Psychosomatic Medicine
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Published online before print March 30, 2007, 10.1097/PSY.0b013e31803139a6
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 Google Scholar
Google Scholar
Right arrow Articles by Cohen, F.
Right arrow Articles by Stites, D. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Cohen, F.
Right arrow Articles by Stites, D. P.
Related Collections
Right arrow Immunology
Right arrow Stress and Coping
Psychosomatic Medicine 69:225-234 (2007)
© 2007 American Psychosomatic Society

ORIGINAL ARTICLES

Immune Function Declines With Unemployment and Recovers After Stressor Termination

Frances Cohen, PhD, Margaret E. Kemeny, PhD, Leonard S. Zegans, MD, Paul Johnson, MPhil, Kathleen A. Kearney, PhD and Daniel P. Stites, MD

From the Health Psychology Program, Department of Psychiatry (F.C., M.E.K, P.J., K.K.), Department of Psychiatry (L.E.Z.), and Department of Laboratory Medicine (D.P.S.), University of California San Francisco School of Medicine, San Francisco, CA.

Address correspondence and reprint requests to Frances Cohen, Health Psychology Program, University of California San Francisco School of Medicine, 3333 California Street, Suite 465, San Francisco CA 94143-0848. E-mail: frances.cohen{at}ucsf.edu


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Objective: To examine the effect of unemployment on natural killer cell cytotoxicity (NKCC) and, in a subsample of persons who become re-employed, to determine if, after termination of the stressor, immune values recover to levels similar to matched controls.

Methods: One hundred unemployed and 100 matched employed healthy men and women, aged 29 to 45 years, were followed for 4 months with monthly blood samples taken to measure NKCC, the ability of NK cells to kill target cells. Twenty-five participants obtained employment before the end of the study, leaving 75 unemployed (and 75 employed) participants in the main sample. For unemployed participants who obtained employment before the end of the study, subsample analyses compared NKCC levels before and after obtaining a new job.

Results: The persistently unemployed sample had significantly lower NKCC levels for all three effector:target ratios (100:1, p = .0004; 50:1, p = .002; and 25:1, p = .02) when compared with the matched employed sample. There were no significant gender effects. In the subsample analyses, NKCC was significantly higher after the participants became employed, compared with their unemployed period, with substantial "recovery" of immune function (44%–72%) compared with values from the steadily employed group.

Conclusions: Chronic stress is associated with persistent NKCC impairment. When the chronic stressor is terminated, however, the immune cell functional capacity quickly begins to recover. We believe this is the first study in humans to document immune function recovery after the definable end of a chronic stressor.

Key Words: chronic stress • natural killer cell cytotoxicity • immunity • unemployment • stress recovery

Abbreviations: NK = natural killer; NKCC = natural killer cell cytotoxicity.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
In recent years, there has been increased evidence that exposure to chronic stressors affects physical health. Numerous animal studies have demonstrated the deleterious health consequences of severe continuing stressors such as crowding or the social stress of changing the group's dominance hierarchy (1,2), although results are not always consistent across different types of stressors (3,4). There is evidence that, among humans, chronic stress is associated with cardiovascular disease (5) and infectious disease outcomes, such as increased likelihood of colds under controlled viral challenge conditions (6) and genital herpes recurrence (7).

McEwen and Seeman (8) argued that increased vulnerability to disease after chronic stressor exposure may be due to the sustained alterations in neurochemical activity elicited by chronic stressors, which can result in long-term, or even permanent, negative consequences for important bioregulatory systems. This persistent physiological dysregulation can result in increased "wear and tear" on the body (referred to as "allostatic load"), which may leave the organism more vulnerable to disease. Long-term changes in the immune system have been viewed as one possible pathway linking chronic stress to susceptibility to infectious, neoplastic, and autoimmune disease (9,10). A growing body of evidence supports the connection between exposure to chronic stressors and immune decrements in animals (11,12) and humans (13).

However, few research studies have investigated what happens to the immune system when a chronic stressor ends. Do immune parameters return to normal levels once the stressor ends or is there a long-term residual effect, as is suggested by the allostatic load framework? To answer this question, we studied the chronic stress of unemployment, a significant life stressor with a myriad of associated consequences. This paradigm provides an ongoing stressor that, for some of our participants, has a defined end-point—a new job—which allows us to answer questions about immune functioning after cessation of the stressor. This is the first study in humans that has examined changes in the immune function in individuals exposed to a chronic stressor where the stressor has a definable end-point without the initiation of a new chronic stressor. To determine the presence and extent of recovery, we compare poststressor immune values with values during the chronic stress.

