Factors associated with high cardiovascular risk in Putumayo

TL;DR

This study found that education level, ethnicity, health affiliation, and urban/rural location are linked to high cardiovascular risk in Putumayo.

Contribution

The study identifies specific sociodemographic factors associated with cardiovascular risk in a chronic care population in Putumayo.

Findings

86.22% of patients had high cardiovascular risk, with no major difference between urban and rural areas.

Higher education and contributory health affiliation were associated with lower cardiovascular risk.

Rural area residence was linked to slightly lower cardiovascular risk compared to urban areas.

Abstract

Sociodemographic factors such as geographic location are associated with high cardiovascular risk. Urban areas are characterized by densely populated areas with access to services, while rural areas have fewer people and limited services.

To determine the factors associated with high cardiovascular risk according to the area of patients enrolled in a chronic care program in Putumayo.

Analytical cross-sectional study. The population was selected through simple random sampling. Information on the main risk factors was collected and included in a multivariate logistic regression model.

A total of 1,190 patients were included (median age 59.5 years [IQR 53–66], 68.74% women). 86.22% of the patients had high cardiovascular risk, with a similar distribution between urban (87.95%, n=628) and rural (83.61%, n=398) areas and a statistically significant difference compared to low/moderate risk (p=0.033). The factors related to cardiovascular risk in this population were primary education (OR: 0.68, CI 95%: 0.38–1.24), secondary education (OR: 0.88, CI 95%: 0.42–1.83), higher education (OR: 0.33, CI 95%: 0.13–0.82), ethnicity (none) (OR: 2.13, CI 95%: 0.98–4.63), rural area (OR: 0.66, CI 95%: 0.47–0.94), and contributory health affiliation (OR: 6.58, CI 95%: 2.75–15.72).

This study revealed that factors such as education level, ethnicity, type of health affiliation, and area were related to cardiovascular risk.

The results showed a high proportion of individuals with elevated cardiovascular risk in Putumayo, with statistically significant differences between areas.

Figures

Cardiovascular risk in patients living in rural areas vs. population living in urban areas χ2 test with a p-value = 0.033.*

Full text

Introduction

Cardiovascular risk (CVR) refers to the individual's probability of developing cardiovascular disease (CVD). It involves the occurrence of events such as myocardial infarction, cerebral hemorrhage, and embolism1, usually caused by obstructions that restrict blood flow to the heart or brain. These obstructions, in turn, result from the formation of atherosclerotic plaque or fatty deposits in the walls of blood vessels2.

When talking about CVR, it is important to recognize that arterial hypertension (AHT) and type 2 diabetes mellitus (DM2) are the most relevant risk factors for cardiovascular events. In Latin America, the prevalence of hypertension is reported to range between 30% and 50%, while DM2 ranges between 8% and 13%3. In the department of Putumayo in the year 2020, the crude incidence rates per 100,000 inhabitants were 1.79 for AHT (n=354), 0.99 (n=354) for DM2, and 2.86 (n=1019) for chronic kidney disease, taking into account that mortality rates per 100,000 inhabitants for these conditions were 93.14 (n=332), 33.10 (n=118), and 46.01 (n=164), respectively4.

Likewise, a frequently used tool recognized worldwide for measuring CVR is available: the Framingham scale, which estimates the 10-year risk in persons over 30 years. It classifies them as low risk if the sum of the scores is below 10%, moderate risk if it falls between 10% and 20%, and high risk if it is >20%5. In Colombia, this scale has been adapted by multiplying the obtained score by 0.756. This tool becomes crucial in programs for patients with chronic conditions, as it helps identify CVR in a population at higher risk of developing cardiovascular disease (CVD) and requiring hospitalization due to their baseline condition, leading to an increased need for medical care7.

On the other hand, despite Colombia’s high-coverage health system, barriers to healthcare access persist, particularly for those living in rural areas of the country8,9. The literature reports that the health situation in rural areas is more complex and is influenced by factors such as nutritional imbalances, high fertility rates, excessive workloads, poor environmental conditions10, and difficult access to education11. Consequently, these factors underscore the relevance of sociodemographic conditions in determining CVR, alongside the previously mentioned clinical factors.

