- Top of page
- Statistical analysis
Objectives: We assessed the independent contribution of C-reactive protein to the risk of cardiovascular disease in Aboriginal Australians.
Methods: High sensitivity CRP levels were measured in 705 Aboriginal participants aged 20–74 years free from CVD at baseline. Participants were followed for a median of 11 years. Cox proportional hazards models were used to assess the association of CRP with the risk of developing CVD events.
Results: A total of 114 participants were diagnosed with CVD. Incidence rates were 5.4 and 21.4 per 1,000 person-years for the lower (<3 mg/l) and the higher (≥3 mg/l) CRP groups, respectively. After adjusting for age, sex, total cholesterol, systolic blood pressure, smoking status, diabetes, BMI and waist circumference, the association between CRP and CVD remained significant, with a hazard ratio of 2.40 (95% CI: 1.25, 4.62) for the higher CRP group relative to the lower CRP group. The population attributable risk was 52% (95% CI: 14%, 74%).
Conclusions: CRP is an independent predictor of CVD in Aboriginal people. A large proportion of CVD cases are associated with elevated CRP levels. Therefore, controlling the conditions that cause inflammation may be beneficial to cardiovascular health in Aboriginal communities.
- Top of page
- Statistical analysis
We used a CRP cutpoint of 3 mg/L to differentiate high-risk and low-risk groups to be consistent with the recommendations from the Centers for Disease Control and Prevention and previous studies.14 We also divided CRP values into quartiles to explore the dose-response relationship. Cardiovascular disease incidence rates were calculated by dividing the number of first-ever either fatal or non-fatal cardiovascular events by the person-years of follow-up according to the baseline CRP levels. Hazard ratios and their 95% conference intervals (CI) were estimated using the Cox proportional hazards model, adjusting for potential confounding factors of age, sex, smoking, blood pressure, body mass index, waist circumference, serum cholesterol, urinary albumin to creatinine ratio and diabetes status. Measurements of those factors have been described elsewhere.15 HDL was not added in the final model because low HDL was not associated with CVD and adding HDL had little impact on the association between CRP and CVD in the study population. The crude and adjusted hazard ratios for the high-risk group (CRP >3 mg/L) were calculated using the low-risk group (CRP ≤3 mg/L) as the reference. We also took the CRP as a continuous variable (transformed by the logarithm base 2) to calculate the hazard ratios for CVD corresponding to the doubling of CRP values. All analyses were performed using Stata version 9.16
The project was approved by the Behavioural and Social Science Ethical Review Committee of the University of Queensland and the community's Health Board and Land Council.
- Top of page
- Statistical analysis
Table 1 shows the baseline characteristics of study participants with high (>3 gm/L) and low (≤3 gm/L) CRP levels. As expected, the high CRP clustered with a number of traditional risk factors. People with high CRP had higher levels of body mass index, waist circumference, total serum cholesterol, triglycerides, diastolic blood pressure, and urinary albumin to creatinine ratio, and they are more likely to be females and to have type 2 diabetes than those with low CRP.
Table 1. Characteristics of study participants by C-reactive protein levels.
| ||CRP <3.0 mg/L||CRP ≥3.0+ mg/L||p-value|
|Age, yearsa||29.1 (10.6)||34.9 (12.1)||<0.01|
|BMI, kg/m2a||21.2 (3.9)||24.3 (5.3)||<0.01|
|Waist circumference, cma||80.1 (10.4)||91.3 (13.8)||<0.01|
|Systolic pressure, mmHga||119.9 (16.5)||121.5 (19.1)||0.28|
|Diastolic pressure, mmHga||72.7 (13.1)||74.7 (14.1)||0.07|
|Total cholesterol, mmol/La||4.4 (1.1)||4.7 (1.2)||<0.01|
|HDL cholesterol, mmol/La||1.15 (0.27)||1.06 (0.24)||<0.01|
|Triglycerides, mmol/Lb||1.65 (1.53, 1.79)||1.89 (1.79, 1.99)||<0.01|
|C-reactive protein, mg/Lb||1.48 (1.38, 1.59)||8.94 (8.36, 9.57)||<0.01|
|Urine ACR, mg/mmolb||1.98 (1.55, 2.53)||5.58 (4.65, 6.68)||<0.01|
|Male (%)c||152 (68.2)||212 (44.0)||<0.01|
|Diabetes (%)c||18 (8.1)||73 (15.1)||<0.01|
|Smoking (%)c||157 (70.4)||354 (73.4)||0.40|
|Drinking (%)c||142 (63.7)||260 (53.9)||0.02|
Of 705 participants, more than two-thirds had CRP levels >3 mg/L, only less than 32% (223) had CRP levels less than 3 gm/L and less than 8% (54) had CRP levels 1 mg/L or less.