Natural killer (NK) cells are a subset of lymphocytes that are seen as efficient effector cells, which can kill a broad array of targets including virally infected cells and cancer cells (10). Impaired NK function has been found to be associated with the development or progression of viral infections, autoimmune diseases, and cancer. For example, in a rodent model of breast cancer, stress-induced reductions in NK cell activity were associated with increases in metastases to the lung (14).

A meta-analysis (13) shows significant associations between exposure to chronic stressors in humans and decrements in functional immune parameters including the ability of NK cells to kill target cells. For example, impaired NK cell functioning was found in bereaved spouses (15), Bosnian prisoners of war (16), and hurricane victims (17). Our prior research with a population of healthy women found that chronic, but not acute, stress was associated with decreased NK cytotoxic activity (18). However, studies on Alzheimer's caregivers reported negative findings (19–21). Evidence remains limited in this area; results are not definitive because most rely on a single blood sampling and confounding factors are not always controlled (e.g., with prisoners of war).

Losing one's job is a highly stressful event because, for most employed adults, work is a central part of one's life and identity and a major source of income. Considerable evidence suggests that involuntary job loss has emotional, cognitive, behavioral, and physiological effects (22–27), including increased mental distress (28,29), and increases in physiological risk factors (such as serum uric acid (30) and blood pressure (31)), cardiovascular disease (32,33), and mortality (34,35).

To examine the impact of the chronic stress of unemployment, the study enrolled 100 individuals who had been involuntarily laid off from their job for ≥2 months but ≤19 months to avoid including a habitually unemployed group that might have more serious psychiatric problems (a criticism addressed to earlier studies). The current study uses a case-control longitudinal design with unemployed participants matched to employed controls. Unlike other studies, a large enough number of men and women participated so gender effects could be examined. Participants were followed over a 4-month time period. Monthly blood samples were obtained to provide more stable indices of immune status than that obtained from a single blood sample. The study design allowed us to examine immune system responses after termination of the chronic stress, if participants became re-employed.

The study addressed a) the impact of the chronic stress of unemployment on NK cell cytotoxicity and b) if, after chronic stressor termination, there is recovery of immune values of the formerly unemployed to levels similar to the employed participants' average levels. Based on our prior study, which investigated the relationship between self-reported persistent stress and immune parameters (18), we hypothesized that over the study period, unemployed participants would show decreased levels of NK cell cytotoxicity (NKCC) compared with employed participants. We further hypothesized that when the chronic stressor terminated, NKCC levels would "recover" to levels similar to the matched sample of employed individuals.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Participants
The sample consisted of 200 participants, 96 men and 104 women between the ages of 29 and 45 years (mean = 37.23, standard deviation (SD) = 4.62). Half the sample were unemployed and half were employed. The study was conducted in the San Francisco Bay Area during 1991 to 1994 when California was in a recession and many unemployed people were having difficulty finding work. Participants were paid ($200 for unemployed participants because they had more interviews, and $125 for employed participants). They were recruited from newspaper ads, radio public service announcements, and flyers (seeking unemployed or working participants for a study focusing on the immune system) posted on bulletin boards at nearby medical centers and community organizations. Potential participants were screened by telephone and inclusion and exclusion criteria were re-verified at the first interview.

To recruit unemployed participants who were at a similar phase in their unemployment history, we established the following inclusion criteria: age 29 to 45 years, had worked full time for at least 1 year before being involuntarily laid off from their job, had been unemployed for at least 2 months but ≤19 months. Participants were excluded if they were attending school full time, were on medical leave, or were involved in litigation (the latter exclusion to ensure the accuracy of the data and protect participants' privacy against records being subpoenaed). Because of the number of questionnaires required for the study and our concern that unskilled workers would be less able to complete these easily, we excluded participants in unskilled occupations (e.g., unskilled labor, counter service workers). We also excluded those in occupations with high employee demand (e.g., nurses, physicians, physical therapists) who were expected to become re-employed easily. For the employed subsample, the inclusion criteria were age 29 to 45 years and had worked full time for at least 1 year. Participants were excluded if they attended school full time, were involved in litigation, worked in unskilled jobs, or if they became unemployed during the 4-month study period.

Participants in the unemployed and employed subsamples were matched on gender (there were 48 men and 52 women in each subsample) and race (with one mismatch due to clerical error) and within 5-year age bands (age 29–34, 35–39, 40–45 years). They were also matched ±1 on educational level (e.g., a college graduate could be matched either with a person with some college, one with a BA, or one with an advanced degree). Participants were excluded for any health condition (e.g., pregnancy, diabetes), medication (e.g., corticosteroids), or medical treatment (e.g., recent surgery) that could influence the immune system (18). Participants were also excluded for cigarette smoking, intravenous drug use, or antidepressive medication. Exclusions were required to be maintained during the 4-month study period. We also excluded participants whose native language was not English.

