“Prevalence, outcome and factors associated with dysglycemia among critically ill children presenting to Fort Portal Regional Referral Hospital: A cross sectional study”

Introduction: Dysglycemia has been shown to influence outcome among critically ill children. We aimed to determine the prevalence, outcome and factors associated with dysglycemia among critically ill children aged one month to 12 years presenting to Fort Portal regional referral hospital. Methods: The study employed a descriptive, cross-sectional design for prevalence and factors associated, and longitudinal observational study design to determine the immediate outcome. Critically ill children aged one month to 12 years were systematically sampled and triaged at outpatient department using World Health Organization emergency signs. The random blood glucose was evaluated on admission and at 24 hours. Verbal and written informed consent/assent were obtained after stabilization of the study participants. Those that had hypoglycemia were given Dextrose 10% and those with hyperglycemia had no intervention. Results: Of the 384 critically ill children, dysglycemia was present in 21.7% (n = 83), of those 78.3% (n = 65) had hypoglycemia and 21.7% (n = 18) had hyperglycemia. The proportion of dysglycemia at 24 hours was 2.4% (n = 2). None of the study participants had persistent hypoglycemia at 24 hours. The cumulative mortality at 48hours was 3.6% (n = 3). At 48 hours 33.2% (n = 27) had stable blood glucose levels and were discharged from the hospital. After multiple logistic regression, obstructed breathing (AOR 0.07(0.02-0.23), inability to breastfeed/drink (AOR 2.40 (1.17-4.92) and active convulsions (AOR 0.21 (0.06-0.74), were the factors that were significantly associated with dysglycemia among critically ill children. The results will guide in the revision of policies and treatment protocols to facilitate better management of children at risk of dysglycemia nationally. Conclusions: Dysglycemia was found to affect one in five critically ill children aged one month to 12 years presenting to Fort Portal Regional Referral Hospital. Dysglycemia outcomes are good with early intervention.

accidental poisoning from hypoglycemic drugs have all been found to be associated with hypoglycemia (7,8).
A study by Mbabazi et al 2011 in ACU, Mulago reported 5.9% prevalence of hypoglycemia and 6.8% of hyperglycemia, however, the prevalence found in this study was lower than what was previously reported by other studies done in the same setting (9). In the tropics, children are particularly prone to developing hypoglycemia and hyperglycemia in a wide variety of diseases, often related to childhood malnutrition, endemic malaria, or infectious diseases (5,10).Therefore, this study aimed to determine the prevalence, outcome and factors associated with dysglycemia among critically ill children aged one month to 12 years presenting to Fort Portal regional referral Hospital.

Aim of the study
To determine the prevalence, factors associated and immediate outcome of dysglycemia among critically ill children aged one month to 12 years presenting to Fort Portal regional referral hospital.

Study design
A descriptive cross sectional study design was employed to determine the prevalence and factors associated with dysglycemia among critically ill children and with an additional assessment of the immediate outcome.

Study setting
Triage of participants was done at OPD using WHO emergency signs and the study was conducted on Paediatric ward of Fort Portal Regional Referral Hospital. This hospital is locally known as The Paediatric ward has a capacity of 62 beds with an average admission of 10 to 20 patients per day. Anecdotally mortality on Paediatric ward is estimated between zero to four per day as a result of different factors. The hospital has a main laboratory that is open throughout the day and has another functional laboratory handling emergencies during the day and night. Laboratory tests routinely done are blood smears for malaria parasites, complete blood counts and hemoglobin concentration. Random blood sugars are not routinely done as a point of care test. In case random blood glucose levels cannot be ascertained, critically ill children with suspected hypoglycemia are usually treated empirically with intravenous dextrose 10%. The commonest conditions seen at the emergency department of OPD include severe malaria, severe pneumonia, severe anemia, and severe dehydration among others.
The pediatric Out Patient department of FPRRH runs throughout the entire week. The patients are seen by the medical officer, Intern doctors and clinical officers. The general paediatric clinic runs every Tuesdays in OPD department, in addition to the fore mentioned there is a pediatrician.
Transfer of patients from OPD to Pediatric ward is always done immediately after clerkship and giving emergency drugs as needed. Pediatric ward has a high dependence room with three beds where critically ill children are admitted. It also has a different ward for malnourished children that belongs to Paediatric ward. Additionally, the ward has 2 government employed pediatricians and two pediatricians from KIU, 13 nurses, three medical officers, four intern doctors, two intern nurses and two cleaners. All the staff for this study were trained in triage before commencing the study. Paediatric ward and OPD were selected for the study because all critically ill patients receive their initial treatment from these two departments and the ward is covered by doctors 24 hours.

