Age and sex differences in diabetic ketoacidosis outcomes in type 1 diabetes: a national cohort study

Abstract

Background

Despite therapeutic advances, diabetic ketoacidosis (DKA) hospitalizations continue to increase. The role of demographic factors, including age and sex in the outcomes of DKA remains underexplored. We aimed to investigate the interaction between age and sex on clinical outcomes, resource utilization and mortality in patients with type 1 diabetes hospitalized for DKA.

Methods

We conducted a retrospective cohort study using the National Inpatient Sample from 2016 to 2021. We identified adult hospitalizations with DKA in patients with type 1 diabetes, stratified into three age groups: 18–44, 45–64 and ≥65 years. Multivariable logistic regression models were used to analyze the association between sex and the primary outcome of in-hospital mortality, and secondary outcomes including acute kidney injury (AKI) and sepsis, adjusting for patient and hospital characteristics.

Results

Across all age groups, female sex was independently associated with significantly lower odds of AKI (aOR 0.56 in ages 18–44; 0.71 in 45–64; 0.79 in ≥65) but higher odds of sepsis (aOR 1.66, 1.31 and 1.17, respectively; all p<0.05). In young adults (18–44), women had significantly lower adjusted odds of mortality (aOR 0.72, 95% CI 0.60 to 0.86). This mortality benefit was not observed in middle-aged or older adults. Prepandemic (2016–2019), mortality trends diverged by sex and age, with rates increasing for young men but decreasing for young women. The pandemic (2020–2021) precipitated a sharp mortality increase across all age–sex groups, most dramatically in young men (from 0.48% to 0.89%).

Conclusion

Our study showed that age and sex were closely linked to acute complications and in-hospital death. Women had lower odds of AKI but higher odds of sepsis than men, with a survival advantage limited to young adults, and mortality rose during the COVID-19 years for both sexes, especially in young men.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Hospitalizations for diabetic ketoacidosis (DKA) are increasing despite advances in diabetes technologies and care. Prior studies suggest outcomes may differ by age and sex, but contemporary nationally representative evidence that explicitly evaluates age–sex interaction—while accounting for patient, socioeconomic and hospital-level factors has been limited.

WHAT THIS STUDY ADDS

• In a national cohort of adults with type 1 diabetes hospitalized for DKA (2016–2021), women had consistently lower adjusted odds of acute kidney injury across all age groups but higher adjusted odds of sepsis. A survival advantage for women was observed only in younger adults (18–44 years). In-hospital mortality increased sharply during 2020–2021 across all age–sex strata, most prominently among younger men.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Risk stratification and quality-improvement efforts for DKA should incorporate both age and sex, with particular attention to younger men as a high-risk group and increased sepsis vigilance among women. These findings should motivate future work to clarify mechanisms and care-delivery factors that drive sex-specific renal and infectious complications, including studies linking clinical severity and outpatient access to in-hospital outcomes.

Introduction

Diabetic ketoacidosis (DKA) is a life-threatening complication of type 1 diabetes and a leading cause of diabetes-related hospitalization in young and middle-aged adults.¹ Despite advances in insulin formulations, delivery systems and glucose monitoring, DKA remains one of the most serious acute complications of type 1 diabetes and a persistent cause of preventable morbidity and mortality.²,⁴ In the USA, population-based surveillance has shown that DKA hospitalization rates rose during 2009–2014 after earlier declines, with the highest burden in younger adults.¹ Rising DKA admission rates have also been reported outside the USA, including in England (1998–2013), suggesting that increasing DKA hospitalization is not uniquely a US phenomenon.⁵ However, factors such as high out-of-pocket insulin costs and cost-related insulin underuse may disproportionately contribute to preventable DKA in some populations in the USA.⁶

The conventional medical paradigm treats DKA as a metabolically uniform emergency, with standardized protocols applied universally regardless of patient demographics. Yet emerging evidence reveals heterogeneity in presentation, complications and outcomes that correlates systematically with age and sex. Women with diabetes face a greater risk of coronary heart disease mortality compared with men with the same condition, while sex-specific differences in acute kidney injury (AKI), sepsis susceptibility and treatment response have been documented across different clinical contexts.⁷,⁹ The relationship between biological sex and chronological age creates even more complex phenotypic variants, as reproductive hormones, comorbidity accumulation and healthcare access patterns evolve across the adult lifespan.

