Hospital Practice
ISSN: 2154-8331 (Print) 2377-1003 (Online) Journal homepage: http://www.tandfonline.com/loi/ihop20
Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population Leigh E. Efird PharmD, Sherita H. Golden MD, MHS, Kanizeh Visram BS & Kenneth Shermock PharmD, PhD To cite this article: Leigh E. Efird PharmD, Sherita H. Golden MD, MHS, Kanizeh Visram BS & Kenneth Shermock PharmD, PhD (2014) Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population, Hospital Practice, 42:1, 101-108 To link to this article: http://dx.doi.org/10.3810/hp.2014.02.1097
Published online: 13 Mar 2015.
Submit your article to this journal
Article views: 3
View related articles
View Crossmark data
Citing articles: 1 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ihop20 Download by: [Central Michigan University]
Date: 12 September 2015, At: 20:01
H e a lt h O u t c o m e s
Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population
Downloaded by [Central Michigan University] at 20:01 12 September 2015
DOI: 10.3810/hp.2014.02.1097
Leigh E. Efird, PharmD 1 Sherita H. Golden, MD, MHS 2 Kanizeh Visram, BS 3 Kenneth Shermock, PharmD, PhD 4 1 Clinical Pharmacy Specialist, Internal Medicine, Department of Pharmacy, The Johns Hopkins Hospital, Baltimore, MD; 2Associate Professor of Medicine and Epidemiology, The Johns Hopkins University School of Medicine and Bloomberg School of Public Health, Baltimore, MD; 3 San Francisco School of Pharmacy, University of California, San Francisco, CA; 4Director, Center for Medication Quality and Outcomes, Division of General Internal Medicine, The Johns Hopkins Hospital, The Johns Hopkins University School of Medicine, Baltimore, MD; Department of Epidemiology, The Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD
Abstract
Objective: A pharmacy-based glucose management program was evaluated to determine whether improved glycemic control could be achieved in an inpatient general medicine patient population. Methods: A retrospective chart review of 151 patients with blood glucose (BG) values outside the range of 70 to 180 mg/dL within a 24-hour period was conducted. Observations for the baseline group with no pharmacy program in place were collected from admissions during July 2010 and for the intervention group in October 2010. The primary goal of the study was to determine if the pharmacy program improved patient days within the BG range of 70 to 250 mg/dL. The odds of poor glycemic control for patients in the intervention versus baseline groups were assessed by multivariate generalized estimating equations. These methods were also used to assess patient characteristics associated with poor glycemic control. Results: No evidence was observed that the pharmacy program decreased the number of days spent out of the targeted blood glucose range (70–250 mg/dL; odds ratio, 1.08; 95% CI, 0.88–1.24). However, the subgroup of patients whose admission BG was , 200 mg/dL (49% of the intervention group) experienced a significant reduction in days spent out of the BG range (70–250 mg/dL; odds ratio, 0.42; 95% CI, 0.22–0.82). No improvement in glycemic control was observed in patients with an admission BG $ 200 mg/dL; these patients had more disease- and social-related factors associated with poor glycemic control. Conclusion: The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values compared with usual care. Patients whose admission glucose was , 200 mg/dL experienced improvement in glycemic control in the pharmacy-based program. Maintaining the BG level of the remaining patients was generally more complicated from a disease-state and social perspective and patients experienced no improvement. These patients may require a more intense, multidisciplinary approach that is better matched to the constellation of factors responsible for their condition. Keywords: glucose management; pharmacy-based; inpatient; general medicine
Introduction
Correspondence: Leigh E. Efird, PharmD, The Johns Hopkins Hospital, 600 North Wolfe Street, Carnegie 180, Baltimore, MD 21287. Tel: 410-502-5703 Fax: 410-955-0287 E-mail: [email protected]
The diabetes mellitus (DM) epidemic continues to grow in the United States. In 2011, DM was estimated to affect approximately 8.3% of the population, or 25.8 million people, including 7 million undiagnosed cases.1 According to the Centers for Disease Control and Prevention (CDC), the incidence of DM has nearly tripled since 1980.2 Optimal care for this growing population of individuals is evaluated through the Centers for Medicare and Medicaid Services quality performance standards. The Joint Commission, which provides advanced certification for hospitals, also implements standards to improve inpatient glucose management.3,4 The first formal consensus
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 101 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 101
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Efird et al
statement regarding target blood glucose (BG) goals in the non–critically ill patient population was issued in 2004 by the American College of Endocrinology and the American Association of Clinical Endocrinologists.5 Subsequently, the American Diabetes Association (ADA) published guidelines that identified a preprandial BG level , 140 mg/dL and a random BG level , 180 mg/dL as reasonable target levels for inpatients with DM.6 Limited evidence exists to support these BG targets or define patient characteristics for increased risk of poor glycemic control in the inpatient setting. However, data do suggest that newly identified hyperglycemia is associated with increased hospital mortality in non–critically ill patients.6,7 There is evidence that glycemic control during a hospitalization is important to achieve positive patient outcomes.7–9 In the non–critically ill patient population, hyperglycemia in patients without a previous diagnosis of DM has been identified as an independent predictor of hospital mortality when compared to patients with a diagnosis of DM (10% vs 1.7%; P , 0.01).7 There are also data suggesting that patients admitted with community-acquired pneumonia and an admission BG . 200 mg/dL have a higher mortality rate compared with patients with an admission BG # 200 mg/dL (13% vs 9%; P = 0.03).8 Baker et al9 showed a 15% increase in the absolute risk of adverse patient outcomes for every 18 mg/dL increase in BG level when adjusted for age, sex, and DM diagnosis in patients with chronic obstructive pulmonary disease and respiratory infection. Finally, it has also been demonstrated that higher BG values and severe diabetic complications in patients upon admission may be associated with higher mean BG values per patient-day in general medicine patients.10 Hyperglycemia in these patient populations may be due to the stress response associated with the severity of their illness on presentation, which could be associated with higher mortality rates. These results have contributed to the recommendation for standardized approaches in managing hospital inpatients in the non–critically ill patient population.6 Our institution formed a multidisciplinary glucose management committee in 2006 to address many of the concerns regarding inpatient glycemic control.11 In conjunction with this committee, the institution implemented a pharmacy-based glucose management program targeting improved glycemic control in general medicine patients during the month of October 2010. This opportunity allowed pharmacists to make antihyperglycemic regimen recommendations and educate providers at the time the treatment plan was being discussed and designed. The primary objective of the study was to assess the impact of the pharmacy-based
glucose management program on glycemic control in an inpatient general medicine population.
