Abstract
Significance
Prior research suggests that Community Engagement and Planning (CEP) for coalition support compared with Resources for Services (RS) for program technical assistance to implement depression quality improvement programs improves 6- and 12-month client mental-health related quality of life (MHRQL); however, effects for clients with multiple chronic medical conditions (MCC) are unknown.
Objective
To explore effectiveness of CEP vs RS in MCC and non-MCC subgroups.
Design
Secondary analyses of a cluster-randomized trial.
Setting
93 health care and community-based programs in two neighborhoods.
Participants
Of 4,440 clients screened, 1,322 depressed (Patient Health Questionnaire, PHQ8) provided contact information, 1,246 enrolled and 1,018 (548 with ≥3 MCC) completed baseline, 6- or 12-month surveys.
Intervention
CEP or RS for implementing depression quality improvement programs.
Outcomes and Analyses
Primary: depression (PHQ9 <10), poor MHRQL (Short Form Health Survey, SF-12<40); Secondary: mental wellness, good physical health, behavioral health hospitalization, chronic homelessness risk, work/workloss days, services use at 6 and 12 months. End-point regressions were used to estimate intervention effects on outcomes for subgroups with ≥3 MCC, non-MCC, and intervention-by-MCC interactions (exploratory).
Results
Among MCC clients at 6 months, CEP vs RS lowered likelihoods of depression and poor MHRQL; increased likelihood of mental wellness; reduced work-loss days among employed and likelihoods of ≥4 behavioral-health hospitalization nights and chronic homelessness risk, while increasing faith-based and park community center depression services; and at 12 months, likelihood of good physical health and park community center depression services use (each P<.05). There were no significant interactions or primary outcome effects for non-MCC.
Conclusions
CEP was more effective than RS in improving 6-month primary outcomes among depressed MCC clients, without significant interactions.
Keywords: Depression, Quality Improvement, Chronic Conditions, Coalition Intervention
Introduction
Depression is associated with high morbidity, health care costs and worse outcomes for chronic conditions such as diabetes.1-6 Chronic medical conditions disproportionately affect individuals with depression and other serious mental illness, contributing to earlier morbidity, premature mortality,7,8 and higher costs of care. Quality improvement (QI) programs for depression in primary care based on the collaborative care model are effective among primary care patients, including those with chronic medical conditions.9-16 Depressed patients in under-resourced communities have lower rates of access to quality care and worse outcomes.17-21 Evidence suggests that implementing QI for depression improves mental health outcomes as much or more for racial/ethnic minority compared with White populations.16,22-25 In under-resourced communities, depressed persons may receive services in alternative community sectors addressing social risk factors, such as homelessness.26-28 Public and private health insurance plans and health care systems increasingly seek to integrate addressing social determinants of health with health care to improve outcomes for patients with comorbid medical and mental disorders (eg, high-risk beneficiaries). One such strategy may be implementing QI for depression, including for clients with multiple medical comorbidities, across health and community-based sectors.29 However, such integrated, community-wide approaches are under-studied.30
Community Partners in Care31 (CPIC), a group-randomized trial, compared two approaches to implement evidence-based QI for depression across health and social-community sectors in under-resourced communities: 1) coalition support (Community Engagement and Planning, CEP) and 2) individual program technical assistance (Resources for Services, RS). Depressed participants were largely African American or Latino (90.8%) with income below federal poverty (73.7%). The study was implemented using Community-Partnered Participatory Research (CPPR).32 CEP, compared with RS, at 6 months significantly reduced the likelihood of having poor mental health-related quality of life (MHRQL), a primary outcome, but not depression by PHQ9; and improved community-prioritized outcomes, including increased likelihood of mental wellness and reduced likelihood of chronic homelessness risk and any behavioral health hospitalization.31 Over a 12-month period, there was evidence of reduced likelihood of poor MHRQL and of behavioral health hospitalization.33 More than half the participants reported ≥ 3 chronic medical conditions,31 and evidence of effectiveness in this subgroup might inform efforts to develop integrated coalition approaches for complex patients. Intervention effects in this subgroup have not previously been reported. A Cochrane Review34 identified CPIC as a main study of the added value of coalitions over an alternative to affect the health of minorities. Accordingly, this study provides a secondary analysis of CPIC intervention effects for depressed participants with multiple chronic medical conditions (MCC) as the main focus.
Objective
Our objective was to explore whether CEP relative to RS conferred similar benefits to participants with MCC as the overall sample, for primary and community-prioritized outcomes and secondary services use. Prior literature9-16 suggests that an effective implementation strategy for depression QI may benefit depressed individuals with MCC; but a coalition approach might not be as effective for MCC clients if it diffuses a clinical focus for a group with complex medical needs. We were also interested in exploring effects of interventions for clients without MCC and differential intervention effects by MCC status to inform future research, but the study has fewer non-MCC than MCC clients and was not specifically designed to test this interaction.
Methods / Design
We provide secondary data analyses of CPIC intervention effects28,31,33,35,36 for participants with and without MCC and explore effects of interactions of intervention and MCC status. Design and interventions are described elsewhere.31,33,36 CPIC was a group-level randomized comparative effectiveness trial comparing two interventions to implement an expanded model of depression collaborative care or QI programs in under-resourced communities. The CPIC protocol is published as an appendix to the 12-month outcomes article by Chung et al.33 Funded in 2007 prior to mandated registration of health services trials, the study was registered post-enrollment at ClinicalTrials.gov (NCT01699789). The human subject protocol was approved by IRBs of RAND and institutions requiring separate review.
Interventions
Both interventions were designed to implement evidence-based QI toolkits across health and community-based services sectors. Toolkits were available online and included manuals and flashdrives with tools for clinician assessment and medication management, individual and group Cognitive Behavioral Therapy for Depression, depression case management and health worker outreach and team management (http://www.communitypartnersincare.org/community-engagement-and-planning/).27,37-40 Resources for Services (RS) offered a prespecified number of webinars by interdisciplinary experts and one site visit to each primary care site, plus outreach by a community-engagement specialist. Community Engagement and Planning (CEP) invited program administrators to 10-12 biweekly meetings using participatory planning32 to adapt toolkits and develop training capacity, followed by oversight of training and program implementation for a year. The number of trainings offered and percentage of staff participating were greater in CEP than RS.36
Sampling
Programs from health care (primary care, public health, mental health, and substance abuse) and social-community sectors (homeless and social services, faith-based, park community centers, hair salons, exercise or other program) were identified from agency lists and community nominations in South Los Angeles and Hollywood-Metro LA. Sixty agencies were invited to participate; 133 of 194 programs within those agencies were potentially eligible (ie, served at least 15 clients per week, at least one staff member and did not focus exclusively on psychotic disorders or home services). Within each community, programs or clusters were paired based on geographic location, service sector, population served, services offered, and funding sources.
One of each pair was randomized to CEP and the other to RS. Following site visits to finalize enrollment, 93 programs from 50 consenting agencies were enrolled (45 RS and 48 CEP). Staff blinded to intervention status approached clients in waiting rooms (consecutively), group meetings or lines (randomly), with 4,649 adults (aged ≥ 18 years) approached over 2–3 days per program; 4,440 (95.5%) agreed to screening. Eligibility was limited to clients providing contact information and scoring ≥ 10 on a modified 8-item Patient Health Questionnaire (PHQ-8).31,41 Of 4,440 adults screened, 1,322 were eligible, 1,246 enrolled and 1,018 completed baseline or 6- or 12-month follow-up telephone surveys conducted by staff blinded to intervention status. MCC status was identified by a baseline measure listing 18 common chronic medical conditions; having ≥3 was classified as MCC. Of 1,018 in the analytic sample, 548 had MCC (270 RS, 278 CEP) and 470 did not (non-MCC) (234 RS, 236 CEP). The CONSORT diagram is presented elsewhere.31,33
Study Variables
Baseline measures included age, sex, education, race/ethnicity, current marital and work statuses, health insurance coverage, family income, MCC status, meeting federal criteria for family poverty, 12-month major depressive or dysthymic disorder from the MINI,42 and alcohol abuse or use of illicit drugs in past 12 months.
