- Email: [email protected]
Diabetes in urban african americans. III. Management of type II diabetes in a municipal hospital setting
Diabetes in Urban African Americans. III. Management Type II Diabetes in a Municipal Hospital Setting
of
David C. Ziemer, MD, Merilyn G. Goldschmid, MD, Victoria C. Musey, MD, William S. Domin, MD, Peter M. Thuk, MD, Daniel L. Gallina, MD, Lawrence S. Phillips, MD, Atlanta, Georgia
OBJECTIVE: Management of type II diabetes is difficult, particularly in urban populations with limited resources and access to care. To evaluate the effectiveness of structured care delivered by non-physician providers, patients were studied prospectively for 6 months in a municipal hospital diabetes clinic. DESIGN AND METHODS: The population was approximately 90% African American and had median known diabetes duration of approximately 1 year, 54% had incomes below the Federal Poverty Guideline. Primary management was provided by nursepractitioners and dietitians, and primary outcome measures were hemoglobin Ale (HbAlc), fasting plasma glucose, and changes in body weight. RESULTS: Responses were analyzed in 325 new patients returning for visits at 2, 4, 6, and 12 months; metabolic profiles at presentation were similar to those of subjects who missed intervening visits. Lean patients largely continued on pharmacologic therapy and improved HbAlc from 9.4% to 7.4% at 2 months (P < O.OOl), remained stable through 6 months, then rose to 7.9% at 1 year. Obese patients (71%) received dietary instruction. Weaning of pharmacologic therapy was attempted for the first 2 months, resulting in a decline of HbAlc from 9.6% to 8.0% (P < O.OOl), with 70% treated with diet alone. In the obese, HbAlc continued to decrease through 6 months (7.7%). Thereafter, providers saw patients at their own discretion and intensified therapy as needed. Although by 1 year, HbAlc had risen to only 8.2%, some patients required reinstitution of pharmacologic therapy; 59%
From the Division of Endocrinology and Metabolism, Emory University School of Medicine. Atlanta. Georgia. This work was supported in part by research and training awards from the National Institutes of Health, Grants DK-07298, DK-08978, DK33475, and DK-48124. Requests for reprints should be addressed to David C. Ziemer, MD, Department of Medicine, Emory University School of Medicine, 69 Butler Street, SE, Atlanta, Georgia 30303. Manuscript submitted June 20, 1995 and accepted in revised form March 25, 1996.
01996 by Excerpta All rights reserved.
Medica,
Inc.
were on diet alone. While 52% lost 4 lb or more (mean 9.3) by 2 months, little additional weight was lost. Interestingly, glycemic control was improved both in those who lost ~8.5 lb in the first 2 months (HbAlc 9.6% to 8.1% at 12 months), and in those who gained weight (HbAlc 10.2% to 8.2%). In the obese p,atients using pharmacologic agents at presentation, 35% were able to discontinue oral agents or insulin by 1 year, with good glycemic control (HbAlc ~8%). For patients who were initially on diet alone, a fasting plasma glucose >177 mg/ dL predicted the need for pharmacologiic therapy with 97% certainty. CONCLUSIONS: In urban African American patients, nonpharmacologic management of type II diabetes substantially improves metabolic control; decreases in HbAlc are compatrable in those who do and do not lose weight. Therapy managed by nonphysician providers can be an effective cornerstone of diabetes care in this socioeconomically disadvantaged population. Am J Med. 1996;101:25-33. iabetes imposes an enormous health burden on African Americans in the United States. In the II analysis, ’ the age-standardized prevalence of diabetes was 10.2% in African Americans but only 6.6% in Caucasians. African Americans consequently suffer disproportionately from diabetes-related complications.* During the 198Os, diabetes-related, age-standardized mortality decreased 1.6% and 4.5% for Caucasian men and women, respectively, yet mortality increased 11% and 5.5% among AfricanAmerican men and women, respectively.3 Among African-Americans with diabetes, the incidlence and prevalence of hypertension4 and retinopathy5 are higher than in Caucasians. In the African-American diabetic population, the incidences of end-stage renal disease,6 lower extremity amputation related to diabetes, 7 and disability due to diabetes’ exceed those for Caucasians by at least 50%. Increased morbidity of diabetes in African Americans may be due to poor metabolic control. The Diabetes Control and Complications Trial (DCCT) has shown that improved metabolic control reduces the development and progression of microvascular complications such as retinopathy, and suggested that
D NHANES
0002-9343/96/$15.00 PII SOOO2-9343(96)00071-X
25
macrovascular disease may improve as well’; Ohkubo et allo have recently reported similar findings in patients with type II diabetes. However, there has been relatively little examination of metabolic control and treatment strategies targeting African Americans, and particularly the inner-city population. The care of urban African Americans presents special problems, including a limitation of funds for medical care, problems of physical access (transportation, and so forth), an absence of a tradition of healthoriented behavior for optimizing long-term health prospects, patients’ feeling a sense of being powerless to alter health outcomes, and a lack of culturally sensitive programs and educational materials I1 ; our studies reveal that many of these patients already have microalbuminuria on lirst presentation, l2 and that many admissions for diabetic ketoacidosis are due to cessation of insulin therapy.‘” The Grady Health System Diabetes Unit serves an urban, largely African-American population, using a cost effective, multidisciplinary team approach established by John K. Davidson14 following similar approaches elsewhere in the United States.‘5,‘6 In this initial study, we evaluated the impact of a management strategy that emphasizes nonpharmacologic approaches to therapy.
METHODS Each year, about 5,000 patients are seen for follow-up care in the Grady Diabetes Unit. An additional 950 new patients are enrolled annually through a full-day evaluation and education visit; patients may be self-referred or referred by inpatient services or other clinics. To determine the economic status of this population, we reviewed the financial counselling records of 394 consecutive patients new to the Diabetes Unit from May 1,1994 to September 151994. All patients are required to submit evidence of residence and of income to determine eligibility for discounts provided to Fulton and DeKalb residents. For those claiming indigency, income evidence may include pay stubs, tax records, unemployment claims from the Georgia Department of Labor, a statement from their homeless shelter, evaluation by Georgia Department of Family & Children’s Services (for Medicaid and food stamps), or evaluation by the Social Security Administration (for disability or social security pension). Patient income was calculated relative to the Federal Poverty Guidelines (FPL). All new patients undergo an in-depth medical interview with standard format, and physical examination including height and weight in light clothing. Percent ideal body weight is calculated from the Hamwi formula.17 Race is assigned by the examining nurse-provider by observation and direct question26
July
1996
The
American
Journal
of Medicine@
Volume
101
ing. Findings are recorded prospectively using data collection sheets, and the data are entered into a microcomputer registry. Similar but more abbreviated data collection is completed at follow-up visits. Laboratory testing is performed at each visit, and these data are subsequently obtained from the department of pathology and laboratory medicine and entered into the registry. The registry is stored in a relational database (FoxPro; Microsoft, Redmond, Washington), using an interface to ease data entry, written by Emory University Computer Center personnel under the direction of the first author. The Diabetes Unit has its own blood1 chemistry and urine laboratory staffed and supervised by the department of pathology and laboratory medicine. Glucose analyses are performed on venous blood using a glucose oxidase method on an AF’IEC glucose analyzer (APEC, Inc., Danvers, Massachusetts). Hemoglobin Ale was measured using an HPLC method (DIAMAT Biorad, Hercules, California). The assay is corrected for the proportion of hemoglobin A and has a normal range of 3.5% to 6.0%. At the first visit, the diagnosis of diabetes is confirmed using standard American Diabetes Association criteria. la Type II, noninsulin-dependent diabetes mellitus (NIDDM) is defined by historical and treatment features including lack of a history of ketosis, lack of absolute requirement for insulin, presence of obesity and/or strong family hisl;ory of diabetes, and lack of diabetogenic medications or recurrent pancreatitis. Each new patient is assigned for follow-up to one health care provider, a nurse practitioner or registered nurse with special interest and training in diabetes care. All patients attend classes i.n diabetes education at the initial visit, and are seen by an assigned dietitian in a one-on-one setting at this and each of their early return visits. Individual patient treatment plans are developed by health care providers, guided by a uniform protocol. At each visit, patients are also seen by an endocrinologist, who reviews the treatment plan. All patients receive instruction on healthy food choices, and obese patients ( > 120%of ideal body weight) receive instruction on a hypocaloric diet with a deficit of approximately 500 calories per day.lg For obese patients with NIDDM, management guidelines have been designed to encourage treatment with diet and weight loss alone, as follows: l All obese patients receive a diet with deficit 500 k&/day. l If fasting plasma glucose (FPG) is ~19.4 mM (350 mg/dl) , pharmacologic therapy is instituted or increased (if already present). l If FPG < 19.4 mM (350 mg/dl) and there is no clinical evidence of decompensation (eg, orthostatic
hypotension), oral agents are discontinued and insulin is decreased 50% for FPG less than 11.1 mM (ZOO mg/dl) or 25% for FPG of 11.1 to 16.7 mM (200 to 300 mg/dL). l Insulin is discontinued if the daily dose is less than 0.2 units/kg and the FPG is less than 16.7 mM (300 mg/dL) . Visits are scheduled frequently (at 1, 2, and 4 weeks, and at 2, 4, and 6 months) to maximize the educational components of therapy and to avoid metabolic decompensation. Fasting plasma glucose levels are obtained at each visit and HbAlc levels are obtained initially and then bimonthly. Data for the 1Smonth visit were obtained from visits occurring between 42 and 62 weeks after the initial visit; numerical data for these visits were averaged, and categorical data (eg, type of therapy) were taken from the last visit in the time period. Data were analyzed using microcomputer-based data instruments including Excel (Microsoft, Redmond, Washington) and Statview 11 (Abacus Concepts, Berkeley, California), with evaluation by contingency table analysis, Mann-Whitney U (rank) test, Student’s t test, or analysis of variance (ANOVA) with repeated measures. No missing data were imputed. Repeated measures ANOVA for HbAlc could be performed only for those 112 patients with HbAlc values at all time points. These 112 were not different from the larger group of 325 (see below) with respect to age, gender, duration of diabetes, or percent of ideal body weight. Although initial fasting plasma glucose was slightly higher (P < 0.03)) initial HbAlc was not, and those with HbAlc at all points were felt to be representative of the group as a whole.