Therefore, assessing potential differences in CVR between rural and urban areas is important, considering the different sociodemographic and clinical characteristics. This assessment allowed us to answer the following question: What factors are associated with high cardiovascular risk in patients from rural areas compared to those residing in urban areas who are enrolled in a chronic disease program at health institutions in Putumayo?

Materials and Methods

Study design and population: A cross-sectional study with an analytical component. The study population comprised adult patients enrolled in a chronic conditions program in one of the health service provider institutions (IPS, for its acronym in Spanish) in the department of Putumayo during the second semester of 2023.

Eligibility criteria: Patients aged 30 to 74 years were included. Pregnant women and individuals with incomplete Framingham equation variables in the database, preventing cardiovascular risk calculation, were excluded.

Sample size and sampling: The sample size was calculated for a logistic regression using the formula

$$[eqn]$$

where n represents the required sample size, k is the number of variables included in the model (6 in this case), which was multiplied by 10 cases per variable, and finally divided by the prevalence of cardiovascular disease (the dependent variable) reported as 5.88%12. Based on these parameters, the estimated sample size was 1,190 patients.

The sample was selected through simple random sampling, from a database provided by two participating IPS from urban low- and medium-complexity care centers, with an eligible population of 6,377 in the two municipalities. No missing values were obtained in the sample data.

Statistical analysis: The analysis was conducted using Stata version 18 with the total sample (1,190). The geographic area of residence (rural or urban), educational level, ethnicity, type of health affiliation, cigarette smoking, history of AHT, history of DM, presence of dyslipidemia, and use of antihypertensive treatment were described using absolute and relative frequencies. According to the Framingham scale, age and CVR scores were reported as medians and interquartile ranges (IQR) after assessing normality with the Shapiro-Wilk test.

CVR analyses were performed considering two categories according to the Framingham scale score: low/moderate and high. First, a comparison between the proportion of these CVR categories and area of residence (urban versus rural) was made using a χ2 test. Subsequently, bivariate analyses were conducted using logistic regression, considering CVR levels and each independent variable. Variables with a p≤0.25 in the bivariate analysis, along with those deemed clinically relevant, were included in a multivariate logistic regression model. For variables that were not statistically significant (educational level and ethnicity), the power calculations were performed, yielding 100% for both; consequently, for this sample, these variables were not associated.

A multivariate logistic regression was conducted, confirming that all variables were significant for inclusion in the multivariate model. Subsequently, possible confounding variables were assessed using the coefficient change method, and interactions were examined by incorporating interaction terms into the logistic regression model. Finally, regression assumptions and model diagnostics were evaluated using the Hosmer-Lemeshow test, which yielded a p-value of 0.96, indicating a good model fit13. The complete dataset collected is freely accessible for consultation on Mendeley Data14.

The study protocol was submitted for approval to the Ethics Committee of the Fundación Universitaria de Ciencias de la Salud (FUCS) and classified as “no-risk research”15 according to Resolution 8430 of 1993 issued by the Colombian Ministry of Health. Additionally, to safeguard privacy and confidentiality, subjects identified during sample construction were assigned sequentially ordered codes in compliance with the guidelines established in Law 1581 of 2012 to protect personal data16. In addition, this study was approved by the Research Ethics Committee of the Universidad Cooperativa de Colombia as per Minute No. 6, dated 06/22/2023.

Results

Results from 1,190 patients identified during the study period are presented. The median age of the study population was 59.5 years (IQR 53-66), with a median of 59 years in urban areas (IQR 52-66) and 60 years in rural areas (IQR 53-65). Regarding sex, a lower proportion of women was observed in urban areas (67.09%, n=479) compared to rural areas (71.22%, n=339). In terms of educational level, most of the population completed primary education, with a slight numerical difference between urban (66.95%, n=478) and rural areas (72.27%, n=344). Most patients had subsidized health affiliation in both urban (97.76%, n=698) and rural areas (98.11%, n=467). In addition, a comparison of variables and area (rural or urban) revealed statistically significant differences in educational level and ethnicity. Other demographic characteristics are shown in Table 1.

