Sex Disparities in Treatment of Cardiac Risk Factors in Patients With Type 2 Diabetes

We conducted a cross-sectional observational study of outpatients with diabetes in five internal medicine practices—two hospital based and three community based—affiliated with an academic medical center between 1 January 2000 and 31 July 2003. Data were collected from the hospital Central Data Repository (laboratory test dates and results), billing claims (hospitalizations and hospital discharge diagnoses), administrative records (patient demographics and insurance status), and directly from the electronic medical record or written medical record (problem lists and prescribed medications at the end of the study period). In addition, we performed manual chart reviews at one community-based health center and one hospital-based clinic to assess changes in blood pressure and medication prescription.

We generated lists of potentially eligible patients using billing claims for nongestational diabetes (ICD-9 codes 250.00–250.90) over a 3-year period. For the two clinics that underwent manual chart review, trained research nurses created diabetes registries. Diabetes was defined based on the diagnosis being listed in the problem list, diabetes-specific medicine listed in the medication list (e.g., sulfonylurea, metformin, insulin, or equipment for insulin injection or home glucose monitoring), or diabetes diagnosis discussed in a progress note. Using these registries as a gold standard, we developed an automated algorithm to identify patients with diabetes using billing claims, laboratory testing, and problem lists and medications from the electronic medical record. Compared with the gold standard chart review, the algorithm had a sensitivity of 98% and specificity of 98% for diabetes. We then used this validated algorithm to develop diabetes registries for the additional clinics. Determination of the type of diabetes was not possible in ∼50% of the sample; in the remainder, 97% of patients were recorded as having type 2 diabetes. We limited our analysis to patients who had at least one visit early (1 January 2000 to 31 August 2001) and one visit late (1 December 2001 to 31 July 2003) in the observation period to capture a population of patients who were attending regular follow-up visits at the practices.

Forty-four percent of the patients in the study cohort were treated at hospital-based primary care practices; the rest were treated at health centers located in different neighborhoods in and around Boston. The population represented a broad sociodemographic sample of outpatients with type 2 diabetes.

Patients with missing values for HbA_1c, total cholesterol, HDL, or LDL (n = 382) were excluded from the original sample of 4,231, leaving a population of 3,849 for analysis (91%). The excluded patients were of similar age, sex, and race as the included patients, but fewer spoke English (81 vs. 87%, P = 0.002).

Demographic variables examined included age, race (white, African American, Hispanic, or other) and whether the patient spoke English (yes/no). All comorbidity and treatment variables were determined at the end of the study period. Active treatment for diabetes and CHD risk factors was defined as follows: treatment for hyperglycemia (prescription of α-glucosidase inhibitors, biguanides, insulin, glitinides, sulfonylureas, or thiazolidinediones); treatment for hypertension (ACE inhibitors, angiotensin receptor blocker [ARB], α-blocker, β-blocker, calcium-channel blocker, or thiazide); and treatment for hyperlipidemia (hydroxymethylglutaryl-CoA reductase inhibitors, cholestyramine, colestipol, fenofibrate, gemfibrozil, or niacin). We also determined whether the nonprescription med-ications aspirin, acetaminophen, and ibuprofen were listed in the medical record.

Hypertension was present if listed on the problem list or if an antihypertensive medication was listed on the medication list. CHD was defined by having CHD (or synonyms) on the problem list or CHD-specific medications such as sublingual nitroglycerin on the medication list. Dyslipidemia was defined as present if listed on the problem list, if a lipid-lowering drug had been prescribed, or if the subject was not taking a lipid-lowering drug and LDL was >100 mg/dl and HDL was <40 mg/dl in men and <50 mg/dl in women. Over 80% of patients identified using these algorithms had hypertension, CHD, or dyslipidemia recorded in the problem list. Current smoking was determined by inclusion on the problem list.

We stratified all analyses by the presence or absence of diagnosed CHD as indicated by CHD entry in the medical record, but we present some collapsed results for analyses where results did not differ by CHD status. Descriptive statistics included t tests for means and the χ² test for proportions comparing differences between women and men. We used logistic regression models to estimate adjusted odds ratios (AORs) and 95% CIs that women, using treatment in men as the referent, were prescribed specific treatments. The models were adjusted for sociodemographic variables that differed by sex in the bivariate analysis including age, racial/ethnic group, language, and smoking status and included a term for clinic site as a random effect to account for possible within-site correlation of treatment patterns.

Evaluation of treatment appropriateness included use of hypoglycemic medicines in all patients and those with HbA_1c >7%, an ACE inhibitor or ARB in all patients, antihypertensive medication in patients with hypertension, lipid-lowering therapy in patients with dyslipidemia, and aspirin in all patients. Using the same modeling strategy, we also estimated the odds of treatment effectiveness in patients who were prescribed medications for hyperglycemia, hypertension, and hyperlipidemia. Treatment effectiveness was defined as meeting recommended risk factor treatment goals defined by the ADA, the National Cholesterol Education Program, and the American Heart Association guidelines or quality-of-care benchmarks established by the National Committee on Quality Assurance (NCQA) (31). Analysis of antihypertensive treatment effectiveness was limited to the subset of patients (n = 1,817, 47% of cohort) for whom blood pressure data were available. All analyses were conducted using the SAS System for Windows, version 8.2 (32). Statistical significance was defined as a two-tailed P value <0.05.