Materials and Methods
All patients admitted to 1 of 4 general medicine teams in July 2010 (baseline group) or October 2010 (intervention group) with $ 2 BG values outside of the range of 70 to 180 mg/dL in any 24-hour period prior to inclusion into the study were included in the data analyses. For the baseline group, there was a clinical pharmacy specialist rounding on only 1 of the 4 general medicine teams consistently to focus on designing medication regimens at the time treatment plans were made. Other routine practices in place during both periods included a hypoglycemia policy, a hospital-wide nursing DM education program, a hyperglycemia policy, and a computerized physician order entry set with recommendations for basal and bolus insulin based on patient characteristics. Prior to the intervention period, a case-based glucose management curriculum was provided to pharmacists to improve their skill and knowledge levels in anticipation of their participation in the program. This curriculum focused on educating the rounding pharmacists on how to appropriately transition patients from an outpatient to an inpatient antihyperglycemic regimen, adjust subcutaneous insulin in the inpatient setting, transition patients off insulin drips, and transition patients from an inpatient to an outpatient antihyperglycemic regimen. Pharmacists were then required to undergo a shadowing phase where recommendations and documentation were reviewed by a clinical pharmacy specialist. Once deemed competent in providing appropriate glucose management recommendations, the pharmacist was assigned an inpatient team to follow during the intervention period. Further details of the program have previously been published.11 During the intervention period, a separate pharmacist was assigned to round on each of 4 medicine teams and provided medication therapy recommendations for patients assigned to those teams on a daily basis Monday through Friday. As a part of providing this service, pharmacists formulated a daily glucose management plan that was communicated to the medical team and documented in the patient’s electronic medical record (EMR). Factors considered in designing the antihyperglycemic regimen included DM diagnosis on admission, insulin needs during the last 24 hours, current nutritional status, glycated hemoglobin (HbA1c) level within the last 3 months, home antihyperglycemic regimen, weight, interacting medications, and renal function. Documentation of the glucose management plan included recommendations
102
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 102
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Pharmacy-Based Glucose Management Program
for basal, nutritional, and correctional insulin as deemed necessary to meet target inpatient BG goals. Entry of orders pertaining to medication regimens was ultimately left to the provider. Other members of the rounding medical team included medicine interns, senior residents, an attending, and a case manager with a nursing background. Baseline patient characteristics were retrospectively abstracted from the EMR by 2 independent reviewers who developed consensus when their assessments were discordant. These included the patients’ age, race, DM diagnosis, admission BG, HbA1c level, home DM regimen, and if the current admission was DM-related (indicated by admission related to DM complications, gastroparesis, diabetic ketoacidosis, hyperosmolar hyperglycemic state, or hyperglycemia). Other patient factors, including the presence of infection on admission, current use of oral steroids, any DSM-IV psychiatric diagnosis, history of medication nonadherence, history of illicit drug use, and unstable social, job, or insurance status (indicated by documentation of unemployment, uninsured status, or homelessness), were also retrospectively abstracted from documentation during the inclusion admission from the EMR. The primary endpoint of the study was the odds of having $ 2 BG values in a given hospital day outside the range of 70 to 250 mg/dL (after the patient met inclusion criteria for the study; ie, the patient already had 2 BG values outside 70–180 mg/dL prior to study entry) for patients in the baseline group compared with the intervention group. Secondary endpoints were the determination of patient characteristics associated with having poor glycemic control and the odds of poor glycemic control in the baseline group compared with the intervention group based on the following criteria: 1) patient BG level outside the range of 70 to 180 mg/dL; 2) patient BG level . 180 mg/dL; 3) patient BG level . 250 mg/dL; 4) patient BG level , 70 mg/dL. Analyses were conducted for the overall groups, and for subgroups of patients based on their admission BG value of , 200 mg/dL, 200 to 300 mg/dL, and . 300 mg/dL. A patient BG value of , 200 mg/dL has been recommended as an appropriate upper limit of the target range in non–critically ill inpatients who are at high risk for developing hypoglycemia, which was the basis of our lower limit in this range.12 Differences in baseline characteristics between the study groups were assessed using the Student’s t test or the Mann-Whitney U test for continuous variables and a Χ 2 or Fisher’s exact test for categorical variables. Bivariate analyses were conducted to identify relationships between each baseline patient characteristic and the odds of experiencing
poor glycemic control. These relationships were assessed statistically at the patient-day level using generalized estimating equations (GEEs) to assess the odds of poor glycemic control on a given day. The GEE techniques account for correlation of response variables within individuals.13 They provide more robust model estimates for longitudinal outcomes, such as the odds of a given patient experiencing poor glycemic control on a given day. For these GEE models, a binomial distribution, logit link function, and autoregressive correlation structure were specified. Multivariate GEE techniques were used to assess the relationship between patient characteristics and the rate of poor glycemic control. Multivariate GEE models were also used to assess differences in glycemic control between the baseline and intervention groups. Final model selection used a forward selection process, starting with the most significant variable from bivariate analysis and included variables that achieved a significance level of P , 0.1 or lower in bivariate analyses. This model building process was guided by Akaike’s Information Criterion. The Χ 2 test for comparison of proportions was used to assess binary outcomes that were not longitudinal. Odds ratios (ORs) and 95% CIs were also calculated for these analyses. A P value , 0.05 was considered statistically significant for all analyses. Analyses were conducted using STATA (Version 11.0). The Johns Hopkins Medicine Institutional Review Board approved this study.
Results
A total of 151 patients, 84 (738 patient-days) in the baseline group and 67 (560 patient-days) in the intervention group, were included in the study. At baseline, the majority of patients were African American and had a previous diagnosis of type 2 DM (Table 1). Between study groups, patients had a similar distribution of baseline HbA1c values and home treatment regimens. The only significant difference detected in baseline characteristics was a higher percent of patients with a history of medication nonadherence in the intervention group compared with the baseline group (27% vs 11%; P = 0.01). Irrespective of study period, a strong majority of patients (91%; n = 137) met the criteria for being outside the range of 70 to 180 mg/dL for $ 1 day during their inclusion in the study. On average, patients were outside the range of 70 to 180 mg/dL for 60% of their hospitalized days (median = 67%). Seventy percent (n = 106) of patients were outside the range of 70 to 250 mg/dL on $ 1 day of their hospitalization. The average and median percent of days (36% and 33%, respectively) spent outside this range were similar. Eighteen percent (n = 27) of patients had a
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 103 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 103
1/31/14 5:19 PM
Efird et al
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Table 1. Demographics and Baseline Characteristics of Study Groups Variable
No pharmacy program (n = 84)
Pharmacy program (n = 67)
P valuea
Age (y), mean (SD) Race, n (%) African American White Other DM diagnosis, n (%) No DM T1DM T2DM Admission BG level (mg/dL), mean (SD) Admission BG category (mg/dL), n (%) , 200 200–300 . 300 HbA1c level, mean (SD) HbA1c category, n (%) ,7 $7 No baseline measure Home DM regimen, n (%) Insulin Oral medication Insulin and oral medication None documented Admitted for DM complications, n (%) Admitted for hyperglycemic episode, n (%) Documented infection on admission, n (%) Documented current oral steroid use, n (%) History of psychiatric diagnosis, n (%) History of medication nonadherence, n (%) Unstable social, job, insurance status, n (%) History of illicit drug use, n (%)
60 (15)
62 (16)
0.37
57 (68) 22 (26) 5 (6)
41 (61) 24 (36) 2 (3)
0.40
28 (33) 10 (12) 46 (55) 228 (171)
20 (30) 5 (7) 42 (63) 283 (260)
0.54
49 (58) 15 (18) 20 (24) 8.2 (2.8)
33 (49) 15 (22) 19 (28) 8.7 (3.1)
0.53
27 (32) 36 (43) 21 (25)
21 (31) 32 (48) 14 (21)
0.79
32 (38) 15 (18) 5 (6) 32 (38) 11 (13) 11 (13) 23 (27) 14 (17) 26 (31) 9 (11) 18 (21) 19 (23)
27 (40) 10 (15) 5 (7) 25 (38) 11 (16) 7 (10) 26 (39) 7 (10) 21 (31) 18 (27) 14 (21) 12 (18)
0.95
0.12
0.31
0.56 0.62 0.16 0.27 0.96 0.01 0.94 0.48
a P value is reported for each characteristic as a whole unless otherwise stated. Abbreviations: BG, blood glucose; DM, diabetes mellitus; HbA1c, glycated hemoglobin; SD, standard deviation; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
BG level , 70 mg/dL and only 2 patients (1.3%) had a BG level , 40 mg/dL during the inpatient stay. Bivariate analysis identified significant associations between several patient characteristics and poor glycemic control (irrespective of study group), including known DM diagnosis on admission, admission BG values of $ 200 mg/dL, baseline HbA1c level $ 7%, admission for a DM-related complication, history of medication nonadherence, and unstable social, job, or insurance status (Table 2). There was no overall difference in glycemic control between the baseline and intervention groups (Table 3). However, in the subgroup of patients with an admission BG level , 200 mg/dL, the intervention was associated with a significant decrease in the odds of poor glycemic control for the BG ranges of 70 to 180 mg/dL (OR, 0.59; 95% CI, 0.38–0.91) and 70–250 mg/dL (OR, 0.42; 95% CI, 0.22–0.82). For patients with an admission BG
level $ 200 mg/dL, there was no observed impact on glycemic control in the intervention period for any of the target ranges. The intervention was associated with a significant increase in hypoglycemic episodes (OR, 2.93; 95% CI, 1.02–8.40) in patients admitted with a BG level . 300 mg/dL. In the multivariate analysis, when the admission BG category and unstable social history were controlled for, patients in the intervention group had significantly lower odds of being outside the glucose range of 70 to 250 mg/dL compared with the baseline group (OR, 0.59; 95% CI, 0.38–0.91; P = 0.02; Table 4).
