Personal attributes that influence the adequate management of hypertension and dyslipidemia in patients with type 2 diabetes. Results from the DIAB-CORE Cooperation

Background

Hypertension and dyslipidemia constitute major public health problems as they increase the risk of cardiovascular diseases (CVDs), especially in patients with concomitant type 2 diabetes (T2D)[1–4].

Nevertheless, several epidemiological studies indicate disappointing deficiencies in the detection and adequate treatment of hypertension and dyslipidemia in Germany with minimal improvement over the last decades[5–13]. About ten years ago, the South German site of the Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) Study, the forerunner to the Cooperative Health Research in the Region of Augsburg (KORA) Study, revealed that the proportion of hypertensive participants with controlled hypertension < 140/90 mmHg was only 7% in men and 13% in women[6]. An earlier analysis of regional differences using data of the two population-based German studies KORA and SHIP (Study of Health in Pomerania) with participants aged 25-74 years found that of all hypertensive participants, 8.4% in SHIP and 10.2% in KORA, were treated and well controlled. Recently, data of the German DIAB-CORE cooperation revealed that 64% of all participants with T2D and 49% of participants without T2D, aged 45-74 years, had untreated or insufficiently treated hypertension[8]. Despite the increased awareness of physicians and the public regarding the beneficial effects of blood pressure control on cardiovascular outcomes such as myocardial infarction and stroke ([5]), the data suggest that further barriers must exist that hinder optimal health care delivery.

Objectives

To our knowledge, no population-based study in Germany has so far dealt with characteristics of patients with T2D and hypertension and/or dyslipidemia who are not sufficiently treated for these serious conditions. In our study, we describe demographic, socio-economic and lifestyle factors in DIAB-CORE participants with diabetes and concomitant hypertension or dyslipidemia and analyse the factors associated with insufficient disease control and absent pharmacotherapy. We therefore aim to identify the group of patients with the most urgent need for intensified care.

Study design and setting

All studies were conducted between 1997 and 2006 and used similar instruments, questionnaires and medical measurements to assess data. Detailed descriptions of study designs, samples and procedures are available elsewhere[14–20]. Ethical approval was obtained for each study. Primary study data of interest were pooled and frequencies compared.

Variables

Participants

The pooled data set included 1,287 participants with T2D aged 45 to 74 years. In the nationwide survey GNHIES98, 3% non-German citizens were included, KORA, SHIP and CARLA focused on participants of German nationality, and the other studies collected information on birthplace only. Two-stage cluster sampling or stratified random sampling were used. Overall response ranged between 56 and 69%.

The hypertension sub-analysis included 1,048 hypertensive participants. N = 17 participants were excluded due to missing information on: study blood pressure measurement (n = 3), physician’s diagnosis of hypertension (n = 3), and/or medication intake (n = 14).

The dyslipidemia sub-analysis included 636 participants with dyslipidemia according to the definition specified above. N = 126 participants had to be excluded due to missing information on laboratory measurements (n = 32), physician’s diagnosis of dyslipidemia (n = 90, all participants of DHS and 3 of HNR), and/or medication intake (n = 14). N = 16 participants were not included in either one of the two sub-analyses.

Statistical analyses

T2D participants with hypertension, with or without the blood pressure target of < 140/90 mmHg, and T2D participants with dyslipidemia with or without the target of total cholesterol/HDL cholesterol < 5, as well as T2D participants with or without treatment were compared with respect to their lifestyle factors, cardiovascular burden, clinical measurements and medications. Continuous variables were characterized by means and standard deviations (SD), categorical variables were described as percentages. Differences between groups were calculated using t-tests and Wilcoxon tests (continuous variables) or chi-square tests and univariable logistic regression models (categorical variables). At first, univariable logistic regression models were programmed to identify factors associated with blood pressure, lipid ratio or medication intake. In a second step, variables found to be significantly associated with the respective outcome were included in multivariable models to examine adjusted effects.

