Study characteristics
Overall, 9263 unique records were identified, of which 9100 records were excluded based on title and abstract screening. 163 records were assessed in full-text for eligibility, among which 92 were excluded because of phenomena of interest, 38 were excluded because they focused on a single disease, 5 were excluded because they were not in primary care settings, 2 were excluded for language, 6 were excluded because only protocols were available, and 3 were excluded because of abstracts only (Figure 1). Therefore, seventeen studies (13 RCTs, 3 cohort studies, and 1 non-randomized trial) were included. Among these studies, seven were conducted in the United States, 4 in the United Kingdom, 1 in Germany, 1 in Taiwan China, 2 in South Africa, 1 in Australia, and 1 in Spain. The sample size of the included studies ranged from 142 to 5 337 377. Eleven studies studied depression comorbid with other diseases, such as hypertension, diabetes, or coronary heart disease, 5 studies studied other chronic diseases, 1 studied multimorbidity in general. Sixteen models of primary care team were identified in the included studies, among which two studies assessed the same intervention at different times. All studies were based on complex and multifaceted interventions or policies, and no study linked quality change in a single intervention component (Table 1).
(Figure 1 inserted here)
(Table 1 inserted here)
Primary care team
The PCT in the included studies is summarized below. A detailed description of the interventions is presented in table S2 (supplementary file 3). Basically, PCT involved in the included studies can be broadly summarized into three categories based on the skill mix of the team: 1) upward collaborative PCT involving primary care workers and specialists from upper-level hospitals34, 2) downward PCT involving primary care workers and lay health workers such as community health workers, and 3) traditional PCT involving primary care physicians and care managers (Figure 2). Some PCT, such as TEAMcare, also emphasized the roles of patients in the team and their collaborative work with other team members35.
(Figure 2 inserted here)
The upward collaborative PCT included 5 models in 7 articles35-41. In this model, health workers from primary care clinics worked together with hospital specialists, such as psychiatrists, psychologists, internalists, etc. The specialists trained primary health workers with essential knowledge and skills, supervised, and gave recommendations for the care process. In the TEAMcare study, a weekly meeting was arranged among nurses, primary care physicians, a psychiatrist, and a psychologist to review new cases and patient progress35, 37. In the multi-site study on the US-Mexico border, the integration and collaboration among multiple healthcare providers were emphasized by introducing warm-handoffs, transportation support, and a transitional nurse41. However, this study was piloted in various places, and the interventions were tailored to the local organizations' context, setting, and population. Multiple professionals emerged in different places, including physicians and nurses from different primary care clinics, community health workers, or specialists from local hospitals.
The downward collaborative PCT included 2 models in 2 articles42, 43. In this model, primary care workers worked collaboratively with lay workers from the community, such as community health workers or lay counselors. A collaborative care model for patients comorbid with mental disorders and chronic conditions in South Africa was composed of primary care nurses, GPs, and lay counsellors42. The primary care nurses were trained to provide person-centered care and supplementary mental health training. The lay counselors were trained and delivered group-based counseling services drawing on cognitive behavioral therapy techniques and also referred patients for further counseling. In the SUCCEED trial targeting patients with transient ischemic attack with multimorbidity, the community health worker (CHW) was the core member of the team who served as a liaison between the patient and the health system43. They reinforced and enhanced health literacy, risk factor control, self-management, and lifestyle changes for patients. The CHW and the physicians meet frequently to discuss patients’ progress and needs.
The traditional PCT included 8 models in 8 articles44-51. In this model, primary care physicians and nurses/medical assistants (care managers) were the main members. Basically, the care managers had close contact with patients to help them manage their conditions within daily life and build a bridge between patients and the healthcare system. In the DROP program, the care managers regularly monitored patients by phone, provided therapeutic advice, and reminded them about upcoming appointments. They also delivered a cognitive-behavioral psychoeducational program for patients, promoting awareness and self-management44. The care managers and primary care physicians were loosely connected by information system. A computerized decision support system, integrated in the clinical electronic medical record system in primary care clinics, can generate general recommendations to support physicians’ decision-making. Care managers annotated these suggestions in the clinical record of each patient as delivery. The Teamlet model was also a typical one. It was a small team composed of three persons—a primary care physician and two health coaches (care managers) who were trained from medical assistants or health workers45. Under this small team, a clinical encounter was extended to four parts: a pre-visit by the coach, a visit by the physician together with the coach, a post-visit by the coach, and between-visit care by the coach. The coach provided guidance on managing their diseases and emotional support to patients. They also monitored patients’ progress, solved their problems, and navigated the healthcare system.
