, we assumed relatively small, standardized effect sizes, z scores ranging from 0.2 to 0.4 for all patient-reported outcomes. The power analysis was conducted using methods described by Donner and Klar30 and implemented in PASS.31 To achieve 80% power with a significance level of .05 for a score test, our per-center sample size requirement ranged from as few as 20 for large effect size (0.4) and a 0 intraclass correlation (ICC) to about 400 for a small effect size and large ICC (0.2 and 0.01, respectively) and assumed equal cluster sizes.30 Given an expectation of a total of 1200 patients across the 6 centers, the study was adequately powered to detect effect sizes as small as 0.2, given that the ICCs were no greater than 0.007.
Statistical Analysis
For the primary analyses, patients were analyzed according to their randomization group and all patients at intervention and control sites were included regardless of participation in screening. Follow-up time was 12 months from the date of the initial oncology consultation for the primary outcome and up to 18 months for secondary outcomes. For patients who died or disenrolled from the health plan, data before disenrollment were used, and the shorter duration of follow-up was incorporated into the analysis. All patients had data on primary outcomes. Patients with unknown race or ethnicity were grouped with the other/missing category. An unknown category was created for those with missing cancer stage. Patients with missing information on partner status were categorized as unpartnered. Patient characteristics were compared between intervention and control sites using means, standard deviations, frequencies, and percentages. Comparisons were made using t tests of Wilcoxon rank sum tests for continuous variables or χ2 or Fisher exact tests for discrete variables, along with reporting of confidence limits. The prespecified primary analysis compared rates of PHQ-9 screening completion and referral to and completion of visits with behavioral health between the intervention and control sites at the patient level using risk differences and Wald asymptotic confidence limits.
The prespecified secondary analysis compared health care utilization between the groups. For utilization, we restricted to participants with 100 days or more of KPSC insurance membership following their initial visit to medical oncology. We used multivariable Poisson regression to assess the association between the intervention and outcomes, accounting for variable length of follow-up with an offset parameter and using robust standard errors to correct for potential variance misspecification. All statistical tests were 2-tailed and considered statistically significant if P ≤ .05. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary end points should be interpreted as exploratory. Analyses were conducted using SAS version 9.3.
Results
Participants
We enrolled a total of 1436 patients with 744 patients in the intervention group and 692 in control. Figure 1 shows the participant flow. The mean age was 61.5 years (SD, 12.9 years), 99% were women, the mean Charlson Comorbidity Index was 2.2 (SD, 2.7), 87% spoke English as their primary language followed by 9% Spanish, and 4% other. Eighteen percent self-reported being Asian or Pacific Islander, 17% Black, 26% Hispanic, and 37% White. Eighty-two percent had stage 0 to stage II breast cancer. Groups were balanced on all characteristics (Table 1).
Table 1. Demographics and Cancer Characteristics of Participants.
| No. (%) of participants | ||
|---|---|---|
| Tailored intervention (n = 744) | Education only (n = 692) | |
| Age, mean (SD), y | 61.1 (12.4) | 62.0 (13.3) |
| Sex | ||
| Women | 740 (99) | 689 (99) |
| Men | 4 (<1) | 3 (<1) |
| Stable partner status, No./total (%)a | 428/740 (58) | 382/681 (56) |
| Race and ethnicityb | ||
| No. | 743 | 691 |
| Asian or Pacific Islander | 149 (19) | 117 (17) |
| Black | 98 (13) | 139 (20) |
| Hispanic | 204 (27) | 172 (25) |
| Multiracial | 2 (<1) | 2 (<1) |
| White | 278 (37) | 255 (37) |
| Other | 12 (2) | 6 (1) |
| Preferred language | ||
| English | 639 (86) | 607 (88) |
| Spanish | 75 (10) | 58 (8) |
| Other | 30 (4) | 27 (4) |
| Insurance type | ||
| Commercial or private pay | 440 (59) | 383 (55) |
| Medicare | 273 (37) | 265 (38) |
| Medicaid | 20 (3) | 25 (4) |
| Dual Medicare and Medicaid | 10 (1) | 15 (2) |
| Non–Kaiser Permanente insurance | 1 (<1) | 4 (1) |
| Census track education | ||
| No. | 723 | 676 |
| ≤High school | 271 (37) | 251 (37) |
| Some college | 210 (29) | 196 (29) |
| ≥College graduate | 242 (33) | 229 (34) |
| Census track family income ≥$50 000 | 450/723 (62) | 420/676 (62) |
| Charlson Comorbidity Index scorec | ||
| No. | 743 | 690 |
| Median (IQR) | 1.0 (0.0-4.0) | 1.0 (0.0-3.0) |
| Breast cancer staged | ||
| No. | 712 | 675 |
| Early stage (0-IIb) | 617 (87) | 559 (84) |
| Late stage (III-IV) | 95 (13) | 106 (16) |
Married or living together as committed partners.
