Collaborative care for the detection and management of depression among adults with hypertension in South Africa: study protocol for the PRIME-SA randomised controlled trial

Cardiovascular disease (CVD) (including coronary heart disease (CHD), hypertension and stroke) is the leading cause of mortality in the world [1]. Even in contexts such as South Africa, which has a large human immunodeficiency virus (HIV) burden, CVD is among the top three causes of years of life lost (YLLs) [2]. In relation to disability-adjusted life years (DALYs), which include quantification of YLLs and years lived with disability (YLD), data from South Africa shows a rapid rise in NCDs [2]. Estimates by the World Health Organisation (WHO) using DALYs suggest that NCDs were responsible for 28% of the total burden of disease in South Africa in 2004 [3]. These conditions are becoming more prevalent with the diffusion of urban lifestyle risk factors, including tobacco smoking, unhealthy diet, alcohol consumption and physical inactivity [4‐6].

The prevalence of hypertension, at 20% of the adult population, is of particular concern [7], with CVD projected to increase by over 40% in the 35–65 years age group by 2030 [6]. A survey of over 18,000 encounters showed that hypertension follow-up was the leading reason for attending primary care clinics, accounting for 10% of all consultations among adults [8]. Similarly, in a study at a primary care clinic in Khayelitsha, Cape Town, hypertension was the most common morbidity (65%) among 14,364 adult chronic disease patients with tuberculosis (TB), HIV, diabetes or hypertension [9].

NCDs often co-exist with one another, as well as with mental disorders. Results from 245,404 participants from the World Health Surveys in 60 countries revealed that co-morbid depression was present in respondents with diabetes (9.3%), arthritis (10.7%), angina (15%) and asthma (18.1%) [10]. Prevalence rates of depression were highest when there was co-morbidity of two or more chronic physical conditions (23%). Relatively high rates of co-morbidity of NCDs are increasingly being reported in patients in South Africa [11], with a local study also revealing a similar pattern where multiple co-morbidities were associated with increased psychological symptoms [12].

Prevalence of depression in respondents from the World Health Surveys without chronic diseases on the other hand was less than 5% [13]. This suggests that people with NCDs are two to five times more likely to suffer from depression than those without NCDs, with the likelihood increasing with multiple co-morbidities. Depression co-existing with other NCDs was also shown to result in greater health decrements than when one or more physical conditions were present [10]. Mental disorders have a mutually reinforcing relationship with NCDs [14], compromising both prevention and treatment through exacerbating modifiable risk factors, and by compromising adherence and self-care, respectively [15]. In addition, through compromising the endocrine and immune systems, depression and anxiety can have a negative interactive effect at the biological level, exacerbating vulnerability to, and the course of, existing NCDs [16].

With respect to CVD in particular, co-morbid depression increases the risk of CHD and stroke [17‐24]. Risk factors for CHD and stroke are related to unhealthy lifestyles which are more common among people with depression or chronic ill-health. Depressed people are more likely to smoke and less likely to exercise, adhere to chronic medication, or eat healthily [25, 26]. It is also possible that there are common causal physiological pathways, such as inflammation, which underlie both CVD and depression [16]. With regard to the impact of common mental disorders on adherence in hypertensive patients specifically, some studies from high-income countries (HICs) show higher non-adherence in patients with co-morbid depression [27]. However, it should be noted that a recent study in Ghana found that there was no relationship between co-morbid depression and/or anxiety and non-adherence, although they did find a relationship between moderate to severe stress symptoms and non-adherence [28].

In response to the rising burden of co-existing chronic diseases that includes HIV, TB and NCDs, the South African National Department of Health has piloted a multi-disease, integrated, chronic services approach [29]. Care for chronic conditions requires collaborative care, which extends beyond the biomedical model that characterises acute care, to incorporate the following elements: self-management support and access to behavioural-change programmes (to help patients have control over their illness and live with their condition); clinical decision support (to improve providers’ expertise and skills to provide the most evidence-informed care); clinical information systems (capturing patient-level information that will help clinical care and follow-up); delivery system design to promote chronic care (promoting planned team-based care and identifying the roles and responsibilities of the different members of the health care team so that they can work synergistically); linkages to community resources such as support groups; and quality improvement to improve the organisation of health care as a whole [30‐32].