Pre-specified primary outcomes were: poor mental health quality of life (MHRQL), using a standard cut-point on the mental health subscale of the Short Form Health Survey (SF-12, MCS-12 ≤ 40)43 over the previous four weeks; and depression by a standard cut-point for mild/moderate depression on the Patient Health Questionnaire (PHQ) (9-item for outcome; 8-outcome for baseline, each ≥ 10)41 in the past two weeks. Following community priorities for outcome, we selected items from existing measures to develop five indicators as secondary outcomes: 1) mental wellness in the prior four weeks based on responding to “all of the time, or most of the time, or a good bit of the time” on any of three items (feeling calm and peaceful, having energy, being happy)31; 2) good physical health44 based on a response to any of five items: a) health not limited at all in daily moderate activities; b) climbing stairs; c) at least fairly physically active in work and leisure time; d) in the prior 4 weeks, health not limited in the kind of work or other activities you do as a result of your physical health; e) not accomplishing less than what one would like; 3) chronic homelessness risk, defined as currently homelessness or living in a shelter or having ≥2 risk factors for future homelessness from life events measures in the past 6 months;45 4) any behavioral health hospitalization (and for sensitivity analysis, ≥four hospital nights) in the past 6 months;31 and 5) any work for pay and if employed, number of work-loss days in the prior 30 days.
Outpatient services in the prior 6 months included: number of visits to health care agencies for “depression” (primary care services for depression or emotional health (ie, assessment, referral, medication, counseling); emergency or urgent care visits for alcohol, drug, or mental health; mental health specialty outpatient visits; and substance abuse program visits with depression services as defined above; number of mental health specialty outpatient visits with advice about medication or with counseling for depression/mental health. We included number of visits to social-community sectors with depression services in the prior 6 months (ie, social services, faith-based programs, parks and recreation community centers, telephone hotlines, and other); and any antidepressant medication use.31,37
Statistical Methods
Our analytic sample included 1018 individuals completing any baseline, 6-, or 12-month survey. We compared baseline characteristics by intervention status overall and stratified by MCC status. We used item-level imputation for missing data and wave-level imputation for missing surveys to adjust findings to the observed analytic sample,46-49 which for each period included clients not known to have died or withdrawn participation (1,018 for 6 months; 1,013 for 12 months). We used weights to account for non-enrollment and attrition (ie, enrolled, but not completing any surveys).50,51 The technical documentation for weighting and multiple imputation is presented elsewhere.33,36 Variance estimation is based on Taylor Series Linearization Method SUDAAN version 11.0.1 (http://www.rti.org/sudaan/), accounting for clustering (clients within programs), weighting, and multiple imputations.
We conducted intention-to-treat analyses using logistic regression for dichotomous variables, or Poisson regression for count outcome variables, estimated MCC-by-intervention (CEP relative to RS) interaction model. With attrition noted as a limitation by a Cochrane Review,34 we present end status at each follow-up for primary analyses, which permits using multiple imputation and attrition weights. Consistent with recommendations for group-level randomization trials,52 all models adjust for baseline status of the dependent variable, age, race/ethnicity, education, 12-month depressive disorder, and community. For PHQ-9 as outcome, we used modified PHQ-8 as covariate, given no suicide item at screening. Comparisons focused on effects of interventions within subgroups (MCC, non-MCC), but including interaction tests.
Significance of comparisons was assessed using contrasts among regression coefficients. Results are presented from logistic regression models as odds ratios (OR) and from Poisson models as incidence rate ratios (IRR) with 95% CIs. We supplement adjusted models with unadjusted raw data to assess robustness.
We conducted sensitivity analyses for intervention effects adjusted for baseline only or without covariates, with no change in conclusions. We include unweighted longitudinal models as sensitivity analyses (Tables 2, 3, 6, 7). For primary outcomes, we applied a Bonferroni adjustment considering two primary outcomes for 2 time periods, ie, P<.0125.53
Results
Baseline
Baseline demographic characteristics among those with and without MCC are similar for those in CEP and RS. Most participants were African American or Latino and had family income below federal poverty level (Table 1). Relative to those without MCC, participants with MCC were older, in worse mental and physical health, less likely to be working for pay, and race-ethnicity distribution differed with a higher percentage of African American (each P<.05, Table 2). For the whole sample, the distribution of specific chronic conditions is given in Table 3, with no significant differences at P<.05 by intervention status.
Table 1. Baseline characteristics of depressed clients in outcomes analysis, by interventiona.
| Overall, N=1018 | with MCC, n=548 | without MCC, n=470 | ||||
| RS, n=504 | CEP, n=514 | RS, n=270 | CEP, n=278 | RS, n=234 | CEP, n=236 | |
| Age, mean (SD), y | 44.9 (12.4) | 46.6 (13.2) | 48.2 (11.0) | 50.4 (11.6) | 40.9 (12.8) | 41.9 (13.6) |
| Female, n (%) | 286 (54.8 ) | 309 (59.1 ) | 151 (54.0) | 166 (58.8) | 135 (55.9 ) | 143 (59.5 ) |
| Race/Ethnicity, n (%) | ||||||
| Latino | 194 (38.8 ) | 215 (43.1 ) | 88 (32.9) | 99 (37.3) | 106 (45.8 ) | 116 (50.1 ) |
| African American | 239 (46.9 ) | 249 (45.0 ) | 135 (49.6) | 149 (49.7) | 104 (43.7 ) | 100 (39.4 ) |
| Non-Hispanic White | 45 (9.7 ) | 41 (8.8 ) | 30 (11.9) | 23 (9.0) | 15 (7.0 ) | 18 (8.6 ) |
| Other | 26 (4.6 ) | 9 (3.1 ) | 17 (5.5) | 7 (4.0) | 9 (3.5 ) | 2 (1.9 ) |
| Married or living with partner, n (%) | 116 (22.6 ) | 115 (22.6 ) | 55 (20.0) | 57 (20.7) | 61 (25.8 ) | 59 (25.0 ) |
| < High school, n (%) | 221 (43.7 ) | 224 (43.5 ) | 115 (42.4) | 112 (40.1) | 106 (45.3 ) | 113 (47.5 ) |
| No health insurance, n (%) | 286 (57.3 ) | 259 (51.1 ) | 151 (56.4) | 127 (46.8) | 135 (58.4 ) | 132 (56.4 ) |
| Income < poverty level, n (%) | 373 (74.5 ) | 377 (73.0 ) | 198 (73.6) | 199 (70.9) | 175 (75.5 ) | 178 (75.6 ) |
| Any work for pay now, n (%) | 105 (20.6 ) | 100 (19.4 ) | 42 (15.2) | 50 (18.0) | 63 (27.0 ) | 49 (21.1 ) |
| Chronic homelessness risk, n (%)b | 283 (58.0 ) | 255 (50.4 ) | 168 (63.7) | 136 (49.8) | 116 (51.2 ) | 119 (51.2 ) |
| 12-month depressive disorder, n (%) | 311 (62.4 ) | 318 (61.4 ) | 189 (70.5) | 180 (65.1) | 122 (52.8 ) | 138 (56.9 ) |
| Alcohol abuse or use of illicit drugs, 12 months, n (%) | 180 (36.3 ) | 218 (42.2 ) | 91 (34.2) | 108 (38.7) | 89 (38.9 ) | 110 (46.4 ) |
| Poor mental health-related quality of life, n (%)c | 271 (53.8 ) | 275 (52.7 ) | 157 (57.5) | 155 (54.5) | 114 (49.4 ) | 120 (50.5 ) |
| Mental wellness, n (%)d | 200 (39.2 ) | 207 (40.1 ) | 89 (32.7) | 101 (36.0) | 111 (47.0 ) | 106 (45.1 ) |
| Good physical health, n (%)e | 380 (75.2 ) | 378 (73.4 ) | 179 (66.6) | 177 (63.8) | 201 (85.4 ) | 201 (85.2 ) |
| PHQ-8, mean (SD) f | 15.1 (4.2) | 14.9 (4.1) | 15.8 (4.1) | 15.5 (4.2) | 14.2 (4.0) | 14.1 (3.8) |
a. Data were multiply imputed and weighted for eligible sample for enrollment; Chi-square test was used for a comparison between the two groups accounting for the design effect of the cluster randomization; P > .10 except homelessness risk within MCC, for which P=.052.
b. Homeless or living in a shelter, or at least two risk factors of four (at least two nights homeless, food insecurity, eviction, financial crisis)
c. Mental Health Composition Score of SF-12 (MCS12) ≤ 40; one standard deviation below population mean.
d. At least good bit of time on any of three items: feeling peaceful or calm, being a happy person, having energy.
e. On any of five items: health not limited in daily activities, climbing stairs, kind of work or other activities from physical health, not accomplishing less than one would like, fairly physically active in work and leisure past 4 weeks.
f. 8-item Personal Health Questionnaire Depression Scale (0 to 24, higher more distress).
