Abstract
Background:
Diabetic ketoacidosis, a severe complication of type 1 diabetes, remains a major cause of morbidity and mortality in children, particularly in resource-limited settings such as Ethiopia. Despite its high burden, data on recovery time and predictors of diabetic ketoacidosis in this population are scarce. Therefore, this study aimed to assess the time to recovery from diabetic ketoacidosis and its predictors among children with diabetic ketoacidosis at selected governmental hospitals in Addis Ababa, Ethiopia.
Methods:
A 5-year retrospective follow-up study was conducted among 391 hospitalized children with diabetic ketoacidosis in selected governmental hospitals in Addis Ababa, from January 1, 2018 to December 30, 2022. Participants and hospitals were selected using a simple random sampling method. A structured data collection checklist was adapted from similar studies and modified. The data were checked for completeness and entered into Stata version 16 for analysis. Variables with p-value < 0.05 in the multivariate Cox proportional hazards model were considered significant predictors of the time to recovery from diabetic ketoacidosis.
Results:
A total of 423 records of children with diabetic ketoacidosis with 391 (92.4%) response rates were included in the final analysis. Out of these, 370 were recovered and discharged. The remaining 21 cases were censored. The overall median time taken to resolution from diabetic ketoacidosis was 27 h, with an interquartile range of 16–38. Diabetic mellitus history (Adjusted Hazard Ratio (AHR) = 0.41, 95% CI: 0.30–0.56), severity of diabetic ketoacidosis (AHR = 2.35, 95% CI: 1.34–6.1), presence of comorbidity (AHR = 1.76, 95% CI: 1.37–2.26), and blood sugar level (AHR = 0.61, 95% CI: 0.39–0.96) were all independent predictors of time to recovery from diabetic ketoacidosis.
Conclusion and recommendation:
The median diabetic ketoacidosis recovery time was 27 h. Key predictors included admission glucose, comorbidities, diabetic ketoacidosis severity, and diabetes history. Early diagnosis, thorough assessment, and optimized management are crucial to reducing risks and improving outcomes.
Keywords: Diabetes, diabetic ketoacidosis, time to recovery, children, Addis Ababa, Ethiopia
Background
Diabetes mellitus (DM) is a prevalent, persistent metabolic disorder characterized by elevated blood glucose levels as a key biochemical feature due to impaired insulin secretion, sensitivity, or both. Type 1 diabetes (T1D), an autoimmune condition resulting in the destruction of pancreatic beta cells, is the most common form of diabetes in children and adolescents. 1 Globally, over 1.2 million children and adolescents under the age of 20 live with T1D, with approximately 108,200 new cases diagnosed annually in children under 15 years. In Africa, 59,500 children and adolescents have diabetes, and 19,700 are newly diagnosed each year. 2
Diabetic ketoacidosis (DKA) is a life-threatening complication of T1D, characterized by hyperglycemia, metabolic acidosis, and ketosis.3,4 It results from absolute or relative insulin deficiency, which can occur at the onset of T1D (primary DKA) or in individuals with established T1D due to insulin omission or inadequate management (secondary DKA). 5 While primary DKA accounts for the majority of cases globally, secondary DKA represents less than 15% of cases in high-income countries. 6 In sub-Saharan Africa, however, the prevalence of DKA at T1D diagnosis is exceptionally high, ranging from 70% to 80%. 7
In Ethiopia, the prevalence of DKA among children with T1D varies widely, with studies reporting rates of 35.8%–78.3% in newly diagnosed DM children and 21.7%–64.2% in established cases.8,9
The time to recovery from DKA varies widely, ranging from 24 h to 8 days.10,11 Studies have also shown that the time to recovery from DKA is affected by different factors, including the severity of DKA, and biochemical parameters such as blood glucose, serum creatinine, and serum electrolytes.12,13
DKA is the leading cause of diabetes-related mortality and mortality in children with T1D and carries a significant risk of potentially fatal complications such as cerebral edema.3,14 Moreover, in emerging nations, DKA-associated death rates vary from 6% to 24%, whereas in western nations it is 0.15% to 0.31%. 15 On the other hand, despite its significant financial impact on healthcare systems, DKA also places an additional financial burden on patients and their families. 16 Due to these severe sequelae, patients with DKA need to be closely monitored and receive sensitive, balanced therapy, most likely in an intensive care unit, with the aim of rehydrating, correcting acidosis, and disappearing of ketosis.12,17 The introduction of insulin to end ketosis and lower hyperglycemia, restoration of dehydration with intravenous (IV) fluids, and correction of electrolyte imbalances with electrolyte replenishment are all part of the management of DKA. 18
Globally, various strategies have been implemented to improve DKA outcomes, including diabetes self-management education, standardized DKA treatment protocols, and preventive measures aimed at reducing morbidity and mortality. 19 However, in Ethiopia, standardized guidelines for DKA management were not available until 2014, when the Federal Ministry of Health adopted treatment protocols from the International Society for Pediatric and Adolescent Diabetes and other international guidelines. Despite this advancement, challenges such as limited healthcare infrastructure, resource constraints, and inadequate access to pediatric intensive care units (PICUs) have contributed to persistently high mortality rates from DKA.9,20,21
Minimizing the recovery time from DKA is vital to reduce or prevent the risk of morbidity and mortality as well as decrease the high expenses associated with treating the severe long-term effects of cerebral edema and other complications that may arise throughout the course of treating DKA. In addition, in countries such as Ethiopia where there is a limitation of PICU access, it decreases avoidable PICU admissions due to DKA complications that occur after starting treatment at the pediatrics emergency unit. While previous studies in Ethiopia have focused on the prevalence and outcomes of DKA, there is limited evidence on the time to recovery and its predictors among children with T1D. This study aims to address this gap by assessing the time to recovery from DKA and identifying its predictors among children with T1D at selected governmental hospitals in Addis Ababa, Ethiopia.