Discussion
The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values (within 70–250 mg/dL in a given hospital day) compared to usual care. Follow-up
104
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 104
1/31/14 5:19 PM
Pharmacy-Based Glucose Management Program
Table 2. Odds of Poor Glycemic Control Compared With Baseline Characteristics of the Overall Study Populationa
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Baseline characteristics Age, y , 54 54–69 $ 70 Race African American White Other DM diagnosis No DM diagnosis T2DM T1DM Admission BG level, mg/dL , 200 mg/dL 200–300 mg/dL . 300 mg/dL Admission HbA1c level ,7% $7% Home DM regimen None documented Oral medication Insulin Both insulin and oral medication Admitted for DM complications Admitted for hyperglycemic episode Documented current oral steroid use History of psychiatric diagnosis History of medication nonadherence Unstable social, job, insurance status History of illicit drug use
Patient days, n (%)
BG concentration ranges, mg/dL 70–180
70–250
. 180
. 250
, 70
442 (34) 467 (36) 389 (30)
1.0b 0.9 (0.6–1.4) 0.8 (0.5–1.2)
1.0b 0.6 (0.4–0.9)c 0.5 (0.3–0.8)c
1.0b 1.0 (0.6–1.5) 0.8 (0.5–1.3)
1.0b 0.6 (0.4–1.0) 0.5 (0.3–0.8)c
1.0b 0.4 (0.2–0.9)c 0.4 (0.2–0.9)c
814 (63) 417 (32) 67 (5)
1.0b 1.2 (0.9–1.8) 0.9 (0.4–2.0)
1.0b 1.0 (0.6–1.4) 0.7 (0.3–1.7)
1.0b 1.2 (0.8–1.8) 1.0 (0.4–2.4)
1.0b 1.0 (0.7–1.6) 0.8 (0.3–2.1)
1.0b 0.5 (0.2–1.0) No events observed
425 (33) 733 (56) 140 (11)
1.0b 4.5 (3.1–6.5)c 13.9 (6.5–26.1)c
1.0b 4.3 (2.6–7.1)c 9.6 (4.9–18.6)c
1.0b 5.1 (3.4–7.5)c 15.0 (7.2–31.0)c
1.0b 6.7 (3.5–12.7)c 13.3 (6.0–29.4)c
1.0b 0.8 (0.4–1.7) 3.0 (1.3–7.0)c
800 (62) 188 (14) 310 (24)
1.0b 5.3 (3.1–9.0)c 4.4 (2.8–6.7)c
1.0b 5.2 (3.2–8.5)c 7.2 (4.7–11.0)c
1.0b 5.6 (3.2–9.8)c 4.7 (3.0–7.3)c
1.0b 6.8 (3.9–12.1)c 9.1 (5.5–15.1)c
1.0b 0.9 (0.3–2.5) 2.3 (1.2–4.5)c
511 (52) 472 (48)
1.0b 4.5 (3.0–6.7)c
1.0b 5.2 (3.5–7.9)c
1.0b 4.8 (3.2–7.3)c
1.0b 6.4 (4.0–10.1)c
1.0b 1.2 (0.6–2.4)
552 (43) 167 (13) 487 (37) 92 (7)
1.0b 5.4 (3.2–9.1)c 5.2 (3.7–7.7)c 8.8 (4.1–18.7)c
1.0b 4.6 (2.6–8.2)c 5.0 (3.2–7.8)c 4.9 (2.4–10.0)c
1.0b 6.6 (3.8–11.4)c 5.7 (3.8–8.5)c 10.1 (4.6–22.2)c
1.0b 7.2 (3.7–14.2)c 6.3 (3.6–10.9)c 6.8 (3.0–15.5)c
1.0b No events observed 1.9 (1.0–3.7) 0.3 (0.0–3.0)
164 (13)
5.6 (2.9–10.7)c
6.9 (4.1–11.6)c
4.8 (2.5–9.0)c
5.8 (3.4–10.0)c
4.9 (2.5–12.8)c
109 (8)
3.2 (1.6–6.4)c
3.3 (1.8–6.0)c
3.6 (1.7–7.3)c
3.6 (1.9–6.8)c
1.1 (0.4–3.3)
210 (16)
0.7 (0.4–1.2)
0.6 (0.3–1.0)
0.8 (0.5–1.3)
0.7 (0.4–1.3)
No events observed
392 (30)
1.0 (0.7–1.4)
1.1 (0.7–1.6)
1.1 (0.7–1.6)
1.2 (0.8–1.8)
0.5 (0.2–1.1)
244 (19)
1.3 (0.8–2.0)
1.7 (1.1–2.6)c
1.3 (0.8–2.1)
1.6 (1.0–2.7)
2.1 (1.1–4.2)c
275 (21)
1.7 (1.1–2.5)c
2.0 (1.3–3.0)c
1.8 (1.2–2.8)c
2.1 (1.3–3.3)c
1.8 (0.9–3.5)
273 (21)
1.5 (1.0–2.3)
2.2 (1.4–3.3)c
1.5 (1.0–2.3)
2.0 (1.3–3.3)c
2.6 (1.4–4.9)c
Bivariate analyses used generalized estimating equations to compare variables. Listed value is for reference. c Indicates significant result of P , 0.05. Abbreviations: BG, blood glucose; DM, diabetes mellitus; HbA1c, glycated hemoglobin; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus. a
b
analyses showed that a significant benefit in glycemic control was achieved for patients whose admission BG level was , 200 mg/dL. For the overall population there was no difference in rates of hypoglycemia between the study groups; however, for patients admitted with a BG level . 300 mg/dL there was an increase in hypoglycemia during the intervention period. Hypoglycemia can have severe implications if left unmanaged, although the overall percentage of patients with hypoglycemia was small (4% in the control group and 5% in the intervention group).
Results of this study suggest that our pharmacy-based program was not designed to optimally meet the needs of our patient population as a whole. These conclusions caused us to think more carefully about our approach to pharmacy practice model development. We noted a tendency to prioritize the most complicated, poorly controlled patients into our care model. These patients often have multiple factors contributing to their uncontrolled disease states and require more resource-intensive care on a longitudinal basis. Our results suggest that this approach may not maximize overall utility, given that patients with the best glucose
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 105 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 105
1/31/14 5:19 PM
Efird et al
Table 3. Poor Glycemic Control by Study Groupa Out of Rangeb Days meeting criteria with no pharmacy program,c n (%) Days meeting criteria with pharmacy program,d n (%) ORe 95% CI P value
Hyperglycemiab
Hypoglycemiab
, 70 or . 180
, 70 or . 250
. 180
. 250
, 70
, 40
417 (57)
220 (30)
387 (52)
185 (25)
32 (4)
1 (0.1)
292 (52)
175 (31)
273 (49)
155 (28)
29 (5)
1 (0.2)
0.85 0.60–1.20 0.34
1.08 0.88–1.24 0.70
0.90 0.62–1.29 0.56
1.19 0.79–1.79 0.42
1.15 0.62–2.15 0.65
1.32 0.08–20.98 0.84
Multiple definitions of poor glycemic control were used, including being outside a given BG range of 70–250 mg/dL, hyperglycemia, and hypoglycemia. BG values are reported as mg/dL. c N = 738. d N = 560. e The OR was calculated using a general estimated equation and is reported as the odds of meeting poor glycemic control criteria in October relative to the odds of meeting the criteria in July. Abbreviations: BG, blood glucose; OR, odds ratio. a
Downloaded by [Central Michigan University] at 20:01 12 September 2015
b
control and least amount of social and behavioral problems on admission appeared to benefit the most. These results suggest that patient care programs should be designed to address the full array of factors (clinical, social, behavioral, etc) contributing to the patient’s current clinical condition. Additionally, our approach was not designed to address the critical transition of care from the inpatient to outpatient setting. Therefore, the program was not explicitly designed to meet the long-term goals of disease state management. To address these programmatic limitations, future designs should consider coordinating with social services and other clinical disciplines and extending glucose management to the outpatient setting. The extent that improved glycemic control in the inpatient setting translates to the outpatient setting is unknown. Many behavioral, social, and clinical factors may have a greater impact in the outpatient setting due to the less controlled nature of this setting. Peyrot et al14 reported younger age, lower income, higher education, diagnosis of type 2 DM, greater injection burden per day, not following dietary recommendations, injections interfering with daily activities, and pain from and embarrassment about injections as
key correlates of missing prescribed insulin injections. Other studies have shown that better medication adherence, increasing age, lower antihyperglycemic medication regimen intensity (fewer oral agents and no insulin use), better reading ability, and being white were associated with improved HbA1c values in the outpatient setting.15,16 The Diabetes Attitudes, Wishes and Needs (DAWN) study found that diabetes selfmanagement is often poor and is related to psychological problems that providers feel unable to manage sufficiently.17 Many of the factors implicated above were present in our population and were associated with poor glycemic control in the inpatient setting. Because of the increased control in the inpatient setting for many factors, such as medication adherence and nutritional status, BG control may deteriorate once patients are transitioned back to the outpatient setting. There are limitations of our work that should be carefully considered. By assessing 1 group of patients prior to and another group following the implementation of our pharmacy-based program, we used a posttest-only design with nonequivalent groups. Perhaps the most important weakness to this quasi-experimental design is that observed effects may be due to patient selection biases rather than
Table 4. Multivariate Regression of the Odds of Poor Glycemic Controla Variable
β coefficient
ORb
95% CI
P value
Intervention group vs baseline group Initial glucose category , 200 mg/dL 200–300 mg/dL . 300 mg/dL Unstable social, job, insurance status Interaction between group and unstable social history
–0.53
