Prevalence of treatment outcomes and associated factors of diabetic ketoacidosis among adult diabetic patients admitted to Debre Markos comprehensive specialized hospital, northwest Ethiopia

Abstract

Background

Diabetic ketoacidosis emergencies are serious acute complications of diabetes mellitus, and their health and economic impacts have been increasing among adult diabetic patients. Despite the growing burden of emergencies from diabetic ketoacidosis among adults with diabetes, its outcomes and factors of treatment have not been well studied in Ethiopia.

Objective

To assess the prevalence of poor treatment outcomes of diabetic ketoacidosis and associated factors among adult diabetic patients admitted to the Debre Markos comprehensive specialized hospital, northwest Ethiopia.

Methods

A hospital-based retrospective cohort study was conducted at Debre Markos comprehensive specialized hospital from April 8, 2025 to April 28, 2025. A total of 360 patients with diabetic DKA who were admitted from August 28, 2019 to September 30, 2024 participated in the study. A simple random sampling technique was used to select study participants. Data were extracted using a pre-tested data collection tool adapted from different literatures. The data were entered into epi-data version 4.6.0 and exported to SPSS version 26 for analysis. Both bivariable and multivariable binary logistic regression analysis were done. Statistical significance was declared at a p-value < 0.05 with an odds ratio of 95% confidence interval.

Results

The prevalence of poor treatment outcomes was 15.5% (95%, CI: 11.5%, 19.3%). Factors, such as severe DKA (AOR: 1.482, 95% CI: 1.324, 4.872), the presence of comorbidities (AOR: 1.752, 95% CI: 1.215, 3.865), and underlying infections (95% CI: 1.362, 4.125), discontinuation of drugs (AOR: 2.115, 95% CI: 1.245, 3.865), treatment complications (AOR: 1.356, 95% CI: 1.253, 4.125), ketone levels > 3 (AOR: 1.213, 95% CI: 1.052, 2.876), and hospital stays of less than 5 days (AOR: 1.29, 95% CI: 1.022, 3.254) were also significant factors associated with poor outcomes in patients with DKA.

Conclusions

This study found that a high number of patients with DKA in the Debre Markos comprehensive specialized hospital experienced poor treatment outcomes. Significant factors included severe DKA, comorbidities, infections, drug discontinuation, treatment complications, high ketone levels and short hospital stays. To improve treatment outcomes, early identification and proactive management of high-risk DKA patients, particularly those who present with severe illness, comorbidity conditions, or infections, shall be prioritized.

Introduction

Diabetes mellitus is a group of metabolic disorders characterized by the presence of hyperglycaemia due to impaired insulin secretion, action, or both, which results in alteration of carbohydrate, protein, and fat metabolism [1]. Diabetes mellitus is most commonly associated with older age, obesity, family history of diabetes mellitus, genetic susceptibility, hypertension, dyslipidaemia, autoimmunity, and physical inactivity [2]. Diabetes is one of the largest global public health emergencies of the twenty-first century. According to the report of the International Diabetes Federation, 536.6 million adults (10.5%) lived with diabetes mellitus worldwide in 2021, and this number is expected to increase to 783.2 million (12.2%) by 2045 [3]. According to the IDF 2021 estimate, 24 million adults aged 20–79 years lived with diabetes in Africa [4].

Diabetic ketoacidosis is an emergency that encompasses the spectrum of preventable acute diabetic complications, which can occur in patients with type 1 and type 2 diabetes mellitus. DKA occurs in the setting of relative or absolute insulin deficiency, excessive counter-regulatory hormone levels, progressive volume depletion, and loss of electrolytes [5]. According to the World Health Organization treatment guideline, diabetic ketoacidosis is a common and potentially fatal acute complication of diabetes mellitus during diagnosis and follow-up [6].

The poor outcome of the DKA treatment can lead to a debilitating and potentially fatal complication, including cerebral oedema and severe hypoglycaemia. Mortality of DKA has been reported to be less than 5% in centres of treatment experienced in the Americas, Europe, and Asia [7] and in Riyadh, Saudi Arabia, King Abdul-Aziz Medical City (2%) [8]. However, in developing countries, high poor treatment outcomes are observed. For instance, in Malaysia 24% of which 17.6% of DKA patients died [9], Tanzania and Ghana between 26 and 29% [10], in Liberia 19.2% [11], Zambia 21% [12], Kenya’s Kenyatta National Hospital 29.8% [13]. Similar studies in Ethiopian countries showed a high prevalence of poor treatment outcomes: Nekemte referral hospital 11.4% [14], Shashemene referral hospital, 26.2% [15], Addis Addis Ababa, at St. Luke Catholic Hospital, 13% [16], Adama General Hospital 15.1% [17], Jima University Hospital 9.8% [18], Hiwot Fana Specialized University Hospital 17.8% [19] and Debre Tabor General Hospital, 20% [20]. This magnitude imposes a high national and individual economic burden associated with medication, bed occupancy, long hospital stays, and mortality. The average treatment and laboratory cost is significantly higher among diabetic ketoacidosis emergencies compared to other emergency cases [21].

