This is a mixed method study combining the results obtained through qualitative and quantitative studies. The study was carried out in a level IV Neonatal Intensive Care Unit (NICU) at a tertiary care teaching hospital located in coastal Karnataka, India from December 2016 to July 2020. To develop the prevention bundle, we needed to identify the factors contributing to neonatal HAIs ³³ ; we did root cause analysis, ³⁴ identifying the failure mode and effect analysis ^{35 , 36} followed by system process mapping. ³⁷ Once we identified the system and its process contributing to HAIs, we observed in other benchmark hospitals ³⁸ the policy, practices and process to prevent neonates from HAIs also carried out a review on the prevention practices. The prevention bundle was developed as the standard processes ^{39 – 41} and to implement the bundles we needed to train the healthcare provider providing direct care to the neonate. Audio-video recording, process flow, various levels of workstations as training method to deliver the behaviour and practice change was selected. Pre and Post training knowledge assessment was also carried out using a validated closed ended questionnaire.

Root cause analysis: Interview and brainstorming of healthcare workers for root-cause analysis was carried out as qualitative component. The healthcare workers providing direct care to the neonates were retrospectively interviewed using an open-ended questionnaire to find the causes leading to neonatal HAIs. Such interviews were conducted until the saturation of responses was met. Eighty different healthcare workers were interviewed to find the mechanism of spread of infection on open-ended questions in this process, until we reached to saturation of responses and similar responses were provided. The periodicity of responses were listed under six cause headings as per Ishikawa ⁴² i.e. A) Men, B) Material, C) Method/Process, D) Machine/Equipment, E) Environment, F) Policy.

The Ishikawa diagram, a diagrammatic representation of the root cause analysis was developed where the sources of infection were categorized under six domains. ^{43 , 44} Later it was recorded and tabulated to obtain descriptive statistics. Based on the periodicity of reporting of the contributor reporting was done through Pareto's chart. ⁴⁵ Pareto's chart was able to highlight the significant contributors that cause HAIs among neonates. A graphical expression of the contributors to the existing problem highlighting the important ones to target was used. In Pareto’s chart, no categorisation was used, however, the contributors with <5% of reporting frequency were summarised together. The top 20% of contributors identified in Pareto’s chart were considered significant to neonatal HAIs. These contributors were taken up further as input to the development of the prevention bundle.

The average score and standard deviation table were developed on major contributors reported during brainstorming and survey. Experiences of healthcare workers were reported using the box plot due to outliers.

Hospital Infection Control Committee’s (HICC) surveillance: Surveillance reports were collected periodically from the air samples from NICU’s environment every three months of the year. The results of these samples were collected from the nurse in charge of NICU. There were two episodes of the suspected outbreak during the study when swab samples from the environment of a few of the neonates clinically acquiring HAIs were captured to identify the contributors. These swab samples were from the cradle side, surface swab, water sample, swab from feeding pallada, and equipment was collected directly by HICC members. The samples were analysed in the central microbiology lab and reports were submitted to NICU. As the official report arrived from a laboratory, the researcher noted the report's results and took it as probable contributors to neonatal HAIs. These contributors were considered critical to preventing HAIs and ensured improvisation through a review of literature for inclusion in the prevention bundle.

System process mapping through observation: The prospective regular random observations were done for the process and techniques used to provide the neonate care. The researcher followed procedural steps from the beginning until the end of the procedure and observations were marked against the checklist, prepared by the researcher, clinician and nursing staff of NICU, considering the standard steps to be followed during a procedure. These checklists were not given to the healthcare workers for marking. After the completion of the observations, the record used was entered in the checklist and on the Excel datasheet. The checklist based on the process are available under Extended data. ⁶⁹

The time selected for observations was morning care time at 07:30 am; clinician’s clinical round time at 09:30 am and 3:00 pm; blood sample collection time at noon; invasive routine procedure time at 02:00 pm; medication/total parenteral nutrition preparation time at 03:30 pm; shift hand overtime at 07:30 pm. Various invasive and non-invasive procedures, care of the neonates and environment cleaning processes were focused during these random observations.

