Assessing capability for implementing mental health counselling within primary care facilities in a middle-income country: A feasibility study

Abstract

Introduction

Differences in primary care facilities’ preparedness for implementing mental health counselling may affect the implementation process but have rarely been studied.

Aim

To assess the feasibility of using a novel methodological approach to explore variations in capability for implementing mental health counselling and factors potentially associated with this variation among primary care services in the Western Cape, South Africa.

Methods

Staff from 26 facilities participated in discussions about their facility’s mental health implementation capability. Three researchers conducted observations of the facility’s environment, staff-patient interactions, and resources. We used qualitative comparative analysis to identify factors potentially associated with implementation capability.

Results

Facilities appeared to vary in their capability for implementing counselling services. The availability of person-centred health services, a therapeutic environment and sufficient human resources may be requirements for implementation preparedness. Other factors that seem to support preparedness include the availability of confidential space for counselling and an adequately managed facility.

Discussion

This study identified several features of well-functioning primary care facilities. Facilities with these features may be better prepared to implement a new counselling service.

Implications for practice

This method may identify facilities that are poorly prepared for implementation that could benefit from preparedness-building interventions. Whether differences in preparedness affect counselling outcomes is yet to be established.

1. Introduction

Like many low- and middle-income countries (LMIC), the health system in South Africa (a middle-income country) is challenged by a high prevalence of untreated mental disorders (McBain, Salhi, Morris, Salomon, & Betancourt, 2012; Whiteford et al., 2013) that contributes to the acquisition and progression of infectious diseases and non-communicable diseases (Ammassari, Trotta, Muri, Castelli, Narciso, & Noto, 2002; Brown et al., 2016). Reducing this mental health treatment gap is likely to improve the overall effectiveness of health systems in LMICs.

The integration of mental health counselling into primary health care services has been proposed as a strategy for reducing this treatment gap (Maya et al., 2015; Patel et al., 2013). To overcome the challenge of the chronic shortage of specialist mental health workers in LMICs (Weinmann & Koesters, 2016), a task sharing approach to the provision of mental health counselling services has been recommended. This involves shifting some of the responsibility for providing mental health counselling (in the form of brief, structured psychological interventions) from specialist providers (such as nurses with an additional specialist qualification in psychiatry and mental health) to trained non-specialist health workers who work under the supervision of specialists. Several systematic reviews and meta-analyses have shown that the provision of evidence-based mental health counselling (for example, cognitive behavioural and interpersonal therapy) for common mental disorders (such as depression, anxiety and alcohol use disorders) by non-specialist health workers can be effective (Clarke, King, & Prost, 2013; Rahman et al., 2013; Singla et al., 2017; van Ginneken et al., 2013). To illustrate, a recent systematic review of 27 trials conducted in LMICs found that psychological interventions for common mental disorders delivered by non-specialist providers had moderate to strong effects in reducing symptoms (pooled effect size 0.49, 95% confidence interval = 0.36–0.62) (Singla et al., 2017).

Since 2013, the South African National Department of Health has supported a task sharing approach to mental health counselling provision, with a focus on training community health workers (CHWs) to provide counselling (South African Department of Health, 2013). In this setting, CHWs are lay health workers trained to provide a narrowly defined set of health services (such as HIV testing, counselling and adherence support services) within their community and the primary care facilities located in their community. Although specialist providers, such as nurses with training in mental health, are available in the primary care system, their main function is the administration of medication to patients with severe mental illness. As they are a scarce resource and often work across several facilities, they have limited time to provide counselling for common mental disorders (Myers et al., 2018a). Task sharing this function with CHWs could potentially reduce the treatment gap in this context.

To facilitate the successful implementation of CHW-delivered mental health counselling within primary health care (PHC) facilities, barriers to implementation should be identified and addressed. In the implementation science literature, organisational preparedness or capability for implementation is viewed as a precursor to the adoption of health service change, with efforts to build preparedness often forming part of the implementation planning process (Aarons, Hurlburt, & Horwitz, 2011; Damschroder et al., 2009; Powell et al., 2012). Although research has shown that variations in organisational preparedness may influence the degree to which proposed service changes are accepted and adopted (Rubenstein et al., 2013; Scaccia et al., 2015), health service preparedness for change has rarely been studied in LMIC settings. To the best of our knowledge, studies have not yet explored whether there are differences in the preparedness of PHC facilities for implementing mental health counselling or the organisational features that may be associated with these differences. Mental health service planners need this information to identify facilities that may be poorly prepared for implementation and to guide the evidence-based selection of strategies to support implementation (Scaccia et al., 2015).