Such integrated collaborative care for clusters of co-existing illnesses described above, especially physical and mental disorders, has been shown to be more effective than usual care in HICs. Exemplary evidence of this is provided by the TEAMcare trial in the United States which targeted patients with depression and poorly controlled diabetes, heart disease or both, using nurse-led, multi-disease collaborative care [33]. The TEAMcare intervention combined pharmacotherapy with psychosocial interventions, supported by self-help materials, to assist patients to solve problems and set goals to improve adherence and self-care.

In South Africa, at the facility level, the integrated chronic care approach which is part of the integrated clinical management service, aims to strengthen the quality of care for chronic conditions through, inter alia: (1) facility re-organisation to improve the quality of care and servicing all chronic care patients at one service point and (2) strengthening clinical decision support through onsite in-service training using an integrated set of algorithms and checklists developed to support nurse-led identification and management of all common chronic diseases whether communicable, NCDs or mental illness – called PC101 [34]. At the community level, well-controlled patients are supported to engage in self-management by community health worker (CHW)-driven, ward-based outreach teams (WBOTs), which are part of the primary health care (PHC) re-engineering framework in South Africa. At a population level, health promotion and population screening is envisaged to promote an informed and activated population [35].

While mental health problems managed at the PHC level are included in the basket of chronic disease categories catered for by this integrated approach, findings suggest that the limited mental health component included in the basic PC101 training is insufficient to improve PHC nurses’ knowledge of mental health conditions [36], and was not associated with increased case detection of depression in a previous trial focussing on NCDs [37]. There is thus a need to strengthen the mental health component of this integrated approach in order to close the large treatment gap for mental disorders in South Africa, particularly in the chronic care population. While one in six adults experience a common mental disorder (depression, anxiety disorders and substance use disorders) over a 12-month period [38], only one in four receives treatment of any kind [39].

The Programme for improving mental health care (PRIME) [40] in South Africa has been working in one of the districts where this integrated model has been piloted for the past 5 years to develop and evaluate the feasibility of an integrated collaborative-care package for depression and alcohol use disorders (AUD). The package includes strengthening of the mental health component of PC101 training and referral pathways for management of AUD and depression [41]; the use of lay behavioural health counsellors to provide manual-based counselling for chronic care patients with depression, shown to have promising outcomes in a pilot trial [42]; strengthened referral pathways for initiation of psychotropic medication by primary health care physicians; as well as referral to mental health specialists for more severe and treatment resistant depression. In addition, systems strengthening innovations to support these efforts have been introduced through the EMERALD (Emerging Mental Health Systems in Low- and Middle-income Countries (LIMCs)) programme [43], including clinical communication skills for nurse-led chronic care and a strengthened employee assistance programme for primary health care providers experiencing personal problems and burn-out.

Development of the strengthened mental health intervention has been iterative, guided by the United Kingdom (UK) Medical Research Council (MRC) framework for the development of complex interventions [44] and refined during two successive pilots, the first in a single clinic, and the second in a further three clinics. The resultant intervention has been scaled up to 10 clinics in the Dr. Kenneth Kaunda (DKK) District and a further 10 clinics in the adjacent Bojanala Platinum (BP) District during the period 2014–2017. The intervention targets depression co-morbid with all common chronic conditions managed at primary care facilities.

This paper describes one of a pair of trials evaluating the effects of this intervention on mental and physical outcomes among chronic care patients. The other trial (Co-morbid Affective Disorders, HIV and Long-term Health – CobALT), focusses on chronically ill patients receiving antiretroviral (ART) treatment. These two conditions covered by this pair of trials account for most chronic care attendances in South Africa. Similarities and differences between the two trials are described in Table 1. The main distinction is the restriction of the trial reported in this paper to one of these districts, the eligibility criteria for patient participants and the choice of physical outcomes.