Of 3,849 patients with diabetes in our analysis cohort, about one-half were women. Thirty percent (1,150) of the cohort had diagnosed CHD. Table 1 shows the baseline characteristics for men and women, stratified by CHD status. In general, women were more likely than men to be older, nonwhite, and non-English speaking and to be treated at a community health center. Women with CHD had significantly higher systolic blood pressure than their male counterparts. Women with and without CHD had higher total HDL and LDL levels than the men.

Compared with national benchmarks established by the NCQA (31), this cohort received high-quality care on most measures (Table 2). Over 60% of the entire population had an HbA_1c <8.0%, compared with the NCQA goal of 55%, although fewer women than men with CHD achieved this goal (62 vs. 69%, P = 0.01). In the subset of the cohort with data on blood pressure, >65% of patients had blood pressure <140/90 mmHg (the NCQA target) with no differences by sex. Lipid levels were also well controlled compared with national standards but were less well controlled in women compared with men. Eighty-one percent of women and 87% of men without CHD (P = 0.002) and 88% of women and 92% of men with CHD (P = 0.02) had LDL levels <130 mg/dl, compared with the NCQA benchmark of 63%. Still, unadjusted measures of HbA_1c, blood pressure, and lipid control in patients prescribed medications for those conditions revealed several differences in levels by sex. In patients without CHD, women had lower diastolic blood pressure and higher levels of total cholesterol and its subfractions; in patients with CHD, women had higher levels of systolic blood pressure, HbA_1c, and cholesterol and its subfractions.

Men and women were equally likely to be treated with hypoglycemic medication (Table 3). Men and women without diagnosed CHD who were treated with hypoglycemic medications had equivalent glycemic control, with about one-third meeting an HbA_1c target of <7% and about two-thirds <8%. Women with diagnosed CHD who were taking hypoglycemic medications had worse glycemic control than their male counterparts. In multivariable analysis adjusted for age, sociodemographic variables, and clinic site, women were less likely than men to have HbA_1c controlled to <7% (AOR 0.84 [95% CI 0.75–0.95], P = 0.005, n = 2,310; Fig. 1B).

In unadjusted analyses stratified for the presence or absence of CHD, about one-half of all patients without CHD and four-fifths of patients with CHD were treated with lipid-lowering medication (Table 3). In general, among patients receiving lipid-lowering therapy, men were more likely than women to have lower LDL and total cholesterol levels, whereas women had higher HDL cholesterol levels. These differences persisted after stratification for the presence of diagnosed CHD. After controlling for age, sociodemographic variables, and clinic site, women with hyperlipidemia appeared less likely than men to be prescribed lipid-lowering medications (AOR 0.82 [95% CI 0.71–0.96], P = 0.01, n = 2,464; Fig. 1A). When they were treated, women remained less likely than men to have LDL <100 mg/dl (0.75 [0.62–0.93], P = 0.007, n = 1,419) or <130 mg/dl (0.58 [0.46–0.70], P < 0.0001, n = 1,419, Fig. 1B).

In both unadjusted and adjusted analyses, women were as likely as men to be treated with antihypertensive medications, including ACE inhibitors or ARBs (Table 3 and Fig. 1A). Among patients without diagnosed CHD, men and women were equally likely to have blood pressure <140/90 mmHg (Table 3 and Fig. 1B). Among patients with CHD, nearly all of whom were treated with antihypertensive agents, women had a higher mean systolic blood pressure than men in unadjusted analysis (133 vs. 129 mmHg, P = 0.01, Table 3) and were less likely to have blood pressure controlled to <140/90 mmHg (AOR 0.88 [95% CI 0.85–0.91], P < 0.0001, n = 587, Fig. 1B).

In unadjusted analyses of patients with diabetes, aspirin was listed less frequently in the medication lists of women than men without CHD (37 vs. 46%, P < 0.0001) and of women than men with CHD (72 vs. 79%, P < 0.01). In contrast, two other nonprescription medicines, acetaminophen and ibuprofen, were listed with equal frequency for women and men with diagnosed CHD and more frequently for women without CHD (Table 3). In multivariable analyses adjusted for age, sociodemographic variables, and clinic site (Fig. 1A), women without CHD were less likely to have aspirin on their medication list than men (AOR 0.63 [0.55–0.72], P < 0.0001, n = 2,699). The results among patients with diagnosed CHD were similar to those for patients without diagnosed CHD: women with diagnosed CHD were 30% less likely than men to have aspirin listed in their medical records.

