From the General Medicine Division, Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts
Over the past decade, clinical trial results have shown the importance of aggressive treatment of hyperglycemia, hypertension, and hyperlipidemia for patients with diabetes.1-3 Evidence-based diabetes care now requires several medications to reduce the risk of microvascular and macrovascular disease complications. During this same period, budget constraints have increased the pressure on physicians to see more patients in less time. We used the National Ambulatory Medical Care Survey (NAMCS) to assess trends in the United States from 1991 to 2000 regarding the complexity of outpatient medical regimens, provision of recommended diabetes-related preventive services, and visit length for patients with diabetes.
Patients and methods
Conducted by the National Center for Health Statistics, the NAMCS collects data from US office-based physicians each year. As part of this program, the physician or a clinic staff member records for each visit such information as diagnosis, medications used, services provided, and length of the visit (defined as time spent by the physician in face-to-face contact with the patient). We analyzed visits to internists, family physicians, and general practitioners by adult patients with diabetes older than 18 years of age. From 1991 to 2000, the annual weighted sample of visits by patients with diabetes ranged from 374 to 631, corresponding to approximately 15 million to 26 million annual visits nationwide.
We evaluated the proportion of visits from 1991 to 2000 during which at least five medicines were listed and categorized medications used to treat hyperglycemia, hypertension, and hyperlipidemia. We also evaluated the proportion of visits during which blood pressure monitoring, cholesterol testing, and counseling regarding diet, exercise, and smoking cessation were provided, as well as the trends in visit length over the decade. We used logistic regression models to control for differences in patient characteristics from year to year when comparing trends in care over time.
Patient characteristics. The sample of 4,708 primary care visits by patients with diabetes from 1991 to 2000 represented a population average of 19.6 million visits per year. Patients had a mean age of 62.4 (± 12) years; 55% of patients were women, 32% were of nonwhite race/ethnicity, and 87% had type 2 diabetes.
Changes in medical regimen complexity. In 2000, 29.9% of patients were prescribed five or more medications, which had increased in
a linear trend from 18.2% in 1991
(P for trend < .001). There was also a marked increase in the number of patients prescribed four or more medications (25.9% in 1991 compared with 43.9% in 2000; P < .001) or at least three medicines (40.4% in 1991 compared with 55.2% in 2000; P < .001; figure 1).
Although the overall proportion of patients taking any glucose-controlling medicines did not change significantly (62.5% of visits in 1991 compared with 65.8% of visits in 2000), annual visits with insulin listed decreased markedly, from 25.3% of visits in 1991 to 15.3% in 2001 (P < .001 for annual trend). The visits listing oral agents, however, increased considerably, from 37.2% to 50.5%
(P < .001). After the 1995 approval of metformin in the United States, the annual proportion of visits listing two or more glucose-controlling agents increased significantly from 1.2% in 1995 to 17.1% in 2000.
Use of antihypertensive medicines increased from 35.9% to 42.3% (P < .001) over the decade. From 1991 to 2000, the percentage of visits listing angiotensin-converting en-zyme (ACE) inhibitors and angiotensin receptor blockers increased from 9.3% to 25.7% (P < .001). The proportion of visits listing more than one antihypertensive agent, however, did not show a significant change (8.7% in 1991 compared with 11.7% in 2000). As shown in figure 2, the percentage of visits listing lipid-lowering medicines (fibrates, niacin, resins, and HMG-CoA reductase inhibitors [statins]) increased from 4.1% to 17.3% (P < .001), a considerable change.
The percentage of patient visits listing five or more medications increased by 10.6% per year (95% confidence interval [CI], 6.0%—15.0%; P < .001) after controlling for diabetes type, physician type, age, sex, race/ethnicity, and insurance status. Annual insulin use decreased (6.3% decline per year; 95% CI, –10% to –2%; P = .002), ACE inhibitor use increased (12.9% increase per year; 95% CI, 9%–16.9%; P < .001), and lipid-lowering medicine use increased significantly (19.1% increase per year, 95% CI, 14%–24%, P < .001), using similar regression models controlling for patient and physician characteristics.