The study design is a pragmatic, two-arm, parallel-cluster randomised controlled trial (RCT) stratified by sub-district. The unit of randomisation is the clinics, and we measure outcomes on individual participants.

The study site is the DKK District in the North West Province of South Africa located adjacent to and west of the populous Gauteng region (see Fig. 2). The South African National Department of Health guided selection of the site as it is a priority district for several initiatives to strengthen primary care, including the integrated chronic care approach and the establishment of ward-based teams of CHWs. DKK is also one of 11 districts where national health insurance is being phased in over a 14-year period [46]. DKK comprises four sub-districts, with a population of 796,823 [41]. The main economic activities are mining and agriculture. Public health facilities include regional hospitals, primary health care facilities and one specialist in-patient mental health facility.

Control arm: the integrated chronic care approach implemented by the Department of Health (described in the introduction) forms the control comparison and has been implemented in both intervention and control clinics in the year prior to the introduction of the study intervention with various levels of technical support provided by the Department of Health. At facility level the main interventions have been the re-organisation of chronic care services and training of clinic staff in PC101. Queues, appointment systems and waiting rooms for communicable disease services (HIV, TB once the intensive phase of treatment has been completed) have been integrated with services catering for NCDs and mental health, resulting in a single chronic care service aimed at reducing fragmentation and promoting integrated care [47]. Nurses in the control facilities are able to refer patients identified as having depression to PHC physicians for the initiation of antidepressants (as they are not authorised to prescribe antidepressant medications in South Africa); as well as to mental health specialists. The latter are, however, limited with only two district-level PHC psychologists available to service all PHC facilities in the DKK District. Limited specialist psychological and psychiatric care is also available at the district hospital.

Intervention arm: in the intervention clinics, we strengthened the training of PHC nurse clinicians in the identification and management of depression through training in clinical communication skills and supplementary PC101 mental health sessions (see Additional file 3 of web-based supplementary material for examples of PC101 guidelines). We trained facility trainers in the intervention facilities to deliver this supplementary training material. Facility-based trainers in the control facilities are not exposed to this training in the delivery of this supplementary material. In addition, we strengthened referral pathways for treatment and counselling (see the stepped-up collaborative-care model depicted in Fig. 3). For treatment, we provided supplementary training in mental health care for PHC physicians. For counselling, we introduced lay behavioural health counsellors into the intervention clinics, with structured supervision from the project psychologist and district-based psychologists that includes individual and weekly group supervision. A comparison of the training provided in the control and intervention clinics is contained in Table 2. The intervention coordinator is responsible for monitoring the intervention to ensure that each component is functioning optimally at the facility level. A detailed monthly monitoring document was developed and is used to monitor coverage of nurse training in the supplementary material given high nurse turnover, nurse referrals to counsellors and counsellor integration within the facility.

We considered all 60 public sector primary care clinics in the DKK District for enrolment in the trial and excluded clinics if they were mobile or satellite units, saw fewer than 10,000 attendances per year, or participated in the piloting of the intervention or data collection. We enrolled the largest 20 eligible clinics in the trial.

Intervention clinic staff cannot be blinded to the clinic’s treatment status given that they have to agree to additional training activities over and above the standard PC101 training; as well as the introduction of behavioural health counsellors. Although the outcome assessors (the fieldworkers) are not informed as to which clinics are in the intervention arm and which are in the control arm, it is not possible to guarantee complete blinding of the fieldworkers given the abovementioned intervention activities at the intervention clinics. To minimise the potential for bias, all patient screening, recruitment, interviewing, follow-up and quality assurance procedures are standardised across both arms to ensure that these data collection activities and procedures are applied equally in all clinics. The data collection team also operates independently of the intervention team, so as to limit exposure to the intervention.