RS, resources for services or individual program technical assistance; CEP, community engagement and planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions).
Table 2. Baseline characteristics of depressed clients by MCC statusa.
| Overall, N=1018 | Participants with MCC, N=548 | Participants without MCC, N=470 | P | |
| Age, mean (SD), y | 45.8 (12.9) | 49.4 (11.3) | 41.4 (13.2) | <.001 |
| Female, n (%) | 595 (57.0 ) | 317 (56.4 ) | 278 (57.7 ) | .749 |
| Race/ethnicity, n (%) | .003 | |||
| Latino | 409 (41.0 ) | 186 (35.2 ) | 223 (48.0 ) | |
| African American | 488 (46.0 ) | 284 (49.7 ) | 204 (41.5 ) | |
| Non-Hispanic White | 86 (9.2 ) | 54 (10.4 ) | 32 (7.8 ) | |
| Other | 35 (3.8 ) | 23 (4.7 ) | 12 (2.7 ) | |
| Married/living with partner, n (%) | 231 (22.6 ) | 112 (20.3 ) | 120 (25.3 ) | .071 |
| < High school, n (%) | 446 (43.6 ) | 227 (41.2 ) | 219 (46.4 ) | .102 |
| Income < poverty level, n (%) | 750 (73.7 ) | 397 (72.2 ) | 353 (75.6 ) | .313 |
| Any work for pay now, n (%) | 205 (20.0 ) | 92 (16.7 ) | 113 (24.0 ) | .017 |
| Chronic homelessness risk, n (%)b | 538 (54.1 ) | 303 (56.5 ) | 235 (51.2 ) | .149 |
| No health insurance, n (%) | 545 (54.1 ) | 278 (51.4 ) | 267 (57.4 ) | .109 |
| 12-month depressive disorder, n (%) | 629 (61.9 ) | 369 (67.7 ) | 260 (54.9 ) | <.001 |
| Alcohol abuse or use of illicit drugs, 12 months, n (%) | 398 (39.3 ) | 199 (36.5 ) | 200 (42.7 ) | .157 |
| Poor mental health-related quality of life, n (%)c | 546 (53.2 ) | 312 (56.0 ) | 234 (49.9 ) | .069 |
| Mental wellness, n (%)d | 407 (39.7 ) | 190 (34.4 ) | 217 (46.0 ) | <.001 |
| Good physical health, n (%)e | 759 (74.3 ) | 357 (65.1 ) | 402 (85.3 ) | <.001 |
| PHQ-8, mean (SD)f | 15.0 (4.1) | 15.6 (4.2) | 14.2 (3.9) | <.001 |
a. Data multiply imputed; percentages and means weighted for enrollment and survey response; Chi-square test for comparing MCC status accounting for cluster-randomized design.
b. Homeless or living in a shelter, or at least two risk factors of four (at least two nights homeless, food insecurity, eviction, financial crisis).
c. Mental Health Composition Score of SF-12 (MCS12) ≤ 40; one standard deviation below population mean.
d. At least good bit of time on any of three items: feeling peaceful or calm, being a happy person, having energy.
e. On any of five items: health not limited in daily activities, climbing stairs, kind of work or other activities from physical health, not accomplishing less than one would like, fairly physically active in work and leisure past 4 weeks.
f. 8-item Personal Health Questionnaire Depression Scale (0 to 24, higher more distress).
RS, Resources for services or individual program technical assistance; CEP, Community engagement and planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions).
Table 3. Percentage of specific chronic conditions at baseline, N=1018a.
| Chronic condition | Overall (N=1018) | RS (N=504) | CEP (N=514) | P |
| Asthma | 223 (21.5%) | 111 (21.6%) | 112 (21.4%) | .963 |
| Diabetes | 191 (19.0%) | 108 (21.5%) | 84 (16.7%) | .217 |
| Hypertension | 384 (38.9%) | 189 (38.6%) | 195 (39.2%) | .892 |
| Arthritis | 375 (37.5%) | 170 (34.1%) | 205 (40.7%) | .076 |
| Physical disability | 143 (14.7%) | 72 (14.8%) | 70 (14.5%) | .874 |
| Trouble breathing caused by emphysema or chronic lung disease | 182 (17.8%) | 87 (17.1%) | 96 (18.4%) | .672 |
| Cancer diagnosed, excluding skin cancer | 35 (3.4%) | 17 (3.2%) | 18 (3.6%) | .791 |
| Neurological condition | 69 (6.5%) | 33 (6.5%) | 36 (6.6%) | .97 |
| Stroke | 83 (8.2%) | 40 (7.9%) | 43 (8.4%) | .799 |
| Angina, heart failure or coronary artery disease | 80 (8.3%) | 38 (7.7%) | 42 (8.8%) | .591 |
| Chronic back problems | 365 (36.5%) | 190 (38.3%) | 175 (34.7%) | .282 |
| Stomach ulcer | 161 (16.3%) | 85 (17.1%) | 76 (15.4%) | .528 |
| Chronic liver disease | 90 (9.3%) | 44 (9.1%) | 46 (9.4%) | .932 |
| Migraine or other chronic severe headaches | 367 (35.3%) | 186 (36.8%) | 181 (33.9%) | .463 |
| Chronic problems urinating or bladder infections | 148 (15.3%) | 70 (14.5%) | 78 (16.0%) | .597 |
| Chronic gynecologic problems or women’s health problems, N=595 | 170 (28.0%) | 78 (26.8%) | 92 (29.1%) | .544 |
| HIV | 83 (8.3%) | 35 (6.8%) | 48 (9.8%) | .647 |
| Other chronic pain condition | 199 (19.9%) | 96 (19.1%) | 104 (20.6%) | .645 |
| 3 or more chronic conditions | 548 (54.8%) | 270 (54.4%) | 278 (55.1%) | .889 |
| Number of chronic conditions, mean (SD) | 3.3 (2.6) | 3.3 (2.6) | 3.4 (2.7) | .829 |
a. Data were multiply imputed; percentages and means were weighted to account for enrollment and survey response; Chi-square test was used for comparing two groups accounting for the design effect of the cluster randomization.
RS, Resources for Services or individual program technical assistance; CEP, Community Engagement and Planning.
Primary Outcome
CEP relative to RS was associated with a lower likelihood of poor MHRQL and depressive symptoms (PHQ-9) at 6-months among MCC participants (Table 4); with no significant 12-month intervention effects. MHRQL remained significant at 6 months after Bonferroni adjustment. There were no significant intervention effects within non-MCC clients, with trends suggesting small effects (for MHRQL, P=.064) but no significant interactions.
Table 4. Intervention effects on primary and community-prioritized outcomes at 6 and 12 month follow-up.