The urgency of addressing these knowledge gaps has been amplified by the COVID-19 pandemic, which has exposed how social determinants of health intersect with biological vulnerabilities to create disparate outcomes. Diabetes is one of the most common comorbidities among hospitalized COVID-19 patients, with a significantly increased risk of death compared with those without diabetes.¹⁰ The pandemic period witnessed fundamental disruptions to diabetes care delivery, including delayed clinic visits, reduced laboratory monitoring and increased reliance on telemedicine.^{11 12} Although prior studies have evaluated associations of age and sex with in-hospital DKA outcomes, contemporary US nationally representative analyses that explicitly test age–sex interaction and incorporate social and hospital-level determinants remain limited.¹³

We aimed to examine the interaction between age, sex and outcomes of DKA in patients with type 1 diabetes mellitus. We hypothesized that age-sex interactions, rather than age or sex alone, may determine DKA outcomes in adults with type 1 diabetes.

Methods

Study design and data source

We retrospectively studied the Healthcare Cost and Utilization Project National Inpatient Sample (NIS) from 2016 to 2021, the largest publicly available, all-payer database of hospitalizations in the USA. The NIS is a 20% stratified sample of discharges from community hospitals, with discharge-level weights that yield national estimates. Each record contains up to 40 diagnosis codes, 25 procedure codes and detailed patient and hospital characteristics. Because the NIS is fully deidentified, the study was exempt from institutional review board review.

Study population

We first identified all adult (≥18 years) hospitalizations with a principal or secondary diagnosis of DKA using ICD-10-CM codes E10.1x, E11.1x or E13.1x and subsequently selected cohort with only type 1 diabetes-related DKA (E10.1x). We restricted the analytic cohort to type 1 diabetes-related DKA to reduce etiologic heterogeneity (such as ketosis-prone type 2 diabetes or sodium-glucose transport 2 inhibitor-associated ketoacidosis). Hospitalizations with missing data on age, sex, race/ethnicity, income, insurance type, elective admission status or mortality are excluded from this study. We first divided the hospitalizations into three different cohorts based on age group: (1) age 18–44, (2) age 45–64 and (3) age 65 years and above. Following that in each cohort, we compared the characteristics between female and male, with male being the reference group. Age was modeled categorically (18–44, 45–64, ≥65 years) to reflect clinically interpretable life-stage groupings, allow non-linear associations and facilitate interaction testing with sex. Covariates were selected a priori on clinical and health-services grounds and included: self-reported race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, Asian/Pacific Islander, Native American, other), primary payer (Medicare, Medicaid, private, self-pay, no charge, other), median household-income quartile for patient ZIP code, hospital bed size, teaching status (rural, urban non-teaching, urban teaching) and hospital region.

Outcomes

The primary outcome was in-hospital mortality. Secondary outcomes were AKI, AKI requiring dialysis or renal replacement therapy, invasive mechanical ventilation, sepsis, vasopressor use, length of stay and total hospitalization cost. AKI, AKI requiring dialysis or renal replacement therapy, sepsis, ventilation and vasopressor therapy were identified with ICD-10 codes.

Statistical analysis

We applied discharge-level weights to produce nationally representative estimates, and robust SEs were calculated with the Taylor linearization method. Baseline characteristics and outcomes were compared between sex and DKA admissions among three different age cohorts with χ² tests for proportions and linearized Wald or design-based F tests for continuous variables. Continuous variables are reported as weighted means with SD while categorical variables as weighted counts with percentages. To explore independent associations with mortality, we fitted separate mixed-effects logistic regression models. All models were screened for race/ethnicity, primary payer, median household-income quartile for patient ZIP code, hospital bed size, hospital teaching status and hospital region as well as comorbidities including congestive heart failure, atrial fibrillation, coronary artery disease, hypertension, smoking, hypothyroidism, chronic kidney disease, chronic liver disease, obesity, alcohol abuse, drug abuse, depression, chronic steroid use and cancer. Variables associated with the outcome in bivariable analyses at threshold of 0.20 were retained as candidate confounders to minimize residual confounding. Multicollinearity was assessed with variance-inflation factors. Adjusted ORs (aORs) or mean differences with 95% CIs were reported. A two-sided p value <0.05 denoted statistical significance. We conducted all statistical analyses using STATA V.18.0 and R V.4.5.