Common causes of DKA are missed insulin doses, illness or infection, and undiagnosed or untreated diabetes. The main clinical features of DKA are hyperglycaemia, dehydration, loss of electrolytes, and acidosis [22–25]. Diabetic ketoacidosis (DKA) has multiple risk factors, with 6–21% of cases presenting as the first manifestation of type 1 diabetes [26, 27]. Common precipitants include infections, intercurrent illnesses, psychological stress, and insulin omission, with infections accounting for 14–58% of cases worldwide [22, 25, 27–30]. Other triggers include stroke, pancreatitis, myocardial infarction, trauma, and substance use [2, 31]. Insulin omission, often driven by socioeconomic and psychological challenges, is a major cause, responsible for over two-thirds of admissions in minority groups [32]. Additional risks include younger age, comorbidities, elevated HbA1c, poverty, food insecurity, and adverse social determinants of health [33, 34]. Ethiopian studies further highlight sociodemographic, behavioural, treatment-related, clinical, and alternative medicine use as key risk factors [15, 20].

The prevalence of diabetic ketoacidosis (DKA) varies considerably across regions, including within Ethiopia, with most studies reporting a high occurrence rate. Despite this, there is limited primary evidence on the prevalence and determinants of DKA among diabetic patients receiving follow-up care at Debre Markos Comprehensive Specialized Hospital. As the only comprehensive hospital in the West Gojjam Zone of Northwest Ethiopia, this facility is better equipped with specialized professionals, advanced medical technologies, and essential medications compared to primary and general hospitals. Therefore, the present study aimed to assess the prevalence of treatment outcomes of DKA and its associated factors among adult diabetic patients admitted to Debre Markos Comprehensive Specialized Hospital. Findings from this study will provide valuable insights for policymakers, health system managers, healthcare providers, and patients in reducing the burden of DKA and its associated factors.

Methods

Study area and period

Debre Markos Comprehensive Specialized Hospital is one of the governmental hospitals located in the Amhara region, East Gojame Zone, in Debre Markos town, which is 300 km from Addis Ababa, the capital city of Ethiopia, and 255 km from Bahirdar (the capital city of the Amhara Regional state) in northwest Ethiopia. It provides preventive, diagnostic, curative and rehabilitative service for millions of people in the region with 9 wards, 29 different outpatient units and 143 beds. Currently there are more than 2301 diabetic patients who receive regular follow-up care at the hospital. From August 28, 2019 to September 30, 2024, the number of patients with DM2 is 1060 and DM1 is 980 and 480 patients were patients with DKA. Outpatient, emergency and surgical care service is provided to diabetes patients. The study was carried out at the Debre Markos Specialized Hospital in northwest Ethiopia from April 8, 2025 to April 28, 2025.

Study design

A hospital-based retrospective cohort study was conducted, involving a review of patient records from admission through discharge.

Population

All adult (≥ 18 years) patients with diabetes who developed ketoacidosis constituted the source population, while those diagnosed with type 1 or type 2 diabetes and admitted with diabetic ketoacidosis during the study period at Debre Markos comprehensive specialized hospital formed the study population.

Inclusion and exclusion criteria

All adult patients (≥ 18 years) diagnosed with type 1 or type 2 diabetes mellitus and admitted with diabetic ketoacidosis to Debre Markos comprehensive specialized hospital between August 28, 2019, and September 30, 2024, were included in the study, whereas individuals with incomplete records of key variables and pregnant women were excluded.

Sample size determination and sampling procedure

Sample size determination

The sample size was calculated using a single population proportion formula for both the outcome of diabetic ketoacidosis treatment and associated factors with a study conducted in Addis Ababa DKA treatment. The improved prevalence was 71.1% [35] of the patients discharged. Assumption of a 95% CI, 5% marginal error (W) and P = 0.711.

The sample size with the formula is:

$n=\frac{Z{\left(\frac{\alpha }{2}\right)}^2\ast PQ}{W^2}=\frac{Z{\left(\frac{\alpha }{2}\right)}^2\ast P(1-P)}{W^2}=\frac{{(1.96)}^2\ast 0.711\ast 0.299}{{(0.05)}^2}=327,$ adding 10% for non-response (incomplete chart), the final sample size for objective one was 360.

For the second objective, the sample size was determined using factors associated with treatment outcomes of DKA patients. Taking into consideration the assumption: Power 80%, one-to-one ratio between exposed and unexposed groups, AOR (lower AOR to get larger sample size), % of outcome in exposed and unexposed groups, were taken. It is computed using EPI-info version 7.2.6 STATCALC dialogue box using the cross-sectional tool bar to auto-calculate the sample size by entering data elements and the final sample size included a 10% non-response rate (List of supplementary files. Supplementary Table 1). Therefore, the larger sample size of the first objective, 360 DKA patients, was the final sample size for this study.

Sampling techniques

In this study, simple random sampling techniques were used to select the study samples. From the emergency and in-patient registration logbook DKA patients were taken and then based on the card number they were selected within a computer-generated random selection. Simple random techniques were used to select the required sample size of 360 from the study period. If a patient experienced two or more episodes of DKA, only the most recent episode was considered to assess treatment outcomes and associated factors.