1994,300 had usable income data (28 were from outside Nton or DeKalb counties, 25 failed to provide data, and 41 had income evaluations pending with state or federal agencies). Of those with1 income data, 54% had incomes below the FPL, 69% were at or below 125% of the FPL, and only 13% had incomes greater than 190% of the FPL (or did not apply for discounted care). From April 15, 1991 to April 15, 1993, the Grady Diabetes Unit enrolled 1,824 new patients, whose characteristics are shown in Table I. There was a predominance of females and African Americans. The median age of patients was 52.5 years and the mean 51.9 years, while median duration of diabetes was approximately 1.2 years, with an avera$e of 5.4 years duration. Although 4 patients were found not to have diabetes, the remaining 1,820 patients were classified as type 1 (5.3%), type 11 (87.4%)) and impaired glucose tolerance ( 1.8%). The remaining 5.5% had diabetes that was secondary or of unclear etiology. For the 1,634 patients with type 11 diabetes, the mean age was 53 years, 59% were female and 41% male, and 88% were African American and 8.8OhCaucasian (Table I). At presentation, the median duration of diabetes was about 1.2 years, with an average of 5.3 years. Patients averaged 142% of ideal body weight, with 88% of the females and 49% of t.he males being above 120% of ideal body weight. For patients with type 11 diabetes, the initial fasting plasma glucose averaged 10.9 mM (197 mg/dL), with HbAlc 9.3%. In order to assess the impact of managem.ent strategies over a 6-month period, we focused on a subpopulation of 325 patients with type 11 diabetes who returned for visits as scheduled, ie, at 2, 4, 6, and 12 RESULTS months. These 325 patients were representative of Patient Characteristics the entire group of 1,634 type 11 subjects, as shown by the lack of a statistically significant difference beThe Grady Diabetes Unit population is composed of adults, 97% of whom reside in either Fulton or tween those 325 patients and the other patients in racial distribution, initial therapy, mean duration of DeKalb County, which contain the city of Atlanta. diabetes, percent of ideal body weight, fasting Fewer than 10% of new patients have commercial health insurance, and approximately 50% have no plasma glucose, or HbAlc. However, patients were more likely to return as scheduled if they were dithird-party health care coverage. Relative to the population of either county or Atlanta, the population of agnosed less than 1 year previously (23% versus 17% return rate, P < O.OOl), older than 55 (27% versus the Diabetes Unit is disproportionately African 14% return rate, P < O.OOl), or female (22% versus American (approximately 88%). In patients who said 17% return rate, P < 0.05). Although 85% of patients they could read, average literacy has been evaluated kept appointments for each subsequent visit, patient as sixth grade level (K. Dobbenstein, personal communication). Thus, the Diabetes Unit population is dropout was high over the first 12 months of followtypical of many served by urban hospitals in the up. By our strict criteria, 80% missed at least one follow-up visit over the la-month period, despite the United States. and There is a high prevalence of homelessness, and availability of frequent routine appointmlents provision of flexibility in scheduling. An informal the population is severely economically disadvansurvey of 30 patients not returning by 6 months retaged. Of the 394 Fulton and DeKalb residents presenting between April 1, 1994 and September 15, vealed that haIf could not be contacted by telephone, July 1996
The American
Journal
of Medicine@
Volume
101
27
TABLE I Grady
Diabetes
Clinic
Type 2 All Patients n = 1,824
Gender (WI Race (%I
Class (%I
Therapy (%I at Presentation
Duration (years) Age (years) Fasting plasma glucose fmM) Fasting plasma glucose fmg/dL) Hemoglobin Ale (%I % lBW-(%, * P 10.03, + P ~0.03, f P ctO.002, 1 P ~0.001,
Female Male Black White Hispanic Other Type 1 Type 2 PancfitOH Other IGT Diet Sulfonylurea Insulin Both Mean Median Mean Mean Mean Mean Mean
Diabetes n = 1,634
57.0 43.0 87.1 9.6 1.4 1.9 5.3 87.4 2.0 2.5 1.8 30.7 22.4 46.4 0.5 5.4 1.2 51.9 11.0 198 9.30 139
58.9 41.1 87.7 8.8 1.4 2.2 100.0 31.5 24.6 43.4 0.6 5.3 1.2 52.9 10.9 197 9.30 142
Type 2 Diabetes Returns n = 325
64.3* 35.7* 88.3 9.0 0.9 1.9 100.0 32.6 25.5 41.5 0.3 4.8+ 0.5t 57.7” 10.7 193 9.30 140
Type
2 Diabetes Returns African-American n = 286
65.4 34.6 100.0 -----
1OCl.O --
-31.8 25.5 42.6 0.0 4.6 0.4 57.5 10.7 192 9.40 141.1
chisquare test, return status versus gender. Mann-Whitney U statistic, Nonreturnees (not shown) versus returnees. chi-square test, return status versus duration -C 1 year. Mann-Whitney U statisttc, nonreturnees (not shown) versus returnees.
with type II diabetes, and were virtually indistinguishable from the non-African Americans in gender distribution, percent of ideal body weiglht, duration of known diabetes, age, and indices of diabetes control or treatment at presentation (Table I).
Glycemic
0
2
4
6
12
0
2
4
6
12
Visit (mo) Figure 1. Metabolic control (HbAlc)for patients returning with HbAlc values at the 2-, 4-, 6-, and 12-month visits. Differences over time are significant for both obese (83) and nonobese (29) patients If < 0.001, ANOVA repeated measures) with no difference or interaction between the two groups. n (obese) = 83; n (non-obese) = 29.
using both hospital records and the local telephone directory. Of those contacted, half stated that they did not return because they “felt well”; only 2 patients were obtaining care from some other source. The 286 African-American patients with type II diabetes comprised 88% of the 325 returning patients 28
July 1996
The American
Journal
of Medicinea
Volume
101
Control
Improvements in glycemic control were seen within 2 months and remained steady through 6 months of follow-up, with a slight rise at 12 months: mean FPG fell by 2 months from 10.9 to 91.6mM (196 to 173 mg/dL) in obese subjects and from 10.7 to 8.6 mM (192 to 154 mg/dL) in nonobese subjects (both P < 0.001). At 1 year, FPG levels were 9.9 and 8.8 mM (179 and 159 mg/dL), respectively. Similarly, HbAlc improved from 9.6%to 8.0%in obese and from 9.4% to 7.4%in nonobese subjects at 2 months (both P < O.OOl), and levels were 7.7% and 7.5%, respectively, at 6 months, and 8.2% and 7.9%, respectively, at 12 months (Figure 1) . The number of patients achieving HbAlc ~8.0% increased by 2 months from 33% to 56% of obese subjects and from 42% to 73% of lean subjects, and fell only slightly thereafter (54% and 62% at 12 months, respectively). Efforts to treat patients with nonpharmacologic approaches were largely successful. At presentation,
NIDDM IN URBAN AFRICAN AMERICANS/ZIEMER
Obese
1
Non-obese
HbAlc
ET AL
Pharmaco-therapy
12-
1-I
‘0°
10
.60
co 80 H
'= s
60
6 8
40 2
20 0
n
Initial
0
2
4
612
0
2
4
612
Visit (mo) Figure 2. Distribution of therapy by visit for all obese (231) and nonobese (94) patients returning at 2, 4, 6, and 12 months. n (obese) = 231; n (non-obese) = 94. k# Insulin; 0 both (OA + Ins); n diet only.
120 100 v)
80
5 'E d
60
-5 8
40 20 0
Diet
Sulfonylurea Initial
Insulin
therapy
Figure 3. Distribution of therapy for obese patients at the 12-month visit versus therapy at presentation (insulin therapy at 12 months includes 4 patients also on sulfonylureas). Number of patients at presentation on diet was 83; sulfonylurea, 56; insulin, 83; and insulin plus sulfonylurea, 1 (not shown). E# Insulin; oral agent; n diet only.
