---
title: "Diabetic ketoacidosis admissions fell 6.9% annually in a 10-year Norwegian cohort"
id: "plos-one-18-trends-and-outcomes-in-diabetic-ketoacidosis-a-10-year-retrospective-study-at-a"
canonical_url: "https://medichelpline.com/clinical-feed/plos-one-18-trends-and-outcomes-in-diabetic-ketoacidosis-a-10-year-retrospective-study-at-a"
content_type: "clinical_feed_article"
specialty: "Endocrinology"
source_name: "PLOS ONE (Medicine)"
source_url: "https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0362629"
published_at: "2026-11-13T14:00:00.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Diabetic ketoacidosis admissions fell 6.9% annually in a 10-year Norwegian cohort
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/plos-one-18-trends-and-outcomes-in-diabetic-ketoacidosis-a-10-year-retrospective-study-at-a
- **Specialty:** [Endocrinology](https://medichelpline.com/clinical-feed/endocrinology.md)
- **Primary Source:** PLOS ONE (Medicine)
- **Source URL:** [Original Journal Publication](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0362629)
- **Published At:** 2026-11-13T14:00:00.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- Retrospective review of adult admissions with **diabetic ketoacidosis** (DKA) at a single Norwegian university hospital from 1 Aug 2012 to 31 Aug 2022. Electronic medical records were systematically reviewed for clinical presentation, treatment and outcomes. - The cohort comprised 178 unique participants and 306 admissions. Median age at admission was 34 years (IQR 22–51). Median **HbA1c** was 88 mmol/mol (IQR 72–105). - Monthly mean admissions fell from 3.20 in 2012 to 1.875 in 2022. Poisson regression estimated a 6.9% annual decline (rate ratio 0.931; 95% CI 0.896–0.968; p 1 admission and accounted for 55% of all admissions. A small subset had very frequent episodes (one patient had 24 admissions). - Psychiatric illness (adjusted OR 2.827; 95% CI 1.070–7.468) and receipt of disability benefits (adjusted OR 4.941; 95% CI 1.576–15.490) were associated with having >1 lifetime DKA admission. Higher age at index admission reduced odds of recurrent lifetime admissions (aOR 0.727 per 10 years). - Prior DKA, suspected treatment non-adherence and psychiatric illness were associated with higher readmission rates in follow-up analyses. No clear association was observed between index-admission **HbA1c** and recurrence. - In-hospital mortality was 1.6%; five-year cumulative mortality from first observed admission was 14.7%. Alcohol or substance misuse (HR 2.499; 95% CI 1.191–5.245) and higher age at index admission (HR 1.874 per 10 years) predicted mortality. - Treatment initiation followed local guidelines in most admissions (isotonic fluids 98%, insulin 96%), but potassium infusion prior to insulin was started in only 55% of eligible admissions. - Limitations included missing ketone measurements for 111 admissions, single-centre retrospective design, and restricted access to individual-level data under local protections.
## Clinical Analysis & Structured Key Points
Trends and outcomes in diabetic ketoacidosis: A 10-year retrospective study at a Norwegian university hospital | PLOS One Browse Subject Areas ? Click through the PLOS taxonomy to find articles in your field. For more information about PLOS Subject Areas, click here . Article Authors Metrics Comments Media Coverage Reader Comments Figures Figures Abstract Aims To describe trends in diabetic ketoacidosis admissions and examine factors associated with repeated admissions and mortality among adults living with type 1 diabetes. Materials and methods Diabetic ketoacidosis admissions among adults at a Norwegian university hospital between 1 August 2012 and 31 August 2022 were retrospectively reviewed using electronic medical records. Results A total of 178 participants with 306 admissions were included. Median [IQR] age at admission was 34 years [22–51]. Median HbA1c was 88 mmol/mol [72–105]. Diabetic ketoacidosis admission frequency decreased by an estimated 6.9% annually (RR 0.931; 95% CI 0.896–0.968; p < 0.001) over the 10-year period. 