Barriers and enablers to delivery of the Healthy Kids Check: an analysis informed by the Theoretical Domains Framework and COM-B model

Focus group discussions with GPs and PNs were undertaken to explore their knowledge-behaviour gaps. Group dynamics were viewed as more likely than individual interviews to reveal attitudes and experiences, and the underlying reasons for specific behaviours [27].

Three groups of GPs and three groups of PNs (total 40 practitioners) were recruited from three socio-culturally diverse urban areas of Melbourne, Victoria, broadly categorised as high income (Bayside), low income (Westgate), and culturally diverse (Dandenong). The study was advertised by newsletter and invitations were faxed to clinics served by Medicare Locals in these areas. To increase responses, phone calls were made to individual practice managers and PNs by one of the researchers (KA), and participants could recommend other practitioners (snowballing), with a limit of one GP and one PN from each clinic.

An interview guide based on the TDF was designed to prompt focus group discussions (Table 3). To avoid ‘group hierarchies’, focus groups were divided by practitioner (except for one attendee, a practice nurse, who opted to attend the GP group) (Table 4). Focus groups took place between June 2011 and October 2011 (three years following introduction of the HKC), lasted approximately 90 minutes and were facilitated by the first two authors (who declared their positions), one a GP, trained in qualitative research methods, the other, an experienced qualitative researcher. A voucher valued at $200 for GPs and $80 for PNs was given to each practitioner in appreciation of their time. This incentive discrepancy reflects differences in average earnings between practitioner groups and known difficulties with recruiting practitioners to research projects [28]. Focus groups were audio-recorded and later transcribed for analysis. A report was emailed to each participant to solicit feedback. Ethics approval was obtained from Monash University, and all participants provided written informed consent.

Data were analysed by applying categories from the TDF in a recursive process that followed the customary steps of thematic analysis [29]. Specifically, after reading through the entire dataset, the first two authors independently coded the data from each transcript and assigned initial ‘code names,’ then collaborated and discussed choices, with a third researcher available to resolve any differences in opinion. Data were imported into NVivo 8 [30] and de-identified. After agreement had been reached, an additional step was taken to match code names to themes represented by the ‘domains’ within the TDF. This required the researchers to re-read data within the codes, then allocate the codes to the appropriate domains. This sometimes meant that the data coded under one code name was categorised into two or three different domains within the TDF. All codes could be applied to at least one domain. From there the domains were mapped to the COM-B system (Table 5).

There were two domains of the TDF that we did not match any data to: the ‘Emotional/Automatic’ aspects of ‘Motivation’ and the ‘Nature of behaviours’. With regard to the former domain, although specific questions had been asked about emotions felt by practitioners when dealing with young children and health screening, responses were captured under the theme ‘GP attitudes and feelings’. This was assigned to the domain, ‘Professional role and identity’, and ultimately mapped to ‘Motivation’ in the COM-B model. The domain ‘Nature of behaviours’, part of the original list of domains within the Framework, could not be assigned to the COM-B model because whilst it described context (current practice), it did not provide a source of behaviour. This domain was subsequently removed in a review of the Framework which tested its validity with a second group of behavioural change experts [31].

There were several limitations to this study. The TDF was originally designed to be accessible and useful to an interdisciplinary audience to understand behaviours around evidence-based guidelines. The researchers had a combined wealth of experience in general practice, preventive care, and qualitative research methods but did not have access to the skills of a behavioural psychologist. Had we had such access, further insights may have been generated, but in this way we have adhered to the original intent of the TDF. Additionally, the fact that preventive healthcare for children, including HKCs, is based on low levels of evidence could have increased the variation in behaviours, so that some discrepancies may have been missed. The 40 practitioners who took part in the focus groups were likely to be more motivated towards prevention or paediatric health, and less motivated practitioners may have additional deterrents to providing preventive healthcare to young children. This study was, however, purposefully aimed at practitioners working in socio-economically diverse metropolitan suburbs and captured a broad range of behaviours around the provision of HKCs. Focus groups run the risk of introducing bias resulting from an individual’s desire to conform to social acceptability, and their perceptions were not actuality. Further studies, using a mix of quantitative and alternative qualitative methods, could be done to address this, and could obtain the views of practitioners from rural areas and other states where variations in health structures and service delivery may produce different results [48]. The fact that common and significant barriers were detected in this engaged group, however, implies that larger gains are likely to be made where the starting base is low. In addition, the participation rate for the focus groups was adequate, and responses were generated in an iterative process that proceeded across each of the study areas with no new data relevant to the topic of interest generated in the last of the six groups, suggesting that saturation had been obtained. Moreover, feedback, solicited from participants, did not amend the study’s findings.

Despite these limitations, there were considerable strengths in this study. This was the first study to apply the TDF to understand preventive healthcare in young children and therefore adds to the body of work that constitutes knowledge translation research. Moreover, the use of the COM-B model as an additional step in the analysis increased the study’s efficiency and proved that the framework was adequate for purpose. An alternative method would have been to analyse the data within the domains of the TDF as a single step. Previous research has used ‘relevance criteria’ to determine which domains could be targeted by potential interventions [49]. In this study, an unwieldy 11 of 12 domains would have had to be considered, making subjective decisions necessary and potentially causing important evidence to be disregarded.

^aChild health surveillance includes measuring growth and promoting healthy weight, developmental assessments including vision, hearing and social and emotional health, assessments of oral health, injury prevention, and other health promotion activities [50].

^bThe original TDF was reviewed, modified and published in 2012 [31], but because data was collected using the original framework (in 2011), analysis was made according to this framework.