We retrieved 4932 articles, of which 1225 were duplicates. After screening records by title and abstract, 165 studies were retained. Papers that focused on Chlorhexidine (CHX) use (n=41) were excluded at this stage on the basis that a large number of existing systematic reviews and meta-analyses have been published specifically on CHX [27-30]. At full text review, a further 85 manuscripts were excluded, as they did not meet the inclusion criteria and four full text versions could not be sourced.
Searches repeated in February 2018, to capture new publications, retrieved another 480 papers, resulting in 26 additional manuscripts included in synthesis. An additional four papers were identified as a result of mining references of relevant manuscripts.
Data were extracted from a total of 110 manuscripts. During synthesis, an additional 32 were excluded based on relevance. A total of 78 papers were included for analysis. See Figure 1 for details.
Study focus and design
Among included studies, 60% (47/78) were observational studies  while 40% (31/78) described an evaluation of intervention activities (referred to from now on as intervention studies). Among the 31 studies that assessed an intervention, approximately half (16/31) included a control or comparison group. Among all studies, 55% were conducted in Asia (43/78), 40% in Africa (31/78) and 4% (3/78) in the Central America and Caribbean. Half of the studies (51%; 40/78) had a household or community focus compared to 28% (22/78) focusing at the health care facility level and 21% of studies (16/78) focused across both environments. The 31 intervention studies were distributed across health facilities (12/31), domestic environments (11/31), or both (8/31).
Across all studies, we identified 31 determinants of clean birthing practices, either explicitly articulated in the published manuscript or extrapolated based on intervention/study design. Determinants identified in the literature are mapped against their associated COM-B category in Additional File 1. We identified behavioural determinants related to general clean birthing practices in 64% of reviewed studies (50/78) and determinants related to one or more of five specific clean behaviours in 50% (39/78): hand hygiene (of the attendant or the caregiver), clean delivery surface, clean cord-cutting instrument, clean cord tie, clean cord care. Determinants were not reported regarding cleaning of the perineum. Of those studies where it was possible to identify determinants for individual clean practices, hand hygiene (13/39) and clean cord care (11/39) were most commonly reported (Table 2).
Determinants identified in observational studies:
The frequency with which determinants were identified across the 47 observational studies reviewed is recorded in Table 3. A more detailed version of this table that includes study references is in Additional File 2. In sum, 49% of the observational studies reviewed (23/47) identify determinants associated with general clean birthing practices. A total of 15 determinants for this set of generalised clean birthing practices were identified across these studies. Six specific determinants related to capability were identified, with knowledge (13/23) and skills (8/23) receiving by far the most attention across studies. Five studies identified inadequate training of birth attendants, reported to result in limited knowledge and skills of clean birthing practices [32-36]. Two studies (one focusing on home births and the other on facility based births) present positive associations between the adoption of clean birthing practices and mothers with higher educational levels , current employment or having 2-3 children . While this is not explicitly stated, it is possible that these determinants are being used as proxies for maternal knowledge or socio-economic status of the mother, all inextricably bound. One study described how the availability of national newborn care guidelines were an important predictor of good newborn care practices by health care workers, including appropriate cord tying and cord care .
A total of six specific determinants related to opportunity were identified in 14 observational studies (14/47). Physical opportunity, more specifically access to adequate materials such as soap or clean delivery kits (7/14) and water, sanitation and/or hygiene infrastructure (4/14), received the greatest attention. One study identified time as a barrier to improved practices . The importance of traditional/cultural beliefs – a component of social opportunity – was described in five studies. One study in south-eastern Nigeria reports that knowledge of and familiarity with cultural practices, as well as the ability to offer comprehensive care, made traditional birth attendants (TBAs) a popular option for care during delivery in their local community . In this study TBAs had good knowledge of the importance of personal hygiene, including hand washing, during the antenatal period. Another study in Karamoja, Uganda, reports that birth preparedness is restricted by local tradition, which forbids women to buy delivery supplies before the baby is delivered .
Determinants associated with motivation receive relatively little mention in the 23 observational studies reporting on general clean birthing practices, with only four studies providing information relative to motivational determinants. In Cambodia, skilled birth attendants viewed attempts at cleanliness as pointless, due to factors such as the unclean clothing of labouring women . In rural Tanzania, women reported the costs of materials for clean birthing practices in a home delivery – razor blade, thread and gloves – as a barrier to adoption of these practices .
