Cross-sectional assessment of cardiovascular risk factors in patients with knee osteoarthritis

Introduction

Osteoarthritis (OA), a degenerative joint disease, is the sixth most predominant cause of disability around the world and a well-established cause of restricted activity, disability, and low quality of life.¹ Its prevalence, currently at 3.5% of the world’s total population and about 35% in adults older than 60 years,² is on the rise. OA has become an important public health issue and is burdensome for both personal health and social well-being.² Both developed and developing nations see it as a major liability on their health care sector with it having severe social and economic impacts.¹

OA is associated with joint pain, functional limitation and deformity and it incurs articular cartilage degeneration, osteophyte formation in joints, reduced joint space and subchondral sclerosis.³ The etiology of OA is idiopathic, yet age, genetics, obesity, menopause, hypertension, and diabetes mellitus have been found to be major contributors.⁴ Some of these risk factors that are directly involved and others indirectly involved in the etiopathogenesis of OA are also seen to increase the risk of cardiovascular diseases (CVDs), namely congestive heart failure (CHF), ischemic heart disease (IHD), transient ischemic attacks (TIA) and stroke.⁴ This group of cardiovascular diseases are also the leading cause of mortality and morbidity worldwide.⁵ Therefore, it becomes important to identify the various cardiovascular risk factors like age, gender, obesity, hypertension, cholesterol, sedentary lifestyle, smoking and nutrition at the earliest opportunity through various screening methods, especially in patients with OA.³

A possible association between OA and cardiovascular risk factors is evident by a study done at Leigh where above average levels of serum cholesterol were observed, especially in women with osteoarthritis of hands.³ Another study found a high prevalence of cardiovascular risk factors in adults with knee OA.⁵ Recently, a new classification for phenotyping of OA, which includes metabolic syndrome, aging, and posttraumatic arthritis, has been developed. Obesity is a well-known contributor for metabolic syndrome.⁶ Observational studies have shown that hypertension is an independent risk factor for knee OA.^6,7 Indirectly, physical inactivity, due to debilitating joint pain, also makes patients more prone to CVD.⁷ Recently, in a systemic review and meta-analysis, it was questioned whether aerobic exercise is being used adequately in patients with knee OA as under-prescription of it could be a potential source of increased risk of CVD.⁸ The concept of systemic inflammation as potential pathogenesis, underlying both OA and CVD has been under scanner for some time now; as opposed to the traditional view of wear and tear.²³ There are studies showing positive correlation of knee OA with metabolic syndrome, hypercholesteremia and DM.^6,12,25

Given that OA and CVD are common conditions in the elderly, a thorough knowledge of the relationship between OA and CVD could help further our understanding of potential biological and behavioral mechanisms, which could in turn help in making informed decisions regarding further management of OA. There is a dearth of literature relating knee OA and cardiovascular risk factors in the Indian population, hence we wanted to study this association among Indian participants. The objectives of the study were (i) to screen patients diagnosed with knee OA for cardiovascular risk factors who don’t have a history of cardiovascular diseases; (ii) to find out the risk of developing a cardiovascular disease in next 10 years in patients suffering from knee OA using a simple standardized risk score calculator; and (iii) to correlate CVD risk with severity of knee OA.

Methodology

Results

A total of 225 patients visiting the orthopaedics OPD with complaints of knee pain took part in the study.³² All participants were over 50 years of age with more than half (53%) of them between the age of 50-60 years (Figure 1). The majority of these patients were females (59%) (Figure 1). As per the BG Prasad socioeconomic scale (SES),¹⁶ the majority of them belonged to either the middle or lower socio-economic group (Figure 1).

Figure 1. Age, Gender, BG Prasad socioeconomic scale, KL grade, BMI and history of smoking distribution of the study population.

Grade 2 K-L classification of knee OA was found to be the most common (45%) among the study population. Higher grades of knee OA, i.e. 3 and 4, were found to be 42% and 13%, respectively (Figure 1). Overall, 57% of the patients belonging to the lower socio-economic class as per the BG Prasad SES had knee OA grade 2, 33% of them had grade 3 and only 10% had grade 4 OA. Among the lower middle and middle-class groups, a decrease in the percentage of patients with grade 2 and increase in percentage of patients with grade 3 and 4 was found as socioeconomic status increased. The same trend was observed, not strictly though, as we go further up the SES. Upper middle and upper-class patients had either grade 3 or 4 knee OA (Figure 2).

Figure 2. Distribution of knee osteoarthritis (OA) severity by socioeconomic group.

Higher socioeconomic groups had a higher proportion of patients with severe grade knee OA.