Although a mixed-methods approach that combines observations of health care with surveys and in-depth interviews of health providers would generate rich data and provide opportunities to examine implementation preparedness using multiple sources, this approach is costly and was beyond the financial scope of this study which used a novel and cost effective evaluation method as a first step. This approach involved observations of primary care delivery that allowed us to observe staff and staff-patient interactions in a natural, non-disruptive manner (Dixson-Woods & Bosk, 2010) and qualitative comparative analysis for identifying factors potentially associated with implementation preparedness across diverse PHC facilities (Kane, Lewis, Williams, & Kahwati, 2014).

1.1. Aims

The aim of this paper is to assess the feasibility of using this method to explore variations in capability for implementing mental health counselling and for identifying factors potentially associated with this variation among primary care services in the Western Cape, South Africa.

2. Methods

This manuscript complies with the Consolidated Criteria for Reporting Qualitative Research (Tong, Sainsbury, & Craig, 2007; see supplementary file 1). As part of a larger study aimed at integrating mental health care into chronic disease services, known as Project MIND (Myers et al., 2018b), we conducted observations of chronic disease service delivery at PHC clinics in the Western Cape Province. Project MIND is a cluster randomised controlled trial (Trial registration number: PACTR201610001825403) that is comparing two different approaches to integrating mental health counselling into HIV and diabetes care, two highly prevalent chronic conditions in South Africa. The choice of these chronic conditions was influenced by the high prevalence of untreated common mental disorders in these patient populations and their impact on treatment adherence and outcomes (Myers et al., 2018b).

2.1. Study sites

PHC clinics were eligible to be study sites if they provided both HIV and diabetes treatment services. The Western Cape Department of Health (WCDOH) purposively selected 26 nurse-led PHC facilities (19 in urban and seven in rural communities) as potential sites for Project MIND. These facilities were selected from a total of 189 clinics (of which 101 were in urban and 88 in rural areas) situated in four of the province’s six health districts. Clinics from two rural health districts were excluded from the study as these districts were demonstration sites for other health system interventions. These 26 sites were selected to broadly reflect the geographic distribution and variability in size, structure and organisation of clinics in the province but the extent to which this sample is representative of the 189 clinics is not known. Table 1 provides a description of the selected sites.

Table 1. Organisational characteristics and implementation capability of participating PHC facilities (n= 26).

Facility	Implementation capability a	Facility Size b	Number of patient visits/ year	Facility location	Comprehensive servicesc
1	high	large	100558	rural	yes
2	low	small	55237	rural	partially
3	moderate-high	large	129348	rural	yes
4	low	small	114342	urban	no
5	low-moderate	medium	281,690	urban	partially
6	high	small	123074	urban	no
7	high	large	228241	urban	yes
8	high	large	97448	urban	yes
9	low	small	73272	urban	no
10	low-moderate	medium	122322	urban	partially
11	low	large	318303	urban	partially
12	low	medium	157492	urban	no
13	low-moderate	large	286161	urban	yes
14	high	large	229356	urban	yes
15	low	large	152853	urban	yes
16	moderate-high	medium	83679	urban	no
17	high	small	83887	urban	no
18	high	medium	145178	urban	no
19	high	large	489307	urban	yes
20	high	large	427561	urban	yes
21	moderate-high	large	420795	urban	yes
22	moderate-high	large	247850	urban	yes
23	high	large	122888	rural	yes
24	high	large	195254	rural	no
25	low	large	195254	rural	no
26	moderate-high	small	195254	rural	no

^afacilities with low implementation capability as unable to implement the programme; low-moderate capability as needing significant external support; moderate-high capability as requiring moderate support; and those with high implementation capability as needing minimal support to implement the programme.

^bsize refers to physical space of facility including size of grounds and number of consultations rooms available. Large facilities were 24 hour facilities that comprised more than 3 buildings and outbuildings, had separate consultation space available for mental health, HIV and diabetes and may also have provided mobile services to outlying facilities; medium facilities comprised 2-3 buildings and had separate consultation space available for HIV and diabetes but not mental health; small facilities comprised a single building had did not have separate consultation space available for HIV, diabetes and mental health care.

