Noise Exposure and Hypertension: A Silent Relationship
Noise Exposure and Hypertension: A Silent Relationship
To address the main question of the association between noise exposure and high blood pressure, we performed a cross-sectional study design of existent secondary health and noise exposure data from 2007 related to workers of a petrochemical and gas refinery in Rio de Janeiro, Brazil.
The studied unit has two types of work force: contractors (workers paid by the government) and sub-contractors (workers paid by outsourced corporations). The two workforces are subject to different labor legislations and all outsourced corporations are required to perform pre-admission and annual health examination in every worker. Subcontracted labor force was working on the plant maintenance and expansion, while government workers were more involved in administrative tasks in the plant. Our sample framework was based only on workers from outsourced corporations (sub-contractor workers) since data was available only to them. We used the 6-month timeframe in order to guarantee a minimum period of noise exposure before the diagnosis of hypertension. The total number of subcontractor workers was 3,023 individuals, but our study was restricted to the 1,729 individuals working at the petrochemical industry for at least 6 months and submitted to annual physical exam in 2007.
To elaborate our dataset we collected information from the electronic clinical forms of workers submitted to the annual mandatory health evaluation.
The variable age was stratified in five-year intervals. Civil status was dichotomized in married and single/divorced individuals, the only available categories. Formal education was reclassified in completed or incomplete elementary school, middle school, high school and college. Physical activity, tobacco use and alcohol consumption were defined according to the clinical form. Thus, regular physical activity included any activity for at least three times a week and a minimum duration of 20 minutes. Regular alcohol consumption was attributed to individuals informing daily alcohol ingestion, occasional consumption to individuals informing alcohol use only on weekends, and individuals informing no alcohol consumption were classified as no consumption. No data were available about the amount, type of alcoholic beverage and history of alcohol consumption. Alcohol consumption data was further categorized in consumption and no consumption, which encompassed those that consumed alcoholic beverages both occasionally and regularly. Body Mass Index (BMI) was calculated from the values of weight and height and categorized as normal (underweight - <18.50 Kg/m and normal – 18.50–24.99 Kg/m), overweight (25.00–29.99 Kg/m) and obese (≥30.00 Kg/m) according to the WHO criteria. Heating exposure was classified as exposed, when an individual was found to be exposed to values equal to or higher than 30 WBGTI (Wet Bulb Globe Thermometer Index) by the time of the measurement, according to a methodology recommended by the Fundacentro.
Traditional socioeconomic variables were not available in the forms from the annual mandatory health evaluation. Therefore, we used all available information that could be linked to the social condition of the workers. Hence, presence of running water, regular trash collection, sewage network and electrical power were used to create a composite indicator in which each of the four variables was classified as absent (0) or present (1), assuming equal importance to all variables. The socioeconomic indicator was classified as higher if all four variables were present and lower if at least one variable was absent.
There were several sectors distributed along the different subcontracted firms and with different levels of exposure to noise, and we were unable to obtain a correct estimate of the location of the workers per sector. Therefore, we opted for grouping the workers by the type of work in the industry as in "industrial maintenance", "cleaning and building maintenance", "civil works" and "others" (meals, transportation and security).
The outcome (high blood pressure) was defined as systolic pressure equal to or higher than 140 mmHg and/or a diastolic pressure equal to or higher than 90 mmHg. Any worker reporting the use of anti-hypertensive medication was also classified as hypertensive, regardless of the measured pressure level. Blood pressure was measured by the attendant physician once at the beginning of the medical consultation, always on the left arm of a patient seated. In cases in which the blood pressure measurement was above the cut-off limits, a second measurement was performed at the end of the consultation. In these situations, only the second measurement was reported in the medical records.
Exposure to noise was measured at Homogeneous Exposure Groups (HEG) by using a digital audio dosimeter affixed next to their aural point to a randomly selected worker within the HEG. The measurement was done at a single moment in cases of continuous exposure to noise and during at least 75% of a workday of 8 hours in cases of intermittent exposure to noise, according to the Fundacentro methodology. We categorized noise exposure in three levels: ≤75 dB(A), from 75 to 85 dB(A), and ≥85 dB(A). We based our decision to use the three level categorization on some studies that also considered lower levels for noise exposure.
