OBJECTIVE

To compare the risks of severe hypoglycemia and mortality between patients with type 2 diabetes (T2D) and discharged with and without acute liver injury.

RESEARCH DESIGN AND METHODS

From 1 January 2000 to 31 December 2010, we identified patients with T2D and hospitalization for acute liver injury and hospitalization for other causes from the National Health Insurance Research Database of Taiwan. Multivariable-adjusted Cox proportional hazards models were used to compare the risks of severe hypoglycemia and mortality between the study and control groups.

RESULTS

The incidence rates and adjusted hazard ratios (aHRs) for severe hypoglycemia within 90 days and 365 days after discharge were 12.28 and 5.59/1,000 person-years (aHR 1.92 [1.30–2.85]) and 7.35 and 2.9/1,000 person-years (aHR 1.98 [1.52–2.58]) for patients discharged with and without acute liver injury, respectively. The incidence rates and aHRs for mortality within 90 days and 365 days after discharge were 82.4 and 27.54/1,000 person-years (aHR 1.73 [1.46–2.05]) and 36.8 and 9.3/1,000 person-years (aHR 1.94 [1.69–2.24]) for patients discharged with and without acute liver injury, respectively. The subgroup analysis of hypoglycemia risk in patients discharged with acute liver injury revealed no significant interaction in risk factors of age, chronic kidney disease, and medications, except for sex difference, which has significant interaction.

CONCLUSIONS

This cohort study demonstrated that patients with T2D and discharged with acute liver injury showed significantly higher risks of severe hypoglycemia and mortality within 90 days and 365 days after discharge than patients discharged with other causes.

As early as 1853, French physiologist Claude Bernard said, “Sugar is manufactured in the liver, which must, therefore, be considered an organ which produces or secretes sugar” (1). The liver is a vital organ for the metabolism of carbohydrates. Gluconeogenesis and glycogenolysis are the main processes in which the liver produces glucose to maintain blood glucose levels in varying physiological states (2). Mann and colleagues experimented with the removal of the liver from dogs resulting in hypoglycemia and cerebral dysfunction (1). Patients with alcoholic liver cirrhosis are susceptible to hypoglycemia due to nutritional imbalances and inadequate hepatic glycogen storage due to long-term chronic liver injury (1). Fulminant hepatitis, caused by the toxic effects of drugs or viral hepatitis with rapid and extensive liver damage, is also reported to result in hypoglycemia and death (3). Medications, toxins, viral hepatitis, ischemia, or other rare causes can lead to acute liver injury or hepatic failure (4). This study aimed to investigate whether patients discharged from the hospital with acute liver injury were susceptible to hypoglycemia and mortality risks.

The liver is an important organ for the metabolism and clearance of drugs (5). Oral medications are absorbed through the intestinal tract, enter the portal system, are metabolized by various intracellular enzymes of hepatocytes, and excreted into the systemic circulation or biliary tracts. Extensive liver damage may affect the clearance and blood level of drugs (5). The effect of medication use on hypoglycemia risk after discharge from acute liver injury remains unclear. Therefore, we also evaluated age, sex, chronic kidney disease (CKD), and medications for the impact of severe hypoglycemia in these patients.

Data Source

The Taiwan government implemented the National Health Insurance (NHI) program in 1995. People pay a small premium for access to health care. Most premiums are paid by employers and the government, and the government is the only health insurance buyer. By 2000, 99% of the 23 million people were enrolled in the NHI program (6). Information about the insured’s health care is recorded in the NHI research data set (NHIRD). This includes age, sex, premium levels, address, outpatient and inpatient diagnoses, medications, and procedures. The diagnostic codes are based on the ICD-9-Clinical Modification. The Health Insurance Authority conducts annual random checks on the medical records of clinics and hospitals to ensure the accuracy of medical diagnosis and the appropriateness of medical practice. This study identified participants from the NHIRD and received approval from the Institutional Review Board of the National Health Research Institutes (EC1060704-E). All information of patients and caregivers was encrypted before release to protect individual privacy. The Institutional Review Board approved a waiver of informed consent.

Study Design

We identified participants diagnosed with type 2 diabetes (T2D) and hospitalized from 1 January 2000 to 31 December 2010 (Supplementary Fig. 1). T2D was defined using the ICD-9-Clinical Modification code 250.xx for at least two outpatient diagnoses in 1 year or one hospitalization. The algorithm for the definition of T2D using ICD-9 coding was validated by a previous Taiwan study, with an accuracy of 74.6% (7). We divided our inpatients into those with acute liver injury and those with other causes. Acute liver injury was defined as a diagnosis of other specified liver disorders (573.8), other sequelae of chronic liver disease (572.8), drug-induced liver injury (573.3), acute and subacute hepatic necrosis (570, 571.1, 070.1, 0.70.30, and 070.51), elevated ALT (790.4), abnormal liver function test (794.8), disorders of bilirubin excretion (277.4), other specified biliary tract disorders (576.8), jaundice (782.4), fulminant hepatitis (070.6, 070.0, 070.20, 070.21, 070.22, 070.23, 070.41, 070.71, 070.42, and 070.43), and hepatic failure (572.2, 572.4, and 572.8) (8,9). We excluded participants with missing information on sex, hospitalization before the diagnosis of T2D, duplicated hospitalization, admission <2 days or >30 days, diabetic initial age <20 or >75 years, diagnosis of type 1 diabetes (250.1), diagnoses of chronic dialysis (V56.0, V56.8, and V45.1), heart failure (428), liver transplantation (V42.7, 996.82, procedure code 50.5), dementia (294 and 331.0), cancers (140–239, except 210–229), and history of severe hypoglycemia before the index date.