Target population
All critically ill children aged one month to 12 years who presented to Fort Portal Regional Referral Hospital.

Accessible population
Children aged one month to 12 years who presented to Fort Portal Regional Referral Hospital between February 2020 and April 2020 and fulfilled the study inclusion criteria.

Actual population
These included all critically ill children who presented to Paediatric Ward and OPD of Fort Portal Regional Referral Hospital during the study period.

Inclusion criteria
a. Children aged one month to 12 years.
b. Children that presented to Fort Portal Regional Referral Hospital with signs of critical illness (WHO emergency signs).
c. Participants with Verbal and written informed consent by the caretakers and assent for children aged eight to 12 years.

Exclusion criteria
Children with a known history of diabetes mellitus, infusion of dextrose containing fluids up to two hours prior to admission, those who had received Dextrose 10%, intake of steroids within 72 hours of admission, children with a known history of hypoglycemic disorder and children who die before getting treatment at admission.

Urgent evaluation and procedures
The recruitment of study participants was done throughout the day and night. Triage was done at OPD of Fort Portal Regional Referral Hospital by the research assistants using WHO emergency signs after obtaining a verbal consent from the patients/guardians. As soon as critically ill children reached the Paediatric ward, a random blood glucose was done, the results were documented, thereafter the findings were shared with the clinical team on Paediatric ward. Those found to be hypoglycemic together with the staff on duty were given Dextrose 10%. After 20 minutes a repeat RBS was done. The participants that were found with hyperglycemia had no intervention.
If a patient declined consent standard of care was initiated immediately according to WHO protocol (1). Once the participants had stabilized on the pediatric ward, the research assistants and/or the principal investigator gave a comprehensive explanation to the parents or guardians of the selected participants, elaborated the purpose of the study, benefits, risks and thereafter request for their participation was sought. Once children eight to twelve years were stable enough and had understood the purpose of the study, they were requested to sign an assent form.
The eligible clients who agreed to participate signed a consent form by writing the initials of their names, signature and/thumb print. In case the guardian/participant could not write, in the presence of the witness a thumb print was considered. Each participant was assigned a study number and the file was labelled with a sticker to enable easy identification of the participants during the study period.

Measurement of blood glucose level
After arrival on the ward, a verbal consent was sought if they agreed to the consent, 0. 6 mL of blood was collected by the investigators through a finger prick to measure the blood glucose concentration using Accu-Chek Performa glucometers from ROCHE Laboratories. A quality control by Accu-Chek control solutions was performed after opening a new box of test strips, when the container was left open and when the strips were subjected to low or extreme temperatures. A range outside 1.7 -3.4 mmol/l was considered out of range. Blood glucose concentrations were recorded in mmol/l (conversion to mg/dl by multiplying by a factor of 18).

Follow up after admission
The enrolled dysglycemic participants had a repeat RBS at 24 hours. The dysglycemic arm was also followed up at 48 hours to determine the proportion deaths and discharges. Normoglycemic study subjects were not followed up. The 24 hours was arrived at because previous studies done to evaluate inpatient dysglycemia showed that mortality was especially high 19/33 (57.6%) during the first 24 hours of admission (43)

Data management
Data was recorded using structured pre-coded case record forms. Completed files were checked for completeness and accuracy, then stored in a lockable cabinet. All complete data was entered into an electronic database using Epidata version 3.1 software package with in-built quality control checks. The computer used for data storage and analysis was a password-protected computer and was only accessible to the PI and later, the statistician. The data was double entered and validated by the principal investigator. The final data were backed up and exported to STATA version 15 for analysis. Chi square or Fisher's exact tests was used to compare categorical variables as appropriate. All data was treated with confidentiality.