36% of obese subjects were being managed with diet, 24% with sulfonylureas, and 40% with insulin (Figure 2). However, after two months, 70%were being managed with diet alone. Thereafter, there was a small decrease in management with diet alone at 6 months (61%)) and a further decrease by 12 months, when obese patients were managed 55% with diet alone, 26% with sulfonylureas, and 19%with insulin (2% with both sulfonylureas and insulin). As expected, lean patients more often presented on pharmacologic agents (75%), and proportions changed relatively little over 12 months of follow-up. After 12
Initial
12mo.
12mo.
Figure 4. Impact of therapy. Metabolic control (HbAlc) in obese patients improved from initial to 12-month visit while percentage of patients on pharmacologic therapy decreased (n = 83 and n = 230, respectively).
months, 27% were managed with diet, 34% with sulfonylureas, and 36% with insulin. As shown in Figure 3, we found that most obese patients could be managed with diet, regardless of their therapy on presentation. In obese patients who presented on dietary therapy alone, 67% were continued on diet alone after 12 months, with another 29% on sulfonylureas, and only 4% were taking insulin. In obese patients who came to the Diabetes Unit on sulfonylureas, after 12 months, 38% were managed with diet alone and 48% returned .to sulfonylureas, with 14% requiring insulin. Even in obese patients who came to the Diabetes Unit on insulin, 54% were managed with diet alone after 12 months. The effectiveness of nonpharmacologic approaches was particularly striking (Figure 4). At presentation, 231 of the 325 subjects were obese, and of these, 83 had HbAlc measurements that permitted evaluation of responses at 2,4,6, and 12 months. While at presentation average HbAlc was 9.6%, and 64% of patients were managed with pharmacologic agents, 6 months later, average HbAlc was 7.7%, and only 39% of patients required pharmacologic agents. At 12 months, average HbAlc was 8.2%, and 45% required pharmacologic agents. Moreover, nonpharmacologic approaches were also effective in obese patients who presented on sulfonylureas or insulin 47%of 148 obese type II patients who presented on pharmacologic therapy were on diet alone after 12 months, with HbAlc decreasing from 9.6% to 8.1%; HbAlc decreased from 9.6% .to 7.2% in obese patients who presented on diet alon’e. Both of these improvements in HbAlc were high.ly signiflcant (P < 0.001). To date, we have found it difficult to predict either eventual glycemic control or need for pharmacologic therapy. However, 85% of type II patients who presented with evidence of reduced pancreatic insulin July 1996
The American
Journal
of Medicine@
Volume
101 29
reserve as shown by a fasting C-peptide 5 1.O ng/mL were either on pharmacologic therapy or had HbAlc greater than 8.0% 1 year later. Moreover, patients with shorter duration of diabetes (
Weight
July 1996
Journal
of Medicine@
Volume
101
4
6
12
Visit (mo.) Figure 5. Metabolic control (HbAlc) in nonobese patients treated with insulin throughout (n = 9) is better at 1 year than in obese patients whose insulin was decreased and then resumed (n = ;3; P < 0.01 by Student’s t test). n Obese (8);
12 10 ‘;; 5 0 2 P
8 6 4 2 0
The American
2
L
Loss
Obese patients lost an average of 1.6 lb over the l-year follow-up period. Within the llrst 2 months of follow-up, approximately 48% of obese patients achieved weight loss of at least 0.5 lb per week. These “responders” to a hypocaloric diet prescription lost an average of 9.7 lb over the first 2 months, but little further weight loss was achieved after that time. While those losing at least 8 lb over the first 2 months (64 subjects) exhibited improved metabolic control (HbAlc decreased from 9.9%to 8.2%, Figure 6), HbAlc also improved from 11.6% to 7.3% in the 54 subjects who gained weight over the first 2 months (between groups P > 0.5, ANOVA repeated measures for all visits). These improvements in metabolic control were largely sustained through 12 months of follow-up, at 8.0% and 8.2%, respectively. Moreover, there was no difference in the number of patients attaining good glycemic control as measured by HbAlc < 8.0% (P = O.SO), nor did weight loss make a difference in the likelihood of being able 30
0
0
2
4
6
12
Visit (mo.) Figure 6. Metabolic control (HbAlc) in obese patients who lost 8 lb or more (n = 19) by 2 months improved no more than in those who gained weight (n = 16; P > 0.1 by ANOVA, repeated measures). n Lost 28 lb by 2 months;
to discontinue insulin with good metabolic control at 12 months; 45% of those on insulin who lost more than 8.0 lb by 2 months were able to discontinue insulin while maintaining HbAlc < 8.0%, and 58% of those who gained weight were able to discontinue insulin with good control (chi-square test, P =
>0.60). Impact
of Missed
Visits
Examination of patients returning for their 6month visit showed that 31 who did not keep their intervening bimonthly appointments had higher
HbAlc at 6 months than 359 who returned for those appointments (8.9% versus 7.5%, P < O.OOl), although the groups had comparable HbAlc at presentation (Figure 7). cost Our approximately 1,500 patients were enrolled over a period of 2 years, and followed up for an additional 12 months. In 1994, costs including estimates for physician and staff salaries, rent, utilities, and administrative overhead were $1.52 million for 25,974 patient visits per year. Thus, average costs were $59 per patient visit. This estimate encompasses a wide variety of patient contacts, including initial visits that lasted 6 or 7 hours, scheduled follow-up appointments, walk-in visits for old or new patients, and brief visits for prescription refills, and so forth.
missed vis’its all visits
Initial
6 months
VISIT Figure 7. Metabolic control improved in patients who returned for all follow-up visits at 2, 4, and 6 months (n = 3591, but did not improve as much if patients missed visits at 2 and 4 months (n = 31). P < 0.001 for difference in 6-month HbAlc levels, adherent versus non adherent, Student’s t test.
DISCUSSION Our objective has been to develop strategies which can be applied effectively by mid-level care providers, and constitute a model for management of patients who have limited access to subspecialty care. We focused on patients with type II diabetes since this is the most common type of diabetes found in the Grady Diabetes Unit and across the United States. As a result of these approaches, patients exhibited improved metabolic control that was largely sustained through a follow-up period of 12 months. Average HbAlc decreased from 9.6% to 8.1%, a fall of 1.5%, similar to the 2.0% difference between the DCCT “standard therapy” levels and “intensive therapy” levels.’ Such improvements were accomplished with decreased use of pharmacologic therapy, as management of obese subjects with diet alone increased from 36% to 70% over the first 2 months of supervision and remained at 55% after 12 months. Surprisingly, the HbAlc at 12 months in obese subjects was similar in those who gained weight over the first 2 months of follow-up, as compared with those losing over 8.5 lb. While highly significant, those improvements are lower than those reported by Banerji et al” and by Umpierrez et alzl in obese African-American patients who present in diabetic ketoacidosis, suggesting that the pathophysiology of decompensation may be different in our more typical patients. In the course of analysis, we recognized several problems. First, glycemic control among obese patients requiring pharmacologic therapy may have been less than satisfactory, in part from attempts to wean pharmacologic agents when such efforts were destined to fail. After review of the data permitted recognition that maximum therapeutic response to diet was generally achieved at 2 months, treatment
plans were revised and now include provisions for aggressive reinstitution of pharmacologic therapy if results with dietary therapy are not encouraging by the 2-month visit. In addition, analysis of d.ata at 12 months suggeststhat in obese patients presenting on diet alone, high FPG may indicate a need for pharmacologic therapy. Second, there was a high rate of patient attrition. Although the average patient return rate from any one appointment to the next averaged 90%,a cumulative effect of this dropout rate was that less than half of the initial patient population returned for the scheduled 6-month visit. Since patients who missed intervening visits had little improvement in HbAlc after 6 months of follow-up, our data suggest that patients who did not return at all would also have had poor metabolic control. Such a conclusion is consistent with other findings that infrequent attenders to clinic and defaulters from clinic have poorer metabolic control than those who return for appointments.2’,23 Examination of demographic features at presentation did not identify useful predictors for loss to follow-up. However, telephone contact with a small number of patients who failed to return indicated that patients felt well and saw no need for continued follow-up. Thus, we are now attempting to educate patients to the need for ongoing medical attention. Such problems may result in part from the particular health belief and behavior systems of urban African-American patients, who may see little need for optimal diet or utilization of medical resources to achieve distant benefit.s.11 However, to address the needs of those patients whose work schedules preclude consistent followup, we are currently designing an evening clinic and a more extensive telephone follow-up program. July 1996
The American
Journal
of Medicinea
Volume
101
31
Even in diabetes units such as ours, care of urban patients is difficult. In studies from Atlanta and Memphis reported in 1984,24 there was no significant improvement in metabolic control over a lo-year period of study, with average random plasma glucose initial 12.1 versus final 12.3 mM (217 versus 221 mg/ dL) in Atlanta, and initial 10.8 versus final 10.3 mM (195 versus 185 mg/dL) in Memphis. While our data indicate improved metabolic control as compared with these earlier studies, only continued follow-up will reveal whether such improvements are sustained. Since the combined direct and indirect health care costs of diabetes exceed $105 billion per ye=, 25,26it is important to develop approaches that are cost effective. Previous examinations revealed that hospitalization for acute diabetes-related problems at Grady Memorial Hospital began to decrease after the Diabetes Unit was established. Admissions for diabetic ketoacidosis and hyperosmolar coma fell 75%, with estimated savings of more than $20 million to Grady and third-party payers from 1971 to 1988.27 Moreover, the addition of a full-time podiatrist reduced lower extremity amputations by 50%.2s Including both decompensated diabetes and leg amputations, estimated savings have been over $43 million due to prevented admissions since 1971.27 Based on the annual budget for the Grady Diabetes Unit, together with costs for physician services, we estimate that the cost per patient visit is approximately $59, exclusive of pharmacy and supplies expenses. If the DCCT and Kumamoto study findings can be extended to our patient population, 2gand improvements in metabolic control at 12 months can be sustained, patients could anticipate decreased development of microvascular complications,’ with attendant decrease in health care costs. It is not clear how to interpret the limited impact of weight loss on metabolic control. Since many patients with NIDDM are overweight, 3oweight loss has been recommended routinely as a goal of management.31 However, many obese Americans have difficulty achieving sustained weight loss, and poor follow-up is common in weight loss prograrns.32’33 Moreover, African-American women are less likely to experience major weight loss than Caucasian women, even after aaustment for education and family income.= While difficulty in achieving weight loss would be understandable, it is harder to explain why many patients achieved improved metabolic control without losing weight.35 Other studies have shown enhanced control with diets containing foods with low glycemic index,36 and therefore our flndings may have resulted in part from broader features of dietary education.37 32
July 1996
The American
Journal
of Medicinea
Volume
101
We conclude that successful management of NIDDM in urban African-American patients often can be achieved with dietary therapy alone, even among individuals presenting on pharmacologic therapy. Improvements can largely be sustained for a la-month period, despite the constraints of limited education, poor insurance coverage, and socioeconomic disadvantage. Likewise, the high prevalence of obesity did not compromise the effectiveness of therapeutic strategies. Reliance on nonpharmacologic therapy and a team approach emph,asizing nonphysician providers likely resulted in cost savings as well. Loss to follow-up was identified as an important barrier to diabetes management, which will require further study to identify particular risk factors and to develop interventions to maximize continuing medical care for urban African Americans who are at increased risk of diabetic complications.
