More than a lack of resources: a mixed-methods assessment of the barriers and facilitators to infection prevention practices in three low

Discussion In this study we used mixed methods of data collection integrated to form a comprehensive picture of adherence to recommended IP practice, exploring barriers and facilitators to recommended practice in three low-resource maternity facilities in Malawi. Significant gaps were identified, including poor hand hygiene, suboptimal aseptic technique and inadequate surgical scrubbing and sterility. Whilst antibiotic prophylaxis and skin preparation were routinely implemented, vaginal cleansing with antiseptic had not been adopted at any of the facilities, though resources were available for its implementation. We found that erratic supply of resources and poor infrastructure were both directly and indirectly linked to IP practice. HCWs demonstrated knowledge of recommended IP practice (a facilitator), but non-adherence was normalised, undermining social opportunity for best practice. Lack of resources hindered automatic and reflective motivation; habit formation and HCW beliefs about consequences, optimism and their personal and collective capability to prevent infection were negatively impacted. Further barriers to recommended IP practices included a mindset of ‘kutayilira’, described as an attitude of doing things without care and attention (‘like riding a bike without holding the handlebars’). The reasons for this attitude of ‘kutayilira’ among HCWs were difficult to pin down, and it is possible that reasons for this attitude overlap with other themes identified from in-depth interviews. For example, does ‘kutayilira’ develop because staff are burnt out, demoralised and disempowered due to the context in which they work? Conversely, we found that ‘passion’ for patients acts as a facilitator to good IP practice, mediated through automatic motivation (emotion). Senior supervision and ‘reminders’ also facilitated good IP practice through improving psychological capability (memory, attention and decision processes, and behavioural regulation), promoting social opportunity and providing automatic motivation (reinforcement). Two domains of the TDF (Cognitive and Interpersonal Skills and Physical Skills) were not represented in the mapped barriers and facilitators. In relation to physical skills, this likely reflects challenges observing physical proficiency in an environment where recommended practices were inconsistently performed, rather than a true lack of capability. Given the content of standard clinical training and because the required procedures were relatively simple, it is reasonable to assume that HCWs possessed the necessary physical skills to complete relevant IP practices. Cognitive and Interpersonal Skills may also have been under-represented because infection prevention practices were largely procedural rather than reliant on complex reasoning or interpersonal coordination. Their absence may partly reflect methodological limitations; unless cognitive or interpersonal processes were verbalised during interviews or visibly enacted during observation, they were unlikely to be captured. Interview participants may also have felt more able to discuss external constraints (resources, norms, workload) than their own internal reasoning or interactional dynamics. Further studies might better assess this by using simulation or asking participants to narrate their activities and thought processes. Overall, our findings suggest that the barriers to good IP practice in this setting were related more closely to opportunity and motivation of HCWs than capability. Strengths of this study include the participation of three maternity departments in data collection, lending validity to findings observed across facilities. Furthermore, the study was set in a district (second level) hospital and two community (first level) hospitals, rather than central teaching hospitals. Our findings are therefore relevant to the experience of ‘typical’ facilities in Malawi and other low-resource settings, where most healthcare is delivered outside of urban centres. The use of mixed methods of data collection enabled triangulation and cross-validation of data collected using different methods. A recognised, evidence-based model to explain and understand HCW practices was used (COM-B/TDF), strengthening and increasing the potential for generalisability of findings between this and similar studies, and directly informing the selection of intervention strategies used in the implementation study which followed. Limitations include a geographical focus around Lilongwe, and exclusion of private facilities (which provide 13% of maternity care).32 Another possible limitation of the study was the employment of HCWs from the facilities as data collection officers, introducing the possibility of bias. To mitigate, study team members regularly visited to cross-check data entry with patient notes and verify resource availability. Data collection regarding resource availability at Hospital A was incomplete, limiting findings from this site. An additional limitation was that observations were made only during daytime working hours between Monday and Friday. It is known that adherence to IP practice is often worse during evenings, night-times and weekend shifts. However, we felt that observations made during daytime practice gave an adequate starting point in appreciation of the context for IP in this setting. Out of hours practice was explored in qualitative interviews with staff to inform our findings. Furthermore, we accept that data collection using participant observation is vulnerable to ‘The Hawthorne effect’, whereby being observed affects the behaviour of participants.29 Given the poor adherence to IP practices we observed, we do not believe that this phenomenon had a significant impact on the data collection. Methodological limitations in assessing two of the domains of the TDF have been discussed above. The lack of resources for basic IP practice found in our study was not surprising; this is a recognised problem in many settings worldwide. In a global report, the WHO noted gaps in the resources and infrastructure for IPC across all regions.33 Previous work in Malawi has determined water, sanitation and hygiene (WASH) facilities in hospitals to be substandard,34 and the most recent assessment confirmed that inadequate and erratic availability of beds, electricity and WASH is a challenge nationally.32 However, this assessment found that 93% of facilities had soap available in maternity areas,32 contrasting with our finding that soap was only intermittently available, particularly at Hospitals B and C. The published literature on IP to prevent healthcare associated infection in low-resource settings primarily focuses on implementation programmes rather than quantifying resource fluctuations. Our findings are novel in describing how erratic availability influences daily practice, challenging the assumption that when supplies exist in a general context of scarcity, they are automatically used. Furthermore, the causes of stock-outs often remained unclear, even after extended observation. The existence of a ‘know-do’ gap—HCWs knowing but not performing IP practices—is well-recognised.35 Even when evidence of benefit is strong, HCWs in all settings may struggle to implement IP practice. There is a paucity of literature (particularly of a qualitative nature) from low-resource settings examining the implementation of IP recommendations. However, similar barriers to IP practice have been described in Uganda36; shortage of HCWs and finances, intermittent supply of water, IP materials and equipment, lack of support and supervision, negative attitudes and lack of training and orientation.36 Facilitators included management support, the presence of an infection prevention and control committee, visual reminders, supportive supervision, continuing medical education sessions, an organised environment, teamwork and cross-organisational collaboration.36 However, extensive exploration of these barriers and facilitators for closer comparison with our study was not available in the published manuscript. Previous studies from the Malawian context have noted similar findings to our study in terms of the ‘know-do’ gap and some of its influencing factors. Merriel et al20 similarly observed that HCWs knew correct procedures but often omitted them, partly due to poor supervision and frequent senior staff absences. The literature on the implementation of hand hygiene recommendations and barriers and facilitators to this practice is more extensive than that exploring other areas of IP. The practice of hand hygiene was noted to be a particular challenge during our study, with no member of staff observed to perform hand hygiene prior to patient contact, including when performing intimate examinations. The finding that HCWs struggle to implement hand hygiene is not novel and has long been a recognised challenge in IP across all global settings; a systematic review of studies from high-income countries found hand hygiene compliance rates of 60–70%.37 Though this intervention is perhaps of greatest impact in preventing hospital-acquired infection in low-resource settings. Tesfaye et al observed hand hygiene compliance as low as 2.8% in Ethiopian delivery wards.34 Contributing factors—limited water, soap and time—mirrored our findings in Malawi, as did lower compliance at night.34 The implications of our findings for clinical practice and healthcare policy include the need to improve the availability of basic resources for infection prevention and for closer supervision and mentorship of HCWs by senior staff at facility level. Furthermore, our findings regarding the context for implementation had critical implications for the development and success of a complex intervention to improve infection prevention at the time of caesarean section in the study setting and later in a multicountry randomised controlled trial.38 We would recommend that future researchers and implementers conduct in-depth mixed-methods evaluations to understand the implementation context before seeking to effect change