Study variables

The dependent variable was the result of the treatment of diabetic ketoacidosis (poor vs. good), while the independent variables included sociodemographic characteristics (age, sex, marital status, residence, occupation and family history of diabetes mellitus); health-related factors (severity of DKA, admission recorded clinical characteristics (respiratory rate, pulse rate, blood pressure, temperature, blood glucose level, urine glucose level and urine ketone level), recent acute illness, infection, duration of diabetes mellitus, type of diabetes mellitus, presence of comorbidities, and chronic diabetic complications); and treatment-related variables (type of medication used, drug discontinuation, treatment complications, glycaemic control, frequency of blood glucose monitoring, and doses of diabetes medications).

Operational definitions

Outcomes of DKA treatment: this study is operationalized as a good and poor treatment of the outcome as follows:

Good outcome of treatment: patients who had improved at the end of treatment at discharge [15].

Poor outcome of treatment: patients who were left without medical advice or referred or died in the hospital [15].

DKA: the diagnostic criteria for diabetic ketoacidosis are defined by the presence of hyperglycaemia (plasma glucose ≥ 11.1 mmol/L [200 mg/dL] or a documented history of diabetes mellitus), evidence of ketosis (serum β-hydroxybutyrate concentration > 3.0 mmol/L or urine ketone level ≥ 2 +), and metabolic acidosis (arterial or venous pH < 7.3 and/or serum bicarbonate < 18 mmol/L) [36].

HHS: the diagnosis of hyperglycaemic hyperosmolar state is established by the presence of profound hyperglycaemia (plasma glucose concentration ≥ 33.3 mmol/L [600 mg/dL]), hyperosmolarity (effective serum osmolality ≥ 300 mOsm/kg, calculated as [2 × Na⁺ (mmol/L) + glucose (mmol/L)], or total serum osmolality > 320 mOsm/kg, calculated as [2 × Na⁺ (mmol/L) + glucose (mmol/L) + urea (mmol/L)]), in the absence of clinically significant ketosis (serum β-hydroxybutyrate < 3.0 mmol/L or urine ketone level < 2 +) and without metabolic acidosis (arterial or venous pH ≥ 7.3 and serum bicarbonate concentration ≥ 15 mmol/L) [36].

Long hospital stay was defined as hospital stay for more than 5 days [20].

Short hospital stay was defined as the patient staying in the hospital for less than or equal to 5 days [20].

Hyperglycaemic emergencies—including diabetic ketoacidosis (DKA), hyperosmolar hyperglycaemic state (HHS)—are defined as random plasma glucose > 200 mg/dL and DKA is diagnosed when blood glucose is > 250 mg/dl, arterial pH < 7.3, bicarbonate < 15 mEq/L, in the presence of ketonuria [2].

Mild DKA is characterized by a glucose level ≥ 11.1 mmol/L (200 mg/dL), ketonaemia with β-hydroxybutyrate 3.0–6.0 mmol/L, pH > 7.25 to < 7.30 or bicarbonate 15–18 mmol/L, and an alert mental status [36, 37].

Moderate DKA presents with glucose ≥ 11.1 mmol/L (200 mg/dL), β-hydroxybutyrate 3.0–6.0 mmol/L, pH 7.0–7.25, bicarbonate 10 to < 15 mmol/L, and a reduced level of mental alertness[36].

Severe DKA is defined by glucose ≥ 11.1 mmol/L (200 mg/dL), β-hydroxybutyrate > 6.0 mmol/L, pH < 7.0, bicarbonate < 10 mmol/L, and impaired consciousness ranging from stupor to coma [36].

Hypoglycaemia is defined as a blood glucose level < 70 mg/D [15]).

Severe hypokalaemia is a plasma potassium level of less than 2.5 mmol/L [38].

Euglycaemia is defined as a serum glucose level between 100 and 200 mg/dL [2].

Drug discontinuation is defined as the premature cessation, omission, or interruption of prescribed medications (such as insulin therapy, oral hypoglycaemic agents, or adjunctive drugs) that are essential for the management of diabetes, either by the patient or healthcare provider, before the intended treatment duration[20, 39].

Treatment complications are adverse outcomes during management. Clinician-related complications include hypoglycaemia, hypokalaemia, cerebral oedema, fluid overload, and treatment failure, while patient-related complications include persistent symptoms, neurological issues, prolonged hospitalization, and psychological distress both during pre-hospital and hospitalization [2, 40]

Data collection tool and procedure

Data were collected from the physical records of patient charts using a pre-tested data extraction checklist. The data extraction checklist is developed by the principal investigator by reviewing previous literature.

The data collection tools, included sociodemographic, disease-related, and treatment-related factor components.

A supervisory health officer, three bachelor science nursing professionals for data collection, and two medical record room workers for chart finding were recruited for the data collection process. Patient charts were retrieved using their medical registration numbers from the medical record room by record room workers. The data collectors collected the necessary information using the data extraction tool in the designated card room area. A supervisor and the principal investigator followed the data collection process.

Data quality assurance

One day of training was given to data collectors and supervisors on the objective of the study, the content, meaning of the questionnaire/checklists and the way to collect data before the actual data collection is started. The data collection tool was pre-tested on 5% of the sample size (17 charts) in Fenote Selam Hospital a week before the actual study. After the pre-test is done all modifications were implemented. All data were daily evaluated by the supervisors and the principal investigator. The supervisor was to check the completed data collection tool daily for its completeness. In addition to this, the principal investigator carefully entered and thoroughly cleaned the data before the beginning of the analysis.

