Introduction

High blood pressure (BP) is the leading risk factor for cardiovascular disease (CVD) [1]. CVD risk is increased by more than twofold in the presence of hypertension [2]. Approximately 60% of individuals will receive hypertensive medication during their lifetime [3]. According to national surveys in Canada and the United States, 20–40% of adults receiving pharmacologic treatment for hypertension have uncontrolled hypertension [4, 5]. Uncontrolled hypertension can lead to serious consequences for the cardiovascular system, including increased incident heart failure [6], chronic kidney disease [7], and cardiovascular mortality [8].

Prospective epidemiological studies have identified psychosocial stressors at work as important risk factors for BP elevation and hypertension [9,10,11,12,13] and for CVD incidence and recurrence [14, 15]. However, little is known about the effect of these stressors on the BP of workers treated for hypertension. Karasek’s demand–latitude model is the most widely used model to define these stressors. This model postulates that a combination of high psychological job demands and low job control (i.e., job strain) increases the risk of developing stress-related health problems. Only one previous study examined the effect of job strain on uncontrolled hypertension, and it found no association [16]. However, this study was conducted exclusively among men and used casual BP measurements [16]. Because casual BP measurements lack validity and precision [17], especially for assessing the effect of psychosocial stressors at work on BP [10], the effect of these stressors on hypertension control in men and women treated for hypertension is yet unknown.

The objective of this study of white-collar workers treated for hypertension was to examine, for the first time, whether those exposed to psychosocial stressors at work had a higher prevalence of uncontrolled hypertension, measured using ambulatory BP (ABP), than unexposed workers.

Methods

Study design and population

The study design and population are described elsewhere [18, 19]. Briefly, the study relied on a repeated cross-sectional design, which is well suited for dynamic populations such as work organizations as it accommodates the addition of new participants over multiple study time points. The study population was composed of white-collar public insurance workers from three institutions in Quebec City, Canada. The workers from each institution were located at the same worksite. Their main professional activities were planning and providing insurance services to the general population. Data were collected at three time points: at baseline (2000–2004), 3 years later (2004–2006), and 5 years later (2006–2009). All eligible employees were invited to participate at each time point. Participation rates for each collection point ranged between 80% and 86%. The study was approved by the Research Center of the CHU de Québec ethical review board, and all participants provided their informed consent.

The current analyses were restricted to participants who reported using antihypertensive medication (n = 480) at one or more time points. A minimum of 20 ABP readings were required, in concordance with current recommendations [20]. A total of 753 observations met these criteria. Seven observations were excluded due to missing information on psychosocial stressors at work, and 25 observations were excluded because of missing information on covariates. Four observations were excluded due to insufficient working hours (≤21 h) to prevent possible misclassification due to insufficient exposure to psychosocial stressors at work.

Complete data were available for 464 workers and were included in the present analyses. The workers comprising our study population participated in one (n = 263), two (n = 149), or all three (n = 52) time points, for a total of 717 observations. The prevalence of uncontrolled hypertension was calculated using the number of observations as the denominator.

Psychosocial stressors at work

Psychosocial stressors at work were evaluated according to the demand–latitude model using validated scales. Both components of the model, psychological demands and decision latitude, were assessed with the French version of the Job Content Questionnaire (JCQ) [21]. The consistency, reliability, and validity of this instrument have been demonstrated [22, 23]. Psychological demands were measured using five items related to the characteristics of the job load, such as the work pace, the difficulty of the tasks performed, and the frequency of interruptions and conflicting tasks. Decision latitude was evaluated using six items related to the level of decision-making autonomy, learning opportunities and use of intellectual skills. The participants rated each item on a 4-point Likert-type scale. Responses were summed to obtain a total score for psychological demands and decision latitude. Tertile-based variables were created for psychological demands and decision latitude [11]. To examine the effect of job strain, we used cut-offs for high psychological demands (≥24) and low decision latitude (≤72), based on the medians for the general working population of Québec [24]. The passive group comprised workers with low demands and low latitude, the active group comprised workers with high demands and high latitude, and the job strain group comprised workers with high demands and low latitude. Other workers were classified as unexposed.