Procedures
The Institutional Review Board at the University of California, San Francisco approved the study protocol, and all participants gave written informed consent. Individuals participated in the study for 4 months. At the beginning of the study, all participants filled out questionnaires and were interviewed to obtain a health history and employment history. Monthly, participants were interviewed and filled out questionnaires to assess work status, and for the unemployed group, to assess job-seeking efforts. At the end of each week, all participants completed and mailed in a weekly health form and other questionnaires.

The participants also had blood samples drawn monthly (beginning with week 5). All blood samples were drawn in the morning between 7:00 and 9:30 AM to control for circadian variation in the immune system. Participants were instructed to refrain that morning from specific activities that could affect the immune system, such as exercise (even rushing to the laboratory) or drinking caffeinated beverages or alcohol. They were also asked to avoid medications except those essential to their health. These restrictions were monitored by questions on the weekly health form; blood samples were rescheduled when participants did not follow the restrictions.

Measures
Weekly Health Form
This form contains questions about alcohol consumption, lack of sleep, cigarette use, medications taken, illnesses and symptoms of illness, exercise, other health variables, and questions about the blood draw restrictions. The weekly health form was used to pinpoint weeks when an infection (e.g., a cold) had occurred, to check for the maintenance of exclusion criteria and blood draw restrictions, and to evaluate sleep adequacy and alcohol consumption. Health behaviors were measured by single questions about participants' weekly alcohol consumption, days of exercise (aerobic and nonaerobic), and nights with inadequate sleep. Infections were considered to be present if on the illness checklist participants reported colds, upper respiratory infections, fever, stomach flu, or other systemic infections.

Job-Related Measures
At the beginning of the study, participants reported on their employment history and provided details about their former/current job. Current job status was assessed monthly through questionnaires. In answer to the question, "How financially strained do you feel?," unemployed participants also rated (on a 7-point scale, with 1 = not at all and 7 = extremely) the amount of financial strain they were experiencing. They also provided information about other sources of income (e.g., unemployment benefits, savings).

Immunological Measurement
We chose to evaluate a functional aspect of natural resistance, NKCC. NK cells are cytotoxic cells thought to be important in defense against viral infections and may play a role in protection against some tumors (10,36,37). The NKCC analysis used the method of Lew, Tsang, Solomon, Selikoff, and Bekesi (38), as described previously (18). The assay determines the capacity of cells to kill tumor targets known to be sensitive to killing by NK cells. Assays were performed at three effector:target ratios (100:1, 50:1, and 25:1).

We also carried out a phenotype determination of the NK cell lymphocyte subset. Simultaneous dual immunofluorescence and flow cytometry using a whole blood lysis method (ImmunoPrep, Coulter Corporation, Hialeah, FL) and processing device (Coulter Q Prep) were used to study the percentages of NK cells (CD56+CD16+/CD3–) in the blood using monoclonal antibodies Leu11/Leu19/T3 (Leu 11 and T3 obtained from Coulter Corporation; Leu 19 from Becton Dickinson, Inc., Milpitas, CA). Antibodies were conjugated to either FITC or phycoerythrin. The staining and lysing procedure are described below.

A 5-ml blood sample was collected from informed study subjects in ethylenediaminetetraacetic acid (EDTA) anticoagulant from a venipuncture and processed within 24 hours of phlebotomy and stored at room temperature until processed. Whole blood samples, 100 µl, were delivered to a series of 12 x 75-mm test tubes. Monoclonal antibodies, 10 µl, especially prepared for this system appropriate to the particular CD values of the NK cell subset, were added and allowed to incubate for 10 minutes at room temperature after vortexing. The tubes were then introduced into the Q-Prep instrument for a 35-second lysing cycle. The lysed blood specimens were then analyzed (EPICS-XL, Coulter). Lymphocytes were selected by using forward and side scatter gating and 5000 events were collected in the lymphocyte gate. Percentages of dually and monostaining cells were determined directly from the flow cytometer output.

Statistical Analysis
Before conducting our main analyses, we did correlation analyses to assess possible relationships between alcohol use, sleep, exercise, length of unemployment, and immune parameters. We also used an extension of linear regression (BMDP 5V) to examine the relationship between infection status and immune parameters (39).

To examine the immune differences between the unemployed group and controls, for each effector:target ratio we obtained mean values over the 4 months for each participant. We then performed multivariate analyses of variance (MANOVAs) to examine if there were differences in immune parameters by unemployment status and gender. We also performed t tests to determine if there were differences between employed and unemployed groups on the number of infectious illnesses reported during the study.