Results

A total of 799 340 DKA hospitalizations involving patients with type 1 DM were identified, of which 52% of these were women (n=413 710)(online supplemental Table S1). We stratified these patients by age into three separate cohorts: (1) age 18–44 years, (2) age 45–64 years and (3) age ≥65 years.

Age 18–44 years

In the youngest age group, significant differences by sex were identified across demographic and clinical variables. Racial distribution varied significantly (p<0.001), with a higher proportion of women being non-Hispanic black (27.1% vs 23.9%) and a slightly lower proportion non-Hispanic white (56.6% vs 59.2%) compared with men. Hospital bed size utilization differed significantly (p=0.0064), with women more frequently admitted to large hospitals (48.0% vs 46.9%) (online supplemental Table S2). Urban teaching hospitals were more commonly used by women (67.8% vs 66.2%, p<0.001). Income quartile distribution also significantly differed (p=0.0004), with more women in the lowest quartile (39.7% vs 38.6%). Regional differences were significant (p<0.001), largely due to higher representation of women in the South (44.8% vs 42.9%). Significant sex disparities existed in primary insurance coverage (p<0.001), notably higher Medicaid coverage among women (48.0% vs 41.5%).

Regarding comorbidities, men had significantly higher prevalence rates of atrial fibrillation (1.0% vs 0.6%), hypertension (29.6% vs 26.7%), smoking (37.6% vs 25.6%), chronic liver disease (4.3% vs 3.6%), alcohol abuse (7.7% vs 3.0%) and drug abuse (23.8% vs 17.2%), all p<0.001. Women had significantly higher prevalence of hypothyroidism (11.0% vs 4.3%, p<0.001), obesity (7.0% vs 4.4%, p<0.001) and depression (22.8% vs 14.8%, p<0.001).

Age 45–64 years

In the middle-aged group, racial distribution differences remained significant by sex (p=0.0015), with a greater proportion of non-Hispanic white women (72.6% vs 70.5%) and fewer non-Hispanic black women (17.8% vs 19.4%) compared with men. Differences in hospital bed size utilization were not significant (p=0.4107), but hospital teaching status differed significantly (p=0.0002), with more men admitted to urban teaching hospitals (68.0% vs 65.8%) (online supplemental Table S2). Income quartile differences approached but did not reach significance (p=0.0618). Significant differences in primary insurance were observed, with women more frequently covered by Medicare (33.3% vs 31.0%) and private insurance (32.1% vs 28.3%), p<0.001.

Significant sex differences in comorbidities persisted. Women had significantly higher prevalence of congestive heart failure (13.4% vs 12.3%), hypothyroidism (26.3% vs 10.8%), obesity (10.5% vs 6.1%) and depression (28.3% vs 18.9%), all p<0.001. Conversely, men had significantly higher prevalence of atrial fibrillation (5.6% vs 4.2%), coronary artery disease (22.3% vs 20.8%), hypertension (67.4% vs 64.1%), smoking (34.3% vs 26.7%), chronic kidney disease (27.6% vs 25.4%), chronic liver disease (6.8% vs 5.4%), alcohol abuse (14.1% vs 5.5%) and drug abuse (15.6% vs 10.9%), all p<0.001.

Age ≥65 years

Among older adults (≥65 years), no significant sex differences were observed in racial distribution (p=0.2882), hospital bed size utilization (p=0.9814) or hospital teaching status (p=0.7053) (online supplemental Table S2). Income quartile distribution was similar (p=0.7661), although regional differences were significant (p=0.0063), primarily due to higher representation of women in the South (38.3% vs 35.5%) and Northeast (18.7% vs 17.6%).