57% of observational studies reviewed (27/47) identify determinants for a specific clean practice, and, of these, the majority focus on hand hygiene (10/27) or clean cord care (10/27), followed by clean cutting or tying of the umbilical cord (7/27) and clean birthing surface (5/27). Among the 10 hand hygiene studies, seven focus on hand hygiene of the birth attendant and three focus on hand hygiene of other caregivers. Studies on birth attendant hand hygiene focus mostly on health care facility staff (5/7) with only two studies providing information specific to hand hygiene among birth attendants in domestic environments. For attendant hand hygiene, identified determinants most commonly relate to physical opportunity, with six studies providing information on four specific physical opportunity determinants, including supply of materials (5/7), convenience (2/7), infrastructure (3/7) and time (1/7). Psychological capability – namely knowledge (4/7) and guidelines (1/7) – was the second most common category of determinant for attendant hand hygiene. Knowledge was the most commonly cited determinant of caregiver hand hygiene (3/3). There were only a limited number of studies that addressed caregiver opportunity or motivation for improved hand hygiene. One study in Indonesia  described the motivational drivers of handwashing behaviour among new mothers, namely disgust, for example after changing a soiled nappy, as well as nurture or the desire to care for their newborn. In a study of new mothers in India, 15% of their interviewees said they had no time for proper handwashing .
Only five observational studies included information related to the determinants of clean birthing surfaces (5/47). A total of four determinants were identified across studies, most commonly related to social opportunity. Specifically, these studies identified the common belief that childbirth – and by extension the mother and neonate – is polluted and/or impure and that it is not worth using clean materials to protect them. Reflective motivation was also a determinant of clean birthing surfaces, specifically considerations about which surfaces are easier to clean and/or dispose of following childbirth. One study in a domestic setting in Bangladesh reports that women delivered on the floor, on a jute bag, or on straw as this made cleaning and disposing of impure blood and placenta easier . Khadduri et al describe how the practice of placing a plastic sheet under the mother in Pakistan was rarely done with clean plastic and was often intended to protect the surface not the mother or child .
Seven observational studies (7/47) identified a total of four determinants for clean cord cutting and/or tying. Determinants of these practices were often presented combined. Psychological capability (in particular insufficient knowledge) is the most commonly reported barrier (5/7). This relates to both mothers, who are responsible for preparing thread and blade prior to delivery, and TBAs, who tie the cord and cut it. One study specified that while the importance of handwashing and germs was widely understood with regards to cord care, the importance of a clean blade was not common knowledge . Another study found a significant association between the use of antenatal care (ANC) and the use of a clean blade and tie, suggesting that exposure to ANC was associated with knowledge of the optimal behaviours . Social opportunity (in particular cultural or traditional beliefs) was described as a barrier to clean cord cutting/tying in three studies. For example, one study reports that TBAs tie the cord in the belief that it prevents air from entering the baby .
10 observational studies (10/47) found a total of eight determinants of clean cord care. The determinants most commonly referred to were traditional or cultural beliefs (5/10) and community influencers (4/10), both of which fall under the social opportunity category. For example, the cultural belief that the cord is harmful to the newborn as a channel for witchcraft or evil spirits is reported in two studies in Tanzania and Haiti as driving the mothers to use sometimes harmful substances on the cord to speed up the drying process [42, 49]. Nurture is often the implied deeper determinant of such culturally informed practices, as it is believed that the longer the cord is attached the more vulnerable the child is to either health or spiritual risks. This is perhaps clearest in a paper reporting that women apply different cord care practices according to the sex of their new offspring. If it is a boy, mothers are more likely to practice cord-care-related behaviours that, while posing a health risk, respond to their heightened desire to nurture . The influence of senior community figures, including grandmothers and traditional healers and birth attendants, is highlighted in four studies [48-51]. One observational study in Nigeria notes how this influence, especially influence by attending nurses, mother or mother-in-law, can supersede a woman’s own knowledge of the risks or benefits associated with a given practice .
Determinants identified in intervention studies
Only 35% (11/31) of the behaviour change intervention studies identified in this review are explicit about which behavioural determinant(s) their intervention targeted. For the remaining studies (20/31), information on targeted behavioural determinants was inferred based on intervention descriptions.
Details of the 20 interventions where the target determinants are not explicitly identified are recorded in Additional File 3, including intervention aim, activities, expected outputs, and implied determinants. Two studies provided information on the determinants of use of maternal health services in Kenya and India. In these studies, clean birthing practices are measured as outcomes associated with accessing maternal health services; however, the mechanisms through which this occurs are not articulated [52, 53].