To determine if there were differences over time in NKCC for unemployed and employed groups, we carried out a repeated-measures analysis of variance (ANOVA) to assess if there were interactions between employment status and time. We assumed a first-order autoregressive (AR1) model for the immune system values; that is, we assumed that immune values which were immediately adjacent in time were correlated more strongly within persons than measurements further apart in time. Schwartz and Stone (40) described the AR approach we used in this analysis and in our prior work (18). The parameters of the regression models were estimated using the methods of maximum likelihood and the BMDP 5V program (39).

We used paired comparison t tests in our subsample of newly employed participants to determine if there were significant differences in immune values during the unemployed and newly employed time periods. Except for the BMDP repeated-measures analyses, all other statistical analyses were carried out with SAS (SAS Institute, Inc., Cary, NC). A probability of .05 was the criterion for statistical significance.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 NOTES
 REFERENCES
 
The sample was 87.5% Caucasian, 16% Asian, and 4.5% African American. The participants averaged 16.39 years of schooling (SD = 2.42). Approximately 71% of the participants were in managerial, professional, or skilled white-collar occupations; the rest held clerical, sales, service, entrepreneurial, or skilled blue-collar positions. Besides the 200 subjects, 41 failed to complete the study.1

To determine if the 41 dropouts differed from the participants, we compared baseline measures (age, education, gender; and, for the unemployed, the number of days after being laid off) between the two groups. Dropouts and participants were similar in age and years of education. A greater percentage of women were in the dropout group (59%) compared with the participants (52% female). Among the unemployed, dropouts averaged fewer days of unemployment before starting the study (mean = 218.67, SD = 95.53) compared with participants (mean = 243.48, SD = 123.11).

To characterize the sample as a whole, the mean values, SD, and ranges for the demographic, health behavior, and job status variables are presented in Table 1. The data are shown separately for the unemployed and employed subgroups and for the sample as a whole. Two-sample t tests showed that there were no significant differences between the unemployed and employed subgroups on any of the health behavior variables (p > .05).


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

  		TABLE 1. Mean Values, Standard Deviations, and Ranges for Demographic, Health, and Job Status Variables

 

Ninety-two percent of the unemployed sample reported income received from other sources. Seventy-nine percent received unemployment benefits. Additional sources of income included income from spouse (18%), others such as family and friends (15%), stock or rentals (16%), part-time work (32%), selling assets or borrowing money (2%), and savings (8%). The unemployed subsample demonstrated a high level of financial strain, averaging 5.24 (SD = 1.52, n = 97) on a 7-point scale.

To determine if unemployed subjects were more chronically stressed than employed subjects, we examined the data from another part of our study, to be reported in a subsequent paper. In that component of the study, we looked at participants' self-identified weekly stressors, using a weekly stress log similar to the one we used previously (7,18). Chronic stressors were defined as those with a duration of ≥4 weeks. Appraised chronic stressor ratings were summed for all self-identified chronic stressors listed on each log. We compared unemployed and employed groups on their chronic appraised stress ratings at study entry and on their mean appraised chronic stress levels over the 4-month study period. Compared with the employed subjects, unemployed subjects had significantly higher levels of appraised chronic stress at study entry (p = .001) and over the study period (p = .01).

Our preliminary analyses showed that there were no significant correlations between alcohol use, lack of sleep, exercise, and immune parameters (p > .05). The number of days that participants had been unemployed before entering the study was not correlated with the average level of NKCC. We also examined if infection status influenced immune parameters. If an infection occurred during the week of or the week before a blood draw, we coded that week as an infection week. Using an extension of linear regression (18), we then related infection status to immune parameters assayed in the concurrent or subsequent week. Infection status was not significantly associated with NKCC (p > .05).

Unemployment Status and NKCC
Of our sample of 100 unemployed participants, 25 obtained full-time employment before the end of our 4-month study period. For the primary analysis, we dropped these 25 participants (and their matched pairs) and carried out MANOVAs to examine if there were differences in NKCC by unemployment status and gender. We found significant differences in NKCC between unemployed and employed groups. Those who were unemployed had lower levels of NKCC for all effector:target ratios (100:1, F(1, 146) = 12.98, p = .0004; 50:1, F(1, 146) = 10.36, p = .006; and 25:1, F(1, 146) = 5.93, p = .016) (Figure 1). There were no significant main effects for gender (p = .51, p = .43, and p = .245 for NK100:1, NK50:1, and NK25:1 respectively) and no significant interactions between unemployment status and gender (p = .13, p = .08, and p = .18 for NK100:1, NK50:1, and NK25:1 respectively), suggesting that men and women did not differ in the relationship between unemployment status and NKCC.