We recruit patient participants from the ambulatory primary care population attending chronic care services at the trial clinics. Participants are considered to be eligible for the trial if they are aged 18 years or older, report receiving treatment for hypertension and screen positive for depression with a total score of 9 or more on the (PHQ-9) [48] (see Fig. 4). We chose the PHQ-9 given that it has been commonly used in other trials on depression [42, 49‐51], and is relatively short and easy to train fieldworkers to administer. Most importantly, it has also been validated on the target population in South Africa [52]. We did consider other depression measures such as the Hamilton Rating Scale for Depression (HAM-D). The HAM-D, however, can take up to 15 to 30 min to administer, is ideally clinician administered and is also complicated to score, requiring substantial training in getting reasonable inter-rater agreement [53]. Given that this is a pragmatic trial, our preference was to use a measure that could be relatively easier to administer and which had been validated in the population of interest. The PHQ-9 is aligned with the Diagnostic and Statistical Manual of Mental Disorders-Text Revision (DSM-IV-TR) diagnostic criteria for major depressive disorder and the cut-off of 9 has been determined for the study population in a preliminary validation study involving 676 chronic care patients also from the DKK District. A cut-off of 9 and above was shown through the validation study to be the optimal cut-off for sensitivity and specificity indices against the ‘gold standard’ of a Structured Clinical Interview for a DSM-IV (SCID) diagnosis of major depression [51]. The trials findings will, therefore, apply to patients with major depressive symptoms at baseline. In this validation study, performance of a Setswana translation of the PHQ-9 with an adapted response to improve understanding of the time period when symptoms were experienced was compared against a structured clinical interview administered by a clinical psychologist [52] (see Table 3 for a comparison of the original and localised validated version). Mean PHQ-9 scores at baseline will be compared between intervention and control arms to assess for any difference which could indicate potential selection bias. We exclude participants if they are in need of urgent medical attention, too ill to provide informed consent or plan to move outside the vicinity of the clinic during the 12 month follow-up period (see Fig. 4). We expect some patients to meet eligibility criteria for both pairs of trials owing to the growing overlap between HIV and hypertension, recognise that it is important to represent this co-morbidity within the trials, and so invite such patients to participate in both trials.

We will enrol roughly equal numbers of depressed hypertensive patients into the study using the PHQ-9, across the 20 clinics with a target of 50 participants per clinic, making up a total of over 1000 participants with over 500 in each arm (see Fig. 4).

In the intervention clinics, we recruit patients following training of nurses and counsellors in the intervention as well as a 2-month embedding period. Trained fieldworkers enrol participants in the study independent of clinical care or participation in the counselling intervention for depression. Screening follows a three-stage process:

If patients show signs of suicidality through a positive response to the suicidal question of the PHQ-9, we repeat the question to confirm the response. We give patients who report having suicidal thoughts in the past 2 weeks psychoeducational material on suicidal thoughts and contact information about where to get help. We refer patients with suicidal thoughts on more than 7 days in the past 2 weeks to the nurse on the same day for assessment and onward referral if necessary.

Participants provide written consent before completing the full baseline questionnaire in the language of their choice, predominantly Setswana although some interviews are also conducted in Afrikaans or English. This questionnaire includes items on demographics and multiple instruments (outlined below, Table 4) (see Additional file 6 of supplementary files for full questionnaire). Anthropometric measurements include height, weight, waist circumference and blood pressure. Trained fieldworkers measure the latter using a digital sphygmomanometer, and take readings while the participant is seated at three time points during the course of the interview. We use the ‘Averaging Rules for Determining Mean Blood Pressure’ from the National Health and Nutrition Examination Survey (NHANES) Health Technology/Blood Pressure Procedures Manual [54] to determine the average blood pressure whereby the first measurement is discarded and the second two readings used to calculate a mean blood pressure for each assessment (baseline, 6 months, 12 months).

The primary outcome measure is a reduction in PHQ-9 score of 50% or more compared with baseline [49‐51, 55, 56], evaluated at 6 months. Thus, we use the PHQ-9 to determine eligibility for the trial as well as to measure mental health outcomes at 6 and 12 months.

