C-Reactive Protein and Glycemic Control in Adults With Diabetes
OBJECTIVE—Recent evidence suggests that poor glycemic control is significantly associated with the development of macrovascular complications of diabetes. Studies have indicated that C-reactive protein (CRP) is an important risk factor for cardiovascular disease. The purpose of this study was to determine the relation between CRP and HbA1c in a large national sample of individuals with diabetes.
RESEARCH DESIGN AND METHODS—A nationally representative sample of noninstitutionalized U.S. adults aged 17 years and over with nongestational diabetes was derived from the National Health and Nutrition Examination Survey III (1988–1994) (n = 1,018). Respondents with diabetes were stratified by HbA1c level. The main outcome measure was elevated (>0.30 mg/dl) CRP.
RESULTS—In unadjusted analyses, respondents with diabetes who had elevated HbA1c levels (≥9.0%) had a significantly higher percent of elevated CRP than people with low (<7%) HbA1c levels (P < 0.001). In adjusted regression analysis, after controlling for age, race, sex, smoking, length of time with diabetes, insulin, and BMI, HbA1c was significantly associated with an increased likelihood of elevated CRP for HbA1c >9.0% (OR 2.15, 95% CI 1.07–4.32) and for HbA1c >11.0% (4.40, 1.87–10.38). Higher HbA1c also predicted elevated CRP in the regression model when HbA1c was analyzed as a continuous variable (1.20, 1.07–1.34).
CONCLUSIONS—In this study, the likelihood of elevated CRP concentrations increased with increasing HbA1c levels. These findings suggest an association between glycemic control and systemic inflammation in people with established diabetes.
- CRP, C-reactive protein
- NHANES III, National Health and Nutrition Examination Survey III
- NSAID, nonsteroidal anti-inflammatory drug
C-reactive protein (CRP), a marker of systemic inflammation, is emerging as an independent risk factor for cardiovascular disease (1–3). High CRP levels have been linked to an increased risk of thrombotic events including myocardial infarction (3–5). Elevated CRP levels have also been linked to an increased risk of later development of diabetes (6,7). Furthermore, CRP levels are higher in people with diabetes compared with those without diabetes (8–10). Less is known about whether CRP in people with diabetes is related to level of glycemic control. Wu et al. (11) found that CRP is associated with HbA1c levels; however, people with diabetes were excluded from the study. HbA1c was 5.4% in people with low CRP and 5.5% in people with medium or high CRP (P < 0.05). Another study found an association between CRP and uncontrolled diabetes in 62 patients, but the study was limited by small a sample size (12).
To provide further insight into the role of inflammation in the development of cardiovascular disease in people with diabetes, we sought to elucidate the link between level of glycemic control and inflammation using a representative national sample. The purpose of the study was to investigate the relation between CRP and HbA1c in adults with diabetes.
RESEARCH DESIGN AND METHODS
A sample of respondents ≥17 years of age was derived from the National Health and Nutrition Examination Survey III (NHANES III), 1988–1994, a cross-sectional study of a nationally representative sample of noninstitutionalized U.S. residents. Respondents with diabetes were identified using the question, “Has your doctor ever told you that you have diabetes?” Respondents who answered positively to either of two questions regarding having diabetes “confined only to pregnancy” were excluded. Respondents with diabetes were not identified using laboratory data because blood was drawn on only one occasion and because people with diabetes may not have had elevated serum glucose at the time the blood was drawn. The respondents that had a usable sample of serum for analysis for CRP form the basis of this report.
People who had used anti-inflammatory drugs or cholesterol-lowering drugs within the previous 30 days were excluded from the analysis, due to the possible effects the drugs might have on CRP levels (13,14). The use of anti-inflammatory drugs was measured by questions regarding the use of prescription medications along with specific questions about the nonprescription use of aspirin and ibuprofen. Prescribed and over-the-counter nonsteroidal anti-inflammatory drugs (NSAIDs) and corticosteroids were classed as anti-inflammatory medications. The use of cholesterol-lowering drugs was measured by a specific question regarding whether any medication was used for the treatment of elevated cholesterol.