0.59
0.38–0.91
0.02
-1.76 -0.12 1.00c -0.29 -1.58
0.17 0.88 1.00c 0.75 4.85
0.11–0.27 0.52–1.51
, 0.001 0.653
0.41–1.39 1.91–12.32
0.36 0.001
The target range for patient BG value was 70–250 mg/dL. The OR is reported as the odds of being outside the given BG range of 70–250 mg/dL. c Listed value is for reference. Abbreviations: BG, blood glucose; OR, odds ratio. a
b
106
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 106
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Pharmacy-Based Glucose Management Program
the intervention. This could result in subgroups having unbalanced characteristics. We attempted to account for this by using multivariate statistical models that adjust for measured differences between patient groups. However, we acknowledge that these models are limited with respect to unmeasured characteristics and balancing between patient subgroups. Our results should be verified by future work, preferably in the setting of a prospective, parallel group experiment. The baseline month in this study was July, which is the month when new medical staff begins caring for patients. This “cohort turnover,” as it has been described, has been associated with increased mortality and decreased efficiency in care.18 Therefore, we are unsure to what extent our results are confounded by this phenomenon. In both months, providers were ultimately responsible for accepting pharmacists’ recommendations and inputting orders for the antihyperglycemic regimen. Although not the ideal situation, without collaborative practice agreements, this is how many pharmacists in the inpatient setting function due to a lack of prescribing authority. The acceptance of pharmacists’ recommendations was not measured, as this would have been very resource and time intensive. A retrospective review of pharmacy documentation, provider order entry, and nursing documentation in the EMR of the dose of medication recommended, ordered, and given would be required to ensure these pharmacy recommendations were accepted by the medical team. In addition, pharmacists were present on rounds only during weekdays; therefore, recommendations were not made during weekends. Another limitation is that baseline characteristics were retrospectively abstracted from admission notes. Even though our independent reviewers came to consensus regarding patients’ baseline characteristics, some information regarding these variables could have been missing or misreported in patient medical records, potentially introducing bias. A significantly higher proportion of patients in the pharmacy group had a history of medication nonadherence at baseline. This population during the pharmacy intervention may have had poorer glycemic control prior to admission, resulting in more adjustments needed during the inpatient stay. We adjusted for this using days out of range by patients who met inclusion criteria in each group as the endpoint. Medication nonadherence in the overall patient population proved to be significant in the bivariate analyses (Table 2) for the odds of poor glycemic control when BG ranges were between 70 to 250 mg/dL and , 70 mg/dL. In the multivariate regression (Table 4), this variable did
not prove to be significant for the odds of poor glycemic control using the target range of 70 to 250 mg/dL between the 2 groups.
Conclusion
The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values compared with usual care. Our analysis identified a subpopulation of patients with admission BG levels , 200 mg/dL who experienced significant improvement in glycemic control associated with the pharmacy-based glucose management program. The remaining patients, who were generally more complicated from a disease state and social perspective, experienced no improvement based on our analysis. These results caused us to think more carefully about the structure and operations of the pharmacy contribution to our inpatient glucose management program for complex patients and our pharmacy practice model in general. We have evidence that some patients benefit significantly from a pharmacy-based model, whereas other patients likely require a more intense, multidisciplinary approach that integrates endocrine-specialized consultative care with social and behavioral interventions to better match the constellation of factors responsible for patient conditions.
Acknowledgments
The authors thank Vi Gilmore, Kim Durand, Matthew Scholenberg, Chris Dooley, and Eric Vogan.
Conflict of Interest Statement
Leigh E. Efird, PharmD, Sherita H. Golden, MD, MHS, Kanizeh Visram, BS, and Kenneth Shermock, PharmD, PhD, have no conflicts of interest to declare.
References 1. Centers for Disease Control and Prevention. National Diabetes Fact Sheet, 2011. http://www.cdc.gov/diabetes/pubs/pdf/ndfs_2011.pdf. Accessed November 1, 2012. 2. Centers for Disease Control and Prevention. Crude and age-adjusted incidence of diagnosed diabetes per 1,000 population aged 18–79 years, United States, 1980–2011. http://www.cdc.gov/diabetes/statistics/ incidence/fig2.htm. Accessed November 1, 2012. 3. The Centers for Medicare and Medicaid Services. Guide to quality performance standards for accountable care organizations starting in 2012: pay for reporting and pay for performance. http://www.cms.gov/ Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/ Downloads/ACO-Guide-Quality-Performance-2012.PDF. Accessed December 31, 2012. 4. The Joint Commission. Advanced certification in inpatient diabetes. http://www.jointcommission.org/certification/inpatient_diabetes.aspx. Accessed December 31, 2012.
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 107 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 107
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Efird et al 5. Garber AJ, Moghissi ES, Bransome ED Jr, et al; American College of Endocrinology Task Force on Inpatient Diabetes and Metabolic Control. American College of Endocrinology position statement on inpatient diabetes and metabolic control. Endocr Pract. 2004;10(Suppl 2):4–9. 6. Umpierrez GE, Isaacs SD, Bazargan N, You X, Taler LM, Kitabchi AE. Hyperglycemia: an independent marker of in-hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002; 87(3):978–982. 7. Schnipper JL, Barsky EE, Shaykevich S, Fitzmaurice G, Pendergrass ML. Inpatient management of diabetes and hyperglycemia among general medicine patients at a large teaching hospital. J Hosp Med. 2006;1(3): 145–150. 8. McAlister FA, Majumdar SR, Blitz S, Rowe BH, Romney J, Marrie TJ. The relation between hyperglycemia and outcomes in 2,471 patients admitted to the hospital with community-acquired pneumonia. Diabetes Care. 2005;28(4):810–815. 9. Baker EH, Janaway CH, Philips BJ, et al. Hyperglycemia is associated with poor outcomes inpatients admitted to hospital with acute exacerbations of chronic obstructive pulmonary disease. Thorax. 2006;61(4):284–289. 10. American Diabetes Association. Standards of medical care in d iabetes— 2012. Diabetes Care. 2012;35(Suppl 1):S11–S63. 11. Munoz M, Pronovost P, Dintzis J, et al. Implementing and evaluating a multicomponent inpatient diabetes management program: putting research into practice. Jt Comm J Qual Patient Saf. 2012;38(5):195–206.
12. Umpierrez GE, Hellman R, Korytkowski MT, et al. Management of hyperglycemia in hospitalized patients in non-critical care setting: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38. 13. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):13–22. 14. Peyrot M, Rubin R, Kruger DF, Travis LB. Correlates of insulin injection omission. Diabetes Care. 2010;33(2):240–245. 15. Schectman JM, Nadkarni MM, Voss JD. The association between diabetes metabolic control and drug adherence in an indigent p opulation. Diabetes Care. 2002;25(6):1015–1021. 16. Odegard PS, Gray SL. Barriers to medication adherence in poorly controlled diabetes mellitus. Diabetes Educator. 2008;34(4):692–697. 17. Peyrot M, Rubin RR, Lauritzen T, Snoek FJ, Matthews DR, Skovlund SE. Psychosocial problems and barriers to improved diabetes management: results of the cross-national diabetes attitudes, wishes and needs (DAWN) study. Diabetic Med. 2005;22(10):1379–1385. 18. Young JQ, Ranjl SR, Wachter RM, Lee CM, Niehaus B, Auerbach AD. “July effect”: impact of the academic year-end changeover on patient outcomes. Ann Intern Med. 2011;155(5):309–315.
