Brief Report

Poor Glycemic control in Diabetic patients seeking care in the ED

Gary Josephsen MD^a,*, Robert Rusnak MD^b,1

^aDepartment of Emergency Medicine, Harbor-UCLA Medical Center, 1000 West Carson Street, Torrance, CA 90509, USA

^bDepartment of Emergency Medicine, Hennepin County Medical Center, 600 Park Avenue, Minneapolis, MN 55415, USA

Received 16 June 2005; revised 17 February 2006; accepted 18 February 2006

Abstract We hypothesized that diabetic patients in the emergency department (ED) have poorer glycemic control than patients seeking care at Primary care clinics. A convenience sample of hemoglobin A1c (HbA1c) values was gathered retrospectively from the ED, Endocrinology, and Family Medicine Clinics. Results were divided into controlled, poorly controlled, and extremely poorly controlled. The only differing pattern of patients (*P b.01) was in the extremely poorly controlled group consisting of 36% of the ED patients (confidence interval [CI], 29.23-42.69; n = 74) vs 18% of the Endocrinology patients (CI, 13.76-22.53; n = 56) and 19% of the Family Medicine patients (CI, 15.98- 22.75; n = 105). A frequency distribution of the ED HbA1c values was bimodal. The first peak represents well-controlled diabetic patients. The second, higher peak comprises a larger number of patients (approximate n = 134, 73% of all 207 ED patients) who have poorer glycemic control. We conclude that a large number of ED diabetic patients have poorer glycemic control than the other clinics. D 2006

Introduction

The emergency department (ED) often serves as a safety net for patients with chronic illnesses, who may be unable to access primary care or specialty clinics in a timely fashion. The result may be a population of ED patients who have poor control of their chronic illness and are at a higher risk for complications. Even more serious is the possibility that the patients may be unaware of their risk. The 2005 annual report from the American Association of Clinical Endo-

Financial Support: This investigation was supported solely by the authors.

* Corresponding author. Tel.: +1 310 740 0865.

E-mail addresses: [email protected] (G. Josephsen)8 [email protected] (R. Rusnak).

¹ Tel.: +1 612 873 5664; fax: +1 612 904 4241.

crinologists revealed that 67% of the nations’ 18 million Type 2 diabetic patients do not have adequate glycemic control. Their report also showed that 84% of those surveyed believed they were in good control. This study was undertaken to determine if patients with a chronic disease, diabetes mellitus, who seek care in the ED have poorer control of their disease than those seen in primary care clinics appropriate for that illness. This investigation represents the first attempt to measure the glycemic control of diabetic patients who seek care in the ED.

It has been well established that poorly controlled diabetes mellitus results in a high rate of microvascular complications (eg, retinopathy, nephropathy, neuropathy, gastropathy, and vasculopathy). Serum glycosylated hemo- globin (HbA1c) is recognized as a standard marker of the long-term control of diabetes mellitus and is easily deter- mined by a peripheral Blood draw. The American Diabetes

0735-6757/$ - see front matter D 2006 doi:10.1016/j.ajem.2006.02.019

Fig. 1 Proportion of patients in each category of glycemic control by location.

Association recommends an HbA1c of less than 7.0 mg. Tight control has been shown to decrease and delay the complications of diabetes [1-3].

In this study, the patients’ control of their diabetes was measured using HbA1c and compared with levels drawn in the Family Medicine Clinic, Endocrinology Clinic, and ED of the same institution. We studied the glucose control of ED patients to try to recognize a population with increased risk of complications. With such information, public health policy and Health care resources may be directed to intervene to decrease the risk of complications in this patient group.

Methods

This is a retrospective cross-sectional study of patients treated in the ED, Endocrinology Clinic, and Family Medicine Clinic of a single institution who had their HbA1c drawn during the period of study. The investigation took place at [blinded] with 416 beds and 342653 annual clinic visits. The ED has 48 beds and 97678 patient visits annually. Prestudy approval with expedited status was obtained from the institutional review board. A search of the databases for the hospital [blinded] revealed values for HbA1c samples drawn from January of 2002 through June of 2003 (18 months). This study did not involve patients who were in DKA or who had a life-threatening illness.

The HbA1c levels from the ED, Endocrinology Clinic, and Family Medicine Clinic were divided into three cate- gories: controlled (HbA1c V7), poorly controlled (HbA1c 7-10), and extremely poorly controlled (HbA1c N10). The

proportion of patients in each group were compared based on location using the Fisher exact test, and average HbA1c values were compared using the Student t test; all results are listed with 95% confidence intervals. Bimodal data were analyzed using previously described methods [4]. Deconvo- lution of these data was done by creating estimates for the mixture of two normal densities based on 1000 bootstrap samples of 201 of the original 207 data points (8 ED patients had an HbA1c level greater than 14 mg%, 7 of these were truncated from the data set to facilitate the deconvolution into two normal distributions). For each bootstrap sample, the penalized EM algorithm was used to calculate the max- imum likelihood estimates for the parameters [4,5]. The final parameter estimates given as modes and 95% confi- dence intervals (CIs) are the mean of these 1000 maximum likelihood estimates. The penalized EM algorithm used prior gamma parameters (a = .4, b = .4). The result is two sets of data with normal distributions. A Kolmogorov-Smirnov goodness-of-Fit test comparing the original data to the mix- ture density determined by these parameter estimates is not significant, indicating that the estimated density matches the data well. The area under each of the curves was used to estimate a number of patients in that group.