Ambulatory blood pressure

ABP was measured using the validated Spacelabs 90207 oscillometric device (Produits Médicaux Spacelabs, St-Laurent, Québec, Canada) [25, 26]. In a research office at the participant’s workplace, the device was placed on the participant’s non-dominant arm if the BP difference measured on both arms was less than 10 mmHg. Otherwise, the device was installed on the arm with the highest BP values. ABP was defined as the mean of all readings taken every 15 min and recorded by the Spacelabs monitor during regular working hours. The participants were white-collar workers whose tasks were mainly performed from a seated position. To minimize motion artifacts, clear instructions were given to participants in the case that the monitor performed a reading while they were in a standing position (i.e., they were instructed to stop walking and rest their arm at their side). Uncontrolled hypertension was defined as a mean daytime systolic ABP ≥135 mmHg and/or diastolic ABP ≥85 mmHg for non-diabetic participants and systolic ABP ≥125 mmHg and/or diastolic ABP ≥75 mmHg for diabetic participants [27].

Sociodemographic characteristics and lifestyle-related risk factors

The participants completed a self-reported questionnaire. This questionnaire measured age, gender, education, smoking status, alcohol intake, family history of CVD, and self-reported diabetes. The following categories for age were created: <40 years, 40–49 years, or ≥50 years. Education was categorized as follows: less than college completed, college completed, or university completed. Occupation was categorized as follows: clerical/technician, professional, and executive. Smoking was defined as smoking at least one cigarette daily. Alcohol intake over the past 12 months was classified into three categories: <1 drink weekly, 1–5 drinks weekly, or ≥6 drinks weekly. Family history of CVD was based on the declaration by the participant of a cardiovascular event, such as angina, myocardial infarction, coronary revascularization, or stroke, suffered by their father, mother, brother, or sister before the age of 60 years. The risk factors listed above were assessed using validated protocols [24, 28]. Weight (kg) and height (cm) were measured by a trained research assistant. Body mass index (BMI) was calculated by dividing the participant’s weight in kilograms by their height in meters squared.

Statistical analyses

First, descriptive statistics were computed. Categorical values were expressed as proportions and continuous values in means with standard deviations (SD). Prevalence ratios of uncontrolled hypertension with 95% CIs were computed with generalized estimating equations (GEE) using robust Poisson regression [29]. GEE takes into account individual changes in job strain exposure, covariates, and hypertension control over time and within-subject correlations due to repeated measurements. Models were adjusted for sociodemographic and cardiovascular risk factors (age, education, and family history of CVD) as well as for lifestyle and BP-related risk factors (smoking, BMI, physical inactivity, alcohol intake, and diabetes). Tertiles of psychological demands and decision latitude were analyzed separately and mutually adjusted. Because a priori evidence suggested the plausibility of a gender interaction in the relationship between job strain and uncontrolled hypertension [10, 30,31,32], results were presented separately for men and women. Data from all three time points were pooled. The stability of the association between job strain and the prevalence of uncontrolled hypertension at each time point was examined by entering interaction terms between exposure and time. The p values suggested a constant effect over time (p = 0.29 for men and p = 0.90 for women). A sensitivity analysis was conducted including work hours as a covariate and yielded similar estimates (Supplementary file). Statistical analyses with a two-tailed significance level were conducted using the statistical software package SAS, version 9.4.

Results

Table 1 presents a description of the study population. The participants’ mean age in years was 51.6 (SD = 6.6) for men and 49.6 (SD = 6.3) for women. The prevalence of job strain was 21.3% and was nearly two times higher among women (27.5% vs 15.0%) than men. The mean ABP was 132.7/84.1 mmHg (SD = 11.2/7.9) in men and 126.8/79.6 mmHg (SD = 11.4/8.4) in women. The prevalence of uncontrolled hypertension based on ABP monitoring was 49% and was higher in men than in women (61.6% vs 36.9%).