Multiple race and Other were self-reporting options in the electronic medical record.
The weighted Charlson Comorbidity Index is a method of mortality prediction based on comorbidities captured in the electronic medical record or administrative data using International Classification of Diseases (ICD) codes. In this study the score is based on ICD codes for the year prior to the initial consult and ranges from 0 to 29; the variable has been categorized into 0, low; 1 to 3, moderate; and 4 or greater, high.
Stage was unknown for 32 patients in the tailored intervention group and 27 in the education-only group.
During the study period, 28 participants died: 19 intervention, 9 control (difference, 1.3%; 95% CI, −0.2% to 2.7%); 93 disenrolled from the health plan: 51 intervention, 42 control (difference, 0.8%; 95% CI, −1.8% to 3.3%). Deaths were due to metastatic cancer or other advanced comorbid conditions. Within the intervention group, there was no significant difference in PHQ-9 score between those who died and those who did not: difference, 0.7 (95% CI, −1.5 to 2.9).
Primary Outcome
Over the study period, 59 out of 744 patients (7.9%) eligible for screening received a referral to behavioral health services at tailored intervention sites; 1 out of 692 patients (0.1%) was referred at education-only sites (difference, 7.8%, 95% CI, 5.8%-9.8%) (Table 2).
Table 2. Percent of Eligible Participants Screened and Referred (Based on 9-Item Patient Health Questionnaire Score) at Tailored Intervention vs Education-Only Sites.
| Primary outcome | Tailored intervention sites (n = 744) | Education-only sites (n = 692) | Risk difference (95% CI) |
|---|---|---|---|
| Received a referral to behavioral health services, No. (%) | 59 (0.079) | 1 (0.001) | 0.078 (0.058-0.097) |
| PHQ-9 score distribution (% based on No. screened)a | |||
| Low | 533 (89) | 2 (67) | |
| Moderate | 57 (10) | 1 (33) | |
| High | 6 (1) | 0 |
Abbreviation: PHQ-9, 9-item Patient Health Questionnaire.
Scores were calculated based on participant responses to each of the 9 items. Each item was scored on a scale of 0 to 3 (0, not at all; 1, several days; 2, more than a week; and 3, nearly every day). The PHQ-9 total score ranges from 0 to 27: 0 to 9, low; 10 to 19, moderate; and 20 or greater, high.
Secondary Outcomes
Behavioral Health Referrals and Utilization
Five hundred ninety-six patients (80%) at the tailored sites were offered PHQ-9 screening during the consultation appointment vs 3 (<1%) at control sites (difference, 79.7%; 95% CI, 76.8%-82.6%). Of the tailored site screenings, 63 patients (11%) scored in the moderate or high range indicating need for immediate referral; 94% received an appropriate referral (patients with moderate scores were referred to oncology LCSW or depression care management; patients with high scores, to the psychology or psychiatry department) and 6% either declined or were not offered a referral. Of those referred, 75% completed a visit with a behavioral health clinician and 25% either declined to schedule, cancelled, or did not show. Of the 3 screened patients at education-only sites, 2 scored in the low and 1 in the moderate range; the patient with a moderate score was referred to and completed a visit with an LCSW.
