Lack of Effect of Guideline Changes on LDL Cholesterol Reporting and Control for Diabetes Visits in the U.S., 1995–2004

The publication of the ADA and ATP III guidelines provides an opportunity to assess the effect of guideline changes on LDL cholesterol reporting and control for diabetes visits. The publication of guidelines does not ensure timely communication to the practicing physician let alone the adoption into patient care. As the treatment strategy for CHD remained unchanged (2), we used CHD visits as the control group for the time trends of LDL cholesterol reporting and control.

We used the National Disease and Therapeutic Index (NDTI) (3), an ongoing survey of U.S. office-based physicians conducted by IMS Health providing nationally representative diagnostic and treatment data, to analyze the national trends of LDL cholesterol reporting and control for diabetes and CHD visits by year between 1995 and 2004. The NDTI has been previously used to report national disease diagnostic and treatment trends (4–6). In one study (6), it was compared with the National Ambulatory Care Medical Survey and yielded similar but less volatile time trends, benefiting from its larger sample size. Variables recorded in the NDTI include demographics, blood pressure, cholesterol levels, diagnoses, and recommended drug therapy. LDL cholesterol was reported as the “most recent result available,” which can be a pre- or posttreatment value.

Diabetes and CHD visits were identified using ICD-9-CM codes 250 (diabetes mellitus) and 410–414 (acute myocardial infarction, other acute or subacute forms of ischemic heart disease, old myocardial infarction, angina pectoris, and other forms of chronic ischemic heart disease), respectively (7). Visits by people <20 years of age or by people diagnosed with both diabetes and CHD were excluded.

We defined the rate of LDL cholesterol reporting as the percent of visits with LDL cholesterol reported. We defined the rate of LDL cholesterol control as the percent of those LDL cholesterol reported that were <100 mg/dl. Using sample weights in the NDTI, these rates were weighted to reflect national patterns of clinical practice, accounting for the probability of sampling based on physician specialty and geographic area as well as adjusting for nonresponse.

After excluding 3,953 visits with both diabetes and CHD, we identified 98,443 diabetes visits (without CHD) and 43,361 CHD visits (without diabetes) in the 1995–2004 NDTI. The latter two represented an estimated 346 million diabetes visits and 158 million CHD visits.

In 1995, the rate of LDL cholesterol reporting was 21% of diabetes visits and 23% of CHD visits (Fig. 1A). Both rates increased similarly over time to 43% of diabetes visits and 48% of CHD visits in 2004. Neither of the guidelines increased LDL cholesterol reporting for diabetes visits relative to CHD.

In 1995, the rate of LDL cholesterol control was 4% higher for diabetes visits than for CHD (Fig. 1B). LDL cholesterol control increased for both diseases over time (but at a slower speed for diabetes) to 44% of diabetes visits and 55% of CHD visits in 2004, with an absolute difference of 11%. Despite the publication of new guidelines in 1998 and 2001, diabetes lagged behind CHD visits in LDL cholesterol control after 1998.

Our results indicate that although LDL cholesterol reporting and control for diabetes and CHD visits improved continuously between 1995 and 2004, neither the 1998 ADA guidelines nor the 2001 ATP III guidelines increased LDL cholesterol control for diabetes relative to CHD. In fact, LDL cholesterol control in patients with diabetes started to lag behind CHD after 1998. This finding is surprising considering public health efforts to promote comprehensive diabetes care (8) and the recognition of ATP III as the gold standard for CHD prevention and its wide dissemination in the medical community. On the other hand, it is consistent with recent reports (9–13) of inadequate LDL cholesterol control in patients with diabetes despite progress in screening, suggesting that treatment is more of a problem than screening in diabetic patients. This finding is also consistent with previous reports (14,15) that the publication of reports by the Joint National Committee on the Detection, Evaluation, and Treatment of High Blood Pressure had little direct impact on clinical practice.

Several limitations associated with the use of the NDTI should be considered in interpreting our results. First, sicker individuals are more likely to see physicians and are therefore overrepresented. Secondly, our defined LDL cholesterol reporting may underestimate LDL cholesterol screening due to underreporting. Thirdly, we could not differentiate between type 1 and 2 diabetes, even though the former may or may not require LDL cholesterol control <100 mg/dl. To address these concerns, we examined the trends of LDL cholesterol reporting and control for diabetes visits over time and used CHD visits as a control group. Finally, although LDL cholesterol reporting and control may be influenced by a number of factors such as demographics, geographic location, and insurance coverage, they were not the focus of this brief report.

This research was supported by AstraZeneca Pharmaceuticals (Y.R.W.), the MacLean Center for Clinical Medical Ethics at the University of Chicago (G.C.A.), and the Centers for Disease Control, Chicago Center of Excellence in Health Promotion Economics (1 P30 CD000147-01 to D.O.M.).

We thank Lisa Croll for administrative assistance.