The study will be conducted in the GKMA which includes Kampala, Wakiso and Mukono districts. According to the Uganda Bureau of Statistics (UBOS), Kampala has a day time population of approximately 4 million people and a night time population of approximately 1.5 million . The rate of urbanisation in Kampala and the neighbouring districts of Wakiso and Mukono is so high, and these areas are experiencing many urban challenges including growth of slums, which are characterized by poverty, poor living conditions and limited access to various services [25, 26]. In partnership with the Ministry of Health (MOH), Ministry of Water and Environment (MWE), WaterAid Uganda and Kampala Capital City Authority (KCCA), we recently conducted a WASH assessment in 63 HCFs in the GKMA, which indicated significant gaps in WASH . This proposed study will be building upon the evidence from that assessment, as well as priorities and needs that have been identified by stakeholders. The intervention will be premised on the understanding that mobile phone usage is very high among HCWs. Indeed, according to the Uganda National Information Technology Agency (NITA), 70.9% of adult Ugandans own a mobile phone, and ownership is even higher in professional groups of people including HCWs . As of 2018, the number of Government and Private not for Profit (PNFP) HCFs in the 3 districts of the GKMA is as provided in Table 1 .
In Uganda, the healthcare system is organised into a four-tier system with hospitals and health centres (HCs) of levels IV, III and II . General hospitals (catchment population 500,000 people) provide preventive, promotive, curative, maternity, and inpatient health services and surgery, blood transfusion, laboratory, and medical imaging services. HC IVs have a target population of 100,000 people and are responsible for preventive, outpatient health services, maternity, inpatient health services, emergency surgery and blood transfusion, and laboratory services. HC IVs provide all the services of HC IIIs except emergency surgery. The study will be carried out in public HCs of level III and IV in Wakiso and Mukono districts, using a cluster-randomized trial (CRT) design. The study will be restricted to public HC IVs and IIIs because these offer affordable Maternal, New-born and Child Health (MNCH) services to the majority of the population in the GKMA , and only HCFs in Wakiso and Mukono will be studied since they are likely to have similar characteristics as they are more rural and receive less funding compared to those in Kampala. The intervention will be preceded by a formative study which will be conducted in the Kampala, Mukono and Wakiso districts. The reason for focusing on HCFs and departments that offer MNCH services is because mothers and neonates are at a greater risk of getting exposed to HCAIs [4,5,6], and there is therefore critical need to improve the status of WASH in maternal and children’s wards.
The study will be guided by the “Behaviour Centred Design (BCD)” model and theory for behaviour change , which guides hand hygiene program design through the “ABCDE” (Assess, Build, Create, Deliver and Evaluate) steps (Fig. 1). The model provides the BCD process on the outside and the theory of change on the inside. The theory suggests that, the right behavioral response depends on the behavior settings (physical, social and temporal context) in which individuals find themselves , and that three causal links must be made: from the environment (modified by an intervention) to psychological change in the target population (body and brain), to performance of the target behaviors (which results in changes to the state-of-the-world).
Prior to the implementation of the CRT, the investigators will obtain a list of HCFs and healthcare providers from the respective district health offices. This list will be used to randomly assign the clusters (HCFs) to the intervention and control groups. Once the clusters have been assigned, the study participants in the maternal and children’s wards within the selected HCFs will be informed about the purpose of the study and recruited in the study after the consenting process.
This study will be implemented following these stages;
Assess - document existing hand hygiene behaviour
In determining the barriers and opportunities for enhancing HCWs’ hand hygiene practices, this study component will involve collating and reviewing available published literature on hand hygiene provisions and drivers among HCWs in HCFs in Uganda and beyond. The literature review will be conducted with the help of a literature review guide. The findings from the literature review will guide the “build” (formative) research phase.