Utilization
Within the utilization cohort, the mean follow up time for the 730 participants in the tailored group was 1.15 years compared with 1.14 years among the 683 in the education-only group (difference, 0.003; 95% CI, −0.028 to 0.034). Participant characteristics for the utilization cohort were similar between the groups (eTable 1 in Supplement 2).
In unadjusted comparisons, the rate difference per person-year of outpatient oncology visits at tailored intervention sites vs education-only sites was −1.81 (95% CI, −2.11 to −1.51); for outpatient primary care, 0.04 (95% CI, −0.17 to 0.24); for urgent care, −0.18 (95% CI, −0.27 to −0.09); and for ED visits, 0.04 (95% CI, −0.04 to 0.12). In models adjusted for age, race and ethnicity, cancer stage, partner status, and Charlson comorbidity index, patients at tailored intervention sites had statistically significantly fewer outpatient visits in medical oncology (adjusted rate ratio (RR), 0.86; 95% CI, 0.86 to 0.89; P = .001) (Figure 2). There was no statistically significant difference in primary care (adjusted RR, 1.07; 95% CI, 0.93 to 1.24), urgent care (adjusted RR, 0.84; 95% CI, 0.51 to 1.38), or ED visits (adjusted RR, 1.16; 95% CI, 0.84 to 1.62).
Figure 2. Adjusted Rate Ratios for Outpatient Utilization of Primary Care, Medical Oncology, Urgent Care, and Emergency Department Visits.
Visits were compared between intervention and control group, restricted to those with at least 100 days of Kaiser Permanente insurance membership from the date of their cancer diagnosis; models were adjusted for age, race and ethnicity, marital status, Charlson Comorbidity Index score, and cancer stage; the median follow-up time per patient in the tailored intervention group was 1.14 years (IQR, 0.89-1.39 years) vs 1.12 years (IQR, 0.89-1.40 years) in the education-only control group.
Post Hoc Outcomes
Regardless of PHQ-9 screening, a significantly greater number of patients in the tailored intervention group received a referral for any behavioral health service compared with patients treated at the education-only clinics during the study period: 135 patients (18%) vs 74 (11%), difference, 7.5% (95% CI, 3.7% to11.2%) (Table 3; referral-level data are included in eTable 2 in Supplement 2). Broken out by referral to psychiatry, depression care management, social services (services provided by LCSWs), and external (non–Kaiser Permanente) behavioral health referral, a greater number of intervention group patients received all referral types with the exception of external referrals: 0 patients in the tailored group vs 13 patients in education-only received external referral for a difference of −1.9% (95% CI, −3.0%, to −0.7%). A significantly greater number of patients treated at the tailored intervention clinics received referrals to behavioral health generated from the oncology department: 97 (59%) vs 23 (26%) for a difference of 32.7% (95% CI, 19.9% to 45.4%) (eTable 3 in Supplement 2).
Table 3. Total Number of Referrals and Visits Made to Behavior Health, Depression Care Management, Psychiatry, and Social Services (Services Provided by Licensed Clinical Social Worker) From Initial Consult to May 2019 by Group.
| Referralsa | Visits | |||||
|---|---|---|---|---|---|---|
| No. (%) | Difference, % (95% CI) | No. (%) | Difference, % (95% CI) | |||
| Tailored intervention (n = 744) | Education only (n = 692) | Tailored intervention (n = 744) | Education only (n = 692) | |||
| Any behavioral healthb | 135 (18) | 74 (11) | 7.5 (3.7 to 11.2) | 75 (10) | 36 (5) | 4.9 (2.0 to 7.7) |
| Depression care management | 70 (9) | 18 (3) | 6.8 (4.3 to 9.4) | 24 (3) | 1 (<1) | 3.1 (1.6 to 4.5) |
| Psychiatry | 29 (4) | 23 (3) | 0.6 (−1.5 to 2.6) | 16 (2) | 10 (1) | 0.7 (−0.8 to 2.2) |
| Social services | 61 (8) | 33 (5) | 3.4 (0.8 to 6.1) | 44 (6) | 27 (4) | 2.0 (−0.4 to 4.4) |
| Behavioral health, external referralc | 0 | 13 (2) | −1.9 (−3.0 to −0.7) | |||
Percent derived from the group total, and row counts are patient level (post hoc analysis).
