The aim of this study was to generate IPC research priorities that could be used by policymakers, funders, and researchers to elucidate important IPC knowledge gaps. We constructed this list using an approach combining (i) a 4-step narrative literature review to identify knowledge gaps and (ii) a validation process with the help of two groups of IPC experts responding to a survey.
Survey results clearly demonstrated the need for research on IPC since several research gaps were scored as high priority areas in both target-groups. Results also support our attempt at building a list of important priorities with (i) no research gap bellow the medium priority (merged groups), (ii) mostly supportive and informative comments on research gaps, (iii) only few requests for additions. According to feedbacks, we only made one additional priority on the interaction between the human and hospital microbiomes (Table 1, ** mark). It was deemed a suitable addition since requested by both expert groups and by 9.1% of the whole expert population. Furthermore, several studies have shown the role of the microbiota in preventing acquisition or expansion of HCAI [15, 16], but with studies limited to murine models or clinical studies with small numbers of patients .
Through our analysis, three research gaps emerged as particularly important from both expert groups.
Assessment of demographic, organisational, socio-economic, and behavioural barriers/facilitators to implement effective IPC programmes
Over the past few years, socio-economic and behavioural sciences have greatly contributed to the fight against HCAI by identifying barriers and facilitators for the implementation of IPC measures. Commonly mentioned barriers include a lack of training/knowledge or awareness [18, 19], and a lack of institutional resources, especially in low- and middle-income countries (LMIC) [19, 20]. On the contrary, close relationships between healthcare workers , positive leadership and role modelling are often seen as facilitating factors [22, 23]. While behavioural determinants have been identified, only few interventions have been proposed and tested to address them. An impactful intervention could be, for instance, the appointment of an IPC champion in institutions to help engage and educate colleagues . Yet, limited quality of evidence failed to generate concrete recommendations. More research is therefore needed to assess innovative interventions and to test organisational frameworks that facilitate the implementation of IPC measures. Interestingly, this research gap is mentioned in the JPIAMR SRIA . The JPIAMR could therefore help funding research in the area.
The impact of overcrowding (staff workload/availability, bed occupancy, …) on the spread of HCAI
There is growing evidence on the impact of overcrowding on HCAI transmission rate. Low staffing and increased workload have been associated with a higher risk of HCAI acquisition [25, 26]. Regarding bed occupancy, the literature is still inconsistent [27,28,29]. These discrepancies could be explained by differences in study settings, monitoring outcomes but mostly by differences in methodologies and bed occupancy definition. More research is therefore needed but with appropriate occupancy parametrization . Overall, there is still insufficient data to generate clear and robust breakpoint thresholds needed by policymakers and hospital managers to implement effective actions (worker/patient and patient/bed ratio for instance). Ideally, breakpoints should be defined for various healthcare settings (intensive care units, short or long stay wards, long-term care facilities) and country settings (high, medium, and low-resource settings). More studies on the impact of visitor frequency and patient movements on HCAI transmission rates would also be beneficial to explore new interventions.
Assessment of the impact of infrastructural changes at facility level on the reduction of infections and resistance
There are little data available on the impact of infrastructural changes on HCAI. In 2016, a meta-analysis concluded that a high density of hand-washing points and single-patient rooms could help reducing HCAI transmission rates in short-term care facilities . However, these conclusions present some major limitations: (i) the small number of studies included in the meta-analysis, (ii) several studies were uncontrolled before and after intervention and (iii) several studies included in the meta-analysis were biased by bundle effects. More research in the area is therefore needed. However, as highlighted by experts, infrastructural changes are rarely considered as research opportunities. Ideally, IPC outcomes should be studied for any new healthcare facility or any facility remodelling. For instance, purchase of sinks, showers or bathtubs in healthcare institutions should include an analysis of evidence of how easily they can be disinfected. Placement and design of hand sanitisers should be based upon evidence on where healthcare personnel are most likely to use them.
Interestingly, not all IPC experts agreed that there is a need for additional IPC research, despite the dearth of high-quality evidence clearly displayed in many of IPC guidelines [12, 32]. The main thrust of this feedback is that these experts would rather have the funds to implement IPC than to research them. We are sympathetic to this argument, as it is a priority to implement effective IPC measures. However, these funding sources are not necessarily competitive, as implementation funds would normally come for the healthcare budget and research funds from the research budget as well as multinational funding sources like JPIAMR. We believe that effective implementation can run in parallel to ongoing research, as it does in other fields.
Other feedback included the difficulty of producing “high quality” IPC research, as determined through the GRADE methodology, which considers randomized control trials (RCTs) as gold standard. While RCTs, especially clustered randomized trials, are appropriate to evaluate some individual elements of IPC programmes (surveillance for instance), they are often limited when assessing IPC programmes containing multiple interventions or relying on qualitative measurements. For example, RCTs are not suited to evaluate organisational or behavioural interventions which rely on measurements such as governance, commitment or compliance. RCTs may also be inappropriate for IPC interventions due to sample size, ethical limitations or even feasibility. However, GRADE does allow for other types of studies to generate high quality of evidence, like cohort, case-control, before-after and time series studies . These study types would be more suited for the three urgent needs identified and could provide meaningful evidence. For instance, there are new developments in data and analytical technologies, offering the opportunity for observational studies to provide much stronger evidence. Propensity score matching, now, allows the assembly of two or more groups such that they appear to have been randomized to a comparator . Improvement in data collection and data linkage techniques also make observational studies easier to undertake . Powerful observational studies could help to provide evidence on barriers/facilitators for IPC measures implementation, thereby tailoring the design of innovative interventions. Implementation studies could then help testing these interventions. Contrary to local and time limited qualitative studies, implementation studies can match qualitative data with measures of success and process indicators over time, generating high quality evidence [36, 37]. In the end, all of these study types could strengthen meta-analyses and provide gold standard evidence related to IPC. In some instances, methodological work is also needed to define appropriate parametrisation or standards to undertake research. This is notably the case for research on the link between bed occupancy and HCAI where different parametrisations of bed occupancy have led to conflicting findings.
There are limitations to our study, specifically the small number of survey respondents (n = 44). However, this small number of respondents also reflects our strategy to target known European IPC experts/expert groups, which drastically reduces sample size. The average level of inter-agreement between both target-groups (Cohen’s κ = 0.21 and Kendall’s τ = 0.43) could also be interpreted as a lack of agreement on the most urgent needs. However, Cohen’s κ have been shown to be naturally lower when computing more than three categories (five in our survey) . Inter-agreement on sorting could therefore be underestimated by our statistical test. Regarding inter-agreement on ranking, while the computer Kendall’s τ remains average, we clearly have a strong expert alignment on the most urgent needs with top two research needs being the same in both groups. Also, we are aware that differences in the number of respondents (11 versus 33) may impact agreement between both target groups. Another limitation is that the experts we interviewed were only from high-income countries. Although we did include LMICs in our literature analysis and proposed research priorities targeted toward them, LMICs were excluded from the validation process. A further study, focusing on LMICs, should be conducted to validate all our findings in this setting. Lastly, we had only one broad research gap focused on IPC measures for agriculture, livestock and the environment, given our panel of human health experts. More focused and detailed IPC research priorities on animals and environment would be beneficial.
Despite its limitations, we believe that this study will inform policymakers and funding agencies regarding important IPC research priorities.