For comorbidities, men had significantly higher prevalence rates of congestive heart failure (24.9% vs 23.1%, p=0.0472), atrial fibrillation (18.2% vs 15.1%), coronary artery disease (44.7% vs 36.4%), smoking (11.9% vs 8.6%), chronic kidney disease (40.0% vs 33.5%), alcohol abuse (6.7% vs 1.8%) and drug abuse (3.6% vs 1.5%), all p<0.001 except congestive heart failure. Women consistently had higher prevalence rates of hypothyroidism (39.9% vs 19.0%), obesity (8.6% vs 6.9%) and depression (22.9% vs 15.2%), all p<0.005.

In-hospital events/outcomes between women and men hospitalized with DKA

Age 18–44 years

For the youngest age group (18–44 years), the population included 286 885 males and 307 345 females. Unadjusted results showed men had a higher mortality rate of 0.5% (1445 cases) compared with 0.4% (1135 cases) in women, with a statistically significant p value of <0.001 (table 1, online supplemental Figure S1). Men also had a significantly higher incidence of AKI, at 40.8% (117 025 cases) vs 27.7% (85 115 cases) in women, with p<0.0001. Conversely, women had a higher rate of sepsis, at 7.2% (22 005 cases) compared with 4.5% (12 890 cases) in men, also with p<0.0001. There were no significant differences in the rates of AKI requiring dialysis, with both groups at 0.6% (1820 cases in men, 1885 in women, p=0.6455) or vasopressor use, both at 0.4% (1255 cases in men, 1175 in women, p=0.1434). Men had a slightly higher need for ventilation, at 2.9% (8320 cases) compared with 2.7% (8305 cases) in women, with p=0.0437, indicating marginal significance.

Table 1. Unadjusted in-hospital events and outcomes stratified by age and sex.

Age group	Variables	Men (n=286 885)	Women (n=307 345)	Total (n=594 230)	P value
Age 18–44	Mortality	1445 (0.5 %)	1135 (0.4 %)	2580 (0.4 %)	0.0004
	AKI	117 025 (40.8 %)	85 115 (27.7 %)	202 140 (34.0 %)	<0.0001
	AKI requiring dialysis	1820 (0.6 %)	1885 (0.6 %)	3705 (0.6 %)	0.6455
	Ventilation	8320 (2.9 %)	8305 (2.7 %)	16 625 (2.8 %)	0.0437
	Sepsis	12 890 (4.5 %)	22 005 (7.2 %)	34 895 (5.9 %)	<0.0001
	Vasopressor use	1255 (0.4 %)	1175 (0.4 %)	2430 (0.4 %)	0.1434
Age 45–64	Mortality	1485 (1.8 %)	1265 (1.6 %)	2750 (1.7 %)	0.1083
	AKI	46 080 (57.1 %)	37 470 (47.5 %)	83 550 (52.4 %)	<0.0001
	AKI requiring dialysis	1340 (1.7 %)	1100 (1.4 %)	2440 (1.5 %)	0.0523
	Ventilation	5570 (6.9 %)	5030 (6.4 %)	10 600 (6.6 %)	0.06
	Sepsis	8305 (10.3 %)	10 070 (12.8 %)	18 375 (11.5 %)	<0.0001
	Vasopressor use	1090 (1.4 %)	1140 (1.4 %)	2230 (1.4 %)	0.4905
Age≥65	Mortality	860 (4.7 %)	1155 (4.2 %)	2015 (4.4 %)	0.2117
	AKI	12 090 (66.7 %)	16 455 (59.8 %)	28 545 (62.5 %)	<0.0001
	AKI requiring dialysis	425 (2.3 %)	360 (1.3 %)	785 (1.7 %)	0.0002
	Ventilation	1650 (9.1 %)	2125 (7.7 %)	3775 (8.3 %)	0.018
	Sepsis	2810 (15.5 %)	4685 (17.0 %)	7495 (16.4 %)	0.0598
	Vasopressor use	450 (2.5 %)	645 (2.3 %)	1095 (2.4 %)	0.6689

AKI, acute kidney injury.