108
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 108
1/31/14 5:19 PM
ISSN: 2154-8331 (Print) 2377-1003 (Online) Journal homepage: http://www.tandfonline.com/loi/ihop20
Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population Leigh E. Efird PharmD, Sherita H. Golden MD, MHS, Kanizeh Visram BS & Kenneth Shermock PharmD, PhD To cite this article: Leigh E. Efird PharmD, Sherita H. Golden MD, MHS, Kanizeh Visram BS & Kenneth Shermock PharmD, PhD (2014) Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population, Hospital Practice, 42:1, 101-108 To link to this article: http://dx.doi.org/10.3810/hp.2014.02.1097
Published online: 13 Mar 2015.
Submit your article to this journal
Article views: 3
View related articles
View Crossmark data
Citing articles: 1 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ihop20 Download by: [Central Michigan University]
Date: 12 September 2015, At: 20:01
H e a lt h O u t c o m e s
Impact of a Pharmacy-Based Glucose Management Program on Glycemic Control in an Inpatient General Medicine Population
Downloaded by [Central Michigan University] at 20:01 12 September 2015
DOI: 10.3810/hp.2014.02.1097
Leigh E. Efird, PharmD 1 Sherita H. Golden, MD, MHS 2 Kanizeh Visram, BS 3 Kenneth Shermock, PharmD, PhD 4 1 Clinical Pharmacy Specialist, Internal Medicine, Department of Pharmacy, The Johns Hopkins Hospital, Baltimore, MD; 2Associate Professor of Medicine and Epidemiology, The Johns Hopkins University School of Medicine and Bloomberg School of Public Health, Baltimore, MD; 3 San Francisco School of Pharmacy, University of California, San Francisco, CA; 4Director, Center for Medication Quality and Outcomes, Division of General Internal Medicine, The Johns Hopkins Hospital, The Johns Hopkins University School of Medicine, Baltimore, MD; Department of Epidemiology, The Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD
Abstract
Objective: A pharmacy-based glucose management program was evaluated to determine whether improved glycemic control could be achieved in an inpatient general medicine patient population. Methods: A retrospective chart review of 151 patients with blood glucose (BG) values outside the range of 70 to 180 mg/dL within a 24-hour period was conducted. Observations for the baseline group with no pharmacy program in place were collected from admissions during July 2010 and for the intervention group in October 2010. The primary goal of the study was to determine if the pharmacy program improved patient days within the BG range of 70 to 250 mg/dL. The odds of poor glycemic control for patients in the intervention versus baseline groups were assessed by multivariate generalized estimating equations. These methods were also used to assess patient characteristics associated with poor glycemic control. Results: No evidence was observed that the pharmacy program decreased the number of days spent out of the targeted blood glucose range (70–250 mg/dL; odds ratio, 1.08; 95% CI, 0.88–1.24). However, the subgroup of patients whose admission BG was , 200 mg/dL (49% of the intervention group) experienced a significant reduction in days spent out of the BG range (70–250 mg/dL; odds ratio, 0.42; 95% CI, 0.22–0.82). No improvement in glycemic control was observed in patients with an admission BG $ 200 mg/dL; these patients had more disease- and social-related factors associated with poor glycemic control. Conclusion: The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values compared with usual care. Patients whose admission glucose was , 200 mg/dL experienced improvement in glycemic control in the pharmacy-based program. Maintaining the BG level of the remaining patients was generally more complicated from a disease-state and social perspective and patients experienced no improvement. These patients may require a more intense, multidisciplinary approach that is better matched to the constellation of factors responsible for their condition. Keywords: glucose management; pharmacy-based; inpatient; general medicine
Introduction
Correspondence: Leigh E. Efird, PharmD, The Johns Hopkins Hospital, 600 North Wolfe Street, Carnegie 180, Baltimore, MD 21287. Tel: 410-502-5703 Fax: 410-955-0287 E-mail: [email protected]
The diabetes mellitus (DM) epidemic continues to grow in the United States. In 2011, DM was estimated to affect approximately 8.3% of the population, or 25.8 million people, including 7 million undiagnosed cases.1 According to the Centers for Disease Control and Prevention (CDC), the incidence of DM has nearly tripled since 1980.2 Optimal care for this growing population of individuals is evaluated through the Centers for Medicare and Medicaid Services quality performance standards. The Joint Commission, which provides advanced certification for hospitals, also implements standards to improve inpatient glucose management.3,4 The first formal consensus
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 101 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 101
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Efird et al
statement regarding target blood glucose (BG) goals in the non–critically ill patient population was issued in 2004 by the American College of Endocrinology and the American Association of Clinical Endocrinologists.5 Subsequently, the American Diabetes Association (ADA) published guidelines that identified a preprandial BG level , 140 mg/dL and a random BG level , 180 mg/dL as reasonable target levels for inpatients with DM.6 Limited evidence exists to support these BG targets or define patient characteristics for increased risk of poor glycemic control in the inpatient setting. However, data do suggest that newly identified hyperglycemia is associated with increased hospital mortality in non–critically ill patients.6,7 There is evidence that glycemic control during a hospitalization is important to achieve positive patient outcomes.7–9 In the non–critically ill patient population, hyperglycemia in patients without a previous diagnosis of DM has been identified as an independent predictor of hospital mortality when compared to patients with a diagnosis of DM (10% vs 1.7%; P , 0.01).7 There are also data suggesting that patients admitted with community-acquired pneumonia and an admission BG . 200 mg/dL have a higher mortality rate compared with patients with an admission BG # 200 mg/dL (13% vs 9%; P = 0.03).8 Baker et al9 showed a 15% increase in the absolute risk of adverse patient outcomes for every 18 mg/dL increase in BG level when adjusted for age, sex, and DM diagnosis in patients with chronic obstructive pulmonary disease and respiratory infection. Finally, it has also been demonstrated that higher BG values and severe diabetic complications in patients upon admission may be associated with higher mean BG values per patient-day in general medicine patients.10 Hyperglycemia in these patient populations may be due to the stress response associated with the severity of their illness on presentation, which could be associated with higher mortality rates. These results have contributed to the recommendation for standardized approaches in managing hospital inpatients in the non–critically ill patient population.6 Our institution formed a multidisciplinary glucose management committee in 2006 to address many of the concerns regarding inpatient glycemic control.11 In conjunction with this committee, the institution implemented a pharmacy-based glucose management program targeting improved glycemic control in general medicine patients during the month of October 2010. This opportunity allowed pharmacists to make antihyperglycemic regimen recommendations and educate providers at the time the treatment plan was being discussed and designed. The primary objective of the study was to assess the impact of the pharmacy-based
glucose management program on glycemic control in an inpatient general medicine population.