Dependent variable: C-reactive protein
CRP was measured between 1988 and 1994 on an ongoing basis as part of the NHANES III physical and laboratory examination. Standard phlebotomy techniques were used to obtain specimens. Serum specimens were frozen to −20°C until analyzed for CRP (Behring Nephelometer Analyzer System; Behring Diagnostics, Somerville, NJ). The lower limit of detection was truncated at 0.30 mg/dl due to the limits of the diagnostic technique used. The value of 0.21 mg/dl was assigned to participants with CRP concentrations below the detection limit of 0.3. This lower limit of detection limited statistical analysis of CRP to elevated or nonelevated, because CRP was noncontinuous across the full range. Further details about the specific methods for laboratory procedures in the NHANES III are available elsewhere (15).
CRP level was evaluated as a dichotomous variable (elevated or not elevated). An elevated level of CRP was defined prospectively, using a cutoff (>0.30 mg/dl) based on previous studies of cardiovascular disease and NHANES III data (11,16).
Independent variable: HbA1c
HbA1c was measured in the NHANES III study by the Diabetes Diagnostic Laboratory at the University of Missouri using the Diamat Analyzer System (Bio-Rad Laboratories, Hercules, CA) (17). This ion-exchange high-performance liquid chromatography system measures HbA1c and has demonstrated excellent, long-term precision (interassay correlation value 2.0). It was standardized to the reference method that was used for the Diabetes Control and Complications Trial. Further information regarding the specific methods used is available elsewhere (15).
Demographic indicators (age, race, and sex) were included as control variables and categorized as in similar epidemiological studies, because CRP levels are known to vary according to these factors (17). In an effort to determine the independent relation between HbA1c level and CRP, additional variables were included that have been shown to influence the level of CRP, including smoking and BMI (8,9,18). Smoking was coded three ways: current smoker, ex-smoker, and never smoker. BMI was calculated using the weight (in kilograms) divided by height (in meters squared) found in the examination file of the NHANES III and dichotomized at above or below 27 according to risk guidelines of the American Diabetes Association (19). Length of time with diabetes was included as a control variable to account for the greater likelihood of cardiovascular disease (and hence greater likelihood of elevated CRP) with a longer duration of diabetes. Fasting insulin levels were available on participants with diabetes and were included in the regression models as a possible confounding variable. Insulin was measured as a continuous variable in micromoles per milliliter.
All analyses were performed using SUDAAN to account for the complex sampling design of the NHANES III. Respondents were stratified into levels of glycemic control according to level of HbA1c in 2% increments (HbA1c <7.0, 7.0–8.9, 9.0–10.9, and >11.0%). We first performed descriptive statistics to compare the demographic characteristics of the HbA1c groups using χ2 and t tests. We then compared the percent of respondents with elevated CRP for each 2% increment of HbA1c using χ2.
Next, an adjusted model was constructed using elevated CRP as the response and HbA1c (continuous) as the predictor, controlling for the following variables: age, race, sex, smoking, length of time with diabetes, and BMI. Standardized β, P values, odds ratios, and 95% CIs were obtained from the logistic regression output. Statistical significance was defined as P ≤ 0.05.
A total of 1,614 adults answered yes to the question regarding diabetes in the NHANES III database. After excluding 152 for gestational diabetes and 444 who were taking excluded drugs, 1,018 adults with nongestational diabetes had CRP levels available and were included in this analysis. The demographic and general health characteristics of the participants are described overall and according to level of HbA1c in Table 1. Mean percent elevated CRP values also are shown for each HbA1c group.
The percent of participants with elevated CRP was compared at different levels of HbA1c. For each level of HbA1c, the percent of elevated CRP was as follows: <7, 48.9%; 7–8.9, 45.4%; 9–10.9, 60.7%; and ≥11, 70.6%. Overall, 51.5% of participants had elevated CRP. In unadjusted analyses, increasing HbA1c was significantly associated with a higher percent of patients with elevated CRP levels (P < 0.05).