| Unadjusted estimatesa | Adjusted analysisb c | ||||||||
| with MCC | without MCC | with MCC | without MCC | Interaction | |||||
| RS | CEP | RS | CEP | CEP vs RS test (95% CI) | P | CEP vs RS test (95% CI) | P | P | |
| 6-month follow-up | |||||||||
| Primary outcomes | n/N (%) | n/N (%) | n/N (%) | n/N (%) | OR | OR | |||
| Poor mental health quality of life | 119/194 (61.3) | 78/181 (43.1) | 67/159 (42.1) | 66/154 (42.9) | .54 (.36, .81) | .005 | 1.06 (.65, 1.71) | .816 | .064 |
| PHQ-9 ≥10 | 157/194 (80.9) | 119/182 (65.4) | 83/159 (52.2) | 85/151 (56.3) | .60 (.40, .90) | .015 | 1.01 (.49, 2.11) | .968 | .125 |
| Community-prioritized (secondary) | |||||||||
| Mental wellness | 51/195 (26.2) | 75/182 (41.2) | 69/159 (43.4) | 79/154 (51.3) | 1.94 (1.25, 3.01) | .004 | 1.51 (.80, 2.84) | .184 | .473 |
| Good physical health | 131/195 (67.2) | 133/182 (73.1) | 136/159 (85.5) | 125/154 (81.2) | 1.49 (.97, 2.29) | .068 | .81 (.47, 1.40) | .438 | .063 |
| Chronic homelessness risk | 97/195 (49.7) | 49/182 (26.9) | 42/158 (26.6) | 45/153 (29.4) | .47 (.30, .72) | .001 | .85 (.36, 2.02) | .675 | .175 |
| Any behavioral health hospitalizations | 16/195 (8.2) | 8/182 (4.4) | 13/159 (8.2) | 8/154 (5.2) | .51 (.24, 1.06) | .071 | .52 (.19, 1.40) | .184 | .967 |
| ≥4 hospital nights for behavioral health | 11/195 (5.6) | 3/182 (1.6) | 4/159 (2.5) | 4/154 (2.6) | .28 (.10, .77) | .015 | .46 (.09, 2.27) | .313 | .595 |
| Work for pay now | 33/195 (16.9) | 42/182 (23.1) | 54/159 (34.0) | 45/154 (29.2) | 1.38 (.76, 2.51) | .269 | .86 (.46, 1.60) | .606 | .229 |
| mean (SD) | mean (SD) | mean (SD) | mean (SD) | IRR | IRR | ||||
| Work days missed last 30 days, if working | 8.4 (10.3) | 2.9 (4.5) | 3.4 (5.7) | 4.2 (7.0) | .41 (.22, .74) | .005 | 1.03 (.44, 2.43) | .934 | .105 |
| 12-month follow-up | |||||||||
| Primary outcomes | n/N (%) | n/N (%) | n/N (%) | n/N (%) | OR | OR | |||
| Poor mental health quality of life | 106/189 (56.1) | 87/177 (49.2) | 66/143 (46.2) | 62/147 (42.2) | .82 (.55, 1.21) | .296 | .77 (.48, 1.23) | .258 | .836 |
| PHQ-9 ≥ 10 | 149/192 (77.6) | 124/177 (70.1) | 72/145 (49.7) | 77/150 (51.3) | .88 (.56, 1.38) | .580 | 1.01 (.57, 1.79) | .973 | .704 |
| Community-prioritized (secondary) | |||||||||
| Mental wellness | 71/194 (36.6) | 72/179 (40.2) | 84/145 (57.9) | 84/151 (55.6) | 1.16 (.71, 1.88) | .542 | .97 (.61, 1.54) | .897 | .509 |
| Good physical health | 121/193 (62.7) | 125/179 (69.8) | 126/145 (86.9) | 128/150 (85.3) | 1.55 (1.04, 2.29) | .031 | 1.25 (.68, 2.30) | .448 | .542 |
| Chronic homelessness risk | 77/192 (40.1) | 63/178 (35.4) | 30/144 (20.8%) | 46/149 (30.9%) | 1.05 (.62, 1.77) | .855 | 1.29 (.75, 2.23) | .344 | .598 |
| Any behavioral health hospitalizations | 10/194 (5.2) | 8/179 (4.5) | 4/145 (2.8) | 8/150 (5.3) | .74 (.34, 1.62) | .447 | 1.01 (.31, 3.24) | .992 | .684 |
| ≥4 hospital nights for behavioral health | 7/193 (3.6) | 7/179 (3.9) | 1/145 (.7%) | 2/150 (1.3) | .87 (.36, 2.09) | .756 | .88 (.13, 6.09) | .894 | .993 |
| Work for pay now | 28/193 (14.5%) | 38/178 (21.3%) | 49/144 (34.0%) | 39/149 (26.2%) | 1.62 (.78, 3.38) | .171 | .77 (.47, 1.26) | .299 | .068 |
| mean (SD) | mean (SD) | mean (SD) | mean (SD) | IRR | IRR | ||||
| Work days missed last 30 days, if working | 6.3 (7.8) | 6.5 (8.7) | 2.9 (4.6) | 4.7 (8.7) | 1.13 (.55, 2.31) | .719 | 1.59 (.74, 3.42) | .220 | .443 |
a. Raw data without weighting and imputation.
b. Intervention-by-MCC interaction models used multiply imputed data, weighted for eligible sample for enrollment; logistic regression models for binary variables (presented as odds ratio, OR) or Poisson regression models for count variables (presented as incidence rate ratios, IRR), interacted of intervention condition by MCC status adjusted for baseline status of the dependent variable, age, education, race/ethnicity, 12-month depressive disorder, and community and accounted for the design effect of the cluster randomization.
c. A sensitivity analysis for “Work days missed last 30 days, if working” based on generalized negative binomial regression models using STATA (svy: gnbreg) provided similar results.
RS, Resources for services or individual program technical assistance; CEP, Community engagement and planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions).
Community-prioritized Outcomes
Among MCC participants at 6 months, CEP relative to RS increased the likelihood of mental wellness and reduced the likelihood of chronic homelessness risk, having ≥4 hospital nights (any hospitalization, P=.07) and work-loss days if working (any work for pay, not significant). At 12 months, CEP relative to RS increased the likelihood of having good physical health. Among non-MCC participants, there were no significant intervention effects, with trends suggesting smaller effects. There were no significant interactions of intervention effect by MCC status.
Outpatient Services
At 6-month follow-up, MCC participants in CEP relative to RS reported greater mean visits to faith-based programs for depression/emotional health services (P=.014). Among those with any faith-based participation, the number of faith-based depression/emotional visits was significantly higher for CEP (mean 1.7, SD 6.0) than RS (mean .8, SD 2.5) (P=.024). Participants in CEP relative to RS had a higher likelihood of any park community center depression service (P=.004). There were no significant intervention effects on service use variables at 12 months among those with MCC. Among non-MCC, CEP relative to RS was associated at 6 months with significantly fewer specialty medication visits (P=.003) and more visits with depression services in faith-based and any community sector; and at 12 months with reduced likelihood of antidepressant use (each P<.05). At 6 months, a significant interaction suggests greater increase in CEP vs RS in having a park depression/emotional health service in MCC than non-MCC. (Table 5)
Table 5. Intervention effects on service utilization at 6 and 12 month follow-up.
| Unadjusted estimatesa | Adjusted analysisb,c | ||||||||
| with MCC | without MCC | with MCC | without MCC | Interaction | |||||
| RS | CEP | RS | CEP | CEP vs RS test (95% CI) | P | CEP vs RS test (95% CI) | P | P | |
| 6-month follow-up | mean (SD) | mean (SD) | mean (SD) | mean (SD) | IRR | IRR | |||
| Health care sector visits for depression | 11.2 (19.1) | 12.2 (25.0) | 13.4 (35.0) | 12.4 (27.9) | .99 (.45, 2.18) | .987 | .90 (.53, 1.54) | .699 | .783 |
| Mental health specialist visits received advice about medication | 4.2 (9.5) | 3.5 (5.1) | 6.0 (25.3) | 2.5 (7.8) | .59 (.33, 1.07) | 079 | .35 (.18, .68) | .003 | .130 |
| Mental health specialist visits received counseling | 7.2 (14.1) | 5.7 (10.4) | 8.6 (29.0) | 5.3 (11.7) | .77 (.38, 1.57) | .421 | .57 (.30, 1.07) | .078 | .429 |
| Community sector visits for depression | 4.2 (20.0) | 3.8 (15.7) | .8 (2.3) | 3.8 (18.8) | 1.20 (.57, 2.50) | .630 | 2.87 (1.14, 7.21) | .027 | .160 |
| Faith-based visits with depression service | .5 (2.1) | 1.0 (4.8) | .3 (1.1) | .5 (3.9) | 2.65 (1.22, 5.74) | .014 | 3.31 (1.32, 8.28) | .011 | .661 |
| n/N (%) | n/N (%) | n/N (%) | n/N (%) | OR | OR | ||||
| Any park services for depression | 4/195 (2.1) | 12/182 (6.6) | 9/159 (5.7) | 3/154 (1.9) | 4.15 (1.60, 10.77) | .004 | .60 (.14, 2.52) | .464 | .014 |
| Took antidepressant | 89/195 (45.6) | 81/181 (44.8) | 51/159 (32.1) | 38/154 (24.7) | .81 (.42, 1.58) | .517 | .60 (.23, 1.61) | .249 | .567 |
| 12-month follow-up | mean (SD) | mean (SD) | mean (SD) | mean (SD) | IRR | IRR | |||
| Health care sector visits for depression | 10.4 (23.1) | 10.3 (22.2) | 7.5 (20.0) | 6.9 (15.8) | .86 (.55, 1.36) | .513 | .99 (.53, 1.86) | .982 | .656 |
| Mental health specialist visits received advice about medication | 3.6 (11.4) | 4.2 (10.6) | 2.2 (6.1) | 1.8 (5.1) | 1.05 (.64, 1.74) | .838 | .98 (.46, 2.11) | .960 | .850 |
| Mental health specialist visits received counseling | 5.1 (12.9) | 4.5 (10.2) | 3.7 (8.0) | 2.6 (7.0) | .86 (.53, 1.41) | .541 | .88 (.48, 1.59) | .657 | .948 |
| Community sector visits for depression | 2.1 (8.6) | 3.4 (18.0) | 1.5 (7.3) | 3.5 (17.9) | 1.61 (.64, 4.07) | .310 | 2.33 (.63, 8.64) | .184 | .652 |
| Faith-based visits with depression service | 1.2 (8.1) | .9 (6.6) | 1.2 (7.1) | .9 (3.8) | .79 (.24, 2.58) | .681 | 1.24 (.39, 3.96) | .710 | .613 |
| n/N (%) | n/N (%) | n/N (%) | n/N (%) | OR | OR | ||||
| Any park services for depression | 5/194 (2.6) | 4/179 (2.2) | 1/145 (.7) | 5/149 (3.4) | .88 (.24, 3.24) | .844 | 3.12 (.57, 16.93) | .181 | .239 |
| Took antidepressant | 75/194 (38.7) | 68/179 (38.0) | 43/145 (29.7) | 26/149 (17.4) | .87 (.51, 1.48) | .592 | .53 (.32, .87) | .015 | .153 |
a. Raw data without weighting and imputation.