40 participants (22.5%) had more than one admission during the study period. Psychiatric illness (aOR 2.827; 95% CI 1.070–7.468; p = 0.036) and receipt of disability benefits (aOR 4.941; 95% CI 1.576–15.490; p = 0.006) were associated with experiencing more than one lifetime admission. Prior diabetic ketoacidosis, suspected treatment non-adherence and psychiatric illness were associated with higher readmission rates during follow-up. In-hospital mortality was 1.6%, and 5-year cumulative mortality was 14.7%. Alcohol or substance misuse (HR 2.499; 95% CI: 1.191–5.245; p = 0.015) and higher age at the index admission (HR 1.874; 95% CI: 1.511–2.324; p < 0.001, per 10 years) were associated with mortality in the follow-up period. Conclusions Diabetic ketoacidosis admissions declined substantially over the 10-year study period despite growth in the hospital catchment area population, suggesting that diabetes care has improved. Repeated admissions were common and associated with psychosocial vulnerability. In-hospital mortality was low, but five-year mortality was considerable, supporting targeted follow-up after an episode of diabetic ketoacidosis. Citation: Soleim SB, Løvås K, Søfteland E, Ueland GÅ (2026) Trends and outcomes in diabetic ketoacidosis: A 10-year retrospective study at a Norwegian university hospital. PLoS One 21(11): e0362629. https://doi.org/10.1371/journal.pone.0362629 Editor: Vipula Rasanga Bataduwaarachchi, University of Colombo Faculty of Medicine, SRI LANKA Received: December 9, 2025; Accepted: October 27, 2026; Published: November 13, 2026 Copyright: © 2026 Soleim et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The individual-level dataset contains sensitive, potentially identifiable health information derived from hospital medical records and cannot be made publicly available under the applicable data protection requirements and the approved arrangements for this study. Data are stored securely at Helse Bergen HF, with a separate linkage key and access restricted to the project management team. Enquiries regarding the ethical requirements for access may be directed to the Regional Committee for Medical and Health Research Ethics, Western Norway (REK vest; rek-vest@uib.no ), quoting reference 511-576. Any disclosure of data would require assessment by Helse Bergen HF and the necessary ethical and legal approvals. Funding: SBS received funding from the “Johan Selmer Kvanes Legat til forskning og bekjempelse av sukkersyken”(website: https://www.legeforeningen.no/om-oss/fond-og-legater/legater/johan-selmer-kvanes-legat-til-forskning-og-bekjempelse-av-sukkersyke/ , grant number: SAK2023005392). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. 1. Introduction Diabetic ketoacidosis (DKA) remains a serious complication of type 1 diabetes despite substantial advances in diabetes care, including continuous glucose monitoring, insulin pumps and automated insulin delivery [ 1 ]. Recent studies have reported increasing DKA admission rates in the UK and USA [ 2 , 3 ], whereas declining rates have been observed in Denmark [ 4 ]. These contrasting trends highlight the importance of examining how DKA admission trends have evolved in different healthcare settings during a period of changing diabetes management. The clinical burden of DKA extends beyond the acute admission. Although in-hospital mortality from ketoacidosis remains low in most high-resource settings [ 1 ], DKA is associated with increased long-term mortality, and repeated episodes account for a substantial portion of admissions and may be a marker of particular vulnerability [ 5 , 6 ]. Understanding the clinical and psychosocial characteristics associated with repeated admissions and mortality may help identify patients who could benefit from targeted follow-up and prevention strategies. To understand the impact and burden of DKA, admission trends and their development over time need to be examined along with the circumstances of individual admissions and subsequent patient outcomes. Detailed clinical cohorts can provide this complementary perspective. We therefore examined trends in ketoacidosis admissions at a Norwegian university hospital from 2012 to 2022 and characterised the clinical presentation and subsequent outcomes. We also assessed factors associated with recurrent admissions and death. 2. Materials and methods We conducted a retrospective cohort study of people with type 1 diabetes admitted with ketoacidosis to the adult service at Haukeland University Hospital in Bergen, between 1 August 2012 and 31 August 2022. The hospital’s catchment area comprised approximately 288 000 residents, 232 000 of whom were 16 or older, in 2022 [ 7 , 8 ]. Applying the national type 1 diabetes prevalence estimate of 0.72% [ 9 ] among adults aged 18 years or older to the catchment population aged 16 or older gives a rough estimate of approximately 1700 people with type 1 diabetes in 2022. Following approval from the Regional Committee for Medical and Health Research Ethics (REK no. 511 576), eligible persons were identified through an index search in the hospital electronic medical records. 