Data processing and analysis

Data were checked and entered into the Epi-Data version 4.6 statistical software and exported to the Statistical Package for Social Science (SPSS) statistical software package, version 26 for analysis. Frequencies and proportions were used to summarize the variables. In addition to frequencies and proportions, each variable was presented using tables. The association between the treatment outcomes of patients with DKA and each categorical variable was assessed using the logistic regression model. The outcome variable categorizes two forms: poor outcome (coded 1) and good outcome (coded 0) of treatment. In multivariable regression modelling, variables with a p-value < 0.25 in the bivariable analysis are often selected for inclusion in the multivariable model to avoid prematurely excluding potentially important factors. The significance of the association was determined at a p-value of < 0.05 in multivariate analysis, while the strength of the association is indicated by the adjusted odds ratio (AOR) with a 95% confidence interval. The model was fitted and checked by the Hosmer–Lemeshow fitness model, p-value 0.89. The variance inflation factor (VIF) values of the independent variables ranged from 0.89 to 2.35, which falls well below the commonly accepted thresholds (VIF < 5 or < 10)[41], indicating no evidence of multicollinearity. The research report was prepared in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guideline, which is provided as supplementary files (Supplementary 2 STROBE Statement checklist completed [42, 43]

Results

Sociodemographic characteristics

A total of 348 DKA patient charts were retrieved from those admitted (360) to DMCSH, yielding a response rate of 96.7%. The mean age of the participants was 36.5 years, with a standard deviation of 12 years. The age distribution revealed that most of the patients were between 35 and 44 years of age, with 27%. Females comprised 61.8% of the participants and the majority of the patients resided in an urban area (67.8%). In terms of marital status, the majority were married (52.6%), and occupationally, the majority of the patients were merchants (47.7%). In particular, 75.0% of the patients reported that they had unknown family histories of diabetes (Table 1).

Table 1.

Sociodemographic characteristics among DKA patients at DMCSH, northwest Ethiopia, from August 28, 2019 to September 30, 2024, (n = 348)

Variable	Category	Frequency	Percent
Age in years	18–24	74	21.3
	25–34	85	24.4
	35–44	94	27.0
	45–54	56	16.1
	≥ 55	39	11.2
Sex	Female	215	61.8
	Male	133	38.2
Residence	Urban	236	67.8
	Rural	112	32.2
Marital status	Single	112	32.2
	Married	183	52.6
	Widowed	34	9.8
	Divorced	19	5.5
Occupational status	Merchant	166	47.7
	Employee	128	36.8
	Farmer	45	12.9
	Student	9	2.6
Family history of DM	Yes	40	11.5
	No	37	13.5
	Unknown	261	75.0

Baseline clinical characteristics among patients with DKA

In Table 2, the majority of patients were diagnosed with known Type 1 diabetes (43.1%) or newly diagnosed Type 1 diabetes (35.6%), while 5.5% were newly diagnosed with Type 2 diabetes and 15.8% had known Type 2 diabetes. Most patients (88.8%) experienced diabetic ketoacidosis (DKA) two times or less, with 11.2% having more frequent occurrences. The mean duration of diabetes was 26.54 years (± 39.2 years). Pulse rates were mostly normal (72.1%), with 27.0% having tachycardia and a small percentage (0.9%) experiencing bradycardia. Mean systolic and diastolic blood pressures were 104.4 ± 15.5 and 67.8 ± 10.5, respectively, and most patients had normal blood pressure (59.8%), although 11.2% had hypotension. Respiratory rates were primarily normal (68.1%), while 31.9% had tachypnea. The average temperature was 36.5 °C (± 0.95 °C). Dehydration was most commonly not documented (86.2%), although 10.1% had mild dehydration, and a small number had moderate (3.4%) or severe dehydration (0.3%). Most of the patients had mild DKA (76.4%), moderate DKA in 18.1% and severe DKA in 5.5% (Table 2).

Table 2.

Baseline clinical characteristics among DKA patients, DMCSH, northwest Ethiopia, from August 28, 2019 to September 30, 2024, (n = 348)

Variable	Category	Frequency	Percent
Types of DM diagnosis	New DM 1	124	35.6
	New DM 2	19	5.5
	Known DM 1	150	43.1
	Known DM 2	55	15.8
Frequency of DKA	≤ 2 times	309	88.8
	> 2 times	39	11.2
Duration of DM	Mean (± SD)	26.54 (± 39.2)
Pulse rate	Bradycardia	3	0.9
	Normal	251	72.1
	Tachycardia	94	27.0
Systolic blood pressure	Mean (± SD)	104.4 (± 15.5)
Diastolic blood pressure	Mean (± SD)	67.8 (± 10.5)
BP status	Hypotension	39	11.2
	Normal	208	59.8
	Elevated	12	3.4
	Stage 1	64	18.4
	Stage 2	25	7.2
Respiration rate	Normal	237	68.1
	Tachypnea	111	31.9
Temperature	Mean (± SD)	36.5 (± 0.95)
Dehydration status	Mild dehydration	35	10.1
	Moderate dehydration	12	3.4
	Severe dehydration	1	0.3
	Not documented	300	86.2
Severity of DKA	Mild DKA	266	76.4
	Moderate DKA	63	18.1
	Severe DKA	19	5.5