b. Intervention-by-MCC interaction models used multiply imputed data, weighted for eligible sample for enrollment; logistic regression models for binary variables (presented as odds ratio, OR) or Poisson regression models for count variables (presented as incidence rate ratios, IRR), interacted of intervention condition by MCC status adjusted for baseline status of the dependent variable, age, education, race/ethnicity, 12-month depressive disorder, and community and accounted for the design effect of the cluster randomization.
c. A sensitivity analysis for count variables based on generalized negative binomial regression models using STATA (svy: gnbreg) provided similar results.
RS, Resources for services or individual program technical assistance; CEP, community engagement and planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions).
Sensitivity Analyses
Unweighted longitudinal models indicated similar patterns of intervention effects on outcomes. The CEP intervention took place during the 12 months post-baseline. As a sensitivity analysis, we conducted a longitudinal analysis using all waves of data (baseline, 6 months, 12 months) with three-way interactions (time, intervention, MCC status) without response weights, adjusting for the same set of baseline covariates as in the primary analysis. Initial explorations of 3-level, random-effects models using the SAS GLIMMIX procedure failed to converge for some variables. As a result, we utilized a generalized estimating equation (GEE) framework with logistic regression models for binary outcomes and Poisson models for count data using the SAS GENMOD procedure, specifying exchangeable correlation at the program level. From the estimated model, we developed a contrast involving a linear combination of coefficients to test intervention effects at each end point (baseline, 6 months, 12 months) and tested differences between intervention groups in change from baseline to 6 months, and 12 months within MCC and nonMCC groups and interaction of intervention by MCC status at each time point. The results of sensitivity analyses are presented in Table 6 and Table 7.
Table 6. Longitudinal analyses for alternative modeling of intervention effects on primary and secondary outcomes a.
| Participants with MCC | Participants without MCC | ||||||||
| CEP vs RS at specific time | CEP vs RS in change from baseline | CEP vs RS at specific time | CEP vs RS in change from baseline | Interaction | |||||
| Test (95% CI) | P | Test (95% CI) | P | Test (95% CI) | P | Test (95% CI) | P | P | |
| Primary outcomes | |||||||||
| Poor mental health quality of life | OR | OR | OR | OR | |||||
| Baseline | .95 (.68, 1.32) | .764 | 1.05 (.74, 1.48) | .8 | .696 | ||||
| 6-month follow-up | .54 (.36, .83) | .007 | .57 (.34, .95) | .032 | 1.01 (.62, 1.64) | .969 | .96 (.54, 1.71) | .901 | .085 |
| 12-month follow-up | .78 (.55, 1.11) | .164 | .82 (.51, 1.32) | .41 | .80 (.51, 1.25) | .318 | .76 (.44, 1.31) | .324 | .935 |
| PHQ-8 ≥10 | |||||||||
| Baseline | 2.16 (.26, 17.95) | .472 | 1.15 (.42, 3.16) | .781 | .599 | ||||
| 6-month follow-up | .59 (.39, .90) | .014 | .27 (.03, 2.23) | .224 | 1.02 (.50, 2.05) | .957 | .88 (.25, 3.12) | .845 | .119 |
| 12-month follow-up | .82 (.55, 1.22) | .328 | .38 (.04, 3.23) | .374 | 1.03 (.57, 1.85) | .914 | .89 (.29, 2.72) | .843 | .506 |
| Community-prioritized | |||||||||
| Mental wellness | |||||||||
| Baseline | 1.11 (.79, 1.55) | .543 | .94 (.64, 1.39) | .773 | .499 | ||||
| 6-month follow-up | 1.84 (1.20, 2.83) | .007 | 1.66 (1.04, 2.66) | .035 | 1.48 (.81, 2.71) | .183 | 1.57 (.79, 3.13) | .187 | .491 |
| 12-month follow-up | 1.19 (.74, 1.92) | .453 | 1.07 (.60, 1.93) | .802 | .96 (.61, 1.49) | .846 | 1.01 (.58, 1.76) | .963 | .401 |
| Good physical health | |||||||||
| Baseline | .89 (.62, 1.28) | .539 | .97 (.58, 1.64) | .914 | .812 | ||||
| 6-month follow-up | 1.42 (.95, 2.12) | .086 | 1.59 (.95, 2.67) | .08 | .83 (.47, 1.47) | .514 | .85 (.44, 1.68) | .645 | .094 |
| 12-month follow-up | 1.46 (.97, 2.19) | .066 | 1.63 (1.03, 2.60) | .038 | 1.28 (.71, 2.33) | .399 | 1.32 (.63, 2.76) | .453 | .709 |
| Chronic homelessness risk | |||||||||
| Baseline | .58 (.37, .93) | .023 | .98 (.63, 1.51) | .913 | .073 | ||||
| 6-month follow-up | .43 (.28, .65) | <.001 | .73 (.46, 1.17) | .194 | .81 (.38, 1.74) | .553 | .83 (.38, 1.80) | .616 | .127 |
| 12-month follow-up | .92 (.56, 1.51) | .742 | 1.58 (.88, 2.83) | .12 | 1.23 (.71, 2.13) | .457 | 1.26 (.67, 2.38) | .475 | .446 |
| Any behavioral health hospitalizations | |||||||||
| Baseline | 1.41 (.83, 2.38) | .198 | 1.53 (.73, 3.18) | .256 | .863 | ||||
| 6-month follow-up | .54 (.27, 1.09) | .085 | .38 (.17, .88) | .024 | .55 (.22, 1.34) | .182 | .36 (.12, 1.11) | .075 | .983 |
| 12-month follow-up | .83 (.36, 1.90) | .659 | .59 (.26, 1.34) | .205 | 1.13 (.37, 3.40) | .828 | .74 (.19, 2.81) | .648 | .68 |
| ≥4 hospital nights for behavioral health | |||||||||
| Baseline | 1.40 (.82, 2.40) | .222 | 1.53 (.75, 3.10) | .24 | .833 | ||||
| 6-month follow-up | .30 (.11, .78) | .014 | .21 (.07, .64) | .006 | .51 (.12, 2.24) | .359 | .34 (.07, 1.68) | .174 | .54 |
a. Intervention-by-MCC interaction models used all waves of multiply imputed data (baseline, 6 months, 12 months). A generalized estimating equation logistic regression model was used for a binary variable (presented as odds ratios) and Poisson regression model was used for a count variable (presented as incidence density ratios), adjusted age, education, race/ethnicity, 12-month depressive disorder, and community.