Procedures

We defined the first day of discharge with acute liver injury as the index date. For participants without acute liver injury, we assigned their index dates as the first day of discharge with other causes. This study included potential confounding variables that might interfere with the results, such as age, sex, alcohol-related disorders (291, 303, 305.0, 571.0, 570.1, 571.3, V11.3, and V79.1), smoking status (305.1, 649.0, and V15.82), obesity (overweight [278.02, 783.1, and V85.2], obesity [278.00, 649.1, V77.8, and V85.3], severe obesity [278.01, 649.2, V45.86, and V85.4]), hepatitis C virus (HCV) infection (070.41, 070.44, 070.51, 070.54, 070.70, 070.71, and V02.62), HCV therapy (interferons), hepatitis B virus (HBV) infection (070.2, 070.3, and V02.61), HBV therapy (lamivudine, tenofovir disoproxil, adefovir dipivoxil, entecavir, and telbivudine), other viral hepatitis (573.1 and 573.2), drug-induced liver injury (573.3), liver cirrhosis (571.5, 571.2, and 571.6), decompensated cirrhosis (789.5, 572.2, and 456.0), CKD (403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 581–588, and A350), chronic obstructive pulmonary disease (491, 492, and 496), Charlson Comorbidity Index (CCI) scores (10) (comorbidities diagnosed within 1 year of the index date), and medications, such as metformin, sulfonylureas (SU), glinides, thiazolidinedione, dipeptidyl peptidase 4 inhibitors (DPP-4i), basal insulin, premixed insulin, short-acting insulin, ACE inhibitor, angiotensin receptor blockers (ARBs), β-blockers, calcium channel blockers, diuretics, statins, fibrate, aspirin, and the enrolled years of 2000–2004, 2005–2008, and 2009–2010. We used the Diabetes Complications Severity Index (DCSI) score (11), duration of diabetes, and the number of oral hypoglycemic agents to reflect the severity of diabetes in this study.

Main Outcomes

We assessed the incidence rates of severe hypoglycemia and all-cause mortality within 90 days and 365 days after discharge and over the follow-up period between the study and control groups. Severe hypoglycemia was defined as participants referred to the emergency department or hospitalized due to hypoglycemia (251.0x, 251.1x, or 251.2x) (12). All-cause mortality was censored, as participants were discharged from the hospital with certified death (the discharge date was defined as the mortality date) or termination of the NHI coverage after hospital discharge due to a critical illness and without further health care utilization in the NHI records for >1 year (the end of NHI coverage was defined as the mortality date).

Statistical Analysis

This study included 39 clinically related variables and the propensity score in the analysis as independent covariates (Table 1). The Student t test was used to determine the statistical difference for continuous variables; for categorical variables, the χ2 test was used to test the statistical difference between the study and control groups. Propensity score matching was used to mitigate the imbalance between participants with and without acute liver injury (13). The propensity score was estimated for every participant by a nonparsimonious multivariable logistic regression with discharge for acute liver injury as the dependent variable. The nearest-neighbor algorithm with the caliper width <0.00001 was used to construct matched pairs, with the standardized differences <0.05 and P value >0.05 as perfect.

Table 1

Basic characteristics of prematched and postmatched data in acute liver injury with diabetes