The goodness-of-fit of adjusted models was assessed by the c-value, and the Hosmer-Lemeshow test. The c-value describes the area under the ROC-curve, ranging from 0.5 (random correlations) to 1.0 (perfect fit). In epidemiologic studies, values between 0.6 and 0.8 are usually regarded as satisfactory. The Hosmer-Lemeshow test compares observed and predicted values with each other. If there is no significant difference (at the significance level of 0.05) the model is characterized as being appropriate.

A two-sided alpha level of 0.05 was chosen as criterion for statistical significance. All analyses were carried out using SAS, version 9.2 (SAS Institute Inc., Cary, NC, USA).

Participants

Of 1,287 T2D participants aged 45-74 years, n = 1,048 (82.5%) had hypertension and n = 636 (54.8%) had dyslipidemia. N = 530 (41.2%) had both conditions and were included in both sub-analyses.

The frequency of hypertension in persons with T2D differed slightly between studies, ranging from 78.5% in HNR to 89.6% in CARLA. Likewise, the number of participants with dyslipidemia ranged from 48.6% in CARLA to 59.6% in GNHIES98. The mean age was 63.1 years (Std. 7.2 years), and was very similar across all studies (Table 1). In total, 45.1% of participants were female, ranging from 40.0% in HNR to 49.8% in GNHIES98. Within the group of hypertensive participants, 46.3% were female, among participants with dyslipidemia, 43.1% were female.

Descriptive data

Of 1,048 participants with T2D and hypertension, n = 240 (22.9%) had controlled blood pressure of < 140/90 mmHg, and n = 808 (77.1%) were uncontrolled. Dyslipidemia was treated to TC/HDL ratio < 5 in 143 (22.5%) of 636 T2D participants and uncontrolled dyslipidemia was observed in 493 (77.5%) T2D participants (Figure 1).

Focus on differences between women and men

Systolic blood pressure (SBP) increased with age in participants with T2D, while diastolic blood pressure (DBP) decreased. Both measurements were consistently lower in women of all age groups than in their male counterparts (mean SBP in women: 149.8 mmHg (SD 20.8) vs. 154.4 mmHg (20.6) in men, mean DBP 83.2 (11.1) and 86.6 (11.2) in women and men, respectively) (Figure 2). Whilst men were more often free from hypertension (19.3% vs. 15.2%), women were significantly more often diagnosed and well controlled if affected: 25.0% of all women had controlled hypertension, compared to 13.9% in men, 43.5% were treated but did not achieve goal levels (compared to 38.8% in men), 5.9% were not treated even though they had a physician’s diagnosis of hypertension (compared to 7.3% in men), and 10.3% had unknown hypertension (compared to 20.6% in men) (see Figure 3).

Total cholesterol and HDL-cholesterol were higher in women than in men (232.2 mg/dl (SD 48.8) vs. 216.4 mg/dl (48.0) and 53.9 (16.6) vs. 46.3 (14.0)) – albeit, 48.7% of women and 35.1% of men had less than 50 or less than 40 mg/dl HDL respectively. However, the TC/HDL ratio was lower in women (4.7 (1.7) vs. 5.0 (1.7)).

More women than men were not affected with dyslipidemia as defined by a TC/HDL ratio < 5 (47.5% vs. 43.4%), 14.9% were treated and well-controlled (vs. 10.2% in men), 7.9% were treated but did not reach goal values (compared to 6.7% in men), 16.3% were not treated despite of a physician’s diagnosis (compared to 17.8% in men) and 13.4% had unknown dyslipidemia (compared to 21.9% in men) (see Figure 3).

Factors associated with uncontrolled hypertension or uncontrolled dyslipidemia

Univariable logistic regression models yielded significant associations of uncontrolled hypertension with sex (OR for men: 2.01, 95% CI 1.50-2.69), BMI < 30 (OR = 1.89, 95% CI 1.40-2.55), and study. Age in five groups was not significantly associated but displayed a steady tendency towards having the greatest odds in the second oldest group (65-69 years) compared to the youngest group (45-54 years). With GNHIES98 as the reference study, HNR had the smallest odds of uncontrolled hypertension (OR = 0.32, 95% CI 0.20-0.50), followed by CARLA (OR = 0.47, 95% CI 0.28-0.79) and KORA (OR = 0.52, 95% CI 0.30-0.91). Multivariable analysis with these significant variables (i.e. sex, BMI and study) showed consistent results as well as the associated factor ‘no previous MI’ which reached statistical significance (OR = 1.70, 95% CI 1.06-2.70). A model including only sex, BMI and study had a c-value of 0.68 and a non-significant Hosmer-Lemeshow test (p = 0.78).

