Implementation and evaluation: A QI team comprising of a faculty, two resident physicians, a nurse educator and two senior staff nurses explored the reasons for poor breastmilk expression using fish-bone analyses. Baseline data collected for initial 7 days included time of first expression of breast milk (EBM) in mothers delivering preterm neonates, daily amount of EBM and percentage of neonates on exclusive mother’s milk feeding in NICU. Poor awareness among healthcare providers and lack of maternal counselling were identified as most important causes for the problem. Hence as a change solution, a comprehensive postnatal counselling was implemented wherein the bed-side nurse was assigned the responsibility to initiate the first milk expression within 6 hours of delivery. The mother was educated with videos on breast feeding and encouraged to pump every 4-hourly as soon as she was stable postpartum. Successive PDSA cycles focused on staff motivation and the primary nurse whose neonate achieved exclusive breast feeding on or before 7th day of life was awarded with a certificate. Run charts were displayed showing percentage of neonates each day on exclusive EBM. This QI also addressed the supply chain issues like the availability of breast pump and its accessories as well as a refrigerator for milk storage both in NICU and postnatal wards.

TABLE I Five Whys Analyses for Deriving the Root Cause of a Problem*

The ‘SMART’ (Specific, Measurable, Applicable, Realistic and Timely) aim was to increase the life span of temperature probe from the current lifespan of 9 days by 80% over a period of 8 weeks. The change ideas involved providing a refresher course for nurses on equipment care, supervising mothers and junior nurses while handling probes and tracking probe breakage. After implementing the PDSA cycle 1, the life of temperature probe increased up to 40 days, which was not as good as expected. Exploring the reasons, it was noted that mothers needed more support and can be taken as partners in QI. In 2nd PDSA cycle, the bedside nurses provided face to face education of mothers on taking care of warmer equipment and made it a point to discuss status of temperature probes during nursing hand-over in each shift. After the 2nd PDSA cycle, the minimum time interval between two temperature probe break-down increased to median of 60 days as indicated by run chart (Fig. 1).

Fig. 1 Annotated run chart showing life span (in days) of temperature probe in the NICU. P1-P8 on the x axis indicates an individual temperature probe. The dashed line indicates median life span.

Learning and Implications for clinical practice: This QI initiative began because one astute nurse recognized that an order for new temperature probe was being placed too often. She had to convince the other staff that something was wrong. Unless people recognize the problems and are involved in initiating the change, there is often resistance. The team used 5-whys analyses to identify the major cause of probe breakdown. They came up with solutions, tested them to see if it worked and then implemented those that worked. Following this QI, recommendations were made in the unit to provide regular in-service training of nurses on proper equipment use and its maintenance. Involvement of mothers as partners in care as suggested by this QI initiative improved the success rate.

Hypothermia is an important contributor of neonatal mortality and morbidity. During an audit, we observed that majority of inborn preterm neonates <32 weeks gestation admitted to our NICU were hypothermic with mean admission temperature of 35.6°C despite the use of plastic wraps, attention to warmth during resuscitation and transport. We wanted to improve the mean admission temperatures (AT) of preterm neonates from the existing baseline of 35.6°C to normal range (36.5-37.5°C) by implementing all the components of an evidence based thermoregulation bundle in a stepwise manner using PDSA cycles over a period of 11 months [12].

Implementation and evaluation: A multidisciplinary QI team was formed who analyzed the causes of hypothermia using fish bone analyzes (Web Fig. 1). During an observation period of three months, a dedicated nurse observed the delivery room activities and a Pareto chart (Web Fig. 2) was constructed to identify those few important practices that needed to change. The Pareto chart revealed that the application of plastic wrap at birth was not proper in a majority of cases. Residents explained that the cling wraps crumpled when they tried to wrap the babies. Thus the changes in PDSA cycle 1, included the use of plastic bags instead of cling wraps for resuscitation. However, a look at the run chart revealed that hypothermia was persisting. The team discussed the issues with resident physicians and nurses and noted that the plastic bag available in market was stiff and the cut ends remained open exposing the neonate. Also the delivery room (DR) nurses were not aware of this QI initiative. So in next cycle, the team reinstituted the use of cling wraps but used a new way of application. The delivery room nurses were educated on the importance of hypothermia and how they could contribute to the initiative. In cycle-3, thermometers were installed in DR to monitor room temperature and the team co-ordinated with nursing assistants to ensure prompt transfer from DR to NICU. Frequent feedback with run charts of AT and appraisal in monthly meetings were done to ensure compliance and staff motivation in the post-intervention phase (4 months).

