Lack of Effect of Guideline Changes on Hypertension Control for Patients With Diabetes in the U.S., 1995–2005

The National Disease and Therapeutic Index (NDTI) is an ongoing survey of office-based physicians in the U.S. conducted by the market research firm IMS HEALTH (3). It surveys >3,500 physicians quarterly and by specialty and region, with each physician reporting all patient encounters in two randomly selected working days. The NDTI provides nationally representative diagnostic and treatment data for outpatient visits in the U.S. It has been previously used to report national disease diagnostic and treatment trends (4–8). In one study (7), it was compared with the National Ambulatory Care Medical Survey and yielded similar, but less volatile, time trends, benefiting from its larger sample size.

In the NDTI, visits diagnosed with hypertension were identified using the three-digit ICD-9-CM code 401 (essential hypertension), and visits diagnosed with both hypertension and diabetes were identified using ICD-9-CM codes 401 and 250 (diabetes) (9). Visits diagnosed with hypertension and without diabetes were then the difference between visits diagnosed with hypertension and visits diagnosed with both hypertension and diabetes. To focus on adult hypertensive patients, visits made by patients <18 years of age were excluded.

The national trends of hypertension control for diabetic and nondiabetic hypertension visits were analyzed by year for the time period of 1995–2005. As the fourth-quarter data of 2005 was not available, only the first three quarters were used in 2005. The rate of hypertension control was defined as the percent of visits with blood pressure <140/90 mmHg, separately for visits with and without diabetes. Both rates were weighted to reflect national patterns of clinical practice using sample weights built in the NDTI. These weights accounted for the probability of sampling based on physician specialty and geographic area and were adjusted for nonresponse.

For hypertensive patients with diabetes, the goal blood pressure was <140/90 mmHg in the JNC VI and VII guidelines (1,2). As a sensitivity analysis, hypertension control was defined as blood pressure <130/85 mmHg (for patients with or without diabetes). Compared with the JNC VI guidelines in 1997, the JNC VII guidelines in 2003 set a more aggressive goal diastolic blood pressure (<80 mmHg) for diabetic subjects. As the study period ended in 2005, or only 2 years after the publication of JNC VII, the incremental effect of JNC VII was not assessed.

Linear regressions were used to assess the immediate or lagged effect of JNC VI guideline changes on hypertension control for patients with diabetes during 1995–2005. There were a total of 22 yearly observations, 11 for the study group of hypertensive patients with diabetes and 11 for the control group of hypertensive patients without diabetes. The dependent variable is the rate of hypertensive control for these groups. The explanatory variables included the indicator variable “diabetics ” (value equal to 1 for diabetic subjects and 0 otherwise), the linear variable “year ” (value equal to 1 in 1995 and 11 in 2005), and an indicator variable representing either the immediate effect of JNC VI on diabetic subjects (for diabetic subjects, the value was equal to 0 during 1995–1997 and 1 during 1998–2005; for patients without diabetes, the value was always equal to 0) or the lagged effect of JNC VI on diabetic subjects (for diabetic subjects, the value was equal to 0 during 1995–2000 and 1 during 2001–2005; for patients without diabetes, the value was always equal to 0). The lag of 3 years was chosen based on the observed difference in hypertension control between hypertensive patient with diabetes and those without diabetes during 1995–2005.

There were 181,288 adult patient visits diagnosed with hypertension in the 1995–2005 NDTI database. The annual rate of blood pressure reporting was similar between diabetic and nondiabetic visits (ranging from 93 to 97%) and increased slightly over time. After the exclusion of 9,000 hypertension visits without blood pressure reported, there were 19,616 diabetic and 152,672 nondiabetic hypertension visits, representing an estimated 76 million diabetic and 638 million nondiabetic hypertension visits in the U.S.

Figure 1 depicts the time trends of hypertension control for diabetic and nondiabetic hypertension visits. Hypertension control (blood pressure <140/90 mmHg) improved for both diabetic and nondiabetic visits over time. The difference in hypertension control between diabetic and nondiabetic visits ranged from 0 to 2% during 1995–1997 and from 2 to 3% during 1998–2000. After 2001, the difference in hypertension control between diabetic and nondiabetic visits widened to 5–6%, or an ∼4% increase for diabetic visits 4 years after the publication of the JNC VI. These time trends continued after 2003 or the publication of the JNC VII.

Alternatively defining hypertension control as blood pressure <130/85 mmHg led to similar findings (Fig. 1). If different goal blood pressure by diabetes status was used, i.e., <130/85 mmHg for patients with diabetes and <140/90 mmHg for those without diabetes (as in JNC VI), diabetic visits lagged behind nondiabetic visits in hypertension control in 1995, and this gap slightly widened over time, from 16% in 1995 to 20% in 2005.

Results from the linear regressions confirmed the above observations (Table 1). Between 1995 and 2005, the annual improvement in hypertension control ranged between 1.5 and 2.5% per year across all models and specifications (all P < 0.001). Compared with patients without diabetes, the publication of the JNC VI resulted in no immediate change in hypertension control (blood pressure <140/90 mmHg) for diabetic subjects and a 3.8% [one-time] increase after 3 years]P < 0.001]). If hypertension control was defined as blood pressure <130/85 mmHg, the publication of JNC VI resulted in no immediate change in hypertension control for diabetic subjects and a 1.7% increase after 3 years (P = 0.12).

Hypertension control improved for both diabetic and nondiabetic patients in the U.S. between 1995 and 2005. However, the more aggressive goal blood pressure set in the JNC VI and VII guidelines for patients with diabetes did not lead to substantially better hypertension control for diabetic subjects compared with nondiabetic patients for whom the goal blood pressure remained the same. This finding is somewhat surprising considering the recognition of JNC guidelines as the gold standard for hypertension treatment, with similar recommendations from other organizations such as the American Diabetes Association, the American College of Physicians, and the American Academy of Family Physicians and public health efforts to promote comprehensive diabetes care (10). On the other hand, it is consistent with recent reports (11–14) on the lack of hypertension control for patients with diabetes and previous reports (7,15,16) that the publication of new guidelines had little direct impact on clinical practice.

Several limitations should be considered in interpreting these data. First, sicker patients are more likely to visit physicians and therefore overrepresented in the visit-based NDTI database. This differs with population-based studies (1,2,17–19), in which a significant portion of hypertensive patients remain undiagnosed. Second, the control group of hypertensive patients without diabetes potentially include those with multiple risk factors, target-organ damage, or renal disease, all needing aggressive blood pressure control similar to diabetic subjects (1,2). Finally, although hypertension control may be influenced by factors such as age, sex, race, geographic location, and health insurance coverage (1,2,18,19), they are not the focus of this brief report examining the national trends of hypertension control before and after guideline changes for diabetic subjects. The continued lack of hypertension control for patients with diabetes in the U.S., despite the publication of more aggressive treatment guidelines, has significant long-term cost implications for the health care system (20). Potential barriers include patient noncompliance with lifestyle recommendations and drug treatment, physician inertia to initiate and intensify treatment, and inadequate access to medical care (21,22). Identification of effective interventions to deal with these challenges and improve diabetes care is a top priority for future research.

Part of the study was completed during Y.R.W.’s previous employment at AstraZeneca Pharmaceuticals in Wilmington, Delaware. No financial support was obtained for this study. Y.R.W. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.