Comorbidity and related factors that precipitate DKA in patients

In this dataset, the majority (91.7%) had no comorbidity. Hypertension was not present in the vast majority (96.0%). Toxic multinodular goitre and HIV/AIDS were rare, affecting only 0.6% and 0.9% of patients, respectively, while tuberculosis in all forms was also reported by 0.9%. Asthma was found in just 0.3% of the patients. Regarding pneumonia, most of the cases were mild (3.4%) or severe (1.4%), while the majority (95.1%) did not experience pneumonia. Precipitation factors were observed in 29.3% of patients, infection being the most common, reported by 13.5%, followed by urinary tract infection (UTI), which occurred in 4.6% of patients. Most of the patients did not have infections (86.5%) or UTIs (95.4%) (Table 3).

Table 3.

Comorbidity and precipitating factors of DKA among DM patients at DMCSH, Ethiopia, from August 28, 2019 to September 30, 2024 (n = 348)

Variable	Category	Frequency	Percent
Comorbidity	Yes	29	8.3
	No	319	91.7
Hypertension	Yes	14	4.0
	No	334	96.0
Goitre	Yes	2	0.6
	No	346	99.4
HIV/AIDS	Yes	3	0.9
	No	345	99.1
TB all forms	Yes	3	0.9
	No	345	99.1
Asthma	Yes	1	0.3
	No	347	99.7
Pneumonia	Severe	5	1.4
	Mild	12	3.4
	No	331	95.1
Infection	Yes	47	13.5
	No	301	86.5
UTI	Yes	16	4.6
	No	332	95.4

Complication profile of patients with DKA

In Table 4, at admission, the majority of patients (65.8%) presented with a Glasgow Coma Scale (GCS) score of 9–15, while the remaining (34.2%) had a severe impairment of consciousness, with a GCS score of 3–8. A small proportion of patients experienced nausea (1.2%) and vomiting (18.1%), while the majority did not report these symptoms (98.9% and 81.9%, respectively). Abdominal pain was reported in 46.8% of the patients, while 53.2% did not. Fatigue was a common symptom, affecting 85.3% of patients, and polyuria and polydipsia were also prevalent, affecting 97.7% and 91.3%, respectively. Diabetic foot ulcers and pyelonephritis were rare, and only 0.6% of patients experienced each. Most of the patients had a short stay (80.2%) and 80.5% did not experience rebound ketones, with 17.5% reporting it once. Regarding the admission ketone status, 59% had < 3 mmol/L, while 41% had levels ≥ 3 mmol/L. Hypoglycaemia occurred in 18.7% of the patients and pneumonia in 3.4%. Urine glucose levels varied, with most patients having a result of + 2 (41.7%) or + 3 (34.2%), and the mean random blood sugar level was 445.5 ± 102.6. Furthermore, at admission, 90 patients (25.8%) presented with hyperkalemia (serum potassium ≥ 5.2 mEq/L), while 127 patients (37%) had hypernatremia (serum sodium ≥ 145 mEq/L). Finally, both the length of stay and the admission ketone status showed a distribution of short stays (80.2%) and ketone levels under 3 (59.0%) (Table 4).

Table 4.

Clinical characteristics of complications of patients with DKA admitted to DMCSH, northwest Ethiopia, from August 28, 2019 to September 30, 2024, (n = 348)

Variable	Category	Frequency	Percent
GCS score	3–8	119	34.2
	9–15	229	65.8
Nausea	Yes	4	1.2
	No	344	98.9
Vomiting	Yes	63	18.1
	No	285	81.9
Abdominal pain	No	185	53.2
	Yes	163	46.8
Fatigue	Yes	275	85.3
	No	51	14.7
Polyuria	Yes	340	97.7
	No	8	2.3
Polydipsia	Yes	299	91.3
	No	28	8.7
DM foot ulcer	No	346	99.4
	Yes	2	0.6
Pyelonephritis	Yes	2	0.6
	No	346	.4
Length of stay	Short	279	80.2
	Long	69	19.8
Rebound ketone	No	280	80.5
	One times	61	17.5
	Two and above times	7	2.0
Admitted ketone status	< 3	208	59.8
	≥ 3	140	40.2
Hypoglycaemia	Yes	65	18.7
	No	283	81.3
Pneumonia	Yes	12	3.4
	No	336	96.6
Urine glucose	Free	2	0.6
	+ 1	38	10.9
	+ 2	145	41.7
	+ 3	119	34.2
	+ 4	8	2.3
	Not measured	36	10.3
Random blood sugar	Mean (± SD)	445.5 ± 102.6
Serum potassium	< 3.5	61	17.6
	3.5–5.2	197	56.6
	> 5.2	90	25.8
Serum sodium	< 135	66	19
	135–145	153	44
	> 145	129	37
Admission ketone status in mmol/L	< 3	193	59.0
	≥ 3	134	41.0