Table 1 Sociodemographic variables of the study populationSociodemographic variable Total n = 1,190 % (n) Rural n=476 % (n) Urban n=714 % (n) p-value** Median age [IQR] 59.5 [53-66] 60 [53-65] 59 [52-66] 0.3432 Sex

0.132 Female 68.74 (818) 71.22 (339) 67.09 (479) Male 31.26 (372) 28.78 (137) 32.91 (235) Educational level

0.001 Illiterate 12.35 (147) 13.03 (62) 11.90 (85) Primary 69.08 (822)72.27 (344) 66.95 (478) Secondary 14.87 (177) 11.97 (57) 16.81 (120) Technicians/University/Other 3.70 (44)2.73 (13) 4.34 (31) Ethnic group

0.003 Ethnic minority 3.28 (39) 14 (2,94) 25 (3.50) None 96.72 (1151) 97.06 (462) 96.50 (689) Type of health affiliation

0.680 Subsidized 97.90 (1165) 98.11 (467) 97.76 (698) Contributory 2.10 (25) 1.89 (9) 2.24 (16) *IQR: Interquartile range; ** χ2 test for categorical variables and Mann-Whitney U test.

Regarding the study population's risk factors and clinical conditions, the proportion of tobacco dependence is low in urban (1.96%, n=14) and rural areas (1.47%, n=7). The proportion of dyslipidemia was similar in both areas (urban 35.15%, n=251 and rural 36.76%, n=175). Type II diabetes was proportionally more frequent in urban areas (64.99%, n=464) than in rural areas (67.65%, n=322). The AHT proportion was high in both urban (91.32%, n=652) and rural (89.50%, n=426) areas. In addition, a comparison of clinical variables by area (rural or urban) revealed statistically significant differences in diastolic blood pressure, antihypertensive treatment, glycosylated hemoglobin, total cholesterol, and LDL cholesterol. Other clinical characteristics are shown in Table 2.

Table 2. Clinical variables Clinical variablesTotal n = 1,190 % (n) Rural n = 476 % (n) Urban n = 714 % (n) p-value Smoker 1.76 (21) 1.47 (7) 1.96 (14) 0.529 Dyslipidemia 35.80 (426) 36.76 (175) 35.15 (251) 0.570 Diabetes Mellitus

0.307 Type I 24.12 (287) 21.85 (104)25.63 (183) Type II 66.05 (786) 67.65 (322) 64.99 (464) No diabetes 9.83 (117) 10.50 (50) 9.38 (67) Arterial hypertension 90.59 (1078) 89.50 (426) 91.32 (652) 0.393 Systolic blood pressure

Median [IQR] 130 [120-144] 130 [120-140] 130 [ 120-146] 0.184 Diastolic blood pressure

Median [IQR] 80 [80-90] 80 [80-90] 80 [80-90] 0.026Antihypertensive Treatment

0.002Non-pharmacological management 3.28 (39) 5.04 (24) 2.10 (15) HCTZ1.60 (19) 1.68 (8) 1.54 (11) ACE inhibitor or ARA 21.60 (257) 22.27 (106) 21.15 (151) HCTZ + ACE inhibitor or ARA 22.77 (271) 21.01 (100) 23.95 (171) HCTZ + ACE inhibitor or ARA + Amlodipine 23.36 (278) 18.07 (86) 26.89 (192) Amlodipine 1.01 (12) 1.47 (7) 0.70 (5) HCTZ + ACE inhibitor or ARA + Amlodipine + Other 12.94 (154) 15.97 (76) 10.92 (78) Other 0.92 (11) 0.84 (4) 0.98 (7) No information 12.52 (149) 13.66 (65) 11.76 (84) Glycated Hemoglobin

0.003 Normal (<5.7)40.08 (477) 45.17 (215) 36.69 (262) Prediabetes (5.7 – 6.4) 4.79 (57) 3.99 (19)5.32 (38) Diabetes (>6.4)55.13 (656) 50.84 (242) 57.98 (414) Blood glucose

0.497 Normal (<99) 46.13 (530) 47.59 (217) 45.17 (313) Prediabetes (100-125) 30.55 (351) 30.04 (137) 30.88 (214) Diabetes (>126)23.32 (268) 22.37 (102) 23.95 (166) Total cholesterol

0.000 Normal (<200 mg/dL) 59.66 (710) 55.04 (262) 62.75 (448) Normal – High (200 - 240 mg/dL) 25.55 (304) 26.89 (128) 24.65 (176) High (>240 mg/dL) 14.79 (176) 18.07 (86) 12.61 (90) HDL