ACKNOWLEDGMENT We thank Gail with statistical programming assistance in
Janes, PhD, Division of Diabetes Translation, CDC, for assistance analysis; Winton Brown, Emory University Computing Center, for assistance; and Sharon DePeaza and Mary 1.0~ Mojonnier for preparing this manuscript.
REFERENCES 1. Harris MI. Epidemiological correlates of NIDDM in Hispanics, whites and blacks in the U.S. population. Dlab Care. 1991;14:639-648. 2. The Carter Center of Emory University. Closing the gap: the problem of diabetes mellitus in the UnIted States. Diab Care. 1985;8:391-406. 3. Diabetes Surveillance, 1980-I 987 PO/icy Program Research, Annual Report. Centers for Disease Control, Division of Diabetes Translation, DHHS; 1990;9-12 (unpublished). 4. Wetterhall SF, Olson DR, DeStefano F, et al. Trends in diabetes and diabetic complications, 1980-1987. Diab Care. 1992;15:960-967. 5. Kahn HA, Hiller R. Bkndness caused by diabetic retinopatlny. Am J Ophthalmol. 1974;78:58-67. 6. Rabb
MF, Gagliano
DA, Sweeney
HE. Diabetic
rebnopathy
in blacks.
Diab
Care. 1990;13:1202-1206. 7. Most RS, Sinnock P. The epidemiology of lower diabetic indiwduals. Diab Care. 1983;6:87-91.
extremity
amputabons
in
8. Diabetes Surveillance, 1991. Centers for Disease Control, Division of Diabetes Translation, DHHS Publication No. 635-150, 1992:41-46 (unpublished). 9. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of longterm complications in insulinzlependent diabetes mellitus. NEJM. 1993;329:
977-986. 10. Ohkubo the progression
Y, Kishikawa of diabetic
H, Araki E, et al. Intensive insulin therapy microvascular complications in Japanese
with non-insulindependent diabetes mellitus: a randomized prospective study. Diab Res Clin Prac. 1995;28:103-117. 11. Anderson RM, Herman WH, Davis JM, et al. Barriers to improving
prevents patients G-year diabetes
care for blacks. Diab Care. 1991;14:605-609. 12. Goldschmid MG, Domin WS, Ziemer DC, et al. Diabetes in urban AfncanAmericans. II. High prevalence of microalbuminuria and nephropathy in AfricanAmericans with diabetes. Oiab Care. 1995;18:955-961. 13. Musey VC, Lee JK, Crawford R, et al. Diabetes in urban African-Americans. I, Cessation of insulin therapy is the major precipitating cause of diabetic ketoacidosis. Diab Care. 1995;18:483-489. 14. Davidson JK, Alogna M, Goldsmith M, Borden J. Assessment of program effectiveness at Grady eds. Educating Diabetic
Memorial Patients.
Hospital-Atlanta. In: Steiner New York: Springer-Verlag;
G, Lawrence
1981:329-348.
PA,
15. Runyan JW, Phillips WE, Herring 0, Campbell L. A program for the care of patients with chronic diseases. JAMA. 1970;211:476-479. 16. Miller LV, Goldstein J. More efficient care of diabetic patients in a countyhospital setting. NEJM. 1972;286:1388-1391. 17. Hamwi GJ. Therapy: changing dietary concepts. In: Danowski TS, ed. Diabetes Mellitus: Diagnosis and Treatment. Vol I. New York: American Diabetes Association; 1964:73-78. 18. American Diabetes Association. Office guide to diagnosis and classification of diabetes mellitus and other categories of glucose tolerance. Drab Care. 1993;16fsuppl 2):4. 19. American Diabetes individuals with diabetes
Association. Nutriional recommendations and principles melliis. Drab Care. 1993;16(suppl 2):22-29.
for
20. Banerji MA, Chaiken RL, Huey H, et al. GAD antibody negative NIDDM in adult black subjects with diabetic ketoacidosis and increased frequency of human leukocyte antigen DR3 and DR4: flatbush diabetes. Diabetes. 1994;43:
741-745. 21. Umpierrez GE, Casals MMC, Gebhart SSP, et al. Diabetic obese African-Americans. Diabetes. 1995;44:790-795. 22. Jacobson control (study
AM, Adler AG, Derby L, et al. Clinic of contrasting groups of patients
ketoacidosis
in
attendance and glycemic with IDDM). Diab Care.
Hammersley
MS, Holland
MR, Walford
S, Thorn
PA. What
happens
to de-
faulters from a diabetic clinic? BMJ. 1985;291:1330-1332. 24. Davidson JK, Zwaag RV, Cox CL, et al. The Memphis and Atlanta continuing care programs for Diabetes. II. Comparative analyses of demographic characteristics, treatment methods, Drab Care. 1984;7:25-31. 25. Bransome ED. Improving Care.
1992;15fsuppl
and outcomes
1):66-72.
the financing
over a 9-10 of diabetes
year follow-up
period.
care in the 1990’s. Diab
WM, Mendelson DN. Health care expenditures 1992. J C/in Endo & Metab. 1994;78:809A-809F.
27. Davidson JK. What does the doctor do when allied health take over? The view of a medical convert. In: Larkins R, Zimmet eds. Diabetes 1988. Amsterdam: Elsevier; 1989:955-958. 28. Davidson JK, Delcher HK, Englund A. Spinoff cost/benefit:s nutritional care. J Am Diet Assoc. 1979;75:250-257,
for people
professionals P, Chisolm
Rifkin H, Porte D, eds. Diabetes Mellitus. New York: Elsevier; 31. Horton ES, Jeanrenaud B. Obesity and diabetes mellitus. D, eds. Diabetes Mellitus. New York: Elsevier; 1990:457.
D,
of expanded
29. Klein R. Hyperglycemia and microvascular and macrovascular disease diabetes (Kelly West Lecture). Diabetes. 1994;43:iv. Abstract. 30. Vinik A, Wing R. Nutritional management of the person with diabetes.
in In:
1990:464-496. In: Rdkin H, Porte
32. Wadden TA. The treatment of obesity: an overview. In: Sturlkard AJ, Wadden TA, eds. Obesity: Theory and Therapy. New York: Raven Press; 1993:197-
217. 33. Wing RR, Blair E, Marcus patients
1991;14:599-601. 23.
26. Rubin RJ, Altman with diabetes mellitus,
with type II diabetes:
M, et al. Year-long weight loss treatrnentfor does including an intermittent very-low-calorie
obese diet
improve outcome? Am J Med. 1994;97:354-362, 34. Kahn HS, Williamson DF, Stevens JA. Race and weight change in U.S. women: the roles of socioeconomic and marital status. Am J Pub Health.
1991;81:319-323. 35. Wing RR, Blair EH, Bononi P, et al. Caloric restriction per se is a significant factor in improvements in glycemic control and Insulin sensitivity during weight loss In obese NIDDM pabents. Diab Care. 1994;17:30-36. 36. Miller JC. Importance of glycemic index in diabetes. Am J Cfin Nutr. 1994;59(suppl 3):7478-7528. 37. Franz MJ, Horton ES, Bantle JP, et al. Nutrition principles fclr the management of diabetes and related complicahons. Diab Care. 1994;1.7:490-518.