These observations were summarised in frequency and percentage under four categories in tabular format. The top 20% of frequent observations were considered as ‘scope for improvement’ for the prevention bundle development. ⁴⁶

A survey on knowledge and practice assessment on hand hygiene practices among healthcare workers in NICU was also captured, analysed and reported, and is published in peer reviewed journal. ⁴⁷ The observations on hand hygiene were classified under three categories i.e. a) time duration for hand hygiene practices, b) opportunities for hand hygiene before the procedures or activity and c) hand hygiene practice after the procedure or activity. Data collection forms for both survey and observation are available under Extended data. ⁶⁹

Failure effect mode analysis (FEMA): FEMA was done on critical contributors identified through brainstorming, observation and system process mapping with the help of eight senior nurses and two clinicians.

We also performed FEMA on 21 identified processes that helped us in identifying the key steps that if missed or overlooked could lead to HAIs also it helped us to add and delete some of the steps that were carrying no value to the process. FEMA helped us to identify the critical key steps to be added to the process with more emphasis. FEMA was done on various steps in the care process that helped to find the critical 20% steps to focus on during the preparation of prevention care bundle however, many other activities scored <240 risk propensity number (RPN), and were not reported.

Visit to benchmark NICUs in state: Study visits to three different NICUs located in distant hospitals setting infection control policies were made. Observations, open ended interviews of healthcare workers and the study of infection control policies were recorded for each NICU. The findings were compared and presented to stakeholders at the study setting to find the best practices.

Based on the findings of all the observations, root cause analysis, process mapping, FEMA and observations at other NICUs the critical steps in the processes and the critical processes were identified. It was decided to develop standard care process for routine critical activities as a prevention bundle.

Development of the prevention bundle: We carried out three rounds of brainstorming meetings with clinicians, microbiologists and nursing staff to formulate a feasible prevention bundle. The presentations on root-cause analysis, observations, and processes were made available through email at least a week prior to the meeting and a presentation was delivered before the meeting.

Four clinicians were identified from the NICU including two senior (minimum 10 years of experience in NICU) and two junior clinicians (minimum 2 years of experience in NICU), they were also given an open-ended questionnaire with the developed process to identify the lacunae and improvise the process.

Each meeting was of a minimum one-hour duration, clinicians were allowed to review each process for one week to provide their critical inputs. The inputs of all the healthcare workers involved in these activities were presented to experts’ i.e. senior neonatologist, senior microbiologist, three senior staff nurse and management representatives ( Figure 1). Their inputs were documented and were considered further for the design of the prevention bundle.

Each process had Input, Process and Outcome as suggested by Dr Donabedian. ⁴⁸ Any surgical, equipment or manpower required to initiate and complete the process was considered under the input (mentioned within the process); the steps involved to carry out the task were arranged in sequential order in process, and the neonates benefiting were considered as an outcome of the process.

As the experts' suggestions were to combine the developed processes into two to three processes, the researcher finally prepared three processes after repeating four consultations with each clinician (four) and nurses (ten) that lasted for three months. The final three processes were identified as: A) Nursing care Process; B) Vascular access process and C) Cardiopulmonary resuscitation management process.

As per the experts' policy suggestions, do’s and don’ts and surveillance charts for NICU were also prepared in the supplement to the prevention bundle.

Development of educational module to implement prevention bundle: The developed three processes were further taken up for audio-video and computer added presentation development. Three senior staff nurses were identified and trained for the process by the researcher. These nurses carried out all the steps of the process as per the instructions on a newborn mannequin in NICU on the cradle. Step by step each task was recorded for two procedure bundles: The nursing care process and the vascular access process. Audio-video recording was carried out during the day shift and three to four shots were taken for each task.

The raw video clips were edited in Windows Movie Maker ver 6.0 with background music obtained from licence free music library in YouTube. A due acknowledgement for the music source was given to the developer and source at the end of the video. The required audio clip was inserted and the length of the movie trimmed from 75 minutes to 15 minutes. Due acknowledgement was given to the video contributors. Final processing was completed in four months. A total of two videos, on the nursing care process and vascular access process was developed.

