Psychosomatic Medicine
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Published online before print December 24, 2007, 10.1097/PSY.0b013e31815c4103
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Smith, P.
Right arrow Articles by Mustard, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Smith, P.
Right arrow Articles by Mustard, C.
Related Collections
Right arrow Other Epidemiology
Psychosomatic Medicine 70:85-91 (2008)
© 2008 American Psychosomatic Society

ORIGINAL ARTICLES

Do Changes in Job Control Predict Differences in Health Status? Results From a Longitudinal National Survey of Canadians

Peter Smith, PhD, John Frank, MD, Susan Bondy, PhD and Cameron Mustard, ScD

From the Institute for Work and Health (P.S., J.F., C.M.), Toronto, Canada; Institute for Medical Science (P.S.), University of Toronto, Toronto, Canada; Department of Public Health Sciences (J.F., S.B., C.M.), University of Toronto, Toronto, Canada; Canadian Institutes of Health Research (J.F.), Institute of Population and Public Health, Toronto, Canada; and the Institute for Clinical Evaluative Sciences (S.B.), Toronto, Canada.

Address correspondence and reprint requests to Peter Smith, Institute for Work and Health, 481 University Avenue, Suite 800, Toronto, ON, Canada M5G 2E9. E-mail: psmith{at}iwh.on.ca


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 NOTES
 APPENDIX
 REFERENCES
 
Objective: To examine the effect of changes in job control on health behaviors, psychological distress and health status.

Methods: Using a path analysis model, we examined the effects of change in job control over a 4-year period on levels of physical activity, smoking, and psychological distress; and on self-rated health over an additional 2 years, among a representative sample of 2221 Canadians.

Results: Over the 4-year period, 280 respondents reported decreases in job control, and 256 reported increases in job control. Health at baseline was not associated with the likelihood of changes in job control. We found a graded relationship between change in job control and levels of physical activity and psychological distress over a 4-year period; and levels of self-rated health over a 6-year period, with positive change in job control associated in higher levels of physical activity and self-rated health and lower levels of distress.

Conclusions: The results of this study suggest that both level of job control and changes in job control have direct and indirect effects on health status over time. Future research should focus on developing precise measures of work exposures, and examine differences between changes in job control due to only changes in perceptions and changes due to work redesign.

Key Words: job control • change • longitudinal • self-rated health • working population

Abbreviations: NPHS = National Population Health Survey; SRH = self-rated health.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 NOTES
 APPENDIX
 REFERENCES
 
The psychosocial work environment is defined as "the range of opportunities given to an individual to meet his or her need of well being, productivity and positive self-experience" (1). To date, the most common measures of the psychosocial work environment are Karasek and Theorell’s Demand-Control (or job-strain) model (2), and Seigrist’s Effort-Reward Imbalance (ERI) model (3,4), although recent attention has also focused on organizational justice (5).

Although there is a large body of research demonstrating within-group differences in health between those exposed to poor psychosocial work environments and those who are not, this does not constitute evidence that changing these conditions would result in better (or worse) health. Yet, from both an organizational and public policy perspective, evidence that changes in psychosocial working conditions are associated with subsequent changes in health status is important if possible health effects are to be incorporated into decisions likely to affect the psychosocial work environment. The objective of the current paper is to examine the effects of changes in job control (one measure of the psychosocial work environment) on health behaviors, household income, psychological distress, and self-rated health (SRH) over a 6- to 8- year period.

Compared with the number of studies examining the impact of job control, measured at one point in time on future health status, few studies have examined whether changes in psychosocial working conditions are associated with differences in health status. Of those examining change in job control in particular, decreases in job control over a 2-year period was associated with higher sickness absence rates over the next 4 years, among 530 Finnish full-time workers, compared with those who reported positive change in job control (6). A study of British civil servants reported that, compared with respondents with no change in decision latitude (one dimension of job control), respondents with decreased decision latitude had more sickness absence spells over the subsequent 2- and 5-year periods (7).