The numbers of CVD events and incidence rates according to the baseline CRP levels are summarised in Table 2. During the 7,111.6 person-year follow-up, 114 first-time primary cardiovascular events including 77 coronary heart disease events were identified. The incidence rates according to baseline CRP quartiles are shown in Figure 1. Participants with higher CRP levels had higher incidence rates. Adjusting for age, sex, total cholesterol, systolic blood pressure, smoking, body mass index and waist circumference, the difference in cardiovascular disease cumulative incidence between the higher CRP group and the lower CRP group remained significant (Figure 2).
Table 2. Cardiovascular events and incidence rates according to baseline C-reactive protein levels.
|CRP||CVD events||Person-years||Rate (95% CI), per 1000pys|
|1st||13||1916.4||6.8 (3.9, 11.7)|
|2nd||29||1832.4||15.8 (11.0, 22.8)|
|3rd||33||1734.7||19.0 (13.5, 26.8)|
|4th||39||1628.1||24.0 (17.5, 32.8)|
|High vs low|
|<3 mg/L||13||2395.6||5.4 (3.2, 9.3)|
|≥3 mg/L||101||4715.9||21.4 (17.4, 25.8)|
|Total||114||7111.6||16.0 (13.4, 19.3)|
Table 3 shows the hazard ratios of cardiovascular events according to baseline CRP levels. High CRP was significantly predictive of the risk of future cardiovascular events. As expected, adjusting for potential confounding factors attenuated the associated between CRP and the risk of cardiovascular disease. However, the association remained positive and significant after adjustment of potential confounding factors of age, sex, total cholesterol, systolic blood pressure, smoking status, diabetes, BMI and waist circumference. The risk of cardiovascular disease for those with CPR >3mg/L was higher than that for those with CRP ≤3mg/L (Hazard ratio = 2.40 and 95% CI: 1.25, 4.62). The corresponding adjusted hazard ratios (95% CI) for conventional risk factors were 2.15 (1.40, 3.28), 2.09 (1.42, 3.09), 1.42 (0.93, 2.17) and 1.40 (0.86, 2.27) for diabetes, hypercholesterolemia, hypertension and smoking, respectively. A significant monotonic trend was observed with a 24% (95% CI: 6%, 44%) increase in the risk of cardiovascular disease corresponding to a doubling of CRP values.
Table 3. Hazard ratios of cardiovascular events according to baseline C-reactive protein levels.
| ||Hazard ratios (95% CI)|| |
|2nd||2.33 (1.21, 4.49)||1.48 (0.72, 3.04)||9 (-0.5, 18)|
|3rd||2.80 (1.48, 5.33)||1.23 (0.59, 2.56)||8 (-3, 18)|
|4th||3.44 (1.83, 6.46)||1.79 (0.88, 3.63)||18 (7, 29)|
|High vs low|
|≥3 mg/L||3.94 (2.21, 7.02)||2.40 (1.25, 4.62)||52 (14, 74)|
|Doubling CRP valuesd||1.34 (1.19, 1.50)||1.24 (1.06, 1.44)||-|
Population attributable risk is the reduction in incidence that would be observed in the whole population if all individuals in the population had CRP values within the reference (unexposed) group. We calculated the population attributable risk with CRP <3 mg/L as the reference. After adjusting for age, sex, total cholesterol, systolic blood pressure, smoking status, diabetes, BMI and waist circumference, as shown in Table 3, the population attributable risk was as high as 52% (95% CI: 14%, 74%).
- Top of page
- Statistical analysis
In this cohort study of Aboriginal Australians, CRP was found to be an important and significant predictor of the risk of future cardiovascular events. Due to the extremely high CRP level, the independent predictive value of CRP for the risk of cardiovascular events in Aboriginal Australians has important public health implications. As the high risk of cardiovascular disease in Aboriginal population cannot be fully explained by traditional risk factors,5 CRP may partly contribute to the disparities in cardiovascular risk among different populations. The combination of the high CRP level and the strong association between CRP and CVD in this population makes CRP an important contributing factor to a large proportion of CVD cases.
