The main results have been presented following the sequence of events from inpatient admission to post-discharge follow-up of malnourished infants u6m as illustrated in Fig. 1. Both the quantitative and qualitative data is reported using the three key areas; i) assessment and admission, ii) inpatient management, iii) discharge and follow-up, each with appropriate sub-headings to provide further clarity.

Challenges in data collection

Locating the data required to compare the guideline recommendations with the recorded practices in both hospitals was challenging. There was no specific area in the Inpatient File allocated for the recording of the diagnosis and nutritional management of infants u6m (data required for guideline implementation)). A nurse in one of the hospitals explained this:

IDI01M
Okay the nurses we normally record in the cardex, though initially we had incorporated into the file a document that was capturing that [diagnosis and nutritional management] just a document for nutrition issues, but then you know when the County came in they started redesigning the files again and that document that we had incorporated was actually removed so we only document in the cardex and then the MO interns also when they review the patients, they also document in their notes.

In both hospitals, the same information was sometimes recorded by two different HWs in two different locations of the Inpatient File. Such practices sometimes led to different information being recorded for the same infant with some contradictory results (see details under nutritional classification sub-heading).

Patient’s description

From the patient files reviewed, infants were slightly older during pre-intervention period with median age of 30 days (IQR 4 to 90 days) compared to during the intervention when the median age was 21 days (IQR 7 to 90 days). The median hospital stay of six days did not vary between the pre-intervention and intervention periods.

1)

Infant nutritional assessment, classification and diagnosis at admission

a)

Anthropometric assessment

In both the pre-intervention and intervention periods, more admission weight measurements were recorded compared to admission length measurements. In the pre-intervention period, admission weight was recorded for 142/170 (84%) of infants while length was recorded for 111/170 (65%) infants u6m. During the intervention period, admission weight was recorded in 47/65 (72%) of admitted infants, length in less than half 30/65 (46%) of the admitted infants u6m.

b)

Nutritional classification

In the pre-intervention period, 35/170 (21%) of the admitted infants records indicated admission diagnosis as acute malnutrition although an additional 11 infants who had a Z score of less than minus 2 at admission did not receive the malnutrition diagnosis. During the intervention period, acute malnutrition was recorded at admission for a slightly larger proportion 18/65 (28%) of the assessed infants. Only four infants with qualifying Z scores were not indicated as malnourished. It is difficult to make any assumptions about why there might have been this slight increase in proportion of infants with malnutrition diagnosis and it’s particularly puzzling in light of the decline in the proportion of infants who had anthropometry recorded on admission. The sample size is small and the period during the intervention lasted only 6 months (compared to the full 12 months for the pre-intervention audit). It is possible that the intervention was being implemented during a particularly challenging nutrition season. Importantly, it became apparent during the audit that the diagnosis of acute malnutrition could be a contested issue. That is, the diagnosis of acute malnutrition recorded in the inpatient file by the clinician sometimes differed from the recorded diagnosis in the nutritionist’s records. For example, in the pre-intervention audit of data from Hospital 1 the nutritionist’s records reported 32 (25%) cases of acute malnutrition at admission while clinicians reported only 20 (16%) of the same infants as acutely malnourished.

The data from interviews with nurses and nutritionists suggests that one reason for these discrepancies could be that some medical interns either didn’t weigh the infant, or they recorded incorrect weights.

IDI05M
Respondent:. .. If it’s the weight, sometimes you can find the clinician maybe is tired he/she will estimate that weight but when we come here we just do it physically, we want to measure the weight we see how is it exactly, the height so that we can get the right measurement for the feedings IDI05M
Respondent: So, we will have to redo and see the calculation if they are okay, we will have to re-ask the nutritionist if this formula is okay, you see, so you can’t just rely on what they (MO interns) have said.. . IDI03M
Respondent: I have gotten that incident like in three times, the MO interns, they change the weights. .. yeah, so that’s the problem. So, I document it in the register so that I may tell them I did it and I got this, so we try to stick with the nutritionist weight, yeah.

This situation was exacerbated by the lack of sufficient weighting scale, with scales being moved between wards so that time could be wasted trying to locate the necessary equipment. A nutritionist in Hospital 1 also expressed concerns that the scales were manual and these took considerably longer to use than a digital scale:

IDI02M
Respondent: Another thing are the resources, like in the Paediatric ward we work with the manual weighing scales, we have that one which you have to push it, yeah, so for that, we prefer the digital scale because when you go there in the morning, let’s say we have 15 children that you have to review and you have the manual weighing scale, so for a child you can take five minutes because in the morning you have to reset it again [. ..] for you to weigh the child, so we prefer the digital because they are giving accurate measurements. We used to borrow from New Born Unit (NBU) then, for the side of infection there is no need

A second possible reason provided by the nurses was that the medical interns did not always correctly calculate the Z score and so were unable to correctly diagnose malnutrition, with their calculations needing to be double checked by the nutritionist.