Secondary outcomes (see Table 4) include: (1) response at 12 months defined as a 50% reduction in the score on the PHQ-9, (2) remission at 12 months defined as a score of < 5 on the PHQ-9, (3) mean PHQ-9 scores at 6 and 12 months, (4) antidepressant treatment, including initiation or intensification of antidepressant medication, (5) referral to a counsellor for depression counselling, (6) referral to a mental health specialist (clinical psychologist, psychiatrist or secondary care services), (7) blood pressure at 6 and 12 months, (8) disability measured at 12 months using the WHO Disability Assessment Schedule version 2.0 (WHODAS 2), (9) stress symptoms measured at 12 months using the Perceived Stress Scale (PSS), (10) patient assessment of quality of chronic illness care received measured at 12 months using the Patient Assessment of Care for Chronic Conditions (PACIC) scale, (11) health care utilisation including clinic visits and hospital admissions, (12) resource use and economic outcomes measured using a Service Use Questionnaire and (13) all-cause mortality measured through follow-up at the clinic and linkage to the South African population register.

The WHODAS 2 [57] is a generic assessment instrument for disability for diseases including mental disorders. We use the 12-item interviewer administered version which has been previously used on a South African ART patient population [58].

The Perceived Stress Measure (PSS) [59] is a 10-item self-reported measure that provides an assessment of the extent to which situations in a person’s life are appraised as stressful. It has been previously used in South Africa with perceptions of high levels of stress associated with higher risk for developing some mental disorders [60].

The Patient Assessment of Care for Chronic Conditions (PACIC) scale is a self-reported measure of quality of patient-centred chronic clinical services received in the past 6 months, which emphasises self-management support in the form of collaborative goal setting, problem-solving and follow-up [61]. The original 20-item scale was adapted for the local context and reduced to 10 items.

Limitations of the trial design include dependence on self-reporting of hypertensive medication for inclusion in the trial, lack of detailed information on prescribed medication for hypertension, and limited data on adherence to medication. While the broad inclusion criteria and light-touch follow-up are consistent with the pragmatic design and are intended to maximise generalisability to an intervention in real-world circumstances, it can be difficult to assess whether all the study participants were eligible for the intervention and to quantify the processes of care in a causal pathway. A qualitative process evaluation will be conducted alongside the RCT to address these gaps and to understand how integrating mental health services for depression affects care of physical conditions like hypertension.

Fieldworkers use handheld devices to collect data that is uploaded onto a secure server. The Trial Management Team manage data quality assurance through weekly telephonic data meetings when uploaded data is reviewed, as well as bimonthly face-to-face meetings of the Data Management Team comprising the data manager, principal investigators (PIs) and invited co-investigators, including the statistician depending on the need. Data is collected alongside the first phase of the CobALT trial in the DKK District, and is subject to the same standard operating procedures for data management for the CobALT trial. These include intensive fieldworker training in the administration of the interview, and onsite supervision by three fieldwork supervisors who perform daily quality checks on interviews conducted, observe interviews weekly to ensure the interviewer administers the questionnaire and performs the measurements as trained, and assists in addressing fieldwork problems that may have impeded data collection.

Patients receive a ZAR50 voucher (just under US$3) in appreciation for their time. We schedule follow-up appointments for 6 and 12 months at the clinic, with reminders and confirmation of follow-up interviews provided using text messaging and, where appropriate, telephone calls.

The PIs have established a Trial Steering Committee comprising key national, provincial and district stakeholders including policy-makers, planners and managers (see Additional file 1 supplementary web-based material for more detail) where trial progress is being reported and provides a platform to disseminate the trial results and partake in discussions around policy implications. These stakeholders have expressed their commitment to implementing the intervention package should the trial results be successful. Scale-up would require, inter alia, the expansion of the PC101 training to include clinical communication skills; a dedicated mental health module; and role clarification of the lay behavioural health counsellors within the collaborative-care model.

Further, the results of the trial will be written up in a main outcome publication within a year of the finalisation of the fieldwork, together with additional papers detailing secondary outcomes and sub-group analyses. Authorship will be determined on the basis of contributions made to the design, intervention, data collection and analysis.

At the time of submission (September 2016) the trial is in follow-up stage. Recruitment and baseline data collection began in April 2015, and will continue until the end of October 2016.