Build - formative phase
The formative research phase will address objectives 1 and 2 of the study, and the findings will inform the design and implementation of the intervention (mhealth and environmental cues on hand hygiene). The purpose of this formative phase is to understand the drivers of hand hygiene among HCWs. During this phase, we shall conduct key informants’ interviews (KIIs) and semi-structured interviews using developed guides. Multi-variation sampling will be conducted to select participants for the interviews and the sample sizes will be determined based on theoretical saturation in the data . Structured observation checklists will be used to establish the status of HCF hygiene infrastructure and hand hygiene behaviour of HCWs.
Create - stakeholders’ workshops to develop and refine the intervention
A stakeholders’ workshop shall be conducted to share findings from the literature review, formative research and discuss the contextual mechanisms of designing and implementing the mhealth and environmental cues intervention. Stakeholders will give their ideas on how best the mhealth intervention should be designed and used to disseminate messages to HCWs. The choice of the environmental cues (colours or pictures to paint around hand hygiene stations, branding of mirrors etc) and the mhealth (mobile texting approach or use of social media such as Facebook or Whatsapp etc) for communicating behavioural change messages will depend on the guidance of the WASH stakeholders, subject area experts and reviewed literature. The stakeholders will include MWE, MOH, Kampala Capital City Authority (KCCA), Non-Government Organisation (NGOs), representatives from local authorities, District and HCFs’ IPCcommittees, WASH consultants as well as HCWs from different levels of HCFs in the GKMA.
During the stakeholders’ workshop, the World Café approach [33, 34] will be used, and 7–8 participants will sit around tables to internalize the findings from the formative research. Within the different groups, participants will use findings from the formative research to develop themes which will be used to develop insights, appealing stories that link the theme to the behaviour (hand hygiene among HCWs). The insights will be formulated to include: the behaviour to change; the character to influence; the motives to encourage the behaviour of that character as well as the direct or indirect reinforcement of social norms to encourage behaviour change. The insights from the different groups will be presented, and participants will choose the best 4–5 insights based on richness (how many of the formative research findings are captured by the insight), power (how strong the logic is in linking the insight to the target behaviour), plasticity (how likely the idea on which the insight is based could be changed by the intervention), novelty (is it a surprise and acceptable to the target population). Using “how might we questions”, participants will identify the campaign activities, tools and touch points that will enable HCWs to practice hand hygiene, change the settings, and create social norms.
The workshop will be used to develop and refine the intervention for implementation in the HCFs, using the workshop guide.
Delivery of the intervention
The study will be implemented through a CRT targeting HCWs working in maternity and children’s wards. The intervention will be implemented over a period of 4 months, through a partnership with WaterAid Uganda, MOH, KCCA, MWE as well as selected HCFs. The design of the study was guided by the checklist for standard protocol items recommended for international trials (SPIRIT), which is provided in Additional file 1.
In both the intervention and control HCFs, simple innovatively designed hand hygiene facilities (tapped plastic jerricans with a stand and basin) will be provided (where needed) in delivery rooms, post-natal wards, and children’s wards. Soap and ABHR will be provided at all hand hygiene stations. Additionally, ABHR will be provided at points of patient care and on medication trolleys. Hand hygiene demonstrations using the “glo germ” gel will be done in both study arms at the beginning of the study. The choice of the departments/wards for the interventions is based on the understanding that mothers and neonates face the greatest risk of infections related to poor WASH and IPC measures in HCFs , but these wards will also allow the research team to maximize the observations of hand hygiene opportunities while standing within a radius of about 5–10 m .
In the intervention HCFs, besides the provisions in the control group, two interventions, mhealth and environmental cues (nudges) that have been successful in enhancing adherence to treatment guidelines and enhancing hand-hygiene in school going children [19, 22] respectively will be used. The choice of the specific mhealth and nudges intervention, and the frequency of exposure to mhealth messages among the study participants will be determined in the insight workshop. The mhealth intervention messages will be designed with a focus on: benefits of hand hygiene; when to do hand hygiene; how to do hand hygiene and how to protect others. However, the messages will also have educational jokes and response prompts where HCWs with the highest number of responses will win hand hygiene supplies. This will be critical in enhancing participant retention and complete follow-up. The educational messages and jokes in the mhealth intervention will be guided by the results from the formative study.