^ccomprehensive services, yes = obstetric and emergency services in addition to basic PHC services, partial= obstetric or emergency services and PHC services; no = only basic PHC services

2.2. Procedures

Observations were conducted at all of the selected facilities between September 2015 and February 2016. Prior to the observation period, two researchers conducted facility-wide meetings where they provided an overview of Project MIND and facilitated a discussion about the proposed programme and the facility’s capacity to implement the MIND counselling intervention. Researchers answered clinic staff’s questions about the proposed counselling programme and how it might work, thereafter clinic staff were asked their opinions as to whether the facility would be able to implement the intervention and the amount of training support, monitoring and external resources they would require for implementation. A researcher took notes of the discussion, with the verbal consent of staff. Discussions were not audio-recorded. To protect the anonymity of participants, their names and designations were not recorded. After these meetings, the researchers met individually with staff working in the HIV and diabetes services to explain the observational components of the study and request their written consent to participate in the study. The staff were given an opportunity to opt-out of the observations, although none did.

Three research workers (two women and one man), who were blind to the discussions of facilities’ implementation capability, independently conducted observations of chronic disease care at each of the clinics. These research workers had training in qualitative research methods, experience in collecting observational data, and were proficient in the local languages spoken in the clinics’ catchment area. On average, research workers spent seven hours (range three to twelve) collecting data from each clinic, with some observations occurring on more than one day. Differences in the size and patient caseloads of these facilities accounted for some of the variability in the time spent conducting observations at each. As research workers observed the process of care until no new information was being collected, less time was spent at smaller facilities that served fewer patients.

Research workers observed the general process of providing HIV and diabetes care in these clinics. They also shadowed clinic staff as they delivered care and patients as they moved through the clinic and received their HIV or diabetes treatment. Before conducting any of these observations, research workers approached patients as they waited to see clinical staff, requested permission to describe the study’s goals and observational activities, and asked if they were willing to be followed as they received care. The research workers had no pre-existing relationship with the clinic staff or patients prior to study commencement. Those that were willing to participate were asked to provide written informed consent. Several patients chose not to have their care observed; we did not record the number of patients who refused consent or their reasons for not providing consent. These patients who declined to have their care observed were thanked for their time.

A semi-structured observation activity form guided the collection of observational data (see supplementary file 2). Three researchers developed an initial version of this form. They chose observation categories that were aligned with Weiner’s (2009) conceptualisation of implementation capability as the product of the task-demands associated with implementing a new service, the availability of resources necessary for implementation, and situational factors within the organisation that may support implementation (Attieh et al., 2013; Shaw et al., 2013; Weiner, 2009). The researchers obtained inputs from stakeholders in the WCDOH on these categories. These stakeholders also assisted in identifying potential sources for these observations. The form was then piloted at one facility by the research team. During the pilot, we noted that the research workers’ observations were superficially described. Based on these findings, we adapted the form to elicit richer descriptions and added prompts to encourage research workers to reflect upon their responses to the interactions they had observed. Research workers used this form to systematically record their impressions of the physical environment, the organisational climate, clinic staff interactions with patients, health service organisation, and the resources available at each facility. They also made detailed field notes about their observations of interactions between patients, clinic staff, and facility management. To supplement these observations, they collected documents detailing the structure, resourcing, available staff mix, and organisation of the facility (such as patient headcounts and flow charts, organograms, and facility maps), where these were available. Facilities were assigned unique identifying numbers which were used instead of facility names (or to replace names in documents) so the researchers responsible for data analysis remained blind to the identity of the facility.

2.3. Ethical considerations

The South African Medical Research Council (EC 004-02/2015), the University of Cape Town (089/2015), and University of Oxford (OxTREC 567-15) in the United Kingdom approved the study. The Western Cape Department of Health approved all procedures and granted permission to access the health facilities (WC 2015_RP 28-480). Facility managers provided consent to the facility hosting the research. Written informed consent was obtained from clinic staff and patients after they had received full information about the project and had an opportunity to ask questions. When research workers observed violations of the National Patient Rights’ Charter (2008) and National Core Standards for Health Care Establishments in South Africa (2011), they reported these incidents to the facility manager as per the National Complaints Management Protocol for the Public Health Sector of South Africa (South African National Department of Health, Office on Standards Compliance, 2013). Researchers observed three adverse incidents involving patients and staff and they reported each of these incidents using the national protocol.