2.1. Study population The search of all discharges with an ICD-10 diagnosis of either E10.0 (type 1 diabetes with coma) or E10.1 (type 1 diabetes with ketoacidosis) yielded 330 unique persons with 504 admissions. Participants were informed by letter and given the opportunity to opt out of participation. Participants were excluded if they lived abroad, withdrew consent or were under the age of 18 at the time of inclusion in 2023 (n = 27). Following review of diabetes classification and evaluation of clinical and biochemical findings, 178 persons and 306 admissions were included. The American Diabetes Association diagnostic criteria for type 1 diabetes and diabetic ketoacidosis [ 1 , 10 ] were employed. Ketone values were unavailable in 111 admissions, and five additional admissions had documented values below the diagnostic threshold. Such admissions were retained only if individual chart review, the treating endocrinologist’s assessment, the clinical presentation and response to DKA treatment were consistent with a DKA diagnosis. Admissions with a plausible alternative explanation for acidosis were excluded. 2.2. Data collection Hospital medical records for individual participants were systematically reviewed by one investigator (SBS) between January and December 2024, with access to identifiable patient information. Clinician-noted demographics, lifestyle factors, diabetes history, complications, somatic and psychiatric comorbidities, history of alcohol or substance misuse, treatment adherence, insulin regimen, presenting clinical parameters, treatment initiated at admission, and short- and long-term outcomes were recorded. The electronic medical records are synchronised with the Norwegian National Population Register, yielding date but not cause of death for deceased participants. The cutoff time for admission ascertainment was August 31 st , 2022, and for death November 26 th , 2024. 2.3. Definitions The most recent diabetes classification recorded in the electronic medical records was used, and C-peptide and antibody levels were not checked. Psychiatric comorbidities were registered in the case of psychiatric illnesses that had required pharmacological or psychological therapy, including anxiety, depression, psychosis and eating disorders, registered in the hospital electronic medical records at the index admission. Alcohol or substance misuse was defined as a previously documented or currently suspected overconsumption of alcohol or psychoactive substances at the index admission. Cognitive impairment was registered if it had been noted in the electronic medical records at any point prior to the index admission, regardless of cause. Suspected treatment non-adherence was defined as several documented missed follow-up appointments or known or suspected medication non-compliance, based on records available at the index admission. Receipt of disability benefits (uføretrygd), irrespective of cause, was noted when explicitly documented at the index admission. The presumed cause of ketoacidosis was estimated by the treating physician and/or by the reviewer based on electronic medical record entries. Persons with and without known diabetes mellitus prior to admission were examined separately and are referred to as having pre-existing or new-onset diabetes mellitus. The ketone criterion was a capillary blood ketone concentration ≥ 3 mmol/L or urine ketones ≥ 2 + . Time to biochemical normalisation was defined as time from first recorded arterial blood gas analysis until the first arterial blood gas analysis with a recorded bicarbonate ≥18 mmol/L. For mortality analyses, time was defined as months from the index admission in the study period to death or censoring. 2.4. Outcomes Outcome measures included temporal trends in DKA admission counts, time to biochemical normalisation, length of hospital stay, recurrent DKA admissions, in-hospital mortality and all-cause mortality following the index admission. 2.5. Statistical analyses Continuous variables are reported as medians with interquartile ranges (IQR), and categorical variables as n (% of available observations). Laboratory measurements are reported in SI units. Effect estimates are reported as rate ratios (RR) for Poisson and negative binomial regressions, odds ratios (OR) for logistic regression, hazard ratios (HR) for Cox regression, and ratios of expected time for gamma regression, all with 95% confidence intervals. A two-sided p-value <0.05 was considered significant. Covariates for each multivariable analysis were selected on clinical grounds and entered simultaneously. Participant-level models used characteristics present at the index admission, whereas admission-level models used characteristics recorded for each individual admission. Sensitivity analyses restricted to admissions meeting the documented ketone criterion were performed to assess robustness of findings. Hospital monthly admission counts were analysed using Poisson regression with a log link and calendar time in years. DKA recurrence analyses included