Incidence and predictors of diabetic ketoacidosis among children with diabetes in west and east gojjam zone referral hospitals north west Ethiopia, 2019.

Recurrent diabetic ketoacidosis in patients with already diagnosed diabetic mellitus remains a relevant problem in pediatric with an incidence of 1–10% per patient. Children may die due to cerebral edema and had a signicant mortality (24%) and morbidity (35%). Objective We assess incidence and predictors of diabetic ketoacidosis among diabetic children at and Methods An institution-based retrospective follow up study was conducted on children who were registered from January 1, 2014, to January 1, 2019. Epi data version 3.1 & Stata 14 were used for data entering and analysis respectively.


Introduction
Diabetic Ketoacidosis (DKA) represents a state of acute metabolic stress, when the body suffers due to absolute or relative insulin de ciency for metabolism of glucose [1] . Globally, recurrent-DKA in patients with already diagnosed DM remains a relevant problem in pediatric. The risk of DKA in established diabetes is 1-10% per patient per year in children [2] . Also, the incidence of DKA varies with geographical variation. The incidence of DKA at diagnosis of T1DM in the UK was 25% [3] and in a study at Iran in 2005-2013, the incidence of DKA was 55.5% [4] . In Africa the data on the incidence of DKA is scarce, but some studies report the frequency of DKA in north-western Nigeria 62.2% [5] , in South Africa 69.8% [6] . Also in Ethiopia, the incidence of DKA is not well studied but a study in Addis Ababa showed that the prevalence of DKA is 35.8% [7] . DKA is the most common cause of death among children and is associated with an increased risk of cerebral edema and cognitive de cits [8,9] . A study showed that the risk of developing cerebral edema was 12.4 per 1000 episodes of DKA which was higher than Non-DKA DM patients (3.8 per 1000). It had a signi cant mortality (24%) and morbidity (35%) [10] . DKA has a crisis in terms of health care costs, missed work and school days. One DKA-related hospital admission ranges from US$4125 to US$11 196 cost was observed [11] . Lack of proper public health education [12] and challenge in self-care because of they are children may have contributed to a high incidence of DKA in Ethiopia. Thus, Self-care of glycemic control can have a great impact on DKA existence in many patients and contribute to most of the increased morbidity and premature mortality [13,14] . Even, many patients seek alternative treatments such as consulting traditional healers, using herbal remedies [15] , prayers and rituals that further complicating the disease process [16] . Many studies showed that being a younger age (< 5 yrs.), infection, lower socioeconomic status and lower parental education contributes to an increased risk of DKA [17] .

Study setting
The study was conducted in the two referral hospitals (Debre-Markos referral hospital and Felege-Hiwot referral hospital) of East and West Gojjam zones in Amara regional state North West Ethiopia. These hospitals serve for more than 3.5 million and 5 million population in their catchment area respectively. Apart from other services, both referral hospitals provide Diabetic treatment services.

Study design
A ve years institution-based retrospective follow up study was conducted Inclusion and Exclusion criteria Children age less than 15 years old and diagnosed with DM with follow up care from January 1, 2014 to January 1, 2019 were included and Child who was developing DKA at the rst diagnosis of DM and charts which was lost during the study period was excluded from the study. Data collection total DM caseload was assessed in the data base on the registered follow up chart/form from the discharge catalog of admitted patient's pediatric ward, emergency and OPD from January 1, 2014 to January 1, 2019. Then medical registration number (MRN) of all diabetic pediatric patients was sorted. After this, simple random method was applied to select the study subject. Finally, trained BSC nurses working at diabetic clinic collect the data by using checklist that measures the socio-demographic characteristics, disease factors, treatment factors and information on glycemic control of the child.

Ethical consideration
After the approval of the proposal, ethical clearance was obtained from the school of nursing and midwifery, college of health sciences, Addis Ababa University. Then permission letter was written to Debre Markos and Felege Hiwot referrals hospitals for data collection purposes. The con dentiality of the data had been kept.
We had taken permission from hospitals medical directors and due to data had taken from chart review only, informed consent was not required

Statistical analysis
The collected data were coded and entered into Epi data version 3.1 and cleaned and transferred to Stata version 14 for further analysis. The incidence rate of DKA was estimated per 100 DM children per month. The Kaplan Meier estimator was used to estimate median time to develop DKA during the treatment period and log-rank tests, to compare survival curves. The predictors of DKA were analyzed by the cox proportional hazard model with hazard ratio, 95% CI. The statistical test was considered signi cant at P value of less than 0.05. Covariates and proportional hazard assumptions were checked using log-log plot and goodness of t by Schoenfeld residual test.

