Outcomes following hospitalization for diabetic ketoacidosis in patients with cardiovascular disease

Abstract

Introduction

Diabetic ketoacidosis (DKA) is a serious complication in patients with diabetes. This study compares the outcomes of hospitalized DKA patients with and without cardiovascular disease (CVD). In addition, we assess outcomes between DKA and hyperosmolar hyperglycemic syndrome (HHS) in diabetes patients with CVD, and between those who developed DKA and those who did not.

Methods

We employed a population-based, retrospective observational design utilizing data sourced from the National Inpatient Sample database for the years 2016–2022. The primary outcome assessed was in-hospital mortality. In addition, various secondary outcomes were examined, including the incidence of acute respiratory failure, acute kidney failure, septic shock, sepsis, acute neurological failure, pulmonary embolism, deep vein thrombosis, acute liver failure, mechanical ventilation, noninvasive ventilation, and the length of hospital stay (LOS) and total hospital charges.

Results

Multivariable regression analysis demonstrated that CVD independently increased mortality and complications, including acute respiratory failure and sepsis, in DKA patients, who also experienced longer LOS and higher medical costs compared to those without CVD. Similar findings were observed when comparing outcomes between DKA and HHS in diabetes patients with CVD, as well as between those who developed DKA and those who did not.

Conclusions

This study demonstrates that CVD significantly affects the outcomes of patients admitted for DKA. Moreover, similar negative outcomes were observed when comparing DKA patients with HHS and those who developed DKA versus those who did not. These findings highlight the need for careful management of DKA in patients with CVD to optimize clinical outcomes.

Introduction

Diabetic ketoacidosis (DKA) is a serious and potentially life-threatening complication of diabetes mellitus, occurring more commonly in individuals with type 1 diabetes but also frequently observed in those with type 2 diabetes, particularly under stressful conditions (1, 2, 3). It is characterized by hyperglycemia, metabolic acidosis, and the accumulation of ketones due to an absolute or relative insulin deficiency, resulting in profound metabolic disturbances (4). Despite advances in diabetes management, the incidence of DKA continues to rise (5, 6). This situation emphasizes the need for attentive care to prevent severe complications such as cerebral edema, kidney failure, and death (7).

Cardiovascular disease (CVD) is a chronic condition that significantly contributes to increased morbidity and mortality. In contrast, DKA is an acute metabolic crisis that can worsen cardiovascular outcomes. Currently, there are a limited number of studies focused on the characteristics and effects of specific CVD subtypes on patients diagnosed with DKA during their hospital stays. Shaka et al. reported that DKA patients with diastolic heart failure had similar in-hospital mortality rates compared to those without diastolic heart failure but experienced higher rates of complications such as non-ST elevation myocardial infarction and acute respiratory failure, resulting in increased hospital costs and longer stays (8). Similarly, Yang et al. found that atrial fibrillation (AF) was associated with higher mortality, prolonged hospitalizations, and greater risks of septic shock and cardiac arrest in DKA patients (9). Agarwal et al. demonstrated that a history of heart failure during hyperglycemic crises was associated with increased mortality, longer hospitalization, and higher rates of discharge to nursing facilities (10). Moreover, diabetes patients with heart failure experiencing DKA were found to have a higher adjusted risk of mortality, ischemic stroke, acute renal failure, and cardiogenic shock (11). Meanwhile, DKA can impact cardiac function through several mechanisms, including the toxic effects of hyperglycemia, acidosis, and electrolyte disturbances (12). These issues can impair myocardial blood flow and induce autonomic dysfunction, disrupting the conduction of electrical impulses within the heart and raising the risk of arrhythmias. In addition, the interplay of hyperglycemia, dehydration, and metabolic stress in DKA can decrease coronary perfusion, which may ultimately lead to myocardial ischemia (13). Studies have also shown that DKA is associated with increased risk of subsequent major adverse cardiovascular events (14).

However, while these findings highlight the risks associated with certain CVD subtypes, the overall impact of CVD on DKA outcomes remains unclear, and the detailed implications of the majority of its subtypes have yet to be clarified. Moreover, the differences in outcomes between DKA and hyperosmolar hyperglycemic syndrome (HHS) in diabetes patients with CVD, as well as between those who developed DKA and those who did not, are largely unexplored. To address these gaps, we conducted a comprehensive analysis using the National Inpatient Sample (NIS) from 2016 to 2022.

Methods

Data source and study population

This study used data from the NIS for the years 2016–2022, obtained from the Healthcare Cost and Utilization Project (HCUP) (15). The NIS includes over 8 million hospitalizations each year and represents a 20% sample of discharges from hospitals across the United States (16). It is the largest all-payer in-hospital database and provides weighted estimates for about 97% of hospitalizations in the country (17).

We identified patients with DKA who were 18 years or older, using primary diagnosis codes from the International Classification of Diseases, 10th Revision (ICD-10) (18). For the identification of diabetes patients with CVD, we selected patients who had both diabetes and CVD based on their corresponding ICD-10 codes. Patients were considered to have CVD if their records included specific diagnoses such as arrhythmia, chronic coronary artery disease, heart disease, cardiomyopathies, heart failure, pulmonary hypertension, or valvular heart disease during their hospitalization for DKA. The relevant codes are listed in Supplementary Table 1 (see section on Supplementary materials given at the end of the article) (19, 20). We excluded patients who had missing important information needed for our analysis. The Elixhauser Comorbidity Software Refined (CMR v2025.1) was employed to categorize comorbidities by grouping ICD-10-CM diagnosis codes into distinct comorbidity classifications (21).