(Table 2 inserted here)
The impact of PCT on quality of care
The impact of PCT on quality of care can be summarized as follows: first, PCT improved patients’ mental and psychological health outcomes significantly. The majority of depression-related measurements like SCL-90, SCL-20, PHQ-9 or GAD-7, showed significant improvements; Second, PCT improved patients’ perceptions towards care. Patients’ satisfaction with care, perceptions of improvement, and patient-centeredness all showed significant improvements; Third, PCT has changed the process of care. Although providing examinations showed mixed results, patients had more consultations with PCT, and the continuity of care also increased. More medication adjustment made by physicians was also observed. Lastly, PCT showed mixed effects on objective and validated outcome measures. The results of changes in BP, HbA1c, or LDL were mixed.
Objective and validated outcome measures
All the 17 studies reported objective and validated outcome measures (Table 2). The effect of PCT on improving blood pressure (BP), hemoglobin A1c (HbA1c), low-density lipoprotein cholesterol levels (LDL) were mixed. Two studies reported statistically significant improvements in BP 37, 51, but the other four studies reported no significant improvements38, 41, 43, 45. Katon et al. found that patients systolic BP decreased by 5.1 mmHg (P<0.001) compared with the control group37, and Wood et al. found that 65% of patients attained the goal of reaching less than 140/90 mmHg compared with 55% in the control group (95%CI=0.6-20.2, P=0.04) 51. Two studies reported significant improvement in HbA1c37, 41, and three studies did not find significant improvements43, 45, 52. One study found significant improvement in BMI 45, two studies found no significant improvements 38, 43, and one study found no significant improvement in waist circumference38. Two studies reported significant improvements in LDL 37, 45, and two studies reported no significant improvements 43, 51.
Ten studies presented data on mental health outcomes (Table 2). All of the ten studies showed significant improvements in depression-related measurements. Four studies reported statistically significant improvements in depression severity39, 44, 52, 53. Aragonès et al. reported the response rate to antidepressant treatment was 18.9% higher than the control group (OR=2.74, 95%CI=1.12-6.67); Katon et al. found that the interventions group was more likely to have a 50% decrease in SCL-90 depression score (OR=1.62 (95%CI=0.98-2.67) at 6 months, and OR=1.47 (95%CI=0.90-2.39) at 12 months; Sharpe et.al. detected a 50% reduction on SCL-20 (OR=8.5, 95%CI=5.5-13.4, P<0.001), and Walker et al. found a 0.62 decrease in SCL-20 score (95%CI=-0.94-0.29). Seven studies presented data on mean depressive scores or PHQ-9 scores, or SCL-20 depression scores. Only one study showed no significant difference in the reduction in PHQ-9 score 38. Six studies showed significant improvements in the reduction of mean depressive scores or PHQ-9 scores or SCL-20 depression scores 37, 41, 42, 44, 46, 49.Three studies showed significant improvements in anxiety39, 46, 53. One study showed a significant reduction of the generalized anxiety disorder scale (GAD-7)46.
Three studies reported results of hospitalizations, mortality, and illness burden (Table 2). One study reported statistically significant lower hospitalization for COPD/asthma (OR=0.91, 95%CI=0.87-0.94) and diabetes or its complications (OR=0.87, 95% CI=0.83-0.92)48. Another two studies both reported no significant difference in the reduction in all-cause hospitalizations47 or the number of hospital admissions/outpatient attendances or the number of deaths 50.
The three types of PCTs showed mixed results on objective and validated outcome measures (Table 3). Despite different measures reported, the three types of PCTs all reported important improvements in measures related to mental and psychological health. The differences between three types of PCTs, however, were not able to lead to a consistent result.
Patient-reported outcomes and experiences
Seven studies presented patient-reported health outcomes and experiences (Table 2). Four studies reported patients’ satisfaction with care, and all found significant greater satisfaction36, 37, 44, 46, for example, Katon et al. found that patients receiving care from PCT were significantly more satisfied with care after 6 months (OR=2.01; 95% CI=1.18-3.43) and 12 months (OR=2.88, 95% CI=1.67-4.97) 36. For perceptions of improvement, four studies reported this measure and all found better results in the intervention group36, 39, 40, 44, although one study found the effect “slight more favorable” because the average point scores lie between categories of “no change” and “a little better” (Mean=3.52 vs 3.97, P=0.011)44. Studies also found care was more patient-centered delivered by PCT46, 50, for example, the standardized difference of patient-centeredness measured by PACIC was 0.39.