Figure 12
View larger version (15K):
[in this window]
[in a new window]

  		Figure 1. Mean natural killer cell cytotoxicity at 3 effector:target ratios for unemployed and employed participants averaged over 4-month study period; n = 150. Error bars represent standard errors of the mean. *p < .05 for unemployed versus employed sample; **p < .01 for unemployed versus employed sample; ***p < .001 for unemployed versus employed sample.

 

Changes in Immune Values Over Time
We also investigated whether NKCC changed over the 4-month study period for the 75 unemployed compared with the 75 employed participants. Using repeated-measures analysis, we found a significant effect for unemployment status for all NKCC effector:target ratios, similar to the findings of the MANOVA. We also found a significant unemployment status X time interaction for the NK100:1 effector:target ratio (Wald test {chi}2 = 9.56, p = .03). There were similar trends for the other two effector:target ratios (for NK50:1, Wald test {chi}2 = 6.96, p = .07; for NK25:1, Wald test {chi}2 = 6.56, p = .09). Figure 2 shows that for the first 3 months of the study, NK100:1 values were lower for unemployed participants. However, after the first 2 months of the study there was a gradual increase in NK100:1 levels for those who were unemployed. We analyzed the data statistically using orthogonal polynomial trend analysis with the BMDP 2V program (39) so that we could partition the dependent variable into orthogonal linear, quadratic (parabolic), or cubic (S-formed) components. For the unemployed subsample, results demonstrated a significant linear trend component for all effector:target ratios (for 100:1, F(1, 74) = 7.35, p = .008; for 50:1, F(1, 74) = 10.10, p = .002; for 25:1, F(1, 74) = 6.30, p = .014). In contrast, in the employed sample, there was no significant linear component. There was a significant quadratic trend component for NK100:1 (F(1, 74) = 4.80, p = .03) but not for NK50:1 (p = .10) or NK25:1 (p = .47).


Figure 22
View larger version (9K):
[in this window]
[in a new window]

  		Figure 2. Natural killer cell cytotoxicity (100:1 effector:target ratio) at 4 monthly time points for employed and unemployed groups; n = 150. Error bars represent standard errors of the mean.

 

We found no relationship between length of unemployment at study onset and mean NKCC levels over the study period. However, it is possible that those unemployed for longer or shorter periods might vary physiologically (41) and could show different NKCC levels over time. To address this question, we carried out three exploratory repeated-measures analyses of NKCC to examine the interaction of time and the length of unemployment (n = 74). One analysis examined the length of unemployment (at study onset) as a continuous variable; in the other two analyses, we divided the length of unemployment into categories. We compared those unemployed ≤1 year (n = 63) with those unemployed >1 year (n = 11). In addition, we divided the length of unemployment into tertiles (<175 days, 175–270 days, >270 days unemployed at study onset). All three repeated-measures analyses of NKCC showed that there were no significant interactions between time and length of unemployment (p > .05). Thus, the relationship between unemployment status and NKCC over the 4-month study period did not depend on how long the study participants had been unemployed when they entered the study.

Infection Findings
Whereas other studies had reported that unemployment status was associated with increased illness reports, we investigated if it was associated with increased reports of infectious illnesses. A t test analysis showed there were no significant differences between employed and unemployed subsamples in the number of infectious illnesses reported during the study (p = .17).

Changes in NKCC After the Chronic Stressor Ended
What happens to unemployed participants after the chronic stressful situation of unemployment ends, and the participant begins to work full time? We averaged the blood sample results taken before the time these participants obtained full-time employment and those taken after, and we compared the immune values for the before and after periods. One participant obtained employment before the first blood sample, leaving a subsample of 24 for this analysis. Paired comparison t tests demonstrated significant differences for two of the three effector:target ratios (50:1 (t = 3.55, p = .002) and 25:1 (t = 4.29, p = .0003), with a trend for 100:1 (t = 1.96, p = .06)). NKCC was higher after the chronic stressor ceased compared with the prior period when participants were unemployed (mean differences = 9.38, 10.59, and 5.80, respectively, for 50:1, 25:1, and 100:1 effector:target ratios). When the same analysis was conducted for the 24 always employed matched pairs using blood samples matched with the ones used for their matched pair, the results showed no significant differences between NKCC values for these two time periods for the always employed (p = .55–.97). Further, t test analyses showed that the recovery ("after") NKCC values for the re-employed are not significantly different from the "after" values of the employed participants (p = .29, .60, and .18). Figure 3 illustrates the 100:1, 50:1, and 25:1 mean values for the unemployed participants before and after they became employed as well as the comparable mean values for the employed matched pairs.