Materials and Methods
All patients admitted to 1 of 4 general medicine teams in July 2010 (baseline group) or October 2010 (intervention group) with $ 2 BG values outside of the range of 70 to 180 mg/dL in any 24-hour period prior to inclusion into the study were included in the data analyses. For the baseline group, there was a clinical pharmacy specialist rounding on only 1 of the 4 general medicine teams consistently to focus on designing medication regimens at the time treatment plans were made. Other routine practices in place during both periods included a hypoglycemia policy, a hospital-wide nursing DM education program, a hyperglycemia policy, and a computerized physician order entry set with recommendations for basal and bolus insulin based on patient characteristics. Prior to the intervention period, a case-based glucose management curriculum was provided to pharmacists to improve their skill and knowledge levels in anticipation of their participation in the program. This curriculum focused on educating the rounding pharmacists on how to appropriately transition patients from an outpatient to an inpatient antihyperglycemic regimen, adjust subcutaneous insulin in the inpatient setting, transition patients off insulin drips, and transition patients from an inpatient to an outpatient antihyperglycemic regimen. Pharmacists were then required to undergo a shadowing phase where recommendations and documentation were reviewed by a clinical pharmacy specialist. Once deemed competent in providing appropriate glucose management recommendations, the pharmacist was assigned an inpatient team to follow during the intervention period. Further details of the program have previously been published.11 During the intervention period, a separate pharmacist was assigned to round on each of 4 medicine teams and provided medication therapy recommendations for patients assigned to those teams on a daily basis Monday through Friday. As a part of providing this service, pharmacists formulated a daily glucose management plan that was communicated to the medical team and documented in the patient’s electronic medical record (EMR). Factors considered in designing the antihyperglycemic regimen included DM diagnosis on admission, insulin needs during the last 24 hours, current nutritional status, glycated hemoglobin (HbA1c) level within the last 3 months, home antihyperglycemic regimen, weight, interacting medications, and renal function. Documentation of the glucose management plan included recommendations
102
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 102
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Pharmacy-Based Glucose Management Program
for basal, nutritional, and correctional insulin as deemed necessary to meet target inpatient BG goals. Entry of orders pertaining to medication regimens was ultimately left to the provider. Other members of the rounding medical team included medicine interns, senior residents, an attending, and a case manager with a nursing background. Baseline patient characteristics were retrospectively abstracted from the EMR by 2 independent reviewers who developed consensus when their assessments were discordant. These included the patients’ age, race, DM diagnosis, admission BG, HbA1c level, home DM regimen, and if the current admission was DM-related (indicated by admission related to DM complications, gastroparesis, diabetic ketoacidosis, hyperosmolar hyperglycemic state, or hyperglycemia). Other patient factors, including the presence of infection on admission, current use of oral steroids, any DSM-IV psychiatric diagnosis, history of medication nonadherence, history of illicit drug use, and unstable social, job, or insurance status (indicated by documentation of unemployment, uninsured status, or homelessness), were also retrospectively abstracted from documentation during the inclusion admission from the EMR. The primary endpoint of the study was the odds of having $ 2 BG values in a given hospital day outside the range of 70 to 250 mg/dL (after the patient met inclusion criteria for the study; ie, the patient already had 2 BG values outside 70–180 mg/dL prior to study entry) for patients in the baseline group compared with the intervention group. Secondary endpoints were the determination of patient characteristics associated with having poor glycemic control and the odds of poor glycemic control in the baseline group compared with the intervention group based on the following criteria: 1) patient BG level outside the range of 70 to 180 mg/dL; 2) patient BG level . 180 mg/dL; 3) patient BG level . 250 mg/dL; 4) patient BG level , 70 mg/dL. Analyses were conducted for the overall groups, and for subgroups of patients based on their admission BG value of , 200 mg/dL, 200 to 300 mg/dL, and . 300 mg/dL. A patient BG value of , 200 mg/dL has been recommended as an appropriate upper limit of the target range in non–critically ill inpatients who are at high risk for developing hypoglycemia, which was the basis of our lower limit in this range.12 Differences in baseline characteristics between the study groups were assessed using the Student’s t test or the Mann-Whitney U test for continuous variables and a Χ 2 or Fisher’s exact test for categorical variables. Bivariate analyses were conducted to identify relationships between each baseline patient characteristic and the odds of experiencing
poor glycemic control. These relationships were assessed statistically at the patient-day level using generalized estimating equations (GEEs) to assess the odds of poor glycemic control on a given day. The GEE techniques account for correlation of response variables within individuals.13 They provide more robust model estimates for longitudinal outcomes, such as the odds of a given patient experiencing poor glycemic control on a given day. For these GEE models, a binomial distribution, logit link function, and autoregressive correlation structure were specified. Multivariate GEE techniques were used to assess the relationship between patient characteristics and the rate of poor glycemic control. Multivariate GEE models were also used to assess differences in glycemic control between the baseline and intervention groups. Final model selection used a forward selection process, starting with the most significant variable from bivariate analysis and included variables that achieved a significance level of P , 0.1 or lower in bivariate analyses. This model building process was guided by Akaike’s Information Criterion. The Χ 2 test for comparison of proportions was used to assess binary outcomes that were not longitudinal. Odds ratios (ORs) and 95% CIs were also calculated for these analyses. A P value , 0.05 was considered statistically significant for all analyses. Analyses were conducted using STATA (Version 11.0). The Johns Hopkins Medicine Institutional Review Board approved this study.
Results
A total of 151 patients, 84 (738 patient-days) in the baseline group and 67 (560 patient-days) in the intervention group, were included in the study. At baseline, the majority of patients were African American and had a previous diagnosis of type 2 DM (Table 1). Between study groups, patients had a similar distribution of baseline HbA1c values and home treatment regimens. The only significant difference detected in baseline characteristics was a higher percent of patients with a history of medication nonadherence in the intervention group compared with the baseline group (27% vs 11%; P = 0.01). Irrespective of study period, a strong majority of patients (91%; n = 137) met the criteria for being outside the range of 70 to 180 mg/dL for $ 1 day during their inclusion in the study. On average, patients were outside the range of 70 to 180 mg/dL for 60% of their hospitalized days (median = 67%). Seventy percent (n = 106) of patients were outside the range of 70 to 250 mg/dL on $ 1 day of their hospitalization. The average and median percent of days (36% and 33%, respectively) spent outside this range were similar. Eighteen percent (n = 27) of patients had a
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 103 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 103
1/31/14 5:19 PM
Efird et al
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Table 1. Demographics and Baseline Characteristics of Study Groups Variable
No pharmacy program (n = 84)
Pharmacy program (n = 67)
P valuea
Age (y), mean (SD) Race, n (%) African American White Other DM diagnosis, n (%) No DM T1DM T2DM Admission BG level (mg/dL), mean (SD) Admission BG category (mg/dL), n (%) , 200 200–300 . 300 HbA1c level, mean (SD) HbA1c category, n (%) ,7 $7 No baseline measure Home DM regimen, n (%) Insulin Oral medication Insulin and oral medication None documented Admitted for DM complications, n (%) Admitted for hyperglycemic episode, n (%) Documented infection on admission, n (%) Documented current oral steroid use, n (%) History of psychiatric diagnosis, n (%) History of medication nonadherence, n (%) Unstable social, job, insurance status, n (%) History of illicit drug use, n (%)
60 (15)
62 (16)
0.37
57 (68) 22 (26) 5 (6)
41 (61) 24 (36) 2 (3)
0.40
28 (33) 10 (12) 46 (55) 228 (171)
20 (30) 5 (7) 42 (63) 283 (260)
0.54
49 (58) 15 (18) 20 (24) 8.2 (2.8)
33 (49) 15 (22) 19 (28) 8.7 (3.1)
0.53
27 (32) 36 (43) 21 (25)
21 (31) 32 (48) 14 (21)
0.79
32 (38) 15 (18) 5 (6) 32 (38) 11 (13) 11 (13) 23 (27) 14 (17) 26 (31) 9 (11) 18 (21) 19 (23)
27 (40) 10 (15) 5 (7) 25 (38) 11 (16) 7 (10) 26 (39) 7 (10) 21 (31) 18 (27) 14 (21) 12 (18)
0.95
0.12
0.31
0.56 0.62 0.16 0.27 0.96 0.01 0.94 0.48
a P value is reported for each characteristic as a whole unless otherwise stated. Abbreviations: BG, blood glucose; DM, diabetes mellitus; HbA1c, glycated hemoglobin; SD, standard deviation; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
BG level , 70 mg/dL and only 2 patients (1.3%) had a BG level , 40 mg/dL during the inpatient stay. Bivariate analysis identified significant associations between several patient characteristics and poor glycemic control (irrespective of study group), including known DM diagnosis on admission, admission BG values of $ 200 mg/dL, baseline HbA1c level $ 7%, admission for a DM-related complication, history of medication nonadherence, and unstable social, job, or insurance status (Table 2). There was no overall difference in glycemic control between the baseline and intervention groups (Table 3). However, in the subgroup of patients with an admission BG level , 200 mg/dL, the intervention was associated with a significant decrease in the odds of poor glycemic control for the BG ranges of 70 to 180 mg/dL (OR, 0.59; 95% CI, 0.38–0.91) and 70–250 mg/dL (OR, 0.42; 95% CI, 0.22–0.82). For patients with an admission BG
level $ 200 mg/dL, there was no observed impact on glycemic control in the intervention period for any of the target ranges. The intervention was associated with a significant increase in hypoglycemic episodes (OR, 2.93; 95% CI, 1.02–8.40) in patients admitted with a BG level . 300 mg/dL. In the multivariate analysis, when the admission BG category and unstable social history were controlled for, patients in the intervention group had significantly lower odds of being outside the glucose range of 70 to 250 mg/dL compared with the baseline group (OR, 0.59; 95% CI, 0.38–0.91; P = 0.02; Table 4).