We studied a total of 79 preterm neonates during this QI initiative and noted AT in the post-intervention period with mean AT of 36.5°C and a significant drop in the incidence of moderate hypothermia (axillary temperature between 32-36°C) from 50% to 12.5%. We used a statistical process control chart for trending admission temperatures over time and demonstrated both a sustained increase in mean AT as well as decreased variability as reflected in the narrowing of the control limits (Web Fig. 3). As a balancing outcome, we did not notice any episode of hyperthermia or increase in mortality among neonates.

Learning and implications for clinical practice: This is an example of a complex QI study where team members implemented a bundle of interventions in different regions of hospital namely labour room, obstetric operation theatre and NICU. When too many etiologies for a problem were noted in the fish-bone diagram, they used the Pareto’s tool to identify those few important causes that contribute most. The Pareto tool also called as ‘80-20 rule’ states that 20% of the causes are responsible for 80% of the problems and can help one to focus on the important few. The first PDSA was not successful but the team innovated a new method in the second cycle. Some problems need a multi-pronged approach or a bundle of interventions and one can simultaneous test 2 or 3 change ideas as opposed to serial testing. The other lesson from this QI was to involve all stake-holders from the multiple areas for success. While we used run charts [13] for tracking and display, we used the statistical process control charts (SPC) to understand the process better. The SPC charts have an upper and lower control line (±3 standard deviations) in addition to the central line which represents the mean as opposed to the run chart where central line represents median [14]. While both run charts and SPC charts can visually display data over time and inform whether changes have resulted in improvement, the SPC charts in addition tell us whether the process is stable over time. A stable process, has most points near central line, few points near control lines and none beyond control lines. If points are above the control line, it indicates a special cause variation due to changes introduced in the system that needs to be investigated. There are other pointers to special cause variation too [14]; and in our example we had eight successive points on the same side of the centre line as an indicator.

Sustaining an improvement is important and in our NICU we have implemented the following – regular monitoring of delivery room temperatures, a checklist to ensure that all resuscitation supplies and equipment are available and we use admission temperature as an indicator of quality resuscitation, which is regularly audited for feedback. Dutta, et al. [15] from India reported a similar QI initiative to improve admission hypothermia in neonates wherein they showed lesser hypothermia as well as reduced mortality and late onset sepsis. The entire QI could be done without major cost, with no additional staff or new equipment. All the change ideas focussed on making the existing system and process of care easier for staff to implement the interventions.

Breastfeeding problems are very common, almost 80% of mothers experience one or more problems related to breastfeeding like sore, cracked or flat nipples, painful or engorged breasts, difficulty latching, feeling of decreased milk output etc. while in hospital [16]. These problems, if not avoided or corrected with proper education and support can even lead to early cessation of breastfeeding.

Breastfeeding in our postnatal mothers is primarily assessed by resident doctors and counselling regarding specific or anticipated problems is done individually at the bedside. Our postnatal follow up clinic that runs all working days for 2 hours specifically caters to neonates who are followed up within a couple of days after discharge. We observed that up to 20-30% of mother-baby dyads presented with issues related to breastfeeding. Hence we designed a QI initiative to address this [17].

Our SMART aim was to reduce the incidence of breast feeding problems at discharge from baseline of 75% to at least 37.5% among mothers delivering term babies by implementing a post-partum education package over a period of four months from September 2016 to December 2016. A QI team was formed and two members (a nurse educator and neonatal resident) objectively assessed breastfeeding among eligible mothers in initial phase (4 weeks) using LATCH breastfeeding assessment scores [18] and collected data on breastfeeding issues. Fishbone analyses was done, and the team planned to introduce a learning package for mothers. The nurses in post-natal wards were involved in teaching activities (phase 2 lasting 4 weeks) which consisted of distributing pamphlets and video demonstration on breastfeeding. Mothers’ perception and acceptability of the education package was then assessed on a 5-point Likert scale at the time of discharge. In the phase 3 (2 weeks), the team stressed on compliance with education package by having a checklist attached to neonate’s file. Focused group discussions with postnatal nurses and mothers were conducted to obtain feedback and to identify barriers in implementation. In phase 4 (2 weeks), a lactation nurse provided one to one support to all postnatal mothers in addition and in phase 5 (4 weeks), videos were uploaded in mother’s cell phone for repeated viewing and other measures were continued.