Figure 32
View larger version (23K):
[in this window]
[in a new window]

  		Figure 3. Mean natural killer cell cytotoxicity at 3 effector:target ratios for unemployed and employed time periods (for the unemployed subsample who became employed) and for the matched employed subsample; n = 24 for each subsample. For matched pairs, the same time periods were used as those in the unemployed subsample. Error bars represent standard errors of the mean. *p < .05 for unemployed time periods versus employed time periods for the unemployed who got new jobs; **p < .01 for unemployed time periods versus employed time periods for the unemployed who got new jobs.

 

We calculated a measure of how much immune values "recovered" by dividing the value for the increase in NKCC after re-employment by the value that represented the amount of the initial gap between the unemployed and the steadily employed participants' immune levels. For NK50:1, there was a 63% recovery (9.38/13.2 = 63%). The recovery was 72% for NK25:1 and 44% for NK100:1.

We also carried out an exploratory analysis to determine if the length of unemployment was related to the magnitude of recovery of immune function after re-employment. Results showed correlations ranging from –0.22 to –0.33 for the three NKCC effector:target ratios; none was significant (p = .13–.33).

We repeated the paired comparison t tests using only data from two time points for each participant, namely, the last "unemployed" time point and the first "employed" time point (Figure 4) to determine if these changes occurred within the first month after stressor termination. The results were similar in direction and magnitude but showed lower mean differences (mean differences = 8.86, 9.73, and 6.18 respectively, for 50:1, 25:1, and 100:1 effector:target ratios; p = .026, .014, and .11), reflecting the lower stability that comes with the comparison of single points rather than averages. The measure of immune recovery for the newly re-employed group was 53% for NK50:1, 61% for NK25:1, and 36% for NK100:1. There were no significant differences for the two time periods for the always employed group (p = .79–.96). Further, t test analyses showed that the recovery ("after") NKCC values for the re-employed were not significantly different from the "after" values of the employed participants (p = .19, .41, and .11).


Figure 42
View larger version (22K):
[in this window]
[in a new window]

  		Figure 4. Natural killer cell cytotoxicity at 3 effector:target ratios for last unemployed time point and first employed time point (for the unemployed subsample who became employed) and for the matched employed subsample; n = 24 for each subsample. For matched pairs, the same time points were used as those in the unemployed subsample. Error bars represent standard errors of the mean. *p < .05 for the last unemployed time point versus the first time point after becoming employed for the unemployed who got new jobs.

 

Because the repeated-measures analysis in the always unemployed group (n = 75) showed a linear increase in NKCC after the first 2 months of the study (Figure 2), we examined the before and after re-employment values separately for each pair of months in the study. Using the single-point data for NK100:1, we separated the participants into groups depending on when they became re-employed. We compared the before re-employment values for month 1 and after re-employment values for month 2 (n = 7), and the before and after re-employment values for month 2 cf month 3 (n = 11) and month 3 cf month 4 (n = 6). Figure 5 shows that no matter which 2 months of the study are compared, after re-employment, there was always an increase in NK100:1 levels (6.47 increase for the month 1/month 2 comparison, 4.98 for the month 2/month 3 comparison, and 8.04 for the month 3/month 4 comparison). Thus, for the subsample of 24 who found new jobs, an increase in NKCC with re-employment occurred no matter when during the study period they became re-employed.


Figure 52
View larger version (16K):
[in this window]
[in a new window]

  		Figure 5. Natural killer cell cytotoxicity (100:1 effector:target ratio) for last unemployed time point and first employed time point shown separately for participants re-employed at months 2, 3, and 4; n = 24. Error bars represent standard errors of the mean.

 

Percentages of NK Cells in the Blood
To determine if the decrease in NKCC associated with unemployment was due to a decrease in the percentage of NK cells in the blood, we repeated the analyses using NK cell percentages as the outcome. We found no significant association in the MANOVA analysis (p = .98) or in the paired t test analyses of our subsample (p = .20 for the averaged results, p = .75 for the single-point comparisons).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
NOTES
 REFERENCES
 
This study examined the effects of a chronic stressor on NKCC by comparing 100 unemployed participants with a matched group of 100 employed. In addition, the impact of stressor termination was evaluated. Our data are derived from a sample of 200 healthy men and women aged 29 to 45 years, evaluated monthly over 4 months. We found that the unemployed group had significantly lower NK cell function compared with the matched employed group. In addition, we demonstrated immune recovery after the end of the unemployment period. Recovery of immune function began very quickly (within 1 month), with substantial "recovery" (44% to 72%) over a period of 1 to 3 months, compared with the levels of the steadily employed. We believe this is the first study to evaluate individuals at multiple time points during a chronically stressful period and follow a subsample of them to a time when the chronic stressor ends (but is not complicated by another significant stressor, such as bereavement). Our findings demonstrate persistent immune alteration during a chronic stressor. The findings also suggest that certain physiological systems may be resilient in the face of chronic stressors that last ≤2 years and can quickly recover most of their functioning after the stressor ends.