Policy document: This was prepared based on a review of the literature and the observations carried out at other healthcare facilities. The policy document included the incorporation of the initiative suggested during the brainstorming exercise and facilitated session. For the development of these policy documents, senior clinicians, a microbiologist, pharmacy expert and five nurses with a minimum of 5 years experience were involved. Prior to the discussion and meeting a draft policy was prepared which was communicated through email, feedback, suggestions and discussion until all agreed to the policy document. These policy documents can be found under Extended data. ⁶⁹

Do’s and Don’ts: These rules were prepared based on the observations carried out at the study setting and other hospitals, review, ^{49 , 50} and the brainstorming exercise with healthcare professionals. Three interviews with the nurse manager and nurse in charge were carried out to make the changes according to the feasibility of carrying out tasks. Meeting and discussion with respiratory therapist staff were carried out to validate the content of the list. Final corrections, validation, and approval were taken from senior neonatology consultants, senior microbiologists, and their team before the study was adopted. The agreement of senior healthcare professionals that includes clinicians, nurses and HICC was sought. These rules were presented to senior nurses, clinicians and respiratory therapists for content validation. It took five rounds of changes during content validation. These documents can be found under Extended data. ⁶⁹

Surveillance charts: The surveillance charts were developed based on the recommendations sought in the review of literature, the observations on practices at other hospitals and recommendations of the senior clinicians. These surveillance charts were presented for three rounds for content validation to nurses and clinicians.

Designing of assessment tools for pre and post-intervention: Ten multiple-choice and closed-ended questionnaires were developed for nurses and medical postgraduate students based on the identified processes and steps. Senior clinicians validated the questions (three) for their content validity and face validity. Internal consistency for the first set of questions was calculated with 15 samples of healthcare providers and cronbach's alpha was 0.812 for these questions.

At each intervention phase, the set of questions administered were different however the intent and purpose of those questions were the same only the language and phrases differ each time and internal consistency with cronbach's alpha was maintained above 0.75, where 15 healthcare providers helped in reviewing and answering these questions for validation. The respondents gave consent only at the beginning of the intervention, however, fifteen nurses were replaced by other staff at the middle of the intervention phase. The researcher included them from the next phase of intervention and took their written consent before the intervention. We anticipate the variation in the result due to these change of nurses in between the interventions. Still, no effort was made to adjust the findings for such variations and results were reported without any modifications considering the real time situations without any interventions.

The answers were evaluated for their correctness and a score of 1 was given to each correct answer whereas the wrong answer did not get any point. The summary of answer scores was prepared and reported as mean ± SD for each time the training was provided. A difference in mean and SD was computed to find the average change in pre and post-training. Students paired t-test with 95% CI at p-value fixed at <0.05 was considered a significant change in knowledge scores.

In training phase I, the training was divided into two stages: implementation of the nursing care process bundle and implementation of vascular access process bundle along with cardiopulmonary resuscitation management process.

The first stage of the experiment began after finalizing the date and week with the nurse manager when all the nurses can be asked to be present to attend training. The stage was divided into two parts for the ease of implementation and considering optimal time for the training effectiveness to achieve. Daily post morning shift change at 02:00 pm all the nurses approximate 7-10 in numbers but occasionally 3-4 reaches to the training room.

One-hour video-based training was conducted for the consented participants. A pre-test was followed by a video show of the procedure. The researcher actively narrated and discussed each step. Concerns and queries of the participants were addressed instantly before continuing further. After active discussion, a post-test was conducted followed by due gratitude to their participation and request for their cooperation and support in the prevention of neonatal HAIs.

The second stage of training was given on the vascular access process to the healthcare providers. One-hour video-based training was conducted for the consented participants. A pre-test was followed by a video presentation of the procedure. The researcher actively narrated and discussed each step. Concerns and queries of the participants were addressed instantly before continuing further. After active discussion, a post-test was conducted followed by due gratitude to their participation and request for their cooperation and support in the prevention of neonatal HAIs.