Other studies have examined if changes in other aspects of the psychosocial work environment (e.g., job strain, ERI, and organizational justice) affect health status. Some have found no association between change in the psychosocial work environment and outcomes of sickness absence (8) or changes in plasma cortisol and plasma prolactin (9). Others have reported that there is an association between their health outcome and change in psychosocial work environment in one direction but not the other; with some reporting no association when change occurs in a positive direction (7); and others reporting no association when change occurs in a negative direction (8,10,11). In addition, one study examining self-reported angina reported an association among men, but not among women (11).

Using path analysis, we aimed to examine a) baseline factors associated with changes in job control over the subsequent 4 years and b) the effects of change in job control on levels of health behaviors, psychological distress, and SRH.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 NOTES
 APPENDIX
 REFERENCES
 
Data
This study is based on a secondary analysis of data from the Canadian National Population Health Survey (NPHS). Starting in 1994, the NPHS has collected information including health, working conditions, and living conditions, from a representative sample of >17,000 Canadians every 2 years. To date, five cycles are available for analysis (1994–95, 1996–97, 1998–99, 2000–01, and 2002–03). Further details about the NPHS are described elsewhere (12–14). For this study, only respondents aged 25 to 60 years, who were nonself-employed labor force participants, working >20 hours per week, without limitations restricting the type or amount of work they could do at baseline (1994–95), were included in the study sample (n = 4886). As perceptions of job control were only conducted in the 1994 and 2000 surveys, only three waves of the NPHS (1994–95, 2000–01, and 2002–03) are included in these analyses.

Main Outcome
Self-Rated Health
SRH is measured in each cycle of the NPHS, using a 5-point scale (excellent, very good, good, fair, poor) (15–17). Level of SRH has been shown to predict future morbidity and mortality in a variety of different populations (15,17–20). In this study, we examined the level of SRH in 2002, at the same time adjusting for level of SRH at baseline. In our models, we chose to define SRH as an ordered categorical variable with higher scores indicating better health.

Intermediate Outcomes
Health Behaviors
We included current smoking status and energy expenditure due to leisure time sport and recreational physical activity (incorporating time, duration, and frequency) in the previous 3 months. Smoking was measured as an ordinal categorical variable (never, former, <5 cigarettes/day, 5–14 cigarettes/day, 15–24 cigarettes/day, and ≥25 cigarettes/day). Physical activity level (expressed as kcal/kg/day) was log transformed because of the skew and kurtosis of scores.

Psychological Distress
The measure of psychological distress used in the NPHS comprises a subset of six questions from the University of Michigan’s revision of the Composite International Diagnostic Interview. Each question is answered on a 5-point Likert scale, giving a respondent a score between 0 and 24, with higher scores indicating greater distress (Cronbach’s {alpha} = 0.77) (21). Questions are listed in the Appendix.

Household Income Adequacy
As changes in job control may be associated with changes in occupations, these changes might also reflect changes in income. The NPHS does not contain a measure of individual income from salaries; however, it does contain a measure of household income adequacy. Household income adequacy, measured in quartiles, was derived based on both the self-reported total household income and the number of people who reside in the household (14) (Appendix 1).

Main Independent Measures
Baseline Job control
Job control was measured using five items from an abbreviated measure of Karasek and Theorell’s Job Content Questionnaire (2). Respondents were asked to state their level of agreement with five statements on a 5-point Likert scale (Appendix 1). The internal consistency for job control within this data set has previously been reported as {alpha} = 0.61 (22). Higher scores indicate greater control at work.

Although items on psychological demands at work are contained within the NPHS, we did not use these items, given concerns regarding both the validity and reliability of the subset of questions used in the NPHS (22), and in the changing and complex meaning of job demands in general (23).

Change in Job Control
Change in job control was assessed using a difference score between time 2 and time 1 estimates, after confirming that respondents had reported changes in job control greater than the level of minimum detectable change in the job control score. Higher scores indicate a more positive change in job control.

Estimating the minimal detectable change in a measure, previously described by Jacobson and colleagues (24,25), incorporates both the variance in scores at time 1 and the estimated day-to-day variability in the measure (i.e., the variability in scores when no change has actually occurred). Using test-retest data over a 2-week period from 48 participants in the Ontario Child Health Survey (26), we estimated that respondents with changes in job control (range 0–20) of <±4 could be considered not to have changed greater than the expected day-to-day variability in the job control instrument (at 95% confidence level).