RS, Resources for Services; CEP, Community Engagement and Planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions)
Table 7. Longitudinal analyses for alternative modeling of intervention effects on primary and secondary outcomes a.
| Participants with MCC | Participants without MCC | ||||||||
| CEP vs RS at specific time | CEP vs RS in change from baseline | CEP vs RS at specific time | CEP vs RS in change from baseline | Interaction | |||||
| Test (95% CI) | P | Test (95% CI) | P | Test (95% CI) | P | Test (95% CI) | P | P | |
| 12-month follow-up | .99 (.40, 2.47) | .981 | .71 (.30, 1.67) | .429 | 1.21 (.20, 7.46) | .834 | .79 (.12, 5.16) | .806 | .844 |
| Any work for pay now | |||||||||
| Baseline | 1.07 (.60, 1.92) | .813 | .71 (.38, 1.32) | .277 | .206 | ||||
| 6-month follow-up | 1.23 (.70, 2.17) | .452 | 1.15 (.65, 2.03) | .623 | .76 (.42, 1.37) | .341 | 1.07 (.56, 2.05) | .842 | .21 |
| 12-month follow-up | 1.33 (.71, 2.48) | .353 | 1.24 (.61, 2.52) | .523 | .71 (.44, 1.14) | .156 | 1.00 (.55, 1.82) | 1 | .054 |
| Work days, N, missed in last 30 days, if working | IRR | IRR | IRR | IRR | |||||
| Baseline | .16 (-.78, 1.11) | .72 | -.03 (-1.20, 1.13) | .952 | .741 | ||||
| 6-month follow-up | -4.48 (-7.67, -1.28) | .006 | -4.64 (-7.68, -1.60) | .003 | .31 (-3.00, 3.62) | .844 | .34 (-2.47, 3.16) | .804 | .049 |
| 12-month follow-up | .90 (-2.48, 4.28) | .593 | .74 (-3.05, 4.53) | .69 | 2.01 (-1.49, 5.50) | .25 | 2.04 (-1.80, 5.88) | .283 | .594 |
| Services utilization | |||||||||
| Health care visits for depression (N) | |||||||||
| Baseline | 1.03 (.75, 1.43) | .848 | .81 (.52, 1.27) | .355 | .236 | ||||
| 6-month follow-up | 1.00 (.48, 2.08) | .99 | .96 (.47, 2.00) | .915 | .81 (.45, 1.43) | .444 | .99 (.54, 1.83) | .987 | .547 |
| 12-month follow-up | .90 (.58, 1.41) | .651 | .87 (.56, 1.37) | .555 | .89 (.46, 1.72) | .718 | 1.10 (.51, 2.38) | .807 | .962 |
| MH outpatient visits with medication (N), advice(N) | |||||||||
| Baseline | 1.27 (.77, 2.12) | .348 | 1.10 (.58, 2.06) | .775 | .732 | ||||
| 6-month follow-up | .63 (.34, 1.15) | .128 | .49 (.24, 1.01) | .053 | .32 (.17, .63) | .001 | .30 (.14, .62) | .002 | .047 |
| 12-month follow-up | 1.16 (.68, 1.99) | .577 | .91 (.52, 1.60) | .751 | .95 (.45, 2.00) | .895 | .87 (.32, 2.33) | .773 | .597 |
| MH outpatient visits with counseling (N) | |||||||||
| Baseline | 1.14 (.77, 1.70) | .501 | 1.19 (.66, 2.13) | .567 | .911 | ||||
| 6-month follow-up | .79 (.40, 1.58) | .472 | .69 (.35, 1.36) | .256 | .54 (.28, 1.05) | .067 | .46 (.24, .89) | .021 | .336 |
| 12-month follow-up | .94 (.56, 1.58) | .811 | .82 (.47, 1.43) | .477 | .86 (.48, 1.56) | .616 | .73 (.36, 1.48) | .371 | .784 |
| Community sector visits for depression (N) | |||||||||
| Baseline | 1.26 (.72, 2.21) | .419 | .74 (.35, 1.60) | .449 | .244 | ||||
| 6-month follow-up | 1.20 (.55, 2.61) | .644 | .95 (.41, 2.22) | .911 | 2.65 (1.00, 7.01) | .05 | 3.56 (1.05, 12.07) | .042 | .225 |
| 12-month follow-up | 1.55 (.59, 4.05) | .369 | 1.23 (.37, 4.09) | .735 | 2.28 (.62, 8.37) | .2 | 3.06 (.67, 14.02) | .144 | .657 |
| Religious services for depression | |||||||||
| Baseline | 1.18 (.64, 2.20) | .591 | .41 (.18, .90) | .026 | .033 | ||||
| 6-month follow-up | 2.53 (1.17, 5.49) | .019 | 2.14 (1.00, 4.55) | .049 | 2.71 (1.12, 6.56) | .027 | 6.69 (2.05, 21.85) | .002 | .896 |
| 12-month follow-up | .68 (.18, 2.56) | .558 | .58 (.15, 2.25) | .414 | 1.01 (.33, 3.04) | .991 | 2.48 (.63, 9.79) | .193 | .688 |
| Any park community center services for depression | OR | OR | OR | OR | |||||
| Baseline | .81 (.45, 1.46) | .48 | .96 (.43, 2.15) | .928 | .722 | ||||
| 6-month follow-up | 3.57 (1.46, 8.73) | .006 | 4.41 (1.47, 13.23) | .009 | .54 (.13, 2.20) | .373 | .56 (.12, 2.55) | .445 | .011 |
| 12-month follow-up | .63 (.20, 1.98) | .428 | .78 (.26, 2.34) | .655 | 2.26 (.51, 10.06) | .279 | 2.35 (.49, 11.22) | .28 | .164 |
| Use antidepressant | |||||||||
| Baseline | 1.15 (.75, 1.77) | .513 | 1.41 (.83, 2.42) | .205 | .457 | ||||
| 6-month follow-up | .99 (.54, 1.80) | .96 | .85 (.49, 1.48) | .556 | .80 (.37, 1.70) | .522 | .56 (.27, 1.19) | .12 | .627 |
| 12-month follow-up | 1.01 (.60, 1.71) | .975 | .87 (.54, 1.41) | .574 | .69 (.41, 1.17) | .167 | .49 (.28, .85) | .012 | .247 |
a. Intervention-by-MCC interaction models used all waves of multiply imputed data (baseline, 6 months, 12 months). A generalized estimating equation logistic regression model was used for a binary variable (presented as odds ratios) and Poisson regression model was used for a count variable (presented as incidence density ratios), adjusted age, education, race/ethnicity, 12-month depressive disorder, and community.
RS, Resources for Services; CEP, Community Engagement and Planning; MCC, medical chronic conditions, defined as presence of 3 or more chronic conditions (among 18 conditions)
Discussion
This secondary analysis was designed to explore whether CPIC intervention findings for the full sample31,36 were confirmed among participants having depression and MCC, an important clinical and policy group, viewed by stakeholders as vulnerable and prioritized for subanalyses. This comparison has potential implications for innovative health insurance and health care programs seeking to integrate clinical and social services to address social determinants of health for complex clients. Consistent with prior studies suggesting that depression collaborative care is effective in medically ill patients,9-16 we found that the CEP relative to RS was associated with greater improvements at 6 months among MCC participants in both primary outcomes of MHRQL and depression, which for MHRQL would survive formal correction for multiple statistical comparisons. We previously reported that intervention effects on depression were not significant in the overall sample.31,36
Further, we found significant effects of CEP over RS for at least one time point among MCC participants for community-prioritized outcomes, ie, mental wellness, homelessness risk factors, ≥ 4 behavioral health hospitalizations, and work-loss days among employed. The sample without MCC was smaller, and we did not observe significant intervention effects on primary or community-prioritized outcomes. For non-MCC clients, we found a pattern under CEP relative to RS of fewer specialty medication visits at 6 months, less use of antidepressants at 12 months, but more faith-based visits and any community visit for depression at 6 months. Most intervention-by-MCC interactions were not significant. Findings were fairly consistent in unweighted longitudinal sensitivity analyses.