Participants were also briefed on the cardiopulmonary resuscitation management process using the flow chart. Participants were briefed on the do's and don’ts. A post-test was conducted followed by due gratitude to their participation and request for their cooperation and support in the prevention of neonatal HAIs.

A total of 49 and 48 nurses were trained in both areas respectively. We excluded 5 and 6 nurses respectively as three nurses were part of the development of the training module and others were on long term leave of more than 30 days.

Training to five respiratory therapists was given on do's and don'ts and cardiopulmonary rehabilitation management process using a flow chart. For them also pre and post-training evaluation was carried out.

5 medical postgraduates were trained using role-play and simulation models along with video teaching aids only on vascular access process, cardiopulmonary rehabilitation process and do's and don'ts. The stimulation and role-play included sterile gowning and sterile gloves donning after surgical handwashing practices that were monitored. During monitoring of the healthcare providers, any deviation or wrong practice if found was corrected followed by carrying out the procedure again in a standard manner.

A review meeting was carried out with the hospital infection control committee (HICC) where the problem statement and its mitigation plan was discussed regarding preventing cross-infection and environmental changes. The review committee agreed to certain environmental changes which were:

Notice at shoe rack area to separate and place street shoe and NICU sleepers in respective racks; Prefilled hand sanitiser and notice to use it before entering to NICU premises. Separation of two different rack positions i.e. street footwear and NICU footwear, placement of hand sanitiser at the door inside to NICU to ease monitoring of hand hygiene practices.

Segregation and labelling of weighing machine as “Only for neonatal weight record” and “Use for diaper weight”. Removal of tissue paper roll for hand drying purpose. AMBU (Air Mask Bag Unit) bag replacement at every 72 hours of use.

Other changes that were requested but could not be implemented due to unavoidable factors were: Separate trolley for invasive procedures at Inborn and Out born area, the indent of fresh AMBU bags to replace AMBU bag every 72 hours.

The training phase II was one stage and was on implementing the nursing care process bundle and the vascular access process bundle along with the cardiopulmonary resuscitation management process.

After finalizing the date and week with nurse manager when all the nurses can be asked to be present to attend training, the dates were fixed to the third week of December 2019. This time a little variation in the training approach was practised where the simulation stations were prepared and the participants were called to practice and show the learning. The corrections in practice were made at the simulation station only. There were five simulation stations, and these were: a)

Handwashing station

Each training time was generally scheduled from 1:00 pm to 02:30 and on two days was scheduled at 09:00 pm to 10:00 pm; two senior nurses were identified and sought for their help in implementing the training. These senior nurses had >10 years of experience in NICU and were also trained and briefed on their role at the station before beginning the training. Each participant was welcomed and gave a pre-test in written format. The consent was obtained only from those who were new to the training. After the pre-test, the participant was shown a video with oral narrations and descriptions by the researcher for 20 min duration. If any doubt arises in the middle of the video, it was discussed by pausing the video and then proceeded further. Comments, queries and discussions of the nurses were welcomed and encouraged during the presentation.

At each training day, three teams were formed among the participants. Post presentation the three groups were taken on three stations where the researcher and the senior nurse helper demonstrated the process and then assessed each participant while they demonstrated the practices. Correction in the practice if any deviation was found were made immediately at the station, and all the queries were resolved that were raised during a demonstration by the researcher. Each team practised and demonstrated their process at all the five stations. Participants were also briefed on cardiopulmonary resuscitation management process using the flow chart. Do’s and don’ts were briefed and handed over to each participant.

Post video and stimulation presentation each participant was given a written post-test to answer. The researcher thanked each participant and requested their cooperation and support in the prevention of the neonatal HAIs. In this stage, a total of 46 nurses were trained. We excluded eight nurses as three nurses were part of the development of the training module; three were utilized in training and others were on long term leave of more than 30 days.

Training for five respiratory therapists was given on Do’s and Don’ts and cardiopulmonary rehabilitation management process using a flow chart. For them, also pre and post-training evaluation was carried out.