Factors associated with no pharmacotherapy for hypertension or dyslipidemia

Using univariable regression models, we found significant odds of non-use of anti-hypertensive medication in younger study participants (OR for 45-54 years: 2.54, 95% CI 1.54-4.20, OR for 55-59 years: 2.15, 95% CI 1.31-3.53, 70-74 years as reference), males (OR = 2.11, 95% CI 1.54-2.87), current smokers (OR: 1.80, 95% CI 1.22-2.64) and participants with BMI < 30 (OR = 2.06, 95% CI 1.52-2.79). Moreover, no previous MI (OR = 3.75, 95% CI 1.79-7.82), no previous stroke (OR = 3.12, 95% CI 1.42-6.88) and study (OR for CARLA: 0.42, 95% CI 0.24-0.71, OR for HNR: 0.60, 95% CI 0.40-0.92, GNHIES98 as reference) were significantly associated with untreated hypertension. In multivariable models adjusted for age, sex, BMI, MI, stroke, and study, the odds remained significant, except for smoking. A model including only these significant variables was characterized by a c-value of 0.71 and a non-significant Hosmer-Lemeshow test result (p = 0.46).

Key results

Uncontrolled blood pressure in T2D participants with (diagnosed or unrecognized) hypertension was associated with male sex, BMI < 30 kg/m2, no previous MI, and study site. Similarly, uncontrolled blood lipid levels in T2D participants with dyslipidemia (diagnosed or unrecognized) were more frequent in men, those who had not suffered MI and attendees of the older DIAB-CORE studies.

Lack of treatment for hypertension was related to younger age, male sex, smoking, BMI < 30, no history of MI or stroke, and study site. Male T2D participants without concomitant hypertension and who had not suffered MI had significantly greater odds of untreated dyslipidemia than other participants in DIAB-CORE.

Socioeconomic features such as educational level and income were not significantly associated with either disease control or pharmacotherapy; lifestyle factors were only associated in some models.

On the one hand, the results of our analyses indicate that the vast majority of patients with T2D in Germany are not adequately treated for hypertension and/or dyslipidemia, even though these conditions have been found to considerably increase the risk of co-morbidities and complications such as MI, stroke, nephropathy and retinopathy.

On the other hand, we confirmed the presumption that patients with additional risk factors, such as advanced age, previous MI, previous stroke or obesity are treated more often for hypertension and dyslipidemia and (apart from those with older age) more frequently achieve adequate blood pressure and lipid target levels. The difference between female and male participants was pronounced and consistent over all sub-analyses.

Strengths and limitations

The essential strength of our study is the large population-based sample drawn from the general German population, aged 45 to 74 years, and the fact that both, laboratory measurements and information on medication intake were available.

Differences in the frequencies of hypertension and dyslipidemia between studies were probably due to the relatively small numbers of affected participants within the individual studies and analyses. Due to the pooling process, only similarly collected and coded data of all six studies could be used and the least common denominator had to be found. Therefore, the definition of diabetes was based on self-report of physician’s diagnosis and treatment with anti-diabetic agents rather than on clinical diagnosis and medical records. Blood pressure was calculated using the mean of the second and third measurements in all studies except for DHS, where only two measurements were performed and used to calculate the mean.

Moreover, measurements of blood pressure and lipids based on a single testing opportunity provide evidence for the respective condition, but are not equal to a clinical diagnosis with repeated measurements. We cannot exclude cases of ‘white coat hypertension’, i.e. elevated blood pressure caused by the excitement of the unfamiliar situation.

Due to compliance issues, women might have reported their medication intake more reliably than men, thus pharmacotherapy in men could have been underestimated and consequently, sex-specific differences overestimated in these analyses.