The overall pattern suggests that findings for clients with MCC at 6 months were generally consistent with that of the overall sample, favoring CEP relative to RS for MHRQL and some community-prioritized outcomes.31,33 Reduced effects at 12 months may partly be due to lower exposure over time to depression services in the assigned intervention.54 Future research is needed to understand mechanisms underlying initial benefits of CEP relative to RS and reduced effects over time, to inform research on determining how to sustain effects.
Limitations
The study has important limitations. The MCC indicator was based on 18 specific conditions, all measures were self-reported and community-prioritized outcomes while using items from standard measures were newly constructed indicators to match stakeholder concepts. The study was limited to two Los Angeles communities. Response rates were moderate for agencies, high for programs and clients enrollment, with moderate client nonresponse at follow-up. Randomization was within community so there could be cross-intervention exposure, a conservative bias limiting observed effects. This subanalysis was exploratory. The non-MCC group had a smaller sample size and the study was not designed to have power for interactions. There was no usual-care control.
Conclusion
Given policy and health care system interest in integrated approaches such as community health homes and accountable communities,55,56 the findings, despite study limitations, suggest that a coalition model relative to program technical assistance to implement collaborative care for depression, holds promise for improving at least short-term outcomes among ethnically diverse groups having both depression and multiple chronic medical conditions.
Acknowledgments
Funding Sources: This study was supported by the Patient Centered Outcomes Research Institute Contract PPRN-1501-26518 and R01MD007721 from the National Institute on Minority Health and Health Disparities and the parent CPIC study by R01MH078853, P30MH082760 and P30MH068639 from the National Institute of Mental Health. The parent CPIC study was also funded by 64244 from the Robert Wood Johnson Foundation, CMCH-12- 97088 from the California Community Foundation, G08LM011058 from the National Library of Medicine, and UL1TR000124 from the NIH/National Center for Advancing Translational Science UCLA CTSI. The RAND Corporation, UCLA Semel Institute and the Los Angeles County Departments of Mental Health, Public Health and Health services provided institutional support.
We thank 95 participating health care and community-based agencies, the CPIC Council and members of the Association of Clinical and Translational Science Team Science Award for CPIC (2014) and Campus-Community Partnerships for Health 2015 Annual Award. We acknowledge posthumously the contributions to study design and implementation of two CPIC Council leaders (Rev. Ron Wright and Rev. Terrance Stone) who passed away during preparation of this manuscript. We thank Erika Orellana for support for manuscript preparation.
References
- 1. Kessler RC, Chiu WT, Demler O, Merikangas KR, Walters EE. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2005;62(6):617-627. 10.1001/archpsyc.62.6.617 10.1001/archpsyc.62.6.617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Unützer J, Park M. Strategies to improve the management of depression in primary care. Prim Care. 2012;39(2):415-431. 10.1016/j.pop.2012.03.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Moussavi S, Chatterji S, Verdes E, Tandon A, Patel V, Ustun B. Depression, chronic diseases, and decrements in health: results from the World Health Surveys. Lancet. 2007;370(9590):851-858. 10.1016/S0140-6736(07)61415-9 [DOI] [PubMed] [Google Scholar]
- 4. Druss BG, Rosenheck RA. Patterns of health care costs associated with depression and substance abuse in a national sample. Psychiatr Serv. 1999;50(2):214-218. 10.1176/ps.50.2.214 [DOI] [PubMed] [Google Scholar]
- 5. Druss BG, Rosenheck RA, Sledge WH. Health and disability costs of depressive illness in a major U.S. corporation. Am J Psychiatry. 2000;157(8):1274-1278. 10.1176/appi.ajp.157.8.1274 [DOI] [PubMed] [Google Scholar]
- 6. Olfson M, Mojtabai R, Merikangas KR, et al. Reexamining associations between mania, depression, anxiety and substance use disorders: results from a prospective national cohort. Mol Psychiatry. 2017;22(2):235-241. 10.1038/mp.2016.64 [DOI] [PubMed] [Google Scholar]
- 7. Whiteman KL, Naslund JA, DiNapoli EA, Bruce ML, Bartels SJ. Systematic review of integrated general medical and psychiatric self-management interventions for adults with serious mental illness. Psychiatr Serv. 2016;67(11):1213-1225. 10.1176/appi.ps.201500521 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Katon WJ. Clinical and health services relationships between major depression, depressive symptoms, and general medical illness. Biol Psychiatry. 2003;54(3):216-226. 10.1016/S0006-3223(03)00273-7PMID:12893098 [DOI] [PubMed] [Google Scholar]
- 9. Koike AK, Unützer J, Wells KB. Improving the care for depression in patients with comorbid medical illness. Am J Psychiatry. 2002;159(10):1738-1745. 10.1176/appi.ajp.159.10.1738 [DOI] [PubMed] [Google Scholar]
- 10. Ell K, Katon W, Xie B, et al. Collaborative care management of major depression among low-income, predominantly Hispanic subjects with diabetes: a randomized controlled trial. Diabetes Care. 2010;33(4):706-713. 10.2337/dc09-1711 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Dwight-Johnson M, Lagomasino IT, Hay J, et al. Effectiveness of collaborative care in addressing depression treatment preferences among low-income Latinos. Psychiatr Serv. 2010;61(11):1112-1118. 10.1176/ps.2010.61.11.1112 [DOI] [PubMed] [Google Scholar]
- 12. Katon W, Fan MY, Unützer J, Taylor J, Pincus H, Schoenbaum M. Depression and diabetes: a potentially lethal combination. J Gen Intern Med. 2008;23(10):1571-1575. 10.1007/s11606-008-0731-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Katon W, Russo J, Lin EH, et al. Cost-effectiveness of a multicondition collaborative care intervention: a randomized controlled trial. Arch Gen Psychiatry. 2012;69(5):506-514. 10.1001/archgenpsychiatry.2011.1548 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Unützer J, Katon W, Callahan CM, et al. ; IMPACT Investigators. Improving Mood-Promoting Access to Collaborative Treatment . Collaborative care management of late-life depression in the primary care setting: a randomized controlled trial. JAMA. 2002;288(22):2836-2845. 10.1001/jama.288.22.2836 [DOI] [PubMed] [Google Scholar]
- 15. Lin EH, Katon W, Von Korff M, et al. ; IMPACT Investigators . Effect of improving depression care on pain and functional outcomes among older adults with arthritis: a randomized controlled trial. JAMA. 2003;290(18):2428-2429. 10.1001/jama.290.18.2428 [DOI] [PubMed] [Google Scholar]
- 16. Katon WJ, Lin EH, Von Korff M, et al. Collaborative care for patients with depression and chronic illnesses. N Engl J Med. 2010;363(27):2611-2620. 10.1056/NEJMoa1003955 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Ell K, Vourlekis B, Nissly J, et al. Integrating mental health screening and abnormal cancer screening follow-up: an intervention to reach low-income women. Community Ment Health J. 2002;38(4):311-325. 10.1023/A:1015901409211 [DOI] [PubMed] [Google Scholar]
- 18. McGuire TG, Miranda J. New evidence regarding racial and ethnic disparities in mental health: policy implications. Health Aff (Millwood). 2008;27(2):393-403. 10.1377/hlthaff.27.2.393 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Miranda J, Cooper LA. Disparities in care for depression among primary care patients. J Gen Intern Med. 2004;19(2):120-126. 10.1111/j.1525-1497.2004.30272.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Williams DR, González HM, Neighbors H, et al. Prevalence and distribution of major depressive disorder in African Americans, Caribbean blacks, and non-Hispanic whites: results from the National Survey of American Life. Arch Gen Psychiatry. 2007;64(3):305-315. 10.1001/archpsyc.64.3.305 [DOI] [PubMed] [Google Scholar]
- 21. Miranda J, McGuire TG, Williams DR, Wang P. Mental health in the context of health disparities. Am J Psychiatry. 2008;165(9):1102-1108. 10.1176/appi.ajp.2008.08030333 [DOI] [PubMed] [Google Scholar]
