Sulfonylurea Use and the Risk of Hospital Readmission in Patients With Type 2 Diabetes

Pamela C. Heaton; Vibha C. A. Desai; Christina M. L. Kelton; Swapnil N. Rajpathak

BMC Endocr Disord. 2016;16(4)

Abstract

Background: Hospital inpatient care for patients with diabetes was estimated to cost $76 billion in 2012. Substantial expense resulted from those patients having multiple hospitalizations. The objective was to compare the risk for diabetes–related hospital readmission in patients with type 2 diabetes treated with sulfonylureas (SUs) compared to those treated with other oral antihyperglycemic agents (AHAs).

Methods: A retrospective cohort analysis was conducted using two-year panels, from 1999 to 2010, from the Medical Expenditure Panel Survey. The study included patients with type 2 diabetes taking an oral AHA who experienced a diabetes-related hospitalization. A Cox proportional hazard regression predicting time to readmission was used to estimate and compare the risks of readmission for SU-monotherapy versus other-AHA-monotherapy patients. Covariates included age, gender, marital status, cardiovascular disease, kidney disease, and eye disease, along with a propensity score to control for selection bias. The lack of clinical data on disease severity and progression limited our ability to estimate causal relationships between drug use and risk of hospital readmission.

Results: From 1999 to 2010, an estimated 13.5 million patients experienced a diabetes-related hospital admission and subsequent AHA treatment. While 23.2 % (n = 746,579) of patients in the SU monotherapy cohort had a readmission, only 16.1 % (n = 881,984) in the other-AHA monotherapy group were readmitted. Average readmission expenditure for readmitted SU users (in 2010 dollars) was $11,148 (±$1,558) compared to $7,673 (±$763) for users of other oral AHAs. The estimated readmission hazard ratio was 1.29 (95 % CI: 1.01–1.65; p-value = 0.04) for SU monotherapy users. If a patient's first hospital admission was during the time period 2008–2010, a readmission was significantly less likely (HR 0.49, 95 % CI: 0.31–0.78; p = 0.003) relative to 2004–2007.

Conclusions: Among patients with type 2 diabetes, SU use was associated with an approximately 30 % increased risk for readmission compared to other-AHA use, while each readmission for an SU user cost on average 45 % more than one for an other-AHA patient. Because of the rapidly rising prevalence of diabetes in the U.S. and the large number of patients with prediabetes, preventing hospital readmissions will continue to be an important cost-saving strategy in the future.

1 2 3 4 5 6

Next Section

Table 1. Hospital Readmission Rates and Average Readmission Costs for Medication-Based Patient Cohorts: 1999–2010. The readmission percentage is found by dividing the number readmitted to the hospital by the total number of patients who experienced an initial diabetes-related hospitalization. Readmission cost is expressed in 2010 U.S. dollars
Patient Cohort	Total number (SE)	Number not readmitted (SE)	Number readmitted (SE)	Readmission percentage	Mean readmission cost (SE)
All Patients	13,537,803 (519,634)	10,959,266 (451,218)	2,578,538 (198,384)	19.1	$8,814 ($580)
All SU Patients^a	7,871,912 (368,351)	6,204,869 (326,273)	1,667,043 (153,735)	21.2	$9,204 ($769)
SU (SU monotherapy)^b	3,217,089 (235,951)	2,470,510 (194,884)	746,579 (106,554)	23.2	$11,148 ($1,558)
SU+ (other AHA)	4,654,823 (274,495)	3,734,359 (247,494)	920,464 (113,487)	19.8	$7,624 ($412)
All Other Oral AHA Patients^a	5,665,891 (291,467)	4,754,397 (256,821)	911,495 (108,690)	16.1	$8,098 ($737)
noSU (monotherapy with non-SU AHA)^b	5,488,379 (288,154)	4,606,396 (254,357)	881,984 (107,289)	16.1	$7,673 ($763)
noSU+ (> 1 non-SU AHA)	177,512 (na)	148,001 (na)	29,511 (na)	16.6	$20,772 (na)

SE standard error
na not available due to too low raw cell count
SU sulfonylurea
AHA antihyperglycemic agent
^aIn a comparison between all SU (SU and SU+) patients and all other oral AHA patients, the relevant p-value for readmission rates was 0.017
^bIn a comparison between SU patients and noSU monotherapy patients, the relevant p-value for readmission rates was 0.003