The mhealth messages will be sent to study participants using either short message service (sms) or whataspp, depending on the decision of the stakeholders in the stakeholders’ workshop. In case stakeholders choose using the sms option, messages will be sent to participants using RapidSMS, which is a free and open-source framework for dynamic data collection, logistics coordination and communication. The service will enable a two-way communication between HCWs using basic SMS mobile phone technology. RapidSMS will be installed on a secure Linux server running Ubuntu 18.04.5 LTS with the following hardware specifications: (1) Intel Core i7-8700T 2.40 GHz processor with 16.0 GB of RAM, and (2) An MTCBA-GF4 modem attached via a serial interface. RapidSMS dependences that include Python and Django web framework will also be installed on the same machine. For compatibility, we shall install Python 3.7.3 and Django 2.2. For communication between the modem and RapidSMS, we shall install Kannel, which is a free and open source SMS gateway compatible with RapidSMS. With Kannel’s SMS Delivery Report functionality, we shall be able to track the status of messages sent to HCWs. We shall configure the system to automatically send mhealth messages to HCWs at specific weekly intervals. Our server will enable HCWs to respond to the messages sent by the system. SMS received by the system from HCWs will be toll free thus HCWs will not incur any costs when responding to mhealth messages. Responses from HCWs will be securely stored in MySQL database on the server; this will enable further analytics on the data. Authorized access to the data will be available via secure login into the system. The data will be available and downloadable in CSV format to enable further processing by third party statistical tools. The software can track message history and delivery, and this will be critical in understanding the proportion of HCWs receiving and responding to sms messages on a daily basis.
Randomization and Trial arms
Using a list of level III and IV HCFs in the study districts , the project coordinator will use computer-generated random numbers to assign HCFs to the intervention and control arms. All HCWs from the same HCF will be allocated to the same group as the randomized HCF in which they are working. HCWs will be enrolled by research assistants (RAs) under the guidance of the principal investigator, and they will be masked off the intervention they are receiving.
The data analysts and the data collectors will be blinded to the group assignment. In order to achieve blinding for the data collection team, independent teams will collect data from either arm of the trial. In addition, data collectors will be trained separately.
Inclusion and exclusion criteria
The study will enrol participants working in the maternal and pediatric wards of selected HCFs, and Table 2 summarizes the inclusion and exclusion criteria.
For quality delivery of the intervention and ensuring that results are acceptable to the different stakeholders, an intervention support team will be in place to provide technical guidance and support. The intervention support team or data monitoring committee (DMC) will be comprised of 6 members, with a representative from the Environmental Health Department at the MOH, KCCA Public Health department, MWE, NGOs, HCFs and private sector. The independent DMC will be chaired by a representative from the Environmental Health division at the MOH.
The primary outcome will be the proportion of utilized hand hygiene opportunities (number of critical times hand hygiene is observed to be done with soap and water or ABHR) out of the total number of observed hand hygiene opportunities. The secondary outcome will be E. coli concentration levels in hand rinsates from HCWs, which indicates recent fecal contamination and hand hygiene failure.
We shall determine the two outcome variables using two approaches. Firstly, observations based on the WHO’s 5 moments for hand hygiene which include: before touching a patient; immediately before performing an aseptic procedure; immediately after an exposure risk to body fluids (and after glove removal); after touching a patient and his or her immediate surroundings when leaving as well as after touching any object or furniture in the patient’s immediate surroundings, when leaving - even without touching the patient . Hand hygiene observations will be conducted during the baseline, midline and end-line surveys. The hand hygiene observations will be done in the mornings (8 am – 12 pm) and evenings (4 pm – 8 pm) since these are considered to be peak treatment hours . Within the targeted time, 20 hand hygiene opportunities will be observed for each of the selected HCWs (corresponding to 10 patients each with 2 hand hygiene opportunities, which will include before touching a patient and after touching a patient), and the used and missed opportunities will be recorded. Therefore, a total of 9000 hand hygiene opportunities will be observed by experienced researchers (HCWs or environmental health officers), who will observe from the least obtrusive point within a radius of 5–10 m of the patient wards .