Patients may have received referrals and visits to more than one behavioral health resource.
External referrals to non–Kaiser Permanente clinicians.
Missing Data
For the primary and utilization outcomes, there were no missing data. For covariates used in the Poisson models, 3 patients were missing information on race and ethnicity, 59 on cancer stage, 15 on partner status.
Discussion
Among patients with breast cancer treated in community-based medical oncology practices, a tailored implementation strategy–guided depression screening program compared with an education-only strategy resulted in a greater proportion of referrals to appropriate behavioral care.
The proportion of eligible patients screened at the tailored intervention sites was high (80%); in Commission on Cancer–accredited institutions, rates of adherence to distress screening protocols varied from 47% to 73% of eligible participants.32 Internationally, similar rates have been documented: approximately 62% of clinicians reported engaging in any distress screening in Australia.33 Forty percent to 60% of eligible patients in Cancer Care Ontario Regional Cancer Centers were screened.1 Given the high burden of depression in patients with breast cancer, effective screening and referral programs are needed.34 In the current era of heightened health-related concerns due to SARS-CoV-2, which may disproportionately affect patients with cancer and survivors, systematic depression screening and referral for patients with cancer may be even more important.35,36
The strategies for the tailored implementation group (facilitation, audit and feedback, adaptability) were selected for feasibility of use during the trial and sustainability using health systems resources. The strategy of facilitation is likely to be replicable and scalable within the KPSC system. However, other studies have noted multiple implementation challenges including intervention complexity, unrealistic workload or workflow, lack of guidance for assessment or management of high scores, and lack of staff engagement.13,16,32,37 It is possible that these issues could be addressed by engaging local stakeholders to codesign a feasible and sustainable workflow adaptive to available resources and context.38 The US Preventive Services Task Force39 recommends depression screening in adult primary care settings, and a 2016 evidence review that includes multiple randomized trials concluded that the evidence supports the benefits of primary care–based screening in the general adult population. However, these programs can have the same implementation-related barriers as seen in the oncology setting.39 Other types of implementation strategies (eg, financial, policy, restructuring) may provide benefit in different settings.21
Analysis of secondary outcomes found significantly less outpatient oncology utilization in the intervention group and no significant difference between the groups for primary care, urgent care, or ED visits. These results suggest that this type of screening program may not lead to increased health care utilization, although the study would need replication in other settings. Few studies have examined utilization in this context; a 2017 study found that cancer centers with high adherence to an oncology distress screening protocol had significantly less ED utilization (RR, 0.82) and fewer hospitalizations (RR, 0.81).40
Limitations
This study has several limitations. First, the analysis did not include mental health clinical outcomes for patients referred to behavioral health services. Thus, it is unknown whether patients referred had an improved clinical outcome compared with those who were not referred. Not all patients who are referred will have a clinically important benefit, so the magnitude of the benefit of screening cannot be inferred from these findings. Second, restricting the primary analysis of behavioral health referrals and visits to patients screened with the PHQ-9 biased the study to have a positive result. However, a post hoc analysis that included all behavioral health referrals and visits regardless of screening status had consistent findings. Third, the focus on patients with breast cancer may limit generalizability to other cancer types. Fourth, the integrated nature of KPSC may limit generalizability to other clinical settings, such as academic centers or stand-alone oncology centers. Fifth, the study did not screen for financial distress, which is increasingly recognized as an important dimension of cancer-related distress. Sixth, fewer patients had high levels of depressive symptoms than what has been reported in other studies (eg, studies using the National Comprehensive Cancer Center Distress Thermometer have found 37% to 64% of patients with cancer experienced high distress32,41); this could be due to variability in the screening instrument (eg, depression screening vs anxiety or distress screening), cancer type, cancer stage, or clinical setting (eg, academic center vs community).
Conclusions
Among patients with breast cancer treated in community-based medical oncology practices, tailored strategies for implementation of routine depression screening compared with an education-only control group resulted in a greater proportion of referrals to behavioral care. Further research is needed to understand the clinical benefit and cost-effectiveness of this program.