In this large cohort of patients with diabetes seen in five academically affiliated primary care clinics, one-third of whom had diagnosed CHD, women were significantly less likely than men to receive recommended treatments for several major modifiable CHD risk factors and, when treated, were less likely to achieve recommended goals of therapy. Even though all of the patients received, on the whole, above-average quality of care by NCQA standards, women did not uniformly receive as good care as men in unadjusted analyses and after adjusting for differences in age, race, clinic site, and other variables. Among patients without CHD, women were less likely than men to be prescribed lipid-lowering medications and aspirin and were less likely to have HbA_1c and LDL at goal levels. Among patients with CHD (and likely to reap the greatest benefits from treatment of CHD risk factors), women were less likely than men to be taking aspirin or to have their HbA_1c, blood pressure, or lipids controlled to recommended levels.

Women with diabetes have not experienced the same improvement in CHD events and mortality as men with diabetes and as women and men without diabetes over the last 30 years (7). Although there are probably biological and other reasons for these differences, our results suggest that differences in clinical treatment may also contribute to the excess burden of CHD in women with diabetes.

Increasingly, the medical profession and the public have been paying particular attention to heart disease in women (33). The American Heart Association recently launched its “Go Red for Women” campaign, published clinical guidelines focused on women (30), and prioritized sex-specific research goals to improve understanding of the diagnosis and treatment of heart disease in women (34). Women are increasingly aware of their own risk of cardiovascular death (35).

However, differences in treatment patterns persist. Lower adherence to treatment guidelines has been demonstrated in patients with diabetes who are older and whose conditions are more medically complicated, but sex-based differences were not examined (36,37). A Swedish study from 2004 examined the treatment of 9,375 patients with diabetes in the primary care setting and found that men had lower blood pressure, HbA_1c, and LDL levels (38), but there have been few other studies of this disparity in diabetes. Although the women in our study were older and less likely to be white than the men (differences for which the models adjusted), they had a similar prevalence of CHD risk factors.

Other studies of patients without diabetes have also found that women with dyslipidemia are less likely than men to be treated with lipid-lowering agents (39). This may be explained by the fact that women frequently have higher HDL levels. Health care providers in general may disregard elevated LDL levels, assuming a protective effect of elevated HDL levels. Emphasis on the protective effects of elevated levels of HDL in patients with diabetes may be misplaced, as some evidence shows that HDL in patients with diabetes may not be as protective as in patients without diabetes (40). Furthermore, results from the Scandinavian Simvastatin Survival Study, the Cholesterol and Recurrent Events (Investigators) Trial (CARE), and the Heart Protection Study have consistently shown a 25–35% reduction in the relative risk of cardiovascular events in patients with diabetes who were treated with statins at all levels of baseline cholesterol. Finally, the consensus recommendations for treating elevated LDL levels are identical for men and women.

No obvious justification can be given to explain the marked differences we observed in aspirin recommendations between diabetic women and men with and without CHD. We have shown in a similar population that hypertension and hyperlipidemia are managed less aggressively than hyperglycemia (41). In this cohort, it appears that female sex may additionally decrease attention to treatment of cardiac risk factors, possibly related to patient preferences or physician perceptions of risk.

Several potential limitations must be addressed. First, the possibility of differential record-keeping by sex cannot be completely eliminated, but the fact that the prevalence of many risk factors and treatments were equal between sexes whereas many of the sex differences were relatively large suggests that differential reporting is unlikely to have confounded our analysis. Similarly, nonprescription medications such as aspirin may have been undercounted and could have been recorded differentially by sex. But because acetaminophen and ibuprofen were listed more frequently for women, it seems that records of nonprescription medication are likely to reflect patient use rather than biased reporting by sex. (Although more women were taking ibuprofen, the additional number taking ibuprofen was not enough to account for the decrement in those taking aspirin.) Another concern is the timing of diagnoses and treatments, which cannot be determined in a cross-sectional study. This cohort of patients with diabetes was seen in follow-up over a 3-year period, and their hypertension, hyperlipidemia, and risk for CHD were long-standing problems. Over 85% were receiving treatment for hypertension and hyperglycemia, and more than half were treated for hyperlipidemia; data were collected at the end of the study period. Consequently, the cohort is likely to reflect patients who were actively being treated rather than those who were about to have medications initiated for newly diagnosed problems. Finally, the five different internal medicine practices, although socioeconomically diverse, were all affiliated with one urban academic health center, limiting the generalizability of this study to other settings.

In summary, this large, clinic-based cohort of patients with diabetes received generally high-quality diabetes care, yet women were less likely than men to be treated with aspirin and lipid-lowering therapy; also, their lipids, blood pressure, and hyperglycemia tended to be treated less effectively. Sex differences in CHD risk reduction may partially explain the persistent increased risk of CHD events in women with diabetes observed in national surveys. More aggressive treatment of CHD risk in women with diabetes appears to offer a specific target for improvement in diabetes care.

This study was supported by a grant from the Aetna Quality Care Research Fund, an unrestricted research grant from Aventis Pharmaceuticals, the Earle P. Charlton, Jr., Charitable Foundation, and a Career Development Award from the American Diabetes Association (to J.B.M.).