After adjustment, females had significantly lower odds of mortality (aOR 0.72, 95% CI 0.60 to 0.86, p<0.001) and AKI (aOR 0.56, 95% CI 0.54 to 0.57, p<0.001), but higher odds of sepsis (aOR 1.66, 95% CI 1.57 to 1.75, p<0.001)(figure 1). There were no significant differences in the adjusted odds for AKI requiring dialysis (aOR 0.94, 95% CI 0.80 to 1.10, p=0.406), ventilation (aOR 0.96, 95% CI 0.90 to 1.04, p=0.34) or vasopressor use (aOR 0.84, 95% CI 0.70 to 1.02, p=0.086).

Figure 1. Forest plot showing adjusted in-hospital events and outcomes of diabetic ketoacidosis (DKA) stratified by age and sex (men as reference group). AKI, acute kidney injury.

Age 45–64 years

In the middle age group (45–64 years), the population included 80 630 males and 78 835 females. Men continued to show a higher incidence of AKI, at 57.1% (46 080 cases) vs 47.5% (37 470 cases) in women, with p<0.0001. Women had a higher rate of sepsis, at 12.8% (10 070 cases) compared with 10.3% (8305 cases) in men, with p <0.0001. Mortality rates were similar between men and women, at 1.8% (1485 cases) vs 1.6% (1265 cases), with p=0.1083, indicating no statistical significance. The rates of AKI requiring dialysis were 1.7% (1340 cases) in men versus 1.4% (1100 cases) in women, with p=0.0523, showing a trend but not reaching conventional significance. Ventilation rates were 6.9% (5570 cases) in men versus 6.4% (5030 cases) in women, with p=0.06, again indicating a trend but not significant at the 0.05 level. Vasopressor use was comparable, at 1.4% for both groups (1090 cases in men, 1140 in women, p=0.4905).

Females had significantly lower odds of AKI (aOR 0.71, 95% CI 0.67 to 0.74, p<0.001) and higher odds of sepsis (aOR 1.31, 95% CI 1.22 to 1.41, p<0.001) after adjustment. There were no significant differences in mortality (aOR 0.89, 95% CI 0.75 to 1.07, p=0.209), AKI requiring dialysis (aOR 0.88, 95% CI 0.73 to 1.07, p=0.207), ventilation (aOR 0.99, 95% CI 0.90 to 1.09, p=0.88) or vasopressor use (aOR 1.07, 95% CI 0.87 to 1.32, p=0.52).

Age ≥65 years

For the oldest age group (≥65 years), there were 18 115 males and 27 530 females. Men had higher rates of AKI, at 66.7% (12 090 cases) vs 59.8% (16 455 cases) in women, with p<0.0001, and higher rates of AKI requiring dialysis, at 2.3% (425 cases) vs 1.3% (360 cases), with p=0.0002, indicating significant differences. Men also had a higher need for ventilation, at 9.1% (1650 cases) compared with 7.7% (2125 cases) in women, with p=0.018, showing statistical significance. Women had a slightly higher rate of sepsis, at 17.0% (4685 cases) compared with 15.5% (2810 cases) in men, but this was not statistically significant (p=0.0598). Mortality rates were comparable, at 4.7% (860 cases) in men versus 4.2% (1155 cases) in women, with p=0.2117. Vasopressor use rates were also similar, at 2.5% (450 cases) in men versus 2.3% (645 cases) in women, with p=0.6689, indicating no significant difference.

After adjustment, females had significantly lower odds of AKI (aOR 0.79, 95% CI 0.72 to 0.87, p<0.001), AKI requiring dialysis (aOR 0.65, 95% CI 0.46 to 0.93, p=0.02) and higher odds of sepsis (aOR 1.17, 95% CI 1.03 to 1.32, p=0.02). There were no significant differences in mortality (aOR 0.98, 95% CI 0.78 to 1.21, p=0.83), ventilation (aOR 0.92, 95% CI 0.78 to 1.08, p=0.29) or vasopressor use (aOR 0.99, 95% CI 0.74 to 1.33, p=0.94).