Preventive services and visit duration. We assessed trends in blood pressure monitoring, cholesterol testing, and lifestyle counseling over the past decade to see whether increasingly complex medical regimens resulted in decreased time spent providing other diabetes-related services. The annual proportion of visits with blood pressure monitoring (84.3% of total visits, ranging from 78.2% of visits in 1992 to 93.8% in 1993) remained stable, as did visits with diet counseling (48.3% of total visits, ranging from 42.8% in 1991 to 53.5% in 1997), smoking cessation counseling (4.4% of total visits, ranging from 2.7% in 2000 to 7.2% in 1994), and cholesterol testing (12.1% of total visits, ranging from 8.0% in 1996 to 18.4% in 2000). Exercise counseling was provided in 22.1% of visits in 1991, compared with 31.9% of visits in 2000 (P = .01).
From 1991 to 2000, there was a modest increase (2.2 minutes) in
visit length for patients with diabetes, from a mean of 17.1 (± 8.0) minutes in 1991 to 19.3 (± 8.2) minutes in 2000 (P < .001). During the study decade, there was no change in the median visit length (15 minutes); 17.8% of visits lasted more than 20 minutes in 1991, whereas 20.9% of visits lasted more than 20 minutes in 2000 (P = .02).
The NAMCS analyzes visit-based information from a broad spectrum of US primary care providers annually, thereby allowing the characterization of nationwide trends in care over time. Using this survey, we found that from 1991 to 2000 the outpatient care of patients with diabetes had become significantly more complex. More patients were prescribed multiple oral hy-poglycemic drugs, and more patients were treated for hypertension
and hyperlipidemia. Despite this increasing complexity in medication management, we found no com-pensatory decrease in visits during which other major diabetes-related services were provided. Blood pressure monitoring, cholesterol testing, and diet and smoking cessation counseling were unchanged, and there was actually an increase in the proportion of visits with exercise counseling.
The development of new medications for the control of cardiovascular risk factors and the adaptation of clinical trial data and practice guidelines to community practice are most likely the reasons for the significant increase in the number of patients taking five or more medications. There is a continuing need, however, for an even further increase in therapy to reach current evidence-based goals of control, based on current rates of cardiovascular risk and glycemic control.4 Despite newer guidelines recommending even lower goals for low-density lipoprotein levels for these patients, cholesterol testing and lipid-lowering medicines, in particular, were each still listed for fewer than 20% of visits in 2000.5 More aggressive treatment of hyperlipidemia is needed because more than half of diabetic patients have this cardiovascular risk factor when tested.4,6
The national rates for effectively controlling hypertension are low.7 When increases in therapy for hypertensive patients are made, they are often inadequate.8 In clinical trials, most hypertensive patients require more than one medication, and over a third require three medications to achieve adequate control.9,10 Our data indicate that most patients are unlikely to be receiving this level of aggressive therapy in community practice.
We also found a marked change in insulin use compared with oral therapy. The introduction of newer oral agents (particularly metformin11), which can be used in conjunction with other glucose-lowering medicines, is likely the reason for the increase in oral therapy. This change, however, has not brought about significant improvement in glycemic control. Current glycosylated hemoglobin (A1c) goal levels are met by fewer than half of patients with diabetes.4
Our study showed no change in median visit length and a slight increase in the proportion of visits over 20 minutes long. Although care is clearly more complex based on prescription of medications, we do not know how many different visits an individual patient made in a given year. It is possible that individual patients are being seen more frequently in later years.
High productivity levels and increasingly complex care are required of primary care physicians who treat patients with diabetes mellitus. Phy-sicians must prescribe more medications and make more frequent adjustments in dose to help their patients achieve lower goals for hemoglobin A1c levels, blood pressure, and low-density lipoprotein cholesterol levels. The increasing complexity of care and the limited time available during clinic visits to address all aspects of this care may be resulting in difficulty in reaching evidence-based goals of diabetes care. Major changes in the current approach to visit-based ambulatory diabetes care may be needed to attain in the community the success of treatment attained in clinical trials.