2.4. Analysis

Data analysis comprised three steps. First, a six-member multidisciplinary project team met to review the information collected during the facility meetings. This team excluded staff who had conducted facility observations. Although this process could potentially have been enriched by the impressions of these staff, we chose to keep the discussion and interpretation of the data provided in the facility meetings separate from the data collection process. This was to ensure that the six-member project team remained blind to the identity and location of facilities and to ensure that the discussions were not influenced by the research workers’ personal views. This project team used all the information provided during the facility staff discussions to classify the selected facilities as having low, low-moderate, moderate-high, or high implementation capability (see Table 1 for facility classification). We defined facilities with low implementation capability as unable to implement the programme; those with low-moderate capability as needing significant external support and intervention in the form of training, resourcing and monitoring; moderate-high capability as requiring moderate support; and those with high implementation capability as needing minimal support to implement the programme.

Second, the framework approach (Pope, Ziebland, & Mays, 2000) was used to analyse the observational activity forms, field notes and supporting documents to identify facility characteristics potentially associated with implementation capability. NVivo 10.0 software (https://www.qsrinternational.com/nvivo) aided the analysis. The researchers reviewed these data, identified emerging themes, and developed a coding frame (see Table 2). They posited that the resources available in the facility would be associated with preparedness for implementation. Next, two qualitative coders (who were members of the project team) independently coded the observational activity forms, documents and field notes gathered during the observational activities. These coders were blinded to the identity of the facilities. A third person was not needed to break coding ties. Although all the data were analysed, no new codes emerged after two-thirds of the observation forms and field notes were coded, suggesting thematic saturation. While analyzing the data, the researchers remained aware of their assumptions about available resources for mental health care within public health facilities and the climate within these facilities. These were discussed and reflected upon when the researchers met to compare how they had coded the materials. After the first two steps of data analysis, researchers (BM and PPW) conducted facility-wide meetings where each facility was provided with individualised feedback on the findings. Clinic staff and managers participating in the meeting were given an opportunity to comment on the findings.

Table 2. Conditions associated with implementation capability in observational data analysis and coding scheme for scoring conditions and outcome at each facility.

Theme	Condition	fsQCA coding scheme
Basic requirements for mental health counselling are present.	Person-centred care (defined as clinic staff treating patients with dignity, warmth and respect).	0- Lack of person-centred care. Most clinic staff treat patients in a disrespectful manner without warmth or empathy. 1- Facility offers person-centred care. Patients are generally treated with respect, warmth and empathy. A team approach is present. Staff team make additional efforts to ensure patients’ needs are met.
	Therapeutic environment (defined as organisational climate characterised by warmth, safety and where patient needs are prioritised)	0- Therapeutic environment largely absent. 0.3- Therapeutic environment somewhat absent. 0.5- Unclear 0.7- Therapeutic environment somewhat present. 1- Therapeutic environment largely present.
	Team approach (defined as clinic staff working together to meet patients’ needs) to delivering chronic disease care	0- Absence of a team approach. Clinic staff work in silos and do not support each other. 0.3- Team approach often absent. Staff rarely support each other in delivering services. 0.5- Unclear 0.7- Team approach is somewhat present. Staff sometimes support each other in delivering services. 1- Team approach is present. Staff clearly support each other and communicate about delivering services.
Availability of resources needed for mental health counselling	Infrastructure resources: enough space for confidential services	0- Severe, infrastructure challenges. Confidentiality of health care encounters is compromised. 1- Infrastructure is largely sufficient. Confidentiality of health care encounters is ensured.
	Human resources to support mental health counselling delivery	0- Visible staff shortages and frozen/unfilled posts in the facility relative to patient count. Staff fulfilling multiple service roles. 1- Limited staff shortages and posts relative to patient count in the facility. Facility appears adequately resourced in terms of human resources.
	Information resources (signage and systems) to aid patient flow and triage	0- Information resources such as clear patient triage system and signage are absent. 1- Information resources to aid patient flow are largely available.
Contextual influences on service delivery	Involved and strong facility management	0- Management is absent or appears uninvolved/disengaged or disinterested in facility operations. 0.3- Management seems somewhat uninvolved, disengaged or disinterested in facility operations. 0.5- Unclear 0.7- Management seems somewhat involved, engaged and interested in facility operations. 1- Management is highly involved, engaged and interested in facility operations.
	Facility organisation	0- Facility appears chaotic with no apparent system for organising services. 0.3- Facility appears somewhat chaotic with services poorly organised. 0.5- Unclear 0.7- Facility is mainly well organised with fairly well-structured services. 1- Facility appears highly organised.
	Outcome	Coding scheme
	Implementation capability	0 - Facilities unable to implement the programme with major changes 0.3-Facilities needing significant support and intervention to implement the programme 0.7- Facilities requiring moderate support to implement the programme 1- Facilities needing minimal or no support to implement the programme.