Trial Protocol
eTable 1. Demographics and cancer characteristics of the utilization cohort, restricted to those with ≥100 days of Kaiser Permanente insurance membership from date of cancer diagnosis, N=1,436
eTable 2. Referrals and visits made to Behavior Health, Depression Care Management, Psychiatry, and Social Services
eTable 3. Referring department for all behavioral health referrals from initial consult to May 31, 2019
Data Sharing Statement
References
- 1.Dudgeon D, King S, Howell D, et al. Cancer Care Ontario’s experience with implementation of routine physical and psychological symptom distress screening. Psychooncology. 2012;21(4):357-364. doi: 10.1002/pon.1918 [DOI] [PubMed] [Google Scholar]
- 2.Adler N, Page A. Cancer Care for the Whole Patient: Meeting Psychosocial Health Needs. National Academies Press; 2008. [PubMed] [Google Scholar]
- 3.Ehlers SL, Davis K, Bluethmann SM, et al. Screening for psychosocial distress among patients with cancer: implications for clinical practice, healthcare policy, and dissemination to enhance cancer survivorship. Transl Behav Med. 2019;9(2):282-291. doi: 10.1093/tbm/iby123 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Suppli NP, Johansen C, Christensen J, Kessing LV, Kroman N, Dalton SO. Increased risk for depression after breast cancer: a nationwide population-based cohort study of associated factors in Denmark, 1998-2011. J Clin Oncol. 2014;32(34):3831-3839. doi: 10.1200/JCO.2013.54.0419 [DOI] [PubMed] [Google Scholar]
- 5.Pilevarzadeh M, Amirshahi M, Afsargharehbagh R, Rafiemanesh H, Hashemi SM, Balouchi A. Global prevalence of depression among breast cancer patients: a systematic review and meta-analysis. Breast Cancer Res Treat. 2019;176(3):519-533. doi: 10.1007/s10549-019-05271-3 [DOI] [PubMed] [Google Scholar]
- 6.Andersen BL, DeRubeis RJ, Berman BS, et al. ; American Society of Clinical Oncology . Screening, assessment, and care of anxiety and depressive symptoms in adults with cancer: an American Society of Clinical Oncology guideline adaptation. J Clin Oncol. 2014;32(15):1605-1619. doi: 10.1200/JCO.2013.52.4611 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Riba MB, Donovan KA, Andersen B, et al. Distress management, version 3.2019, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2019;17(10):1229-1249. doi: 10.6004/jnccn.2019.0048 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.American College of Surgeons . Psychosocial distress screening. In: Cancer Program Standards: Ensuring Patient-Centered Care. 2016. ed. American College of Surgeons; 2016:56–57. Accessed April 9, 2021. https://www.facs.org/quality-programs/cancer/coc/standards
- 9.Stanton AL. Psychosocial concerns and interventions for cancer survivors. J Clin Oncol. 2006;24(32):5132-5137. doi: 10.1200/JCO.2006.06.8775 [DOI] [PubMed] [Google Scholar]
- 10.Mitchell AJ, Lord K, Slattery J, Grainger L, Symonds P. How feasible is implementation of distress screening by cancer clinicians in routine clinical care? Cancer. 2012;118(24):6260-6269. doi: 10.1002/cncr.27648 [DOI] [PubMed] [Google Scholar]
- 11.McCarter K, Britton B, Baker AL, et al. Interventions to improve screening and appropriate referral of patients with cancer for psychosocial distress: systematic review. BMJ Open. 2018;8(1):e017959. doi: 10.1136/bmjopen-2017-017959 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Carlson LE, Waller A, Mitchell AJ. Screening for distress and unmet needs in patients with cancer: review and recommendations. J Clin Oncol. 2012;30(11):1160-1177. doi: 10.1200/JCO.2011.39.5509 [DOI] [PubMed] [Google Scholar]