Outcomes assessments

The primary outcome was comparing in-hospital mortality in DKA based on the presence or absence of CVD. Secondary outcomes included rates of acute respiratory failure, sepsis, septic shock, acute kidney failure, acute neurological failure, pulmonary embolism, deep vein thrombosis, acute liver failure, mechanical ventilation, and noninvasive ventilation, as well as length of hospital stay (LOS) and total hospital charges. The relevant codes are listed in Supplementary Table 2 (8, 9, 22).

Statistical analysis

This study utilized weighted samples from the NIS database. Descriptive statistics were employed to summarize the demographic, clinical, and socioeconomic characteristics of the study cohorts. Categorical variables were represented as counts and percentages, while continuous variables were expressed as means with standard deviations (SD). Statistical comparisons of categorical variables were conducted using the chi-square test or Fisher’s exact test as appropriate, while the Wilcoxon rank-sum test was applied for continuous variables confirmed to follow a non-normal distribution through the Kolmogorov–Smirnov test.

We analyzed the associations between CVD and its various subtypes with in-hospital mortality and aforementioned secondary outcomes using multivariable logistic regression. This analysis was adjusted for potential confounders, including demographic factors (age, gender, and race), socioeconomic variables (income, geographic location, and primary expected payer), hospital characteristics (teaching hospital status and hospital bed size), clinical characteristics (type of admission and the number of comorbidities identified via CMR v2025.1), and admission year. An initial univariate logistic regression was performed to identify potential confounders.

Statistical analyses were conducted using SAS software (version 9.4, SAS Institute Inc., USA) and R software (version 4.5.0, R Foundation for Statistical Computing, Austria), with a two-sided P-value of less than 0.05 considered statistically significant.

Ethical considerations

This research employed publicly available de-identified data, which exempted it from the necessity of formal approval from an institutional review board (IRB). Throughout the study, we adhered to ethical guidelines and reporting standards for observational research.

Results

Demographics and prevalence of CVD among DKA hospitalizations

Between 2016 and 2022, the NIS database recorded a total of 283,599 hospitalizations with a primary diagnosis of DKA for patients aged 18 and older. After excluding individuals with missing key demographic data, 268,473 DKA hospitalizations remained. Among these hospitalizations related to DKA, the prevalence of CVD was found to be 45,234 cases, which accounts for 16.8% of the total.

In the analysis of DKA hospitalizations, CVD was significantly more common among White individuals compared to other racial and ethnic groups (Table 1). Patients diagnosed with CVD tended to be older and were more likely to have Medicare, as well as a higher number of medical comorbidities (Table 1).

Table 1.

Baseline characteristics of patients with diabetic ketoacidosis hospitalization, stratified by cardiovascular disease status in the national inpatient sample (2016–2022).

Demographics	Non-CVD	CVD	P-value
	n = 223,239	n = 45,234
Age (mean SD)	39.4 (15.7)	57.0 (15.4)	<0.001
Gender			<0.001
Male	116,142 (52.0)	23,097 (51.1)
Female	107,097 (48.0)	22,137 (48.9)
Race
White	119,839 (53.7)	28,832 (63.7)	<0.001
Black	61,982 (27.8)	11,095 (24.5)
Hispanic	30,491 (13.7)	3,626 (8.0)
Asian or Pacific Islander	2,716 (1.2)	526 (1.2)
Native American and others	82,111 (2.7)	1,155 (2.5)
Admission type
Non-elective	219,235 (98.2)	44,517 (98.4)	0.002
Elective	4,004 (1.8)	717 (1.6)
Income quartile by zip code
Quartile 1 (lowest)	87,879 (39.4)	17,274 (38.2)	<0.001
Quartile 2	60,766 (27.2)	12,289 (27.2)
Quartile 3	46,462 (20.8)	9,559 (21.1)
Quartile 4 (highest)	28,132 (12.6)	6,112 (13.5)
Patient location
Large central metropolitan	67,492 (30.2)	12,661 (28.0)	<0.001
Large fringe metropolitan	47,498 (21.3)	10,047 (22.2)
Medium metropolitan	49,958 (22.4)	10,159 (22.5)
Small metropolitan	21,990 (9.9)	4,708 (10.4)
Micropolitan	21,768 (9.8)	4,497 (9.9)
Noncore	14,533 (6.5)	3,162 (7.0)
Primary expected payer
Medicare	38,013 (17.0)	23,130 (51.1)	<0.001
Medicaid	81,820 (36.7)	10,275 (22.7)
Private insurance	64,802 (29.0)	8,177 (18.1)
Self-pay	29,834 (13.4)	2,411 (5.3)
No charge and others	8,770 (4.0)	1,241 (2.7)
Hospital region
Northeast	32,080 (14.4)	7,471 (16.5)	<0.001
Midwest	45,751 (20.5)	10,484 (23.2)
South	102,129 (45.7)	19,940 (44.1)
West	43,279 (19.4)	7,339 (16.2)
Hospital bed size
Small	55,571 (24.9)	10,426 (23.0)	<0.001
Medium	66,904 (30.0)	13,409 (29.6)
Large	100,764 (45.1)	21,399 (47.3)
Hospital location/teaching status
Rural	55,571 (24.9)	10,426 (23.0)	<0.001
Urban non-teaching	66,904 (30.0)	13,409 (29.6)
Urban teaching	100,764 (45.1)	21,399 (47.3)
Number of medical comorbidities
≤ 3	206,550 (92.5)	33,708 (74.5)	<0.001
>3	16,689 (7.5)	11,526 (25.5)

Data are n (%) or mean ± SD.