IDI05M
Respondent: “So we will have to redo and see the calculation if they are okay, we will have to re-ask the nutritionist if this formula is okay, you see, so you can’t just rely on what they (MO interns) have said.” c)

Breastfeeding assessment

In the pre-intervention period, 157/170 (92%) infants had breastfeeding information recorded in their inpatient file. Of the 157 infants, 96 (61%) were reported to be exclusively breastfeeding at admission. However, of the 35 infants diagnosed with acute malnutrition, only 28 (80%) had breastfeeding status recorded at admission, and this was primarily through self-reporting, not by observation as recommended by the guidelines. One of the nurses described that the lack of observations was due to shortages of staff:

IDI01K
Respondent: because of the shortage of staffs, sometimes you find that even if they give it, they don’t give it as a whole, maybe someone will just a mother, are you breastfeeding? are you practicing exclusive breastfeeding? If she says yes, they don’t give further information. .. .

In the audit of records from the intervention period, we observed no change in recording of breastfeeding status of infant u6m at admission. While the BFPS were present, a similarly high proportion 59/65 (91%) of admissions had breastfeeding status indicated in their records. Additionally, 43/59 (73%) infants were recorded as being exclusively breastfeeding at admission. Of the 21 infants diagnosed with acute malnutrition, breastfeeding status was recorded for 19/21 (90%) of them.

d)

Maternal/Caregiver’s health evaluation

The audit of records from the pre-intervention period, indicated that fewer than a quarter of caregivers 38/170 (22%) received some form of maternal health assessment. By contrast, in the audit of records completed during the intervention almost a half of care givers 29/65 (45%) received some form of physical and mental assessment. Of the 29 caregivers assessed during the intervention implementation, two were recorded as experiencing mental health challenges requiring specialized treatment. However, in neither case was there any record of the maternal mental health assessment tools used in the diagnosis and no information was recorded on the treatment or referral recommended for the caregiver.

Though not recorded, the interview data suggests that the nutritional status of the mothers was also assessed and if a mother was viewed as being undernourished, they were registered into a supplementary nutrition support programme.

IDI02M
Respondent: We counsel them (mothers). .. if it happens that the nutritional status of the mother is not that conducive or convincing, so we have to put the mother into supplementary program to support the mother with nutritional supplements 2)

Infant nutritional management during admission

In the pre-intervention record audit, only a fifth (36/170; 21%) of inpatient files contained an infant nutrition management plan. The audit of inpatient files completed during the intervention period revealed a considerable improvement in recording of a nutritional management plan with over four fifths of the infants (55/65; 85%) having a nutritional management plan recorded in their admission file.

From the interviews, we learn that the BFPS worked alongside the nutritionist who used information collected by the BFPS to populate inpatient documents. It seems the nutritionists used information recorded by the BFPS in her tool to populate the nutrition register and inpatient file as illustrated in the quotes below.

IDI02M
Moderator: Alright,. .. you had earlier on mention about some of the changes in practice that you experienced through the breastfeeding peer supporter’s intervention. Was there a change now in the data recorded when you’ve been having the breastfeeding peer supporter? Respondent: Yeah, there is, there is because like apart from her using the counselling forms in for example in pediatrics, the register, the register is also capturing these young infants who are below 6 months and they’ve been having challenges with breastfeeding and the came with severe acute malnutrition, yeah that was actually being documented on a daily basis. Moderator: So, she was documenting all that information? Respondent: The nutritionist were the ones who actually used to document the register, but her she used to have those forms where she actually documents the counselling forms together with those other research forms. a)

Feeding plan/re-establishing exclusive breastfeeding

In the pre-intervention period, 104/170 (61%) infants had a detailed breastfeeding management plan recorded. In the records, details were given in reference to demonstrating breastfeeding techniques such as positing and attachment, enrolling mothers into a nutrition support program or prescription of lactogogue to mothers to initiate breastmilk production. Further, among the 35 infants diagnosed with acute malnutrition, there was detailed records indicating that for 21 of these mother’s (71%) their milk was supplemented with formulae (F-100, F75, or Prenatal nan formulae).

Prior to the intervention, the data from the interviews suggested that breastfeeding support was not a central part of the feeding plan for acutely malnourished infants with nutritionists reporting that would give supplemental milk to infant as they “waited” for breastmilk to increase:

IDI02K
Respondent: Okay, when we assess we have said that we assess the mother’s milk by expressing, so when we express two to three times and we don’t attain the quantity that we need the child to feed per maybe after three ours or two, then we know the child is being underfed, so we look for an alternative, as the mother we supplement with either porridge so that we increase the milk, we do have something to support the child as we wait the milk to increase, so we use the F-100 dilute

By contrast, in the intervention period, 55/65 (85%) infants had a detailed breastfeeding management plan recorded in their inpatient file. The majority of these plans included details on how to support caregivers improve breastfeeding techniques, and how to support cup feeding of hand expressed breastmilk. Of the 21 infants diagnosed with acute malnutrition, only a third (7/21:33%) were recorded as having received Dilute F100 milk while the majority were recorded to have been supported to retain exclusive breastfeeding.