Because of concerns with the reliability of the difference score, we also grouped respondents into three categories based on the minimum detectable change: a) those who decreased in job control; b) those with no change in job control; and c) those who increased in job control. We compared results using the difference score with results using change as a three-level categorical variable. Given we found limited differences between methods, for ease of interpretation, we have only reported results with change measured as a difference score. Results using change as a three-level categorical variable are available on request.

Other Covariates
The pathways between job control and change in job control, and each outcome (physical activity, smoking, household income, psychological distress, and SRH) were adjusted for the following baseline variables: education level (five groups); gender; age (continuous); body mass index (continuous); presence of self-reported hypertension, chronic back pain, and heart disease (yes/no); and SRH status at baseline (categorical).

Analysis
Data preparation was performed using SAS software (SAS Institute, Cary, North Carolina) (27). Path analysis models were performed using Mplus software (Mplus User’s Guide, Los Angeles, California) (28). The advantages of using path analysis (compared with a traditional regression model) are: a) it allows concurrent examination of covariates at baseline (e.g., health status) and the likelihood of changes in job control, and the effects of these changes on the level of health behaviors, household income, and psychological distress in 2000, and health status in 2002; and b) it allows covariances between independent variables in a model to be specified (e.g., baseline job control and level of SRH). Path models provide estimates of both the direct effects and the indirect effects of variables on the outcomes. Indirect effects are calculated by multiplying the unstandardized coefficients of the paths between a given variable and an outcome; e.g., the indirect path between change in job control and SRH, via physical activity, is calculated by multiplying the coefficient between change in job control and physical activity and the pathway between physical activity and SRH (29). Estimates in Mplus for categorical outcomes, such as SRH status and smoking, are reported as probit coefficients. We chose to estimate these variables in a probit model under the assumption that each categorical variable is a proxy for a true underlying continuous normal distribution. In general, {chi}2 statistics, p values, and general conclusions are usually similar for probit and logit models (30,31).

For continuous outcomes, such as physical activity and distress, coefficients from Mplus are normal linear regression coefficients (28). T statistics presented (b/se) can be interpreted as similar to z scores (e.g., a t stat of 1.96 = p = .05).

Of the initial study sample (n = 4886), 588 (12%) respondents did not report information on job control at baseline. Male respondents and those who were interviewed in French were more likely to have missing responses. An additional 225 respondents were missing information on other baseline covariates, leaving an eligible sample of 4053. Respondents who were younger and female were more likely to be missing baseline covariates (in particular, body mass index). No relationship was found between missing responses and baseline education or health status.

Of this sample, 2434 (60%) respondents were still working >20 hours/week, were nonself-employed, and were without activity limitations that prevented the type of work they could do in the 2000 survey. Of those who were no longer working: 63 (1.5%) respondents had died or were institutionalized; 243 (6%) respondents were unable to work because of an illness or disability; 205 (5%) respondents had retired; 315 (8%) respondents were either now self-employed or were working <20 hours/week; 79 (2%) respondents were no longer working because of "reasons other than illness" or retirement (e.g., caring for children); 63 (2%) respondents were looking for work; and 651 (16%) respondents did not respond to the 2000 survey, or did not answer questions on labor force status. Low levels of education were associated with nonresponse to the 2000 survey and not working because of reasons other than illness. High baseline job control was associated with moving into self-employment or working <20 hours/week. Older age, higher levels of smoking, high psychological distress, and poorer SRH were associated with no longer being able to work because of illness or disability.

Of those respondents still working in 2000 (n = 2434), 21 were pregnant and were removed as this may affect levels of health behaviors, psychological distress, and health; and 149 (6%) were missing information on job control, physical activity, smoking, psychological distress, or household income in 2000. No relationship was found between missing responses and any baseline variable. An additional 164 respondents did not respond to the 2002 survey, and three more died. As we did not have information on SRH for these respondents, they were removed, leaving a total sample of 2097.

Relationships between baseline variables and change in job control, and change in job control and the intermediate outcomes, and the final health outcome were examined using one path model (Figure 1).