Third, we used qualitative comparative analysis (QCA) to compare data from these diverse facilities. QCA is a method of comparing data from several case studies that allows one to identify multiple factors, and configurations of these factors, potentially associated with an outcome (Kane et al., 2014). We used fuzzy set QCA (fsQCA), a specific type of QCA, to explore how different combinations of these facility characteristics (termed “conditions”) were associated with perceived capability for implementing mental health counselling across the various PHC sites. In contrast to crisp set QCA which treats the outcome as a dichotomous variable that is either in (1) or out (0) of the set, fuzzy set QCA treats the outcome as a continuous variable with values ranging between 0 (poor) and 1 (good), thereby allowing for a granulated understanding of perceived implementation capability (Backhans, Mosedale, Bruce, Whitehead, & Burstrom, 2016; Kane et al., 2014). To prepare the data for analysis, we calibrated it by assigning each facility (case) a value between 0 (completely out of the set) and 1 (completely in the set) for every condition and the outcome. Findings from step one and two guided the assignment of these values (see Table 2). Two authors (BM, KS) independently conducted the value assignment, which they then discussed and compared to ensure that the calibrated data reflected the cases studied (see supplementary file 3 for calibrated data). There were instances where the observed characteristics of the facility did not correspond with the facility’s perceived capability for implementing mental health counselling (e.g. facilities 5 and 25 in Supporting file 1). fsQCA understands these outliers as possible indicators that the models are under-fitted.

We used fsQCA software version 2.5 (https://fsqca.software.informer.com/2.5) to generate conditions or groups of conditions that must be present but are not sufficient on their own for an outcome to occur, known as “necessary conditions” (Backhans et al., 2016). Conditions are necessary if they have a consistency score greater than 0.9, indicating that at least 90% of cases with this condition produced the outcome (Garson, 2016). We also identified conditions that on their own or together produced the outcome but were not always present when the outcome was present, termed “sufficient conditions” (Rihoux & Ragin, 2009). Boolean minimisation was used to remove redundant conditions and produce solutions that reflect pathways or combinations of conditions associated with either poor or good implementation capability. We assessed the goodness of fit for each solution by assessing consistency and coverage (the extent to which the condition explains the outcome). We used a consistency cut-off of 0.8 to identify sufficient conditions (Garson, 2016).

3. Results

Of the twenty-six clinics, seven had low, three had low-moderate, five had moderate-high, and eleven had high implementation capability. Research workers conducted 286 hours of observation of facilities, patients (n= 78) and clinic staff (n= 156). We identified several factors that could possibly account for this variability, which we grouped into three themes: availability of basic requirements for mental health counselling; availability of resources for mental health counselling; and contextual influences on service delivery. These themes are described below and illustrated with extracts from the observation activity forms and field notes.

3.1. Availability of basic requirements for mental health counselling

Facilities varied in the extent to which they were observed to (i) provide person-centred services (where patients are treated with dignity, respect and warmth), (ii) offer a therapeutic environment with a calm, safe climate conducive to mental health care, and (iii) utilise a team approach to service delivery where clinic staff worked collaboratively to meet patients’ needs. In facilities with low implementation capability, person-centred care was observed to be largely lacking. Research workers witnessed three incidents where clinic staff verbally abused, belittled or mistreated patients at these facilities.

The patient said something that irritated the nurse and the nurse threw insults and used vulgar language … she also shouted at another patient. Another sister [nurse] she works with also spoke harshly to the patient in my presence. [Facility 11]

In contrast, staff at high implementation capability facilities seemed to treat their patients with respect and showed a genuine interest in their patients’ well-being. For example, research workers recorded instances where clinic staff delivered medication to patients’ homes if they were unable to come to the facility. They also observed how staff at these facilities seemed to understand their patients’ personal circumstances.

We also observed that the availability of a therapeutic environment varied among facilities. At facilities with negative climates, staff did not appear to prioritise patients’ needs. One research worker noted that staff from an urban facility “do not seem bothered by the business of the facility- instead they are having conversations in the corridors about their weekends.” At other facilities, research workers observed that clinic staff seemed “bored” and “to be working for the sake of earning an income.” Safety and security concerns may have contributed to the climate of the facility. Six of the urban facilities located in communities riddled with gang violence, had a very strong security presence. These facilities had low or low-moderate implementation capability. The climate in these facilities reflected the palpable tension present in the communities. Despite this, most of the staff at these facilities thought that their patients would benefit from mental health counselling as “their patients just need to talk.”