Among the remaining studies, the majority (15/20) involve educational messaging to the mother or TBA about maternal and newborn clean care behaviours. Of interventions that provided education, 40% (6/15) also distributed essential materials, most commonly a clean delivery kit, to enable desired behaviours. This intervention design implies that the barriers to adoption of clean birthing practices are insufficient knowledge and inadequate access to necessary materials. One intervention distributed materials (including soap and clean delivery kits) at ANC visits, and these products were found to successfully incentivise mothers to obtain ANC, suggesting the importance of motivational determinants . Four studies (4/15) described interventions that included a community engagement component in addition to the standard educational messaging to mother or TBA. These included efforts to engage local leaders and/or community elders through community health committees or existing community structures [53-56]. The inference here is that the interventions were specifically targeting determinants related to social opportunity.
Five studies (5/20) described multi-modal interventions within a health care facility, either for general quality improvement [57-59] or specifically to improve adherence to the WHO Safe Childbirth Checklist [60, 61]. The interventions typically included training of health care staff, engagement of senior facility and district personnel, continual monitoring, feedback and action cycles, and mentoring, coaching and supervision of staff. These components suggest the significance of knowledge and skills, job motivation and the fear of repercussions as main drivers of behaviour.
Of the eleven studies that are explicit about which determinant(s) the interventions targeted (11/31), seven report on target determinants of clean birthing practices in general and only four identify target determinants specific to a particular practice. Details of the interventions explicitly targeting identified determinants for clean practices in general are recorded in Additional File 4. There are three studies (3/7) of community-based interventions [62-64] and three studies (3/7) of healthcare facility-based interventions [65-67]. One study (1/7) evaluated a community and facility-based intervention . The most commonly targeted determinant was knowledge of the mother or birth attendant, with four interventions (4/7) aiming to improve knowledge through increased use of maternal and newborn services , home-based maternal counselling by community volunteers , improved quality of facility-based maternal counselling through use of visual job-aids , or facility-based training of care providers on essential newborn care practices . Two focused on participatory approaches, specifically the active involvement of women, families, and community members in effecting behaviour change [63, 64]. Two interventions engaged powerful community members in influencing maternal or attendant behaviours. Job motivation emerges as a minor theme, featuring in one healthcare facility-based study in India that included peer coaching and performance monitoring . Another minor theme, featuring in one community-based study, was utilizing a woman’s life-stage – namely pregnancy – as a teachable moment .
Details of the four interventions (4/31) explicitly targeting identified determinants for specific clean birthing practices are recorded in Additional File 5. The four interventions targeted clean hands of the attendant (2/4) [69, 70] and of the carer (1/4) , clean blade (1/4) , and cord care (1/4) . Knowledge is identified as a determining factor for attendant handwashing with soap in a healthcare facility setting in Nicaragua  and for using a clean blade to cut the umbilical cord in a community setting in India . In Nicaragua, the absence of alcohol gel was also highlighted as a barrier, leading to an intervention based on guideline development, training of medical staff and revision of the basic medical supply list. In India, the importance of social norms and collective behaviours, the influence of community influencers and decision-makers was also raised, leading to a community-based intervention delivered by community health workers with a multi-level strategy for engagement of individuals with key roles as influencers, decision makers, supporters, practitioners of newborn care and normative behaviour within community. This community approach also targets a further determinant identified in this study; the role of cultural beliefs in shaping hand hygiene behaviour among attendants. The delivery process and the newborn are reportedly unclean and ‘polluting’ so birth attendants usually do not see the value in handwashing. Cultural beliefs also emerge as an important barrier to clean cord care practices in a Maasai community in East Africa, which reports that the application of cow dung to the cord is culturally symbolic, underlining the close connection between the Maasai way of life and the tending of cattle . The importance of participatory approaches in effecting caregiver hand hygiene behaviour change, as well as the significance of the pregnancy period as a teachable moment, were identified in one facility-based study , which responded with an interactive educational intervention to promote handwashing, using behaviour change communicators trained on motivational interviewing to encourage active engagement from participants and their families.
Behavioural Determinants Summary
Table 4 illustrates the frequency with which COM-B categories feature across the entire body of literature included in this review. Generally, determinants cluster around psychological capability (most commonly knowledge) and physical opportunity (most commonly access to resources).
However, the picture is more nuanced when we disaggregate according to specific behaviours of interest. For example, among 13 studies that focus on determinants of hand hygiene the most investigated determinants are knowledge (n = 8, 62%), materials/supplies (n = 6, 46%) and infrastructure (n = 3, 23%), which is in line with studies that identify determinants for clean birthing practices in general. However, when the analysis is restricted to studies that identify determinants for clean cord care (n=11), the importance of social hierarchies/community influencers (n=4) and traditional or cultural beliefs (n=6) gain markedly in prominence relative to knowledge.