This was a cross-sectional study with a follow up component carried out from August 2015 to March 2016.
The study was conducted at the MNU of Mulago National Referral Hospital in Kampala, Uganda. Mulago hospital has a 1500 bed capacity and comprises of two main divisions; − New Mulago and old Mulago. MNU is found in old Mulago where the main paediatric wards are located. MNU is the nutrition ward in Mulago national referral hospital and receives patients from all over the country. It admits all children older than 6 months with severe acute malnutrition who have associated complications and infants less than 6 months who have features of severe acute malnutrition (nutritional oedema or Z scores < -3SD) including those with ineffective breastfeeding, medical complications or other social issues that compromise the nutrition of the infant. The children are admitted through the emergency paediatric unit (Acute Care Unit) where they receive the initial care in the first 24 h. Later, those with severe acute malnutrition are transferred to MNU. Most come from the central region while the rest are referred from hospitals and health centres upcountry. The MNU has 80 beds and on average it admits about 120 patients per month. The unit has 5 paediatricians and 1 medical Officer. There are 17 nurses and, of these, 4 are enrolled nurses, 9 are nursing officers and 4 are nursing assistants. There are 3 nutritionists attached to the unit.
Children on the ward receive feeds at stipulated intervals. In the first 48 to 72 h (critical phase), they receive feeds (F75 for those > 6 months with severe acute malnutrition) and specially diluted therapeutic milk (SDTM) for infants < 6 months with severe acute malnutrition who are not breast feeding adequately) 2 hourly. When children are clinically stable, an appetite test (for those > 6 months) is done to identify children to transit to “Ready to Use Therapeutic Feeds” (RUTF). Specially Diluted Therapeutic Milk (SDTM) is made from F100 and contains 69Kcal/100 ml. It is given to infants less than 6 months who have severe acute malnutrition and are unable to get sufficient breast milk. The critical phase usually lasts 7 to 10 days. Children who pass the appetite test are then transferred to the rehabilitation phase for a few days where the care takers are also taught how to prepare nutritionally rich food for use at home. Children who are stable and feeding well are then transferred to the Outpatient therapeutic Clinic for outpatient management which has been demonstrated to be a better environment for rehabilitation of these children.
We included children aged 1–60 months admitted to MNU for SAM and who had been on the ward for less than 3 days.
We excluded children who had no care givers to give information and children with a known or suspected diagnosis of diabetes mellitus.
Sample size calculation
The sample size was calculated using Kish Leslie formula for prevalence studies .
N = ZP (1-P)/D2- where N is the required sample size, Z is the value corresponding to 95% confidence interval, D is the absolute error equal to 0.05 and P is the prevalence of hyperglycaemia among children with SAM  equal to 11%.
A sample size of 165 was obtained for the first objective assuming 10% for incomplete data.
The sample size for factors associated with hyperglycaemia was calculated using the formula for proportions. N = (Zα/2 + Zβ)2 ∗ (p1(1 − p1) + p2(1 − p2))/(p1 − p2)2.
Where N = the required sample size, P1- the prevalence of stress hyperglycemia among children with hyperthermia (a factor associated with hyperglycemia) was assumed to be 9.3%and P2- prevalence among those without hyperglycemia was 2.8% . Hence, a sample size of 208 was attained. Since this was a bigger sample size, it was used for the study.
The study had 2 research assistants a nurse and a medical officer and one of us (AT). The nurse was responsible for screening the children and obtaining informed written consent from the care givers and also doing the random blood sugar tests and Rapid diagnostic test for malaria. The medical officer carried out physical examination, anthropometry, collected the blood samples and delivered them to the laboratories. AT was in charge of the whole process and was involved in patient recruitment and care.
All children admitted to MNU were screened for eligibility every morning between 8 and 10 am. Informed oral and written consent was obtained from the care givers of the eligible children. A random blood sugar was done using a finger prick at least 2 h from the previous therapeutic feed. The study assistant was always on the ward before 8 am to screen the patients for eligibility and be able to do the random blood sugar measurements just the children received the 8 am feeds. The study assistant also took note of whether the child had received intravenous dextrose and how long they had received it before doing the random blood glucose level. From the data collected, all the children who received 10% dextrose had received it more than 12 h prior to blood testing. Children with hyperglycaemia were referred to the attending team on the ward for follow up. An interviewer administered questionnaire was used to take the history from care takers about the patient such as the demographic information, clinical symptoms and information about the care takers. Patients’ charts were reviewed for treatment received such as antibiotics and intravenous fluids/ dextrose. The final outcome was obtained by reviewing the patient charts on death or discharge from the ward.
Physical examination was done for all the enrolled children and this involved doing anthropometry, general and systemic examination.
Anthropometry was done by taking the weights of the children using a digital scale. The height was measured using a stadiometer for children above 24 months who could stand. For the younger children less than 24 months who could not stand, length was measured using the stadiometer with a child in supine position and stabilised in a straight position with the help of the care taker. The length was reported to the nearest 0.1 cm. Children who were more than 24 months but could not stand had their length measured and an adjustment of 0.7 cm added to the length so as to get an equivalent of their height if they were able to stand.  Their nutritional status was determined using the WHO Z score charts taking into the count the presence or absence of nutritional oedema. Children were considered to be SAM if their Weight/height Z score was ≤ -3SD or had nutritional oedema or both.