A total of 330 mother-infant dyads were enrolled in all the phases of QI. Incidence of breastfeeding problems at discharge gradually decreased from phase 1 (baseline) to phase 5 from 72.6% to 6.8% (P<0.001) (Web Fig. 4). Compared to baseline, the proportion of mothers with LATCH score >8/10 at end of final phase (RR 3.8; 95% CI 2.7-5.5, P< 0.001) as well as each of the individual phases were significantly greater. The compliance with discharge checklist increased from 29.8% to 100% and mothers felt that the educational package had utility as well as acceptability, with 43-65% of mothers strongly agreeing in favor of the education bundle.

Learning and implications for clinical practice: This QI initiative addressed breastfeeding problems in postnatal mothers by successful implementation of educational intervention driven by stepwise rapid cycle PDSA based approach. As outcome measures of improvement, the team used percentage of mother-infant dyads with problems at discharge and LATCH score to objectively assess breastfeeding. As process measures, they tracked the compliance with the interventions using a checklist. The team used p-chart (p stands for proportion) to monitor the proportion of mothers with breastfeeding problems. Similar to a control chart, p-chart also has a central line which corresponds to the mean proportion and 2 control lines corresponding to three standard deviations (SD) around the mean. A special-cause variation that requires investigation is indicated by point (s) outside the control limits.　 Web Table 1 lists the various QI projects done at AIIMS NICU.

After multiple QI efforts, the NICU nurse educator emerged as a champion who further supervised more QI work like, increasing the duration of kangaroo care for preterm neonates in NICU, decreasing medication errors, antibiotic stewardship and streamlining of post-natal follow up clinic. QI had also been topics of dissertation work for postgraduates in the department; the QI work on preventing admission hypothermia, resolution of breast feeding problems in the post-natal ward (both discussed above) and improving oxygen saturation targeting among neonates on oxygen therapy are dissertation work done by postgraduates. An ongoing QI dissertation work focuses on implementing potentially best practices in the first week of life among preterm neonates to reduce the incidence of bronchopulmonary dysplasia.

Dissemination and capacity-building: AIIMS, a WHO Collaborating Center for Newborn health disseminates QI to other organizations through its on-line resources, face-to-face workshops and the Point of Care Continuous Quality Improvement (POCQI) module. A dedicated website (http://aiimsqi.org/) on quality improvement provides free resources for teaching and learning QI, and as a platform for capacity building of teams and sharing QI work. The site also has short videos and posters of various QI projects done in the Institute. Many SEARO (South East Asian Regional Organization) countries participated in a Regional Workshop for Improving Quality of Hospital Care for Maternal and Newborn Health in May 2016 at New Delhi and have initiated QI work in their areas. QI project power-point template is freely available in the site that teams can use as a framework for QI work.

Teaching and Research in QI: QI is taught to undergraduates and nursing students at AIIMS. Postgraduates are encouraged to be part of QI teams in clinical areas and as topics for research or dissertation work. A good starting point for a dissertation or thesis on QI work is a single-centre prospective study with a pre-post design. The focus should be on implementation rather than testing whether certain practices are effective or not. So careful planning is essential to understand the existing processes, to identify how and why the current practices differ from recommended or evidence-based practice and to involve all stake-holders who may influence implementation. The team should identify the data that needs to be collected to measure improvement before study begins. Broadly this includes process measures (measurement of processes to see if the changes are being implemented), outcome measures (impact of the changes on patient or provider outcomes) and balancing measures (unexpected outcomes or changes both beneficial or harmful that are introduced either in the same system or other parts of system due to improvement efforts) [19]. For example, if a team plans to implement skin-to-skin care (STS) at birth for neonates, exclusive breastfeeding rates at discharge can be taken as outcome measure, the percentage of eligible mother-baby dyad experiencing STS, number of STS checklists filled per month and time to first breastfeeding can be process measures and incidence of acute life threatening event in neonate (desaturation or bradycardia) during STS can be chosen as balancing outcome. Published guidelines are available for reporting QI work in order to improve the completeness and accuracy of reporting so that improvement efforts can be reproduced at other suitable settings [20]. It is equally important to publish QI work to share one’s improvement experiences, so that others working on similar problems can adopt or adapt the change ideas without wasting time, effort and money testing the same ideas.