In-hospital mortality rates in DKA patients with CVD

The in-hospital mortality rate among all patients diagnosed with DKA was 0.51%. When comparing patients with and without CVD, the unadjusted mortality rates were 1.6 and 0.3%, respectively (P < 0.001) (Table 2). A multivariable regression analysis revealed that CVD was independently associated with increased mortality rates in DKA patients, with an adjusted odds ratio (aOR) of 26.68 and a 95% confidence interval (CI) ranging from 17.68 to 40.27 (Table 2). The risk of in-hospital mortality varied among different subtypes of CVD during DKA hospitalizations (Fig. 1). Arrhythmias had the highest risk of in-hospital mortality (aOR 73.27, 95% CI 47.20–113.73), followed by pulmonary hypertension (aOR 26.59, 95% CI 7.57–93.44), cardiomyopathies (aOR 15.73, 95% CI 6.79–36.44), and heart failure (aOR 8.36, 95% CI 4.50–15.53). Conversely, the associations between chronic coronary artery disease, valvular heart disease, and congenital heart disease with in-hospital mortality were not statistically significant, with respective aORs of 1.86 (95% CI 0.92–3.74), 1.25 (95% CI 0.22–6.92), and 1.09 (95% CI 0.03–34.98).

Table 2.

Comparison of clinical outcomes of diabetic ketoacidosis hospitalization patients with and without cardiovascular disease in the national inpatient sample (2016–2022).

Morbidity	Non-CVD	CVD	Crude OR (95% CI)	aOR (95% CI)*
	n = 223,239	n = 45,234
In-hospital mortality	647 (0.3)	723 (1.6)	5.59 (5.02–6.22)	26.68 (17.68–40.27)
Acute respiratory failure	5,114 (2.3)	4,164 (9.2)	4.32 (4.15–4.51)	6.61 (5.58–7.81)
Septic shock	1,163 (0.5)	535 (1.2)	2.29 (2.06–2.53)	4.69 (3.18–6.91)
Acute kidney failure	88,952 (39.8)	24,272 (53.7)	1.75 (1.71–1.78)	1.24 (1.14–1.35)
Sepsis	3,109 (1.4)	1,427 (3.2)	2.31 (2.16–2.46)	3.35 (2.59–4.33)
Acute neurological failure	4,257 (1.9)	1917 (4.2)	2.28 (2.15–2.40)	2.11 (1.67–2.66)
Pulmonary embolism	440 (0.2)	236 (0.5)	2.66 (2.27–3.11)	6.37 (3.50–11.58)
Deep vein thrombosis	521 (0.2)	282 (0.6)	2.68 (2.32–3.1)	6.47 (3.65–11.48)
Acute liver failure	346 (0.2)	252 (0.6)	3.61 (3.07–4.25)	9.22 (5.17–16.46)
Mechanical ventilation	3,162 (1.4)	1863 (4.1)	2.99 (2.82–3.17)	8.06 (6.47–10.03)
Noninvasive ventilation	788 (0.4)	727 (1.6)	4.61 (4.17–5.10)	2.84 (1.80–4.46)

CVD, cardiovascular disease; OR, odds ratio; aOR, adjusted odds ratio.

Data are n (%) unless otherwise specified.

^*Model adjusted for age, race, admission type, insurance status, income quartile, patient location, hospital bed size, hospital location/teaching status, number of medical comorbidities, and admission year.

Figure 1.

Impact of various cardiovascular conditions on clinical outcomes. This figure presents the adjusted odds ratios (aOR) and 95% CIs for adverse clinical outcomes, including in-hospital mortality (death), acute respiratory failure (ARF), septic shock (SS), acute kidney failure (AKF), sepsis, acute neurological failure (ANF), pulmonary embolism (PE), deep vein thrombosis (DVT), acute liver failure (ALF), mechanical ventilation (MV), and noninvasive ventilation (NV) associated with different cardiovascular conditions. Each panel depicts a specific condition: chronic coronary artery disease, heart failure, pulmonary hypertension, cardiomyopathies, valvular heart diseases, arrhythmia, and congenital heart disease. The blue dots represent the point estimates of the aOR, while the red lines indicate the corresponding 95% CI. An aOR of 1.0 denotes no effect, with values greater than 1.0 indicating increased risk, and values less than 1.0 indicating decreased risk for the specified outcomes.

Clinical outcomes in DKA patients with CVD

Our data revealed significant differences in clinical outcomes for DKA patients with and without CVD, as shown in Table 2. CVD patients exhibited higher incidences of acute respiratory failure (crude OR: 4.32, 95% CI: 4.15–4.51; aOR: 6.61, 95% CI: 5.58–7.81), acute neurological failure (crude OR: 2.28, 95% CI: 2.15–2.40; aOR: 2.11, 95% CI: 1.67–2.66), and acute kidney failure (crude OR: 1.75, 95% CI: 1.71–1.78; aOR: 1.24, 95% CI: 1.14–1.35). The rates of sepsis (crude OR: 2.31, 95% CI: 2.16–2.46; aOR: 3.35, 95% CI: 2.59–4.33) and septic shock (crude OR: 2.29, 95% CI: 2.06–2.53; aOR: 4.69, 95% CI: 3.18–6.91) were also elevated in those with CVD. Furthermore, pulmonary embolism (crude OR: 2.66, 95% CI: 2.27–3.11; aOR: 6.37, 95% CI: 3.50–11.58) and deep vein thrombosis (crude OR: 2.68, 95% CI: 2.32–3.10; aOR: 6.47, 95% CI: 3.65–11.48) were significantly more common among CVD patients. Acute liver failure was observed more frequently (crude OR: 3.61, 95% CI: 3.07–4.25; aOR: 9.22, 95% CI: 5.17–16.46). Mechanical ventilation rates were markedly increased (crude OR: 2.99, 95% CI: 2.82–3.17; aOR: 8.06, 95% CI: 6.47–10.03), as was noninvasive ventilation use (crude OR: 4.61, 95% CI: 4.17–5.10; aOR: 2.84, 95% CI: 1.80–4.46). These findings indicate that DKA patients with CVD experience significantly worse clinical outcomes.