The interview data suggest that the presence of the BFPS in the ward provided the workforce and time needed to implement the feeding guidelines more effectively reducing overdependence on diluted F100.

IDI02 K
Respondent: Ok. the peer supporter was of great help in the management of malnourished children in the ward especially the under 6 months children coz after admission, we do calculate the feeds but also, we give the time to the peer supporter to at least talk to the mother. So far for the mothers of the children that we had in the ward while we gave the peer supporter time to discuss with the mother, we found that after two to three days we found that the milk the breast milk started flowing very good and we could even stop the formula milk and just continue with the breastfeeding only IDI02 M
Respondent: Yeah, of course there was effect because we saw children who actually came in with very low weight actually finally gaining weight and finally getting discharged when the mothers have really established breastfeeding effectively. Yeah, so that was very, very important and like initially, initially it was a bit of a challenge because the first thing that we would ever thought about was any time a mother came in would first think about supplementary feeds that is F100, yeah so, F100 diluted and then see whether we can. .. we used that’s what we used to do and then support the mother with that as we do counselling to see whether they can actually establish breast milk, but when we started then we realized that it was very much easy to achieve not just by using artificial feeds, but just counselling and showing the mother how to attach it was actually working very effectively. 3)

Infant discharge and follow-up

a)

Discharge

Both prior to and during the intervention, important discharge information was missing from the records of the infants u6m. In the pre-intervention period, three infants died after admission. Discharge anthropometry was recorded for less than a quarter 39/167 (23%) of the discharged infants with a similar percentage having their breastfeeding status recorded 36/167 (22%). During the intervention period, six infants died after admission. Discharge weight was recorded for 18/59 (31%) and length recorded for 6/59 (10%) of infants. Exclusive breastfeeding at discharge was recorded for 18/59 (31%) while 6/59 (10%) infants were recorded to have been discharged while still on supplemental dilute F100 feed.

The nutritionist in both hospitals suggested that a potential reason for the lack of anthropometric and breastfeeding status data from the discharge records was that they were often not on duty when the discharge process was taking place. The clinical team were in charge of discharge decisions, and it became clear that the primary criteria for discharge was clinical rather than nutritional rehabilitation. This was particularly true when there was pressure for beds. Consequently, infants could be discharged without their nutritional status being adequately assessed and recorded:

IDI03M
Respondent: Now what happens it’s the MO interns, you know like today we were having pressure, they wanted to discharge the patient, they say we have very many patients in the wards, so they say let’s discharge so that we have less patients, so that we may work. So, in a situation like that like today I have held two children under six, I have told them do not discharge until tomorrow we see how this child will be, because since yesterday they have not been adding weight, they didn’t add weight yesterday, today they have not added weight, so I told them let’s wait until tomorrow we see if there is a gain. So, in incident like that for now am here, when I go back maybe they are already discharged, because the MO wants to discharge and am telling him not to discharge, so when I go and most of the time am not in the ward, they discharge, especially with the MO interns.

By contrast, the status of clinical conditions of the infants at discharge were recorded in 133/167 (80%) of the infant’s pre-intervention and in 50/59 (85%) of the infants during the intervention.

b)

Follow-up

The guidelines recommend follow-up after discharge through community-based breastfeeding support programmes or out-patient clinics. In this audit there were no records of the linkage of caregivers to community-based breastfeeding support programmes, either prior to, or during the intervention. During the intervention period, follow-up with breastfeeding support at the hospital level was recorded for 29/59 (49%) of the infants.

In the pre-intervention interviews, nurses and nutritionists expressed concerns about the lack of support for effective post-discharge care and follow-up, reporting that many of the infants just ended up back in hospital:

IDI05M
Respondent: Yeah, coz the mother. .. maybe after we discharge the mother you give them the right information as per the guidelines but when they go at home, they do their own things, and the baby still comes here with the same, same acute malnutrition, yeah
Moderator: Does this happen so often?
Respondent: Yeah, it does, [[I: Mm]] mm, when you ask, they will always tell you ooh, I did give cow’s milk, because I was not able to produce enough, such like things, so it’s challenging, you can’t be able to implement the WHO because of that

However, during the intervention when the BFPS were in place, they sometimes took the initiative to follow-up on infants’ post-discharge to check on how they were progressing:

IDI04K
Moderator: How was this follow up as in when?
Respondent: Aah what she (BFPS) would do is if like she has identified the place where the clients come from she would even visit them to their homes to see how they are progressing and if not that they would be, they were given a returning date to come to the facility to see 202how they are progressing.

This follow-up, together with the higher success rate of relactation, gave the nurses and nutritionists the impression that the intervention decreased the number of days in hospital stay for the infants.

IDI03K
Respondent: First, it showed that when there is proper implementation or maybe a mother is well supported to breastfeed, then. .. we decrease that stay in the hospital.. .