Socio-demographic characteristics
Out of 376 children's clinical pro le reviewed, 354 were enrolled in the study. The rest of the sample 22 (5.8) was incomplete data. From 354 children, more than half 159 (55.1%) were males and more than half 189 (53.4) of them were from a rural area. The mean age of the children at the time of DM diagnosis was 8.21 years with SD 3.94 years (Table 1).   Fig. 4).
The median survival time for those who had a history of medication adherence was 44.3 months with 95% CI (36.9, 50.5) and the mean survival time for those who had a history of medication non-adherence was 27.5 months with 95% CI (23.6, 33.9). The survival time difference between the groups was found statically signi cant with P < 0.001 (Fig. 5).
Predictor of Diabetic ketoacidosis   [18] , international society of pediatric and adolescent diabetes 2014 report which was 1-10% per patient per year [2] and Austria which was 8.4 to 18.4 per 100,000 per year [19] . These controversies might be due to in studies done at Saudi Arabia the age of participants was 14 to 40, but in this study, participants were < 15 years old and study done in Austria was population-based study and for a long time (twenty years) ago.
Regarding cumulative incidence, this nding is consistent with studies done in north-western Nigeria which was 62.2% [5] and 55.5% in Iran [4] . However, this nding is much lower than studies done in Tikur-Abesa specialized hospital which was 80% [20] , 77.1% in Benin teaching hospital [21] . This discrepancy might be due to methodological differences, in those two studies the incidence was calculated based on newly diagnosed DM, but in the current study, the incidence was estimated based on known diabetes diagnosed children. This nding is much higher than studies done in the US which was 25.5% [22] , 40.3% in Italy [23] , 27% in New Zealand [24] in Poland 28% [25] , 40% in southern Iraq [26] . This discrepancy might be due to methodological differences, in the US the incidence was calculated based on newly diagnosed DM, in others the incidence was calculated based on both newly diagnosed DM and known diabetes children but in the current study, the incidence was estimated based on known Diabetes. Another explanation might be due to different population characteristics and qualities of health care service.
Children age < 5 years were more likely to develop DKA compared to age > 10 years. This is consistent with other previous studies conducted in the US [22] , Italy [23] , Southern Iraq [26] . This might be the age group < 5year might be more dependent on their caregiver thus not take on time and not collaborate to take medication. Children who have medication non-adherence were more likely to develop DKA as compared to adhere to medication. This nding is supported by other previous studies conducted in sub-Saharan Africa [27] , north India [28] , Saudi Arabia [29] , New Zealand [24] , Southern Iraq [26] . This might be since DM is chronic illness after taking medication symptoms may disappear for some times so, the children may not take their medication on time. Also, children having a history of inappropriate insulin storage at home were more likely to develop DKA. This might be most of the participants were from rural areas and may not have an appropriate storage material like refrigerator and may have inadequate knowledge about the storage of insulin. Children who have preceding gastroenteritis and upper respiratory tract infection at the time of DKA development were more likely to develop DKA as compared to those who have not. This is supported by study in Tikur Ambesa hospital [32] , Nigeria [33] , sub-Saharan Africa [27] , Malaysia [32] , north India [28] and Saudi Arabia [29] . This might be infection can cause high levels of counteracting hormones which triggering an episode of DKA

Limitations
Since the data were collected from medical records, patients' charts lost and incomplete data were found.
These may affect the outcome of the study. Also, this study did not include the recurrence of diabetic ketoacidosis (trend) and the lack of some variables like parental factors that can't be addressed through card review

Conclusion And Recommendation
In conclusion, the incidence of DKA in known diabetes children was found to be high. Children who have age < 5-year, medication none adherence, inappropriate insulin placement at home, presence of upper respiratory tract infections at the time of diabetic ketoacidosis development and presence of preceding gastroenteritis were predictors of DKA development at East and West Gojjam zone referral hospitals, Northwest Ethiopia. Therefore, Diabetic care clinics needs to be strengthened as well as assessing, close monitoring and strengthened diabetic education for patients as well as for caregiver/families/ should be given to children who have age < 5 years, non-adherence to medication, inappropriate insulin placement at home, and upper respiratory tract infection and gastroenteritis. Finally, we recommend those variables that cannot be assessed through card review and recurrence diabetic ketoacidosis will be investigated with another study design.

Declarations
Ethical approval and consent to participate After the approval of the proposal, ethical clearance was obtained from the school of nursing and midwifery, college of health sciences, Addis Ababa University. Then permission letter was written to Debre Markos and Felege Hiwot referrals hospitals for data collection purposes. The con dentiality of the data had been kept. We had taken permission from hospitals medical directors and due to data had taken from chart review only, Availability of data and materials All materials and data are available in the man author without any restriction.

Consent for publication
Not applicable