Pre–propensity score matchedPost–propensity score matched
With acute liver injuryWithout acute liver injurySIDP valueWith acute liver injuryWithout acute liver injurySIDP value
n 18,498 126,584   3,937 3,937   
Age group (years)         
 20–39 1,635 (8.8) 10,455 (8.3) 0.021 0.008 359 (9.1) 389 (9.9) 0.026 0.25 
 40–64 12,577 (68) 84,618 (66.8) 0.024 0.002 2,598 (66) 2,590 (65.8) 0.004 0.85 
 65–74 4,286 (23.2) 31,511 (24.9) 0.04 <0.001 980 (24.9) 958 (24.3) 0.013 0.56 
 Mean (SD) 55.2 (11.1) 56 (11) 0.073 <0.001 55.6 (11.4) 55.5 (11.3) 0.004 0.85 
Sex       
 Male 12,207 (66) 69,990 (55.3) 0.22 <0.001 2,253 (57.2) 2,300 (58.4) 0.024 0.28 
 Female 6,291 (34) 56,594 (44.7) 0.22 <0.001 1,684 (42.8) 1,637 (41.6) 0.024 0.28 
Comorbidity       
 Alcohol 3,367 (18.2) 2,473 (2) 0.561 <0.001 90 (2.3) 95 (2.4) 0.008 0.71 
 Obesity 264 (1.4) 2,022 (1.6) 0.014 0.08 67 (1.7) 63 (1.6) 0.008 0.72 
 HBV 4,719 (25.5) 4,867 (3.8) 0.643 <0.001 170 (4.3) 186 (4.7) 0.02 0.39 
 HBV therapy 229 (1.2) 113 (0.1) 0.142 <0.001 10 (0.3) 1 (0) 0.061 0.007 
 HCV 4,126 (22.3) 2,491 (2) 0.655 <0.001 78 (2) 65 (1.7) 0.025 0.27 
 HCV therapy 107 (0.6) 111 (0.1) 0.085 <0.001 6 (0.2) 2 (0.1) 0.032 0.16 
 CKD 2,545 (13.8) 13,167 (10.4) 0.103 <0.001 444 (11.3) 403 (10.2) 0.034 0.14 
 Other viral liver disease 50 (0.3) 77 (0.1) 0.052 <0.001 3 (0.1) 1 (0) 0.023 0.32 
 Drug-induced liver disease 3,997 (21.6) 3,331 (2.6) 0.608 <0.001 138 (3.5) 112 (2.8) 0.038 0.09 
 Cirrhosis 4,813 (26) 2,475 (2) 0.74 <0.001 77 (2) 60 (1.5) 0.033 0.14 
 COPD 2,662 (14.4) 12,598 (10) 0.136 <0.001 482 (12.2) 431 (10.9) 0.04 0.07 
CCI scores        
 Mean (SD) 5.2 (2.2) 2.9 (1.9) 1.116 <0.001 3.2 (1.7) 3.2 (1.9) 0.02 0.16 
DCSI         
 0 7,745 (41.9) 54,454 (43) 0.023 0.003 1,641 (41.7) 1,739 (44.2) 0.05 0.026 
 1 4,373 (23.6) 28,578 (22.6) 0.025 0.001 880 (22.4) 861 (21.9) 0.012 0.61 
 ≥2 6,380 (34.5) 43,552 (34.4) 0.002 0.82 1,416 (36) 1,337 (34) 0.042 0.06 
Diabetic duration, years 2.6 (2.4) 2.6 (2.4) 0.003 0.73 2.5 (2.4) 2.6 (2.4) 0.023 0.32 
Prescription        
 Metformin 6,707 (36.3) 50,400 (39.8) 0.073 <0.001 1,534 (39) 1,472 (37.4) 0.032 0.15 
 SU 7,763 (42) 55,890 (44.2) 0.044 <0.001 1,722 (43.7) 1,646 (41.8) 0.039 0.08 
 TZD 1,030 (5.6) 8,201 (6.5) 0.038 <0.001 233 (5.9) 239 (6.1) 0.006 0.78 
 DPP-4i 119 (0.6) 849 (0.7) 0.003 0.67 30 (0.8) 25 (0.6) 0.015 0.5 
Oral antidiabetic drugs        
 0 9,106 (49.2) 57,638 (45.5) 0.074 <0.001 1,818 (46.2) 1,922 (48.8) 0.053 0.019 
 1 3,067 (16.6) 23,676 (18.7) 0.056 <0.001 725 (18.4) 663 (16.8) 0.041 0.07 
 2 4,814 (26) 34,858 (27.5) 0.034 <0.001 1,085 (27.6) 1,050 (26.7) 0.02 0.37 
 ≥3 1,511 (8.2) 10,412 (8.2) 0.002 0.79 309 (7.8) 302 (7.7) 0.007 0.77 
Insulin         
 Basal 131 (0.7) 512 (0.4) 0.041 <0.001 15 (0.4) 18 (0.5) 0.012 0.6 
 Premixed 328 (1.8) 759 (0.6) 0.109 <0.001 19 (0.5) 35 (0.9) 0.049 0.029 
 Fast-acting 416 (2.2) 791 (0.6) 0.137 <0.001 33 (0.8) 19 (0.5) 0.044 0.051 
ACEi/ARBs 4,066 (22) 33,549 (26.5) 0.106 <0.001 1,007 (25.6) 973 (24.7) 0.02 0.38 
β-Blocker 2,226 (12) 18,573 (14.7) 0.078 <0.001 574 (14.6) 537 (13.6) 0.027 0.23 
CCB 4,143 (22.4) 31,643 (25) 0.061 <0.001 974 (24.7) 936 (23.8) 0.023 0.32 
Diuretics 3,376 (18.3) 16,603 (13.1) 0.142 <0.001 513 (13) 545 (13.8) 0.024 0.29 
Statin 2,074 (11.2) 21,806 (17.2) 0.173 <0.001 633 (16.1) 610 (15.5) 0.016 0.48 
Fibrate 1,305 (7.1) 9,948 (7.9) 0.031 0.001 332 (8.4) 300 (7.6) 0.03 0.18 
Aspirin 2,219 (12) 18,640 (14.7) 0.08 <0.001 566 (14.4) 543 (13.8) 0.017 0.46 
Enrolled research year        
 2000–2004 5,102 (27.6) 33,703 (26.6) 0.022 0.006 1,107 (28.1) 1,024 (26) 0.047 0.035 
 2005–2008 7,996 (43.2) 55,661 (44) 0.015 0.06 1,701 (43.2) 1,720 (43.7) 0.01 0.67 
 2009–2010 5,398 (29.2) 37,220 (29.4) 0.005 0.54 1,127 (28.6) 1,193 (30.3) 0.037 0.1 
Propensity score 0.42331 (0.3048) 0.08427 (0.1274) 1.451 <0.001 0.09098 (0.1193) 0.09098 (0.1193) >0.99 <0.001 