Discussion
The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values (within 70–250 mg/dL in a given hospital day) compared to usual care. Follow-up
104
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 104
1/31/14 5:19 PM
Pharmacy-Based Glucose Management Program
Table 2. Odds of Poor Glycemic Control Compared With Baseline Characteristics of the Overall Study Populationa
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Baseline characteristics Age, y , 54 54–69 $ 70 Race African American White Other DM diagnosis No DM diagnosis T2DM T1DM Admission BG level, mg/dL , 200 mg/dL 200–300 mg/dL . 300 mg/dL Admission HbA1c level ,7% $7% Home DM regimen None documented Oral medication Insulin Both insulin and oral medication Admitted for DM complications Admitted for hyperglycemic episode Documented current oral steroid use History of psychiatric diagnosis History of medication nonadherence Unstable social, job, insurance status History of illicit drug use
Patient days, n (%)
BG concentration ranges, mg/dL 70–180
70–250
. 180
. 250
, 70
442 (34) 467 (36) 389 (30)
1.0b 0.9 (0.6–1.4) 0.8 (0.5–1.2)
1.0b 0.6 (0.4–0.9)c 0.5 (0.3–0.8)c
1.0b 1.0 (0.6–1.5) 0.8 (0.5–1.3)
1.0b 0.6 (0.4–1.0) 0.5 (0.3–0.8)c
1.0b 0.4 (0.2–0.9)c 0.4 (0.2–0.9)c
814 (63) 417 (32) 67 (5)
1.0b 1.2 (0.9–1.8) 0.9 (0.4–2.0)
1.0b 1.0 (0.6–1.4) 0.7 (0.3–1.7)
1.0b 1.2 (0.8–1.8) 1.0 (0.4–2.4)
1.0b 1.0 (0.7–1.6) 0.8 (0.3–2.1)
1.0b 0.5 (0.2–1.0) No events observed
425 (33) 733 (56) 140 (11)
1.0b 4.5 (3.1–6.5)c 13.9 (6.5–26.1)c
1.0b 4.3 (2.6–7.1)c 9.6 (4.9–18.6)c
1.0b 5.1 (3.4–7.5)c 15.0 (7.2–31.0)c
1.0b 6.7 (3.5–12.7)c 13.3 (6.0–29.4)c
1.0b 0.8 (0.4–1.7) 3.0 (1.3–7.0)c
800 (62) 188 (14) 310 (24)
1.0b 5.3 (3.1–9.0)c 4.4 (2.8–6.7)c
1.0b 5.2 (3.2–8.5)c 7.2 (4.7–11.0)c
1.0b 5.6 (3.2–9.8)c 4.7 (3.0–7.3)c
1.0b 6.8 (3.9–12.1)c 9.1 (5.5–15.1)c
1.0b 0.9 (0.3–2.5) 2.3 (1.2–4.5)c
511 (52) 472 (48)
1.0b 4.5 (3.0–6.7)c
1.0b 5.2 (3.5–7.9)c
1.0b 4.8 (3.2–7.3)c
1.0b 6.4 (4.0–10.1)c
1.0b 1.2 (0.6–2.4)
552 (43) 167 (13) 487 (37) 92 (7)
1.0b 5.4 (3.2–9.1)c 5.2 (3.7–7.7)c 8.8 (4.1–18.7)c
1.0b 4.6 (2.6–8.2)c 5.0 (3.2–7.8)c 4.9 (2.4–10.0)c
1.0b 6.6 (3.8–11.4)c 5.7 (3.8–8.5)c 10.1 (4.6–22.2)c
1.0b 7.2 (3.7–14.2)c 6.3 (3.6–10.9)c 6.8 (3.0–15.5)c
1.0b No events observed 1.9 (1.0–3.7) 0.3 (0.0–3.0)
164 (13)
5.6 (2.9–10.7)c
6.9 (4.1–11.6)c
4.8 (2.5–9.0)c
5.8 (3.4–10.0)c
4.9 (2.5–12.8)c
109 (8)
3.2 (1.6–6.4)c
3.3 (1.8–6.0)c
3.6 (1.7–7.3)c
3.6 (1.9–6.8)c
1.1 (0.4–3.3)
210 (16)
0.7 (0.4–1.2)
0.6 (0.3–1.0)
0.8 (0.5–1.3)
0.7 (0.4–1.3)
No events observed
392 (30)
1.0 (0.7–1.4)
1.1 (0.7–1.6)
1.1 (0.7–1.6)
1.2 (0.8–1.8)
0.5 (0.2–1.1)
244 (19)
1.3 (0.8–2.0)
1.7 (1.1–2.6)c
1.3 (0.8–2.1)
1.6 (1.0–2.7)
2.1 (1.1–4.2)c
275 (21)
1.7 (1.1–2.5)c
2.0 (1.3–3.0)c
1.8 (1.2–2.8)c
2.1 (1.3–3.3)c
1.8 (0.9–3.5)
273 (21)
1.5 (1.0–2.3)
2.2 (1.4–3.3)c
1.5 (1.0–2.3)
2.0 (1.3–3.3)c
2.6 (1.4–4.9)c
Bivariate analyses used generalized estimating equations to compare variables. Listed value is for reference. c Indicates significant result of P , 0.05. Abbreviations: BG, blood glucose; DM, diabetes mellitus; HbA1c, glycated hemoglobin; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus. a
b
analyses showed that a significant benefit in glycemic control was achieved for patients whose admission BG level was , 200 mg/dL. For the overall population there was no difference in rates of hypoglycemia between the study groups; however, for patients admitted with a BG level . 300 mg/dL there was an increase in hypoglycemia during the intervention period. Hypoglycemia can have severe implications if left unmanaged, although the overall percentage of patients with hypoglycemia was small (4% in the control group and 5% in the intervention group).
Results of this study suggest that our pharmacy-based program was not designed to optimally meet the needs of our patient population as a whole. These conclusions caused us to think more carefully about our approach to pharmacy practice model development. We noted a tendency to prioritize the most complicated, poorly controlled patients into our care model. These patients often have multiple factors contributing to their uncontrolled disease states and require more resource-intensive care on a longitudinal basis. Our results suggest that this approach may not maximize overall utility, given that patients with the best glucose
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 105 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 105
1/31/14 5:19 PM
Efird et al
Table 3. Poor Glycemic Control by Study Groupa Out of Rangeb Days meeting criteria with no pharmacy program,c n (%) Days meeting criteria with pharmacy program,d n (%) ORe 95% CI P value
Hyperglycemiab
Hypoglycemiab
, 70 or . 180
, 70 or . 250
. 180
. 250
, 70
, 40
417 (57)
220 (30)
387 (52)
185 (25)
32 (4)
1 (0.1)
292 (52)
175 (31)
273 (49)
155 (28)
29 (5)
1 (0.2)
0.85 0.60–1.20 0.34
1.08 0.88–1.24 0.70
0.90 0.62–1.29 0.56
1.19 0.79–1.79 0.42
1.15 0.62–2.15 0.65
1.32 0.08–20.98 0.84
Multiple definitions of poor glycemic control were used, including being outside a given BG range of 70–250 mg/dL, hyperglycemia, and hypoglycemia. BG values are reported as mg/dL. c N = 738. d N = 560. e The OR was calculated using a general estimated equation and is reported as the odds of meeting poor glycemic control criteria in October relative to the odds of meeting the criteria in July. Abbreviations: BG, blood glucose; OR, odds ratio. a
Downloaded by [Central Michigan University] at 20:01 12 September 2015
b
control and least amount of social and behavioral problems on admission appeared to benefit the most. These results suggest that patient care programs should be designed to address the full array of factors (clinical, social, behavioral, etc) contributing to the patient’s current clinical condition. Additionally, our approach was not designed to address the critical transition of care from the inpatient to outpatient setting. Therefore, the program was not explicitly designed to meet the long-term goals of disease state management. To address these programmatic limitations, future designs should consider coordinating with social services and other clinical disciplines and extending glucose management to the outpatient setting. The extent that improved glycemic control in the inpatient setting translates to the outpatient setting is unknown. Many behavioral, social, and clinical factors may have a greater impact in the outpatient setting due to the less controlled nature of this setting. Peyrot et al14 reported younger age, lower income, higher education, diagnosis of type 2 DM, greater injection burden per day, not following dietary recommendations, injections interfering with daily activities, and pain from and embarrassment about injections as
key correlates of missing prescribed insulin injections. Other studies have shown that better medication adherence, increasing age, lower antihyperglycemic medication regimen intensity (fewer oral agents and no insulin use), better reading ability, and being white were associated with improved HbA1c values in the outpatient setting.15,16 The Diabetes Attitudes, Wishes and Needs (DAWN) study found that diabetes selfmanagement is often poor and is related to psychological problems that providers feel unable to manage sufficiently.17 Many of the factors implicated above were present in our population and were associated with poor glycemic control in the inpatient setting. Because of the increased control in the inpatient setting for many factors, such as medication adherence and nutritional status, BG control may deteriorate once patients are transitioned back to the outpatient setting. There are limitations of our work that should be carefully considered. By assessing 1 group of patients prior to and another group following the implementation of our pharmacy-based program, we used a posttest-only design with nonequivalent groups. Perhaps the most important weakness to this quasi-experimental design is that observed effects may be due to patient selection biases rather than
Table 4. Multivariate Regression of the Odds of Poor Glycemic Controla Variable
β coefficient
ORb
95% CI
P value
Intervention group vs baseline group Initial glucose category , 200 mg/dL 200–300 mg/dL . 300 mg/dL Unstable social, job, insurance status Interaction between group and unstable social history
–0.53
0.59
0.38–0.91
0.02
-1.76 -0.12 1.00c -0.29 -1.58
0.17 0.88 1.00c 0.75 4.85
0.11–0.27 0.52–1.51
, 0.001 0.653
0.41–1.39 1.91–12.32
0.36 0.001