Figure 1 presents the aOR with 95% CI for a range of clinical outcomes in patients with various types of CVD. The data indicate that most conditions are associated with varying degrees of increased aOR. Of note, patients with arrhythmia show significantly elevated aORs for multiple adverse outcomes, implying the considerable morbidity associated with this condition. This is indicative of the elevated vulnerability of arrhythmia patients to complications arising from DKA. Conversely, congenital heart diseases and valvular heart diseases reveal consistently lower aORs across the assessed outcomes, suggesting a relatively lower risk profile compared to other cardiovascular conditions. This differential impact highlights the varied clinical implications of different CVD types and suggests the need for customized management strategies based on the specific cardiovascular condition present in DKA patients.

The incidence of clinical outcomes during DKA hospitalization was analyzed across patients with various CVD subtypes. As shown in Fig. 2 and Supplementary Table 3, acute kidney failure was notably prevalent among these patients. The incidence of acute kidney failure was observed in 58.72% of cases involving arrhythmia.

Figure 2.

Distribution of clinical outcomes by cardiovascular conditions. This figure illustrates the percentage of various clinical outcomes associated with different cardiovascular conditions, including arrhythmia, coronary artery disease (CAD), chronic heart disease (CHD), cardiomyopathies (CM), heart failure (HF), non-cardiovascular disease (non-CVD), pulmonary hypertension (PH), and valvular heart diseases (VHD). The outcomes measured include acute respiratory failure (ARF), acute kidney failure (AKF), sepsis, septic shock, acute neurological failure (ANF), pulmonary embolism (PE), deep vein thrombosis (DVT), acute liver failure (ALF), mechanical ventilation (MV), noninvasive ventilation (NV), and in-hospital mortality (death). The bar heights represent the percentage of each outcome for the respective cardiovascular conditions, providing insights into the impact of these conditions on patient morbidity and mortality.

Impact of CVD on LOS and costs in DKA hospitalizations

We next analyzed hospital outcomes related to DKA among patients with and without various subtypes of CVD. The findings, summarized in Table 3, indicate significant differences in both the LOS and hospitalization costs associated with specific CVD subtypes. Overall, patients with any form of cardiovascular disease had a longer average LOS by 1.03 days (95% CI: 0.99–1.07, P < 0.001) and experienced higher hospitalization costs, averaging $12,931.72 (95% CI: $12,496.25–$13,367.18, P < 0.001). Among the specific CVD subtypes, pulmonary hypertension was associated with the highest LOS increase of 2.10 days (95% CI: 1.69–2.52, P < 0.001), leading to hospitalization costs of $26,160.35 (95% CI: $21,279.83–$31,040.86, P < 0.001). Patients with heart failure demonstrated a significantly increased LOS of 1.35 days (95% CI: 1.19–1.51, P < 0.001) and incurred average costs of $17,515.92 (95% CI: $15,622.25–$19,409.59, P < 0.001). Patients with valvular heart diseases experienced an increased LOS of 1.70 days (95% CI: 1.36–2.03, P < 0.001), with associated costs of $18,769.20 (95% CI: $14,843.28–$22,695.13, P < 0.001). In addition, those with arrhythmia had a notable LOS extended by 1.00 day (95% CI: 0.84–1.15, P < 0.001) and hospitalization costs averaging $17,331.64 (95% CI: $15,503.85–$19,159.43, P < 0.001). However, patients with congenital heart disease showed no significant difference in LOS (mean difference: 0.12 days, 95% CI: −0.48–0.71, P = 0.70) or hospitalization costs ($998.64, 95% CI: −$5,996.60–$7,993.87, P = 0.78). These results highlight the substantial impact of cardiovascular comorbidities on hospitalization outcomes in patients presenting with DKA.

Table 3.

Comparison of hospital outcomes in diabetic ketoacidosis hospitalization patients with and without cardiovascular disease subtype in the national inpatient sample (2016–2022).

Variable	Adjusted mean difference*	95% CI	P value
Cardiovascular disease
Length of stay (days)	1.03	0.99–1.07	<0.001
Hospitalization costs ($)	12,931.72	12,496.25–13,367.18	<0.001
Chronic coronary artery disease
Length of stay (days)	0.39	0.24–0.54	<0.001
Hospitalization costs ($)	7,715.03	5,641.80–9,488.25	<0.001
Heart failure
Length of stay (days)	1.35	1.19–1.51	<0.001
Hospitalization costs ($)	17,515.92	15,622.25–19,409.59	<0.001
Pulmonary hypertension
Length of stay (days)	2.10	1.69–2.52	<0.001
Hospitalization costs ($)	26,160.35	21,279.83–31,040.86	<0.001
Cardiomyopathies
Length of stay (days)	0.86	0.60–1.12	<0.001
Hospitalization costs ($)	16,604.50	13,561.77–19,647.23	<0.001
Valvular heart diseases
Length of stay (days)	1.70	1.36–2.03	<0.001
Hospitalization costs ($)	18,769.20	14,843.28–22,695.13	<0.001
Arrhythmia
Length of stay (days)	1.00	0.84–1.16	<0.001
Hospitalization costs ($)	17,331.64	15,503.85–19,159.43	<0.001
Congenital heart disease
Length of stay (days)	0.12	−0.48–0.71	0.70
Hospitalization costs ($)	998.64	−5,996.60–7,993.87	0.78

CI, confidence interval.