Methods
Study design, area, and period
This study was a 5-year retrospective follow-up study conducted in Addis Ababa, which has a population of over 5.4 million people. Addis Ababa is officially divided into eleven sub-cities and 121 districts and is served by 13 government hospitals. Among these, six are governed by the Addis Ababa health bureau, five are federal hospitals, one is owned by the military, and one is run by the police force.
The study was conducted in five randomly selected governmental hospitals: Tikur Anbessa Specialized Hospital, Zewditu Memorial Hospital, Saint Paul Medical Millennium College Hospital, Saint Peter Specialized Hospital, and Yekatit 12 Hospital. The study involved a comprehensive review of the medical records of children aged 1 month to ⩽15 years who were diagnosed with T1DM and experienced DKA between January 1, 2018, and December 31, 2022.
Inclusion and exclusion criteria
Children aged 1 month to 15 years diagnosed with T1DM and admitted with DKA to the pediatric emergency room, pediatric ward, or pediatric intensive care unit between January 1, 2018, and December 31, 2022, were included in the study. Patients with incomplete medical records lacking essential clinical or laboratory data necessary for the analysis were excluded to ensure the reliability and completeness of the study findings.
Sample size determination
The sample size was calculated using the single-population proportion formula, based on the assumptions of a 95% confidence level (α/2 = 1.96), 5% margin of error, and a 50% proportion since there is no study done in Ethiopia regarding the time to recovery from DKA among children. Hence, after adding 10% for missing or incomplete data, the final sample size was 423.
Sampling technique and procedures
The study was conducted in 5 government hospitals in Addis Ababa, selecting 423 children diagnosed with DKA from January 1, 2018 to December 31, 2022. The hospitals were chosen randomly, and the total number of children admitted with DKA during this period was identified using a list of medical records of children with DKA recorded on the health management information system of each institution. The study involved a total of 1398 children admitted during this time. By proportionally dividing the total sample by each institution, 423 cards were sampled using a simple random sampling technique using a computer-generated method. Finally, data were extracted from the selected medical records.
Data collection tools, procedures, and quality control
A structured checklist was adapted to extract information from medical records, as it appears in similar studies in the literature.22–24 The checklist was prepared in English and consisted of four sections: sociodemographic characteristics, clinical parameters at admission, biochemical profile, and treatment-related factors (Supplemental file 1). The checklist template was created in Kobo Toolbox and shared via a mobile phone URL with five trained BSc nurses responsible for data collection. The time period from the initiation of fluid management to the absence of two consecutive ketones in the urine was considered the time to recover from DKA. 25 The information was extracted from both electronic and non-electronic medical records. Recurrent cases of DKA episodes were excluded from the study to avoid selection bias. Patient’s registration numbers were used to maintain anonymity. The checklist was pretested on 5% (21 charts) of the sample size in the Tirunesh Beijing Hospital to assess its consistency and completeness. The experts verified the validity of the checklist, and their suggestions and comments were incorporated. Data collectors and supervisors were trained for 1 day.
Operational definition
Event: Recovery from DKA.
Time to recovery from DKA: Time period in hours from which fluid management of DKA is started until the absence of two consecutive ketones in the urine.22,25
Censored: Children transferred to other facilities, discontinuation of therapy for DKA, or died for any reason before recovery from DKA, and children with unknown status.
Mild DKA: Patients who meet the DKA criteria and who are alert and oriented but fatigued with no or some dehydration sign. 12
Moderate DKA: A patient who meets the DKA criteria and has Kussmaul respiration, is lethargic, and exhibits shock or dehydration sign. 12
Severe DKA: Patient who fulfills the DKA criteria and exhibits Kussmaul’s or depressed breathing, and shows signs of shock or severe dehydration and sensorium depression to coma. 12
Data processing and analysis
Data were checked for completeness and imported to Stata version 16 for analysis. The clinical and sociodemographic characteristics of the study participants were described using descriptive statistics including median, percentage, and frequency. To calculate the DKA mean free survival time, the Kaplan–Meier survival curve was employed. The DKA-free survival time was compared with several categorical explanatory variables using the log-rank test, along with Chi-square and p-value. A Cox proportional hazard model with a hazard ratio and 95% CI was used to analyze the predictors of the DKA recovery time. Variables with a p-value of less than 0.25 in the bivariate analysis were added to the multivariate Cox proportional hazards model, and variables with a p-value of less than 0.05 in the multivariate model were considered significant predictors of the time to recovery from DKA. The proportional hazard assumption was assessed graphically using a log-log plot, and the goodness of the model fit was tested based on the Schoenfeld residual (pH test) test. The overall global test p-value was 0.0507. To calculate the recovery time, hours were used as the time scale, and the status of the study participants was dichotomized as either an event or censored.
Results
Sociodemographic characteristics of the study participants
Of the 1398 DKA admissions from January 1, 2018 to January 30, 2022, 423 charts were randomly selected for this study. Three charts were absent, and 29 charts were incomplete and excluded from this study. Three hundred ninety-one (92.4%) fulfilled the inclusion criteria and were included in the study. The median age of the participants was 8 years, with an interquartile range (IQR) of 3.5–11 years. More than half (56.3%) of the study participants were males (Table 1).