In cases where the researchers noted a poor therapeutic environment, patients’ trust in providers appeared to be negatively affected. In one busy, rural facility where observers did not detect a therapeutic environment, patients asked the observing team whether they had documented how they had to wait for services. In this facility, the research workers witnessed openly conflictual relationships between patients and staff. The presence of a therapeutic environment also seemed to impact whether the providers implemented a “team approach” to health service delivery where staff members support each other in their shared purpose of caring for patients. In facilities with high implementation capability, we observed that all facility personnel played a role in ensuring patients received the care they needed and worked together to resolve service delivery challenges. For example, in one large rural facility with chronic staff shortages, observation notes documented how staff fulfilled multiple roles in a facility and “come together to help each other make sure it is a functioning facility.” Facilities with low implementation capability seemed to lack this “team approach.”

The chronic disease nurse shared with me that there is no team work in this facility … the CHWs also expressed unhappiness that they work like slaves. [Facility 4]

3.2. Availability of resources for mental health counselling

We observed a relationship between information, infrastructure and human resource availability, and facilities’ capability for implementing counselling. In facilities with low implementation capability, patient-oriented information resources about where and how to access health services seemed either outdated, inaccurate or lacking. At these facilities, patients seemed unclear about how to access care. Research workers observed that in an urban facility located in a predominantly Xhosa-speaking community:

There is a lack of information about where patients must go. All the health materials, posters and signage are available in English only. [Facility 15]

Infrastructure availability (particularly private consultation rooms for counselling) also appeared related to facilities’ implementation capability. Many of the facilities with low implementation capability seemed to lack this infrastructure, often due to very high patient caseloads that placed the available space under strain. To illustrate, at one urban facility with low implementation capability and very high patient headcounts, the research team observed how the “nurse does foot exams of the patients in the waiting area as consultation rooms are full.” At a rural facility, a research worker documented how “all patients can also hear others’ consultation and the information that gets conveyed during this conversation.”

Third, human resource factors seemed associated with implementation capability. Although we collected information on staff qualifications and mix, absolute shortages of staff rather than absence of suitably qualified staff were common-place in facilities with low implementation capability. These facilities seemed to lack sufficient clinic staff to treat the number of patients seeking care. Staff at such facilities worked under extreme pressure and we believe were unlikely to support the introduction of additional services. In one rural facility with chronic staff shortages, a research worker documented how in her informal interactions with clinic staff, they expressed concerns about their ability to implement the project MIND counselling programme due to

the severe constraints they were working under. The facility is quite under-staffed, and this was visible. [Facility 2]

3.3. Contextual influences on service delivery

Two contextual factors, namely the strength of facility management and degree of facility organisation, seemed to influence implementation capability. Many of the facilities with high implementation capability also had strong management teams and were well organised. We observed that managers who were actively involved in finding solutions to the daily challenges associated with service delivery, often through the re-allocation of space and other scarce resources, were able to cushion the impact of resource shortages on health service delivery. We also observed that person-centred care was more likely to occur at facilities where managers appeared invested in and committed to supporting their staff and meeting the needs of their patient population. In such facilities, managers seemed to set the standard for how staff interacted with each other and their patients. For example, the observation notes from one large urban facility with high implementation capability show that “management is engaged and supportive and this seems to infiltrate throughout the facility and how staff interact with patients.”

In contrast, we observed that many facilities with low implementation capability had management teams that were “disengaged and absent.” This seemed to impact on whether there was a team approach to care. In one urban clinic, researchers documented that

Management is disengaged. There is a bad atmosphere at the clinic, staff are not happy to be at work, poor morale, and management is missing. [Facility 9]

In a smaller urban clinic, research workers noted that the facility manager did not know the names of her staff. The research worker described staff as “bored and uninterested.” She also described an interaction with patients who had noticed her making field notes and took the opportunity to “express their discontent with the quality of service and the way things are communicated with them.” The strength of facility leadership also seemed linked to the organisation of services. Facilities where managers were less involved were more poorly organised. One research worker described such a facility as a “street market where anyone can come here and hide and no-one would notice.”

3.4. Components of good capability for implementing mental health counselling

QCA identified three conditions that may be necessary, but not sufficient, for good implementation capability: the availability of person-centred care (consistency 0.96, coverage 0.84), the provision of a therapeutic environment (consistency 0.94, coverage: 0.79), and having sufficient human resources in relation to patient caseload (consistency 0.91, coverage 0.76; Table 3). In addition, three different combinations of components emerged, each representing a combination of conditions that together seem sufficient for good implementation capability. In one combination of conditions, implementation capability seemed present at facilities when infrastructural resources and patient-related information resources were available. In the other combinations, implementation capability seemed present at facilities when infrastructural resources were available, a team approach to patient care was adopted and either strong facility management was present or the facility was well organised. The consistency of this solution was high (0.88) and the solution coverage was moderate (0.70; Table 3).