Healthcare resource utilization

Across the age groups, healthcare resource utilization showed distinct patterns. In the 18–44 age group, women had significantly longer hospital stays and higher costs compared with men. In contrast, the 45–64 age group showed no significant differences, indicating similar resource utilization between sexes in middle age. For those aged 65 and older, men had significantly longer stays and higher cost of hospitalization than women.

Trend of in-hospital mortality

From 2016 to 2019, mortality rates for type 1 patients with DKA showed a slight upward trend for men across all ages (from 0.74% to 0.86%), while women exhibited a decline (from 0.80% to 0.64%)(figure 2). In the 18–44 age group, men’s mortality increased steadily (0.32 to 0.48), whereas women’s decreased (0.34% to 0.27%), maintaining lower rates than men. For ages 45–64, men’s rates fluctuated but trended downward (1.68% to 1.40%), while women’s rates varied (1.27% to 0.98%), with women briefly surpassing men in 2018 (1.56% vs 1.45%). In the ≥65 group, men’s mortality rose (3.24% to 4.56%), while women’s declined after a peak in 2016 (5.34% to 3.77%), with women having higher rates in 2016 but lower by 2019. Both sexes experienced a notable increase in mortality in 2020 and 2021, with men reaching 1.54% and women 1.29% across all ages by 2021, likely influenced by the impact of the COVID-19 pandemic, particularly pronounced in the 18–44 and 45–64 age groups for men (0.89% and 2.72% in 2021, respectively) and to a lesser extent for women (0.59% and 2.60%). Overall, men generally had higher mortality rates in younger and middle-aged groups, while women showed higher rates in the oldest group in 2016, with diverging trends before a sharp rise for both in 2020–2021.

Figure 2. Trend of in-hospital mortality among hospitalizations with diabetic ketoacidosis (DKA).

Discussion

In this national analysis of more than 799 000 hospitalizations for DKA among patients with type 1 diabetes, we found pronounced and age-dependent sex differences that persisted after multivariable adjustment. Across every adult age stratum, women experienced significantly lower odds of AKI but a consistently higher odds of sepsis, whereas men carried a greater burden of traditional cardiometabolic comorbidities (hypertension, smoking, coronary artery disease) and substance-use disorders. These disparities were most striking in younger adults, where men’s excess mortality and renal injury contrasted with women’s greater resource use and sepsis risk, and they remained evident, though attenuated in middle and older age. Notably, the divergent temporal trends in mortality from 2016 to 2019, followed by the parallel surge during the first two pandemic years, suggest sex-specific resilience patterns that were ultimately overwhelmed by the systemic stresses of COVID-19.

Demographic patterns and healthcare access

The interaction between age and sex creates distinct demographic vulnerabilities that evolve across the lifespan. Among young adults (18–44 years), women demonstrate a concerning concentration of socioeconomic disadvantage, with 39.7% residing in the lowest income quartile compared with 38.6% of men, and substantially higher Medicaid dependency (48.0% vs 41.5%). This disparity is further compounded by their higher representation in the Southern USA (44.8% vs 42.9%), a region historically characterized by limited Medicaid expansion and reduced healthcare infrastructure.^{14 15} These socioeconomic vulnerabilities diminish with advancing age, as middle-aged women (45–64 years) show improved insurance profiles with higher Medicare (33.3% vs 31.0%) and private insurance coverage (32.1% vs 28.3%), while older adults (≥65 years) achieve demographic parity across most^{16 17} socioeconomic indicators.