Table 2. Baseline characteristics for patients receiving antidiabetic monotherapy. Characteristics include demographic characteristics; insurance coverage; period of first hospital admission; comorbidities; medical care received; and disease severity. In order to achieve balance between the cohorts in characteristic distributions, propensity scores were estimated and used as a covariate in the Cox proportional hazard regression following a statistical check to see whether balance was achieved
Characteristic	All Patients (Percent)	SU Cohort (Percent)	noSU Cohort (Percent)	p-value^a	Adjusted p-value^b
Total	8,705,468	3,217,089	5,488,379
Mean Age (95 % CI)	63.4 (62.5–64.4)	68.3 (67.0–69.7)	60.5 (59.3–61.7)	<0.0001	0.951
Male	3,738,700 (42.9)	1,514,251 (47.1)	2,224,449 (40.5)	0.093	0.985
Race
White	6,859,909 (78.8)	2,525,415 (78.5)	4,346,796 (79.2)
African American	1,305,820 (15.0)	466,478 (14.5)	856,187 (15.6)	0.784	0.912
Other	539,739 (6.2)	225,196 (7.0)	285,396 (5.2)
Region
Northeast	1,647,688 (18.9)	465,284 (14.5)	1,182,404 (21.5)
Midwest	2,007,585 (23.1)	777,258 (24.2)	1,230,327 (22.4)	0.115	0.994
South	3,290,396 (37.8)	1,278,818 (39.8)	2,011,578 (36.7)
West	1,534,591 (17.6)	526,591 (16.4)	1,008,000 (18.4)
Married	4,530,506 (52.0)	1,578,073 (49.1)	2,952,432 (53.8)	0.193	0.992
Insurance status
Public	3,410,507 (39.2)	1,455,344 (45.2)	1,955,164 (35.6)
Private	4,442,060 (51.0)	1,457,269 (45.3)	2,984,792 (54.4)	0.0881	0.998
Other	223,194 (2.6)	136,278 (4.2)	86,916 (1.6)
Uninsured	629,707 (7.2)	168,199 (5.2)	461,508 (8.4)
Period of first hospital admission
1999–2003	1,796,930 (20.6)	1,012,088 (31.5)	784,842 (14.3)
2004–2007	3,932,720 (45.2)	1,475,349 (45.9)	2,457,371 (44.8)	<0.0001	0.937
2008–2010	2,975,819 (34.2)	729,652 (22.7)	2,246,166 (40.9)
Cardiovascular disease	5,272,543 (60.6)	1,695,922 (52.7)	3,576,621 (65.2)	0.0017	0.966
Renal disease	823,544 (9.5)	354,703 (11.0)	468,840 (8.52)	0.298	0.979
Eye disease	1,454,104 (16.7)	547,914 (17.7)	886,190 (16.2)	0.624	0.995
Mean number of HbA1c tests (95 % CI)	3.5 (2.4–4.7)	3.1 (1.4–4.92)	3.8 (2.3–5.2)	0.602	0.443
Perceived health status
Excellent/Very good	1,782,916 (20.5)	616,432 (19.2)	1,166,485 (21.3)
Good	2,674,290 (30.7)	792,801 (24.6)	1,881,489 (34.3)	0.012	0.973
Fair/Poor	4,161,730 (47.8)	1,767,732 (55.0)	2,393,999 (43.6)
Perceived mental health status
Excellent/Very good	3,830,005 (44.0)	1,219,758 (37.9)	2,610,246(48.2)
Good	2,678,886 (30.8)	998,646 (31.0)	1,680,241 (30.6)	0.017	0.919
Fair/Poor	2,110,045 (24.2)	958,560 (29.8)	1,150,485 (21.0)
Physical limitations	4,119,188 (47.3)	1,634,470 (50.8)	2,484,718 (45.3)	0.153	0.991
Difficulty in getting care	4,541,484 (52.2)	1,554,111 (48.3)	2,987,374 (54.4)	0.107	0.989

CI confidence interval
^ap-value for the test of equality (t- or chi-squared) of percentage distributions between cohorts
^bp-value for the test of treatment effect on patient characteristic, controlling for patient propensity score

Table 3. Estimated hazard ratios from a Cox proportional hazard regression with time to hospital readmission as the dependent variable. After propensity-score adjustment, and controlling for demographic and other patient characteristics, patients on a sulfonylurea were 29 % more likely to be readmitted to the hospital than patients on another oral antihyperglycemic agent. Unmarried patients and patients with eye disease were also more likely to be readmitted
Independent variable^a	Hazard ratio	95 % confidence interval	p-value
Sulfonylurea	1.29	1.01–1.65	0.042
Age	1.01	0.98–1.03	0.749
Male	0.96	0.61–1.28	0.838
Race
African American	0.89	0.66–1.28	0.512
Other	0.96	0.61–1.52	0.867
White	1.00
Region
Northeast	1.00	0.65–1.44	0.864
Midwest	0.86	0.62–1.18	0.349
West	1.28	0.94–1.74	0.110
South	1.00
Unmarried	1.44	1.04–1.99	0.030
Period of first hospital admission
1999–2003	0.69	0.37–1.28	0.235
2008–2010	0.49	0.31–0.78	0.003
2004–2007	1.00
Cardiovascular disease	1.30	0.89–1.89	0.147
Renal disease	1.34	0.90–1.98	0.589
Eye disease	1.45	1.06–2.00	0.022
Perceived health status
Good	0.76	0.49–1.18	0.221
Fair or poor	1.09	0.70–1.67	0.709
Excellent or very good	1.00
Physical limitations	1.09	0.79–1.51	0.589

^aThe patient propensity score was also included in the regression. Its estimated coefficient was not statistically significant (p > 0.100)

Table 4. ICD-9 Codes for diabetes-related conditions
Condition	ICD-9 code(s)
Diabetes	250
Cardiovascular disease	410–414, V458, 398, 428,426,427,785, V450, V533, 430–438, 437, 415–417, 440–414
Renal disease	585–586, V420, V451, V56, 580–584, 590, 595,597, 598, 599
Lower extremity disease	337,342–344, 354,355,356,357,358, 440, 442, 443, 444, 445, 711, 718, 727, 739, 735, 736, 784, 020, 021, 022, 031, 032, 035, 039, 680–682, 684–686, 690, 694–698, 700–703, 707, 709, V133, V423
Eye disease	361,362, 365–369,V431, V410
Mycoses	110–112, 114, 115, 116–118
Fluid and electrolyte disorders	276
Fractures	800–829
Syncope	780
Other injuries	959
Hypoglycemia	251
Other disorders of soft tissue	729

Table 5. Medication classification codes for oral antihyperglycemic agents
Drug class	Therapeutic classification code
Sulfonylureas	213
Other oral antihyperglycemic agents
Biguanides	214
Thiazolidinediones	271
Alpha-glucosidase inhibitors	216
Meglitinides	282
Dipeptidyl peptidase 4 inhibitors	371
Antidiabetic combinations	314