Table 1.
Sociodemographic characteristics of children admitted with DKA at selected public hospitals in Addis Ababa from January 2018 to December 2022 (n = 391).
| Variables | Category | Survival status | |
|---|---|---|---|
| Recovered, n (%) | Censored, n (%) | ||
| Age in years | Under 5 | 115 (94.3) | 7 (5.7) |
| 5–10 | 128 (92) | 11(8) | |
| >10 | 127(97.7) | 3 (2.3) | |
| Sex | Male | 206 (93.6) | 14 (6.4) |
| Female | 164 (96) | 7 (4) | |
| Residence | Addis Ababa | 263 (96.7) | 9 (3.3) |
| Out of Addis Ababa | 107 (90) | 12 (10) | |
| Family history of T1D | Yes | 138 (94) | 9 (6) |
| No | 224 (95) | 12 (5) | |
| Unknown | 8 (100) | 0 (0) | |
Clinical characteristics and patient baseline information
More than half (51.2%) of the participants admitted with DKA had no previous history of DM. The median duration of illness in previously diagnosed DM subjects was 2 years. Most of the participants (93.7%) admitted with known diabetes were on mixed (neutral protamine Hagedorn and regular) insulin. Noncompliance with insulin was the most common (51.3%) precipitating factor in known DM patients, followed by infection (26.2%). The most frequently diagnosed acute comorbid illness was respiratory tract infection (11.5%), followed by urinary tract infection (5.4%). Regarding the severity of DKA, 18.2% were severe (Table 2).
Table 2.
Clinical characteristics of children admitted with DKA at selected governmental hospitals in Addis Ababa from January 2018 to December 2022, (n = 391).
| Clinical characteristics | Categories | Survival status | |
|---|---|---|---|
| Recovered, n (%) | Censored, n (%) | ||
| History of DM | Newly diagnosed | 187 (93.5) | 13 (6.5) |
| Known | 183 (95.8) | 8 (4.2) | |
| Type of insulin (n = 191) | Mixed (NPH and regular insulin) | 171 (95.5) | 8 (4.5) |
| NPH | 11 (100) | 0 (0) | |
| Regular and Lente | 1 (100) | 0 (0) | |
| Precipitating factor among known DM children (n = 191) | Total omission of insulin | 79 (95.2) | 4 (4.8) |
| Inadequate dosage of insulin | 15 (100) | 0 (0) | |
| Infection | 47 (94) | 3 (6) | |
| Consumption of high-carbohydrate content food and fluid | 22 (95.7) | 1 (4.3) | |
| Others | 3 (100) | 0 (0) | |
| Unknown | 17 (100) | 0 (0) | |
| Weight loss | Yes | 171 (94.5) | 10(5.5) |
| No | 199 (94.8) | 11 (5.2) | |
| Kussmaul breathing | Yes | 176 (92.6) | 14 (7.4) |
| No | 194 (96.5) | 7 (3.5) | |
| Changed in mental status | Yes | 164 (93.7) | 11 (6.3) |
| No | 206 (95.4) | 10 (4.6) | |
| Shock | Yes | 59 (96.7) | 2 (3.3) |
| No | 311 (94.2) | 19 (5.8) | |
| Dehydration | Yes | 183 (93.4) | 13 (6.6) |
| No | 128 (95.5) | 6 (4.5) | |
| Severity of DKA | Mild | 174 (96.7) | 6 (3.3) |
| Moderate | 127 (90.7) | 13 (9.3) | |
| Severe | 69 (97.2) | 2 (2.8) | |
| Comorbidity | Yes | 98 (96.0) | 4 (4.0) |
| No | 272 (94.0) | 17 (6.0) | |
| Types of comorbidities | RTI | 43 (95.6) | 2 (4.4) |
| UTI | 21 (100) | 0 (0) | |
| Gastroenteritis | 18 (94.7) | 1 (5.3) | |
| Others* | 16 (94) | 1 (6) | |
RTI: respiratory tract infection; UTI: urinary tract infection.
Others: stress, trauma.
Biochemical profiles of the study participants
The majority of the study participants, 370 (94.6%), were found to have high blood sugar levels (hyperglycemia). Although DKA is typically associated with hyperglycemia, we identified 21 cases of euglycemic DKA, defined as DKA with blood glucose levels < 200 mg/dL. In addition, 30.9% of the participants had a high level of glucose in their urine (+3 on arrival), and approximately half of the subjects presented with a high level of ketones in their urine (ketonuria) at plus three. Furthermore, 169 participants (43.2%) had a blood sodium level below 135 mmol/l, and 22 participants (5.6%) had hypernatremia. The median chloride and potassium level on admission were 100.4 mmol/l (range: 68–137), and 4.54 mmol/l (range: 2.3–7.69), respectively (Table 3).
Table 3.