Table 1 Characteristics of the study population (Quebec, Canada), 2000–2009 (n = 717 observations)
Full size table

Table 2 presents the prevalence of uncontrolled hypertension according to job strain exposure. Among men, the prevalence of uncontrolled hypertension was higher in those exposed to job strain (PR = 1.46, 95% CI: 1.07–1.98). The prevalence of uncontrolled hypertension was also higher in those with active jobs (PR = 1.47, 95% CI: 1.12–1.94). In women, there was no association between job strain and the prevalence of uncontrolled hypertension.

Table 2 Adjusted prevalence and prevalence ratio of uncontrolled hypertension according to job strain exposure (Quebec, Canada), 2000–2009
Full size table

Table 3 presents the prevalence of uncontrolled hypertension according to exposure to psychological demands and decision latitude. Both men and women in the highest tertile of psychological demands had a higher prevalence of uncontrolled hypertension after adjustment for decision latitude and sociodemographic and lifestyle-related risk factors. Men in the highest tertile of exposure for psychological demands had a 60% increase in the prevalence of uncontrolled hypertension (PR = 1.60, 95% CI: 1.25–2.06). Women with the same level of exposure to psychological demands also had a 60% increase in the prevalence of uncontrolled hypertension (PR = 1.60, 95% CI: 1.03–2.47). Exposure to low decision latitude was not associated with a statistically significant increase in uncontrolled hypertension in either men or women.

Table 3 Adjusted prevalence of uncontrolled hypertension according to psychological demands and decision latitude at work (Quebec, Canada), 2000–2009
Full size table

Discussion

In the present study of middle-aged workers treated for hypertension, men exposed to job strain and those with active jobs had a higher prevalence of uncontrolled hypertension than unexposed men. High demands at work were associated with a higher prevalence of uncontrolled hypertension in both men and women.

Uncontrolled hypertension among treated patients is a major public health issue. In the present study, nearly one out of two treated participants (49%) had this condition. This prevalence is noticeably higher than that reported in a previous Canadian study, in which 18% of patients being treated for hypertension had uncontrolled hypertension [5]. The elevated prevalence observed in the present study could be explained by the use of ABP measurements. Indeed, a previous large-scale European study showed that 31% of apparently well-controlled hypertensives at the clinic have uncontrolled hypertension according to ABP monitoring [33]. Therefore, ABP monitoring should be considered the preferred method for accurately assessing uncontrolled hypertension among treated patients.

The findings of the present study suggest that psychosocial stressors at work are important contributors to the alarmingly high prevalence of uncontrolled hypertension among treated patients. Prior to the present study, evidence on this topic was scarce. Mezuk et al. [16] found no association between job strain and BP control in a US sample of aging working men. In that study, the effect of job strain exposure was examined in combination with other work stressors, which could have diluted the effect and thus limits comparison with our results. Moreover, the use of casual BP could also have led to an underestimation of the effect of job strain [10, 19]. A previous study conducted by our group used ABP measurements and reported a higher prevalence of uncontrolled hypertension among workers exposed to psychosocial stressors according to Siegrist’s effort-reward imbalance model [19]. Karasek’s demand–latitude model and the effort-reward imbalance model cover different aspects of the psychosocial environment. The demand–latitude model focuses on task-level characteristics, while Siegrist’s model focuses on broader socioeconomic conditions, such as salary, promotion prospects, and job stability. The present findings provide complementary evidence supporting the effect of adverse psychosocial stressors at work on poor hypertension control among treated workers.

Karasek’s demand–latitude model poses that workers exposed to job strain (a combination of high psychological demands and low decision latitude) are the most vulnerable to developing work-stress-related adverse health effects [34]. The findings of the present study are consistent with this hypothesis, showing a 46% increase in the prevalence of uncontrolled hypertension among men exposed to job strain after sociodemographic and lifestyle-related risk factors were controlled. A higher prevalence of uncontrolled hypertension was also observed among men in active jobs, which are characterized by high psychological demands and high decision latitude. This finding does not fully correspond to Karasek’s proposition [34]. This result may partly be explained by the fact that in some situations, higher autonomy over work may not compensate for the pressure of increased psychological demands at work [35]. The worse cardiovascular profile for active workers observed in the present study is consistent with some previous evidence [31, 36,37,38].