Pre–propensity score matchedPost–propensity score matched
With acute liver injuryWithout acute liver injurySIDP valueWith acute liver injuryWithout acute liver injurySIDP value
n 18,498 126,584   3,937 3,937   
Age group (years)         
 20–39 1,635 (8.8) 10,455 (8.3) 0.021 0.008 359 (9.1) 389 (9.9) 0.026 0.25 
 40–64 12,577 (68) 84,618 (66.8) 0.024 0.002 2,598 (66) 2,590 (65.8) 0.004 0.85 
 65–74 4,286 (23.2) 31,511 (24.9) 0.04 <0.001 980 (24.9) 958 (24.3) 0.013 0.56 
 Mean (SD) 55.2 (11.1) 56 (11) 0.073 <0.001 55.6 (11.4) 55.5 (11.3) 0.004 0.85 
Sex       
 Male 12,207 (66) 69,990 (55.3) 0.22 <0.001 2,253 (57.2) 2,300 (58.4) 0.024 0.28 
 Female 6,291 (34) 56,594 (44.7) 0.22 <0.001 1,684 (42.8) 1,637 (41.6) 0.024 0.28 
Comorbidity       
 Alcohol 3,367 (18.2) 2,473 (2) 0.561 <0.001 90 (2.3) 95 (2.4) 0.008 0.71 
 Obesity 264 (1.4) 2,022 (1.6) 0.014 0.08 67 (1.7) 63 (1.6) 0.008 0.72 
 HBV 4,719 (25.5) 4,867 (3.8) 0.643 <0.001 170 (4.3) 186 (4.7) 0.02 0.39 
 HBV therapy 229 (1.2) 113 (0.1) 0.142 <0.001 10 (0.3) 1 (0) 0.061 0.007 
 HCV 4,126 (22.3) 2,491 (2) 0.655 <0.001 78 (2) 65 (1.7) 0.025 0.27 
 HCV therapy 107 (0.6) 111 (0.1) 0.085 <0.001 6 (0.2) 2 (0.1) 0.032 0.16 
 CKD 2,545 (13.8) 13,167 (10.4) 0.103 <0.001 444 (11.3) 403 (10.2) 0.034 0.14 
 Other viral liver disease 50 (0.3) 77 (0.1) 0.052 <0.001 3 (0.1) 1 (0) 0.023 0.32 
 Drug-induced liver disease 3,997 (21.6) 3,331 (2.6) 0.608 <0.001 138 (3.5) 112 (2.8) 0.038 0.09 
 Cirrhosis 4,813 (26) 2,475 (2) 0.74 <0.001 77 (2) 60 (1.5) 0.033 0.14 
 COPD 2,662 (14.4) 12,598 (10) 0.136 <0.001 482 (12.2) 431 (10.9) 0.04 0.07 
CCI scores        
 Mean (SD) 5.2 (2.2) 2.9 (1.9) 1.116 <0.001 3.2 (1.7) 3.2 (1.9) 0.02 0.16 
DCSI         
 0 7,745 (41.9) 54,454 (43) 0.023 0.003 1,641 (41.7) 1,739 (44.2) 0.05 0.026 
 1 4,373 (23.6) 28,578 (22.6) 0.025 0.001 880 (22.4) 861 (21.9) 0.012 0.61 
 ≥2 6,380 (34.5) 43,552 (34.4) 0.002 0.82 1,416 (36) 1,337 (34) 0.042 0.06 
Diabetic duration, years 2.6 (2.4) 2.6 (2.4) 0.003 0.73 2.5 (2.4) 2.6 (2.4) 0.023 0.32 
Prescription        
 Metformin 6,707 (36.3) 50,400 (39.8) 0.073 <0.001 1,534 (39) 1,472 (37.4) 0.032 0.15 
 SU 7,763 (42) 55,890 (44.2) 0.044 <0.001 1,722 (43.7) 1,646 (41.8) 0.039 0.08 
 TZD 1,030 (5.6) 8,201 (6.5) 0.038 <0.001 233 (5.9) 239 (6.1) 0.006 0.78 
 DPP-4i 119 (0.6) 849 (0.7) 0.003 0.67 30 (0.8) 25 (0.6) 0.015 0.5 
Oral antidiabetic drugs        
 0 9,106 (49.2) 57,638 (45.5) 0.074 <0.001 1,818 (46.2) 1,922 (48.8) 0.053 0.019 
 1 3,067 (16.6) 23,676 (18.7) 0.056 <0.001 725 (18.4) 663 (16.8) 0.041 0.07 
 2 4,814 (26) 34,858 (27.5) 0.034 <0.001 1,085 (27.6) 1,050 (26.7) 0.02 0.37 
 ≥3 1,511 (8.2) 10,412 (8.2) 0.002 0.79 309 (7.8) 302 (7.7) 0.007 0.77 
Insulin         
 Basal 131 (0.7) 512 (0.4) 0.041 <0.001 15 (0.4) 18 (0.5) 0.012 0.6 
 Premixed 328 (1.8) 759 (0.6) 0.109 <0.001 19 (0.5) 35 (0.9) 0.049 0.029 
 Fast-acting 416 (2.2) 791 (0.6) 0.137 <0.001 33 (0.8) 19 (0.5) 0.044 0.051 
ACEi/ARBs 4,066 (22) 33,549 (26.5) 0.106 <0.001 1,007 (25.6) 973 (24.7) 0.02 0.38 
β-Blocker 2,226 (12) 18,573 (14.7) 0.078 <0.001 574 (14.6) 537 (13.6) 0.027 0.23 
CCB 4,143 (22.4) 31,643 (25) 0.061 <0.001 974 (24.7) 936 (23.8) 0.023 0.32 
Diuretics 3,376 (18.3) 16,603 (13.1) 0.142 <0.001 513 (13) 545 (13.8) 0.024 0.29 
Statin 2,074 (11.2) 21,806 (17.2) 0.173 <0.001 633 (16.1) 610 (15.5) 0.016 0.48 
Fibrate 1,305 (7.1) 9,948 (7.9) 0.031 0.001 332 (8.4) 300 (7.6) 0.03 0.18 
Aspirin 2,219 (12) 18,640 (14.7) 0.08 <0.001 566 (14.4) 543 (13.8) 0.017 0.46 
Enrolled research year        
 2000–2004 5,102 (27.6) 33,703 (26.6) 0.022 0.006 1,107 (28.1) 1,024 (26) 0.047 0.035 
 2005–2008 7,996 (43.2) 55,661 (44) 0.015 0.06 1,701 (43.2) 1,720 (43.7) 0.01 0.67 
 2009–2010 5,398 (29.2) 37,220 (29.4) 0.005 0.54 1,127 (28.6) 1,193 (30.3) 0.037 0.1 
Propensity score 0.42331 (0.3048) 0.08427 (0.1274) 1.451 <0.001 0.09098 (0.1193) 0.09098 (0.1193) >0.99 <0.001 