0.177 Normal79.50 (946) 78.36 (373) 80.25 (573) Low 20.50 (244) 21.64 (103) 19.75 (141) LDL

0.000 Optimal (<100) 46.10 (544)40.80 (193) 49.65 (351) Near optimal (100-130) 27.88 (329) 27.48 (130) 28.15 (199) Upper limit of the normal range (>130 – 160) 16.69 (197)19.24 (91) 14.99 (106) High (>160 – 189) 6.53 (77) 8.03 (38) 5.52 (39) Very high (>189)2.80 (33) 4.44 (21) 1.70 (12) Triglycerides

0.146Normal (<150) 39.83 (474) 42.02 (200) 38.38 (274) Borderline high (150 – 200) 19.24 (229) 18.91 (90) 19.47 (139) High (>200) 40.92 (487) 39.08 (186) 42.16 (301) *IQR: Interquartile range; HCTZ: Hydrochlorothiazide; ACE: Angiotensin-Converting Enzyme; ARA: Angiotensin Receptor Antagonists. HDL: High-Density Lipoprotein LDL: Low-Density Lipoprotein

The most frequent CVR level for the study population was high, with a numerically higher proportion in the urban areas (87.95%) compared to the rural areas (83.61%). When comparing CVR level with geographic areas of residence, a statistically significant difference was found (p=0.033) (See Figures 1).

Figure 1. Cardiovascular risk in patients living in rural areas vs. population living in urban areas** χ2 test with a p-value = 0.033.*

Finally, within the results observed in the multiple logistic regression, the final model included the following variables: primary education (OR: 0.68, CI 95%: 0.38–1.24), secondary education (OR: 0.88, CI 95%: 0.42–1.83), higher education (OR: 0.33, CI 95%: 0.13–0.82), rural area (OR: 0.66, CI 95%: 0.47–0.94), subsidized health affiliation (OR: 6.58, CI 95%: 2.75–15.72), and ethnicity (none) (OR: 2.13, CI 95%: 0.98–4.63). The latter did not reach statistical significance (see Table 3). Confounding and interaction variables were assessed, and none were identified.

Table 3. Sociodemographic factors associated with cardiovascular risk (Framingham*)Variable Unadjusted OR (95%CI) p-value Adjusted OR** of the multivariate model (95% CI) p-valueEducation level

Illiterate 1 1 Primary0.64 (0.36 –1.16) 0.1450.68 (0.38 -1.24) 0.216 Secondary0.82 (0.40 –1.69) 0.6040.88 (0.42 -1.83) 0.736 Technicians/University/Other0.20 (0.08 –0.46) 0.000 0.33 (0.13-0.82) 0.017 Ethnic group

Ethnic minority1 1 None1.92 (0.89 – 4.13) 0.092 2.13 (0.98- 4.63) 0.054 Area

Urban1 1 Rural 0.69 (0.50 – 0.97) 0.034 0.66 (0.47-0.94) 0.021Type of health affiliation

Contributory1 1 Subsidized 8.61 (3.83-19.32)0.000 6.58 (2.75- 15.72) 0.000* Output/Framingham Variable: Age, total cholesterol, HDL, systolic blood pressure, smoking, and diabetes mellitus; **Multivariate odds ratios adjusted for other variables.

Discussion

This study's results suggest that there is a statistically significant difference in cardiovascular risk levels among patients enrolled in a chronic condition care program in the department of Putumayo, depending on their geographic location. This finding aligns with a 2013 study17 conducted in an urban population, which reported high CVR in 98.7% of the study population. However, it contrasts with the rural population findings of that study, where only 2.3% exhibited high CVR, a numerical difference compared to the 83.61% identified in this study. The high proportion of cardiovascular risk observed in Putumayo may be attributed to the fact that individuals who visit these IPS health centers are patients with symptoms or a history of cardiovascular risk-related conditions, as opposed to healthy individuals who also access healthcare services.

We found that educational level, area, and type of health affiliation were associated with higher cardiovascular risk. In the final logistic regression model, educational level was statistically significant in the technical/university/other category, indicating that the possibility of having a higher level of education among patients with high cardiovascular risk is one-third that of patients who are not at high cardiovascular risk. This finding is consistent with a study published in April 2019, which related low educational levels to higher cardiovascular mortality and suggested that individuals with higher education are more knowledgeable about the disease, its risk factors, and preventive measures5,18.