Comorbidity differences

We observed different patterns of co-existing comorbidities, with cardiovascular and renal disease clustering in men and autoimmune-endocrine and mood disorders predominating in women. These differences may arise from a confluence of biologic and behavioral mechanisms. Male patients with type 1 diabetes accumulate cardiometabolic injury earlier, a pattern linked to higher rates of tobacco and alcohol use as well as attenuated estrogen-mediated vasoprotection, driving the steeper age-related rise in hypertension, coronary artery disease and chronic kidney disease.¹⁸ In women, the exponential increase in hypothyroidism and persistent excess of depression likely reflects their greater predisposition to organ-specific autoimmunity, facilitated by X-chromosome gene dosage and immune-stimulatory effects of estrogens. A seminal cell study showed that the female-specific long-non-coding RNA Xist scaffolds a ribonucleoprotein complex studded with autoantigenic proteins; introducing this complex into male mice was sufficient to trigger lupus-like autoantibodies and multiorgan pathology, implicating Xist-RNP-driven loss of tolerance as a causal engine for the female excess of organ-specific autoimmunity.¹⁹ Additionally, bidirectional relationship between diabetes and affective illness has been documented, which could be amplified by glycemic variability, inflammatory signaling and psychosocial stressors among women.^{16 17 20}

Clinical outcomes and mechanistic insights

The most compelling findings emerge from the age-sex stratified analysis of acute complications, revealing consistent protective effects for women against AKI across all age groups, with adjusted ORs of 0.56 (95% CI 0.54 to 0.57) in young adults, 0.71 (95% CI 0.67 to 0.74) in middle-aged adults and 0.79 (95% CI 0.72 to 0.87) in older adults. This finding aligned broadly with multiple existing literatures that showed that female sex was found to be associated with a significant reduction in the risk of AKI, irrespective of the etiology.²¹,²³ Neugarten and colleagues studied the pattern of hospital-acquired AKI in a meta-analysis and found that the risk of AKI was significantly lowered in women compared with men.²⁴ Likewise, in critically ill patients admitted with septic shock, there was a lower odds of AKI and AKI requiring dialysis in women compared with men in both unadjusted and adjusted analysis.⁹

Preclinical work has begun to explain this phenomenon. In a murine ischemia–reperfusion and rhabdomyolysis model, Miao et al revealed that female kidneys preserve mitochondrial integrity by maintaining expression of the epigenetic regulator Sirtuin-6; androgen-receptor signaling in males suppresses Sirtuin-6, causing PGC-1α acetylation, mitochondrial dysfunction and tubular apoptosis, thereby amplifying renal injury.²⁵ Moreover, accumulating evidence indicates that circulating gonadal hormones gate these sex gaps in renal vulnerability. Experimental and translational work shows estrogen preserves endothelial nitric-oxide signaling, blunts tubular oxidative stress and sustains mitochondrial integrity, whereas testosterone amplifies inflammation and apoptotic pathways, rendering male kidneys more susceptible to ischemic and toxic injury.²²²⁶,²⁸ Supporting this endocrine paradigm, Aufhauser and colleagues demonstrated that female mice preserve renal function after ischemia-reperfusion injury through intact estrogen-receptor-α signaling, an advantage lost with ovariectomy but rescued by estrogen supplementation and that the recipient’s sex, not the donor kidney’s sex, dictates post-transplant ischemia tolerance.²⁶ In >46 000 human deceased-donor transplants, the same group found male recipients had significantly higher odds of delayed graft function than female recipients after multivariable adjustment, underscoring the clinical relevance of estrogen-mediated renal resilience. The gradual attenuation of protection with advancing age (ORs increasing from 0.56 to 0.79) may reflect the progressive loss of estrogen-mediated vasodilation and anti-inflammatory effects following menopause, yet the persistence of significant protection in postmenopausal women suggests additional non-hormonal mechanisms.