ACEi, ACE inhibitor; CCB, calcium channel blocker; COPD, chronic obstructive pulmonary disease; SID, standardized difference; TZD, thiazolidinedione.

Crude and multivariable-adjusted Cox proportional hazards models with robust sandwich SE estimates were used to compare the outcomes between participants with and without acute liver injury. The results were presented as hazard ratios (HR) with a 95% CI. As death may interfere with hypoglycemic risk calculation, we used the competing risk analyses for adjustment (Table 2) (14). We censored participants at the time of mortality or the end of the study, whichever came first, to assess the incidence of mortality. We censored participants on the day of hypoglycemia or the end of follow-up on 31 December 2010 to evaluate the risk of hypoglycemia.

Table 2

The risk for patients with diabetes with or without acute liver injury in multivariate Cox regression analysis

With acute liver injuryWithout acute liver injuryCrude modelAdjusted modelaAfter propensity-matched adjusted model
nIRnIRHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P value
Hypoglycemia 341 1,751.4 983 708.41 2.47 (2.18–2.79) <0.001 2.14 (1.83–2.49) <0.001 1.76 (1.16–2.66) 0.008 
 Competing riskb       2.07 (1.78–2.42) <0.001 1.73 (1.16–2.58) 0.008 
 Competing riskc       2.14 (1.83–2.49) <0.001 1.76 (1.16–2.66) 0.008 
Hypoglycemia in 90 days 54 12.28 170 5.59 2.19 (1.61–2.98) <0.001 1.92 (1.30–2.85) <0.001 2.08 (0.71–6.06) 0.18 
 Competing riskb       1.90 (1.29–2.82) 0.001 2.07 (0.70–6.16) 0.19 
 Competing riskc       1.92 (1.30–2.85) 0.001 2.08 (0.71–6.06) 0.18 
Hypoglycemia in 365 days 122 7.35 336 2.9 2.52 (2.05–3.11) <0.001 1.98 (1.52–2.58) <0.001 1.99 (1.02–3.90) 0.045 
 Competing riskb       1.95 (1.49–2.57) <0.001 1.97 (1.01–3.82) 0.046 
 Competing riskc       1.98 (1.52–2.58) <0.001 1.99 (1.02–3.90) 0.045 
All-cause mortality 1,332 18.49 1,961 3.85 4.75 (4.43–5.10) <0.001 2.44 (2.22–2.69) <0.001 2.48 (1.84–3.35) <0.001 
All-cause mortality in 90 days 363 82.4 838 27.54 2.97 (2.63–3.36) <0.001 1.73 (1.46–2.05) <0.001 1.59 (0.97–2.63) 0.068 
All-cause mortality in 365 days 613 36.8 1,078 9.3 3.92 (3.55–4.33) <0.001 1.94 (1.69–2.24) <0.001 1.80 (1.20–2.71) 0.005 
With acute liver injuryWithout acute liver injuryCrude modelAdjusted modelaAfter propensity-matched adjusted model
nIRnIRHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P value
Hypoglycemia 341 1,751.4 983 708.41 2.47 (2.18–2.79) <0.001 2.14 (1.83–2.49) <0.001 1.76 (1.16–2.66) 0.008 
 Competing riskb       2.07 (1.78–2.42) <0.001 1.73 (1.16–2.58) 0.008 
 Competing riskc       2.14 (1.83–2.49) <0.001 1.76 (1.16–2.66) 0.008 
Hypoglycemia in 90 days 54 12.28 170 5.59 2.19 (1.61–2.98) <0.001 1.92 (1.30–2.85) <0.001 2.08 (0.71–6.06) 0.18 
 Competing riskb       1.90 (1.29–2.82) 0.001 2.07 (0.70–6.16) 0.19 
 Competing riskc       1.92 (1.30–2.85) 0.001 2.08 (0.71–6.06) 0.18 
Hypoglycemia in 365 days 122 7.35 336 2.9 2.52 (2.05–3.11) <0.001 1.98 (1.52–2.58) <0.001 1.99 (1.02–3.90) 0.045 
 Competing riskb       1.95 (1.49–2.57) <0.001 1.97 (1.01–3.82) 0.046 
 Competing riskc       1.98 (1.52–2.58) <0.001 1.99 (1.02–3.90) 0.045 
All-cause mortality 1,332 18.49 1,961 3.85 4.75 (4.43–5.10) <0.001 2.44 (2.22–2.69) <0.001 2.48 (1.84–3.35) <0.001 
All-cause mortality in 90 days 363 82.4 838 27.54 2.97 (2.63–3.36) <0.001 1.73 (1.46–2.05) <0.001 1.59 (0.97–2.63) 0.068 
All-cause mortality in 365 days 613 36.8 1,078 9.3 3.92 (3.55–4.33) <0.001 1.94 (1.69–2.24) <0.001 1.80 (1.20–2.71) 0.005 

IR, incidence rate per 1,000 person-years; n, number of cases.

a

Adjusted model adjusted age, sex, smoking, alcohol, overweight, obesity, severe obesity, HBV infection and therapy, HCV infection and therapy, CKD, chronic obstructive pulmonary disease, CCI score, DCSI, diabetic duration, metformin, SU, thiazolidinedione, DPP-4i, insulin basal, insulin premixed, insulin basal and bolus, antidiabetic number, ACE inhibitor/ARB, β-blocker, calcium channel blocker, diuretics, statin, fibrate, aspirin, and enrolled research year.

b

Taking death as a competing risk for subdistribution competing risk analysis of hypoglycemia.

c

Taking death as a competing risk for cause-special competing risk analysis of hypoglycemia.

We calculated the cumulative incidences of severe hypoglycemia and all-cause mortality within 90 and 365 days after discharge between participants with and without acute liver injury using the Kaplan-Meier method; we used a log-rank test to determine the statistical differences between the study and control groups. We performed subgroup analyses for the risk of severe hypoglycemia within 90 and 365 days after discharge and a likelihood ratio test to identify significant interactions between acute liver injury and age, sex, CKD, SU use, glibenclamide, glipizide, gliclazide, glimepiride, number of oral hypoglycemic agents, insulin, β-blocker, and fibrate. We did a sensitivity test by excluding patients with nonspecific diagnosis of acute liver injury, such as other specified liver disorders, other sequelae of chronic liver disease, elevated ALT, and abnormal liver function test, to further validate robustness of our study results. A two-sided P value of <0.05 was considered a significant difference. SAS version 9.4 and Stata SE version 15.1 were used for analyses.

From 1 January 2000 to 31 December 2010, 18,498 patients were diagnosed with T2D and hospitalized for acute liver injury, while 126,584 patients were diagnosed with T2D and hospitalized for other causes (Supplementary Fig. 1). Before propensity score matching, few differences were noted between the study and control groups (Table 1). After matching, 3,937 pairs of patients with T2D with hospitalization were selected. The matched pairs were similar for all variables. The mean age of the matched cohort was 55.6 years; the mean duration of diabetes was 2.6 years; the prevalence of alcohol-related disorders, drug-induced liver diseases, HBV, and HCV infections were 2.4%, 3.2%, 4.5%, and 1.9%, respectively. The mean follow-up time (SD) was 4.0 (2.8) and 3.9 (2.8) years in patients discharged with and without acute liver injury, respectively.

In this cohort study, 54 (0.29%) patients discharged with acute liver injury and 170 (0.13%) patients discharged without acute liver injury had severe hypoglycemia within 90 days after discharge (incidence rate 12.28 vs. 5.59/1,000 person-years). The adjusted multivariable models showed that patients discharged with acute liver injury had a higher risk of severe hypoglycemia; the prematched adjusted HR (aHR) was 1.92 (1.30–2.85), and the postmatched aHR was 2.08 (0.71–6.06) (Table 2). The subdistribution competing risk analysis showed the prematched aHR of hypoglycemia as 1.90 (1.29–2.82), taking death as a competing risk; for cause-special competing risk analysis, the aHR of hypoglycemia was 1.92 (1.30–2.85). Patients discharged with acute liver injury also had a higher risk of severe hypoglycemia within 365 days after discharge (aHR 1.98 [1.52–2.58]) and over the follow-up period (aHR 2.14 [1.83–2.49]) than those discharged without acute liver injury (Table 2).