Treatment-related factors of DKA patients

Among the patients studied, 81% had not experienced treatment complications, and 84.2% continued their drug treatment. Regarding the glycaemic status at discharge, 33.9% had hyperglycaemia, 60.1% achieved euglycaemia, and 6.0% had hypoglycaemia. Ceftriaxone was used in 7.5% of the patients, azithromycin was 4.0%, metronidazole 9.2% and other antibiotics, including vancomycin and amoxicillin, were administered to 10.3%. Enalapril, an ACE inhibitor, was used in 6.9%, while β-blockers, calcium channel blockers, and diuretics were prescribed in 13.2%. Moreover, thiamine (vitamin B1) was administered to 35.9% of patients with DKA, while potassium chloride (KCl) supplementation was provided to 80.5% of DKA patients. Fluid replacement with 0.9% normal saline was provided in 100% of the cases. All patients (100%) received an initial dose of regular insulin (RI), intravenously/intramuscularly, while 37.1% received RI combined with NPH insulin, 10.3% took metformin alone, 5.2% were treated with glibenclamide, 6.9% were on both metformin and glibenclamide, and 36.5% received NPH insulin alone (Table 5).

Table 5.

Treatment-related characteristics of patients with DKA in DMCSH, Ethiopia, from August 28, 2019 to September 30, 2024, (n = 348)

Variable	Category	Frequency	Percent
Treatment complication	Yes	66	19.0
	No	282	81.0
Drug discontinuation	Yes	55	15.8
	No	293	84.2
Glycaemia level at discharge	Hyperglycaemia	118	33.9
	Euglycaemia	209	60.1
	Hypoglycaemia	21	6.0
Antibiotics	Ceftriaxone 1gm IV BID	26	7.5
	Azithromycin 500 mg	14	4.0
	Metronidazole 50 mg IV BID	32	9.2
	Others*	36	10.3
ACE inhibitors	Enalapril 5 mg per oral daily	24	6.9
Others1	β-blockers, calcium channel blockers and diuretics	46	13.2
Thiamine administration	Yes	125	35.9
	No	223	64.1
KCL given	Yes	280	80.5
	No	68	19.5
Fluid replacement	0.9% normal saline solution	348	100
Anti-diabetic medications	RI 10 units IV and 10 units IM stat	348	100
	RI + NPH 0.5 units/kg/day BID	129	37.1
	Metformin 500 mg PO BID	36	10.3
	Glibenclamide 5 mg PO daily	18	5.2
	Metformin 500 mg PO BID + glibenclamide 5 mg PO daily	24	6.9
	NPH 0.5 units/kg/day BID	127	36.5

N.B BID, bis in die or twice a day, KG kilogram, NPH Neutral Protamine Hagedorn, it is intermediate-acting insulin, PO per os or peroral; RI, regular insulin, it is short-acting insulin

Others*: vancomycin, cloxacillin, amoxicillin, norfloxacin, metformin, augmentin

Treatment outcome of diabetic ketoacidosis

The results of this study showed that the prevalence of poor treatment outcomes among diabetic patients with DKA at Debre Markos General Specialized Hospital was 15.5% (95% CI: 11.5%, 19.3%). Among the 54 patients who experienced poor outcomes, 19 (5.5%) died, 11 (3.2%) left against medical advice, and 24 (6.9%) were referred to other institutions.

Factors associated with treatment outcome of DKA

In the binary logistic regression analysis, all variables were assessed, and those with a p-value < 0.25 including sex, residence, severe DKA, hypoglycaemia, comorbidities, infections, drug discontinuation, elevated ketone levels, random blood sugar (RBS), and treatment complications were considered candidates for multivariable analysis. In the multivariable logistic regression, seven factors were significantly associated with poor treatment outcomes in DKA: severity of DKA, presence of comorbidities, infections, drug discontinuation, elevated ketone levels, treatment complications, and prolonged hospital stay.

Patients who developed severe DKA were 1.48 times more likely to have poor DKA treatment outcomes compared to those with mild DKA status (AOR: 1.482, 95% CI: 1.324, 4.872). Similarly, the presence of comorbidities was associated with 1.8 times higher odds of poor outcomes compared to patients without comorbidities (AOR: 1.752, 95% CI: 1.215, 3.865). Patients who developed infections were also more likely to experience poor outcomes, with an AOR of 1.521 (95% CI: 1.362, 4.125). Regarding drug adherence, those who discontinued their prescribed diabetes medication were 2.12 times more likely to have poor treatment outcomes compared to those who continued their medication (AOR: 2.115, 95% CI: 1.245, 3.865). Treatment complications increased the likelihood of poor outcomes by 1.4 times compared to patients without complications (AOR: 1.356, 95% CI: 1.253, 4.125). Furthermore, patients with a ketone level ≥ 3 mmol/L at admission had a 21.3% higher risk of poor outcomes compared to those with levels < 3 mmol/L (AOR: 1.213, 95% CI: 1.052, 2.876). Finally, a hospital stay of 5 days or less was associated with a 1.3-fold increase in the odds of poor outcomes compared to stays longer than 5 days (AOR: 1.29, 95% CI: 1.022, 3.254) (Table 6).

Table 6.