Secondly, hand rinsates from HCWs participating in the CRT will be collected immediately after the hand hygiene observations during baseline and end-line. The samples (hand rinsates) will be collected by a team of trained environmental health officers. During sampling of hand rinsates, HCWs will put their hands, one at a time in Whirl-Pak bags containing distilled water. The HCWs will wash their hands by rubbing the inner hands for about 2 min and the enumerator massaging their hands from outside the whirl pak bag to remove any potential pathogens. Aseptic techniques will be used to collect duplicate samples from each of the study participants (HCWs) during baseline, midline and end-line. The samples will be stored in sterile plastic bottles and transported on ice to the laboratory for further analysis within 4 h. The hand rinsates will be analysed for Escherichia coli (E.coli), to establish if there are differences in the levels of contamination on the hands of the HCWs before and after the intervention. During analysis, hand rinsates will be diluted using the ratios of 1:1, 1:10 and 1:100. The membrane filtration method (with Chromocult agar) will be used to concentrate samples, and incubation will be done at 37 °C for 24 h. Colonies of E.coli will be identified by their dark blue to violet colour, and they will be counted and recorded per 100 ml of sample.
Data collection methods, respondents and sample size
Key informants’ interviews (KIs)
A total of 30 KIs selected based on their knowledge, position and experience on WASH aspects in HCFs will be conducted. KIs will include supervisors/managers of HC IIIs and IVs, nurses, administrators, Environmental Health Officers as well as officials from the MOH and district health office. KIs interviews will be used to explore the barriers and facilitators of hand hygiene in HCFs, and this will be done using a KIs’ interview guide, which is provided in Additional file 2.
A total of 30 semi-structured interviews with HCF managers will be conducted, to assess barriers and facilitators to hand hygiene, motives of hand hygiene, social norms related to hand hygiene, behavioural settings and touch points. This will be done using a semi-structured interview guide provided in Additional file 3.
Structured observations in HCFs
As part of the formative study, observations using a structured observation checklist (Additional file 4) will be conducted in 100 HCFs to assess the hand hygiene infrastructure, roles, presence of hand hygiene supplies, functionality of facilitates, and norms in maternity and children’s wards.
Cluster randomized trial
A sample size of 30 clusters [HC IIIs and IVs] (15 HCFs under the intervention and 15 HCFs under the control arm) with 15 HCWs per HCF will allow us to detect a 20% increase in the proportion of HCWs that practice hand hygiene (hand washing/hand rub) at the 5 critical moments at 4 months between the control and the intervention groups. The calculation of the sample size is based on the following assumptions: a 25% and 5% increase in hand hygiene in intervention and control group respectively, a standard normal deviate of at 95% confidence level (1.96), a standard normal deviate at 80% power level (0.84), the proportion of HCWs practising hand hygiene in the control group (0.747) from an infection control study in HCFs in Arua district, Northern Uganda  and a design effect of 2.0 to account for the fact that there would be clustering of HCWs within HCFs. This gives a sample size of 426, however, in order to achieve a consistent cluster size, the sample size will be rounded off to 450 HCWs participants, with 225 in each of the study arms. The study will recruit 450 HCWs (doctors, clinical officers, nurses and midwives) who work in maternity and children’s wards. In each HCF, the study will target 13 nurses/midwives and 2 Doctors/Clinical Officers.
We shall conduct process and outcome evaluations. The process evaluation will be conducted at the midline to understand how the intervention has been implemented. The process evaluation will seek to establish: whether the intervention is being implemented according to plan; whether the intervention is working or not; how many HCWs will be getting exposed to the touch points; how many HCWs will be responding to the mhealth intervention as well as the aspects of the intervention that participants will have liked. The baseline, midline and end-line evaluations will be conducted using a tool provided as Additional file 5.