Since the available dataset was not created to be used in a research project, not all-important variables were readily available to be used in the statistical analyzes. We decided to analyze confounding and interaction terms on any available variable with some indication in the biomedical literature of association to either noise exposure or high blood pressure. Therefore, age, gender, marital status, education, socioeconomic condition, physical activity, tobacco use, alcohol consumption, body mass index (BMI), type of service and heat exposure were the co-variables used in our analysis.
We began our data analysis by describing the frequency distribution of all variables according to the level of noise exposure. Thereafter, we performed a bivariate analysis for the association between all co-variables and both the exposure and the outcome variables in order to identify possible confounding variables for the studied association. All variables were categorized and a Chi-squared test used to test the associations. Any variable associated to the outcome with a p-value of at least <0.15 was considered to be included in the final multivariate model.
First order interactions terms between noise and age, gender, socioeconomic condition, physical activity, tobacco use, consumption of alcohol, body mass index and exposure to heat were individually tested and considered for the inclusion in the final multivariate model. Therefore, the baseline multivariate model was composed by all variables with a p-value of less than 0.15 in the exploratory analysis and statistically significant interaction terms. A logistic regression model with backward Wald elimination was used to model the independent association between noise exposure and high blood pressure, while considering other possible confounding variables. We present the final multivariate model only with variables statistically associated with high blood pressure since addition of non-significant variables did not add to the fitness of the final logistic model. The final odds ratio represents the relative odds of presence of high blood pressure. Statistical analysis was performed with the statistical package SPSS (version 17; SPSS Inc. software products).
The study was approved by the Ethics Committee at the National School of Public Health (CEP/ENSP). We analyzed a secondary dataset with no identifiers and the study was granted with a consent form waiver by the CEP/ENSP.
Methods
Study Design and Population
To address the main question of the association between noise exposure and high blood pressure, we performed a cross-sectional study design of existent secondary health and noise exposure data from 2007 related to workers of a petrochemical and gas refinery in Rio de Janeiro, Brazil.
The studied unit has two types of work force: contractors (workers paid by the government) and sub-contractors (workers paid by outsourced corporations). The two workforces are subject to different labor legislations and all outsourced corporations are required to perform pre-admission and annual health examination in every worker. Subcontracted labor force was working on the plant maintenance and expansion, while government workers were more involved in administrative tasks in the plant. Our sample framework was based only on workers from outsourced corporations (sub-contractor workers) since data was available only to them. We used the 6-month timeframe in order to guarantee a minimum period of noise exposure before the diagnosis of hypertension. The total number of subcontractor workers was 3,023 individuals, but our study was restricted to the 1,729 individuals working at the petrochemical industry for at least 6 months and submitted to annual physical exam in 2007.
Health, Socioeconomic and Environmental Data
To elaborate our dataset we collected information from the electronic clinical forms of workers submitted to the annual mandatory health evaluation.
The variable age was stratified in five-year intervals. Civil status was dichotomized in married and single/divorced individuals, the only available categories. Formal education was reclassified in completed or incomplete elementary school, middle school, high school and college. Physical activity, tobacco use and alcohol consumption were defined according to the clinical form. Thus, regular physical activity included any activity for at least three times a week and a minimum duration of 20 minutes. Regular alcohol consumption was attributed to individuals informing daily alcohol ingestion, occasional consumption to individuals informing alcohol use only on weekends, and individuals informing no alcohol consumption were classified as no consumption. No data were available about the amount, type of alcoholic beverage and history of alcohol consumption. Alcohol consumption data was further categorized in consumption and no consumption, which encompassed those that consumed alcoholic beverages both occasionally and regularly. Body Mass Index (BMI) was calculated from the values of weight and height and categorized as normal (underweight - <18.50 Kg/m and normal – 18.50–24.99 Kg/m), overweight (25.00–29.99 Kg/m) and obese (≥30.00 Kg/m) according to the WHO criteria. Heating exposure was classified as exposed, when an individual was found to be exposed to values equal to or higher than 30 WBGTI (Wet Bulb Globe Thermometer Index) by the time of the measurement, according to a methodology recommended by the Fundacentro.