^*Model adjusted for age, race, admission type, insurance status, income quartile, patient location, hospital bed size, hospital location/teaching status, number of medical comorbidities, and admission year.

Impact of CVD on hospital and clinical outcomes in DKA patients with T1DM and T2DM

We further investigated the outcomes of patients diagnosed with DKA concerning the presence of CVD in both T1DM and T2DM. Detailed subgroup demographics were provided in Supplementary Table 4. Our findings revealed that T2DM patients exhibited significantly higher mortality rates compared to T1DM patients (2.2 vs 0.9%, P < 0.001) (Fig. 3A). In T1DM patients, those with CVD experienced a significantly longer LOS (5.05 vs 4.67 days for T1DM, P < 0.001) (Fig. 3B), along with increased hospitalization costs ($55,900 vs $48,700 for T1DM, P < 0.001) (Fig. 3C). Among T2DM patients, those with CVD had their length of stay extended by 0.74 days (95% CI: 0.51–0.96, P < 0.001), and hospitalization costs increased by an average of $13,839.31 (95% CI: 11,049.86–16,628.72, P < 0.001) (Table 4). For T1DM patients, the presence of CVD was associated with a significant adjusted mean difference in LOS of 0.98 days (95% CI: 0.86–1.10, P < 0.001), along with increased hospitalization costs amounting to $13,152.69 (95% CI: 11,790.11–14,515.28, P < 0.001) when compared to their counterparts without CVD (Table 4).

Figure 3.

Comparison of mortality risk and healthcare utilization between type 1 and type 2 diabetes mellitus. This figure displays the differences in mortality proportion, length of hospital stay, and hospitalization costs between patients with type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM). (A) The mortality proportion for T1DM and T2DM patients is presented; T2DM have a significantly higher mortality proportion compared to T1DM (***, P < 0.001). Error bars represent the standard error of the mean. (B) The length of stay in days for T1DM and T2DM patients is presented; T2DM patients have a longer hospital stay than their T1DM counterparts (***P < 0.001). Boxplots indicate the interquartile range, with the median marked within each box. (C) The hospitalization costs for T1DM and T2DM patients are shown. T2DM patients have higher hospitalization costs compared to T1DM patients (***P < 0.001). Boxplots indicate the interquartile range, with the median marked within each box.

Table 4.

Comparison of hospital outcomes in DKA hospitalization patients with and without CVD in the T1DM and T2DM.

Variable	Adjusted mean difference*	95% CI	P value
T1DM
Length of stay (days)	0.98	0.86–1.10	<0.001
Hospitalization costs ($)	13,152.69	11,790.11–14,515.28	<0.001
T2DM DKA
Length of stay (days)	0.74	0.51–0.96	<0.001
Hospitalization costs ($)	13,839.31	11,049.86–16,628.72	<0.001

^*Model adjusted for age, race, admission type, insurance status, income quartile, patient location, hospital bed size, hospital location/teaching status, number of medical comorbidities, and admission year.

Furthermore, as illustrated in Fig. 4, both T1DM and T2DM DKA patients exhibited considerable disparities in clinical outcomes based on their CVD status. This suggests that the presence of CVD is linked to poorer clinical outcomes for patients with DKA, implying the importance of considering CVD when managing DKA in both types of diabetes.

Figure 4.

Adverse outcomes in patients with type 1 and type 2 diabetes mellitus. This figure presents the incidence of various clinical outcomes in patients with T1DM and T2DM. (A) The aOR with 95% CI for adverse outcomes in T1DM patients is displayed. The outcomes include in-hospital mortality (death), noninvasive ventilation (NV), mechanical ventilation (MV), deep vein thrombosis (DVT), pulmonary embolism (PE), septic shock (SS), sepsis, acute liver failure (ALF), acute neurological failure (ANF), acute kidney failure (AKF), and acute respiratory failure (ARF). The blue circles represent the aOR, while the red bars indicate the 95% CI. (B) The same outcomes for T2DM patients following the same format are illustrated. For (A) and (B), an aOR of 1.0 indicates no effect. Values exceeding 1.0 suggest an increased risk, while values below 1.0 indicate a reduced risk for the specified outcomes.