To assess the outcome of the intervention, a structured questionnaire will be administered at baseline, midline (2 months after starting intervention) and end-line (4 months) as indicated the participants’ timeline in Fig. 2. During surveys, social demographic, knowledge and attitudes questions will be collected from all participating HCWs through an interviewer administered questionnaire. The questionnaires will be administered to the HCWs before the hand hygiene observations. In order to maintain systematic consistency for measurement indicators, the same research tools used at baseline will be used for mid-term, end-line and impact evaluation. In addition, observations of hand hygiene practice among the HCWs will be done at baseline, midline and end-line using an observation tool. Hand rinsates will be collected and analysed, and the results will be entered in a hand rinsates form. All the data collection tools to be employed will be guided by the WHO’s 5 moments for hand hygiene .
Data management and analysis plan
Data will be collected electronically using tablets or phones preloaded with Open Data Kit software. Data capture forms with in-built restrictions plus logical checks to minimize errors and missing data at collection will be designed by the study data management team. The data collected from the field will be submitted to the cloud server on a daily basis, and only accessible by the data management team for data security purposes. The data will be downloaded in a comma-separated value (csv) format, and then exported to STATA version 14 after which consistency checks programs will be run daily and error reports produced. The generated errors will be sent back to the field team for clarification and cleaning. The data manager will be responsible for the security of the data and will back it up on the cloud server on a daily basis to avoid its loss.
Study progress will be monitored by comparing numbers accrued against those expected. Numbers accrued will further be verified against the physical counts from the field team. Progress reports will be provided to the study team on a daily basis for purposes of monitoring the progress. There will be a data quality team which will monitor the data collection activity and carry out re-interviews on a randomly selected percentage of the records to further ensure quality data is collected. Data will be treated as confidential by the study personnel and all records will be kept secure in locked filling cabinets and offices will be treated with utmost confidentiality.
Qualitative analysis (formative stage)
The KIIs will be transcribed verbatim by two experienced transcribers. The typed transcripts will then be read several times by all members of the study team who will then develop codes. Code book definitions will be based on the objectives of the study while integrating in emerging themes from the data. The code book will be discussed and agreed upon by the study team. Then two experienced research assistants (RAs) will code the articles using ALTAS-ti software to ease further analysis. The code reports will then be read and discussed by all the investigators who will agree on both codes and categories. Then, codes will be grouped into categories, themes and subthemes .
Quantitative analysis (effect of intervention)
The effect of the intervention will be evaluated at two time-points: midway the intervention, and immediately after the completion of the intervention period to assess the short-term effects on hand hygiene. Data from all surveys will be double entered into Open Data Kit (ODK) collect, and statistical analysis will be carried out using Stata version 14. We will calculate the prevalence of cluster level hand hygiene at baseline and follow-up for each HCF. Mixed sub-group analysis will be stratified by sex, cadre, experience, and level of training. We shall run multivariable logistic regression under the generalised estimating equations (GEE) framework to account for the dependence of hand hygiene on the intervention in a marginal way. This approach will give population average estimates for effect of intervention while accounting for other covariates in the model. An exchangeable correlation structure will be considered. Table 3 provides a summary of the study objectives, methods and data analysis procedures.
Quality assurance/ quality control of data and collection
Experienced RAs including observers will be recruited from a well-established network of RAs that have participated in successful research projects. These will be trained on research ethics, research design and how to minimize bias. Prior to data collection, there will be a pre-test of the data collection tools to ensure they are valid and reliable. This will also enable RAs to familiarise with the data collection tools and also correct any errors if discovered. Pretesting will be in Nyimbwa health centre IV, Luweero district. This HCF has been purposively selected because it shares similar characteristics with some of the HCFs in the GKMA. A quality control team will be instituted whose role will be to ensure that RAs adhere to the approved study procedures. To ensure quality in quantitative data entry, the data entry screen will be designed with skips and restrictions. In case of samples for hand rinsates, a duplicate sample of the hand rinsates will be obtained for validation of results.
Study findings will be presented at appropriate international conferences and submitted for peer-reviewed publication. Besides, study findings will be disseminated in stakeholders’ meetings and to the HCWs and managers in study HCFs.