Traditional socioeconomic variables were not available in the forms from the annual mandatory health evaluation. Therefore, we used all available information that could be linked to the social condition of the workers. Hence, presence of running water, regular trash collection, sewage network and electrical power were used to create a composite indicator in which each of the four variables was classified as absent (0) or present (1), assuming equal importance to all variables. The socioeconomic indicator was classified as higher if all four variables were present and lower if at least one variable was absent.
There were several sectors distributed along the different subcontracted firms and with different levels of exposure to noise, and we were unable to obtain a correct estimate of the location of the workers per sector. Therefore, we opted for grouping the workers by the type of work in the industry as in "industrial maintenance", "cleaning and building maintenance", "civil works" and "others" (meals, transportation and security).
Definition of Exposure and Outcome
The outcome (high blood pressure) was defined as systolic pressure equal to or higher than 140 mmHg and/or a diastolic pressure equal to or higher than 90 mmHg. Any worker reporting the use of anti-hypertensive medication was also classified as hypertensive, regardless of the measured pressure level. Blood pressure was measured by the attendant physician once at the beginning of the medical consultation, always on the left arm of a patient seated. In cases in which the blood pressure measurement was above the cut-off limits, a second measurement was performed at the end of the consultation. In these situations, only the second measurement was reported in the medical records.
Exposure to noise was measured at Homogeneous Exposure Groups (HEG) by using a digital audio dosimeter affixed next to their aural point to a randomly selected worker within the HEG. The measurement was done at a single moment in cases of continuous exposure to noise and during at least 75% of a workday of 8 hours in cases of intermittent exposure to noise, according to the Fundacentro methodology. We categorized noise exposure in three levels: ≤75 dB(A), from 75 to 85 dB(A), and ≥85 dB(A). We based our decision to use the three level categorization on some studies that also considered lower levels for noise exposure.
Statistical Analysis
Since the available dataset was not created to be used in a research project, not all-important variables were readily available to be used in the statistical analyzes. We decided to analyze confounding and interaction terms on any available variable with some indication in the biomedical literature of association to either noise exposure or high blood pressure. Therefore, age, gender, marital status, education, socioeconomic condition, physical activity, tobacco use, alcohol consumption, body mass index (BMI), type of service and heat exposure were the co-variables used in our analysis.
We began our data analysis by describing the frequency distribution of all variables according to the level of noise exposure. Thereafter, we performed a bivariate analysis for the association between all co-variables and both the exposure and the outcome variables in order to identify possible confounding variables for the studied association. All variables were categorized and a Chi-squared test used to test the associations. Any variable associated to the outcome with a p-value of at least <0.15 was considered to be included in the final multivariate model.
First order interactions terms between noise and age, gender, socioeconomic condition, physical activity, tobacco use, consumption of alcohol, body mass index and exposure to heat were individually tested and considered for the inclusion in the final multivariate model. Therefore, the baseline multivariate model was composed by all variables with a p-value of less than 0.15 in the exploratory analysis and statistically significant interaction terms. A logistic regression model with backward Wald elimination was used to model the independent association between noise exposure and high blood pressure, while considering other possible confounding variables. We present the final multivariate model only with variables statistically associated with high blood pressure since addition of non-significant variables did not add to the fitness of the final logistic model. The final odds ratio represents the relative odds of presence of high blood pressure. Statistical analysis was performed with the statistical package SPSS (version 17; SPSS Inc. software products).
The study was approved by the Ethics Committee at the National School of Public Health (CEP/ENSP). We analyzed a secondary dataset with no identifiers and the study was granted with a consent form waiver by the CEP/ENSP.