Results of bivariable and multivariable logistic regression analyses of factors associated with treatment outcomes among DKA patients at DMCSH, northwest Ethiopia, from August 28, 2019 to September 30, 2024, (n = 348)

Variable	Category	Treatment outcome of DKA		COR 95%CI	AOR 95% CI
		Poor	Good
Sex	Male	17	116	0.705 (0.37, 1.31)	1.851(0.925, 4.124)
	Female	37	178
Residence	Urban	38	198	1.152 (0.611, 2.168)	1.845(0.725, 3.253)
	Rural	16	96
Severity of DKA	Mild	42	224
	Moderate	8	55	0.776(0.345, 2.747)	0.923(0.914, 2.912)
	Severe	4	15	1.422(1.150, 4.497)	1.482(1.324, 4.872)**
Hypoglycaemia	Yes	11	54	1.137 (0.551, 2.348)	1.25(0.864, 3.231)
	No	43	240
Comorbidity	Yes	5	24	1.148 (0.918, 3.153)	1.752(1.215, 3.865)*
	No	49	270
Infection	Yes	5	42	1.633(1.235, 3.521)	1.521(1.362, 4.125)**
	No	49	252
Drug discontinuation	Yes	13	42	1.902 (0.941, 3.847)	2.115(1.245, 3.865)**
	No	41	252
Ketone status at admission in mmol/L	< 3	31	177
	≥ 3	23	117	1.122(0.924, 2.020)	1.213(1.052, 2.876)*
RBS in mg/dl	< 500	15	102
	≥ 500	39	192	1.381 (0.72, 2.625)	1.923(0.876, 2.956)
Treatment complication	Yes	12	54	1.270 (0.927, 2.573)	1.356(1.253, 4.125)*
	No	42	240
Length of stay in hospital	≤ 5 days	48	231	2.182 (0.83, 5.330)	1.29 (1.022, 3.254) *
	> 5 days	6	63

1: reference, *p < 0.05, **p < 0.01

Discussion

This study found that the prevalence of poor treatment outcomes among diabetic patients with diabetic ketoacidosis (DKA) at the Debre Markos comprehensive specialized hospital was 15.5% (95% CI: 11.5%, 19.3%). This finding aligns with similar research carried out in Ethiopia (12%) [14], Kenya (17.2%) [13], Nigeria (16%) [44], Zambia16.66% [45] and Malaysia 17.6% [9]. However, the prevalence observed in Ethiopia is significantly lower than that reported in other countries, such as, Kenya (29.8%) [46], Cameroon (21.7%)[47]. These findings may differ due to several factors, such as variations in clinical presentation, the effectiveness of DKA detection and management, the prevalence of precipitating factors, and the capacity to address complications in various healthcare settings. It is important to note that the outcomes remain higher than those reported in some other regions with a well-established medical infrastructure. For example, studies conducted in Thailand (4.3%) [48] and Saudi Arabia (1.83%) reported significantly lower DKA-related mortality rates. Furthermore, in treatment-experienced facilities in Asia, Europe, and the Americas, a 5% mortality rate for DKA has been documented [49]. These findings suggest that poor outcomes remain a significant concern in the management of DKA within low-resource settings, although rates can vary depending on the population and the health facility. Among the 54 patients with poor treatment results, 19 (5.5%) died, 11 (3.2%) left against medical advice, and 24 (6.9%) were referred to other institutions. The mortality rate observed in this study is comparable to findings in Lusaka, Zambia [45], where the mortality rate in patients with DKA was 7.5%, but lower than the 9% mortality rate reported in a large U.S. cohort study [50]. The discrepancy in mortality rates could be attributed to factors such as differences in healthcare infrastructure, early intervention capabilities and patient management protocols.

One of the significant findings of this study was that patients with severe DKA had significantly higher odds of poor treatment outcomes compared to those with mild DKA. This is consistent with existing literature in Debretabour [20] and Saudi Arabia [8]which consistently shows that severe DKA is associated with worse outcomes, and also reports that severe DKA, defined by higher blood glucose and ketone levels, correlates with increased mortality and morbidity rates. The severity of DKA often reflects a delayed diagnosis and inadequate treatment, factors that can contribute to a worsened prognosis [49].

The presence of comorbidities was another strong predictor of poor treatment outcomes, with patients having 1.8 times higher odds of poor outcomes compared to those without comorbidities. This aligns with findings from [14, 38] who demonstrated that comorbid conditions such as hypertension, cardiovascular disease, and kidney dysfunction exacerbate the course of DKA and negatively impact patient recovery.

In addition, infections were found to be significantly associated with poor outcomes in this study, reinforcing findings from [20] where infection was a key factor contributing to DKA-related complications and mortality. Residing in a developing country where hygiene is a major concern may contribute to an increased risk of urinary tract infections (UTI) in people with diabetes mellitus (DM).

The study also identified discontinuation of drugs as a major risk factor for poor outcomes, with patients who discontinued their prescribed diabetes medications being 2.12 times more likely to experience poor treatment outcomes. This is consistent with the [50] guidelines, which highlight medication adherence as a crucial factor in preventing episodes of DKA and improving the prognosis of the patient. Patients who do not adhere to treatment protocols often have inadequate glycaemic control, increasing the risk of metabolic derangements such as DKA.