The age–sex stratification of sepsis risk reveals a more complex pattern that challenges simple hormonal explanations. Among type 1 diabetes with DKA, women demonstrate consistently elevated sepsis risk across all age groups (adjusted ORs: 1.66 in young adults, 1.31 in middle-aged adults and 1.17 in older adults), with the magnitude of risk paradoxically decreasing with advancing age. Despite this higher infection burden, rates of vasopressor use and mechanical ventilation were indistinguishable between sexes in every age stratum; indeed, in the 18–44-year cohort, women’s odds of vasopressor use were numerically lower and narrowly missed statistical significance. This DKA-specific pattern contrasts with the general population, where large epidemiologic studies consistently show men to have both a higher incidence of sepsis and a steeper age-related rise in sepsis-related ICU admission and mortality.²⁹ For example, a Norwegian population-based cohort found men had a 41% greater risk of bloodstream infection and an 87% higher risk of death from such infections than women, while US and European registries likewise report male predominance in septic shock episodes and related deaths.^{9 30} The discordance suggests that the metabolic and immunologic phenomenon related to DKA may override the usual male susceptibility to severe infection, perhaps through sex-specific differences in cytokine responses to hyperketonemia or diagnostic thresholds that favor earlier sepsis coding in women, a hypothesis that merits mechanistic and health-services investigation.

Age–sex mortality trajectories and pandemic impact

Prepandemically, while literature often points to higher overall DKA mortality in males, our data show a more detailed picture with transiently higher or declining mortality in young and older women, respectively. The initially higher mortality in young women, which later declined, may reflect a higher baseline DKA incidence in this group, potentially linked to behavioral factors like insulin omission, followed by improvements in the behavior. The divergent trends in older adults, with rising male and falling female mortality, likely reflect sex-specific differences in the accumulation and management of comorbidities, a known risk factor for DKA mortality.

The COVID-19 pandemic introduced a profound shift, drastically increasing mortality across all demographics. This aligns with extensive reports on the synergistic and often fatal interaction between COVID-19 and DKA. The particularly sharp mortality increase in young men to 0.9% suggests a specific vulnerability, possibly reflecting the known male disadvantage in COVID-19 outcomes. Conversely, the uniform mortality surge in middle-aged individuals supports the hypothesis that the severe systemic stress of the virus overwhelmed any pre-existing sex-specific protective factors. The less pronounced, though still significant, increase among older adults might reflect cautious shielding behaviors or the statistical effect of a higher baseline mortality and competing risks, warranting further investigation into the specific care patterns for this vulnerable population during the pandemic.

Limitations

This study has important limitations related to the NIS. The database is discharge-based and does not include patient identifiers, so we could not assess outcomes after discharge. Identification of type 1 diabetes-related DKA, comorbidities and in-hospital complications relied on ICD-10-CM codes, which may introduce misclassification bias. Furthermore, the database does not provide laboratory data, vital signs, measures of DKA severity, coronavirus infection status, insulin or fluid regimens or information on intensive care versus ward level care, so we could not fully account for illness severity or treatment differences between groups. Although we adjusted for many demographic, clinical and hospital characteristics, unmeasured factors such as diabetes duration, social context, health behaviors and access to outpatient care may still confound our findings. Costs were derived from hospital charges and do not capture expenses after discharge or in other settings, and our results may not be generalized to population outside the USA. Additionally, we did not conduct a prespecified pre-COVID-19 vs COVID-19–era comparison of sex-associated differences in outcomes in the current study, and we are pursuing this question in a separate prespecified analysis among adults with type 1 diabetes.

Conclusion

In this national cohort of adults with type 1 diabetes and DKA, clinical profiles and in-hospital outcomes varied meaningfully by age and sex. Across all age groups, women had lower adjusted odds of AKI but higher odds of sepsis than men, while differences in the need for dialysis, mechanical ventilation and vasopressors were small. A survival advantage for women was evident only in young adults, with no clear sex difference in mortality in middle-aged and older patients. Patterns of resource use also differed by age and sex. Before the pandemic, mortality among women with DKA declined, whereas mortality in men changed little; during the COVID-19 years, mortality rose for both sexes, most sharply in young men. Taken together, these findings suggest that risk assessment and management of DKA should account for both age and sex, and they highlight young adults, particularly men, as a group at growing risk. Within the constraints of an administrative database, our results provide a benchmark for designing targeted interventions and for future work that links biological mechanisms with health services factors to reduce preventable complications and deaths.

Data availability statement

Data may be obtained from a third party and are not publicly available.