July 1996
The American
Journal
of Medicine@
Volume
101
33
of
David C. Ziemer, MD, Merilyn G. Goldschmid, MD, Victoria C. Musey, MD, William S. Domin, MD, Peter M. Thuk, MD, Daniel L. Gallina, MD, Lawrence S. Phillips, MD, Atlanta, Georgia
OBJECTIVE: Management of type II diabetes is difficult, particularly in urban populations with limited resources and access to care. To evaluate the effectiveness of structured care delivered by non-physician providers, patients were studied prospectively for 6 months in a municipal hospital diabetes clinic. DESIGN AND METHODS: The population was approximately 90% African American and had median known diabetes duration of approximately 1 year, 54% had incomes below the Federal Poverty Guideline. Primary management was provided by nursepractitioners and dietitians, and primary outcome measures were hemoglobin Ale (HbAlc), fasting plasma glucose, and changes in body weight. RESULTS: Responses were analyzed in 325 new patients returning for visits at 2, 4, 6, and 12 months; metabolic profiles at presentation were similar to those of subjects who missed intervening visits. Lean patients largely continued on pharmacologic therapy and improved HbAlc from 9.4% to 7.4% at 2 months (P < O.OOl), remained stable through 6 months, then rose to 7.9% at 1 year. Obese patients (71%) received dietary instruction. Weaning of pharmacologic therapy was attempted for the first 2 months, resulting in a decline of HbAlc from 9.6% to 8.0% (P < O.OOl), with 70% treated with diet alone. In the obese, HbAlc continued to decrease through 6 months (7.7%). Thereafter, providers saw patients at their own discretion and intensified therapy as needed. Although by 1 year, HbAlc had risen to only 8.2%, some patients required reinstitution of pharmacologic therapy; 59%
From the Division of Endocrinology and Metabolism, Emory University School of Medicine. Atlanta. Georgia. This work was supported in part by research and training awards from the National Institutes of Health, Grants DK-07298, DK-08978, DK33475, and DK-48124. Requests for reprints should be addressed to David C. Ziemer, MD, Department of Medicine, Emory University School of Medicine, 69 Butler Street, SE, Atlanta, Georgia 30303. Manuscript submitted June 20, 1995 and accepted in revised form March 25, 1996.
01996 by Excerpta All rights reserved.
Medica,
Inc.
were on diet alone. While 52% lost 4 lb or more (mean 9.3) by 2 months, little additional weight was lost. Interestingly, glycemic control was improved both in those who lost ~8.5 lb in the first 2 months (HbAlc 9.6% to 8.1% at 12 months), and in those who gained weight (HbAlc 10.2% to 8.2%). In the obese p,atients using pharmacologic agents at presentation, 35% were able to discontinue oral agents or insulin by 1 year, with good glycemic control (HbAlc ~8%). For patients who were initially on diet alone, a fasting plasma glucose >177 mg/ dL predicted the need for pharmacologiic therapy with 97% certainty. CONCLUSIONS: In urban African American patients, nonpharmacologic management of type II diabetes substantially improves metabolic control; decreases in HbAlc are compatrable in those who do and do not lose weight. Therapy managed by nonphysician providers can be an effective cornerstone of diabetes care in this socioeconomically disadvantaged population. Am J Med. 1996;101:25-33. iabetes imposes an enormous health burden on African Americans in the United States. In the II analysis, ’ the age-standardized prevalence of diabetes was 10.2% in African Americans but only 6.6% in Caucasians. African Americans consequently suffer disproportionately from diabetes-related complications.* During the 198Os, diabetes-related, age-standardized mortality decreased 1.6% and 4.5% for Caucasian men and women, respectively, yet mortality increased 11% and 5.5% among AfricanAmerican men and women, respectively.3 Among African-Americans with diabetes, the incidlence and prevalence of hypertension4 and retinopathy5 are higher than in Caucasians. In the African-American diabetic population, the incidences of end-stage renal disease,6 lower extremity amputation related to diabetes, 7 and disability due to diabetes’ exceed those for Caucasians by at least 50%. Increased morbidity of diabetes in African Americans may be due to poor metabolic control. The Diabetes Control and Complications Trial (DCCT) has shown that improved metabolic control reduces the development and progression of microvascular complications such as retinopathy, and suggested that
D NHANES
0002-9343/96/$15.00 PII SOOO2-9343(96)00071-X
25
macrovascular disease may improve as well’; Ohkubo et allo have recently reported similar findings in patients with type II diabetes. However, there has been relatively little examination of metabolic control and treatment strategies targeting African Americans, and particularly the inner-city population. The care of urban African Americans presents special problems, including a limitation of funds for medical care, problems of physical access (transportation, and so forth), an absence of a tradition of healthoriented behavior for optimizing long-term health prospects, patients’ feeling a sense of being powerless to alter health outcomes, and a lack of culturally sensitive programs and educational materials I1 ; our studies reveal that many of these patients already have microalbuminuria on lirst presentation, l2 and that many admissions for diabetic ketoacidosis are due to cessation of insulin therapy.‘” The Grady Health System Diabetes Unit serves an urban, largely African-American population, using a cost effective, multidisciplinary team approach established by John K. Davidson14 following similar approaches elsewhere in the United States.‘5,‘6 In this initial study, we evaluated the impact of a management strategy that emphasizes nonpharmacologic approaches to therapy.
METHODS Each year, about 5,000 patients are seen for follow-up care in the Grady Diabetes Unit. An additional 950 new patients are enrolled annually through a full-day evaluation and education visit; patients may be self-referred or referred by inpatient services or other clinics. To determine the economic status of this population, we reviewed the financial counselling records of 394 consecutive patients new to the Diabetes Unit from May 1,1994 to September 151994. All patients are required to submit evidence of residence and of income to determine eligibility for discounts provided to Fulton and DeKalb residents. For those claiming indigency, income evidence may include pay stubs, tax records, unemployment claims from the Georgia Department of Labor, a statement from their homeless shelter, evaluation by Georgia Department of Family & Children’s Services (for Medicaid and food stamps), or evaluation by the Social Security Administration (for disability or social security pension). Patient income was calculated relative to the Federal Poverty Guidelines (FPL). All new patients undergo an in-depth medical interview with standard format, and physical examination including height and weight in light clothing. Percent ideal body weight is calculated from the Hamwi formula.17 Race is assigned by the examining nurse-provider by observation and direct question26
July
1996
The
American
Journal
of Medicine@
Volume
101
ing. Findings are recorded prospectively using data collection sheets, and the data are entered into a microcomputer registry. Similar but more abbreviated data collection is completed at follow-up visits. Laboratory testing is performed at each visit, and these data are subsequently obtained from the department of pathology and laboratory medicine and entered into the registry. The registry is stored in a relational database (FoxPro; Microsoft, Redmond, Washington), using an interface to ease data entry, written by Emory University Computer Center personnel under the direction of the first author. The Diabetes Unit has its own blood1 chemistry and urine laboratory staffed and supervised by the department of pathology and laboratory medicine. Glucose analyses are performed on venous blood using a glucose oxidase method on an AF’IEC glucose analyzer (APEC, Inc., Danvers, Massachusetts). Hemoglobin Ale was measured using an HPLC method (DIAMAT Biorad, Hercules, California). The assay is corrected for the proportion of hemoglobin A and has a normal range of 3.5% to 6.0%. At the first visit, the diagnosis of diabetes is confirmed using standard American Diabetes Association criteria. la Type II, noninsulin-dependent diabetes mellitus (NIDDM) is defined by historical and treatment features including lack of a history of ketosis, lack of absolute requirement for insulin, presence of obesity and/or strong family hisl;ory of diabetes, and lack of diabetogenic medications or recurrent pancreatitis. Each new patient is assigned for follow-up to one health care provider, a nurse practitioner or registered nurse with special interest and training in diabetes care. All patients attend classes i.n diabetes education at the initial visit, and are seen by an assigned dietitian in a one-on-one setting at this and each of their early return visits. Individual patient treatment plans are developed by health care providers, guided by a uniform protocol. At each visit, patients are also seen by an endocrinologist, who reviews the treatment plan. All patients receive instruction on healthy food choices, and obese patients ( > 120%of ideal body weight) receive instruction on a hypocaloric diet with a deficit of approximately 500 calories per day.lg For obese patients with NIDDM, management guidelines have been designed to encourage treatment with diet and weight loss alone, as follows: l All obese patients receive a diet with deficit 500 k&/day. l If fasting plasma glucose (FPG) is ~19.4 mM (350 mg/dl) , pharmacologic therapy is instituted or increased (if already present). l If FPG < 19.4 mM (350 mg/dl) and there is no clinical evidence of decompensation (eg, orthostatic
hypotension), oral agents are discontinued and insulin is decreased 50% for FPG less than 11.1 mM (ZOO mg/dl) or 25% for FPG of 11.1 to 16.7 mM (200 to 300 mg/dL). l Insulin is discontinued if the daily dose is less than 0.2 units/kg and the FPG is less than 16.7 mM (300 mg/dL) . Visits are scheduled frequently (at 1, 2, and 4 weeks, and at 2, 4, and 6 months) to maximize the educational components of therapy and to avoid metabolic decompensation. Fasting plasma glucose levels are obtained at each visit and HbAlc levels are obtained initially and then bimonthly. Data for the 1Smonth visit were obtained from visits occurring between 42 and 62 weeks after the initial visit; numerical data for these visits were averaged, and categorical data (eg, type of therapy) were taken from the last visit in the time period. Data were analyzed using microcomputer-based data instruments including Excel (Microsoft, Redmond, Washington) and Statview 11 (Abacus Concepts, Berkeley, California), with evaluation by contingency table analysis, Mann-Whitney U (rank) test, Student’s t test, or analysis of variance (ANOVA) with repeated measures. No missing data were imputed. Repeated measures ANOVA for HbAlc could be performed only for those 112 patients with HbAlc values at all time points. These 112 were not different from the larger group of 325 (see below) with respect to age, gender, duration of diabetes, or percent of ideal body weight. Although initial fasting plasma glucose was slightly higher (P < 0.03)) initial HbAlc was not, and those with HbAlc at all points were felt to be representative of the group as a whole.