The 5 medical postgraduates were trained using role-play and a stimulation model along with video teaching aids only on vascular access process, cardiopulmonary rehabilitation process and Do’s and don’ts. The stimulation and role-play included sterile gowning and sterile gloves donning after surgical handwashing practices that were monitored. During monitoring of all the healthcare providers, any deviation or wrong practice was corrected there itself, followed by carrying out the procedure again in a standard manner.

Phase III was one stage and was only on implementing the nursing care process bundle and the vascular access process bundle along with the cardiopulmonary resuscitation management process.

After finalising the date and week with the nurse manager, when all the nurses will be able to be present to attend training, the dates were fixed. This time a little variation in the training approach was practised; we identified seven nurse leaders with >5 years of experience, each nurse leader could choose any six nurses from their shift team to train further. The simulation stations were prepared, and the participants were called to practice and show the learning. The corrections in the practice were made at the simulation station only. There were two simulation stations and these were: Medication preparation stations; and vascular catheter insertion station.

The training time was scheduled as per the convenience of these senior nurses. The nurse assessed the handwashing, medication delivery, and neonatal routine care in real-time while providing the care to the neonate. The researcher trained each team leader and briefed on the training protocol with a pre and post-assessment questionnaire. Two to three staff nurses were welcomed to participant for pre-test and were trained by team leader each day completing their assigned six nurses in 3 to 6 days duration depending upon their shift duty. A video with oral narrations and explanation by the team lead for 10 minute duration was delivered. The nurses' comments, queries, and discussions were welcomed and encouraged during the presentation and resolved before moving further.

The team lead demonstrated the process and then assessed each participant while they demonstrated the practices at workstations. If found, corrections in the practice were made immediately at the station and the team leader resolved all the queries. Each team practised and demonstrated the process at two stations. Participants were also briefed on the cardiopulmonary resuscitation management process using the flow chart followed by do’s and don’ts in NICU.

Post video and stimulation presentation each participant was given a written post-test to answer. The team leader thanked each participant and requested their cooperation and support in the prevention of neonatal HAIs.

In this stage, 40 nurses were trained. We excluded 12 nurses as three nurses were part of the development of the training module, seven were utilized in training, and others were on long-term leave of more than 30 days.

Training to respiratory therapists and medical postgraduates could not be carried out at this phase due to the complete lock-down of the country owing to the surge of COVID-19 and as directed by the Institutional Ethics Committee. However, the prevalence of neonatal HAIs was captured until the completion of three months.

Categorical variables were reported using descriptive statistics, a time series graph was used for reporting changes in HAIs. Paired students T-test was performed on findings of the pre and post training.

Written consent from all the participants of training were obtained before the training. They were informed that the data will be used for academic and research purposes. In no way would their personal information be disclosed or identity revealed either in the final report, publication or anywhere. Confidentiality will be maintained at all times, and privacy is respected. Institutional Ethics Committee (IEC) approval was taken, approval ID: MUEC/014/2016-17. CTRI (Clinical Trial Registry India) registration was done before starting the project, the confirmation ID was: CTRI/2017/08/009538.

Healthcare workers were interviewed on an open-ended written questionnaire, a total of 80 healthcare workers participated, and the majority of them were staff nurses with more than two years of experience (88%). Healthcare workers were identified as major contributors for HAIs ( Table 1).

The ranking of the contributors to HAIs by these healthcare workers showed that prime contributors are healthcare workers (27%), followed by equipment (21%), material (21%), process/protocol (19%), and last is infrastructure (12%). Detailed root-cause analysis and multiple observations showed that non-compliance to handwashing practices, equipment disinfection/cleaning, IV line insertion, many invasive procedures, and improper aseptic techniques (>50%) are prime contributors causing HAIs among neonates.

As the number and variation in responses of healthcare workers were very high the Ishikawa diagram was plotted to find any inference out of the responses ( Figure 2). To simplify further, the causes were ranked ordered on a scale of 1-5 by healthcare providers that helped for further analysis of responses in the form of Pareto’s chart ( Figure 3). As per the rank order the hand hygiene practices (87%), improper aseptic procedures (83%), improper handling of venous line (57%) and unable to provide timely isolation to sick/infected neonates (53%) were highlighted ( Figure 3).