- 13.Jacobsen PB, Norton WE. The role of implementation science in improving distress assessment and management in oncology: a commentary on “screening for psychosocial distress among patients with cancer: implications for clinical practice, healthcare policy, and dissemination to enhance cancer survivorship”. Transl Behav Med. 2019;9(2):292-295. doi: 10.1093/tbm/ibz022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lazenby M, Ercolano E, Grant M, Holland JC, Jacobsen PB, McCorkle R. Supporting commission on cancer-mandated psychosocial distress screening with implementation strategies. J Oncol Pract. 2015;11(3):e413-e420. doi: 10.1200/JOP.2014.002816 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.McCarter K, Fradgley EA, Britton B, Tait J, Paul C. Not seeing the forest for the trees: a systematic review of comprehensive distress management programs and implementation strategies. Curr Opin Support Palliat Care. 2020;14(3):220-231. doi: 10.1097/SPC.0000000000000513 [DOI] [PubMed] [Google Scholar]
- 16.Smith SK, Loscalzo M, Mayer C, Rosenstein DL. Best practices in oncology distress management: beyond the screen. Am Soc Clin Oncol Educ Book. 2018;38:813-821. doi: 10.1200/EDBK_201307 [DOI] [PubMed] [Google Scholar]
- 17.Koebnick C, Langer-Gould AM, Gould MK, et al. Sociodemographic characteristics of members of a large, integrated health care system: comparison with US Census Bureau data. Perm J. 2012;16(3):37-41. doi: 10.7812/TPP/12-031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Curran GM, Bauer M, Mittman B, Pyne JM, Stetler C. Effectiveness-implementation hybrid designs: combining elements of clinical effectiveness and implementation research to enhance public health impact. Med Care. 2012;50(3):217-226. doi: 10.1097/MLR.0b013e3182408812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Damschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC. Fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science. Implement Sci. 2009;4:50. doi: 10.1186/1748-5908-4-50 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Damschroder LJ, Hagedorn HJ. A guiding framework and approach for implementation research in substance use disorders treatment. Psychol Addict Behav. 2011;25(2):194-205. doi: 10.1037/a0022284 [DOI] [PubMed] [Google Scholar]
- 21.Powell BJWT, Waltz TJ, Chinman MJ, et al. A refined compilation of implementation strategies: results from the Expert Recommendations for Implementing Change (ERIC) project. Implement Sci. 2015;10(21):21. doi: 10.1186/s13012-015-0209-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Loudon K, Treweek S, Sullivan F, Donnan P, Thorpe KE, Zwarenstein M. The PRECIS-2 tool: designing trials that are fit for purpose. BMJ. 2015;350:h2147. doi: 10.1136/bmj.h2147 [DOI] [PubMed] [Google Scholar]
- 23.Thorpe KE, Zwarenstein M, Oxman AD, et al. A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers. J Clin Epidemiol. 2009;62(5):464-475. doi: 10.1016/j.jclinepi.2008.12.011 [DOI] [PubMed] [Google Scholar]
- 24.Pirl WF, Fann JR, Greer JA, et al. Recommendations for the implementation of distress screening programs in cancer centers: report from the American Psychosocial Oncology Society (APOS), Association of Oncology Social Work (AOSW), and Oncology Nursing Society (ONS) joint task force. Cancer. 2014;120(19):2946-2954. doi: 10.1002/cncr.28750 [DOI] [PubMed] [Google Scholar]
- 25.Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606-613. doi: 10.1046/j.1525-1497.2001.016009606.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ivers NM, Grimshaw JM, Jamtvedt G, et al. Growing literature, stagnant science? systematic review, meta-regression and cumulative analysis of audit and feedback interventions in health care. J Gen Intern Med. 2014;29(11):1534-1541. doi: 10.1007/s11606-014-2913-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Berta W, Cranley L, Dearing JW, Dogherty EJ, Squires JE, Estabrooks CA. Why (we think) facilitation works: insights from organizational learning theory. Implement Sci. 2015;10(1):141. doi: 10.1186/s13012-015-0323-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Smith AB, Rush R, Wright P, Stark D, Velikova G, Sharpe M. Validation of an item bank for detecting and assessing psychological distress in cancer patients. Psychooncology. 2009;18(2):195-199. doi: 10.1002/pon.1423 [DOI] [PubMed] [Google Scholar]
