Hospital Readmission and Emergency Department Use Based on Prescribing Patterns in Patients with Severely Uncontrolled Type 2 Diabetes Mellitus

Abstract

Background:

Patients with uncontrolled diabetes are more likely to be readmitted to the hospital. The study objective was to determine the risk of hospital admission or emergency department (ED) use in patients with severely uncontrolled type 2 diabetes mellitus based on whether their diabetes medication regimen was intensified at discharge.

Subjects and Methods:

A retrospective cohort study of patients admitted to the medicine services at an academic medical center was conducted during a 9-month period. Medical records were reviewed to identify patients with type 2 diabetes mellitus and hemoglobin A1c >10% who were admitted for more than 24 h. Primary exclusion criteria included pregnancy, age >65 or <18 years, life expectancy of <12 months, hypoglycemia at admission, and new diagnosis of type 2 diabetes. Admission medication lists for patients were compared with discharge medications to determine whether the diabetes regimen was intensified. Patients whose regimen was intensified were then compared with patients whose regimen was not intensified for admissions and ED visits within 3 months after discharge. The primary end point was 90-day all-cause hospital and ED visits.

Results:

Patients who had their regimens intensified had significantly fewer average all-cause hospital and ED visits at 90 days (0.41 vs. 0.85; P=0.044). Statistical differences were also found in 30-day all-cause visits, readmissions within 90 days, and hospital-free time.

Conclusions:

Patients whose home diabetes regimen was intensified at discharge were less likely to return to the hospital within 90 days compared with patients whose regimen was not intensified.

Background

Type 2 diabetes mellitus is a common comorbid condition of hospitalized patients. Patients with type 2 diabetes are more likely to be admitted to the hospital and to have longer lengths of stay. The number of patients with type 2 diabetes admitted to the hospital is likely to grow as the prevalence of type 2 diabetes continues to climb. An inpatient admission provides an opportunity to address a patient's diabetes therapy.¹

Patients who have diabetes have an increased cost of medical care and rate of readmission than patients without diabetes. The Centers for Disease Control and Prevention has estimated that expenditures for health care for patients that have diabetes are more than two times greater than those without a diagnosis of diabetes.² A cross-sectional analysis showed that patients with diabetes were almost twice as likely to visit the emergency room or to be admitted to the hospital as those without diabetes.³ Furthermore, increased rates of readmission have been noted in patients with diabetes and other comorbid conditions, including congestive heart failure or liver disease or those who underwent coronary artery bypass grafting.^4

–7

Also, patients with uncontrolled diabetes have more costly care and are at a greater risk of readmission than those with diabetes that is controlled. The American Diabetes Association (ADA) defines severely uncontrolled diabetes with catabolism as fasting glucose plasma glucose levels of >250 mg/dL, a random plasma glucose level of >300 mg/dL, hemoglobin A1c (HbA1c) level of >10%, the presence of ketonuria, or the occurrence of symptomatic polyuria, polydipsia, or weight loss.⁸ In an analysis performed by Menzin et al.,⁹ patients with uncontrolled diabetes, defined as an HbA1c level ≥10%, were shown to have an almost $4,000 increased cost of hospitalization compared with patients that were at the ADA HbA1c goal of <7%. Also, patients with uncontrolled diabetes mellitus were more than twice as likely to be admitted to the hospital with diabetes mellitus-related complications.⁹ A study by Dungan et al.¹⁰ found that both inpatient glucose control and HbA1c were predictive of hospital readmission between 30 and 90 days of discharge in patients with congestive heart failure; the authors suggested long-term glycemic control contributes to the risk of readmission even in cases when inpatient blood glucose readings are at acceptable levels.

Changes in healthcare payment in the United States may force practitioners to ensure that patients' readmission risk is as low as possible after discharge. The Hospital Readmission Reduction Program was developed as a result of the U.S. Patient Protection and Affordable Care Act. The Hospital Readmission Reduction Program uses risk standardized readmission rates to assess hospital performance when discharging patients with acute myocardial infarction, heart failure, and pneumonia. Beginning in 2013, hospitals that do not meet acceptable readmission standards will be penalized; moreover, this program is set to be expanded to other diagnoses after 2015.¹¹ It has been suggested that diabetes is one disease state where quality of care can be assessed by rate of readmission.¹² One study suggested that one in seven readmissions in patients with diabetes was associated with substandard care.¹³ Not addressing chronic diabetes management during an inpatient visit may become very costly, with the possibility of reduced reimbursement based on readmission.

The objective of this study was to determine the risk of hospital admission or emergency department (ED) use in patients with an HbA1c level of >10% based on whether their diabetes medication regimen was intensified at discharge.