A comparison of the characteristics and outcomes of diabetes patients with CVD who experienced DKA versus those who experienced HHS

DKA and HHS represent two critical acute complications of diabetes mellitus, frequently requiring emergency intervention. To explore whether the interplay between DKA and CVD leads to compounded risks that are less pronounced in other acute conditions, we examined the clinical differences between diabetes patients with CVD who developed DKA and those who experienced HHS. To ensure diagnostic clarity, cases with co-occurring DKA and HHS were excluded from the final cohort. Among diabetes patients with CVD, those who developed DKA exhibited distinct demographic and clinical differences compared to those with HHS (Supplementary Table 5). The DKA cohort was younger, had a higher proportion of females, and included more White individuals. Notably, DKA patients were less likely to reside in low-income zip codes and were more frequently admitted to urban teaching hospitals. Clinically, DKA was associated with significantly worse outcomes (Table 5). After adjusting for confounders, DKA patients had 2.30-fold higher odds of in-hospital mortality (95% CI 2.11–2.51) and were more likely to develop sepsis (aOR 2.00, 95% CI 1.91–2.09), septic shock (aOR 1.71, 95% CI 1.49–1.96), acute respiratory failure (aOR 1.41, 95% CI 1.35–1.48), acute kidney failure (aOR 1.15, 95% CI 1.10–1.19), acute neurological failure (aOR 1.16, 95% CI 1.08–1.25), pulmonary embolism (aOR 1.58, 95% CI 1.32–1.89), deep vein thrombosis (aOR 1.52, 95% CI 1.29–1.80), and acute liver failure (aOR 2.71, 95% CI 2.29–3.22) compared to HHS. In addition, DKA was associated with increased utilization of mechanical ventilation (aOR 1.91, 95% CI 1.79–2.03) and noninvasive ventilation (aOR 1.15, 95% CI 1.06–1.26) compared to HHS (Table 5). DKA was also associated with prolonged hospitalization (+1.55 days, 95% CI 1.42–1.68) and higher costs (+$31,352.90, 95% CI 29313.82–33,391.98) compared to HHS (Supplementary Table 7). In summary, compared to HHS, DKA patients exhibit worse clinical outcomes and greater healthcare resource utilization.

Table 5.

Comparison of clinical outcomes between diabetes patients with CVD who experienced DKA and those who experienced HHS.

Morbidity	HHS	DKA	Crude OR (95% CI)	aOR (95% CI)*
	n = 16,367	n = 62,052
In-hospital mortality	632 (3.9)	4,675 (7.5)	2.03 (1.86–2.219)	2.30 (2.11–2.51)
Acute respiratory failure	3,306 (20.2)	15,765 (25.4)	1.35 (1.29–1.40)	1.41 (1.35–1.48)
Septic shock	255 (1.6)	1,498 (2.4)	1.56 (1.37–1.79)	1.71 (1.49–1.96)
Acute kidney failure	9,853 (60.2)	37,229 (60.0)	0.99 (0.96–1.03)	1.15 (1.10–1.19)
Sepsis	2,843 (17.4)	17,398 (28.0)	1.85 (1.77–1.94)	2.00 (1.91–2.09)
Acute neurological failure	1,078 (6.6)	4,197 (6.8)	1.03 (0.96–1.10)	1.16 (1.08–1.25)
Pulmonary embolism	143 (0.9)	864 (1.4)	1.60 (1.34–1.91)	1.58 (1.32–1.89)
Deep vein thrombosis	177 (1.1)	1,012 (1.6)	1.52 (1.29–1.78)	1.52 (1.29–1.80)
Acute liver failure	149 (0.9)	1,575 (2.5)	2.83 (2.39–3.36)	2.71 (2.29–3.22)
Mechanical ventilation	1,229 (7.5)	8,223 (13.3)	1.88 (1.77–2.00)	1.91 (1.79–2.03)
Noninvasive ventilation	689 (4.2)	2,780 (4.5)	1.07 (0.98–1.16)	1.15 (1.06–1.26)

^*Model adjusted for age, race, admission type, insurance status, income quartile, patient location, hospital bed size, hospital location/teaching status, number of medical comorbidities, and admission year.

Outcomes and characteristics of diabetes patients with CVD: a comparison between those who developed DKA and those who did not develop DKA-HHS

Next, we aimed to compare the clinical characteristics, outcomes, and healthcare utilization between diabetes patients with CVD who developed DKA and those who did not develop DKA-HHS. As shown in Supplementary Table 6, those who developed DKA were younger, more likely to be female, and had higher proportions of Black individuals and Medicaid recipients compared to non-DKA-HHS patients. DKA was associated with substantially worse clinical outcomes (Table 6), including higher in-hospital mortality (aOR 2.28, 95% CI 2.21–2.35), increased sepsis (aOR 2.68, 95% CI 2.63–2.73), septic shock (aOR 2.15, 95% CI 2.04–2.26), acute respiratory failure (aOR 1.04, 95% CI 1.02–1.06), acute kidney failure (aOR 3.66, 95% CI 3.60–3.73), acute liver failure (aOR 2.58, 95% CI 2.45–2.71), and deep vein thrombosis (aOR 1.23, 95% CI 1.16–1.31). In addition, DKA patients required mechanical ventilation more frequently (aOR 2.64, 95% CI 2.58–2.71) (Table 6). Resource utilization was significantly higher in the DKA group, with prolonged hospital stays (+1.66 days, 95% CI 1.62–1.70) and increased costs (+$22,287.93, 95% CI $21,621.72–$22,954.15) (Supplementary Table 7). These comparative analyses confirm that DKA places a significant additional clinical and economic burden on diabetic patients with CVD, exceeding those experienced by their counterparts without DKA and HHS who also have CVD.

Table 6.

Comparison of clinical outcomes between diabetes patients with CVD who experienced DKA and those who did not develop DKA-HHS.

