Summary of findings 12. Primary health professional‐led or collaborative care compared to specialist‐led care for people with severe mental disorder in low‐ and middle‐income countries.
| What are the effects of primary health professional‐led or collaborative care vs specialist‐led care for people with severe mental disorders? | ||||||
|
Patient or population: people with severe mental disorder Setting: low‐ and middle‐income countries (China (5 studies), Iran (2 studies)) Intervention: primary health professional‐led or collaborative care Comparison: specialist‐led care | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect (95% CI) | №. of participants (studies) | Certainty of the evidence (GRADE) | Comments | |
| Risk with specialist‐led care | Risk with primary health professional‐led or collaborative care | |||||
|
Clinical recovery from severe mental disorder (immediately post intervention) Defined by BPRS decreased by ≥ 80% (RR > 1 denotes greater likelihood of recovery) |
684 per 1000 participants | 739 per 1000 participants (554 to 985) | RR 1.08 (0.81 to 1.44) | 76 (1 RCT)a | ⨁⊝⊝⊝ VERY LOWb | It is uncertain whether PHP‐led collaborative care compared to specialist‐led care for people with severe mental disorder has any effect on clinical recovery immediately post intervention |
|
Relapse of severe mental disorder (immediately post intervention) Defined by re‐appearance of symptoms or worsening of symptoms necessitating adjustment of medicationa; Li 2002; based on 1 item scoring ≥ 5 or 2 items scoring ≥ 4 in items 4, 7, 11, 12, and 15 of the BPRSc; determined clinicallyd,e (RR > 1 denotes higher risk of relapse) |
211 per 1000 participants | 63 per 1000 participants (34 to 116) | RR 0.30 (0.16 to 0.55) | 492 (4 RCTs) | ⨁⊝⊝⊝ VERY LOWf | It is uncertain whether PHP‐led or collaborative care compared to specialist‐led care for people with severe mental disorder have any effect on relapse |
| Prevalence of severe mental disorder | No studies that reported on this outcome were identified | |||||
|
Schizophrenia symptoms severity (immediately post intervention) Assessed with BPRSa,c,d,g; PANSSh,i (higher score = greater severity) |
Mean PANSS score with usual care was 80.6i | 836 (6 RCTs) | ⨁⊝⊝⊝ VERY LOWj | Scores estimated based on an SMD ‐0.30 (95% CI ‐0.71 to 0.11). It is uncertain whether PHP‐led or collaborative care compared to specialist‐led care for people with severe mental disorder has any effect on schizophrenia symptom severity immediately post intervention | ||
|
Depression symptom severity (immediately post intervention) Assessed with HDRSb; SCL‐90d (higher score = higher symptom severity) |
Mean Hamilton Rating Scale for Depression score with usual care was 11.9b | Mean Hamilton Rating Scale for Depression score in the intervention group was 3 (8.2 lower to 2.3 higher) lower | 270 (2 RCTs) | ⨁⊝⊝⊝ VERY LOWk | Scores estimated based on an SMD of ‐0.41 (95% CI ‐1.13 to 0.32). It is uncertain whether PHP‐led or collaborative care compared to specialist‐led care for people with severe mental disorders has any effect on severity of depression symptoms immediately post intervention | |
|
Quality of life (immediately post intervention) Assessed with SF‐36i; WHOQOL BREFe,h (higher score = higher quality of life) |
Mean WHOBREF score with usual care was 84.4 | Mean WHOBREF score in the intervention group was 9.04 (8.36 lower to 26.4 higher) higher | 536 (3 RCTs) | ⨁⊝⊝⊝ VERY LOWl | Scores estimated based on an SMD of 0.40 (95% CI ‐0.37 to 1.17). It is uncertain whether PHP‐led or collaborative care compared to specialist‐led care for people with severe mental disorders has any effect on quality of life immediately post intervention | |
|
Functional impairment ‐ immediately post intervention Assessed with GAF (results were multiplied by ‐1)h; KELSi; SDSSa,c,d,g; self care ADL and Instrumental ADLe (lower score = less functional impairment) |
Mean function score with usual care was 9.58i | Mean function score in the intervention group was 5.0 (7.8 to 2.1) lower | 874 (7 RCTs) | ⨁⨁⊝⊝ LOWm | Scores estimated based on an SMD of ‐1.13 (95% CI ‐1.78 to ‐0.47). PHP‐led or collaborative care for people with severe mental disorders may reduce functional impairment immediately post intervention compared to specialist‐led care | |
|
Service utilisation ‐ hospital re‐admission (during intervention) (RR > 1 denotes greater risk) |
360 per 1000 | 216 per 1000 (101 to 461) | RR 0.60 (0.28 to 1.28) | 441 (3 RCTs)e,h,i | ⨁⊝⊝⊝ VERY LOWn | It is uncertain whether PHP‐led or collaborative care compared to specialist‐led care for people with severe mental disorders has any effect on hospital re‐admission immediately post intervention |
| Adverse events | No studies that reported on this outcome were identified | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADL: activities of daily living; BPRS: Brief Psychiatric Rating Scale; CI: confidence interval; GAF: Global Assessment of Functioning Scale; GP: general practitioner; HDRS: Hamilton Depression Rating Scale; KELS: Kohlman Evaluation of Living Skills; LMIC: low‐ to middle‐income country; PANSS: Positive and Negative Syndrome Scale; PHP: primary health professional; RCT: randomised clinical trial; RR: risk ratio; SCL‐90: Symptom Checklist‐90; SDSS: Social Disability Screening Schedule; SF‐36: 36‐Item Short‐Form Health Survey; SMD: standardised mean difference; WHOQOL BREF: World Health Organization Quality of Life assessment abbreviated. | ||||||
| GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | ||||||
aLi 2002. Community rehabilitation (medications, counselling, requested work or social activities) vs inpatient care for patients with first episode of late‐onset schizophrenia.