- 29.Mitchell AJ. Short screening tools for cancer-related distress: a review and diagnostic validity meta-analysis. J Natl Compr Canc Netw. 2010;8(4):487-494. doi: 10.6004/jnccn.2010.0035 [DOI] [PubMed] [Google Scholar]
- 30.Donner A, Klar N. Statistical considerations in the design and analysis of community intervention trials. J Clin Epidemiol. 1996;49(4):435-439. doi: 10.1016/0895-4356(95)00511-0 [DOI] [PubMed] [Google Scholar]
- 31.Hintze J. PASS 13: Power Analysis and Sample Size. NCSS Statistical Software; 2014. Accessed April 9, 2021. http://www.ncss.com
- 32.Zebrack B, Kayser K, Sundstrom L, et al. Psychosocial distress screening implementation in cancer care: an analysis of adherence, responsiveness, and acceptability. J Clin Oncol. 2015;33(10):1165-1170. doi: 10.1200/JCO.2014.57.4020 [DOI] [PubMed] [Google Scholar]
- 33.Fradgley EA, Byrnes E, McCarter K, et al. A cross-sectional audit of current practices and areas for improvement of distress screening and management in Australian cancer services: is there a will and a way to improve? Support Care Cancer. 2020;28(1):249-259. doi: 10.1007/s00520-019-04801-5 [DOI] [PubMed] [Google Scholar]
- 34.Burgess C, Cornelius V, Love S, Graham J, Richards M, Ramirez A. Depression and anxiety in women with early breast cancer: five year observational cohort study. BMJ. 2005;330(7493):702. doi: 10.1136/bmj.38343.670868.D3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Liang W, Guan W, Chen R, et al. Cancer patients in SARS-CoV-2 infection: a nationwide analysis in China. Lancet Oncol. 2020;21(3):335-337. doi: 10.1016/S1470-2045(20)30096-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Nekhlyudov L, Duijts S, Hudson SV, et al. Addressing the needs of cancer survivors during the COVID-19 pandemic. J Cancer Surviv. 2020;14(5):601-606. doi: 10.1007/s11764-020-00884-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Donovan KA, Grassi L, Deshields TL, Corbett C, Riba MB. Advancing the science of distress screening and management in cancer care. Epidemiol Psychiatr Sci. 2020;29:e85. doi: 10.1017/S2045796019000799 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Fitch MI, Ashbury F, Nicoll I. Reflections on the implementation of screening for distress (sixth vital sign) in Canada: key lessons learned. Support Care Cancer. 2018;26(12):4011-4020. doi: 10.1007/s00520-018-4278-y [DOI] [PubMed] [Google Scholar]
- 39.O’Connor E RR, Henninger M, Groom HC, Burda BU, Henderson JT, Bigler KD, Whitlock EP. Screening for depression in adults: an updated systematic evidence review for the US Preventive Services Task Force Agency for Healthcare Research and Quality; 2016. Vol Evidence Synthesis No. 128. AHRQ Publication No. 14-05208-EF-1. [PubMed]
- 40.Zebrack B, Kayser K, Bybee D, et al. A practice-based evaluation of distress screening protocol adherence and medical service utilization. J Natl Compr Canc Netw. 2017;15(7):903-912. doi: 10.6004/jnccn.2017.0120 [DOI] [PubMed] [Google Scholar]
- 41.Mehnert A, Hartung TJ, Friedrich M, et al. One in two cancer patients is significantly distressed: Prevalence and indicators of distress. Psychooncology. 2018;27(1):75-82. doi: 10.1002/pon.4464 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
eTable 1. Demographics and cancer characteristics of the utilization cohort, restricted to those with ≥100 days of Kaiser Permanente insurance membership from date of cancer diagnosis, N=1,436
eTable 2. Referrals and visits made to Behavior Health, Depression Care Management, Psychiatry, and Social Services
eTable 3. Referring department for all behavioral health referrals from initial consult to May 31, 2019
Data Sharing Statement