In this cohort, 363 (1.96%) patients discharged with acute liver injury and 838 (0.66%) patients discharged without acute liver injury died within 90 days after discharge (incidence rate 82.4 vs. 27.54/1,000 person-years). The adjusted multivariable models show that patients discharged with acute liver injury had a higher mortality risk; the prematched aHR was 1.73 (1.46–2.05), and the postmatched aHR was 1.59 (0.97–2.63) (Table 2). Patients discharged with acute liver injury also had a higher mortality risk within 365 days after discharge (prematched aHR 1.94 [1.69–2.24]; postmatched aHR 1.80 [1.20–2.71]) and over the follow-up period (prematched aHR 2.44 [2.22–2.69]; postmatched aHR 2.48 [1.84–3.35]) than those discharged without acute liver injury (Table 2). After adding severe hypoglycemia (+ vs. −) as a covariate to minimize the possible confounding effect caused by hypoglycemia, the multivariable-adjusted analyses showed the prematched aHRs (95% CI; P value) for mortality within 90 days and 365 days after discharge as 1.71 (1.40–2.09; P < 0.001) and 1.85 (1.58–2.18; P < 0.001), respectively, in patients discharged with versus without acute liver injury. Additionally, we have assessed the mortality risk in patients with acute liver injury, stratified by hypoglycemia, to see whether hypoglycemia might be a risk factor for the mortality risk for these patients (Supplementary Table 1). Through these evaluations, our study demonstrated that hypoglycemic events were not a risk factor for the high mortality risk in patients with acute liver injury.

In the sensitivity tests, we excluded the patients with the four nonspecific diagnoses (including other specified liver disorders, other sequelae of chronic liver disease, elevated ALT, and abnormal liver function test) of acute liver injury. The adjusted multivariable models showed that patients discharged with acute liver injury had a higher risk of severe hypoglycemia within 90 days after discharge (aHR 2.34 [1.40–3.91]), within 365 days after discharge (aHR 2.12 [1.50–2.98]), and over the follow-up period (aHR 2.29 [1.87–2.82]) than those discharged without acute liver injury; patients discharged with acute liver injury had a higher risk of mortality within 90 days after discharge (aHR 2.28 [1.83–2.83]), within 365 days after discharge (aHR 2.29 [1.92–2.72]), and over the follow-up period (aHR 2.65 [2.36–2.98]) than those discharged without acute liver injury (Supplementary Table 2). The sensitivity tests showed similar results.

The cumulative incidences of severe hypoglycemia within 90 days and 365 days after discharge in matched patients with acute liver injury showed a significantly higher risk than those discharged without acute liver injury (Fig. 1). The cumulative incidences of mortality within 90 days and 365 days after discharge in matched patients with acute liver injury also showed a significantly higher risk than those without acute liver injury (Supplementary Fig. 2).

Figure 1

The cumulative incidence of severe hypoglycemia between patients with and without acute liver injury within 90 days (A) and within 365 days (B) after discharge and after propensity score matching.

Figure 1

The cumulative incidence of severe hypoglycemia between patients with and without acute liver injury within 90 days (A) and within 365 days (B) after discharge and after propensity score matching.

Close modal

The forest plot for subgroup analysis of hypoglycemia risks in patients discharged with acute liver injury versus those without acute liver injury showed no significant interaction in the risk factors of age, CKD, SU, glibenclamide, glipizide, gliclazide, glimepiride, number of oral hypoglycemic agents, insulin, β-blocker, and fibrate, except that sex difference had significant interactions in hypoglycemia risk within 90 days after discharge. Sex difference and CKD versus no CKD had significant interaction in hypoglycemia risk within 365 days after discharge (Fig. 2).

Figure 2

Subgroup analysis of basic demographics, comorbidity, and medications on hypoglycemia risks in patients with T2D with versus without acute liver injury within 90 days (A) and within 365 days (B) after discharge and before propensity score matching. OAD, oral antidiabetic drug.

Figure 2

Subgroup analysis of basic demographics, comorbidity, and medications on hypoglycemia risks in patients with T2D with versus without acute liver injury within 90 days (A) and within 365 days (B) after discharge and before propensity score matching. OAD, oral antidiabetic drug.

Close modal

This study demonstrated that patients with diabetes hospitalized for acute liver injury showed a higher risk of hypoglycemia within 90 days and within 365 days of discharge and over the follow-up period than patients hospitalized for other causes. Patients hospitalized for acute liver injury also had a higher risk of death within 90 days and 365 days of discharge and over the follow-up period than patients hospitalized for other causes. Moreover, patients discharged with acute liver injury had a higher risk of hypoglycemia, regardless of age, CKD, insulin, SU and different types of SU, number of oral hypoglycemic agents, β-blocker, and fibrate use.

Hypoglycemia can affect the mood and daily life of the patient. Severe hypoglycemia can further alter the patient’s mental state and increase accident risk, even mortality (12,15). Approximately 7.1% of patients with type 1 diabetes on insulin injections have incidence of severe hypoglycemia (16). About 0.1 to 0.7% (11) (or 0.5 events/1,000 person-years) (15) of patients with T2D, ˜0.8 (9 episodes/1,000 person-years) (16) to 7% (100 episodes/1,000 patient-years) of patients with T2D on SU (12,15), and 1.2 to 7.3% (or 118 episodes/1,000 person-years) (15) of patients with T2D on insulin therapy have severe hypoglycemia incidence (12,16). It has been known for a long time that chronic alcohol drinkers are prone to hypoglycemia (1). A retrospective study of 87 patients with cirrhosis showed that 11 patients (12.6%) had low glucose levels (<70 mg/dL) or substantiated hypoglycemia (17). Felig et al. (18) reported a case series of 15 patients with viral hepatitis. Among them, eight patients (53.3%) had glucose levels <60 mg/dL but without classical features of hypoglycemia. O’Grady and Williams (19) treated 620 patients with acute liver failure and hepatic encephalopathy and found that hypoglycemia was “almost invariable” in these patients. Fischer et al. (20) noted hypoglycemia (blood glucose ≤49 mg/dL) in 18 patients hospitalized for different liver diseases. Hospitalized patients with acute liver failure had a higher risk of hypoglycemia, which could be prevented by glucose infusion (21).