Figure 113
View larger version (12K):
[in this window]
[in a new window]

  		Figure 1. Path model examining change in job control and level of smoking, physical activity, psychological distress, and household income in 2000, and self-rated health in 2002. Level of the intermediate outcomes (smoking, physical activity, psychological distress, and household income) have been adjusted for the baseline level of each outcome, as has the final outcome of self-rated health.

 

In addition to measuring change as a difference score and an ordered categorical variable, we also performed a series of models examining the effects of change in job control in each direction separately (i.e., examining if the effects found were stronger in a positive or negative direction). As the results of this analysis were similar to those using change as a difference score, they are not presented here but are available from the authors on request.

We examined if the pathways between job control and change in job control on the intermediate health outcomes, household income, and SRH differed by gender. As they did not, results are reported adjusting for gender. As baseline measures of smoking, physical activity, psychological distress, household income, and SRH status are included in the path model, positive estimates can be interpreted as having a higher than expected level (or categories) of these outcomes, after taking baseline levels of each into account, and vice versa for negative estimates.

Although the NPHS survey is conducted using a complex stratified cluster sampling design, limited information about the sampling procedure is supplied; therefore, adjustment of standard errors for the cluster design was not possible. We did compare our results with separate probit and linear regression models, adjusted using a bootstrap procedure supplied by Statistics Canada (32). Although this comparison showed limited changes in variance estimates, we recommend caution when interpreting results, particularly those with a marginal levels of probability (p = .05). All analyses were weighted to account for the probability of selection, and for nonresponse, in the initial survey.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 NOTES
 APPENDIX
 REFERENCES
 
Table 1 describes the distribution of the study sample across changes in job control between 1994 and 2000, using the minimum detectable change. Table 2 presents the coefficients for baseline variables on change in job control as a difference score. Respondents with higher levels of education and household income were more likely to have positive changes in job control. Respondents with higher job control and those who were female were less likely to have positive changes in job control. No strong associations were found between measures of health and likelihood of changes in job control.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Distribution of Study Variables at Baseline by Change in Job Control Between 1994 and 2000 (n = 2097)

 

View this table:
[in this window]
[in a new window]

  		TABLE 2. Unstandardized Coefficients (b) for Baseline Variables and Change in Job Control as a Difference Score (n = 2097). All Estimates Are Adjusted for All Other Variables in the Table

 

Table 3 presents the regression coefficients for both low job control at baseline, and change in job control between 1994 and 2000, on levels of physical activity, smoking, psychological distress, and household income in 2000. Positive change in job control was associated with higher than expected levels of physical activity and household income, and lower than expected levels of psychological distress, after adjusting for the baseline level of each outcome. No relationship was found between baseline job control or change in job control and level of smoking.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Unstandardized Regression Coefficients (b) for Job Control, and Change in Job Control, on Intermediate Health Behaviors, Psychological Distress, and Household Income in 2000 (n = 2097)

 

Table 4 presents the direct and indirect regression coefficients for both low job control and change in job control between 1994 and 2000, and SRH in 2002. Higher job control and positive change in job control were directly associated with higher than expected SRH in 2002. Indirect effects on SRH were also present between job control and change in job control, via physical activity and psychological distress levels in 2000. No indirect effects were present between job control and change in job control via smoking or household income, although a direct relationship was present between smoking in 2000 and health status in 2002.


View this table:
[in this window]
[in a new window]

  		TABLE 4. Unstandardized Regression Coefficients (b) for Direct and Indirect Paths Between Job Control, and Change in Job Control, on Self-Rated Health in 2002 (n = 2097)

 


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
NOTES
 APPENDIX
 REFERENCES
 
The objectives of this paper were to examine a) baseline factors associated with changes in job control over the next 4 years; b) the effects of change in job control on subsequent levels of health behaviors, psychological distress, and household income; and c) the direct and indirect effects of job control and change in job control on SRH. This study adds several findings to the literature examining the relationship between changes in job control and subsequent health status. We found no evidence that health status at baseline was related to the likelihood of positive change in job control over the next 4 years; both higher job control at baseline and positive change in job control were associated with higher levels of physical activity and lower levels of psychological distress in 2000, which in turn had an indirect effect on SRH in 2002; and both higher job control and positive change in job control were directly associated with higher levels of SRH in 2002.