1994,300 had usable income data (28 were from outside Nton or DeKalb counties, 25 failed to provide data, and 41 had income evaluations pending with state or federal agencies). Of those with1 income data, 54% had incomes below the FPL, 69% were at or below 125% of the FPL, and only 13% had incomes greater than 190% of the FPL (or did not apply for discounted care). From April 15, 1991 to April 15, 1993, the Grady Diabetes Unit enrolled 1,824 new patients, whose characteristics are shown in Table I. There was a predominance of females and African Americans. The median age of patients was 52.5 years and the mean 51.9 years, while median duration of diabetes was approximately 1.2 years, with an avera$e of 5.4 years duration. Although 4 patients were found not to have diabetes, the remaining 1,820 patients were classified as type 1 (5.3%), type 11 (87.4%)) and impaired glucose tolerance ( 1.8%). The remaining 5.5% had diabetes that was secondary or of unclear etiology. For the 1,634 patients with type 11 diabetes, the mean age was 53 years, 59% were female and 41% male, and 88% were African American and 8.8OhCaucasian (Table I). At presentation, the median duration of diabetes was about 1.2 years, with an average of 5.3 years. Patients averaged 142% of ideal body weight, with 88% of the females and 49% of t.he males being above 120% of ideal body weight. For patients with type 11 diabetes, the initial fasting plasma glucose averaged 10.9 mM (197 mg/dL), with HbAlc 9.3%. In order to assess the impact of managem.ent strategies over a 6-month period, we focused on a subpopulation of 325 patients with type 11 diabetes who returned for visits as scheduled, ie, at 2, 4, 6, and 12 RESULTS months. These 325 patients were representative of Patient Characteristics the entire group of 1,634 type 11 subjects, as shown by the lack of a statistically significant difference beThe Grady Diabetes Unit population is composed of adults, 97% of whom reside in either Fulton or tween those 325 patients and the other patients in racial distribution, initial therapy, mean duration of DeKalb County, which contain the city of Atlanta. diabetes, percent of ideal body weight, fasting Fewer than 10% of new patients have commercial health insurance, and approximately 50% have no plasma glucose, or HbAlc. However, patients were more likely to return as scheduled if they were dithird-party health care coverage. Relative to the population of either county or Atlanta, the population of agnosed less than 1 year previously (23% versus 17% return rate, P < O.OOl), older than 55 (27% versus the Diabetes Unit is disproportionately African 14% return rate, P < O.OOl), or female (22% versus American (approximately 88%). In patients who said 17% return rate, P < 0.05). Although 85% of patients they could read, average literacy has been evaluated kept appointments for each subsequent visit, patient as sixth grade level (K. Dobbenstein, personal communication). Thus, the Diabetes Unit population is dropout was high over the first 12 months of followtypical of many served by urban hospitals in the up. By our strict criteria, 80% missed at least one follow-up visit over the la-month period, despite the United States. and There is a high prevalence of homelessness, and availability of frequent routine appointmlents provision of flexibility in scheduling. An informal the population is severely economically disadvansurvey of 30 patients not returning by 6 months retaged. Of the 394 Fulton and DeKalb residents presenting between April 1, 1994 and September 15, vealed that haIf could not be contacted by telephone, July 1996
The American
Journal
of Medicine@
Volume
101
27
TABLE I Grady
Diabetes
Clinic
Type 2 All Patients n = 1,824
Gender (WI Race (%I
Class (%I
Therapy (%I at Presentation
Duration (years) Age (years) Fasting plasma glucose fmM) Fasting plasma glucose fmg/dL) Hemoglobin Ale (%I % lBW-(%, * P 10.03, + P ~0.03, f P ctO.002, 1 P ~0.001,
Female Male Black White Hispanic Other Type 1 Type 2 PancfitOH Other IGT Diet Sulfonylurea Insulin Both Mean Median Mean Mean Mean Mean Mean
Diabetes n = 1,634
57.0 43.0 87.1 9.6 1.4 1.9 5.3 87.4 2.0 2.5 1.8 30.7 22.4 46.4 0.5 5.4 1.2 51.9 11.0 198 9.30 139
58.9 41.1 87.7 8.8 1.4 2.2 100.0 31.5 24.6 43.4 0.6 5.3 1.2 52.9 10.9 197 9.30 142
Type 2 Diabetes Returns n = 325
64.3* 35.7* 88.3 9.0 0.9 1.9 100.0 32.6 25.5 41.5 0.3 4.8+ 0.5t 57.7” 10.7 193 9.30 140
Type
2 Diabetes Returns African-American n = 286
65.4 34.6 100.0 -----
1OCl.O --
-31.8 25.5 42.6 0.0 4.6 0.4 57.5 10.7 192 9.40 141.1
chisquare test, return status versus gender. Mann-Whitney U statistic, Nonreturnees (not shown) versus returnees. chi-square test, return status versus duration -C 1 year. Mann-Whitney U statisttc, nonreturnees (not shown) versus returnees.
with type II diabetes, and were virtually indistinguishable from the non-African Americans in gender distribution, percent of ideal body weiglht, duration of known diabetes, age, and indices of diabetes control or treatment at presentation (Table I).
Glycemic
0
2
4
6
12
0
2
4
6
12
Visit (mo) Figure 1. Metabolic control (HbAlc)for patients returning with HbAlc values at the 2-, 4-, 6-, and 12-month visits. Differences over time are significant for both obese (83) and nonobese (29) patients If < 0.001, ANOVA repeated measures) with no difference or interaction between the two groups. n (obese) = 83; n (non-obese) = 29.
using both hospital records and the local telephone directory. Of those contacted, half stated that they did not return because they “felt well”; only 2 patients were obtaining care from some other source. The 286 African-American patients with type II diabetes comprised 88% of the 325 returning patients 28
July 1996
The American
Journal
of Medicinea
Volume
101
Control
Improvements in glycemic control were seen within 2 months and remained steady through 6 months of follow-up, with a slight rise at 12 months: mean FPG fell by 2 months from 10.9 to 91.6mM (196 to 173 mg/dL) in obese subjects and from 10.7 to 8.6 mM (192 to 154 mg/dL) in nonobese subjects (both P < 0.001). At 1 year, FPG levels were 9.9 and 8.8 mM (179 and 159 mg/dL), respectively. Similarly, HbAlc improved from 9.6%to 8.0%in obese and from 9.4% to 7.4%in nonobese subjects at 2 months (both P < O.OOl), and levels were 7.7% and 7.5%, respectively, at 6 months, and 8.2% and 7.9%, respectively, at 12 months (Figure 1) . The number of patients achieving HbAlc ~8.0% increased by 2 months from 33% to 56% of obese subjects and from 42% to 73% of lean subjects, and fell only slightly thereafter (54% and 62% at 12 months, respectively). Efforts to treat patients with nonpharmacologic approaches were largely successful. At presentation,
NIDDM IN URBAN AFRICAN AMERICANS/ZIEMER
Obese
1
Non-obese
HbAlc
ET AL
Pharmaco-therapy
12-
1-I
‘0°
10
.60
co 80 H
'= s
60
6 8
40 2
20 0
n
Initial
0
2
4
612
0
2
4
612
Visit (mo) Figure 2. Distribution of therapy by visit for all obese (231) and nonobese (94) patients returning at 2, 4, 6, and 12 months. n (obese) = 231; n (non-obese) = 94. k# Insulin; 0 both (OA + Ins); n diet only.
120 100 v)
80
5 'E d
60
-5 8
40 20 0
Diet
Sulfonylurea Initial
Insulin
therapy
Figure 3. Distribution of therapy for obese patients at the 12-month visit versus therapy at presentation (insulin therapy at 12 months includes 4 patients also on sulfonylureas). Number of patients at presentation on diet was 83; sulfonylurea, 56; insulin, 83; and insulin plus sulfonylurea, 1 (not shown). E# Insulin; oral agent; n diet only.
36% of obese subjects were being managed with diet, 24% with sulfonylureas, and 40% with insulin (Figure 2). However, after two months, 70%were being managed with diet alone. Thereafter, there was a small decrease in management with diet alone at 6 months (61%)) and a further decrease by 12 months, when obese patients were managed 55% with diet alone, 26% with sulfonylureas, and 19%with insulin (2% with both sulfonylureas and insulin). As expected, lean patients more often presented on pharmacologic agents (75%), and proportions changed relatively little over 12 months of follow-up. After 12
Initial
12mo.