Studies of downward PCT did not present the patient-reported outcomes and experiences, which was subjected to risk of reported bias. The upward PCT and downward PCTs both performed better in this aspect. Patients were found to have better perceived quality of care39, 40, and greater satisfaction with care36, 37. The traditional PCT seemed to enhanced patient-centeredness better, with two studies reported improvements in this measure46, 50.
Changes in patient behaviors
Nine studies reported outcomes on patient behaviors (Table 2). Four studies reported medication adherence, and only one study found significant improvement (OR=2.18, 95%CI=1.32-3.62) 36. Seven studies provided results of patients’ lifestyle changes. The three studies that reported smoking status found no significant improvement 38, 43, 51. Findings on exercises were mixed, Morgan et al. reported a 19% increase in exercise49, while the other two studies did not find significant improvement37, 43. As for diet habits, one study found improvement in self-reported salt intake (Difference=15.4, 95% CI=4.4-26.0, P =0.004)43. Another study found that the intervention group reduced saturated fat intake by 17.3% (95%CI=6.4-28.2, P=0.009) and increased intake of fruit and vegetables by 37.3% (95%CI=18.1-56.5, P=0.004) 51.
Traditional PCTs and downward PCTs had better performance than upward PCTs for patient behaviors. In the traditional PCTs and down PCTs, patients changed their diet habits and exercise. Wood et al. found patients reduced consumption of saturated fat and increased consumption of fruit, vegetables and oily fish, and Towfighi et al. also found patients reported they reduced their salt intake43, 51.
Process measures
Five studies reported process measures (Table 2). Chen et al. reported that the change in the testing for HbA1c did not differ between the intervention and control group, but testing for LDL was significantly lower in the intervention group (difference=-5.8%, P=0.001) 45. For the intervention group, measurement of BMI (+85%, P<0.001), assessing smoking status (+82.8%, P<0.001), making self-management plan (+35.6%, P<0.001) increased significantly compared with baseline. Lin et al. found that initiation and adjustment of medication increased significantly in the intervention group, and the RR was 6.2 (P<0.001), 1.86 (P<0.001), and 2.97 (P<0.001) for antidepressants, antihypertensive drugs, and insulin, respectively. Salisbury reported consultations and continuity of care for process measure50. The continuity of care was significantly higher in the intervention group (Difference=0.08, 95%CI=0.02-0.13, P=0.0045). Patients in the intervention group also had more consultations with primary care nurses (Difference=1.37, 95%CI=1.17-1.61, P<0.001) and primary care physicians (Difference=1.13, 95%CI=1.02-1.25, P=0.021), but not in hospital admissions (Deference=1.04, 95%CI=0.84-1.30, P=-0.71) and hospital outpatient attendances (Difference=1.02, 95%CI=0.92-1.14, P=0.72). Katon et al. also found that patients in the intervention group were more likely to revive medication adjustment, including insulin (P=0.006), antihypertensive agents (P<0.001), and antidepressants (P<0.001) 37. Morgan et at. also detected more consultations to mental health workers in the intervention group (+17% vs -3%).
Two studies of upward PCTs reported medication adjustments, for example, insulin and antihypertensive medications35, 37. For traditional PCTs, Wood et al. found more prescriptions of statins, angiotensin-converting enzyme inhibitors were provided51. Studies of traditional PCTs had more process measures reported than the other two types of PCTs, and also found a great deal of changes in processes. Some of the changes, such as increased self-management plan formulation, showed a promising effect for patients with multimorbidity45.
Mechanisms PCT on quality of care
The discussions of the mechanisms of PCT on quality of care in the seventeen studies were extracted and reported in Table S1 (supplementary file 3). The articles highlighted several common mechanisms by which PCTs achieved effective or ineffective results. First, support for team members and patients from leadership was fundamental and foremost for PCT implementation and effectiveness45. Chen et al. considered that active participation and support from departmental leadership was critical in implementing and sustaining the Teamlet intervention. Second, changing the organization of care played an important role in increasing coordination of care, for example, integrating mental care41, 46 or cancer care40 with primary care. Integrating mental care for patients with chronic conditions comorbid with depression reduced depressive symptoms in patients with chronic conditions, but how to integrate mental and physical healthcare in patients with broader multimorbidity was not certain46. Third, timely follow-up and medication adjustment were important to enhance medication adherence and achieve treatment goals35, 37. Adjusting medication portfolio timely could guarantee a higher chance of improvements in clinical goals and also serve as a reminder for patients’ adherence. Fourth, strong support for patient self-management provided by PCT was reported to contribute strongly to improving quality of care35, 37, 43, 49. Nurses educated patients about essential skills and knowledge in managing their conditions and changing their lifestyle36.