Our results should be interpreted given the following limitations. All measures included in this study were based on self-report; therefore, there is a possibility the associations presented here are inflated due to common method variance in survey responses (e.g., due to negative affectivity among respondents) (33–35). In particular, some concerns have been raised over the consistency between SRH and true health status among particular population subgroups. As we measured both change in job control and differences in levels of physical activity and psychological distress over the same time period (1994 and 2000), it is possible that the associations we observed may operate in the opposite direction. For example, increases in the level of physical activity may lead to perceptions of higher job control (rather than positive changes in job control leading to higher levels of physical activity). Unfortunately, there is no way for us to test this hypothesis in our data. However, baseline physical activity had no effect on changes in job control over the next 4 years; yet, baseline level of job control did have an effect on the level of physical activity in 2000, supporting a casual relationship between changes in job control on physical activity, and not vice versa.

Given our data source, we cannot be sure if the changes in job control we observed were due to the actual change in working environments, or due to changes in perceptions (within-person shifts). As such, our results cannot be extended to evidence that intentionally changing working environments will result in differences in levels of health. This will require well-designed intervention studies with reliable and valid, subjective and objective measures of job control.

This study also has a number of strengths. These include the large representative sample, our ability to ensure that changes in job control were greater than day-to-day variability, and an analytical technique which allowed us to simultaneously examine different pathways through which both low job control and change in job control were associated with health behaviors, psychological distress, and SRH status.

We found that positive changes in job control were associated with lower levels of psychological distress, similar to previous findings linking changes in organizational justice to psychiatric morbidity (36). However, this association was no longer present when we restricted our sample to those who did not change occupations between 1994 and 2000 (results not presented, but available on request), suggesting that lower psychological distress levels were due to changes in occupation, which in turn caused changes in level of job control. Similar to previous work by Head et al. (7), we found no evidence of a drift hypothesis between lower health at baseline and increased probability of subsequent decreases in job control. However, as outlined previously in our attrition analysis, we did find that lower levels of SRH and higher levels of psychological distress were associated with increased probability of being unable to work because of illness or injury, or having a health problem which limited the amount or type of work that the respondent could do at follow-up.

Only a small number (24%) of respondents in our study reported changes in job control greater than the estimated day-to-day variability in our job control measure. This is less than the 37% of respondents who changed in level of job control in Head’s study (7); the only other study of which we are aware has taken the day-to-day variability of job control scores into account when estimating change (7). The study by Head et al. used a more precise measure of job control (15 items) compared with the five items used here. If we had a more precise measure of job control in our study, we may have found a larger number of respondents reporting changes in job control due to both greater variance in scores from a larger number of job control dimensions being measured and a higher reliability of job control estimates.

Positive changes in job control between 1994 and 2000 predicted both higher than expected levels of physical activity in 2000, lower than expected psychological distress, and higher than expected SRH in 2002. Similarly, negative changes in job control predicted lower than expected physical activity and lower than expected SRH. For example, when using categories of change in job control rather than a difference score, after adjusting for various baseline factors and level of job control at baseline, we compared those respondents with positive changes in job control and found those with negative changes had 22% higher incidence of inactivity and 23% higher incidence of good, fair, or poor SRH1 (Figure 2).


Figure 213
View larger version (14K):
[in this window]
[in a new window]

  		Figure 2. Percent of respondents who were inactive in 2000, and who reported good, fair, or poor self-rated health in 2002, by change in job control between 1994 and 2000 (N = 2,097). Adjusted for baseline: job control education, age, gender, self-rated health, hypertension, BMI, heart disease, and level of physical activity in 1994; self-rated health outcome also adjusted for psychological distress, physical activity, household income, and smoking in 2000.

 

The pathways through which low control at work results in lower physical activity levels may include a diffusion of lack of control at work to lack of control in general, or feelings of helplessness, making participation in physical activity more challenging; or respondents with low control at work may not have time to plan opportunities for activity participation. More research should examine if changes in job control also affect other activities outside of work that have also been associated with higher health status (e.g., social engagement). Research should also examine if changes in job control, through work redesign, have similar effects on differences in health status as changes in job control created through changes in perception.

