Table 1. Cohort Characteristicsa.
| Characteristic | Patients, No. (%) (N = 16 178) |
|---|---|
| Demographic and comorbidity | |
| Age, median (IQR), y | 73 (67-80) |
| Female | 283 (2) |
| Race/ethnicity | |
| White | 13 138 (81) |
| Black | 2336 (14) |
| Hispanic | 225 (1) |
| Other | 479 (3) |
| Income, mean (SD), $ | 26 653 (46 601) |
| Selected comorbidities | |
| Congestive heart failure | 6722 (42) |
| Kidney disease | 5061 (31) |
| Cerebrovascular accident | 4217 (26) |
| Prior myocardial infarction | 3024 (20) |
| Malignant neoplasm | 3574 (22) |
| Dementia | 519 (3) |
| Estimated life expectancy, yb | |
| >10 | 7065 (44) |
| 5-10 | 4652 (29) |
| <5 | 4461 (28) |
| Hospitalization | |
| Year of hospitalization | |
| 2011 | 6370 (39) |
| 2012 | 5101 (32) |
| 2013 | 4707 (29) |
| Training hospital | 14 551 (90) |
| Length of stay, median (IQR), d | 4 (2-6) |
| Discharge diagnosis | |
| Arrhythmia | 217 (1) |
| Asthma | 84 (1) |
| Chronic obstructive pulmonary disease | 1531 (10) |
| Chest pain | 664 (4) |
| Conduction disorders | 1617 (10) |
| Coronary artery disease | 1795 (11) |
| Acute coronary syndrome | 971 (6) |
| Congestive heart failure | 2553 (16) |
| Heart valve disorder | 312 (2) |
| Pneumonia | 1893 (12) |
| Sepsis | 285 (2) |
| Skin infection | 1377 (9) |
| Stroke | 791 (5) |
| Transient ischemic attack | 263 (2) |
| Urinary tract infection | 1429 (9) |
| Venous thromboembolism | 396 (3) |
| Estimated glomerular filtration rate, mean (SD), mL/min/1.73 m2 | |
| Preadmission | 66 (24) |
| Discharge | 68 (27) |
| Medication use | |
| No. of admission medications, median (IQR) | 9 (6-12) |
| No. of admission diabetes medications | |
| 0 | 4153 (26) |
| 1 | 8263 (51) |
| 2 | 3350 (21) |
| ≥3 | 412 (3) |
| Admission diabetes medications | |
| Sulfonylureas | 7830 (48) |
| Biguanides | 7328 (45) |
| Thiazolidinediones | 535 (3) |
| α-Glucosidase inhibitors | 265 (2) |
| Dipeptidyl peptidase-4 inhibitors | 145 (1) |
| Meglitinides | 28 (0) |
| Glucagon-like peptide agonists | 5 (0) |
| Prior diabetes medication adherence, proportion of days covered, %c | |
| <80 | 3700 (23) |
| ≥80 | 8325 (52) |
| Not taking diabetes medications at admission | 4153 (26) |
| Any receipt of inpatient corticosteroids | 1810 (11) |
Abbreviation: IQR, interquartile range.
Multiple imputation was used to account for missing data, which included inpatient estimated glomerular filtration rate (1300 patients), outpatient estimated glomerular filtration rate (1356 patients), and inpatient glucose recordings (673 patients).
Life expectancy was calculated based on number of comorbidities and age.
Adherence was calculated from electronic pharmacy dispensing data as the proportion of days covered for each admission diabetes medication in the year prior to index hospital admission. A threshold of 80% is a commonly used criterion to determine clinically significant nonadherence. To account for patients taking multiple diabetes mediations prior to hospitalization, a composite proportion of days covered was calculated as the mean of each individual diabetes medication proportion of days covered.