The target range for patient BG value was 70–250 mg/dL. The OR is reported as the odds of being outside the given BG range of 70–250 mg/dL. c Listed value is for reference. Abbreviations: BG, blood glucose; OR, odds ratio. a
b
106
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 106
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Pharmacy-Based Glucose Management Program
the intervention. This could result in subgroups having unbalanced characteristics. We attempted to account for this by using multivariate statistical models that adjust for measured differences between patient groups. However, we acknowledge that these models are limited with respect to unmeasured characteristics and balancing between patient subgroups. Our results should be verified by future work, preferably in the setting of a prospective, parallel group experiment. The baseline month in this study was July, which is the month when new medical staff begins caring for patients. This “cohort turnover,” as it has been described, has been associated with increased mortality and decreased efficiency in care.18 Therefore, we are unsure to what extent our results are confounded by this phenomenon. In both months, providers were ultimately responsible for accepting pharmacists’ recommendations and inputting orders for the antihyperglycemic regimen. Although not the ideal situation, without collaborative practice agreements, this is how many pharmacists in the inpatient setting function due to a lack of prescribing authority. The acceptance of pharmacists’ recommendations was not measured, as this would have been very resource and time intensive. A retrospective review of pharmacy documentation, provider order entry, and nursing documentation in the EMR of the dose of medication recommended, ordered, and given would be required to ensure these pharmacy recommendations were accepted by the medical team. In addition, pharmacists were present on rounds only during weekdays; therefore, recommendations were not made during weekends. Another limitation is that baseline characteristics were retrospectively abstracted from admission notes. Even though our independent reviewers came to consensus regarding patients’ baseline characteristics, some information regarding these variables could have been missing or misreported in patient medical records, potentially introducing bias. A significantly higher proportion of patients in the pharmacy group had a history of medication nonadherence at baseline. This population during the pharmacy intervention may have had poorer glycemic control prior to admission, resulting in more adjustments needed during the inpatient stay. We adjusted for this using days out of range by patients who met inclusion criteria in each group as the endpoint. Medication nonadherence in the overall patient population proved to be significant in the bivariate analyses (Table 2) for the odds of poor glycemic control when BG ranges were between 70 to 250 mg/dL and , 70 mg/dL. In the multivariate regression (Table 4), this variable did
not prove to be significant for the odds of poor glycemic control using the target range of 70 to 250 mg/dL between the 2 groups.
Conclusion
The primary finding of this analysis was that there was no global benefit of the pharmacy-based glucose management program for improving BG values compared with usual care. Our analysis identified a subpopulation of patients with admission BG levels , 200 mg/dL who experienced significant improvement in glycemic control associated with the pharmacy-based glucose management program. The remaining patients, who were generally more complicated from a disease state and social perspective, experienced no improvement based on our analysis. These results caused us to think more carefully about the structure and operations of the pharmacy contribution to our inpatient glucose management program for complex patients and our pharmacy practice model in general. We have evidence that some patients benefit significantly from a pharmacy-based model, whereas other patients likely require a more intense, multidisciplinary approach that integrates endocrine-specialized consultative care with social and behavioral interventions to better match the constellation of factors responsible for patient conditions.
Acknowledgments
The authors thank Vi Gilmore, Kim Durand, Matthew Scholenberg, Chris Dooley, and Eric Vogan.
Conflict of Interest Statement
Leigh E. Efird, PharmD, Sherita H. Golden, MD, MHS, Kanizeh Visram, BS, and Kenneth Shermock, PharmD, PhD, have no conflicts of interest to declare.
References 1. Centers for Disease Control and Prevention. National Diabetes Fact Sheet, 2011. http://www.cdc.gov/diabetes/pubs/pdf/ndfs_2011.pdf. Accessed November 1, 2012. 2. Centers for Disease Control and Prevention. Crude and age-adjusted incidence of diagnosed diabetes per 1,000 population aged 18–79 years, United States, 1980–2011. http://www.cdc.gov/diabetes/statistics/ incidence/fig2.htm. Accessed November 1, 2012. 3. The Centers for Medicare and Medicaid Services. Guide to quality performance standards for accountable care organizations starting in 2012: pay for reporting and pay for performance. http://www.cms.gov/ Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/ Downloads/ACO-Guide-Quality-Performance-2012.PDF. Accessed December 31, 2012. 4. The Joint Commission. Advanced certification in inpatient diabetes. http://www.jointcommission.org/certification/inpatient_diabetes.aspx. Accessed December 31, 2012.
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 107 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 107
1/31/14 5:19 PM
Downloaded by [Central Michigan University] at 20:01 12 September 2015
Efird et al 5. Garber AJ, Moghissi ES, Bransome ED Jr, et al; American College of Endocrinology Task Force on Inpatient Diabetes and Metabolic Control. American College of Endocrinology position statement on inpatient diabetes and metabolic control. Endocr Pract. 2004;10(Suppl 2):4–9. 6. Umpierrez GE, Isaacs SD, Bazargan N, You X, Taler LM, Kitabchi AE. Hyperglycemia: an independent marker of in-hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002; 87(3):978–982. 7. Schnipper JL, Barsky EE, Shaykevich S, Fitzmaurice G, Pendergrass ML. Inpatient management of diabetes and hyperglycemia among general medicine patients at a large teaching hospital. J Hosp Med. 2006;1(3): 145–150. 8. McAlister FA, Majumdar SR, Blitz S, Rowe BH, Romney J, Marrie TJ. The relation between hyperglycemia and outcomes in 2,471 patients admitted to the hospital with community-acquired pneumonia. Diabetes Care. 2005;28(4):810–815. 9. Baker EH, Janaway CH, Philips BJ, et al. Hyperglycemia is associated with poor outcomes inpatients admitted to hospital with acute exacerbations of chronic obstructive pulmonary disease. Thorax. 2006;61(4):284–289. 10. American Diabetes Association. Standards of medical care in d iabetes— 2012. Diabetes Care. 2012;35(Suppl 1):S11–S63. 11. Munoz M, Pronovost P, Dintzis J, et al. Implementing and evaluating a multicomponent inpatient diabetes management program: putting research into practice. Jt Comm J Qual Patient Saf. 2012;38(5):195–206.
12. Umpierrez GE, Hellman R, Korytkowski MT, et al. Management of hyperglycemia in hospitalized patients in non-critical care setting: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. 2012;97(1):16–38. 13. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):13–22. 14. Peyrot M, Rubin R, Kruger DF, Travis LB. Correlates of insulin injection omission. Diabetes Care. 2010;33(2):240–245. 15. Schectman JM, Nadkarni MM, Voss JD. The association between diabetes metabolic control and drug adherence in an indigent p opulation. Diabetes Care. 2002;25(6):1015–1021. 16. Odegard PS, Gray SL. Barriers to medication adherence in poorly controlled diabetes mellitus. Diabetes Educator. 2008;34(4):692–697. 17. Peyrot M, Rubin RR, Lauritzen T, Snoek FJ, Matthews DR, Skovlund SE. Psychosocial problems and barriers to improved diabetes management: results of the cross-national diabetes attitudes, wishes and needs (DAWN) study. Diabetic Med. 2005;22(10):1379–1385. 18. Young JQ, Ranjl SR, Wachter RM, Lee CM, Niehaus B, Auerbach AD. “July effect”: impact of the academic year-end changeover on patient outcomes. Ann Intern Med. 2011;155(5):309–315.
108
© Hospital Practice, Volume 42, Issue 1, February 2014, ISSN – 2154-8331 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.
12_Efird.indd 108
1/31/14 5:19 PM