Dr. John Frank is supported through Scientific Director’s Grant IOP-44972 from the Canadian Institutes of Health Research.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 NOTES
 APPENDIX
REFERENCES
 


View this table:
[in this window]
[in a new window]

  		TABLE A1. Study Questions

 


View this table:
[in this window]
[in a new window]

  		TABLE A2. Household Income Adequacy: Description of Household Income and Persons in Household Within Each Group of Household Income Adequacy

 


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 NOTES
APPENDIX
 REFERENCES
 
1These bottom three levels of self-rated health are usually grouped when dichotomizing self-rated health into good or poor, with excellent and very good combined to form the good category. Back

Received for publication January 4, 2007; revision received August 28, 2007.

DOI:10.1097/PSY.0b013e31815c4103


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 NOTES
 APPENDIX
 REFERENCES
 

  1. Siegrist J, Marmot M. Health inequalities and the psychosocial environment—two scientific challenges. Soc Sci Med 2004;58:1463–73.[CrossRef][Medline]
  2. Karasek R, Theorell T. Healthy Work: Stress Productivity and the Reconstruction of Working Life. New York: Basic Books Inc.; 1990.
  3. Siegrist J, Peter R. Measuring Effort-Reward Imbalance at Work: Guidelines. Dusseldorf, Germany: Heinrich Heine University; 1996.
  4. Siegrist J. Adverse health effects of high-effort/low-reward conditions. J Occup Health Psychol 1996;1:27–41.[Medline]
  5. Kivimaki M, Elovainio M, Vahtera J, Ferrie J. Organisational justice and health of employees: prospective cohort study. Occup Environ Med 2003;60:27–34.[Abstract/Free Full Text]
  6. Vahtera J, Kivimaki M, Pentti J, Theorell T. Effect of change in the psychosocial work environment on sickness absence: a seven year follow up of initially healthy employees. J Epidemiol Community Health 2000;54:484–93.[Abstract/Free Full Text]
  7. Head J, Kivimaki M, Martikainen P, Vahetera J, Ferrie JE, Marmot MG. Influence of change in psychosocial work characteristics on sickness absence: the Whitehall II study. J Epidemiol Community Health 2006;60:55–61.[Abstract/Free Full Text]
  8. de Lange AH, Taris TW, Kompier MA, Houtman I, Bongers P. Effects of stable and changing demand-control histories on worker health. Scand J Work Environ Health 2002;28:94–108.[Medline]
  9. Theorell T, Perski A, Akerstedt T, Sigala F, Ahlberg-Hulten G, Svensson J, Enroth P. Changes in job strain in relation to changes in physiological state. Scand J Work Environ Health 1988;14:189–96.[Medline]
  10. Schnall PL, Schwartz J, Landsbergis P, Warren K, Pickering T. A longitudinal study of job strain and ambulatory blood pressure: results from a three-year follow-up. Psychosom Med 1998;60:697–706.[Abstract/Free Full Text]
  11. Chandola T, Siegrist J, Marmot M. Do changes in effort-reward imbalance at work contribute to an explanation of the social gradient in angina. Occup Environ Med 2005;62:223–30.[Abstract/Free Full Text]
  12. Tremblay L, Catlin G. Sample design of the national population health survey. Health Rep 1995;7:29–38.[Medline]
  13. Swain L, Catlin G, Beaudet MP. The national population health survey—its longitudinal nature. Health Rep 1999;10:69–82.[Medline]
  14. Statistics Canada. 1994 User Guide to the National Population Health Survey. Ottawa, Canada: Statistics Canada; 1994.
  15. Bjorner J, Kristensen T, Orth-Gomer K, Tibblin G, Sullivan M, Westerholm P. Self-Rated Health, A Useful Concept in Research, Prevention and Clinical Medicine. Stockholm: Forshingsradsnamnden. Swedish Council for Planning and Coordination of Research; 1996.
  16. Ware JEJ, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey. Manual and Interpretation Guide. Boston: The Health Institute, New England Medical Center; 1993.
  17. Idler E, Benyamini Y. Self-rated health and mortality: a review of twenty seven community studies. J Health Soc Behav 1997;38:21–37.[CrossRef][Medline]