- 22. Wells K, Miranda J, Bruce ML, Alegria M, Wallerstein N. Bridging community intervention and mental health services research. Am J Psychiatry. 2004;161(6):955-963. 10.1176/appi.ajp.161.6.955 [DOI] [PubMed] [Google Scholar]
- 23. Areán PA, Ayalon L, Hunkeler E, et al. ; IMPACT Investigators. Improving depression care for older, minority patients in primary care. Med Care. 2005;43(4):381-390. https:// doi.org/ 10.1097/01.mlr.0000156852.09920. b1 PMID:15778641 [DOI] [PubMed]
- 24. Gilbody S, Bower P, Fletcher J, Richards D, Sutton AJ. Collaborative care for depression: a cumulative meta-analysis and review of longer-term outcomes. Arch Intern Med. 2006;166(21):2314-2321. 10.1001/archinte.166.21.2314 [DOI] [PubMed] [Google Scholar]
- 25. Davis TD, Deen T, Bryant-Bedell K, Tate V, Fortney J. Does minority racial-ethnic status moderate outcomes of collaborative care for depression? Psychiatr Serv. 2011;62(11):1282-1288. 10.1176/ps.62.11.pss6211_1282 [DOI] [PubMed] [Google Scholar]
- 26. Gamble VN. Under the shadow of Tuskegee: african Americans and health care. Am J Public Health. 1997;87(11):1773-1778. 10.2105/AJPH.87.11.1773 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Miranda J, Chung JY, Green BL, et al. Treating depression in predominantly low-income young minority women: a randomized controlled trial. JAMA. 2003;290(1):57-65. 10.1001/jama.290.1.57 [DOI] [PubMed] [Google Scholar]
- 28. Miranda J, Ong MK, Jones L, et al. Community-partnered evaluation of depression services for clients of community-based agencies in under-resourced communities in Los Angeles. J Gen Intern Med. 2013;28(10):1279-1287. 10.1007/s11606-013-2480-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Dobransky-Fasiska D, Brown C, Pincus HA, et al. ; RNDC-Community Partners . Developing a community-academic partnership to improve recognition and treatment of depression in underserved African American and white elders. Am J Geriatr Psychiatry. 2009;17(11):953-964. 10.1097/JGP.0b013e31818f3a7e [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Alexander L, Druss B. Behavioral Health Homes for People with Mental Health and Substance Use Conditions: The Core Clinical Features. Washington, DC: SAMHSA-HRSA Center for Integrated Health Solutions; 2012. [Google Scholar]
- 31. Wells KB, Jones L, Chung B, et al. Community-partnered cluster-randomized comparative effectiveness trial of community engagement and planning or resources for services to address depression disparities. J Gen Intern Med. 2013;28(10):1268-1278. 10.1007/s11606-013-2484-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Jones L, Wells K. Strategies for academic and clinician engagement in community-participatory partnered research. JAMA. 2007;297(4):407-410. 10.1001/jama.297.4.407 [DOI] [PubMed] [Google Scholar]
- 33. Chung B, Ong M, Ettner SL, et al. 12-month outcomes of community engagement versus technical assistance to implement depression collaborative care: a partnered, cluster, randomized, comparative effectiveness trial. Ann Intern Med. 2014;161(10)(suppl):S23-S34. 10.7326/M13-3011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Anderson LM, Adeney KL, Shinn C, Safranek S, Buckner-Brown J, Krause LK. Community coalition-driven interventions to reduce health disparities among racial and ethnic minority populations. Cochrane Database Syst Rev. 2015;(6):CD009905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Chung B, Jones L, Dixon EL, Miranda J, Wells K; Community Partners in Care Steering Council . Using a community partnered participatory research approach to implement a randomized controlled trial: planning community partners in care. J Health Care Poor Underserved. 2010;21(3):780-795. 10.1353/hpu.0.0345 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Chung B, Ngo VK, Ong MK, et al. Participation in training for depression care quality improvement: a randomized trial of community engagement or technical support. Psychiatr Serv. 2015;66(8):831-839. 10.1176/appi.ps.201400099 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Wells KB, Sherbourne C, Schoenbaum M, et al. Impact of disseminating quality improvement programs for depression in managed primary care: a randomized controlled trial. JAMA. 2000;283(2):212-220. 10.1001/jama.283.2.212 [DOI] [PubMed] [Google Scholar]
- 38. Miranda J, Duan N, Sherbourne C, et al. Improving care for minorities: can quality improvement interventions improve care and outcomes for depressed minorities? Results of a randomized, controlled trial. Health Serv Res. 2003;38(2):613-630. 10.1111/1475-6773.00136 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Unützer J, Katon W, Williams JW Jr, et al. Improving primary care for depression in late life: the design of a multicenter randomized trial. Med Care. 2001;39(8):785-799. 10.1097/00005650-200108000-00005 10.1097/00005650-200108000-00005 [DOI] [PubMed] [Google Scholar]
- 40. Springgate BF, Wennerstrom A, Meyers D, et al. Building community resilience through mental health infrastructure and training in post-Katrina New Orleans. Ethn Dis. 2011;21(3)(suppl 1):S1-S20, 9. PMID:22352077 [PMC free article] [PubMed]
- 41. Kroenke K, Strine TW, Spitzer RL, Williams JB, Berry JT, Mokdad AH. The PHQ-8 as a measure of current depression in the general population. J Affect Disord. 2009;114(1- 3):163-173. https://doi.org/ 10.1016/j jad.2008.06.026 PMID:18752852 [DOI] [PubMed]
- 42. Sheehan DV, Lecrubier Y, Sheehan KH, et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry. 1998;59(suppl 20):22-33. [PubMed] [Google Scholar]
- 43. Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30(6):473-483. 10.1097/00005650-199206000-00002 [DOI] [PubMed] [Google Scholar]
- 44. Lam CA, Sherbourne C, Tang L, et al. The impact of community engagement on health, social, and utilization outcomes in depressed, impoverished populations: secondary findings from a randomized trial. J Am Board Fam Med. 2016;29(3):325-338. 10.3122/jabfm.2016.03.150306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Sherbourne CD, Edelen MO, Zhou A, Bird C, Duan N, Wells KB. How a therapy-based quality improvement intervention for depression affected life events and psychological well-being over time: a 9-year longitudinal analysis. Med Care. 2008;46(1):78-84. 10.1097/MLR.0b013e318148478d [DOI] [PubMed] [Google Scholar]
- 46. Little RJ. Missing-data adjustments in large surveys. J Bus Econ Stat. 1988;6(3):287-296. [Google Scholar]
- 47. Rubin D, ed. Multiple Imputation for Nonresponse in Surveys. Hoboken, NJ: John Wiley & Sons; 1987. 10.1002/9780470316696 [DOI] [Google Scholar]
- 48. Schafer JL. Analysis of Incomplete Multivariate Data. Online: Chapman and Hall/CRC Press; 1997. https://doi. org/ 10.1201/9781439821862 [DOI]
- 49. Lavori PW, Dawson R, Shera D. A multiple imputation strategy for clinical trials with truncation of patient data. Stat Med. 1995;14(17):1913-1925. 10.1002/sim.4780141707 [DOI] [PubMed] [Google Scholar]
- 50. Groves RM, Dillman DA, Eltinge JL, Little RJA, eds. Survey Nonresponse. Vol 326 Hoboken, NJ: Wiley-Interscience; 2002. [Google Scholar]
- 51. Korn E, Graubard B. Analysis of Health Surveys. Hoboken, NJ: Wiley- Interscience; 1999. 10.1002/9781118032619 [DOI] [Google Scholar]
- 52. Murray DM. Design and Analysis of Group-randomized Trials. Oxford, UK: Oxford University Press; 1998:29. [Google Scholar]
- 53. Dunn OJ. Multiple comparisons among means. J Am Stat Assoc. 1961;56(293):52-64. https://doi.org/ 10.1080/01621459.1961.104 82090 [DOI]
- 54. Ong MK, Jones L, Aoki W, et al. A community-partnered, participatory, cluster-randomized study of depression care quality improvement: three-year outcomes. Psychiatr Serv. 2017;68(12):1262-1270. 10.1176/appi.ps.201600488 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Dougherty D, Conway PH. The “3T’s” road map to transform US health care: the “how” of high-quality care. JAMA. 2008;299(19):2319-2321. 10.1001/jama.299.19.2319 [DOI] [PubMed] [Google Scholar]
- 56. Alley DE, Asomugha CN, Conway PH, Sanghavi DM. Accountable health communities- addressing social needs through Medicare and Medicaid. N Engl J Med. 2016;374(1):8-11. 10.1056/NEJMp1512532 [DOI] [PubMed] [Google Scholar]