Table 3. Necessary and sufficient conditions for good and poor capability for implementing mental health counselling.

Conditions		Good implementation capability						Poor implementation capability
		Necessary causes			Sufficient conditions			Sufficient conditions
					1	2	3	1		2
Context	Facility organisation					+
	Management						+
	Person-centred care	+						_		_
Counselling requirements	Therapeutic environment		+
	Team approach					+	+			_
	Infrastructure				+	+	+	_
Resources	Human resources			+						_
	Information resources				+			_
Condition	Consistency	0.96	0.94	0.91	0.60	0.93	0.97	0.91		1
	Coverage	0.84	0.79	0.76	0.08	0.60	0.63	0.58		0.49
Solution	Consistency					0.88			0.93
	Coverage					0.70			0.70

+ condition is present in pathway; — condition is absent in the pathway

3.5. Components of poor capability for implementing mental health counselling

QCA found no necessary conditions for poor mental health implementation capability at PHC facilities. Two different sets of components seemed sufficient for poor implementation capability. The first suggests that capability for implementing mental health counselling may be poor when there is an absence of person-centred care, space (infrastructure) and patient-oriented information resources. The second combination suggests that capability for implementing mental health counselling may be poor when there is an absence of person-centred care and a team approach to care, weak management, and insufficient human resources (Table 3). The consistency of this solution was high (0.93), suggesting that the combination of these conditions may lead to poor capability for implementation in 93% of cases. The solution coverage was moderate (0.70), indicating that 70% of cases with poor implementation capability appear to arise from this combination of conditions (Table 3).

4. Discussion

In the past decade, there has been a drive to scale up the task sharing of mental health counselling in primary care services within LMICs (Lund et al., 2012; Petersen et al., 2017). While these efforts have identified systemic and policy barriers to implementation (Hanlon et al., 2017; Petersen et al., 2017), less attention has been given to the preparedness of PHC facilities for counselling implementation. This study is among the first to use observational methods and QCA to explore variations in preparedness for counselling implementation among a diverse set of PHC facilities in South Africa and to test the feasibility of using this method for identifying factors potentially associated with these variations.

Using this approach to identify inter-facility differences in preparedness for implementing mental health counselling and factors potentially associated with this variation seems feasible in that we were able to identify facilities with high and low implementation capability. We found that a substantial proportion of PHC facilities seemed well-functioning and potentially prepared for counselling implementation. However, more than a third of the facilities in this sample had relatively low perceived capability for implementing counselling.

These facilities may benefit from preparedness-building interventions prior to the implementation of this new service (Scaccia et al., 2015).

We were also able to use this method to identify factors potentially associated with variations in preparedness. Findings suggest that PHC facilities offering person-centred health services and a therapeutic environment with sufficient human resources to serve their patient population may be more prepared for the implementation of mental health counselling. These findings are not surprising as patients need to feel cared for and safe to accept the offer of counselling (Myers et al., 2018a) and clinic staff must have sufficient capacity to meet the task-demands of providing a new service (Weiner, 2009). A quarter of PHC facilities appeared to lack person-centred health services with staff shortages seemingly contributing to the lack of person-centred care. This could be a result of compassion fatigue associated with elevated patient caseloads (Khamisa, Oldenburg, Peltzer, & Ilic, 2015). These findings suggest that current initiatives to train clinic staff in person-centred care strategies (Malan, Mash, & Evertt-Murphy, 2015) should be accompanied by efforts to reduce human resource shortages.

In addition, the availability of confidential space emerged as an important consideration. This is not surprising as confidential space is a pre-requisite for the delivery of mental health counselling and impacts on the uptake and delivery of counselling services (Myers, Carney, & Wechsberg, 2016; Myers et al., 2018a). Other sub-Saharan African countries address this infrastructural challenge through delivering counselling on the grounds of (instead of within) the facility (see for example Chibanda et al., 2016). The use of m-Health platforms to deliver counselling (Farrington, Aristidou, & Ruggeri, 2014) and provision of home-based treatment (Sukhato et al., 2017) may also be viable alternatives to facility-based counselling services. In South Africa, brief mental health counselling could be incorporated into the package of care delivered by nurse-led ward-based outreach teams (WBOTs) who are tasked with providing a range of community-based health promotion services and support to vulnerable individuals and families as part of South Africa’s PHC re-engineering strategy (Assegaai, Reagon, & Schneider, 2018). As CHWs form an important part of these WBOTs, they could be trained to provide community-based psychosocial interventions for people with common mental disorders. Although a recent systematic review of community-based psychosocial interventions in LMICs provided evidence of their feasibility, acceptability and effectiveness for reducing symptom severity and hospital readmissions among people with schizophrenia (Asher, Patel, & De Silva, 2017), the extent to which community-based mental health counselling is feasible to implement and acceptable to patients with common mental disorders must still be established for the South African context.