The Fishers exact test was used to find out the statistical significance of the relationship between socioeconomic status and knee OA grade, which was found to be highly significant (p value<.01). Chi square test, when applied to compare knee OA grade with age {X²(4, 225) = 3.801, p = .434} or gender {X²(2, 225) = 1.016, p = .602}, was not found to be statistically significant. A one-way ANOVA test was applied to K-L grade of knee OA with BMI and found to be highly statistically significant {F(4,225) = 60.652, p < .01}. An increasing trend in the BMI of patients was noted with a higher grade of knee OA, e.g. 60% of grade 2 knee OA patients had a BMI of less than 25 as compared to 40% in grade 3 and only 7% in grade 4. In addition, 57% of patients with grade 4 knee OA had a BMI of more than 30 as compared to 13% in grade 3 and only 4% in grade 2 (Table 1).

The mean BMI of patients with grade 2 knee OA was 24.82, grade 3 was 26.39 and grade 4 was 30.96, with each knee OA grade showing statistically significant data when compared with the BMI of patients (Table 2). In post hoc analysis using the Bonferroni test with BMI as the dependent variable, it was found that multiple comparisons of each of grade of knee OA with the other were found to be statistically significant, as depicted in Table 3.

Of the 225 participants, 18.2% (n = 41) were smokers. All the smokers were male, and 39% of them had grade 2 knee OA, 44% of them had grade 3 and 17% of them had grade 4. However, this data was not statistically significant on chi square test {X²(2, 225) = 0.964, p = .617} (Table 1).

The mean systolic blood pressure (SBP) of participants was 128mmhg, 139mmhg and 150mmhg for grade 2, grade 3 and grade 4 knee OA, respectively. Table 2 shows the representation of mean SBP for each K-L grade. On comparing each grade with each other using post hoc analysis with SBP as the Bonferroni dependent variable, the comparisons came out to be highly significant (Table 3).

Of the 225 participants, 88 had a SBP reading of 140 or more, out of which 28 (33%) were not taking any antihypertensive medication. In total, 25.8% of the entire sample had a history of diabetes mellitus (DM), which is much lower than more than half (53%) of patients with grade 4 knee OA. Of grade 3 patients, 40% had DM, yet of grade 2, only 4% had DM. This was found to be statistically significant {X²(2, 225) = 47.574, p < .01} as per the chi square test (Table 1).

Similar trends were seen in the history of CVD among the relatives of the study population. A total of 35 (15.5%) patients gave a history of their relatives suffering from CVD in the past. Compared by grade, 46.7% of the patients with grade 4 knee OA gave a positive relative CVD history as compared to only 17% and 5% of the patients with grade 3 and 2 knee OA, respectively (Table 1). The chi square test for this difference was statistically significant {X²(2, 225) = 30.907, p < .01}.

None of the patients in the study population suffered atrial fibrillation (AF) or gave a history of chronic kidney disease in the past. Only 11 patients gave a history of rheumatoid arthritis which was found to be statistically significant by Fisher exact test (p < .01) (Table 1).

Serum total cholesterol (TC) and high-density lipoprotein (HDL) were tested for all the patients. Table 2 shows the mean levels of TC and HDL with respect to each of the KL grades of knee OA. An increase in mean value of TC and decrease in mean value of HDL were noted as K-L grades of knee OA increase. Patients with grade 2 knee OA had a mean TC value of 188.34 and mean HDL value of 53.74. Similarly, patients with grade 3 knee OA had a mean TC and HDL value of 213.66 and 51.99, respectively. Patients with grade 4 knee OA had a mean TC and HDL value of 234.90 and 49.30, respectively (Table 2).

In post hoc analysis using the Bonferroni test with TC as the dependent variable, it was found that multiple comparisons of each of grade of knee OA with the other were found to be highly statistically significant (Table 3). The same was not true, however, with HDL, which showed statistically significant data only when K-L grade 2 was compared to grade 4 knee OA (Table 3).

This JBS3 risk score calculator gave us three parameters about the cardiovascular condition of the heart, i.e. physiological heart age, 10-year risk in percentage of developing a CVD and life expectancy provided any other cause of death is ruled out. Statistically significant results were obtained when studying the severity of knee OA with heart age and life expectancy. Physiological heart age came out lesser but life expectancy higher for patients with a lower grade of knee OA. As we go up the K-L grade of knee OA, the physiological heart age went up and the life expectancy came down (Figure 3).

Figure 3. As we go up the K-L grade of knee OA, the physiological heart age went up and the life expectancy came down.