Since Ridker et al. reported that baseline plasma concentration of CRP predicted the future risk of cardiovascular events in apparently healthy men in 1997,6 the phenomenon has been observed in several populations.17–21 However, few studies have examined the association between CRP and cardiovascular disease in Indigenous populations. Plasma levels of CRP <1 mg/L, 1 to 3 mg/L, and >3 mg/L have been established as representing lower, average and higher cardiovascular risk for western populations, respectively. The relative risk for CVD among those with CRP >3 mg/L is much higher in Aboriginal people than the corresponding relative risk estimates in other populations.22 In addition, the CRP level in this population is substantially higher than in other populations. In this study, more than 68% study participants had CRP >3 mg/L while corresponding value is less than a quarter in the Physicians’ Health Study6 and about 35% males and 41% females in the Framingham cohort.23 The number of participants with a CRP level <1 mg/L was very small, constituting only 8% of the study participants, while in the Framingham cohort it constituted over 38%. We therefore used a combined group of CRP ≤3 mg/L (CRP <1 gm/L and CRP 1 to 3 gm/L) as the reference group. Although the reference group in this study has a higher cardiovascular risk than the reference of CRP <1 mg/L used in other studies, the high CRP group still had 2.4 times the risk of cardiovascular events as the reference group. For example, the 10 year absolute risk for a person with a low CRP is 5%. This person's risk could be as high as 12% if he or she had a high CRP given other conventional risk factors remain the same.
A unique phenomenon in the study population is the combination of a strong CRP-CVD association and an extremely high CRP level. As more than two-thirds of Aboriginal people are in high risk group (CRP >3 mg/L), the population attributable risk is substantially high given the association is causal. Due to a relative small sample size in our study, the 95% confidence intervals of the population attributable risk in this study were wide. Nevertheless, more than half of cardiovascular events are associated with a high CRP.
Although more research is needed to establish a causal relationship between CRP and cardiovascular disease, it has been shown that CRP directly participates in the process of atherogenesis by modulating endothelial function.24 Whether lowering CRP concentration in human populations will result in reduction in cardiovascular disease risk remains to be investigated. Effective interventions targeting CRP concentration is not clear. Using animal models, Pepys et al. designed a specific small-molecule CRP inhibitor that blocks the adverse effects of human CRP in rats with acute myocardial infarction.25 Their study suggests that early therapeutic inhibition of CRP might be beneficial for heart attack patients. Lifestyle changes including exercise and diet can also decrease CRP levels.26,27
For the purposes of developing effective intervention programs at the population level, one of our priorities should be to understand why this population has such a high CRP level. Skin infection is common in Aboriginal people living in tropical regions, in which the prevalence of skin infection in adults was 25%.28 The most important skin infection is Aboriginal community in northern Australia is streptococcal pyoderma. Multiple strains of streptococcal pyogenes in skin sores have been reported by Carapetis et al.29 Further research is required to examine if the high level of CRP in this population is due to chronic skin infection or other conditions. Preventing high CRP rather than reducing CRP levels particularly among children and young adults may have long term effects in preventing cardiovascular events.
There are some other possible explanations of the observed association alternative to the causal relationship between CRP and cardiovascular disease. Since the CRP measurements and cardiovascular event diagnoses were conducted at different times and places, it is unlikely that knowing baseline CRP values by the researchers could have influenced the diagnosis of cardiovascular events and vice versa. The observed hazard ratios are more likely to underestimate the true association due to the random measurement errors in both CRP and CVD. Confounding is another possible explanation of the observed association. The CRP level is clustered with a number of risk factors for CVD. After adjustment for known potential confounding factors in this study, the association remained positive and significant. If there were confounding effects, it is more likely due to some unknown factors. It is also possible that CRP is a marker of one or more true factors that drive CVD risk rather than CRP being the true cause itself. Even if this is the case, CRP is still a useful predictor of the CVD risk until we have identified those unknown true factors.
In summary, Aboriginal Australians have both a high risk of CVD and an extremely high level of CRP. High CRP levels not only independently predict the future risk of CVD, but also contribute to a large proportion of CVD cases in the population. Therefore, controlling the risk factors related to elevated CRP levels is helpful for improving cardiovascular health in Aboriginal people, and lowering CRP levels may also be beneficial.