Biochemical characteristics of children admitted with DKA at selected public hospitals in Addis Ababa from January 2018 to December 2022, (n = 391).
| Variable | Category | Survival status | |
|---|---|---|---|
| Recovered (%) | Censored (%) | ||
| RBS | Hyperglycemic | 349 (94.3) | 21(5.7) |
| Euglycemic | 21 (100) | 0 (0) | |
| Urine glucose | Negative | 46 (100) | 0 (0) |
| +1 | 81 (97.6) | 2 (2.4) | |
| +2 | 103 (95.4) | 5 (4.6) | |
| +3 | 114 (94.2) | 7 (5.8) | |
| +4 | 26 (78.8) | 7 (21.2) | |
| Urine ketone | +1 | 15 (100) | 0 (0) |
| +2 | 106 (96.4) | 4 (3.6) | |
| +3 | 186 (95.9) | 8 (4.1) | |
| +4 | 63 (87.5) | 9 (12.5) | |
| Sodium (mmol/l) | Below 135 | 160 (94.7) | 9 (5.30) |
| 135–145 | 189 (94.5) | 11 (5.5) | |
| >145 | 21 (95.5) | 1 (4.5) | |
| Chloride (mmol/l) | Low | 106 (93) | 8 (7) |
| Normal | 204 (95.8) | 9 (4.2) | |
| High | 60 (93.8) | 4 (6.3) | |
| Potassium (mmol/l) | Below 3.5 | 32 (94.1) | 2 (5.9) |
| 3.5–5.5 | 292 (94.8) | 16 (5.2) | |
| >5.5 | 46 (93.9) | 3 (6.1) | |
| Age-specific creatinine level (mg/dl) | Normal | 267 (94) | 17 (6) |
| High | 103 (96.3) | 4 (3.7) | |
Treatment-related characteristics of the study subjects
Among the 391 children with DKA, 278 (71.10%) were treated with IV fluid. All children were injected with regular insulin. Two hundred and eighty-six (73.1%) have taken IV boluses. The RBS was checked every hour, and the urine ketone was assessed every 2 h. Potassium chloride was added for 237 (60.61%) of the study subjects. The most frequently used fluids were normal saline and dextrose in normal saline. Children with comorbid illnesses were managed with specific treatments for each disease (Table 4).
Table 4.
Treatment-related characteristics of children admitted with DKA at selected governmental hospitals in Addis Ababa from January 2018 to December 2022 (n = 391).
| Variable | Category | Survival status | |
|---|---|---|---|
| Recovered (%) | Censored (%) | ||
| Fluid management | IV fluid | 262 (94.2) | 16 (5.8) |
| Oral fluid | 92 (95.8) | 4 (4.2) | |
| Both IV and oral | 16 (94.1) | 1 (5.9) | |
| IV bolus | Received | 270 (94.4) | 16 (5.6) |
| Not received | 100 (95.2) | 5 (4.8) | |
| KCL | Added | 224 (94.5) | 13 (5.5) |
| Not added | 146 (94.8) | 8 (5.2) | |
| Other medication | Antibiotics | 64 (98.5) | 1 (1.5) |
| Others | 14 (82.4) | 3 (17.6) | |
| Both antibiotics and others | 9 (81.8) | 2 (18.2) | |
KCL: potassium chloride; IV: intravenous; IV fluid: NS, DNS, ringer lactate; Oral fluid: ORS, water and KCL.
Time to recovery from DKA and overall Kaplan–Meier survivor function
Among the 391 study subjects, 370 (94.6%) were recovered and discharged from the hospital, whereas 21 (5.4%) were censored. More than half 11 of the censored cases had unknown status; one had gone against medical therapy; six were absent on call; two were transferred to another facility; and one had died. The median length of hospital stay was 4 days. There was a total of 11,289 h of analysis time at risk and under observation. As shown in Figure 1, the overall median time to recovery from ketosis was 27 h, with an IQR of 16–38 h.
Figure 1.
Overall Kaplan–Meier estimation of survivor functions of children with DKA followed at selected governmental hospitals, Addis Ababa, Ethiopia, from January 2018 to December 2022, (n = 391).
Comparison of the survival status
The log-rank test was used to assess the differences in survival time among various groups of predictors, such as diabetes history, severity of DKA, presence of comorbidities, and blood glucose level. The test indicated that the median time to recover from DKA varied across these independent categorical variables (Table 5). For example, in children with newly diagnosed T1D, the median time to recover from DKA was longer than that of children with previously diagnosed DM, with median times of 32 and 23 h, respectively (Figure 2).
Table 5.
Median time to recovery and log-rank test according to different characteristics of children admitted with DKA at selected governmental hospitals from January 2018 to December 2022, Addis Ababa, Ethiopia (n = 391).
| Variables | Category | Median time to recovery (95% CI) | Log-rank test | |
|---|---|---|---|---|
| p-Value | X 2 | |||
| Diabetic history | Newly dx | 32 (30–36) | 0.000 | 41.75 |
| Known DM | 23 (19–24) | |||
| Polysymptoms | Yes | 28 (26–31) | 0.0001 | 15.43 |
| No | 23 (18–27) | |||
| Nausea and vomiting | Yes | 30 (28–32) | 0.0001 | 16.29 |
| No | 22 (24–28) | |||
| Weight loss | Yes | 31 (28–36) | 0.0000 | 26.24 |
| No | 23 (21–26) | |||
| Abdominal pain | Yes | 30 (28–33) | 0.0000 | 21.82 |
| No | 23 (21–24) | |||
| Kussmual breathing | Yes | 35 (32–37) | 0.0000 | 83.55 |
| No | 27 (24–28) | |||
| Shock | Yes | 42 (37–47) | 0.0000 | 42.03 |
| No | 23 (22–26) | |||
| Severity | Mild | 19 (17–22) | 0.0000 | 121.95 |
| Moderate | 30 (28–33) | |||
| Severe | 42 (38–47) | |||
| Comorbidity | Yes | 36 (31–38) | 0.0000 | 22.84 |
| No | 23 (22–26) | |||
| Blood glucose level | Hyperglycemic | 28 (26–39) | 0.0000 | 18.98 |
| Euglycemic | 17 (13–22) | |||
| Fluid management | Taken IV fluid | 31 (29–35) | 0.0000 | 93.56 |
| Taken oral fluid | 15 (13–19) | |||
| Taken both | 28 (14–31) | |||
| IV bolus | Yes | 31 (29–34) | 0.0000 | 72.50 |
| No | 16 (14–19) | |||
Figure 2.