12mo.
Figure 4. Impact of therapy. Metabolic control (HbAlc) in obese patients improved from initial to 12-month visit while percentage of patients on pharmacologic therapy decreased (n = 83 and n = 230, respectively).
months, 27% were managed with diet, 34% with sulfonylureas, and 36% with insulin. As shown in Figure 3, we found that most obese patients could be managed with diet, regardless of their therapy on presentation. In obese patients who presented on dietary therapy alone, 67% were continued on diet alone after 12 months, with another 29% on sulfonylureas, and only 4% were taking insulin. In obese patients who came to the Diabetes Unit on sulfonylureas, after 12 months, 38% were managed with diet alone and 48% returned .to sulfonylureas, with 14% requiring insulin. Even in obese patients who came to the Diabetes Unit on insulin, 54% were managed with diet alone after 12 months. The effectiveness of nonpharmacologic approaches was particularly striking (Figure 4). At presentation, 231 of the 325 subjects were obese, and of these, 83 had HbAlc measurements that permitted evaluation of responses at 2,4,6, and 12 months. While at presentation average HbAlc was 9.6%, and 64% of patients were managed with pharmacologic agents, 6 months later, average HbAlc was 7.7%, and only 39% of patients required pharmacologic agents. At 12 months, average HbAlc was 8.2%, and 45% required pharmacologic agents. Moreover, nonpharmacologic approaches were also effective in obese patients who presented on sulfonylureas or insulin 47%of 148 obese type II patients who presented on pharmacologic therapy were on diet alone after 12 months, with HbAlc decreasing from 9.6% to 8.1%; HbAlc decreased from 9.6% .to 7.2% in obese patients who presented on diet alon’e. Both of these improvements in HbAlc were high.ly signiflcant (P < 0.001). To date, we have found it difficult to predict either eventual glycemic control or need for pharmacologic therapy. However, 85% of type II patients who presented with evidence of reduced pancreatic insulin July 1996
The American
Journal
of Medicine@
Volume
101 29
reserve as shown by a fasting C-peptide 5 1.O ng/mL were either on pharmacologic therapy or had HbAlc greater than 8.0% 1 year later. Moreover, patients with shorter duration of diabetes (
Weight
July 1996
Journal
of Medicine@
Volume
101
4
6
12
Visit (mo.) Figure 5. Metabolic control (HbAlc) in nonobese patients treated with insulin throughout (n = 9) is better at 1 year than in obese patients whose insulin was decreased and then resumed (n = ;3; P < 0.01 by Student’s t test). n Obese (8);
12 10 ‘;; 5 0 2 P
8 6 4 2 0
The American
2
L
Loss
Obese patients lost an average of 1.6 lb over the l-year follow-up period. Within the llrst 2 months of follow-up, approximately 48% of obese patients achieved weight loss of at least 0.5 lb per week. These “responders” to a hypocaloric diet prescription lost an average of 9.7 lb over the first 2 months, but little further weight loss was achieved after that time. While those losing at least 8 lb over the first 2 months (64 subjects) exhibited improved metabolic control (HbAlc decreased from 9.9%to 8.2%, Figure 6), HbAlc also improved from 11.6% to 7.3% in the 54 subjects who gained weight over the first 2 months (between groups P > 0.5, ANOVA repeated measures for all visits). These improvements in metabolic control were largely sustained through 12 months of follow-up, at 8.0% and 8.2%, respectively. Moreover, there was no difference in the number of patients attaining good glycemic control as measured by HbAlc < 8.0% (P = O.SO), nor did weight loss make a difference in the likelihood of being able 30
0
0
2
4
6
12
Visit (mo.) Figure 6. Metabolic control (HbAlc) in obese patients who lost 8 lb or more (n = 19) by 2 months improved no more than in those who gained weight (n = 16; P > 0.1 by ANOVA, repeated measures). n Lost 28 lb by 2 months;
to discontinue insulin with good metabolic control at 12 months; 45% of those on insulin who lost more than 8.0 lb by 2 months were able to discontinue insulin while maintaining HbAlc < 8.0%, and 58% of those who gained weight were able to discontinue insulin with good control (chi-square test, P =
>0.60). Impact
of Missed
Visits
Examination of patients returning for their 6month visit showed that 31 who did not keep their intervening bimonthly appointments had higher
HbAlc at 6 months than 359 who returned for those appointments (8.9% versus 7.5%, P < O.OOl), although the groups had comparable HbAlc at presentation (Figure 7). cost Our approximately 1,500 patients were enrolled over a period of 2 years, and followed up for an additional 12 months. In 1994, costs including estimates for physician and staff salaries, rent, utilities, and administrative overhead were $1.52 million for 25,974 patient visits per year. Thus, average costs were $59 per patient visit. This estimate encompasses a wide variety of patient contacts, including initial visits that lasted 6 or 7 hours, scheduled follow-up appointments, walk-in visits for old or new patients, and brief visits for prescription refills, and so forth.
missed vis’its all visits
Initial
6 months
VISIT Figure 7. Metabolic control improved in patients who returned for all follow-up visits at 2, 4, and 6 months (n = 3591, but did not improve as much if patients missed visits at 2 and 4 months (n = 31). P < 0.001 for difference in 6-month HbAlc levels, adherent versus non adherent, Student’s t test.
DISCUSSION Our objective has been to develop strategies which can be applied effectively by mid-level care providers, and constitute a model for management of patients who have limited access to subspecialty care. We focused on patients with type II diabetes since this is the most common type of diabetes found in the Grady Diabetes Unit and across the United States. As a result of these approaches, patients exhibited improved metabolic control that was largely sustained through a follow-up period of 12 months. Average HbAlc decreased from 9.6% to 8.1%, a fall of 1.5%, similar to the 2.0% difference between the DCCT “standard therapy” levels and “intensive therapy” levels.’ Such improvements were accomplished with decreased use of pharmacologic therapy, as management of obese subjects with diet alone increased from 36% to 70% over the first 2 months of supervision and remained at 55% after 12 months. Surprisingly, the HbAlc at 12 months in obese subjects was similar in those who gained weight over the first 2 months of follow-up, as compared with those losing over 8.5 lb. While highly significant, those improvements are lower than those reported by Banerji et al” and by Umpierrez et alzl in obese African-American patients who present in diabetic ketoacidosis, suggesting that the pathophysiology of decompensation may be different in our more typical patients. In the course of analysis, we recognized several problems. First, glycemic control among obese patients requiring pharmacologic therapy may have been less than satisfactory, in part from attempts to wean pharmacologic agents when such efforts were destined to fail. After review of the data permitted recognition that maximum therapeutic response to diet was generally achieved at 2 months, treatment
plans were revised and now include provisions for aggressive reinstitution of pharmacologic therapy if results with dietary therapy are not encouraging by the 2-month visit. In addition, analysis of d.ata at 12 months suggeststhat in obese patients presenting on diet alone, high FPG may indicate a need for pharmacologic therapy. Second, there was a high rate of patient attrition. Although the average patient return rate from any one appointment to the next averaged 90%,a cumulative effect of this dropout rate was that less than half of the initial patient population returned for the scheduled 6-month visit. Since patients who missed intervening visits had little improvement in HbAlc after 6 months of follow-up, our data suggest that patients who did not return at all would also have had poor metabolic control. Such a conclusion is consistent with other findings that infrequent attenders to clinic and defaulters from clinic have poorer metabolic control than those who return for appointments.2’,23 Examination of demographic features at presentation did not identify useful predictors for loss to follow-up. However, telephone contact with a small number of patients who failed to return indicated that patients felt well and saw no need for continued follow-up. Thus, we are now attempting to educate patients to the need for ongoing medical attention. Such problems may result in part from the particular health belief and behavior systems of urban African-American patients, who may see little need for optimal diet or utilization of medical resources to achieve distant benefit.s.11 However, to address the needs of those patients whose work schedules preclude consistent followup, we are currently designing an evening clinic and a more extensive telephone follow-up program. July 1996
The American
Journal
of Medicinea
Volume
101
31
Even in diabetes units such as ours, care of urban patients is difficult. In studies from Atlanta and Memphis reported in 1984,24 there was no significant improvement in metabolic control over a lo-year period of study, with average random plasma glucose initial 12.1 versus final 12.3 mM (217 versus 221 mg/ dL) in Atlanta, and initial 10.8 versus final 10.3 mM (195 versus 185 mg/dL) in Memphis. While our data indicate improved metabolic control as compared with these earlier studies, only continued follow-up will reveal whether such improvements are sustained. Since the combined direct and indirect health care costs of diabetes exceed $105 billion per ye=, 25,26it is important to develop approaches that are cost effective. Previous examinations revealed that hospitalization for acute diabetes-related problems at Grady Memorial Hospital began to decrease after the Diabetes Unit was established. Admissions for diabetic ketoacidosis and hyperosmolar coma fell 75%, with estimated savings of more than $20 million to Grady and third-party payers from 1971 to 1988.27 Moreover, the addition of a full-time podiatrist reduced lower extremity amputations by 50%.2s Including both decompensated diabetes and leg amputations, estimated savings have been over $43 million due to prevented admissions since 1971.27 Based on the annual budget for the Grady Diabetes Unit, together with costs for physician services, we estimate that the cost per patient visit is approximately $59, exclusive of pharmacy and supplies expenses. If the DCCT and Kumamoto study findings can be extended to our patient population, 2gand improvements in metabolic control at 12 months can be sustained, patients could anticipate decreased development of microvascular complications,’ with attendant decrease in health care costs. It is not clear how to interpret the limited impact of weight loss on metabolic control. Since many patients with NIDDM are overweight, 3oweight loss has been recommended routinely as a goal of management.31 However, many obese Americans have difficulty achieving sustained weight loss, and poor follow-up is common in weight loss prograrns.32’33 Moreover, African-American women are less likely to experience major weight loss than Caucasian women, even after aaustment for education and family income.= While difficulty in achieving weight loss would be understandable, it is harder to explain why many patients achieved improved metabolic control without losing weight.35 Other studies have shown enhanced control with diets containing foods with low glycemic index,36 and therefore our flndings may have resulted in part from broader features of dietary education.37 32
July 1996
The American
Journal
of Medicinea
Volume
101
We conclude that successful management of NIDDM in urban African-American patients often can be achieved with dietary therapy alone, even among individuals presenting on pharmacologic therapy. Improvements can largely be sustained for a la-month period, despite the constraints of limited education, poor insurance coverage, and socioeconomic disadvantage. Likewise, the high prevalence of obesity did not compromise the effectiveness of therapeutic strategies. Reliance on nonpharmacologic therapy and a team approach emph,asizing nonphysician providers likely resulted in cost savings as well. Loss to follow-up was identified as an important barrier to diabetes management, which will require further study to identify particular risk factors and to develop interventions to maximize continuing medical care for urban African Americans who are at increased risk of diabetic complications.