Subsidized health affiliation was identified as a risk factor for developing severe cardiovascular problems compared to contributory health affiliation. As observed in previous studies, no statistically or clinically significant association was found between cardiovascular risk and type of health affiliation19. However, according to a recent study, individuals with a subsidized health affiliation exhibited better glycemic index control than those with a contributory health affiliation20. Given that diabetes and hypertension contribute to or increase CVR, glycemic control can be considered an indirect indicator of cardiovascular risk. According to the healthcare system's functioning, high cardiovascular risk levels are expected to be distributed independently of the type of health affiliation. However, in this study sample, the high proportion of individuals with subsidized health affiliation may indicate a possible association. This finding could also be explained by selection bias.

Ethnicity in the final logistic regression model was not statistically significant, with an OR of 2.13 and a 95% CI of 0.98–4.63. However, the literature shows that ethnicity is a predisposing factor for cardiovascular disease. A document published by the Spanish Heart Society cites multiple studies identifying a greater predisposition to AHT among individuals of black ethnicity, with evidence suggesting a worse prognosis in this population21,22. However, in Colombia, there is a mixture of cultures and races; therefore, these characteristics are measured by official entities using patient self-reporting, which poses challenges at the moment of objectively classifying them for data analysis. Additionally, for this analysis, the ethnicity variable was analyzed by grouping individuals at risk —Romanies, Gypsies, Black, and Afro-Colombian populations—compared to White individuals.

This study made it possible to know cardiovascular risk levels among patients attending chronic care units and living in rural and urban areas, revealing statistically significant differences between these areas in the department of Putumayo. These findings, in turn, provide the basis for formulating hypotheses in future research, as the study offers an overview of potential relationships between the variables of interest.

It is important to mention that the main limitation of this study is the apparent selection bias, as the sample consisted of patients enrolled in chronic care programs. This is why these individuals are more likely to have risk factors, which could lead to overestimating cardiovascular risk. It is relevant to recognize that not all patients with risk factors fall into the high-risk category. Therefore, conducting risk classifications during patient follow-up is important for improving the accuracy of cardiovascular risk assessment, particularly for subclinical cardiovascular disease, silent ischemia, and future cardiovascular events. Additionally, risk classifications could prevent the unnecessary use of treatments in low-risk patients, thereby increasing adverse events23. It should also be mentioned that these programs' recruitment rates of chronic patients do not reach the entire population with CVR. However, an attempt was made to mitigate this bias by selecting participants through simple random sampling.

he possibility of measurement bias cannot be ruled out, as the main source of information was medical records, which in these municipalities are transcribed manually. Additionally, misclassification bias may have occurred, as errors in the variables used in the equation for calculating cardiovascular risk could have led to a change in the categorization of patients (low, moderate, or high risk). Consequently, such misclassification of information could also impact the interpretation of the results.

The importance of cardiovascular risk assessment and the identification of risk factors has long been a subject of research. Although many risk-enhancing factors have been identified, it is necessary to determine whether this risk is the same in different populations. Current health patterns, urbanization, dietary changes, access to healthcare services, and medical care may also influence the development of cardiovascular disease. This will enable the identification of potential socioeconomic factors to be considered in populations with specific characteristics, allowing the implementation of long-term, sustainable strategies to improve health outcomes. In turn, this could help mitigate the rising prevalence of cardiovascular diseases and reduce health costs18.

**Implications for practice **

The implications for clinical practice of this study lie in its potential to enhance the accuracy of cardiovascular risk classification and, consequently, optimize medical care. The findings of this study reveal significant differences in cardiovascular risk between rural and urban areas of Putumayo, highlighting the importance of adapting prevention and treatment strategies to the specific characteristics of each population. In clinical practice, these insights would allow health professionals to implement more personalized interventions, tailored to patients' socioeconomic and geographic conditions. Ultimately, this study establishes a starting point for future research and changes in clinical practice that will allow more efficient management of cardiovascular diseases, contributing to better public health and reducing the associated healthcare costs.

Conclusions

This study determined that there is a high proportion of individuals with elevated cardiovascular risk in Putumayo, with statistically significant differences between rural and urban areas. We declare that, due to selection bias, the population with which the statistical analyses were performed is not representative of the entire population of Putumayo. Therefore, the data can only be extrapolated to populations with chronic conditions. Cardiovascular risk was associated with educational level, ethnicity, type of health affiliation, and area.