Morbidity	Non-DKA-HHS	DKA	Crude OR (95% CI)	aOR (95% CI)*
	n = 5,955,439	n = 62,052
In-hospital mortality	241,201 (4.1)	4,675 (7.5)	1.93 (1.87–1.99)	2.28 (2.21–2.35)
Acute respiratory failure	1,434,559 (24.1)	15,765 (25.4)	1.07 (1.05–1.09)	1.04 (1.02–1.06)
Septic shock	71,973 (1.2)	1,498 (2.4)	2.02 (1.92–2.13)	2.15 (2.04–2.26)
Acute kidney failure	1,821,285 (30.6)	37,229 (60.0)	3.40 (3.35–3.46)	3.66 (3.60–3.73)
Sepsis	752,538 (12.6)	17,398 (28.0)	2.69 (2.65–2.74)	2.68 (2.63–2.73)
Acute neurological failure	279,756 (4.7)	4,197 (6.8)	1.47 (1.43–1.52)	1.83 (1.78–1.89)
Pulmonary embolism	78,984 (1.3)	864 (1.4)	1.05 (0.98–1.12)	0.89 (0.83–0.95)
Deep vein thrombosis	78,066 (1.3)	1,012 (1.6)	1.25 (1.17–1.33)	1.23 (1.16–1.31)
Acute liver failure	51,408 (0.9)	1,575 (2.5)	2.99 (2.84–3.15)	2.58 (2.45–2.71)
Mechanical ventilation	296,947 (5.0)	8,223 (13.3)	2.91 (2.84–2.98)	2.64 (2.58–2.71)
Noninvasive ventilation	324,863 (5.5)	2,780 (4.5)	0.81 (0.78–0.84)	0.75 (0.72–0.78)

^*Model adjusted for age, race, admission type, insurance status, income quartile, patient location, hospital bed size, hospital location/teaching status, number of medical comorbidities, and admission year.

Discussion

In this large-scale national study, we analyzed hospitalization data from 2016 to 2022 and demonstrated the significant impacts of CVD on clinical outcomes, healthcare utilization, and in-hospital mortality in patients admitted with DKA. In addition, comparable adverse results were noted when comparing patients with DKA to those with HHS, as well as between those who experienced DKA and those who did not. Our findings show that CVD is not merely a comorbidity but a crucial factor determining the severity of the disease and the prognosis for hospitalized patients with DKA.

Our analysis found that the prevalence of CVD among hospitalized DKA patients was 16.8%. This relatively high prevalence highlights the substantial burden of cardiovascular comorbidities in individuals with diabetes experiencing acute metabolic decompensation. This elevated prevalence suggests the need for careful cardiovascular risk assessment and management in patients hospitalized for DKA.

The presence of CVD had a striking impact on in-hospital mortality rates among DKA patients. In the current cohort, patients with CVD had over a fivefold higher risk of in-hospital mortality compared to those without CVD (1.6 vs 0.3%, P < 0.001). Multivariable analyses confirmed that CVD was independently associated with increased odds of mortality, with specific subtypes such as arrhythmias showing a remarkable aOR of 73.27. This finding is in line with previous data suggesting that cardiovascular events worsen the metabolic disruptions of DKA (8, 9, 10), further complicating patient management. Furthermore, pulmonary hypertension and cardiomyopathy also contributed significantly to mortality. Chronic coronary artery disease, valvular heart diseases, and congenital heart disease did not exhibit significant associations with in-hospital mortality. These findings reflect the complex and varied mechanisms through which different types of CVD influence outcomes in DKA.

Notably, patients with arrhythmias exhibited significantly elevated aORs for various adverse outcomes, particularly in-hospital mortality, implying the significant impact of arrhythmias on worsening the prognosis of hospitalized DKA patients. The specific factors leading to this worsening prognosis in DKA patients with arrhythmias remain poorly understood. AF, the most common type of arrhythmia, is associated with considerable morbidity and mortality (23). Research indicates that diabetic patients are approximately 40% more likely to develop AF compared to non-diabetic patients (24). Among patients hospitalized for DKA, the presence of comorbid AF has been associated with higher in-hospital mortality rates and prolonged LOS (9). There are bidirectional associations between AF and insulin resistance (IR), overactivation of stress hormones, and vascular endothelial dysfunction. IR causes alterations in atrial structure and disrupts intracellular calcium balance, both of which increase susceptibility to AF (25). Chronic stress stimulates the overactivation of stress hormones through the hypothalamic–pituitary–adrenal axis, negatively affecting cardiac tissue and electrical conduction, while also promoting inflammation and oxidative stress, all of which contribute to the onset and progression of AF (26). Emerging evidence indicates a significant relationship between AF and endothelial dysfunction in the atria, coronary arteries, and systemic circulation, suggesting a potential bidirectional relationship that may exacerbate both conditions (27). In addition, AF may lead to elevated stress hormone levels and a relative deficiency of insulin, which can worsen the severity of DKA (9). In our current study, we found that CVD was associated with an increased risk of various complications, including acute respiratory failure, acute neurological failure, acute kidney failure, sepsis, septic shock, pulmonary embolism, acute liver failure, and deep vein thrombosis, along with a higher demand for mechanical ventilation and the use of noninvasive ventilation. The presence of these complications is also likely to raise the overall risk of mortality. Despite this, it is not possible to establish causality in this retrospective study.

Acute respiratory failure and septic shock, among the most prominent complications, highlight the vulnerability of DKA patients with CVD to multi-organ dysfunction (28). The elevated risk of pulmonary embolism and deep vein thrombosis in these patients may stem from a combination of immobilization during hospitalization, inflammation, hypercoagulability associated with DKA, and underlying cardiovascular dysfunction (29, 30).