Generalization

Patient-centered studies in Germany and other countries have so far examined awareness, treatment, control and factors associated with insufficient control of co-morbidities in patients with and without diabetes. A systematic review by McLean et al. from 2008, which included 26 studies from different countries with 66,833 diabetes patients with co-morbid hypertension, concluded that 83% (range 32-100%) of patients were treated, yet only about 29% (range 5-59%) had their blood pressure controlled to < 140/90 mmHg. The proportions of treatment to control were similar and thus equally insufficient between studies and countries. Unfortunately, the authors did not report person-related factors associated with disease control[24].

Differences by gender

There are very few studies that have reported factors associated with cardiovascular disease control in patients with diabetes, as opposed to adults in general, irrespective of diabetes status. However, a number of recent studies that focused on gender differences[25–30] found that female patients with diabetes had a worse cardiovascular risk profile and were less controlled compared to their male counterparts. Our study confirmed these results in part; women in DIAB-CORE had higher total cholesterol values than men. However, since HDL-cholesterol values were also higher, the TC/HDL ratio was lower and thus more beneficial in women. In all studies[25–30], blood pressure was higher in women, especially in patients with cardiovascular disease. Although this sex difference was abrogated in some analyses after adjustment for associated variables[27, 28]. There was either no sex difference in the amount of antihypertensive and lipid-lowering medication, or women took more medication, which is in agreement with our study. Furthermore, our results indicate that men are about twice as often unaware of their hypertension and dyslipidemia as women, which is well in line with known gender differences concerning health behaviour. Especially in the age group of about 30 to 60 years, men use health care services less than women and tend to undervalue health care and health behaviour[31, 32]. Several authors hypothesize that societally dominant ‘traditional masculinity’ leads men to adopt beliefs and behaviours that increase health risks and support the ideal of the ‘bulletproof superhero’ who would be embarrassed to check his cholesterol level[33–35].

Influence of age

Findings on the influence of age on hypertension control are controversial (e.g.[36, 37]). An explanation for less control with simultaneously intensified treatment, as seen in our study, might be that resistant hypertension occurs more often in older persons[38]. Accordingly, this association was not detected for dyslipidemia. A cross-sectional study in Sweden examined the assumption that the excess cardiovascular risk of persons with diabetes compared to persons without diabetes decreases with increasing age. However, the authors found that the burden of CVD risk factors clustered over the entire life span with increasing glucometabolic disturbance, especially in older women. Likewise, self-rated health decreased with increasing cardio-metabolic risk and age. The authors suggest that a decreased burden of risk in older patients with diabetes might be due to a survival bias[39].

Association with body mass index

In our study obese participants (BMI ≥ 30 kg/m2) received medication more frequently and were more often well treated for hypertension, although they were more often affected with hypertension than leaner individuals (data not shown). A Swedish population-based study in hypertensive 60-year-old persons with and without diabetes by Carlsson et al. found an inverse, direct association of high waist circumference with uncontrolled hypertension[40]. The same seems to be true for patients in primary care irrespective of diabetes status[41]. Bramlage et al. found that the odds of good blood pressure control in diagnosed and treated individuals was significantly smaller in overweight and obese primary care patients than in patients with normal weight. Using data of physician diagnosed hypertensive DIAB-CORE participants without diabetes (n = 5012), we found that 29.2% of those with BMI < 25 had controlled hypertension (41.2% were uncontrolled and 29.6% untreated) compared to 28.6% of those with BMI ≥ 30 (50.9% uncontrolled and 20.5% untreated). These results indicate that obese people may be more difficult to treat and obtain goal blood pressure than lean persons; however, they appear to be more aware of their disease. The same seems to be true for a comparison of hypertensive people with and without T2D. Those with T2D are more often treated and well-controlled, much more often treated but not controlled, half as often untreated and half as often unrecognized, irrespective of BMI group (data not shown) than those without T2D. All in all, obese individuals do more often present with a blood pressure ≥ 140/90 mmHg, irrespective of treatment. Thus, people with additional risk factors (such as obesity and diabetes) may be more aware of their blood pressure and more often treated but at the same time, they are more difficult to effectively treat.