What started as small projects in a few departments mushroomed into several projects wherein several other departments like Rajendra Prasad Ophthalmic Sciences Centre, Emergency medicine, Pediatrics, Obstetrics, Cardio-Neuro centre and Jai Prakash Narayan Apex Trauma Institute also participated. QI made people recognize the problems in their system, utilize various tools (process mapping, fish-bone analyses, 5 whys) to understand the cause, come up with changes or solutions tailored for their system, test the changes in small scale over a short period or on few patients (PDSA cycles), and adapt/adopt or abandon the changes and finally implement them. People understood that not all changes lead to improvement and collecting data to assess baseline performance and reassessing performance over time using annotated run charts or statistical process control charts help in differentiating day-to-day variations from improvement. Importantly, the teams learnt the scientific way of doing QI activities. There was a lot of learning from each project (both success and failure) that people shared in common meetings. Salient learning points are given below:

Bottom up approach: We created a culture for QI in the unit by building conviction among front-line workers. The human element is the most important component of any QI success story and staff motivation and engagement is essential to work towards a common goal [21]. Each team member contributes to improvement and in many situations, we observed practical change ideas not from leaders but from front-line staff. In our case, NICU nurses identified the problem of frequent temperature probe breakdown and came up with change ideas like empowering the bedside nurse to help mothers with milk expression; the housekeeping staff volunteered in hypothermia QI and offered timely and quick transport of neonates to NICU, and mothers of admitted neonates joined as partners in some QI initiatives.

Understand the system and eliminate waste: Traditionally, the solution to most problems are handled by increasing human or material resources; but in healthcare system in which resources are limited, focussing one’s attention to understanding the system to identify the root cause and eliminating the redundant and wasteful processes can help to utilise resources effectively [22]. In our case, breastfeeding counselling by multiple team members with different messages, lack of supplies in the delivery room despite their availability etc. could be rectified when the counselling became more standardized using pamphlets/videos and a checklist was used for arranging supplies in the delivery room. Behavioral change of healthworkers is also important in eliminating waste and facilitating this change will require education, training, motivation and feedback.

Step-up and then scale-up: It is very essential that initial projects be simple, doable, patient-centric and under the control of the team members. Disease processes or problems with multifactorial etiologies take time or produce no improvement at all. Undertaking such projects at the outset would demotivate staff who may feel over-burdened and incapacitated. Our team felt comfortable to use the Institute for Healthcare Improvement’s Plan-Do-Study-Act (PDSA) cycle for rapid cycle improvement and changes were accomplished through small and frequent PDSAs [23]. Only change ideas that were tested on a small population and adapted for local context were implemented on a large scale. Building sustainability in each project is so important [24], that improvements are maintained in long run and the changes become a new norm in the unit. This involves making changes in the system itself, use of visual reminders, posters, score cards and identifying champions who would lead the change movement. In our NICU, the following system changes were implemented to sustain the improvements; regular in-service training on equipment handling for nurses and residents, installation of room thermometers in delivery room and tracking of temperatures each shift, charting admission temperatures in the NICU admission register as a quality indicator, and use of videos for breastfeeding counselling.

Celebrate and share success: Having QI meetings both within and between departments helps to show case activities and rally slow movers to gain pace. The QI teams from various departments attended learning sessions and workshops organized by AIIMS from time to time where they had the opportunity to learn from experts, presented their work and shared their experiences. The teams also met face-to-face with other teams, participated in webinars and conference calls, and presented their work at conferences. Teams learnt that they could learn and achieve more by collaboration than working in silos [25].

Time and patience surpasses hurdles: We understood early that problems have to be solved by a team of people and getting people on board could sometimes be challenging. Such factors include staff misunderstanding the principles of QI, feeling over-burdened with data collection, fear of being monitored or blamed for poor performance or just an inertia to change. In our hand hygiene QI, we noticed nurses’ resentment against being monitored, in our exclusive breast feeding QI, only a couple of nurses from one NICU were involved. Thus initially like-minded ones teamed up but with time even the disinterested become passionate about change. Nurses participating in QI mentored fellow nurses, and with time came up with ways to collect or add essential data in the nursing flow sheet itself thus decreasing additional paperwork and burden.

Organizations striving to improve their outcomes should initiate and support quality improvement activities. The important requisites for successful QI ventures are listed in Box I.