The Kaplan–Meier survival curves showing the survival status of children with DKA at selected governmental hospitals by history of diabetes, Addis Ababa, Ethiopia, from January 2018 to December 2022, (n = 391).
There was also variation in the median time among children admitted with DKA regarding the severity of DKA. Children with mild DKA had a quicker median recovery time (19 h), than moderate (30 h) and severe (42 h) DKA (Figure 3).
Figure 3.
The Kaplan–Meier survival curves showing the survival status of children admitted with DKA at selected governmental hospitals by severity of DKA, Addis Ababa, Ethiopia, from January 2018 to December 2022, (n = 391).
Besides this, compared with children who presented without acute comorbid illness, children with acute comorbid illness had delayed median recovery time from DKA (23 and 36 h respectively; Figure 4).
Figure 4.
The Kaplan–Meier survival curves showing the survival status of children admitted with DKA at selected governmental hospitals by the presence of comorbid diseases, Addis Ababa, Ethiopia, from January 2018 to December 2022 (n = 391).
Predictors of recovery time from DKA
After controlling for possible confounders, multivariable Cox proportional hazard regression analysis showed that the diabetic history, severity of DKA, presence of comorbidity, and random blood glucose level were all found to be the significant predictors of the time to recover from DKA among children at p-value ⩽ 0.05.
According to this study, children with new-onset diabetes had a 59% longer recovery time from DKA (AHR = 0.41; 95% CI: 0.30–0.56) than children with known (established) diabetes. Compared to children admitted with severe DKA, those with mild and moderate DKA had 1.73- and 2.35-times faster recovery time from DKA (AHR = 1.73, 95% CI: 1.13–2.63) and (AHR = 2.35, 95% CI: 1.36–4.10), respectively. Similarly, the recovery time was 1.76 times faster in children without comorbid diseases when compared to those presenting with other comorbid illnesses (AHR = 1.76, 95% CI: 1.37–2.26). In addition, compared to euglycemic DKA, children with hyperglycemic DKA had a 39% longer recovery time from DKA (AHR = 0.61; 95% CI: 0.39–0.96; Table 6).
Table 6.
Bivariable and multivariable Cox proportional hazards regression analysis of time to recovery from DKA among children admitted with DKA at selected governmental hospitals in Addis Ababa from January 2018 to December 2022 (n = 391).
| Variables | Categories | Survival status | CHR (95% CI) | AHR (95% CI) | |
|---|---|---|---|---|---|
| Recovered (%) | Censored (%) | ||||
| Diabetic history | New onset | 187 (93.5) | 13 (6.5) | 0.51 (0.42–0.63) | 0.41 (0.30–0.56)* |
| Known DM | 183 (95.8) | 8 (4.2) | 1 | 1 | |
| Polysymptoms | Yes | 275 (94.5) | 16 (5.5) | 0.63 (0.50–0.8) | 1.1 (0.81–1.45) |
| No | 95 (95) | 5 (5) | 1 | 1 | |
| Nausea and vomiting | Yes | 224 (93.72) | 15 (6.28) | 1 | 1 |
| No | 146 (96.05) | 9 (3.95) | 1.52 (1.23–188) | 1.14 (0.87–1.50) | |
| Weight loss | Yes | 171 (94.48) | 10 (5.52) | 1 | 1 |
| No | 199 (94.76) | 11 (5.24) | 1.70 (1.38–2.10) | 0.99 (0.73–1.36) | |
| Abdominal pain | Yes | 169 (92.86) | 13 (7.14) | 1 | 1 |
| No | 201 (96.17) | 8 (3.83) | 1.62 (1.31–20) | 1.3 (1.0–1.64) | |
| Kussmaul breathing | Yes | 176 (92.63) | 14 (7.37) | 1 | 1 |
| No | 194 (96.52) | 7 (3.48) | 2.61 (2.10–3.25) | 1.40 (0.95–2.0) | |
| Shock on admission | Yes | 59 (96.72) | 2 (3.28) | 0.41 (0.31–0.54) | 0.88 (0.60–1.32) |
| No | 311 (94.24) | 19 (5.76) | 1 | 1 | |
| Severity of DKA | Mild | 174 (96.67) | 6 (3.33) | 4.86 (3.55–6.65) | 2.35 (1.36–4.1)* |
| Moderate | 127 (90.71) | 13 (9.29) | 2.36 (1.72– 3.2) | 1.73 (1.13–2.63)* | |
| Severe | 9 (97.18) | 2 (2.82) | 1 | 1 | |
| Presence of comorbidity | Yes | 98 (96.1) | 4 (3.9) | 1 | 1 |
| No | 272 (94.12) | 17 (5.88) | 1.73 (1.37–2.20) | 1.76 (1.37–2.26)* | |
| Blood glucose level | Hyperglycemic | 349 (94.32) | 21 (5.68) | 0.39 (0.25–0.61) | 0.61 (0.39–0.96)* |
| Euglycemic | 21 (100) | 0 (0) | 1 | 1 | |
| IV bolus | Taken IV bolus | 270 (94.41) | 16 (5.59) | 1 | 1 |
| Not taken IV bolus | 100 (95.24) | 5 (4.76) | 2.65 (2.1–3.37) | 0.74 (0.39–1.43) | |
| Fluid management | IV fluid | 262 (94.24) | 19 (6.7) | 1 | 1 |
| Oral fluid | 92 (95.83) | 4 (4.17) | 3.18 (2.47–4.08) | 2.0 (1.0–3.92) | |
| Both | 16 (94.12) | 1 (5.88) | 1.53 (0.92–2.54) | 1.14 (0.66–1.95) | |
1-reference category.
p-value < 0.005.