Subjects and Methods

A retrospective cohort study was completed using medical records from a 581-bed academic medical center in the United States. This study was approved by the University of Tennessee Graduate School of Medicine Institutional Review Board. A report was populated with all patients who had an HbA1c level of >10% during a hospital admission at our institution from October 2010 through June 2011. The report was then screened for inclusion and exclusion criteria.

Eligible patients included those patients admitted to one of the three medicine services at the medical center who had an HbA1c level >10% during their admission. The medical center supports both an internal medicine and a family medicine residency program that have inpatient medicine services. The third service is a private non-teaching hospitalist medicine service.

Exclusion criteria included pregnancy, admission for <24 h, age >65 or <18 years, life expectancy of <12 months, blood glucose level of <70 mg/dL at admission, type 1 diabetes, and new diagnosis of type 2 diabetes during admission. Patients older than 65 years of age were excluded to account for the potential that the patient may have a higher HbA1c goal. Patients were also excluded if continuation of their home medication regimen resulted in no episodes of hyperglycemia during the admission (suggesting the patient was nonadherent to medications at home) or the patient had a documented hemoglobin variant because of its ability to affect the HbA1c assay. To address multiple admissions of the same patient, if a patient was admitted twice during the study period more than 3 months apart and an HbA1c was measured during both admissions that was >10%, each admission was analyzed separately.

Eligible patients were then divided into two cohorts: patients whose diabetes regimen was intensified at discharge and patients whose medication was not intensified at discharge. This determination was made by comparing the patient's home medication list at admission with the discharge medication list. A patient's diabetes regimen was considered intensified if any of the following changes was made: (1) increase of a medication's dose, (2) addition of a new non-insulin medication, or (3) addition of insulin therapy.⁸ The patient's medical record was then examined to identify readmissions to the hospital and ED visits within 90 days of discharge.

During the study period medication reconciliation was performed by the nursing staff. Medication lists were obtained through many methods, including patient interview, patient-provided medication lists, and contact with the patient's dispensing pharmacy. These lists were then either added to the paper chart or included in the patient's electronic medical record for provider review. We collected the lists retrospectively from the patient's medical record.

Baseline characteristics of each cohort were analyzed using descriptive statistics as well as both independent-samples t tests and χ ² analyses, as appropriate. The primary outcome examined was whether hospital admission and ED visit rates were associated with whether a patient's chronic diabetes medication regimen was intensified at discharge. Statistical normality of this outcome variable was reviewed, and if the data were found to be non-normal, the Mann–Whitney U test was used to account for violating this assumption. Time to first ED visit or hospital readmission (hospital-free time) was also compared using a Kaplan–Meier analysis with log-rank statistics as well as a Cox proportional hazards regression to control for possible cofounders.

Secondary outcomes including 90-day ED visits, 90-day readmissions, diabetes-related visits, 30-day hospital readmissions, and 90-day total visits based on whether insulin was added to a patient's chronic diabetes regimen were all compared using Mann–Whitney U analysis. Admissions were considered diabetes-related if they were consistent with the definition used by Menzen et al.⁹ All statistical tests were two-tailed and assumed an α of P<0.05 for reaching statistical significance.

Results

In total, 238 patients during the time period examined had an HbA1c level of >10% and were screened for inclusion and exclusion. The two most common reasons for exclusion were patients not being admitted by one of the inpatient medicine services and age greater than 65 years. No patients were excluded for normoglycemia during admission while only on home medications. Of these patients, 83 met the criteria to be included in the study. Forty-nine patients were determined to have had their regimen intensified, and 34 patients did not have their regimen intensified. Table 1 gives the demographic and baseline characteristic comparisons between the two groups. HbA1c was found to be statistically greater in the intensified cohort (12.12±1.72% vs. 11.47±1.05%; P=0.037), whereas all other demographics and baseline characteristics were not found to be statistically different. Of the 49 patients who had their regimen intensified, 21 had insulin initiated, 18 had doses of current medications increased, and 10 had additional medications added. Of the 21 patients initiated on insulin, 10 were started on basal insulin, and three of those also had non-insulin therapy initiated. Ten patients had basal and bolus insulin initiated, with one of those patients also having non-insulin therapy initiated, and one patient started on insulin therapy was started on bolus insulin and non-insulin therapy. Of patients who were included for increased dose, 16 had the insulin dose increased, one had the dose of non-insulin therapies increased, and one had both insulin and non-insulin doses increased. For patients included for new medication initiation, six were started on additional insulin therapy, three were started on sulfonylureas, and one was started on a sulfonylurea plus metformin.

Table 1.