The mean physiological heart age was found to be 70 years; 10 years older than the mean chronological age (60) of the study population.

The JBS3 risk score also increased with the increase in the K-L grade of knee OA. Patients with grade 2 knee OA had a mean risk of around 11% of developing a CVD in the next 10 years and this increased with K-L grades, i.e. there was a 23% risk among the patients with grade 3 knee OA and 38% risk among those with grade 4 knee OA (Table 2). Chi square test found this to be statistically significant {X²(4,225) = 70.776, p < .01}.

The Bonferroni post hoc analysis of K-L grade with the JBS3 risk score as the dependent variable showed multiple comparisons of each K-L grade with the others as statistically significant (Table 3). The ROC curve showing area under the curve of sensitivity vs specificity of grade 4 vs 2 and 3 for risk score is depicted in Figure 4. According to ROC analysis for the JBS3 risk score, it was determined that patients with K-L grade 4 knee OA were more likely to run a 10-year risk of developing CVD at minimum 14% with sensitivity of 72.3% and specificity of 66%. On analyzing the data using multiple regression with dependent variable as K-L Grade of knee OA (Table 4), we found that variables like history of hypertension treatment, diabetes mellitus, serum levels of total cholesterol and body mass index were found to be significant, rest all variables were found to be not significant.

Figure 4. ROC analysis for the JBS 3 risk score showed that a patient with K-L grade 4 of knee OA is more likely to be running a 10 years risk of developing CVD at minimum 14% with sensitivity of 72.3% and specificity of 66%.

Discussion

Osteoarthritis is a common disease in the elderly population, with its higher prevalence in the 50+ year age group well documented in the literature.^1,3,12 The current study’s entire sample of 225 patients were above the age of 50 years. Higher prevalence of knee OA in the 50-60-year age group was seen but with no statistical significance when compared with groups of higher ages. This could be attributed to the fact that this study did not include the entire population of the given area but rather only the patients coming to orthopaedic OPD with pain in the knee joint. Women are more likely to have knee pain above the age of 50 years¹⁰ and with no surprise our study also had women forming almost two thirds (60%) of the sample size. The estimated global prevalence of OA is around 10% in men and 20% in women.^3,5,12,13,17 One study on epidemiology of knee OA in India showed a prevalence of 28.7%.¹²

American and European studies have found OA to be a disease of the upper class.¹⁷ The COPCORD study from Bangladesh also showed the same.¹⁸ On the contrary, a study conducted in south Delhi found the prevalence of OA was higher in perimenopausal women of lower classes than upper classes.¹⁹ Our study participants of lower socioeconomic status mostly had grade 2 knee OA but upper classes had the more severe grades of 3 or 4 as per K-L grading. BMI was found to be higher in groups with severe grades of knee OA as compared to milder grades, and there was a statistically significant association found between BMI and K-L grade of knee OA. Upper class people tend to have more modifiable risk factors,¹⁹ which is quite evident in the current study, showing increased prevalence of these modifiable risk factors (BMI, TC, HDL, SBP, DM) in these groups.

Atherosclerosis, hypertension, diabetes and hypercholesterolemia cause systemic inflammation in the body, which can lead to CVD.²⁴ Association between knee OA and CVD has been studied and linked to underlying systemic inflammation, however, it is unsure if this relationship is direct or indirect. For example; A meta-analysis of 299 publications found a high prevalence of OA in patients with DM.⁶ They also emphasized that metabolic OA phenotypes needs to be studied further. Another meta-analysis to explore the correlation between metabolic syndrome and knee OA, found a significant odds ratio (OR) even after adjusting for many risk factors.²⁵ The aforementioned Korean study displayed a significantly higher knee OA prevalence in patients with hypertension and impaired glucose tolerance.¹² Veronese et al. while studying elderly participants without CVD found that OA at baseline was associated with subsequent incident CVD.^24,26 In the current study, the mean SBP was significantly higher in patients with K-L grade 3 and 4 of knee OA. Similar trends were seen while studying the prevalence of DM in these patients. In in the total sample, 25.8% patients had DM, yet the largest percentage (53%) was seen in patients with K-L grade 4 knee OA, showing the risk of DM increases with severity of knee OA, as found by Louati K et al.⁶ Hypercholesterolemia was also studied here, being another systemic inflammatory marker. It showed a statistically significant relationship with K-L grades of knee OA; mean serum TC was found to be higher in patients with severe grades of knee OA, whilst mean serum HDL was found to be lower in patients with severe grades of OA.