Discussion
In this study, we aimed to determine the time to recovery from DKA and its predictors in children with T1D who were admitted to a public hospital in Addis Ababa. This study revealed that the median survival time to recover from DKA was 27 h, with an IQR of 16–38. The median survival time is consistent with the study conducted in India 26 but longer than those reported in the USA, 27 Japan, 22 and Indonesia. 28 In contrast, the median survival time to recover from DKA is shorter than the study conducted in China. 29 These variations may be attributed to differences in the definition of DKA resolution, the severity of DKA at presentation, the study design, sample size, and treatment protocols.
For instance, high-income countries such as the USA and Japan have often standardized DKA management protocols and advanced healthcare infrastructure, which may contribute to shorter recovery times. In contrast, low- and middle-income countries (LMICs) such as Ethiopia and India may face challenges such as delayed presentation, limited access to insulin, and resource constraints, which could prolong recovery. The use of urine ketones as a marker for DKA resolution in Ethiopia may also have led to an overestimation of the recovery time due to the persistence of ketones after acidosis resolution. These disparities highlight the need for improved diabetes care infrastructure and public health interventions in LMICs to reduce the burden of DKA.
Regarding predictors of time to recovery from DKA, the present study showed that children with new-onset DM have a slower recovery time compared with children with established diabetes, which is comparable with studies conducted in Colombia, 24 Israel, 30 and Japan. 22 This could be attributed to the newly diagnosed type 1 children having a lower index of suspicion and hence delaying seeking medical care. As a result, they were subjected to prolonged insulin deficiency, resulting in severe DKA. Besides that, newly diagnosed children may not tolerate or develop resistance to exogenous insulin, which is used to treat DKA.
The current study revealed that the time to resolve DKA in children with other comorbid illnesses was significantly longer than that for those admitted with DKA alone, which is supported by a study conducted in Japan 22 in which children with gastroenteritis had a prolonged recovery time from DKA when compared to others without the condition. Likewise, a study done in India 31 showed that children with AKI had a longer recovery time from DKA than those without AKI. The possible explanation may be that acute comorbid illness may markedly increase the amount of counterregulatory hormones, which results in a decrease in the action of insulin by increasing blood sugar through promoting gluconeogenesis, glycogenolysis, and ketogenesis, which in turn worsens DKA.
The present study further revealed that children with mild and moderate DKA had a faster recovery time than those admitted with severe DKA, which is comparable with two studies conducted in China32,33 and the USA. 27 The likely explanation is that in severe DKA, the body has more widespread metabolic abnormalities and has undergone major changes because of insulin shortage and hyperglycemia. As a result, the body needs more time to gradually replace fluids, electrolytes, and insulin to repair metabolic abnormalities. In contrast to severe DKA, the metabolic derangement in mild and moderate DKA is minor, requiring less time to rebalance and correct the acidosis.
Another predictor of time to recovery from DKA was the blood glucose level. Children with blood glucose levels below 250 recovered quickly. A study conducted in Turkey aligns with our findings. 23 The explanation for this could be that high blood sugar causes increased osmotic diuretics, resulting in severe dehydration, severe electrolyte imbalances, and severe acidosis, which takes longer to correct. High blood glucose can also result in hyperinflammation, which in turn causes insulin resistance. 34
Strength and limitation of the study
This study provides valuable insights into the predictors of DKA recovery in a resource-limited setting. The research involved multiple hospitals, making it more representative than a study conducted at a single center. It is also the first study in Ethiopia to use Cox regression analysis to identify factors that impact the recovery time from DKA in children with T1D. However, there are limitations to the study. First, the resolution of DKA was determined based on the absence of urine ketones in two consecutive measurements, as blood gas analysis (pH and bicarbonate levels) is not routinely available in our setting due to resource constraints. Although this method was the most feasible and practical approach in our context, it may have led to an overestimation of the recovery time, as urine ketones can persist for several hours after the resolution of metabolic acidosis. Additionally, the study relied on secondary data, which meant that information about some children and their families had to be excluded, potentially affecting the time to recovery from DKA. Furthermore, the heterogeneity of censored cases (transfers, treatment discontinuations, deaths, and unknown outcomes), may affect the interpretation of predictors of DKA recovery.
Conclusion
The overall median time to recovery from DKA in children admitted with type 1 DM was 27 h. Diabetic history, severity of DKA, presence of comorbidity, and blood glucose level on admission were statistically significant predictors of the time to recovery from DKA in children. Early detection and intervention, comprehensive clinical and biochemical evaluation, and effective management are crucial for improving outcomes in children with DKA.