Finally, various aspects of adequately managed facilities (including the availability of patient information resources, a team approach to service delivery, well-organised services or a strong management team) also emerged as factors potentially associated with better preparedness for counselling implementation. These findings imply that strengthening the management, and consequently the organisation, of PHC facilities may potentially enhance their capability for implementing mental health counselling.

4.1. Limitations

This study has several limitations. First, despite efforts to minimize the subjectivity of the observations by ensuring there was more than one research worker conducting observations at each site, observing care at more than one point in time, and triangulating observations with documentary data, the observational data remains subjective. Confidence in the validity of these findings could have been enhanced through the addition of patient and clinic staff surveys. Qualitative interviews with clinic staff and patients also would have enriched the observational data. Second, we relied on clinic staff’s own views of their facility’s implementation capability. These views may not be accurate, and some staff may have found it hard to disagree with their colleagues in a group setting. Clinic staff may have felt more comfortable sharing their opinions in individual interviews. Similarly, although participation in the discussions and observations was voluntary, none of the clinic staff declined the offer of participation, suggesting that some may have felt pressurised to participate. Future studies may wish to consider alternative ways of obtaining consent from clinic staff that provides them with ample time to consider the request and multiple opportunities to decline the offer of participation or withdraw from the study. As such, there is no evidence to suggest that these views of implementation capability are correct and whether the factors that seem to support implementation preparedness are indeed helpful. Future research should consider using more objective measures to assess implementation capability. Since this study, we have embarked upon additional work to validate a measure of organisational readiness for change (Brooke-Sumner et al., 2018) that we plan to use to validate our ratings of implementation capability. Finally, we do not know the extent to which our sample of clinics was representative of the 189 clinics in the province. Future studies should consider selecting facilities purposively to facilitate generalisation of findings.

5. Conclusion

This study explored the feasibility of using observational methods to examine variations in the preparedness of PHC facilities to implement mental health counselling and features of these facilities that may be associated with preparedness. Through using this approach, we were able to provide rich descriptions of aspects of PHC facilities that may impact on preparedness for counselling implementation. If future research can demonstrate that findings from these observations predict the extent to which counselling services are implemented and the effectiveness of these services, then this may be a low-cost method to identify health facilities with greater readiness for implementing mental health counselling.

Accessible summary.

What is already known about this topic?

• ▯Integrating mental health counselling into primary care services is a recommended strategy for reducing the mental health treatment gap in low- and middle-income countries.
• ▯To support this strategy, potential barriers to counselling integration must be identified and addressed.
• ▯Organisational preparedness for implementation may influence the extent to which the introduction of counselling is successful.
• ▯Features of primary care facilities associated with preparedness for the implementation of mental health counselling have not been explored.

What the paper adds to existing knowledge?

• ▯This study uses a novel approach to explore variations in preparedness of primary care services to implement counselling and factors potentially associated with these variations.
• ▯Findings suggest there is considerable variation in the preparedness of facilities to implement counselling.
• ▯Organisational factors such as resource availability, management style and facility environment are potentially associated with capability for implementing mental health counselling.

What are the implications for practice?

• ▯Health planners could use this method to identify poorly functioning services that may benefit from additional interventions to build preparedness for counselling implementation.
• ▯Future research should examine whether differences in facility preparedness impact on the implementation and outcomes of this service.

Relevance Statement.

Registered nurses play a potentially critical role in facilitating good mental health implementation capability in this context. Like other middle-income countries, primary care facilities in South Africa are primarily managed by nurses. Training, mentoring and supporting nurse-managers to identify and address potential facility-level barriers to mental health counselling provision may support the implementation of this service. These nurse-managers could also draw in nurses with an additional specialist qualification in psychiatry and mental health to help train and support lay counsellors, also referred to as community health workers, in the delivery of mental health counselling and to create environments conducive to the delivery of therapeutic services.