Results

Of the three categories controlled (HbA1c V7), poorly controlled (HbA1c 7-10), and extremely poorly controlled (HbA1c N10), the only significant difference (*P b .01) in the proportion of patients by location was in the extremely

Fig. 2 Frequency distribution of HbA1c values from Primary Care Clinics.

poorly controlled group consisting of 36% of the ED patients (CI, 29.23-42.69; n = 74) vs 18% of the Endo- crinology patients (CI, 13.76-22.53; n = 56) and 19% of the Family Medicine patients (CI, 15.98-22.75; n = 105). There was no difference by location in the other categories of control nor between the proportion of patients with any level of control between the two primary care clinics (Fig. 1). The Endocrinology Clinic HbA1c averaged 7.91 (CI, 7.72-8.10; n = 314; median, 7.5). Family Medicine averaged 7.98 (CI,

7.79-8.18; n = 547; median, 7.4) (Fig. 2). Each of these averages differed significantly ( P b .01) from the average ED HbA1c of 9.00 (CI, 8.67-9.36; n = 207; median, 9.0).

The ED data were bimodal (Fig. 3). The first subgroup was estimated to represent approximately 73 of the 207 patients (CI, 72-74), with an average HbA1c of 6.09 (CI, 6.07-6.11). The second group comprised the remaining 134 patients (CI, 133-135), with an average HbA1c of 10.33 (CI, 10.30- 10.35). There was no significant difference for average

Fig. 3 Bimodal distributions of HbA1c levels in the ED after 1000 bootstrap samples of 201 of the original 207 data points (approximated by solid curve A) and resulting normal distributions after deconvolution of bimodal data (broken curves B and C).

HbA1c between patients who did (9.38 [CI, 8.75-10.02]; n = 73) and patients who did not (8.82 [CI, 8.36-9.29]; n = 134) have insurance in the ED.

Discussion

These results suggest that compared with both the Family Medicine and the Endocrinology Clinic patients, the population of diabetic patients seeking care in the ED represents a group of patients with higher average HbA1c values and thus a higher risk of microvascular complications from diabetes [2,6].

Perhaps more interesting than the averages are the comparative distributions of the data drawn from the ED vs data collected elsewhere. Specifically, both the Endocri- nology Clinic and the Family Medicine Clinic data approximate a Gaussian distribution with a heavy right tail indicating a larger distribution of values greater than the average. However, the bimodal ED data had both a peak near the average found in the primary care facilities and another higher peak representing a larger patient group.

The first peak (similar to those treated at a primary care facility) represents a well-controlled diabetic popula- tion. The second, higher peak comprises a larger number of patients (approximate n = 134, 73% of all 208 ED patients) who have poorer control of their diabetes; these patients have a much higher risk of developing microvas- cular complications.

Possible factors related to the poor control of this group include the following: (1) insufficient education regarding their disease; (2) lack of resources (to purchase insulin, meter strips, needles, etc); (3) failure of the current primary care system to accommodate these patients; (4) lack of patients’ appreciation of the importance of tight control;

(5) differences in the appreciation of the serious nature of this illness.

Why is this important for the specialty of EM? The higher HbA1c values for the sample patients in the ED vs samples from primary care clinics suggest a group of poorly controlled diabetic patients seeking care in the ED. This group presented to the ED for a variety of non-acute issues such as those routinely treated in a primary care clinic. One reason for their use of the ED may be that they do not have access to primary care. The observation that their visits are not directly related to their diabetes suggests that the higher prevalence is not due to sampling bias. This group of patients with a poorly controlled chronic disease is at higher risk for microvascular complications. If these patients use the ED in lieu of primary care, then the only contact point for intervention is the ED. Our preliminary results indicate that most diabetic patients presenting to EDs similar to our own have poor control of their disease. Perhaps armed with this knowledge, practitioners will be more likely to

recommend expeditious follow-up care specifically target- ing their diabetes in addition to follow-up related to the presenting complaint.

In addition, current health insurance initiatives that demand higher co-pays by patients might have untoward consequences for this population (inability to purchase glucometer, glucose test strips, insulin, and needles) [7]. This has the potential to lead to deterioration of the long- term health of these patients and a rise in nationwide Health care costs.

Limitations

Weaknesses of this study included its small sample size, retrospective nature, and the potential for selection bias in the pattern of ordering HbA1c levels in ED patients. Also, we did not ascertain each patient’s access to primary care. To facilitate the deconvolution of the bimodal data, we truncated 7 patients with Hba1cs greater than 14; if included, this would result in a larger average HbA1c level and a larger sample size estimate for ED patients with poorly controlled diabetes.

Conclusions

Physicians practicing in an environment similar to that of our study should recognize that many diabetic patients have poor control of their disease and should arrange expeditious follow-up. Identifying patients with poor control early in their disease (and on an ongoing basis) creates opportunities to significantly reduce the Economic burden of this and other chronic diseases.

References

1. Clark Jr MJ, Sterrett JJ, Carson DS. Diabetes guidelines: a summary and comparison of the recommendations of the American Diabetes Association, Veterans Health Administration, and American Associa- tion of Clinical Endocrinologists. Clin Ther 2000;22(8):899 - 910 [discussion 898].
2. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med 1993;329(14):977 - 86.
3. American Diabetes Association. Standards of medical care for patients with diabetes mellitus. Diabetes Care 2001;24:s33 - 43.
4. Schomig A, Kastrai A, Elezi S, et al. Bimodal distribution of angiographic measures of restenosis six months after coronary stent placement. Circulation 1997;96(11):3880 - 7.
5. Ciuperca G, Ridolfi A, Idier J. Penalized maximum likelihood estimator for normal mixtures. Scand J Stat 2003;30:45 - 59.
6. Klein R, Klein BE. Relation of glycemic control to diabetic complications and health outcomes. Diabetes Care 1998;21 (Suppl 3): C39 - 43.
7. Goldman DP, Joyce GF, Escarce JJ, et al. Pharmacy benefits and use of drugs by the chronically ill. JAMA 2004;291(19):2344 - 50.