Unpacking the enabling factors for hand, cord and birth-surface hygiene in Zanzibar maternity units

Recent national surveys in The United Republic of Tanzania have revealed poor standards of hygiene at birth in facilities. As more women opt for institutional delivery, improving basic hygiene becomes an essential part of preventative strategies for reducing puerperal and newborn sepsis. Our collaborative research in Zanzibar provides an in-depth picture of the state of hygiene on maternity wards to inform action. Hygiene was assessed in 2014 across all 37 facilities with a maternity unit in Zanzibar. We used a mixed methods approach, including structured and semistructured interviews, and environmental microbiology. Data were analysed according to the WHO ‘cleans’ framework, focusing on the fundamental practices for prevention of newborn and maternal sepsis. For each ‘clean’ we explored the following enabling factors: Knowledge, infrastructure (including equipment), staffing levels and policies. Composite indices were constructed for the enabling factors of the ‘cleans’ from the quantitative data: Clean hands, cord cutting, and birth surface. Results from the qualitative tools were used to complement this information. Only 49% of facilities had the ‘infrastructural’ requirements to enable ‘clean hands’, with the availability of constant running water particularly lacking. Less than half (46%) of facilities met the ‘knowledge’ requirements for ensuring a ‘clean delivery surface’; six out of seven facilities had birthing surfaces that tested positive for multiple potential pathogens. Almost two thirds of facilities met the ‘infrastructure (equipment) requirement’ for ‘clean cord’; however, disposable cord clamps being frequently out of stock, often resulted in the use of non-sterile thread made of fabric. This mixed methods approach, and the analytical framework based on the WHO ‘cleans’ and the enabling factors, yielded practical information of direct relevance to action at local and ministerial levels. The same approach could be applied to collect and analyse data on infection prevention from maternity units in other contexts. Our first aim was to produce actionable information, meaning information that (1) is organized by the WHO ‘clean’ practices necessary to reduce maternal and newborn infection acquired at the time of delivery; (2) clearly identifies the behavioural factors from the WHO IPC guidelines that enable these clean practices and that can be addressed through MoH interventions; and (3) allows the root causes of the IPC gaps to be identified, using a mixed methods approach. We investigated four out of the six ‘cleans’: clean hands, clean cord (clamping and cutting), and a clean birth surface. The clean perineum of the mother at birth was excluded because of the weak evidence base for this clean (Blencowe et al. 2011) and the postpartum skincare of the newborn was excluded because we were focused on intrapartum care for data collection The WHO IPC guidelines for facilities identified eight core components.(WHO 2016) We collected data in Zanzibar that allowed us to investigate four of these components that we refer to as behavioural factors in relation each of the four cleans we chose to investigate. These enabling factors and their definition in this paper are: Three tool sets were used during the assessment: (1) a structured facility questionnaire, administered to the maternity in-charge or equivalent at the time of the interview in all facilities providing delivery services (n = 37), (2) a ‘walkthrough’ tool set (described below) and (3) semi-structured interviews conducted in a purposively selected sample of facilities in Zanzibar (n = 7). The seven facilities were selected by the Zanzibar MoH to represent the variation in facility type, volume of deliveries, location and levels of service quality. The tools described below were based on the WASH & CLEAN toolkit, adapted with the collaboration of key MoH stakeholders and administered in Swahili. The toolkit, previously used in India, Bangladesh and the Gambia, was developed by the Soapbox Collaborative from existing tools from international organizations to assess IPC on maternity units and is publically available online (Cross et al. 2016). The facility questionnaire was initially piloted in five facilities, and the walkthrough tools and the semi-structured interviews were piloted in four. The tools were administered between 19 May and 10 September 2014. We conducted 26 semi-structured interviews with healthcare staff including in-charges (7), care providers in the maternity (7), orderlies (7) and maintenance staff (5) present in the facility at the time of the visit. One member per cadre per facility was invited to be interviewed. Staff selection was based on who was available at the time. The facility questionnaire and the semi-structured interviews focused on guidelines, training and infrastructure for IPC, WASH and solid waste management; barriers to maintaining good practice; and the actions needed to overcome them. Qualitative interviews were also conducted with 20 women attending vaccination services for their newborns at the seven facilities, who had delivered within the past 8 weeks. The team aimed to interview a minimum of two women at each facility visited; one who delivered at the facility under assessment and one who delivered at home but who was living around the facility catchment area. The first woman presenting in the relevant facilities during the assessment period who consented to participate in the study was interviewed. These interviews sought to capture women’s perception of an appropriate delivery environment, and their experiences during their most recent childbirth, particularly in relation to hygiene at the delivery unit. Interviews were conducted in Swahili and were tape recorded. Two types of data were collected with the walkthrough tool set: (1) observations recorded in the walkthrough checklist, noting the availability and conditions of specific areas and equipment (e.g. labour ward room, toilets and cleaning equipment); and (2) microbiological samples taken using swabs of high-risk hand touch sites such as bedside lockers, delivery beds, cleaning equipment, and of water used for hand washing in the maternity unit. See Supplementary Material S1 for more details on the water sampling and microbiological swabs. For each ‘clean’ we built a composite index, using the facility questionnaire data (n = 37), that aimed to be represent each of the four enabling factors