  18. Kopec JA, Williams JI, To T, Austin PC. Measuring population health: correlates of the health utilities index among English and French Canadians. Can J Public Health 2000;91:465–70.[Medline]
  19. Power C, Matthews S, Manor O. Inequalities in self rated health in the 1958 birth cohort: lifetime social circumstance or social mobility? BMJ 1996;313:449–53.[Abstract/Free Full Text]
  20. Larsson D, Hemmingsson T, Allebeck P, Lundberg I. Self-rated health and mortality among young men: what is the relation and how may it be explained? Scand J Public Health 2002;30:259–66.[CrossRef][Medline]
  21. Marchand A, Demers A, Durand P. Do occupation and work conditions really matter? A longitudinal analysis of psychological distress experiences among Canadian workers. Social Health Illn 2005;27:602–27.[CrossRef]
  22. Wilkins K, Beaudet MP. Work stress and health. Health Rep 1998;10:47–62.[Medline]
  23. Kristensen TS, Bjorner JB, Christensen KB, Borg V. The distinction between work pace and working hours in the measurement of quantitative demands at work. Work and Stress 2004;18:305–22.[CrossRef]
  24. Jacobson NS, Follette WC, Revenstorf D. Psychotherapy outcome research: methods for reporting variability and evaluating clinical significance. Behav Ther 1984;15:336–52.[CrossRef]
  25. Jacobson NS, Truax P. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol 1991;59:12–9.[CrossRef][Medline]
  26. Boyle MH, Offord DR, Hofmann HG, Catlin GP, Byles JA, Cadman DT, Crawford JW, Links PS, Rae-Grant NI, Szatmari P. Ontario Child Health Study: Methodology. Arch Gen Psychiatry 1987;44:826–31.[Abstract/Free Full Text]
  27. The SAS Institute. The SAS System for Windows, Release 8.0. Cary, NC: The SAS Institute; 2000.
  28. Muthen LK, Muthen BO. Mplus User’s Guide. 3rd ed. Los Angeles: Muthen and Muthen; 2004.
  29. Kline RB. Principles and Practice of Structural Equation Modeling. New York: The Guilford Press; 1998.
  30. Greene WH. Models with discrete dependent variables. In: Econometric Analysis. 4th ed. New Jersey: Prentice-Hall; 2000.
  31. Allison PD. Binary logit analysis: details and options. In: Logistic Regression Using the SAS System: Theory and Application. Cary, NC: The SAS Institute; 1999.
  32. Yeo D, Mantel H, Liu TP. Bootstrap variance estimation for the national population health survey. In: Proceedings of the Annual Meeting of the American Statistical Association, Survey Research Methods Section, August 1999. Baltimore: American Statistical Association; 1999.
  33. Macleod J, Davey Smith G, Heslop P, Metcalfe C, Carroll D, Hart C. Are the effects of psychosocial exposures attributable to confounding? Evidence from a prospective observational study on psychological stress and mortality. J Epidemiol Community Health 2001;55:878–84.[Abstract/Free Full Text]
  34. Watson D, Pennebaker JW. Health complaints, stress, and distress: exploring the central role of negative affectivity. Psychol Rev 1989;96:234–54.[CrossRef][Medline]
  35. Macleod J, Davey Smith G, Heslop P, Metcalfe C, Carroll D, Hart C. Psychological stress and cardiovascular disease: empirical demonstration of bias in a prospective observational study of Scottish men. BMJ 2002;324:1247–51.[Abstract/Free Full Text]
  36. Ferrie JE, Head J, Shipley MJ, Vahtera J, Marmot MG, Kivimaki M. Injustice at work and incidence of psychiatric morbidity: the Whitehall II study. Occup Environ Med 2006;63:443–50.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Am J EpidemiolHome page
P. M. Smith and D. E. Beaton
RE: "CHANGES IN PERCEIVED JOB STRAIN AND THE RISK OF MAJOR DEPRESSION: RESULTS FROM A POPULATION-BASED LONGITUDINAL STUDY"
Am. J. Epidemiol., July 1, 2009; 170(1): 131 - 132.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Smith, P.
Right arrow Articles by Mustard, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Smith, P.
Right arrow Articles by Mustard, C.
Related Collections
Right arrow Other Epidemiology

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS