Study setting and participants
The study was conducted in Kampong Chhnang Province, Cambodia. The study included eight purposively selected healthcare facilities (HCF), six primary health centres (PHC) and two referral hospitals, that were initially observed in the formative phase of the project (18, 19). Four HCFs were assigned to receive the intervention and the remaining four assigned to serve as comparison facilities. The six participating PHCs represented three different settings; rural/low facility-birth volume, rural/high facility-birth volume, and peri-urban/high facility-birth volume. One facility from each setting was assigned to receive the intervention. The referral hospital with the higher number of monthly deliveries was selected to receive the intervention. In the six PHCs, women were recruited until either five births per facility had been observed or the 14-day observation period ended. In the referral hospitals as many women as possible were recruited and observed over a period of 14 days. The sample size was considered sufficient for the exploratory nature of the study.
Any woman who presented to the HCF for delivery prior to entering the second stage of labour and was not already in excess pain and distress was eligible for recruitment. Patients considered by clinical staff to have complicated labour or delivery, or those under 18 and unaccompanied by a parent/guardian were excluded from the study. Written consent was obtained from the women, health care workers and all accompanying members who were present with the woman at any point during the observation period. We obtained verbal consent in addition to a witness signature in the case of a participant with low literacy. Participation was voluntary and the recruitment was done in a private area and the women were encouraged to have someone else with them during the recruitment process. To minimise reactivity, the explicit mention of handwashing was avoided and participants were informed that the aim of the study was to observe care giving practices during childbirth and postnatal care. The data collector discussed and agreed verbal or non-verbal cues with each participant that they could use to pause or terminate the observation at any time.
Intervention package
The design and development of the intervention was informed by earlier formative research conducted in the 8 HCFs (18, 19). The Behaviour Centred Design (BCD) theoretical framework was used to guide the collection and analysis of the formative research data (30). Following the eco – evo theory understanding of human behaviours, the BCD approach provides a systematic means of categorising behavioural determinants and translate these determinants into specific intervention components.
Potential interventions were tested and refined through a participatory creative process led by 17 Triggers, an in-country creative agency. Following a three-day co-creation workshop with key stakeholders in Phnom Penh, rapid field testing and prototyping was conducted with 15 mothers, 10 midwives, two facility directors and seven family members in two non-study HCFs and catchment communities over three weeks in November 2019. The final intervention design was based on the level of acceptability by users and HCF management, theoretical considerations and logistic and financial constraints. The final intervention was a multimodal intervention targeting midwives, mothers, fathers and non-parental caregivers that included physical environment restructuring, provision and improved access to hand hygiene infrastructure and materials, visual cues and reminders, social influence, and participatory training.
The intervention was delivered in two locations within the maternity ward; the labour and delivery (LD) room and PNC room. The intervention components in the LD room aimed to improve hand hygiene practices of birth attendants, primarily midwives, during birth and the intervention components in the PNC room aimed to improve the hand hygiene practices of primarily mothers and caregivers providing early newborn care in both the post-natal care ward and the household following discharge. Table 1 provides an overview of the intervention components and content.
Table 1
Overview of intervention components and content
Intervention location
|
Component
|
Content and purpose
|
Labour and Delivery room
|
Nudges and reminders
|
A set of bright coloured hand hygiene icon stickers strategically placed by midwives at key points in the room where hand hygiene opportunities most often occurred during birth.
Painting the labour and delivery hand washing station area and installing a similarly coloured soap dispenser to visibly link the reminder to the required practice
|
Social influence
|
A group commitment/pledge made by midwives to always practice proper hand hygiene and hold each other accountable
Hand hygiene badges for midwives to wear on their uniforms meant to encourage midwives to ask each other about their hand washing activities at key points during their shifts.
A token jar and counters placed in a visible area for midwives to tally how often they reminded each other per week
|
Participatory hygiene training
|
The training was conducted in the labour room and delivered digitally/remotely by a facilitator from the creative agency with experience in both participatory techniques and overseen by a technical lead from National Institute of Public Health – Cambodia (NIPH) and WaterAid. Midwives were trained using a combination participatory group discussions, creative exercises, scripts and role play. All training was based on Ministry of Health and WHO guidelines for technical validity.
Midwives training aimed to improve and refresh knowledge on adequate hygiene protocol specific to LD events (labour, delivery and immediate newborn care) including identifying recontamination events and the corresponding hand hygiene protocol.