Baseline Demographics of the Study Population

	Intensified (n=49)	Not intensified (n=34)	P value ^a
Age (years) (mean±SD)	50.33±9.28	52.47±9.96	0.318
IBW (kg) (mean±SD)	68.71±12.17	64.18±10.66	0.083
BMI (kg/m²) (mean±SD)	31.87±8.77	33.64±9.49	0.385
HbA1c (%) (mean±SD)	12.12±1.72	11.47±1.05	0.037
Male (%)	67.3%	52.9%	0.185
Ethnicity (%)
White	87.8	79.4	0.308
African American	10.2	20.6
Asian	2.0	0
Comorbid conditions (%)
Heart failure	12.2	23.5	0.177
Previous MI	6.1	17.6	0.097
Pancreatitis	24.1	11.8	0.184
Diabetes-related admission diagnosis (%)	55.1	41.2	0.212
Admission diabetes regimen
None	—	15 (30.6%)	—
Insulin	18 (52.9%)	14 (28.6%)	Reference
Non-insulin therapy	5 (14.7%)	12 (24.5%)	0.079
Insulin+non-insulin therapy	11 (32.4%)	8 (16.3%)	0.909
Discharge regimen
None	2 (5.9%)	—	—
Insulin	21 (61.8%)	27 (55.1%)	Reference
Non-insulin therapy	4 (11.8%)	7 (14.3%)	0.655
Insulin+non-insulin therapy	7 (20.6%)	15 (30.6%)	0.346

Groups were compared using independent-samples t test or χ ² test of independence, as appropriate.

BMI, body mass index; HbA1c, hemoglobin A1c; IBW, ideal body weight; MI, myocardial infarction.

Four patients were readmitted during the study period and were included a second time in the study based on the study protocol. One patient was included twice in the non-intensified group, and one was included twice in the intensified group. Two other patients were included once in each cohort. To ensure that this did not affect study outcomes of the study, readmissions were removed from data, and outcomes were reanalyzed. No statistical differences were found in the outcomes after removing the readmissions.

The mean all-cause readmissions and ED visits in patients who had their diabetes regimen intensified were significantly lower than in patients whose regimen was not intensified. Patients who had their regimen intensified averaged 0.41 total visits (ED visits plus readmissions) during the follow-up period compared with 0.85 total visits for patients whose regimen was not intensified (P=0.044) (Table 2).

Table 2.

Hospital Readmission and Emergency Department Use

Variable	Intensified (n=49)	Not intensified (n=34)	P value ^a
Primary outcome
90-day total visits	0.41±0.86	0.85±1.16	0.044
Secondary outcomes
90-day hospital readmissions	0.12±0.53	0.56±1.00	0.002
90-day ED visits	0.29±0.65	0.29±0.58	0.767
90-day diabetes-related total visits	0.18±0.53	0.44±0.80	0.168
30-day total visits	0.08±0.028	0.41±0.66	0.004

	Insulin added (n=21)	Insulin not added (n=62)	P value ^a
90-day total visits	0.33±0.66	0.69±1.01	0.077

Data are mean±SD values. Total visits were defined as emergency department (ED) visits plus hospital readmissions.

Comparisons were assessed using Mann–Whitney U test analysis.

Patients in the intensified cohort had a significantly longer hospital-free period (P=0.023). The average hospital-free time was 77.45 days (95% confidence interval, 70.55–84.35) in the intensified group versus 62.12 days (95% confidence interval, 49.34–74.89) in the non-intensified group (Fig. 1). Using a Cox proportional hazards model intensification of regimen did not reach statistical significance (hazard ratio, 0.55; 95% confidence interval, 0.25–1.21; P=0.136) when controlling for body mass index, HbA1c, and age (Table 3).

FIG. 1.

Readmission rates for intensified and non-intensified patients.

Table 3.

Cox Proportional Hazards Models for 90-Day Total Visits

Variable	HR	95% CI	P value
Regimen intensified	0.55	0.25–1.21	0.136
BMI	1.02	0.97–1.06	0.467
HbA1c	0.74	0.53–1.05	0.089
Age	1.00	0.96–1.04	0.927

Total visits were defined as emergency department visits plus hospital readmissions.

BMI, body mass index; CI, confidence interval; HbA1c, hemoglobin A1c; HR, hazard ratio.