Descriptive statistics
To achieve objective one of this study which seeks to determine the prevalence of dysglycemia among critically ill children aged one month to 12 years presenting to Fort Portal Regional Referral Hospital, data was analyzed using descriptive analysis thereafter was summarized using proportions and percentages. Continuous variables were summarized using medians (IQR) and categorical data were summarized using proportions and percentages. To further achieve objectives 2 and 3 data was summarized using tables and figures.

Bivariate variables
Logistic regression was performed individually to assess the relationship between the outcome (dysglycemia) and each independent variable. The outcome was classified as: hypoglycemia: RBS ˂ 2.5 mmol/l (45 mg/dl) in well-nourished children or 3mmol/l (˂ 54 mg/dl) in severely malnourished children and hyperglycemia: RBS over 8.3 mmol/l (≥150 mg/ dl). Independent variables with P-value ≤ 0.2 were considered for the multivariate analysis.

Multivariate variables
At multivariable analysis, a multivariable logistic regression model was used to obtain factors that were independently associated with dysglycemia at p<0.05 and 95% confidence interval.
Those factors with a P-value of less than 0.05 were considered to be statistically significant factors.

Description of the study population
The study was conducted between February 2020 to April 2020. Four hundred and three (403) critically ill children presented to OPD of Fort Portal Regional Referral Hospital during this period.
Of these, 393 children were screened for eligibility using WHO emergency signs while at OPD, and random blood sugars were taken off from 384 critically ill children immediately after arrival on Paediatric ward where stabilization is routinely done (Figure1).

critically ill children presented to the OPD of Fort
Portal RRH 393 were screened for eligibility using WHO emergency signs while at OPD 10 critically ill children died on admission before any intervention.
384 were enrolled into the study.

Clinical characteristics of the study participants
Ninety-three of (24.2%) had severe anaemia, of those 40(10.4%) had a hemoglobin level of ˂5g/dl.

Bivariable analysis of clinical characteristics associated with dysglycemia.
None of the socio-demographic characteristics were significantly associated with dysglycemia at bivariable analysis. However, clinical characteristics that were significant at this level included: obstructed breathing, lethargy, active convulsions, fever, severe pneumonia, inability to feed and positive HIV status and these were considered for multivariable analysis. Other findings are shown in table 6 .

Factors associated with dysglycemia at multivariable analysis
As shown in Table 7, critically ill children with inability to feed were at risk of getting dysglycemia whereas obstructive breathing and active convulsions were protective in this study.

Multivariable analysis of clinical characteristics associated with hypoglycemia among 384
critically ill children presenting to Fort Portal Regional Referral Hospital between February 2020 to April 2020.

NOTE:
No factors were found to be significantly associated with hyperglycaemia in this study.

Introduction
This was a descriptive cross-sectional study design that set out to determine the prevalence, outcome and factors associated with dysglycemia among critically ill children aged one month to 12 years presenting to Fort Portal Regional Referral Hospital between February 2020 to April 2020.

Prevalence of dysglycemia among critically ill children.
Overall, the prevalence of dysglycemia among critically ill children was 83(21.7%). This translates to one in five critically ill children with dysglycemia among children presenting to Fort Portal

