What do we think we are doing? How might a clinical information network be promoting implementation of recommended paediatric care practices in Kenyan hospitals?

A full explanation of how the proposed strategy for intervention might tackle identified challenges is provided elsewhere [11]. The broad strategy was based on considerable prior experience of the context, the clinical focus of care and our understanding of theory that resonated with challenges diagnosed before initiating the current intervention approach. In brief, constraints operating to undermine the provision of care in accordance with evidence-based clinical practice guidelines were felt to include failures of expert professionals (paediatricians) to take a more holistic responsibility for paediatric service delivery beyond their personal technical contribution to individuals’ care. With typically only one or two of these personnel in a hospital and with a form of training that emphasizes expert knowledge, this failure resulted, in part, from their being inadequately prepared and poorly supported to act as leaders of service units (sometimes referred to as hybrid clinical managers [15]). This was compounded by their often being professionally and geographically isolated. Further exacerbating these system deficiencies was an almost complete lack of useful information for routinely monitoring or understanding care processes or outcomes within hospitals [16]. We therefore proposed bringing together efforts to disseminate national policies, the national professional association and implementers (a research group) to develop a collaborative network to tackle these constraints. As part of this, we aimed to provide paediatricians with simple skills to act as effective hybrid managers when engaging with senior managers and frontline staff to improve services. In summary, our proposed network would act as a system oriented intervention targeting the pivotal role of leaders of service units [15], while also addressing the wider professional context (or norms) likely to be an important influence on practice [17]. The network was also to generate improved information on the process of care being provided and share this through episodic feedback. This feedback is linked to discussions with senior clinicians and is intended to promote adherence to national clinical guidelines, thus attempting to make use of likely positive benefits of feedback and outreach [18, 19]. Having articulated what we intended, the actual evolution of network activities is now described.

In Table 1, we summarise the activities that have been conducted across the network between its inception, in September 2013, to April 2016. At the centre of the network was a senior clinician coordinator who, in addition to the events outlined in Table 1, made three further trips to each hospital within the first 15 months to provide face-to-face feedback and discuss hospital feedback reports. This process also acted as a form of coaching to local paediatricians expected to deliver the feedback to local teams themselves as the CIN continued. A data coordinator who worked with the one data entry clerk at each hospital also visited each site every 3 months to reinforce good data collection practices. Every second visit (6 monthly) this data coordinator conducted a verification of the data being entered by independently entering data from a random set of records and contrasting entries with those of the local clerk providing feedback and reinforcing training as required. (A full description of the data collection process is provided elsewhere [21, 22]). The clinical coordinator and the data coordinator also maintained phone and email contact (approximately every 2 weeks) with the paediatricians and data clerks, respectively. Between January 2014 and March 2016 quality of care reports spanning a summary of admission diagnoses, outcomes and reporting on the quality of documentation in the medical record have been sent to hospitals 10 times (approximately every 3 months). These reports also provide information on whether the care provided (documented) adheres to key aspects of national guidelines for malaria, pneumonia, diarrhoea and dehydration, severe malnutrition, and meningitis.

We briefly outlined a rationale for the network we intended to deliver as an intervention strategy in the introduction. In the first part of these results, we presented a brief description of the activities conducted as part of the network intervention. Here, we examine the intervention efforts to identify the specific implementation strategies encompassed by these activities [12]. To do this, we tabulate these strategies, organised within conceptually related groups as defined by Waltz et al. [20], and indicate our rationale for considering these to be active ingredients of the network intervention being delivered (Table 2). This process suggests that we are employing 22/73 defined implementation strategies [12]. Amongst these we would consider 12 to be major active components and 10 to be minor.

Amongst the implementation strategies as defined by Powell et al. [12] and conceptually grouped by Waltz et al. [20], we did not feel the CIN actively employed strategies in the domains of ‘Adapt and Tailor to the Context’, ‘Engage Consumers’ or ‘Utilize Financial Strategies’. In the case of Tailor Strategies (defined as tailoring to ‘address barriers and leverage facilitators that were identified through earlier data collection’) this decision was debated. However, our rationale was that, although the CIN was developed based on considerable prior work spanning over a decade [1, 5, 10, 23] that informed initial theorising on intervention design [11], we did not collect new data to support this theorising before initiating the intervention.

Within CIN, we felt that ‘Alter incentive structures’ was a major component implementation strategy. In the typology developed by Powell et al. [12] this is defined as “Work to incentivize the adoption and implementation of the clinical innovation”. However, its further description makes it clear that it refers to purely financial incentives and, in the classification by Waltz et al. [20], this strategy is conceptually grouped in the domain ‘Utilise Financial Strategies’. We believe this points to a potential gap in the existing typology that seems to exclude socially constructed incentives such as recognition and promoting professional values that may have important effects on behaviour [24]. We therefore retain ‘Alter incentive structures’ as a major component but do not classify it within the currently proposed domain structure (Table 2).

Given the broad nature of our network approach judgements had to be made about whether or not we were actively delivering specific components. This was of particular relevance to our identification of minor intervention components and those not felt to meet the criteria of being actively and sufficiently employed. For example, we felt the network had little role in ‘Change physical structure and equipment’ as any organisational changes were left to local hospitals and no direct financial support was provided. This decision was made although we did supply a single desktop computer for data entry and advocate with national health sector partners for provision of mid-upper arm circumference (MUAC) tapes to screen for malnutrition. However, the single desktop was used solely for data collection within the records department and we considered this to be part of the strategy ‘Develop and implement tools for quality monitoring’. We did not consider advocating for delivery of a job aide for a very specific task as a sufficient criterion of ‘Change physical structure and equipment’ in hospital paediatric units, although for this very particular task it may have supported change (as described below). Similarly, we did not feel our approach helped ‘Identify and prepare champions’ as the leaders we worked with were selected by virtue of their existing, formal position and we were engaging them in a process where they might reconsider the scope and responsibilities associated with their established role. Instead, we felt our efforts to support clinical leaders was best captured within a minor intervention component as ‘Recruit, designate, and train for leadership’. This component was considered minor as we had no role in recruitment or designation, we worked with clinical leaders already found in participating hospitals, but we did offer some training in a role as clinical leaders (Table 2). At a higher level, although the CIN intervention is being undertaken in partnership with the national paediatric association, it did not have the authority to ‘Change accreditation or membership requirements’. Instead, our efforts to change professionals’ behaviour we feel is being mediated through two minor intervention strategies, the creation of a learning collaborative and use of social influences to revise professional roles (Table 2).