Supplemental Material
Supplemental material, sj-docx-1-smo-10.1177_20503121251343175 for Time to recovery from diabetic ketoacidosis and its predictors among children with type 1 diabetes at selected governmental hospitals in Addis Ababa, Ethiopia; A five-year retrospective follow-up study by Shimeles Tefera Mamo, Tigistu Gebreyohannis Gebretensaye, Feven Mulugeta and Gemechu Gelan Bekele in SAGE Open Medicine
Acknowledgments
We would like to express our heartfelt gratitude to Addis Ababa University for the financial support to carry out the research. We are also grateful to each hospital’s pediatric department heads, unit heads, coordinators, and members of the medical records departments, whose contributions were indispensable to the success of this study. It is also our pleasure to acknowledge data collectors and supervisors for giving their time during the study period.
Footnotes
ORCID iDs: Shimeles Tefera Mamo 
https://orcid.org/0009-0003-1906-0919
Feven Mulugeta
https://orcid.org/0000-0002-4868-1875
Gemechu Gelan Bekele
https://orcid.org/0000-0002-8476-5320
Ethical considerations: Ethical approval was received from Addis Ababa University’s research ethical review board, College of Health Science, School of Nursing and Midwifery (Ref. No pm 23/681). The letter of support was also received from the Department of Nursing. The copy of the ethical clearance and supportive letter was delivered immediately to tertiary hospitals regulated by the Ministry of Health, as well as to the Addis Ababa health bureau. This retrospective study analyzed existing medical records of children with DKA. Due to the nature of the research, which involved only the review of medical records and no direct patient contact, the ethical review board granted a waiver of written informed consent. The information acquired was kept confidential; no names or other personal identifiers were utilized throughout data collection and analysis.
Consent for publication: Not applicable.
Author contributions: STM drafted the proposal and made a significant contribution to the write-up of the methods, results, and discussion section. TGG and FM reviewed the initial drafting of the proposal to the final manuscript draft in all sections of this paper. GGB reviewed the methodology and results and drafted the manuscript. All authors reviewed the draft and approved the final manuscript.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Addis Ababa University provided financial support for this study. The funders had no role in study design, data collection and analysis, the decision to publish, or the preparation of the manuscript.
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement: The data used in this study will be available from the corresponding author on reasonable request.
Supplemental material: Supplemental material for this article is available online.
References
- 1. Wolosowicz M, Lukaszuk B, Chabowski A. The causes of insulin resistance in type 1 diabetes mellitus: is there a place for quaternary prevention? Int J Environ Res Public Health 2020; 17(22): 8651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Aschner P, Karuranga S, James S, et al. The International Diabetes Federation’s guide for diabetes epidemiological studies. Diabetes Res Clin Pract 2021; 172: 108630. [DOI] [PubMed] [Google Scholar]
- 3. Aye T, Mazaika PK, Mauras N, et al. Impact of early diabetic ketoacidosis on the developing brain. Diabetes Care 2019; 42(3): 443–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kuppermann N, Ghetti S, Schunk JE, et al. Clinical trial of fluid infusion rates for pediatric diabetic ketoacidosis. N Engl J Med 2018; 378(24): 2275–2287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Michal C, Smadar S, Nehama Z-L, et al. Diabetic ketoacidosis in the pediatric population with type 1 diabetes. In: Kenia Pedrosa N. (ed.) Major topics in type 1 diabetes. Rijeka: IntechOpen; 2015, p. Ch. 5. [Google Scholar]
- 6. Maahs DM, Hermann JM, Holman N, et al. Rates of diabetic ketoacidosis: international comparison with 49,859 pediatric patients with type 1 diabetes from England, Wales, the U.S., Austria, and Germany. Diabetes Care 2015; 38(10): 1876–1882. [DOI] [PubMed] [Google Scholar]
- 7. Majaliwa E, Mohn A, Chiavaroli V, et al. Management of diabetic ketoacidosis in children and adolescents in sub-Saharan Africa: a review. East Afr Med J 2010; 87(4): 167–173. [DOI] [PubMed] [Google Scholar]
- 8. Gebeyehu K, Tefera M, Melaku B, et al. Assessment of clinical profiles, and treatment outcomes for children with diabetic ketoacidosis, in two hospitals selected from Addis Ababa, Ethiopia, 2020. Ethiop J Health Develop 2022; 36(2): 1–9. [Google Scholar]
- 9. Atkilt HS, Turago MG, Tegegne BS. Clinical characteristics of diabetic ketoacidosis in children with newly diagnosed type 1 diabetes in Addis Ababa, Ethiopia: a cross-sectional Study. PLoS One 2017; 12(1): e0169666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Shenoy S, Upadhya N. The factors affecting resolution of acidosis in children with diabetic ketoacidosis—a retrospective study from a tertiary care center in India. Indian J Child Health 2017; 4(3): 294–297. [Google Scholar]
- 11. Mendez Y, Surani S, Varon J. Diabetic ketoacidosis: treatment in the intensive care unit or general medical/surgical ward? World J Diabetes 2017; 8(2): 40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wolfsdorf JI, Glaser N, Agus M, et al. ISPAD Clinical Practice Consensus Guidelines 2018: diabetic ketoacidosis and the hyperglycemic hyperosmolar state. Pediatr Diabetes 2018; 19: 155–177. [DOI] [PubMed] [Google Scholar]