The NK cells of the unemployed participants showed a lower average ability to kill tumor targets compared with the employed sample over the 4-month study period. Thus, our first hypothesis was confirmed. The relationship between unemployment and NK function was not due to smoking (because smokers were excluded from the study) and there were no associations between NKCC and alcohol use, lack of sleep, or exercise. The relationship was not due to differences in the percentage of NK cells in the blood because there were no differences in NK cell percentages as a function of unemployment. Thus, the lower NKCC associated with unemployment reflects differences in the cytotoxic capacity of the cell. These data may suggest one pathway through which chronic stressors can affect physical disease.

Our study found that the length of unemployment was not significantly related to mean levels of NKCC. Repeated-measures analyses of NKCC examining the length of unemployment (as a continuous and as a categorical variable) did not show any significant interaction of time and days of unemployment. Thus, these data suggest that for the time periods examined in this study, there may not be set "phases" of unemployment, with some phases more stressful than others, as Polakoff (41) and others have suggested. Because our study did not follow a single cohort of unemployed participants from layoff forward over time, there was considerable variability in the length of unemployment for participants (2–19 months at study onset). Only 11 participants were unemployed for >1 year and our design excluded anyone unemployed for an extended period of time. Thus, our data may not offer a good test of this question.

Our findings are consistent with other studies that found decreased NKCC in response to chronic stress (13). Our previous study (18), in a population of healthy women aged 21 to 45 years, found lower NKCC for those experiencing high levels of subjectively defined persistent stress (for stressors persisting 2–12 weeks). Lowered NK cell activity was also found for prisoners of war (16), elderly residents anticipating housing relocation (42), and for Hurricane Andrew victims 2 to 4 months after the hurricane hit (17). However, studies examining elderly caregivers of patients with Alzheimer's disease (19–21) and caregivers of very low birthweight infants (43) did not find decreased NKCC in the caregivers compared with controls. Thus, results may depend on the nature of the chronic stressor studied (caregiving versus other chronic stressors), its duration, or the age of participants.

The repeated-measures analyses of immune status over the 4 months in the study showed a different immune pattern for employed and unemployed groups. This difference was significant for NK100:1. Figure 2 shows that the unemployed sample had a gradual increase in NK100:1 levels after the first 2 months of the study; the employed sample did not. The time variable in the analysis represents the effect of the months that participants were in the study, not how long they were unemployed. The data were collected over a 3-year time period, which suggests that the effects were not due to changes in the laboratory or to external community stressors. We could speculate that the unemployed sample's increase in NKCC for months 3 and 4 might be a consequence of having completed a weekly stress log for 16 weeks. Perhaps the process of doing so offered unemployed participants a way of gaining a new perspective on their unemployment experiences and validated that their stressors were being taken seriously in a research study. Diary writing studies have shown that brief episodes of diary writing about important life events can have psychological and physiological effects (44). Alternatively, contacts with research staff may have had more of a supportive effect on the unemployed than the employed sample.

Those who became re-employed had significantly higher NKCC values after the chronic stressor ceased compared with the period when those participants were unemployed. This effect was found when the single blood samples before and after re-employment were compared and also when all blood samples before and after re-employment were separately averaged and then compared. Figure 5 shows that the changes in immune status after re-employment did not appear to be simply a function of the linear increase in NKCC that was found in the sample of 75 consistently unemployed participants (Figure 2). The increase in NKCC with re-employment occurred at all assessment periods for the subsample of 24, whereas for those unemployed throughout the entire study period, there was a pronounced increase in NKCC only from the third to the fourth assessment point.

The only other human studies that examined stressor termination and immunity were two studies on caregivers of Alzheimer's patients that evaluated different immune parameters. In those studies, the termination of caregiving occurred when the Alzheimer's patient died; immune decrements persisted for up to 3 years after caregiving ended (19,45). However, it may be inappropriate to compare our results with these two studies because they used an elderly population and the end of caregiving coincided with the death of the patient and a prolonged bereavement for the caregiver (46). In contrast, the end of unemployment constitutes an end to a stressor without the occurrence of a related new chronic stressor. Although new short-term stressors could occur when starting a new job, the chronic stress of unemployment has ended. Our results suggest that NK function can recover after chronic stress if another major chronic stressor is not immediately initiated.