Systolic BP, DM and hypercholesterolemia are three of the major contributors to CVD.¹² As per the results of this study it would be safe to say that patients with severe knee OA have a higher risk of developing a CVD in the future. On the contrary, Hoeven et al.²⁷ believe that OA-related disability and not OA predicts CVD. This is along the lines of belief of many other researchers who postulate that OA doesn’t directly lead to a CVD, rather it renders a patient physically more inactive and disabled than those without OA, which ultimately increases the CV risk factors. Recently, in a systemic review and meta-analysis, it was questioned whether aerobic exercise is being used adequately in patients with knee OA as under-prescription of it could be a potential source of increased risk of CVD.⁸ Whether OA causes CVD directly or indirectly, it is hard to ignore the fact that they often exist co-dependently in a patient’s body. Assessment of a patient with knee OA for CV risk factors would be of no harm and rather provide the doctor a systemic approach to the patient’s morbid condition and also make the patient aware about their unknown CV risk factors and risk of developing a CVD. Korean National Health Survey and Nutrition Examination Survey (KNHANES) found a higher prevalence of hypertension, DM, dyslipidaemia, angina and myocardial infarction in patients with OA compared with healthy individuals.²⁸

Contrary to traditional belief, smoking was not found to have any statistically significant effect on the K-L grade of knee OA.²⁹ Other variables such as atrial fibrillation, chronic kidney disease and rheumatoid arthritis were not found to be prevalent in our study population, preventing us from analysing any relationship between them and knee OA.

This study has attempted to collaborate the basic mechanical, causal and shared risk factors (age, obesity, gender) between knee OA and CVD. It has also tried to study various other CV risk factors (hypertension, DM, dyslipidaemia, smoking, family history) in patients with OA. Ho Sun Kim et al.¹² used the Framingham risk score (FRS) in south Koreans to study the association between OA and CV risk factors; however, it was found that the JBS3 risk score calculator is a better tool than FRS when applied to the Indian population.³⁰ We used the JBS3 risk score to calculate 10-year risk of developing CVD, physiological heart age and life expectancy. Statistically significant data revealed the mean physiological heart age of the patients to be as much as 10 years higher than their actual mean age and their life expectancy shortened with more CV risk factors. The mean JBS3 risk was 11% for patients with grade 2 OA, 23% for patients with grade 3 OA and 38% with grade 4 OA. This was along the same lines as studied in Korea.¹²

Our study had 225 patients with a prevalence of grade 4 knee OA of 13%. With the help of the ROC curve, the cut-off value of JBS3 was calculated to be 14%, which implies the amount of risk patients with grade 4 knee OA had of developing a CVD in the next 10 years, with a sensitivity of 72.3% and specificity of 66%. It would not be an overstatement to say that a patient with grade 4 knee OA undergoing total knee replacement could have a CV risk of as much as 14%.

India, although one of the countries where both OA and CVD is most prevalent, has very little literature stating the relationship between the two. The strength of this study lies in the different variables (demographic and CV risk factors) that were studied. With the genetic and phenotypic variations in the population in this part of the world, BMI plays a major role in understanding various pathologies in human body. The BG Prasad scale has helped us to divide the study population into various socio-economic strata and study its relation to both OA and CV risk factors. In addition, the diagnosis of knee OA has been made with the help of standard radiographs using K-L classification. The potential confounding factors (CV risk factors) are clubbed together with the help of a standardized risk score calculator devised by the Joint British Society. JBS3 also tells us about the physiological heart age and life expectancy. Although these factors have an individual effect on the patho-mechanisms of knee OA, when they are used as a part of this scale, various limitations are eliminated, and they are studied as a whole as well. The risk factors thus are studied both individually as well as co-dependently. However, this study has its limitations too. It is a cross-sectional study, which comes with its own inherent limitations. The study population only has patients with K-L grade of ≥2, so the risk assessment of patients with K-L grade 1 was not done here. The JBS3 risk score calculator used here could only come close to being ideal for the study population of our country. A risk score calculator developed in our country for Indian genotypes and phenotypes would be ideal. Also, it doesn’t include stress³¹ and activity level as variables, which has a known effect on cardiovascular status. The SBP values entered here are of single reading, which can be affected by the hemodynamic status of the patient at one particular moment. The readings of TC and HDL values may have differed depending on the investigator and equipment available from place to place. The JBS3 risk score calculator is not a tool to initiate treatment for a patient, rather a tool that gives a fair idea of the patient’s general condition and CV status. Some variables of this tool depend upon the history given by the patient, which has its own limitations.

Conclusion

Our study concluded that there is a strong relation between knee OA and CVD with CV risk score being positively correlated to the severity of OA.

Data availability