- 13. Praveen PA, Hockett CW, Ong TC, et al. Diabetic ketoacidosis at diagnosis among youth with type 1 and type 2 diabetes: results from SEARCH (United States) and YDR (India) registries. Pediatr Diabetes 2021; 22(1): 40–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Bialo SR, Agrawal S, Boney CM, et al. Rare complications of pediatric diabetic ketoacidosis. World J Diabetes 2015; 6(1): 167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Poovazhagi V. Risk factors for mortality in children with diabetic keto acidosis from developing countries. World J Diabetes 2014; 5(6): 932–938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Icks A, Strassburger K, Baechle C, et al. Frequency and cost of diabetic ketoacidosis in Germany—study in 12,001 pediatric patients. Exp Clin Endocrinol Diabetes 2013; 121(1): 58–59. [DOI] [PubMed] [Google Scholar]
- 17. Dovc K, Neuman V, Gita G, et al. Association of diabetic ketoacidosis at onset, diabetes technology uptake, and clinical outcomes after 1 and 2 years of follow-up: a collaborative analysis of pediatric registries involving 9269 children with type 1 diabetes from nine countries. Diabetes Care 2025; 48(4): 648–654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Castellanos L, Tuffaha M, Koren D, et al. Management of diabetic ketoacidosis in children and adolescents with type 1 diabetes mellitus. Pediatr Drugs 2020; 22: 357–367. [DOI] [PubMed] [Google Scholar]
- 19. Benoit SR, Zhang Y, Geiss LS, et al. Trends in diabetic ketoacidosis hospitalizations and in-hospital mortality—United States, 2000–2014. MMWR Morb Mortal Wkly Rep 2018; 67(12): 362–365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Hadgu FB, Sibhat GG, Gebretsadik LG. Diabetic ketoacidosis in children and adolescents with newly diagnosed type 1 diabetes in Tigray, Ethiopia: retrospective observational study. Pediatr Health Med Ther 2019; 10: 49–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Bogale H. Precipitating factors and clinical-laboratory features of diabetic ketoacidosis (DKA) at Tikur Anbesa Specialized Hospital Emergency Department, Addis Ababa, Ethiopia. Addis Ababa: Addis Ababa University, 2014. [Google Scholar]
- 22. Yuyama Y, Kawamura T, Nishikawa-Nakamura N, et al. Relationship between bedside ketone levels and time to resolution of diabetic ketoacidosis: a retrospective cohort study. Diabetes Ther 2021; 12: 3055–3066. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Ongun EA, Çelik N. Risk factors associated with resolution of diabetic ketoacidosis in pediatric critical care units. Cumhur Med J 2019; 41(1): 42–50. [Google Scholar]
- 24. Valero-Guzmán L, Vásquez-Hoyos P, Camacho-Cruz J, et al. Difference in the duration of pediatric diabetic ketoacidosis: comparison of new-onset to known type 1 diabetes. Pediatr Diabetes 2020; 21(5): 791–799. [DOI] [PubMed] [Google Scholar]
- 25. World Health Organization. Guideline: updates on pediatric emergency triage, assessment and treatment: care of critically-ill children. Guideline: updates on pediatric emergency triage, assessment and treatment: care of critically-ill children. Geneva: World Health Organization, 2016. [PubMed] [Google Scholar]
- 26. Varshney GA, Varshney D, Mehr V, et al. Clinical profile and outcome of diabetic ketoacidosis in children at tertiary care hospital. J Evol Med Dental Sci 2015; 4(31): 5329–5334. [Google Scholar]
- 27. von Oettingen JE, Rhodes ET, Wolfsdorf JI. Resolution of ketoacidosis in children with new onset diabetes: evaluation of various definitions. Diabetes Res Clin Pract 2018; 135: 76–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Pulungan A, Batubara J, Tridjaja B, et al. Profile of diabetic ketoacidosis in children with diabetes mellitus in a tertiary care hospital in Jakarta: a retrospective study. Acta Sci Paediatr 2019; 2(2): 24–29. [Google Scholar]
- 29. Yuan X, Wang J, Chen X, et al. Effects of the timing of the initiation of dietary intake on pediatric type 1 diabetes for diabetic ketoacidosis. BMC Pediatr 2022; 22(1): 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Smuel-Zilberberg K, Shalitin S, Yackobovitch-Gavan M, et al. Diabetes ketoacidosis recovery in youth with newly diagnosed and established type 1 diabetes. Pediatr Res 2022; 91(5): 1272–1277. [DOI] [PubMed] [Google Scholar]
- 31. Baalaaji M, Jayashree M, Nallasamy K, et al. Predictors and outcome of acute kidney injury in children with diabetic ketoacidosis. Indian Pediatr 2018; 55: 311–314. [PubMed] [Google Scholar]
- 32. Liu Q, Yin X, Li P. Clinical, hormonal, and biochemical characteristics of 70 Chinese children with moderate to severe type 1 diabetic ketoacidosis. BMC Endocr Disord 2022; 22(1): 301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Wei Y, Wu C, Su F, et al. Clinical characteristics and outcomes of patients with diabetic ketoacidosis of different severity. Medicine (Baltimore) 2020; 99(45): e22838. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Glaser N, Fritsch M, Priyambada L, et al. ISPAD clinical practice consensus guidelines 2022: diabetic ketoacidosis and hyperglycemic hyperosmolar state. Pediatr Diabetes 2022; 23(7): 835–856. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material, sj-docx-1-smo-10.1177_20503121251343175 for Time to recovery from diabetic ketoacidosis and its predictors among children with type 1 diabetes at selected governmental hospitals in Addis Ababa, Ethiopia; A five-year retrospective follow-up study by Shimeles Tefera Mamo, Tigistu Gebreyohannis Gebretensaye, Feven Mulugeta and Gemechu Gelan Bekele in SAGE Open Medicine