Regional Referral Hospital
The most prevalent dysglycemic anomaly was hypoglycemia of 65(78.3%) compared to hyperglycemia of 18(21.7%). The prevalence of dysglycemia increased with decreasing age. The possible reasons for the high prevalence of hypoglycemia in the study could be due to various pathophysiological mechanisms like poor oral intake, increased losses in vomiting, diarrhea and poor glucose reserve mechanisms for critically ill children that may not allow them to fast for longer periods as adults. Studies have shown that children have a limited tolerance to fasting because glycogen storage capability is limited and therefore, they are only able to maintain normal plasma glucose levels for a fasting period of 12 hours (5, 15).
These findings are comparable to those of Soro-Coulibaly et al 2018, in Côte d'Ivoire that found the prevalence of dysglycemia at 19% (12). The close similarity could be due to both studies used a descriptive, cross sectional study design, recruited critically ill children in relatively similar settings (referral hospitals) and both studies excluded critically ill children that had diabetes and those that had received dextrose within 2 hours prior to admission. Nonetheless, the observed dysglycemic rate is lower than that reported in Madagascar by Sambany et al 2013, of 34% (5).
The discrepancies may be explained by variations in thresholds used to define hypoglycemia that is ˂ 2.2 mmol/l for their study Vs ˂ 2.5mmol/l for this study, different age ranges of above 12 years for their study participants and longer duration of five months for their study compared to three months for this study.
In contrast, Mbabazi et al 2011, observed a lower dysglycemic rate of 12.7% among critically ill children as compared to this study (2). These variations could be attributed to the shorter period of the study, different cutoffs for blood glucose values and relatively different study settings.
Additionally, the later study was done in urban national referral center where children may have received therapy in previous centers as compared to this study that was conducted in a hospital that predominantly serves a rural community where children are brought from far off facilities with no standardized clinical services provided enroute to the hospital.
There are no well-defined criterion for diagnosing hyperglycemia at admission in non-diabetic critically ill children (5,44). However, the prevalence of hyperglycemia of 21.7% in this study was much higher than that reported in the study by Soro-Coulibaly et al 2018 of 11.7% in Côte d'Ivoire (12). The differences in the findings could be because their study was conducted during the pre-COVID 19 era where patients had easy access to medical facilities before the onset of stress hyperglycemia unlike this study that was done during the COVID 19 era with stricter transport restrictions and its presumed that with this delay, critically ill children could have accessed care when the stress hyperglycemic events had already set in. However, the findings are closely similar to the 23.9% in Ethiopian study reported by Sime et al 2021. The possible explanation could be because both studies enrolled the study participants in three months and also used similar blood glucose cutoff values for hyperglycemia of ˃ 8.3mmol/L (150mg/dL). In summary, different dysglycemic cutoff levels could have modified the results in this study in comparison to the work done by other authors (2,5,12).

Outcome of dysglycemia among critically ill children.
The study evaluated the proportion of dysglycemia at 24 hours and the proportion of deaths and discharges at 48 hours. Majority 81(97.5%) of the study participants that were dysglycemic at admission had normal blood glucose levels at 24 hours compared to two (2.4%) that had persistent dysglycemia. It's clear from this study that majority of the study participants had hypoglycemia.
The fact that all hypoglycemic children were given Dextrose 10% at admission, this could have improved the outcome.
In contrast, hyperglycemia could have resolved following the treatment of the primary medical condition since its believed to be induced by critical illness (5). Only a few studies have explored blood glucose levels beyond admission and throughout hospitalization. A study by Madrid et al 2017, in Mozambique recruited critically ill children less than 15 years with malaria and had continuous blood glucose monitoring up to 72 hours but found only one child among those with hypoglycemia on admission presented subsequently with hypoglycemic episodes (45). Their findings were closely similar to the findings in this study because both used similar blood glucose cutoff values for hypoglycemia and an intervention with dextrose 10% was the standard of care basing on WHO guidelines.
The cumulative mortality was generally low among the dysglycemic children considering that critically ill children even without dysglycemia have higher mortality than this. The hospital usually reports zero to four deaths in a day during the pre COVID era, since the study was conducted in the first phase of the lockdown this could have underestimated the mortality in this study. Additionally, delayed access to care due to transport lockdown and the hierarchy of accessing travel permit to the health facilities as a result of COVID 19 pandemic could have underestimated the cumulative mortality and hindered timely access to healthcare services. Also, the fact that ten critically ill children died before admission, this may not be the true prevalence as many unregistered deaths could have occurred in the communities, thus this underrated the mortality in the study as well. Additionally, study participants with hypoglycemia had an intervention of dextrose 10%, this could have modified the outcome.  (5,28). In the current study and in previously mentioned studies majority of the dysglycemic study participants were under 5 years, thus with the background evidence of reduced immune system in the setting of critical illness, this could have lengthened the period of hospital stay.