Effect of socio-economic differences

We expected to find a significant negative association of high socio-economic status with uncontrolled co-morbidities, however no significant associations were observed with any socio-economic characteristics and the outcome variables. This might, in the case of school education, be due to the unequal frequencies of low (about 80%) vs. high and middle status, which owns to the high mean age of the study population (63 years) and the fact that older people mostly attended junior high school only. In the previously mentioned Swedish study, lack of health care due to low income was independently associated with uncontrolled hypertension in men (OR = 2.71, 95% CI 1.09-6.78) but not in women. In contrast, living in an apartment instead of a house (as an indicator of lower socio-economic status) remained a significantly protective factor in an adjusted model in women (OR = 0.55, 95%CI 0.35-0.85). The authors stated that the finding was puzzling and offered no explanation[40].

Association with previous cardiovascular disease

Carlsson et al. confirmed that previous cardiovascular disease or coronary heart disease has a protective effect on uncontrolled hypertension, probably because of more intense treatment, better medication adherence and/or symptom relief of angina pectoris[40].

Influence of more complex factors

There are a number of possible influences on uncontrolled hypertension and dyslipidemia that we could not consider. For example, the Swedish study by Carlsson et al. included data on nutrition and found that daily intake of fruit had an independent, protective effect in men but not in women[40].

Steckelings, 2004 et al.[5] used data from primary care patients, irrespective of diabetes status (the HYDRA Study: Hypertension and Diabetes Screening and Awareness Study), to describe possible determinants of unsatisfactory hypertension control in Germany. Less than 30% of treated and 19% of all patients, treated or untreated, had controlled blood pressure <140/90 mmHg. The frequency of diagnosis was particularly low in young people, probably due to insufficient blood pressure screening. The authors found that physicians used outdated guidelines and based treatment on diastolic pressure. The great majority of participants (94%) stated that they knew that hypertension is an important risk factor for serious diseases, and most of them (63%) occasionally measured their own blood pressure. However, the physicians participating in the study often misclassified their patients as ‘well controlled’ even though they had BP measurements ≥ 140/90 mmHg.

According to a review by Düsing et al. 2006[42], more complex, hindering factors could also be insufficient education and motivation provided to the patient by physicians, patients’ reluctance to change lifestyle factors or commence/modify drug treatment, lack of awareness of the risks associated with hypertension, and poor compliance. The latter is a complex and thoroughly explored concept and challenge (e.g.[43]). Apparently, many people with hypertension do not seem to recognize high blood pressure as a progressive chronic illness, but rather misinterpret it as a risk factor in a gamble with a potentially positive outcome[44, 45].

Finally, the sex of the attending physician has been shown to play a role in the quality of risk factor control in patients with hypertension and dyslipidemia. In a cross-sectional study by Journath et al., diabetic men and women achieved goals for blood pressure control, and men achieved goals for cholesterol control more often if they were treated by female physicians[46]. Likewise, a German study including 51,053 diabetes patients treated by 3,096 physicians concluded that female physicians achieved a better quality of care than their male counterparts, especially in risk management important for future disease prognosis[47].

Differences between Europe and the United States/Canada

Interestingly, there appears to be an intriguing difference in hypertension prevalence and control, irrespective of diabetes status, between Europe and the United States/Canada that has not been appreciated and examined sufficiently. Wolf-Maier et al.[48, 49] compared sample surveys conducted in the 1990s in Germany, Finland, Sweden, England, Spain, Italy, Canada, and the United States and found that even though mean BMI was very similar across these countries, the prevalence of hypertension, defined as blood pressure > 140/90 mmHg or intake of anti-hypertensive medication, differed remarkably. The European average was 44.2% compared with 27.6% in North America. Germany had the highest prevalence with 55.3%[48]. Treatment and control of hypertensive participants within the individual studies was also significantly better in the United States than in Europe with 7.8% hypertension control in the population (29.9% in treated hypertensive participants) in Germany compared to 28.6% (54.5% in treated hypertensives) in the United States[49].

In order to improve awareness, creative new approaches have been successfully implemented in the United States, e.g. by addressing the issue and counseling black men during a visit at a barbershop[50]. Such innovative ideas might also work in Germany – possibly for younger men with diabetes and no previous cardiovascular complications.