patients with previously known diabetes. A binary logistic regression model examined risk of multiple lifetime admissions, while a Cox regression model was used to examine time from discharge to first recurrent admission in the study period, censoring at death or August 31 st 2022. In addition, a negative binomial regression analysis was performed to model all subsequent readmissions using log-transformed person-time from index discharge to death or August 31 st 2022 as an offset. The Cox and negative binomial regression analyses of recurrence were restricted to participants who survived their index admission. Five-year mortality was estimated with a Kaplan-Meier analysis from the first observed admission. Cox regression of all-cause mortality used all available follow-up information until the 26 th of November 2024. An additional Cox model examined recurrence status as a time-varying covariate for mortality. Time to biochemical normalisation and length of stay were analysed using gamma regression with a log link and generalised estimating equations to account for repeated admissions. Covariate sets, missingness documentation, model diagnostics and sensitivity analyses are detailed in the Supplementary appendix. For the recurrence analyses, multiple imputations were performed, while the rest of the analyses used complete cases. All analyses were performed using IBM SPSS Statistics v31. 3. Results In total, 178 unique participants with 306 admissions were included. Baseline characteristics and diabetes history are summarised in Table 1 . Download: PNG larger image TIFF original image Table 1. Baseline characteristics and diabetes history at first observed DKA admission. https://doi.org/10.1371/journal.pone.0362629.t001 3.1. Clinical features Clinical features during the admission are summarised in Table 2 and Fig 1 . The most common identified precipitating cause was forgotten or voluntarily omitted insulin dose(s). Notably, a history of psychiatric illness was present in 40% of admissions in which insulin had been forgotten or intentionally omitted. In 83 admissions (30.1% of admissions among participants with pre-existing T1D) the patient was using an insulin pump, and 24 admissions were due to pump failure. In three of the admissions due to pump failure, the participant was using an automated insulin delivery device. Download: PNG larger image TIFF original image Table 2. Clinical characteristics, treatment and outcomes of DKA admissions. https://doi.org/10.1371/journal.pone.0362629.t002 Download: PNG larger image TIFF original image Fig 1. Precipitating causes of diabetic ketoacidosis. Distribution of presumed precipitating causes across 306 diabetic ketoacidosis admissions among 178 participants. Percentages were calculated using all 306 admissions and rounded for presentation. https://doi.org/10.1371/journal.pone.0362629.g001 3.2. Biochemistry and treatment Median HbA1c at admission was 88 mmol/mol (IQR 72–105). In only 10 admissions (3.3%) HbA1c was 53 mmol/mol or lower. Using a 300 mOsm/kg cutoff, concomitant hyperosmolality was present in 44% of admissions. 58% of ketoacidoses were characterised as severe. In the majority of admissions, the diagnosis was recognised in the emergency room and treatment with isotonic fluids (98%) and insulin (96%) was initiated according to local treatment guidelines. Hospital guidelines required potassium infusion prior to insulin when serum potassium was ≤ 5 mmol/L, but this was only initiated in 55% of eligible admissions. For biochemical parameters at admission, see S1 Appendix , Table S7. 3.3. Temporal trend in DKA admissions Mean monthly DKA admission counts decreased from 3.20 in 2012 to 1.875 in 2022. In a Poisson regression, the decline was estimated at 6.9% per year (RR 0.931; 95% CI 0.896–0.968; p < 0.001), corresponding to a model-estimated reduction of approximately 51% over the 10-year follow-up period ( Fig 2 ). Download: PNG larger image TIFF original image Fig 2. Temporal trend in diabetic ketoacidosis admissions. Grey circles show the observed mean monthly number of admissions in each calendar year. Data for 2012 cover August-December, and data for 2022 cover January-August. The red line shows the fitted trend from Poisson regression using all 121 monthly admission counts. The estimated annual rate ratio was 0.931 (95% CI 0.896-0.968; p < 0.001). DKA, diabetic ketoacidosis. https://doi.org/10.1371/journal.pone.0362629.g002 3.4. Time to biochemical normalisation Each successive calendar year was associated with a 2.8% shorter time until resolution (p = 0.025) for a cumulative relative decrease of 24% during the 10-year period. Male sex, a higher pH and higher glucose at admission were associated with shorter time to resolution ( S1 Appendix , Table S5A). 