Results of adjusted logistic regression analyses using HbA1c as the predictor variable are shown in Table 2. After controlling for demographic variables, smoking, BMI, fasting insulin level, and length of time with diabetes, HbA1c >9% and >11% remained a significant predictor of elevation of CRP (Table 2). We also evaluated HbA1c in the same model using it as a continuous variable. Again we found that HbA1c was a significant predictor of elevated CRP (1.20 OR, 95% CI 1.07–1.34), which means that for every 1.0% increase in HbA1c there was a 20% increase in the likelihood of having an elevated CRP.
The present study demonstrated that a higher HbA1c is significantly associated with a greater likelihood of higher CRP among adults with diabetes. The relation was significant in unadjusted comparisons of the percent of people with elevated CRP according to HbA1c level and in logistic regression models to predict elevation of CRP after controlling for age, race, sex, smoking, BMI, insulin level, and length of time with diabetes.
Inflammatory markers such as CRP have been related to the development of insulin resistance and type 2 diabetes (6,7,11,20–22). Previous research has also established that CRP levels are higher in people with diabetes (8) and associated with HbA1c in people without diabetes (11). The results of the current study go a step further with the finding that among people with established diabetes, at successively higher levels of HbA1c the percent of people with CRP >0.30 mg/dl is significantly higher. The main implication of these findings is that inflammation may not only be implicated in the development of diabetes, but also in ongoing levels of hyperglycemia once diabetes is established.
Due to the cross-sectional design of the study, we cannot infer from these results a cause and effect relation, i.e., whether poor glycemic control leads to inflammation or whether inflammation leads to higher glucose levels (or whether a third factor influences both). Prospective studies are needed to evaluate that question. However, either direction of causality would have important implications. If poor glycemic control leads to inflammation, then better glycemic control should lower inflammation and therefore lower the risk of cardiovascular complications. If inflammation leads to poor glycemic control, then treatment of inflammation with NSAIDs or hydroxymethylglutaryl-CoA reductase inhibitors may help improve glycemic control. In light of recent findings of an association among inflammatory proteins, endothelial dysfunction, and insulin resistance (23–25), the results of the current study provide additional support for a relation between glycemic control and systemic inflammation in people with established diabetes.
Recent research evidence supports a link between hyperglycemia and inflammation. CRP is known to be higher in people with impaired glucose tolerance and frank diabetes (8,11). Furthermore, increased CRP has been found to be a risk factor for later development of diabetes (6,20,21). Festa et al. (22) found links between CRP and insulin resistance. Other studies have related hyperglycemia to inflammation by demonstrating simultaneous inflammation, endothelial dysfunction, and insulin resistance at the physiologic level (23,24). One of the several mechanisms proposed is oxidative stress on the endothelium, which promotes inflammation and is enhanced by hyperglycemia (26–28). Such evidence is consistent with the findings in the current study, which further documents the association between hyperglycemia and inflammation in adults with diabetes.
Limitations of this study include the fact that much of the data were by self-report, including the diagnosis of diabetes, use of anti-inflammatory medications, and smoking status. However, self-report questions have shown good agreement with other measures in previous studies and have proved useful (29,30). Furthermore, the CRP measure used is a different and older technique than the highly sensitive CRP assay developed more recently (2,5). However, the method used was also used and validated in the Diabetes Control and Complications Trial (16). In addition, the current study focused on elevated levels rather than the lower levels detectable by newer methods.
A further limitation of this study is that we were unable to control for the use of thiazolidinedione drugs that can affect CRP (31). However, such medications were not in common use at the time of the survey and were not included in the questions in the NHANES III about medication use.
The current study demonstrates that higher HbA1c is significantly associated with elevation of CRP. These results imply a significant relation between inflammation and glycemic control in people with established diabetes. Prospective studies should be conducted to determine the direction of this association; such research would have important implications for the treatment of adults with diabetes.
This study was supported in part by Grant 1D12HP00023-01 from the Health Resources and Services Administration and by an Advanced Research Training grant from the American Academy of Family Physicians.
Address correspondence and reprint requests to Dana E. King, MD, Department of Family Medicine, P.O. Box 250192, 295 Calhoun St., Medical University of South Carolina, Charleston, SC 29425. E-mail:.
Received for publication 5 September 2002, and accepted in revised form 16 January 2003.
A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.
- DIABETES CARE