The results showed that CVD substantially increased healthcare resource utilization among DKA patients. Patients with CVD required an average of 1.03 additional hospital days and spent $12,931.72 more in treatment costs, implying the burden this comorbidity imposes on healthcare systems. Among specific CVD subtypes, pulmonary hypertension was associated with the highest increases in LOS and costs, likely reflecting the complexity and the need for intensive care. Interestingly, congenital heart disease was not associated with significant differences in LOS or costs, warranting further investigation into the mechanisms.

Several studies have examined DKA outcomes in T1DM versus T2DM. A large retrospective cohort of 1,244,184 hospitalized DKA patients found that T2DM was associated with higher mortality rates, longer hospital stays, and increased costs (31). Similarly, research by Shaka et al. indicated that T2DM patients had greater mortality risk and higher odds of septic shock, along with elevated hospital charges and extended hospitalizations (32). Eledrisi et al. analyzed 1,330 DKA cases in Qatar, finding significantly higher ICU admission rates, longer stays, and mortality in T2DM patients (33). These findings highlight substantial disparities in DKA outcomes, particularly the higher risk for T2DM patients. Here, we investigated whether similar trends exist in DKA patients with CVD. Our stratified analysis revealed that both T1DM and T2DM patients experienced significant adverse effects from CVD, but T2DM patients demonstrated a greater impact, particularly regarding in-hospital mortality, healthcare costs, and LOS. Nevertheless, CVD increases the burden of DKA in both types of diabetes, implying the need for interventions regardless of diabetes classification.

Our study also provides important insights into the differential clinical impact of DKA and HHS among diabetes patients with CVD, as well as the substantial burden associated with DKA compared to non-DKA-HHS patients in this population. The findings demonstrate that DKA is associated with significantly worse clinical outcomes, higher mortality, and greater healthcare resource utilization than both HHS and non-DKA-HHS cases, underscoring the severity of DKA in this high-risk group.

The present findings indicate critical implications for both the prevention and management of DKA in patients with CVD. Given the markedly worse clinical outcomes observed once DKA develops, strict glycemic control in patients with pre-existing CVD is essential to prevent DKA occurrence and its associated burdens of prolonged hospitalization, increased healthcare costs, and higher risk of death (7, 34).

In cases where DKA does occur, early identification and comprehensive evaluation of cardiovascular comorbidities are essential, as their presence independently contributes to adverse in-hospital outcomes. Optimizing multidisciplinary care pathways may reduce complication rates, shorten LOS, and improve survival (35). Furthermore, implementing proactive strategies to prevent and manage major complications is vital for improving recovery (36). Finally, the heterogeneity of outcomes across different CVD subtypes highlights the need for individualized therapeutic approaches. Patients with arrhythmias, pulmonary hypertension, heart failure, and cardiomyopathy may benefit from more intensive management and monitoring, whereas those with stable chronic coronary artery disease, valvular heart disease, or congenital heart disease may require less aggressive intervention.

This study has several limitations. First, as an observational study, we can only identify associations rather than prove causality (37). In addition, the reliance on ICD-10 codes for disease classification in the NIS database may lead to errors and misclassifications. We also lacked access to important information regarding the severity of CVD and its underlying causes, both of which are crucial for understanding their effects on outcomes (38). Furthermore, severe DKA is known to be associated with poor outcomes (39). If instances of severe DKA were unevenly distributed between the comparison groups, this could bias our estimates. The NIS database does not provide information on medication adherence or detailed treatment regimens. Poor adherence to guideline-directed therapies may worsen both CVD and DKA outcomes (40). If adherence levels differed systematically between groups, it could confound the observed associations. In addition, the NIS lacks data on lab results (such as pH and ketone levels) and disease duration, which could introduce further confounding factors. Longitudinal glycemic markers (e.g., HbA1c, glucose variability) are also unavailable in NIS. Chronic hyperglycemia may accelerate the progression of CVD and predispose patients to DKA, creating a potential confounding pathway. The introduction of SGLT2 inhibitors has significantly improved the management of type 2 diabetes, particularly in patients with CVD. However, their use has been associated with an increased risk of DKA (41, 42). This paradoxical effect presents an important safety consideration in clinical practice. A significant limitation of this study is the failure to account for the use of SGLT2i among participants, which may affect our understanding of the complex relationship between CVD and DKA outcomes. While our study accounted for key demographic and clinical variables, missing data could still introduce residual confounding. Finally, our analysis emphasizes hospital admissions rather than the specifics of individual patients, potentially overstating the admission rates as the same patients may be hospitalized multiple times. Despite these limitations, our findings offer valuable insights into the interplay between DKA and cardiovascular conditions, highlighting the need for more extensive, controlled prospective studies in the future.

Conclusion

This study highlights the substantial impact of CVD on both clinical outcomes and the economic burden of DKA hospitalizations. CVD not only increases mortality risk but also contributes to higher complication rates, prolonged hospital stays, and greater healthcare costs. These findings underscore the importance of close cardiac monitoring in DKA patients with comorbid CVD, as well as proactive glycemic control in diabetic patients with CVD to prevent DKA onset. A deeper understanding of the interplay between metabolic and cardiovascular pathophysiology may ultimately support more effective risk stratification and improve outcomes in this high-risk population.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

Funding

The study is funded by the Young Scientists Fund of the National Natural Science Foundation of China (Grant No. 82100930), the Guangdong Basic and Applied Basic Research Foundation (No. 2025A1515012538, No. 2025A1515010231, and No. 2023A1515010452), and the Guizhou Provincial Health Commission Science and Technology Fund (No. gzwkj2025-003).