To the best of our knowledge, our study is the first to investigate whether the risk of hypoglycemia is higher following patient discharge from the hospital due to acute liver injury. This study showed that the incidence of severe hypoglycemia was 12.28 (within 90 days after discharge) and 7.35 (within 365 days after discharge) per 1,000 person-years in patients with T2D hospitalized for acute liver injury. It showed that patients with T2D discharged for acute liver injury had a 1.92 (aHR 1.92 [1.30–2.85] within 90 days after discharge) or 1.98 times (aHR 1.98 [1.52–2.58] within 365 days after discharge) higher risk of hypoglycemia than those discharged for other causes. This finding implies that patients discharged from the hospital with acute liver injury did not show full hepatic recovery within 90 days and 365 days of discharge, or some patients possibly had preexisting chronic liver diseases, which hindered normal glucose metabolism. The finding of no statistical difference in the postmatched aHR for hypoglycemia risk within 90 days after discharge could be due to few event rates with wide CIs. Studies have shown that age, renal impairment, SU, insulin, β-blocker, or fibrate use are associated with a higher risk of severe hypoglycemia (12,15,22). The subgroup analysis of hypoglycemia risk in our patients discharged for acute liver injury revealed no significant interaction in these risk factors, except that sex differences had significant interaction, although the cause is unclear. Patients with T2D (specifically male patients) discharged from the hospital with acute liver injury should be advised to monitor their blood glucose closely to detect hypoglycemia immediately.

There are several possible explanations for the higher risk of severe hypoglycemia after discharge with acute liver injury. First, the liver is a vital organ for maintaining glucose homeostasis. Although a patient is discharged from the hospital with acute liver injury, the liver may not have fully recovered from the damage or may have underlying diseases that prevent effective gluconeogenesis and glycogenolysis (1,3). Second, patients discharged with acute liver injury may be inadequately nourished to provide the necessary substrates (lactate, amino acids, and glycerol) for gluconeogenesis during hypoglycemia (22,23). Third, patients discharged with acute liver injury may have insufficient glycogen storage for glycogenolysis during hypoglycemia (3,18). Fourth, the damaged liver may be unable to metabolize and extract insulin and insulin secretagogues, leading to increased blood levels and hypoglycemia in patients with T2D (24). Further research is needed to determine whether gluconeogenesis or glycogenolysis causes hypoglycemia in patients with acute liver injury. Moreover, we need to examine the mechanisms responsible for abnormal gluconeogenesis and glycogenolysis in these patients.

People with chronic liver diseases have about four times higher mortality rates than the general population (25). Patients with acute liver injury or acute liver failure also have a higher mortality rate (4,26), and those with more organ failure have a higher risk of death (25). Our study revealed that the risks of all-cause mortality within 90 days and within 365 days of discharge and over the follow-up period were higher in patients discharged with acute liver injury than those discharged with other causes. Our study also showed that hypoglycemia had no significant influence on mortality risk in patients with acute liver injury. This finding may indicate that although patients have some remission from hospitalization for acute liver injury, liver damage and physical health may not show full recovery within 90 days and 365 days after discharge or that some patients may have underlying chronic liver diseases with a higher risk of mortality than those discharged with other causes.

This study has some limitations. First, this administrative data set lacked complete information on family history, dietary patterns, alcohol drinking, and physical activity, which may influence the measured outcomes. Second, we could not obtain data on liver function, bilirubin, prothrombin time, renal function, ultrasound, and pathology. Therefore, we could not determine the severity of liver damage or whether the liver function was restored at hospital discharge. We used the clinical diagnosis to separate patients with acute liver injury into acute liver injury, severe acute liver injury with jaundice, and acute liver failure. We also used the clinical diagnosis to discriminate the possible causes of acute liver injury, such as medications, hepatitis viral infection, or alcohol. Third, the database lacked information on glucose and hemoglobin A1c levels; therefore, we could not ascertain the treatment status and severity of patients with diabetes. However, we matched sex, age, DCSI scores, comorbidities, CCI, and medications to maximally balance the condition between the study and comparison groups and increase their comparability. Fourth, we defined severe hypoglycemia as patients sent to the emergency department or hospitalized for hypoglycemia (corresponding to the American Diabetes Association’s definition for severe hypoglycemia) (27). This definition may not capture all severe hypoglycemic episodes and underestimate the incidence of severe hypoglycemia. However, because this definition is applied to the study and comparison groups, it may have an undifferentiated impact on the outcomes. Finally, a retrospective cohort study is affected by unobserved or unknown confounding factors, and prospective studies are needed to confirm our results.

In conclusion, patients with diabetes hospitalized for acute liver injury have a higher risk of severe hypoglycemia and death at 3 months and 1 year after discharge than those hospitalized for other causes. In clinical care, when a patient has hypoglycemia, the doctor needs to identify the cause of hypoglycemia, including acute liver injury. In patients with acute liver injury, regular blood glucose testing is required for early detection and management of hypoglycemia.

This article contains supplementary material online at https://doi.org/10.2337/figshare.21346746.

Acknowledgments. This study used data from the NHIRD, provided by the NHI Administration of Taiwan. The interpretation and conclusion of data reported in this article do not represent the position of the NHI administration.

Funding. This work was supported by grants from the Taipei Veterans General Hospital (V105C-204 and V110C-175) and the Ministry of Science and Technology, Republic of China, Taiwan (MOST 110-2314-B-075-027-MY3).