Factors associated with dysglycemia among the study participants
In general inability to feed increased the risk of dysglycemia whereas active convulsions and obstructed breathing were protective. No factor was found to be independently associated with hyperglycemia.
Critically ill children with inability to breastfeed and drink were 2.4 times more likely to get dysglycemia. The children that had longer duration of illness of above seven days and prolonged fasting of greater than 2 days were more predisposed to dysglycemia ( Table 7). The possible cause for the findings could be because these children are too ill to feed orally in the presence of loss of consciousness and anorexia induced by critical illness. Furthermore, prolonged fasting depletes the glycogen stores and heightens the risk of dysglycemia.
In addition, studies have shown that long travel to the health facility deprives critically ill children a chance to be fed along the way thus prolonging the fasting period (5,15). In this study 46% of the participants were from other neighboring districts. The findings of this study are higher than those of Sambany et al, 2013 that found 10.4% had this symptom (5). This could be because they excluded many patients in their study and also had a 24 hour break every after 48 hours of data collection, this could have underestimated the prevalence in their study.
Obstructed breathing was protective for dysglycemia in this study. The reason behind this could be because obstructed breathing is taken as a sign of critical illness in the community and children are brought to the health facility timely. Additionally, it is a common practice for children with obstructed breathing to have a nasal gastric tube inserted because of the feared risk of aspiration pneumonia. However, there are very few studies that have used this emergency sign at triage. A prospective observational study by Mawji et al in Kenya 2018 found obstructed breathing was not statistically significant in children aged two to 60 months seeking treatment for an acute illness (46). Nevertheless, they recruited critically ill children below the age of five years unlike in this study where study participants were recruited every day and included critically ill children from one month to twelve years.
Also, critically ill children with convulsions were less likely to be dysglycemic in this study. This could be because it's a common practice for children with convulsions to receive dextrose from health worker's pre-referral because convulsions are considered a sign of hypoglycemia and as well as an indicator of loss of consciousness. The dysglycemic findings of 5(6.0%) are much lower than those reported by Sambany et al, 2013 and Barennes et al 2016 that found 31.6% and 51.4% respectively had dysglycemia with convulsions (5,15). These variations could be due to the different thresholds that were used to classify dysglycemia.

Strengths of the study.
This is one of a few studies that have used the WHO cutoffs for hypoglycemia in both wellnourished and malnourished critically ill children. Furthermore, it's the first study in our setting that has been conducted from a peripheral health facility to evaluate the two disorders simultaneously (hypoglycemia and hyperglycemia).
Limitations of the study.
1. The study did not specifically screen participants for Diabetes mellitus and Diabetic ketoacidosis, however, the history and clinical evaluation helped to exclude their clinical presentation for the children that had hyperglycemia.
2. Additionally, the study was conducted during the first phase of the lockdown due to COVID 19 pandemic (March to April 2020), this therefore may have underestimated the mortality and the prevalence rate of dysglycemia.
3. Furthermore, due to vehicle restrictions especially motorcycles which are the commonest mode of transport to facilities, access to healthcare was limited and thus fewer children from distant places were reaching the referral centers like Fort Portal Hospital, this therefore, could have underrated the prevalence of dysglycemia during the study period. 4. Likewise, children from distant locations from the hospital, had higher possibilities of delay, often associated with poor health outcomes, and these might have been underrepresented in the study leading to the low prevalence of dysglycemia and a low cumulative mortality. 5. Also, comparisons with other studies are difficult due to the different thresholds used for hypoglycemia and hyperglycemia. In the current study this was overcome by use of the WHO cutoff values for hypoglycemia and also by using of International Society for Pediatric and Adolescent Diabetes (ISPAD) guidelines for the cutoff of hyperglycemia.
6. Lastly the study was unable to assess the long term outcome of dysglycemia among critically ill children.
On the other hand, limitations 2, 3, and 4 were difficult to overcome because we had no control on the government lockdown guidelines.