Observations against a checklist to find causes of HAIs in NICU

Physical observations for 248 days and 1761 various infection control opportunities at the study site was made ( Table 2) and forty interviews on twenty-five variables were carried out for healthcare workers at NICU. The observations were carried out on the following practices: a) hand hygiene practices, b) care of intubated and ventilated patients, c) intubation process and ET (Endotracheal tube) suctioning, d) care of NIV (Non-invasive ventilation) patients (Oral suction and fixing of the nasal mask), e) insertion and care practice of PICC (Peripherally inserted central catheter) line, central line, umbilical line, arterial line and IV lines, f) practices on infant feeding g) cleaning and care practices at NICU, h) daily clinical rounds, I) medication preparation, sample collection, and blood transfusion practices

There were two occasions when HICC collected the environment sample for the identification of microorganisms. In its first occasion, only air samples were collected at different parts of the NICU that did not grow any microorganism on culture. On the second occasion, the swab samples were collected from various environments of the NICU as well as nasal swabs of healthcare workers.

A total of twelve swab samples, including one drinking water sample, were also sent to the microbiology lab for further analysis. There was a growth of Klebsiella sp. from the hand washbasin and neonatal cradle. The culture grew Serratia marsescens in swab culture of the oral cavity of one patient and feed container, Klebsiella pneumoniae in swab samples of cradle side, and no growth in a drinking watersample. There was no other growth of microorganisms on any of the other samples.

System process mapping and FEMA on processes: Multiple observations and interviews of healthcare providers were carried out to identify and understand the processes critical to infection control practices.

We identified the 21 processes and its process validation with the help of healthcare providers at NICU. We found site cleaning with 4% chlorhexidine, disinfection of IV port before infusion, the sterile container for feed preparation, preparation of medication under laminar flow and strict hand hygiene practices were few critical steps of the processes ( Figure 4).

All the four NICU settings had an average of 80% occupancy and any two to three months per year, the occupancy goes up to 100%.

The observations were made and compared with existing practices in the study hospital ( Table 3). These observations were categorised further under the following headings:

Environment-related practices

Human related practices

Practices/processes

Process monitoring

Material/supply related practices

Development of prevention bundle and training module

Printed three bundle processes i) Vascular Access and care process ii) Nursing care and iii) cardiopulmonary resuscitation management process were developed. Audio-Video training & hands-on workstation training was also developed as prevention bundle module for healthcare workers to prevent neonatal healthcare association infections.

Two videos, a PowerPoint presentation, and five hands-on stations were developed as educational modules to implement the prevention bundle.

There were four individual sets of training given to nursing staff and where there was a change of overall 21% knowledge score ( Table 4). Other healthcare workers were given training for only one time due to their unavailability and resistance to participating ( Table 5). Housekeeping workers were trained, but their knowledge assessment was not carried out. The prevalent microorganism identified during the interventional period was Klebsiella pneumoniae (41.3%), followed by Acinetobacter baumanii (27.5%). All these microorganisms were identified in blood samples. The rate of HAIs at the beginning of the intervention was 9.6, which at the end of the intervention was noted to be 7.0 as of April 2020 ( Figure 5).

At the beginning of the study, we aimed to bring at least a 5% change from the baseline rate of HAIs; the current interventional study was able to bring about a 26% reduction in the rate of HAIs reducing it from 9.6 to 7.0 per 100 admissions >48 hours. The bacteraemia rate fell from 5.2 per 1000 patient days to 2.6 per 1000 patient days and was statistically significant on two-tailed student t-test with 95% CI at p=<0.05 with p-value=0.00073

The following method was used for the calculation: ARR = Control event rate − Experimental event rate ARR = 0.096 − 0.07 = 0.026 NNT = 1 / ARR = 1 / 0.026 = 38.5

On average, 39 patients would have to receive this prevention bundle (instead of standard care) for one additional patient to prevent from HAIs.