With the exception of the painting, all labour and delivery room intervention components were delivered and set up by the midwives as part of the training.
|
Post-natal care ward
|
Nudges and reminders
|
Visual demarcation of the post-natal care ward as a “clean hands” zone by painting the ward door and wall section bright green
Placing hand hygiene icon stickers illustrating proper hand hygiene technique and nurture-evoking hygiene messaging posters across all post-natal care ward hand hygiene infrastructure and toilets
Provision of a brightly coloured hat to each baby with visible hygiene-related messaging
|
Provision of hand hygiene infrastructure and materials
|
Installation of a matching coloured hand washing station, liquid soap dispensers and soap at the ward entrance
Installation of matching coloured alcohol-based handrub station at the end of each bed
Provision of personal alcohol-based handrub bottle for each mother
|
Behaviour Change Communication
|
Newborn hygiene – related information would be passed on to mother along with the personal alcohol-based handrub bottle and baby hat during routine care advice in the labour and delivery room following birth
Newborn hygiene – related information would be passed on to the present visitors during the routine post-natal care advice given upon admission of the mother-newborn pair to the post-natal care ward. The ‘clean hands’ zone concept was to be explained to the visitors at this time or earlier during routine cervical checks when the admitted mother and caregivers were in PNC ward during stage 1 labour.
|
Participatory hygiene training
|
Similar to LD training component, the training was delivered digitally to the midwives via participatory group discussions, creative exercises, scripts and role play in the post-natal care ward. All training was based on National Ministry of Health and WHO guidelines for technical validity
Training included:
• Adequate hygiene protocol specific to newborn aftercare for all caregivers in the ward and at home
• Familiarisation with the post-natal care ward intervention components including deciding how and when to introduce the intervention components and pass on the behaviour change communication to the family members and mother
|
The intervention was delivered to selected facilities in July 2020 and ran for 24 weeks. Throughout the 6-month implementation period, the intervention HCFs were responsible for refilling the liquid soap and maintaining the handwashing sinks. The study project team was responsible for restocking and maintaining all other intervention components. Final data collection ran from November to December 2020.
Study design and evaluation
Outcomes
The study was designed as a non-randomised controlled before-and-after feasibility study. The behavioural outcomes of interest were hand hygiene practices of 1) Birth attendants during labour, delivery and immediate newborn care in the delivery room; 2) Health care workers and other caregivers in post-natal care facility ward during newborn care and 3) Caregivers in the home environment during newborn care. Previously reported observational data on hand hygiene (18, 19) served as the baseline measures for this evaluation.
Data collection procedures
Data (baseline) was collected in all eight participating facilities (February – July 2019) as part of the formative research study and were repeated at the end of the implementation period (endline). All study methods, data collection procedures, tools and baseline findings are detailed in earlier publications (18, 19). Structured assessments were conducted to assess the facility-level and household hygiene conditions. Hand hygiene practices were assessed via direct observation over three periods; childbirth in the delivery ward, post-natal period in the facility PNC ward and post-natal period in the home.
Labour and delivery observations began when the woman was admitted to the facility and the first vaginal examination occurred and terminated either after six hours or when the woman-newborn pair was discharged from the delivery room, whichever came first. Data collectors recorded key events including birth attendants’ handwashing and gloving practices and any observed contact of the birth attendants with the mother, newborn, objects and surfaces and other individuals during the observation period.
Observations during the post-natal period in the facility began when the mother-newborn pair was transferred to the PNC room of participating facilities and were terminated after a period of four continuous hours. Home observations were conducted within 72 hours following discharge from the HCF and began after the household structured assessment was completed and lasted one hour. Home observations were only completed for women who delivered in the six PHCs. During both the PNC period at the facility and the home, newborn care practices included diaper changes, cord care, breastfeeding and general newborn handling and were recorded along with any corresponding hand hygiene practices of any individuals providing newborn or maternal care.
All women were given a 15-minute break from observations every two hours, but any of the participants could use the agreed verbal or nonverbal cues to pause or terminate the observations at any time.
Data were collected on tablets used pre-coded observation tools by trained data collectors. Across all observations, data collectors positioned themselves in an unobtrusive location and recorded key events of all individuals present in each respective period. Five of the six study observers had prior experience with the study protocol and data collection methods having participated in the baseline data collection. All observers received a 3-day in house refresher training of the study protocol, ethical research practices and role play sessions. All observations tools and protocols were piloted in the field prior to data collection, however due to COVID-19 risk measures, the pilots were conducted in the study referral hospitals where the midwives observed four deliveries. Study observers were paired up during the pilot period to increase interrater reliability. Given the small sample size, rather than use statistical techniques to formally test reliability, observers worked independently to gather data and the observation data was compared at the end of each delivery. Any discrepancies between the pairs were discussed with the study manager for clarification and discussion.