3.5. Length of hospital stay New-onset diabetes and macrovascular disease were associated with 151% and 87% longer hospital stays. For each successive calendar year, length of stay decreased by 3.7% ( S1 Appendix , Table S5B). 3.6. Recurrent admissions Twenty-two percent of the participants had more than one admission during the 10-year follow-up period, and this group accounted for 55% of all admissions. A small subset of five persons (2.8%) had six or more episodes; one experienced 24 admissions. Among the 30 participants in whom a ketoacidosis admission marked the debut of their type 1 diabetes, two were admitted twice with ketoacidosis, and two were admitted thrice in the study period. Among 148 participants with previously known diabetes, 63 had more than one DKA admission at any point prior to or in the study period. In a multivariable binary logistic regression, psychiatric illness (aOR 2.827; 95% CI 1.070–7.468; p = 0.036) and receipt of disability benefits (aOR 4.941; 95% CI 1.576–15.490; p = 0.006) were associated with experiencing more than one lifetime admission, while higher age at the index admission in the study period was associated with decreased odds (aOR 0.727; 95% CI 0.565–0.935; p = 0.013, per 10 years of age) ( Fig 3A ). Download: PNG larger image TIFF original image Fig 3. Factors associated with multiple lifetime diabetic ketoacidosis admissions and all-cause mortality. (A) Adjusted odds ratios from binary logistic regression for at least two versus one documented lifetime diabetic ketoacidosis admission among participants with pre-existing type 1 diabetes (n = 148). Estimates were pooled across 50 imputed datasets. (B) Adjusted hazard ratios from Cox regression for all-cause mortality from the first observed admission through 26 November 2024 (n = 178). Covariates were assessed at the first observed admission. Squares indicate point estimates, and horizontal lines indicate 95% confidence intervals. The horizontal axis uses a logarithmic scale. Age and HbA1c effects are expressed per 10 years and 10 mmol/mol, respectively. DKA, diabetic ketoacidosis; OR, odds ratio; HR, hazard ratio; CI, confidence interval; HbA1c, glycated haemoglobin. https://doi.org/10.1371/journal.pone.0362629.g003 Of the 143 participants with previously known diabetes who survived their index admission, 36 had a readmission in the study period, accounting for 122 readmissions in total. A secondary analysis with a Cox regression model analysing time to first readmission in the study period showed that documented DKA prior to the index admission was associated with a higher hazard of first readmission (HR 3.058; 95% CI 1.469–6.365; p = 0.003). A negative binomial model accounting for all subsequent DKA admissions in the study period showed that higher rates of recurrence within the 10-year period were associated with having experienced a previous episode (RR 3.755; 95% CI 1.790–7.877; p < 0.001), as well as with suspected treatment non-adherence (RR 4.402; 95% CI 1.835–10.570, p < 0.001) and psychiatric illness (RR 2.643; 95% CI 1.096–6.373; p = 0.030). No clear association with index-admission HbA1c was observed. Model results are reported in full in the S1 Appendix , Tables S3A and S3B. 3.7. Short-term outcomes and mortality Following 50 (16.6%) admissions, the participant was readmitted for any cause within 30 days of discharge. Five admissions had a fatal outcome: in four, active treatment was withdrawn; in three due to anoxic brain injury caused by pre-hospital cardia
## Related Clinical Research

- [PMOS/PCOS and Risk Signals for Alzheimer’s Disease and Dementia: Evidence Review](https://medichelpline.com/clinical-feed/endocrine-news-0-could-pmos-pcos-be-linked-to-alzheimer-s-disease-and-dementia.md)
- [Undiagnosed type 2 diabetes presenting as DKA with hypertriglyceridemia-induced acute pancreatitis](https://medichelpline.com/clinical-feed/cmaj-1-undiagnosed-type-2-diabetes-presenting-as-diabetic-ketoacidosis-with.md)
- [Insulet and Calm Partner to Offer Mindfulness Tools for People Living with Diabetes](https://medichelpline.com/clinical-feed/endocrine-news-0-pharma-friday-july-17-2026.md)
- [Diabetes and Suicide Risk in Black Males, Including Sexual Minority Males: Clinical Review and Pri](https://medichelpline.com/clinical-feed/pubmed-42803858.md) (DOI: 10.1007/s11892-026-01647-0)
- [Higher Ultra-Processed Food Intake Linked to Greater Risk of Multiple Chronic Diseases](https://medichelpline.com/clinical-feed/medical-news-today-0-every-extra-100-grams-of-ultra-processed-foods-tied-to-greater-disease-risk.md)

## Navigation
- [← Back to Endocrinology Feed](https://medichelpline.com/clinical-feed/endocrinology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.