bDowngraded by one level for risk of bias: high risk of selection (lack of allocation concealment) and detection (lack of blinding of outcome assessors) bias. Downgraded by one level for indirectness: the only study in this analysis was conducted in a lower‐middle‐income country, on people age 50 and older experiencing their first episode of illness (schizophrenia); intervention comprised antipsychotics and weekly home visits. Study population, setting, and intervention are not generalisable to all patients with serious mental disorders in LMICs. Downgraded by one level for imprecision: small event numbers.
cTan 2005. Community observation (medications, symptom monitoring, psychoeducation, rehabilitation) vs hospitalisation as needed for patients with schizophrenia.
dLing 1999. Family intervention (education on medication side effects and adherence, symptom monitoring, psychoeducation, counselling, family communication training) vs community psychiatric nurse‐led care for patients with schizophrenia.
eWu 2016. Self‐care model combined with collaborative care (medications, counselling, family communication training, requested work or social activities, self‐care training) vs community psychiatric nurse‐led care for patients with chronic stable schizophrenia.
fDowngraded by one level for risk of bias: high risk of detection bias (lack of blinding in outcome assessments) in all four studies, and selection bias (lack of allocation concealment) in one study (Li 2002). Downgraded by one level for indirectness: only one country represented among these studies. Downgraded by one level for imprecision: Small total event numbers.
gYao 2014. Community day rehabilitation (medications, symptom monitoring, psychoeducation, counselling, social skills training, rehabilitation) vs community psychiatric nurse‐led care for patients with chronic stable schizophrenia.
hBarfar 2017. Aftercare service (medications, education on medication side effects and adherence, symptom monitoring, psychoeducation, telephone reminders to attend outpatient clinics, social skills training) vs usual specialist care in outpatient clinics or inpatient services for patients with schizophrenia, bipolar disorder, or schizoaffective disorder.
iMalakouti 2015. Home visits by nurse or GP (medications, education on medication side effects and adherence, symptom monitoring, psychoeducation) vs usual specialist‐led outpatient clinic or hospitalisation during exacerbation for patients with schizophrenia or bipolar disorder with difficult‐to‐treat disease.
jDowngraded by one level for risk of bias: high risk of detection bias (lack of blinding in outcome assessments) in four studies (Barfar 2017; Li 2002; Ling 1999; Tan 2005), selection bias (lack of allocation concealment) in one study (Li 2002), and attrition bias in one study (Malakouti 2015). Downgraded by one level for inconsistency: large unexplained statistical heterogeneity (I² = 87%). Downgraded by one level for imprecision: confidence level of SMD ranges from moderate clinical effect favouring PHP‐led or collaborative care to no clinical effect. Note that a small clinically appreciable benefit was set at SMD 0.2 to 0.5, a moderate benefit at SMD 0.5 to 0.8, and a large benefit at > 0.8 (Cohen 1988).
kDowngraded by one level for risk of bias: high risk of detection bias in Barfar 2017 and Ling 1999. Downgraded by one level for inconsistency: large unexplained statistical heterogeneity (I² = 89%). Downgraded by one level for imprecision: low total number. Confidence interval of SMD ranged from favouring PHP‐led or collaborative care to favouring specialist‐led care.
lDowngraded by one level for risk of bias: high risk of detection bias in Barfar 2017 and Wu 2016, and attrition bias in Malakouti 2015. Downgraded by one level for inconsistency: large unexplained statistical heterogeneity (I² = 94%). Downgraded by one level for imprecision: confidence interval of effect size ranges from large clinical effect favouring PHP‐led or collaborative care to favouring specialist‐led care.
mDowngraded by one level for risk of bias: high risk of detection bias (lack of blinding in outcome assessments) in five studies (Barfar 2017; Li 2002; Ling 1999; Tan 2005; Wu 2016), selection bias (lack of allocation concealment) in one study (Li 2002), and attrition bias in one study (Malakouti 2015). Downgraded by one level for inconsistency: large unexplained statistical heterogeneity (I² = 95%).
nDowngraded by one level for risk of bias: high risk of detection bias in Barfar 2017 and Wu 2016, and attrition bias in Malakouti 2015. Downgraded by one level for inconsistency: moderate unexplained statistical heterogeneity (I² = 74%). Downgraded by one level for imprecision: small event numbers. Confidence interval of risk ratio ranged from favouring PHP‐led or collaborative care to favouring specialist‐led care.