Hospital readmission contributed greatly to total visits and was also found to be statistically different. Patients who did not have their regimens intensified were much more likely to have been readmitted to the hospital within 90 days (mean readmissions, 0.56 vs. 0.12; P=0.002). The rate of ED visits within 90 days was similar for each group. A statistically significant difference was also found for 30-day hospital use between groups. Patients whose regimen was not intensified visited the hospital within 30 days an average of 0.41 times versus 0.08 visits for those whose regimen was intensified (P=0.004). Although diabetes-related visits were more common in the non-intensified group, no statistically significant difference was found (mean diabetes-related visits, 0.44 vs. 0.18; P=0.168). Finally, when comparing the subset of patients who were initiated on insulin versus those who were not, initiation of insulin was not found to be predictive of readmission (Table 2).

Discussion

This study suggests that adult patients with severely uncontrolled type 2 diabetes who have their chronic diabetes regimens intensified at discharge are less likely to return to the hospital at 30 and 90 days, have fewer readmissions, and have a longer period between hospital visits than patients who do not have their regimens intensified. Based on these findings, it is imperative that a patient's home diabetes regimen be addressed at discharge when a patient is found to have an HbA1c level of >10%. Furthermore, it may be prudent to assess a patient's blood glucose control during every admission by measuring the HbA1c level and making changes to the patient's diabetes therapy in an effort prevent future admissions.

When accounting for confounders including age, body mass index, and baseline HbA1c using the Cox proportional hazards model, intensification was not shown to be statistically significant, although the odds ratio favored intensification. The lack of significance likely resulted from the limited sample size.

Although no statistical difference was found in diabetes-related visits in this study, a larger sample may provide a definitive difference; however, focusing only on diabetes-related visits may limit the interpretation of these results. Diabetes mellitus leads to many other conditions that may have resulted in visits that may not be considered diabetes related. Examining the difference in all-cause visits allowed us to include the negative effects that uncontrolled type 2 diabetes may have on one's general health.

Another interesting finding was that there was no difference shown in hospital visits for patients who initiated insulin therapy at discharge. The ADA strongly encourages initiation of insulin in the patient population studied, as this approach to management will quickly and effectively lower the glucose level.¹⁴ This could also lead to decreases in hospital admissions as the patient approaches glycemic control. In addition, it has also previously been shown that continuation of insulin therapy upon discharge is associated with reduced readmissions.^15,16 The concern for hypoglycemia after initiation of insulin may be a roadblock to discharging a patient on appropriate therapy for uncontrolled type 2 diabetes mellitus; however, the results of this study and previous research suggest that initiating insulin in a patient at discharge may be accomplished without increasing the risk of hospital readmission.

The results of the present study are likely the consequence of continued uncontrolled hyperglycemia following discharge leading to increased morbidity and comorbidity of type 2 diabetes. Previous data have suggested that diabetes itself is predictive of increased hospital admission.^3

–7 When diabetes is uncontrolled this risk is increased to a greater extent.^9,10 Results of the present study are consistent with previous findings and should add to the argument that a patient's diabetes regimen must be addressed when a patient is admitted to the hospital. Addressing this issue is of even greater importance if the patient is shown to have severely uncontrolled type 2 diabetes.

There are many barriers that may prevent a provider from intensifying a patient's regimen at discharge, including (1) the fear of hypoglycemia after discharge, (2) the possibility of offending the patient's primary care provider, (3) limited time to address primary care issues at discharge, (4) lack of confidence to effectively address these therapies at discharge, (5) lack of effective transition of care processes, and (6) patient-specific factors such as mental or physical disabilities or financial and social barriers. Our findings suggest these barriers must be overcome to ensure that patients receive the best care. An appropriate transition of care process from inpatient to outpatient could address many of these issues. Processes could include medication reconciliation at discharge by a specialized practitioner or team to ensure that diabetes care is addressed appropriately. This practitioner or team could also communicate with primary care providers both before and at discharge to ensure collaboration and appropriate follow-up.

There are some limitations to the study that may limit the interpretation of the results. This study was retrospective in nature and could not be controlled for several variables. There are several factors that may limit the external validity of our results, including a lack of randomization, the single-center nature of the study, inclusion of patients only admitted to a general medicine service, and restricting enrollment to patients with an HbA1c level of >10%. Patients over 65 years of age were excluded from the study, as this age group may be more likely to have higher HbA1c goals than the general population. However, this also limits the generalizability of the study results to patients in this age group. Moreover, patients were only followed up for use of the study site during the follow-up period; therefore, admissions to other institutions were not captured. In addition, information regarding additional diabetes care (nutritional counseling, consultations for diabetes-specific care, etc.) while inpatient, follow-up with primary care provider following discharge, and follow-up HbA1c values were also not collected. Intensification of therapy was determined using patient-reported home medication lists, which may not reflect the medications that the patient was actually taking. On the other hand, this limitation provides authenticity to the study because the patient-reported home medication list is the same information that the provider uses on which to base decisions regarding therapy. The small sample size studied may have also limited the interpretation of the results. Patient adherence was not assessed either prior to admission or during the follow-up period. Finally, HbA1c testing is an average of blood glucose measurement over 3 months and may not always reflect the current control of a patient's diabetes mellitus. Although these limitations were present, the study indicates that intensification of therapy may provide a beneficial impact on hospital readmission and ED use in this population. A prospective, randomized, multicenter study would be needed to confirm these findings.