investigated: ‘knowledge and training, infrastructure, staffing levels’ and ‘policies’. The choice of index components was informed by published IPC international guidelines for each topic (EngenderHealth 2003, 2011; WHO 2015). This allowed us to standardise the analysis of the ‘cleans’’ enabling factors with relevant data from the facility questionnaire. Table 1 describes the information used to build these indices. For the ‘knowledge and training’ index, we used questions that explored the topics discussed during IPC training received in the past year and questions around maternal and newborn care practices. With regards to the latter, interviewees were asked about their care practices but discussion with our data collectors led us to believe that their answers reflect knowledge of expected practices rather than actual staff behaviour and thus are best considered a proxy for knowledge. We aimed to interview the maternity in-charge or equivalent in each facility; this information therefore represents their knowledge. For the ‘infrastructure’ index, we used questions on the availability of, and access to key infrastructure and equipment in the maternity unit. Indices’ components by ‘clean’ and for each enabling factor For the ‘policies’ determinant, we present data on whether policies or posters of key protocols i.e. IPC, hand hygiene and decontamination of areas soiled by blood and other body fluids were available in the maternity unit. For ‘human resources’, at least one skilled SBA should be present in the maternity during the morning and night shifts; this ensures that someone formally trained in IPC is available on site capable of cleaning their hands adequately at appropriate times and capable of performing clean cord care. Since it was unusual in Zanzibar, especially in small facilities, that orderlies were allocated to night shifts, for clean birth surface the variable we referred to was whether an orderly was present on the previous morning shift. The indices were all binary, with facilities either meeting all the conditions prescribed by the index or not. Similar composite indices have been used previously to describe key markers of the quality of maternal healthcare facilities (Nesbitt et al. 2013; Campbell et al. 2016). The key assumption was that the components chosen to construct the indices were fundamental for performing the ‘cleans’. The variety of tools used produced quantitative, qualitative and microbiological data. Results from all three tool sets were organised thematically using the frameworks discussed: the WHO cleans and the enabling factors. The water analysis, using conventional pour plate and membrane filtration techniques, focused on the total bacterial count in the water samples, as well as looking at the presence of Enterococcus and fecal coliforms—standard indicators for assessing water quality (Ashbolt et al. 2001). Swabs collected from surfaces were directly inoculated onto selective media and screened using standard biochemical techniques to identify and characterize potential pathogens. The analysis of the microbiology swab data focused first on whether Staphylococcus aureus (S. aureus), one of the most common pathogens linked to healthcare associated infections (Allegranzi et al. 2011), was present at the touch site. Opportunistic pathogens such as S. aureus are frequently shed by patients and staff in healthcare environments and can persist on surfaces for months on dry surfaces, posing a significant transmission risk to new patients admitted to the facility—thus, we used this as an indicator for cleanliness (Kramer et al. 2006). The second indicator examined was whether multiple pathogenic organisms were identified on the touch site. Two or more such pathogens found on a hand touch site indicate a lack of effective cleaning or long durations between cleans. For more details see Supplementary Material S1. We began our analysis of the qualitative materials with word-for-word transcriptions of the audio files in their original language. Transcripts were later translated into English and analysed manually using a qualitative ‘content analysis’ method to extract manifest and latent content from the interviews (Vaismoradi et al. 2013). We used an inductive process for analysis whereby all codes and themes were derived from data. No software was used, a research assistant coded the data manually and the senior qualitative researcher reviewed the codes to check their quality (all codes are available on request). Using facility questionnaire responses, indices representing each of the four enabling factors were constructed for each ‘clean’ and described by facility type. In our dataset, we distinguished between three types of facilities: with an operating theatre or without, and those which the MoH had not deemed appropriate to perform deliveries because they lacked key equipment and infrastructure. Since facility questionnaire data came from all facilities providing maternity services in Zanzibar, no survey weights were applied. The walkthrough checklist data produced counts of the infrastructure and equipment available, cleaned, and according to state of repair. Data were double entered into EpiData v3.1 and analysed using STATA v13 SE. We obtained ethical approval from the Zanzibar Medical Research and Ethics Committee and the Observational/Interventions Research Ethics Committee at the London School of Hygiene and Tropical Medicine for this study. The women interviewed gave their individual consent, while the MoH granted permission to interview healthcare staff, and collect and analyse microbiology samples in the facilities. Women who gave birth recently—respondents were informed about the purpose of the survey before the start of the interview, informed that their participation was voluntary, and that all information provided was confidential and would be de-identified. The respondent’s consent, if obtained, was in written form. Facility data—prior to commencing the facilities questionnaire, an official letter was sent by the MoH to all facilities to inform them of the study aims and that the information collected might be used by the MoH or other organizations seeking to improve the planning and delivery of health services, and that the identity of the facility would be anonymized. For each of the seven facilities selected for the semi-structured interviews and the walkthrough this information was also provided in person by the enumerator to the facility in-charge, the maternity in-charge and the orderlies in-charge.