ACKNOWLEDGMENT We thank Gail with statistical programming assistance in
Janes, PhD, Division of Diabetes Translation, CDC, for assistance analysis; Winton Brown, Emory University Computing Center, for assistance; and Sharon DePeaza and Mary 1.0~ Mojonnier for preparing this manuscript.
REFERENCES 1. Harris MI. Epidemiological correlates of NIDDM in Hispanics, whites and blacks in the U.S. population. Dlab Care. 1991;14:639-648. 2. The Carter Center of Emory University. Closing the gap: the problem of diabetes mellitus in the UnIted States. Diab Care. 1985;8:391-406. 3. Diabetes Surveillance, 1980-I 987 PO/icy Program Research, Annual Report. Centers for Disease Control, Division of Diabetes Translation, DHHS; 1990;9-12 (unpublished). 4. Wetterhall SF, Olson DR, DeStefano F, et al. Trends in diabetes and diabetic complications, 1980-1987. Diab Care. 1992;15:960-967. 5. Kahn HA, Hiller R. Bkndness caused by diabetic retinopatlny. Am J Ophthalmol. 1974;78:58-67. 6. Rabb
MF, Gagliano
DA, Sweeney
HE. Diabetic
rebnopathy
in blacks.
Diab
Care. 1990;13:1202-1206. 7. Most RS, Sinnock P. The epidemiology of lower diabetic indiwduals. Diab Care. 1983;6:87-91.
extremity
amputabons
in
8. Diabetes Surveillance, 1991. Centers for Disease Control, Division of Diabetes Translation, DHHS Publication No. 635-150, 1992:41-46 (unpublished). 9. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of longterm complications in insulinzlependent diabetes mellitus. NEJM. 1993;329:
977-986. 10. Ohkubo the progression
Y, Kishikawa of diabetic
H, Araki E, et al. Intensive insulin therapy microvascular complications in Japanese
with non-insulindependent diabetes mellitus: a randomized prospective study. Diab Res Clin Prac. 1995;28:103-117. 11. Anderson RM, Herman WH, Davis JM, et al. Barriers to improving
prevents patients G-year diabetes
care for blacks. Diab Care. 1991;14:605-609. 12. Goldschmid MG, Domin WS, Ziemer DC, et al. Diabetes in urban AfncanAmericans. II. High prevalence of microalbuminuria and nephropathy in AfricanAmericans with diabetes. Oiab Care. 1995;18:955-961. 13. Musey VC, Lee JK, Crawford R, et al. Diabetes in urban African-Americans. I, Cessation of insulin therapy is the major precipitating cause of diabetic ketoacidosis. Diab Care. 1995;18:483-489. 14. Davidson JK, Alogna M, Goldsmith M, Borden J. Assessment of program effectiveness at Grady eds. Educating Diabetic
Memorial Patients.
Hospital-Atlanta. In: Steiner New York: Springer-Verlag;
G, Lawrence
1981:329-348.
PA,
15. Runyan JW, Phillips WE, Herring 0, Campbell L. A program for the care of patients with chronic diseases. JAMA. 1970;211:476-479. 16. Miller LV, Goldstein J. More efficient care of diabetic patients in a countyhospital setting. NEJM. 1972;286:1388-1391. 17. Hamwi GJ. Therapy: changing dietary concepts. In: Danowski TS, ed. Diabetes Mellitus: Diagnosis and Treatment. Vol I. New York: American Diabetes Association; 1964:73-78. 18. American Diabetes Association. Office guide to diagnosis and classification of diabetes mellitus and other categories of glucose tolerance. Drab Care. 1993;16fsuppl 2):4. 19. American Diabetes individuals with diabetes
Association. Nutriional recommendations and principles melliis. Drab Care. 1993;16(suppl 2):22-29.
for
20. Banerji MA, Chaiken RL, Huey H, et al. GAD antibody negative NIDDM in adult black subjects with diabetic ketoacidosis and increased frequency of human leukocyte antigen DR3 and DR4: flatbush diabetes. Diabetes. 1994;43:
741-745. 21. Umpierrez GE, Casals MMC, Gebhart SSP, et al. Diabetic obese African-Americans. Diabetes. 1995;44:790-795. 22. Jacobson control (study
AM, Adler AG, Derby L, et al. Clinic of contrasting groups of patients
ketoacidosis
in
attendance and glycemic with IDDM). Diab Care.
Hammersley
MS, Holland
MR, Walford
S, Thorn
PA. What
happens
to de-
faulters from a diabetic clinic? BMJ. 1985;291:1330-1332. 24. Davidson JK, Zwaag RV, Cox CL, et al. The Memphis and Atlanta continuing care programs for Diabetes. II. Comparative analyses of demographic characteristics, treatment methods, Drab Care. 1984;7:25-31. 25. Bransome ED. Improving Care.
1992;15fsuppl
and outcomes
1):66-72.
the financing
over a 9-10 of diabetes
year follow-up
period.
care in the 1990’s. Diab
WM, Mendelson DN. Health care expenditures 1992. J C/in Endo & Metab. 1994;78:809A-809F.
27. Davidson JK. What does the doctor do when allied health take over? The view of a medical convert. In: Larkins R, Zimmet eds. Diabetes 1988. Amsterdam: Elsevier; 1989:955-958. 28. Davidson JK, Delcher HK, Englund A. Spinoff cost/benefit:s nutritional care. J Am Diet Assoc. 1979;75:250-257,
for people
professionals P, Chisolm
Rifkin H, Porte D, eds. Diabetes Mellitus. New York: Elsevier; 31. Horton ES, Jeanrenaud B. Obesity and diabetes mellitus. D, eds. Diabetes Mellitus. New York: Elsevier; 1990:457.
D,
of expanded
29. Klein R. Hyperglycemia and microvascular and macrovascular disease diabetes (Kelly West Lecture). Diabetes. 1994;43:iv. Abstract. 30. Vinik A, Wing R. Nutritional management of the person with diabetes.
in In:
1990:464-496. In: Rdkin H, Porte
32. Wadden TA. The treatment of obesity: an overview. In: Sturlkard AJ, Wadden TA, eds. Obesity: Theory and Therapy. New York: Raven Press; 1993:197-
217. 33. Wing RR, Blair E, Marcus patients
1991;14:599-601. 23.
26. Rubin RJ, Altman with diabetes mellitus,
with type II diabetes:
M, et al. Year-long weight loss treatrnentfor does including an intermittent very-low-calorie
obese diet
improve outcome? Am J Med. 1994;97:354-362, 34. Kahn HS, Williamson DF, Stevens JA. Race and weight change in U.S. women: the roles of socioeconomic and marital status. Am J Pub Health.
1991;81:319-323. 35. Wing RR, Blair EH, Bononi P, et al. Caloric restriction per se is a significant factor in improvements in glycemic control and Insulin sensitivity during weight loss In obese NIDDM pabents. Diab Care. 1994;17:30-36. 36. Miller JC. Importance of glycemic index in diabetes. Am J Cfin Nutr. 1994;59(suppl 3):7478-7528. 37. Franz MJ, Horton ES, Bantle JP, et al. Nutrition principles fclr the management of diabetes and related complicahons. Diab Care. 1994;1.7:490-518.
July 1996
The American
Journal
of Medicine@
Volume
101
33