These funding agencies had no role in the study design, data collection and analysis, decision to publish, or manuscript preparation. No organization provided funds to assist with manuscript preparation, and data analysis was not performed by employees of funders or any author who received funding. The funders did not offer writing support.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.-S.Y. conceived and designed this study, participated in interpretation of data, and wrote and edited the manuscript. M.-C.H. conceived and designed this study and critically revised the manuscript for important intellectual content. C.-W.P. participated in data interpretation and manuscript preparation. J.-S.L. contributed to the data acquisition, statistical analysis, and interpretation of data and critically revised the manuscript for important intellectual content. C.-C.H. contributed to the data acquisition, administrative, technical, or material support and critically revised the manuscript for important intellectual content. C.-M.H. conceived and designed this study, contributed to the data acquisition, obtained funding, and critically revised the manuscript for important intellectual content. C.-C.H. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Arky
RA
.
Hypoglycemia associated with liver disease and ethanol
.
Endocrinol Metab Clin North Am
1989
;
18
:
75
90
2.
Nordlie
RC
,
Foster
JD
,
Lange
AJ
.
Regulation of glucose production by the liver
.
Annu Rev Nutr
1999
;
19
:
379
406
3.
Hepatic hypoglycaemia
.
BMJ
1971
;
2
:
416
417
4.
Koch
DG
,
Speiser
JL
,
Durkalski
V
, et al
.
The natural history of severe acute liver injury
.
Am J Gastroenterol
2017
;
112
:
1389
1396
5.
Gangopadhyay
KK
,
Singh
P
.
Consensus statement on dose modifications of antidiabetic agents in patients with hepatic impairment
.
Indian J Endocrinol Metab
2017
;
21
:
341
354
6.
Cheng
TM
.
Taiwan’s new national health insurance program: genesis and experience so far
.
Health Aff (Millwood)
2003
;
22
:
61
76
7.
Lin
CC
,
Lai
MS
,
Syu
CY
,
Chang
SC
,
Tseng
FY
.
Accuracy of diabetes diagnosis in health insurance claims data in Taiwan
.
J Formos Med Assoc
2005
;
104
:
157
163
8.
Chan
KA
,
Truman
A
,
Gurwitz
JH
, et al
.
A cohort study of the incidence of serious acute liver injury in diabetic patients treated with hypoglycemic agents
.
Arch Intern Med
2003
;
163
:
728
734
9.
Huerta
C
,
Zhao
SZ
,
García Rodríguez
LA
.
Risk of acute liver injury in patients with diabetes
.
Pharmacotherapy
2002
;
22
:
1091
1096
10.
Meduru
P
,
Helmer
D
,
Rajan
M
,
Tseng
CL
,
Pogach
L
,
Sambamoorthi
U
.
Chronic illness with complexity: implications for performance measurement of optimal glycemic control
.
J Gen Intern Med
2007
;
22
(
Suppl. 3
):
408
418
11.
Young
BA
,
Lin
E
,
Von Korff
M
, et al
.
Diabetes complications severity index and risk of mortality, hospitalization, and healthcare utilization
.
Am J Manag Care
2008
;
14
:
15
23
12.
Amiel
SA
,
Dixon
T
,
Mann
R
,
Jameson
K
.
Hypoglycaemia in type 2 diabetes
.
Diabet Med
2008
;
25
:
245
254
13.
D’Agostino
RB
Jr
.
Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group
.
Stat Med
1998
;
17
:
2265
2281
14.
Lau
B
,
Cole
SR
,
Gange
SJ
.
Competing risk regression models for epidemiologic data
.
Am J Epidemiol
2009
;
170
:
244
256
15.
Inkster
B
,
Zammitt
NN
,
Frier
BM
.
Drug-induced hypoglycaemia in type 2 diabetes
.
Expert Opin Drug Saf
2012
;
11
:
597
614
16.
Leese
GP
,
Wang
J
,
Broomhall
J
, et al.;
DARTS/MEMO Collaboration
.
Frequency of severe hypoglycemia requiring emergency treatment in type 1 and type 2 diabetes: a population-based study of health service resource use
.
Diabetes Care
2003
;
26
:
1176
1180
17.
Gundling
F
,
Seidl
H
,
Strassen
I
, et al
.
Clinical manifestations and treatment options in patients with cirrhosis and diabetes mellitus
.
Digestion
2013
;
87
:
75
84
18.
Felig
P
,
Brown
WV
,
Levine
RA
,
Klatskin
G
.
Glucose homeostasis in viral hepatitis
.
N Engl J Med
1970
;
283
:
1436
1440
19.
O’Grady
JG
,
Williams
R
.
Management of acute liver failure
.
Schweiz Med Wochenschr
1986
;
116
:
541
544
20.
Fischer
KF
,
Lees
JA
,
Newman
JH
.
Hypoglycemia in hospitalized patients. causes and outcomes
.
N Engl J Med
1986
;
315
:
1245
1250
21.
Bernal
W
,
Wendon
J
.
Acute liver failure
.
N Engl J Med
2013
;
369
:
2525
2534
22.
Cryer
P
.
Hypoglycemia in Diabetes: Pathophysiology, Prevalence and Prevention
.
Alexandria, VA
,
American Diabetes Association
,
2012
, pp.
37
125
23.
Frizzell
RT
,
Campbell
PJ
,
Cherrington
AD
.
Gluconeogenesis and hypoglycemia
.
Diabetes Metab Rev
1988
;
4
:
51
70
24.
Scheen
AJ
.
Pharmacokinetic and toxicological considerations for the treatment of diabetes in patients with liver disease
.
Expert Opin Drug Metab Toxicol
2014
;
10
:
839
857
25.
Fleming
KM
,
Aithal
GP
,
Card
TR
,
West
J
.
All-cause mortality in people with cirrhosis compared with the general population: a population-based cohort study
.
Liver Int
2012
;
32
:
79
84
26.
Arroyo
V
,
Moreau
R
,
Jalan
R
.
Acute-on-chronic liver failure
.
N Engl J Med
2020
;
382
:
2137
2145
27.
International Hypoglycaemia Study Group
.
Glucose concentrations of less than 3.0 mmol/L (54 mg/dL) should be reported in clinical trials: a joint position statement of the American Diabetes Association and the European Association for the Study of Diabetes
.
Diabetes Care
2017
;
40
:
155
157
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.