Above we have articulated what implementation components are included as part of our CIN intervention. However, such a process of deconstruction may not help us understand how these components work together to effect change. Change that is essentially the consequence of individual and group behaviours. Here, we use the Behaviour Change Wheel framework to explore how these behaviours may be influenced [13]. In line with this framework, we consider policy level levers and forms of intervention and their effect on capability, opportunity and motivation. At the policy level the network promoted and helped distribute clinical guidelines with hospital paediatricians themselves engaged as members in at least one national guideline development panel in 2015. This exposed network partners to the process of evidence-based guideline development [25] and may have built trust in their value. The network also acted as a means of communicating and marketing these guidelines and feedback within the network focused on hospitals’ compliance with them, feedback that was shared across the peer group involved in the network.

The Behaviour Change Wheel describes a number of forms of intervention. We feel the network employs Incentivisation, Persuasion, Education, Enablement, Modelling and Environmental Restructuring. Perhaps the main form of incentivisation (as in Table 2) is the effort to use recognition of successful change and improvement efforts and link this to promoting professional pride in these successes. Persuasion is employed across the collaborative approach spanning engagement in guideline development, links to authoritative organisations, including the research institute, the university and the national paediatric association, and the use of processes that encourage development of shared goals and peer-to-peer benchmarking and interaction. The engagement of multiple institutions and peers across the network provided opportunities for education, particularly focused on developing understanding of new roles and areas requiring improvement (as distinct from formal skill oriented training). Modelling by providing examples for people to imitate and enablement such as that provided by the continued support of the clinical coordinator were also employed. No restructuring of the physical environment in participating hospitals was conducted by the network team, but we did change the medical record systems and furnish information where it did not exist. This information on adherence to guidelines is aimed at creating a professional and social expectation to improve. Linked to a sharing of this information across groups and peers we were also aiming to create new norms and to revise socially constructed professional roles.

Ultimately, these strategies are anticipated to promote motivation, enhance psychological capability and provide the social opportunity that will create change. The effect on motivation of the intervention described is mainly perhaps through the reflective pathway with participants engaged in developing plans and evaluating progress against goals that are shared with experts and peers (in common with more specific quality improvement strategies [26]). However, we believe automatic motivation may be triggered by a link to an innate desire to provide good care encompassed in the idea of vocation amongst health workers [27]. The engagement in creating guidelines, understanding indicators, and involvement in reflection on and construction of approaches to change practice might improve the psychological capability of both individuals and the network as a group. Meanwhile, an overarching theme of the network is to change the social milieu in which the clinical leaders in hospitals operate. A broad professional focus on adoption of guidelines and improvements in care that is endorsed by recognised institutions and the professional association, as well as an effort to create ownership of this agenda, are at the heart of the network strategy.

An early focus of the network was to improve the availability and quality of accessible data to characterise care quality for inpatient children. The challenges to be overcome and the approach taken have been described elsewhere [21]. In brief, medical records themselves were often poorly kept and medical documentation was limited in scope and quality. Routine health information systems focus only on aggregate workload and morbidity and mortality statistics and are not designed to provide any information on quality of care. Furthermore, weaknesses in the information system [16] and its predominant role as a means to send data to a national repository with little use of data locally have created a system of care that is rarely informed by data analysis. By fostering introduction of (but not providing) structured forms to standardise the content of inpatient records [28], collating and analysing the data they contain, and providing feedback on data quality, clinical documentation has considerably improved [22].

Early efforts to promote adoption of guideline recommendations included focusing attention on ensuring all admitted children had their HIV status ascertained. Knowing the HIV status is important in determining treatment of acute illnesses (e.g. pneumonia) and has become increasingly important as now all children who are positive, irrespective of immune status, should be started immediately on antiretroviral medication. Documenting HIV status was discussed at network meetings and feedback on hospitals’ performance was provided in regular reports (Tables 1 and 2) with paediatricians encouraged to overcome local barriers to making testing more available and to promote documentation of status by their clinical team members to avoid missed or multiple testing. This has seen a steady rise in clearly documented HIV status over time (Fig. 1a). In 2013, new national guidelines recommended that sick children be screened for malnutrition using measurement of MUAC. It was clear from the data, discussions at network meetings and with the clinical coordinator that this new policy was not being enacted. The initial problem being one of supply of simple MUAC tapes designed for the purpose. Network partners began to advocate with the national Ministry of Health and its partners to make these tapes available. Once this happened, feedback on use of this screening tool, linked to ongoing discussion at network meetings and through the coordinator, was used to promote its uptake (Fig. 1b). These examples we hope help illustrate some of our decisions on what we consider active implementation strategies encompassing ‘Change record systems’, ‘Develop and organize quality monitoring systems’, ‘Develop and implement tools for quality monitoring’, ‘Audit and provide feedback’, ‘Create a learning collaborative’, ‘Build a coalition’, ‘Revise professional roles’, and ‘Mandate change’ (Table 2).