Data analysis
All quantitative data was analysed using StataSE 15 (Stata Corp, College Station, TX, USA). All Qualitative notes recorded during the observations were reviewed and where applicable, recoded using STATA. Data from the home and facility level assessments were analysed descriptively and triangulated to provide context to the structured observations.
Analysis of Labour and delivery structured observations
We defined labour, delivery and newborn aftercare flows according to the analysis described in the methods by Nalule et. al (18). For each flow, we assigned time-specific hand hygiene categories (Table 2) to each birth attendant around hand hygiene opportunities prior to the initiation of the flow and within the flow when invalidation of aseptic technique occurred. Detailed definitions and descriptions of flows and aseptic procedures used for the analysis have been previously described (18).
Table 2
Hand hygiene categories used in analysis
Observation period
|
Hand Hygiene Category
|
Hand hygiene action
|
Labour and delivery
(all flows)
|
Adequate
|
Handwashing with soap and new gloves (multiple or single) worn at each hand hygiene opportunity
No potential recontamination of gloved and/or washed hands observed
|
Inadequate
|
Gloves (multiple or single) are changed without intermediate handwashing with soap
|
Aseptic technique invalidated
|
No hand hygiene actions taken at observed hand hygiene opportunity
|
Post-natal care in the facility
Household
|
Adequate
|
Handwashing with soap and/or use of alcohol-based handrub;
Handwashing with soap and glove use for any aseptic/clean procedures
|
Inadequate
|
Handwashing with water only;
Wearing gloves without intermediate handwashing with soap;
No hand hygiene actions taken at observed hand hygiene opportunity
|
Descriptive statistics were used to calculate the frequency and proportion of flows initiated and aseptic procedures within the flows that were conducted under each respective hand hygiene category. Results were provided by data collection rounds (endline, baseline) and study group assignment (intervention, comparison).
Multilevel logistic regression models accounting for clustering at the facility-level were used to calculate difference-in-difference (DID) estimates to assess the effect of the intervention between intervention and comparison facilities after adjusting for baseline observations. The outcome measures for this analysis were:
1). Adequate hand hygiene prior to the initiation of a flow or prior to any key event related to newborn care (coded as adequate hand hygiene category = 1; aseptic technique invalidated or inadequate hand hygiene categories = 0)
2). Invalidated hand hygiene prior to the initiation of a flow (coded as aseptic technique invalidated hand hygiene category = 1; inadequate or adequate hand hygiene categories = 0).
3). Adequate hand hygiene during aseptic procedures within the flow (coded as adequate hand hygiene category = 1; aseptic technique invalidated or inadequate hand hygiene categories = 0)
We ran models separately for each outcome measure. The first level of DID models assessed the differences in outcomes without adjustment for any potential confounding variables. The second set of DID models for outcome measures 1 and 2 adjusted for working shift time, facility type (referral hospital vs primary health facility) and professional qualification (midwife vs Doctor + Nurse vs Midwife intern) and for outcome measures 3 adjusted for facility type (referral hospital vs primary health facility) and professional qualification (midwife vs Doctor + Nurse vs Midwife intern). We present the effect estimates from the DID analyses as odds ratios.
Analysis of PNC and home observation structured observations
Only hand hygiene opportunities pertaining to newborn care were analysed for the PNC and home observations. Detailed descriptions of newborn-care related hand hygiene opportunities used for this analysis are previously described (19). Caregivers of the newborns were categorised into four groups; mothers, fathers, healthcare workers (midwives, nurses, doctors and interns) or non-parental caregivers (all other individuals observed providing care to the newborn). For each caregiver, hand hygiene actions associated with each hand hygiene opportunity were coded into two categories for the analysis; adequate and inadequate (Table 2).
Descriptive statistics were used to calculate the frequency and proportion of hand hygiene opportunities under each respective hand hygiene category by caregiver and treatment group at endline only. For the PNC in the facility observation period, we ran multilevel logistic regression models; unadjusted and adjusted for working shift time, facility type (referral hospital vs primary health facility) and professional qualification (midwife vs Doctor + Nurse vs Midwife intern). Due to low rates of observed hand hygiene measures at baseline (19), we based our estimate of effect on endline observations only. For home observations, the limited numbers of observations and zero baseline hand hygiene compliance (19) did not warrant significance testing.