In our study, low decision latitude had no effect on hypertension control in either men or women. This is consistent with some [31, 37, 38], but not all [10], previous studies documenting the effect of this particular exposure on BP. In comparison, high psychological demands were associated with a 60% increase in the prevalence of uncontrolled hypertension among men and women. Comparisons with previous studies should be made with caution because the current study is the first to examine the independent effect of decision latitude and psychological demands on uncontrolled hypertension in a sample composed exclusively of treated workers.

Gender differences have been observed in terms of the adverse impact of high job strain on BP elevation [11] and CVD incidence [39], which is consistent with our results. A recent study conducted among treated workers suggests that job strain could be associated with the rate pressure product (systolic BP × heart rate) in women [40]. Gender differences may be explained by different occupational trajectories among women than among men, which result in less continuous exposure to adverse psychosocial stressors at work [10]. In addition, factors such as marital strain and family responsibilities may interact with psychological demands at work and contribute to explaining the relative importance of this specific work factor on women’s BP control. One previous study showed an interaction between low marital cohesion and high job strain in the elevation of BP among women [41]. Similarly, previous studies have observed that combined exposure to high family responsibilities and high job strain tended to have a greater effect on BP than job strain alone [42, 43].

Limitations should be mentioned. First, the duration of exposure was not taken into account. However, previous studies have shown that, if anything, the deleterious effect of psychosocial exposures is stronger when duration of exposure is considered [10, 15]. Moreover, the possibility of selection bias cannot be entirely ruled out despite the high participation rate at each time point. In the present study, information on the type and number of antihypertensive agents being taken and on treatment adherence was not available. Psychosocial stressors at work could reduce adherence to health recommendations among workers with CVD [44]. However, controlling for adherence would lead to overadjustment because adherence could result from the psychosocial stressors under study (intermediate factor). Finally, because the study population was composed of white-collar workers, generalization of our findings should be restricted to workers sharing similar occupations. It should, however, be noted that the deleterious effect of psychosocial stressors at work on BP was previously observed in different occupational groups (i.e., nurses, bus drivers) [10]. In addition, the proportion of workers exposed to job strain in the present study (21% in the total sample) was comparable to that in other industrialized countries [45]. The composition of our study population had the advantage of minimizing potential confounding from physical work constraints that may influence BP at work.

This study also has several strengths. First, the present study examined, for the first time, the association between psychosocial stressors at work based on the demand–latitude model and uncontrolled hypertension among both men and women. In addition, BP was measured using ambulatory monitoring, which is the gold standard method for assessing hypertension [17, 27]. Daytime ABP was used, which had the advantage of minimizing biases due to low participation or incomplete BP data [46], as supported by the high participation rates obtained at multiple measurement points. It should be noted that a recent systematic review suggested that the deleterious effect of psychosocial stressors at work on BP is of a comparable magnitude whether BP was measured at work, at home, over 24 h or during the daytime [10]. A large set of potential confounders, including known hypertension risk factors, were taken into consideration, minimizing the possibility of strong confounding. Finally, validated scales were used to assess job strain [22, 23], minimizing the possibility of information bias.

The current study has both clinical and public health implications. First, there is a need for increased clinical awareness regarding the negative effect of these work stressors on BP control. The identification of treated patients who are exposed to work stressors could contribute to improving their BP management. For example, ABP monitoring would give a more accurate assessment of BP control than office or home BP measurement. Second, at the population level, preventive interventions that aim to reduce exposure to job strain through organizational changes could be relevant strategies for decreasing the burden of uncontrolled hypertension.

Conclusion

The current study is the first to evaluate the effect of psychosocial stressors at work, based on the demand–latitude model, on the prevalence of uncontrolled hypertension in treated workers using ABP measures. The prevalence of uncontrolled hypertension was higher among workers exposed to psychosocial stressors at work, as described by this model, than among unexposed workers. Addressing these frequent and modifiable adverse workplace exposures should be considered for at-risk and hypertensive patients. Workplace interventions that reduce the prevalence of psychosocial stressors at work could improve BP control at the population level.