Furthermore, treatment complications significantly increased the odds of poor DKA treatment outcomes, which is consistent with findings [8], where complications such as electrolyte imbalances and fluid overload were associated with worse patient outcomes.

Another important factor positively significant with the poor outcome of DKA treatment in patients with DKA was ketone levels ≥ 3 at admission with a 21.3% increased risk of poor outcomes, which aligns with previous studies [51, 52] that demonstrate that high ketone levels correlate with severe metabolic acidosis and poorer recovery in patients with DKA.

Lastly, the relationship between length of hospital stay (LOS) and outcomes in patients with diabetic ketoacidosis (DKA) is multifaceted. This study indicated that patients with shorter hospital stays (≤ 5 days) have 1.3 times higher odds of poor outcomes compared to those with longer stays. This study was supported by Mekonnen et al. [20]. The possible explanations might be that patients with very short LOS may experience early deterioration or death shortly after admission which means that a short stay does not reflect uncomplicated recovery but rather indicates severe or rapidly progressing illness [53, 54]. Additionally, brief hospitalizations may occur when patients are discharged soon after perceived stabilization, without complete metabolic control, adequate diabetes education, or structured follow-up, increasing the likelihood of complications or readmission [10, 28, 48]. Conversely, longer hospital stays are often associated with more complex cases, multiple comorbidities, or hospital-acquired complications such as infections or electrolyte disturbances, all of which can negatively affect treatment outcomes [2]. Together, these patterns suggest a U-shaped relationship between LOS and DKA outcomes, where both very short and very long stays are linked to adverse results, though through different mechanisms. Achieving an optimal LOS long enough for stabilization and education, yet short enough to minimize complications is essential for improving patient outcomes [32].

The strength of this study offers valuable empirical evidence on the prevalence and associated factors of poor treatment outcomes among DKA patients in a real-world hospital setting in northwest Ethiopia, addressing a significant gap in regional data. The application of both bivariable and multivariable logistic regression analyses allowed the control of potential confounders and facilitated the identification of independent factors. In particular, the study highlights several clinically important risk factors including severe DKA, comorbidities, infections, discontinuation of treatment, elevated ketone levels, and treatment complications that can support the development of targeted interventions and improve hospital management strategies for patients with DKA.

However, the study has several limitations. The main limitation lies in its retrospective nature, which relies on existing medical records and may be subject to missing data, measurement errors, and unmeasured confounding, introducing the risk of information bias. Additionally, the study was conducted in a single comprehensive hospital, which may limit the generalizability of the findings to broader populations or different healthcare settings. Furthermore, certain potentially influential variables such as socioeconomic status, educational background, and long-term follow-up outcomes were not assessed, which could have provided a more comprehensive analysis. Finally, the classification of poor outcomes included heterogeneous categories (death, referral, and leave without medical advice), which may vary in clinical severity and underlying causes, but were analysed as a single group.

Conclusions

This study revealed that the prevalence of poor DKA treatment outcomes among diabetic patients at Debre Markos' comprehensive specialized hospital was high. The findings of this study underscore several critical factors that influence treatment outcomes in patients with DKA, including the severity of the condition, comorbidities, infections, drug adherence, complications during treatment, high levels of ketones (3 < 3) at admission, and a short hospital stay (5 days). Hence it is necessary to develop strategies to strengthen early detection, manage comorbidities and infections, ensure proper drug adherence, and effectively monitor complications. Also, it is imperative to improve hospital protocols, patient education, and structured follow-up to improve treatment outcomes.

Abbreviations

COR

Crude odds ratio

DKA

Diabetic ketoacidosis

DMCSH

Debre Markos Comprehensive Specialized Hospital

FBG

Fasting blood glucose level

HHS

Hyperglycaemic hyperosmolar state

IDF

International Diabetes Federation

NPH

Neutral Protamine Hagedorn insulin

Author contributions

MBY: Writing—original draft; writing—review & editing; conception; investigation; software; data curation; methodology; formal analysis; resources; and visualization. FL: Writing—original draft; writing—review & editing; conceptualization; investigation; data curation; methodology; supervision; project administration; funding acquisition; resources; and visualization. FM: Writing—original draft; writing—review & editing; investigation; data curation; supervision; validation; and visualization. MT: Writing—original draft; writing—review & editing; investigation; methodology; supervision; validation; and visualization. All authors read and approved the final manuscript.

Funding

There is no funding.

Availability of data and materials

The primary data contributed in this study were included in the article; Any clarity issues can be directly communicated with the corresponding author.

Declarations

Ethics approval and consent to participate

The current study included human respondents because this ethical clearance was obtained from the Research Ethics Review Committee, Department of Medicine and Health Sciences, Debre Markos University with reference number RCSTTD/522/01/17. A permission letter was written by the Amhara Public Health Institute and a cooperation letter was obtained from Debre Markos comprehensive specialized hospital to collect the data. To keep confidentiality, names and unique card numbers were not included in the data collection tool. After entering the data into the computer, the data were locked by password and were not disclosed to any person other than the principal investigator. All information collected from the patient chart was kept strictly confidential.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.