A systemic physical examination was carried for clinical signs pointing to any co-morbid conditions such as presence of pallor, jaundice, respiratory distress, tachycardia, signs of shock, lethargy, alertness and body temperature was measured. The findings were summarised in the questionnaire.
Collection and analysis of blood samples
An ‘On Call Plus’ glucometer with matching test strips; an international brand from the ACON Laboratories in the United States of America was used for the blood sugar tests. The glucometer was calibrated using a control solution every after 20 tests or whenever a different batch of test strips was used.
The random blood sugar level was done using a finger prick. The left index finger was used for the test. The side of the finger was swabbed with sterile alcohol swabs containing 70% alcohol and a prick was made with a sterile lancet then a drop of blood was put onto the test strip that was already mounted into the glucometer. A reading of the blood sugar was then obtained after 30 s and recorded in mg/dl. The same finger prick used for random blood sugar was also used to obtain 2 drops of blood to do the RDT.
A blood sample for the other tests was collected from the children using peripheral veins.. The area was first swabbed with sterile cotton swabs containing 70% alcohol and then allowed to air dry for about 60 s. A tourniquet was then applied for a brief period of about 2 min then a sterile autodestruct syringe was used to collect about 5mls of blood. One millilitre(ml) was put into an EDTA container for CBC, 2mls in a general container for serum albumin, urea, creatinine and electrolytes, 2mls was used for blood culture. For all children below 18 months of age, the HIV serology of the mother was done by taking off 1 ml of blood from the mother, if the results came back negative; the child was considered HIV negative. If the mother’s test came back positive, 1.5mls of blood was taken off from the child and sent to Baylor-Paediatric Infectious Diseases Clinic (PIDC) for HIV DNA-PCR to get the HIV status of the child. For children above 18 months of age, HIV serology was using the rapid testing kits (Determine and Stat pak). Children or care givers who were HIV positive were connected to Baylor-PIDC for specialised care. The samples for CBC, electrolytes, serum albumin, urea and creatinine were analysed in the Mulago National Referral hospital laboratories. The CBCs were analysed in the haematology laboratory using the SysmexXs 800i machine. The blood cultures were done in the Microbiology laboratory. The electrolytes were analysed in the clinical chemistry laboratory using a Cobas R 6000c501clinical chemistry analyser (Roche Diagnostics, Indianapolis, IN). All the results were recorded in the questionnaires. A structured questionnaire was used to collect information about the patients. It was pretested on 5 patients before the actual data collection to determine its applicability and suitability. The pretesting was done by one of us (AT). The study instrument was used to collect data on the socio-demographic characteristics, patient’s history, physical examination, laboratory results and to record the final outcome of the study participants.
Patient management and feedback
Results of the patients were promptly availed to the clinical team on the ward. Where an urgent intervention was required, the study team initiated management in consultation with the attending clinicians. Currently, there are no guidelines on the management of hyperglycaemia in non-diabetic children because the burden and clinical significance of the problem among children with severe malnutrition in our setting is not known due to limited studies in the field. Thus, for children who had hyperglycaemia, the results were communicated to the attending team for follow up.
The primary outcome variables included a random blood sugar more than 150 mg/dl and death. Independent variables include patient’s general clinical condition, type of feed given (F75/SDTM), co-morbid conditions such as fever, convulsions, septicaemia, HIV serology of the patient and treatment received such as antibiotics and intravenous fluids (Dextrose).
Data management and analysis
Data was entered into EPI DATA version 3.1 and exported into STATA version 12. The data was checked for errors, cleaned and coded in preparation for analysis. Descriptive analysis was done using medians and inter-quartile ranges. Frequencies were also tabulated for other demographic and background characteristics. To answer objective 1, a new binary variable for hyperglycaemia was created. A patient was considered to have hyperglycaemia if he or she had a random blood sugar greater than 150 mg/dl.
To determine the factors associated with hyperglycaemia, bivariable analysis was done by cross tabulating the hyperglycaemia variable with the demographic characteristics, clinical symptoms of the children with severe malnutrition on admission to MNU, feeding history and co-morbid conditions of the children with SAM admitted to MNU and other factors. A simple chi square test was done at bivariate level and odds ratios and p-values were tabulated.
Factors with a p-value less than 0.05 were considered to be statistically significant.
For determination of the final outcome of the study participants:
A binary outcome variable was created (0 = no death, 1 = death). A cross tabulation was done for the binary outcome variable of death with that of stress hyperglycaemia, to get the frequencies of those with and without stress hyperglycaemia who died. Then bivariate analysis was done, to compare mortality between those with stress hyperglycaemia and those without.
Approval for the study was obtained from the Department of Paediatrics and Child Health of Makerere University; the School of Medicine Research and Ethics Committee (SOMREC) and the National Council for Science and Technology. Informed written consent was obtained from the care givers of the children.