Footnotes

Acknowledgments

The authors thank B. Blake Poole, University of Tennessee Health Science Center College of Pharmacy, for his assistance in data collection and Brian Hamblen, iDesign, for his assistance with graphics. At the time of this study, P.H.L. was a PGY-1 Pharmacy Resident at University of Tennessee Medical Center and University of Tennessee Health Science Center College of Pharmacy, Knoxville, TN.

Author Disclosure Statement

M.Z.F. was a member of the Merck & Co, Inc. Speaker Bureau (2009–2011). P.H.L., A.S.F., and P.B.B. declare no competing financial interests exist.

References

Moghissi

, Korytkowski

, DiNardo

, Einhorn

, Hellman

, Hirsch

, Inzucchi

, Ismail-Beigi

, Kirkman

, Umpierrez

; American Association of Clinical Endocrinologists; American Diabetes Association: American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Diabetes Care, 2009; 32:1119–1131.

Centers for Disease Control and Prevention: National diabetes fact sheet. 2011. www.cdc.gov/diabetes/pubs/pdf/ndfs_2011.pdf (accessed July 10, 2012 ).

Natarajan

, Nietert

: Hypertension, diabetes, hypercholesterolemia, and their combinations increased health care utilization and decreased health status. J Clin Epidemiol, 2004; 57:954–961.

Berman

, Tandra

, Forssell

, Vuppalanchi

, Burton

Jr , Nguyen

, Mullis

, Kwo

, Chalasani

: Incidence and predictors of 30-day readmission among patients hospitalized for advanced liver disease. Clin Gastroenterol Hepatol, 2011; 9:254–259.

Sun

, Zhang

, Lowery

, Petro

, Hill

, Haile

, Garcia

, Bafi

, Boyce

, Corso

: Early readmission of low-risk patients after coronary surgery. Heart Surg Forum, 2008; 11:E327–E332.

Stewart

, Campos

, Jennings

, Lollis

, Levitsky

, Lahey

: Predictors of 30-day hospital readmission after coronary artery bypass. Ann Thorac Surg, 2000; 70:169–174.

Aranda

Jr , Johnson

, Conti

: Current trends in heart failure readmission rates: analysis of Medicare data. Clin Cardiol, 2009; 32:47–52.

Nathan

, Buse

, Davidson

, Ferrannini

, Holman

, Sherwin

, Zinman

; American Diabetes Association; European Association for Study of Diabetes: Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care, 2009; 32:193–203.

Menzin

, Korn

, Cohen

, Lobo

, Zhang

, Friedman

, Neumann

: Relationship between glycemic control and diabetes-related hospital costs in patients with type 1 or type 2 diabetes mellitus. J Manag Care Pharm, 2010; 16:264–275.

10.

Dungan

, Osei

, Nagaraja

, Schuster

, Binkley

: Relationship between glycemic control and readmission rates in patients hospitalized with congestive heart failure during implementation of hospital-wide initiatives. Endocr Pract, 2010; 16:945–951.

11.

Kocher

, Adashi

: Hospital readmissions and the Affordable Care Act: paying for coordinated quality care. JAMA, 2011; 306:1794–1795.

12.

Benbassat

, Taragin

: Hospital readmissions as a measure of quality of health care: advantages and limitations. Arch Intern Med, 2000; 160:1074–1081.

13.

Ashton

, Kuykendall

, Johnson

, Wray

, Wu

: The association between the quality of inpatient care and early readmission. Ann Intern Med, 1995; 122:415–421.

14.

Inzucchi

, Bergenstall

, Buse

, Diamant

, Ferrannini

, Nauck

, Peters

, Tsapas

, Wender

, Matthews

; American Diabetes Association (ADA); European Association for the Study of Diabetes (EASD): Management of hyperglycemia in type 2 diabetes: a patient-centered approach: position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care, 2012; 35:1364–1379.

15.

, Zhou

, Yu

, Lu

, Sharma

, Gill

, Graf

: Outcomes associated with insulin therapy disruption after hospital discharge among patients with type 2 diabetes mellitus who had used insulin before and during hospitalization. Endocr Pract, 2012; 18:651–659.

16.