The number of infectious illnesses reported during the study did not differ for the employed and unemployed subsamples. Previous studies of unemployment had found increased self-reports of physical illnesses and increased physician visits (24,47,48); however, others had not (49,50).

One possible mediator of the effects of unemployment on NK cell function is sympathetic nervous system activity. Norepinephrine (NE) is increased in response to stress and has been shown to reduce NKCC in vitro (51). In humans, the effects of acute stressors on NKCC appear to be mediated by increased levels of NE and/or epinephrine (52). In chronically stressed individuals, exposure to acute stressors is associated with both increased epinephrine and decreased NKCC; this pattern is not observed in those not experiencing a chronic stressor (53). In animals, the effects of physical stress on NKCC have been shown to be mediated by sympathetic nervous system activity (54).

In this study, we recruited a sample that was restricted in various ways to control for important sources of confounding. We limited our age range to ages 29 to 45 to control for age-related differences in immune responses and eliminate those at the beginning of their career pathway. We restricted our unemployed sample to include only those with previous steady employment and focus on a group unemployed for ≤19 months, avoiding samples with mental health problems that could lead to chronic unemployment. We excluded participants with illnesses, medications, treatments, or behaviors (e.g., smoking) that might alter their immunological parameters. A limitation of the study is that results cannot be generalized to younger or older samples, those who never held a steady job, or those unemployed for longer periods. Other limitations include using only single-item measures of alcohol consumption, exercise, and sleep, and not collecting physiological data to assess menstrual phase for women. Further, we were not able to explore personal qualities differentiating our re-employed sample of 24 and the seven who dropped out of the study after getting a new job, qualities that might have contributed to their NKCC recovery.

The findings here add to our theoretical knowledge of how chronic stresses affect immune parameters and highlight the ameliorating effect of stressor termination. Future research is needed to determine the factors that moderate the long-term physiological impact of a chronic stressor, such as the characteristics of the stressor, the timeframe, the age of the individual, as well as other psychological, genetic, and contextual factors. For example, it is important to determine the effects of longer periods of unemployment than the 2 to 19 months studied in this sample, to discover if the immune effects return to baseline over time or become more pronounced if the duration of the stressor is substantially extended. Such studies could also determine the time to recovery for chronic stressors of different lengths. Longer follow-ups could also determine if recovery can become complete over an extended period of time. Further, it is also critical to determine the physiological mechanisms that control immune recovery from chronic stress.

We thank our project coordinator, Wendy K. Berland, research assistants, and secretarial staff who made important contributions to the project and Marilyn Vella for her assistance in preparing the manuscript.


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
REFERENCES
 
1Among the 25 in the unemployed subsample who failed to complete the study, 15 withdrew from the study before completion. Five of the 15 indicated that they were not interested in continuing and seven got a job and found it too difficult to continue with the study. Eight additional participants were dropped for failure to maintain exclusions; two were dropped from the study for failure to complete all study components. Among the 16 in the employed subsample who failed to complete the study, seven withdrew from the study before completion, stating that they were too busy to continue. Seven additional participants were dropped for failure to maintain exclusions. Of the two additional participants dropped from the study, one was dropped because the participant became unemployed. Back

Received for publication June 13, 2006; revision received September 19, 2006.

This study was supported by Grant MH46788 (F.C.) from the National Institute of Mental Health.

DOI:10.1097/PSY.0b013e31803139a6


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 

  1. Capitanio JP, Lerche NW. Social separation, housing relocation, and survival in simian AIDS: a retrospective analysis. Psychosom Med 1998;60:235–44.[Abstract/Free Full Text]
  2. Edwards EA, Dean LM. Effects of crowding of mice on humoral antibody formation and protection to lethal antigenic challenge. Psychosom Med 1977;39:19–24.[Abstract/Free Full Text]
  3. Cohen S, Line S, Manuck SB, Rabin BS, Heise ER, Kaplan JR. Chronic social stress, social status, and susceptibility to upper respiratory infections in nonhuman primates. Psychosom Med 1997;59:213–21.[Abstract/Free Full Text]
  4. Padgett DA, Sheridan JF, Dorne J, Berntson GG, Candelora J, Glaser R. Social stress and the reactivation of latent herpes simplex virus type 1. Proc Natl Acad Sci U S A 1998;95:7231–5.[Abstract/Free Full Text]
  5. Black PH, Garbutt LD. Stress, inflammation and cardiovascular disease. J Psychosom Res 2002;52:1–23.[CrossRef][Medline]
  6. Cohen S, Frank E, Doyle WJ, Skoner DP, Rabin BS, Gwaltney JM Jr. Types of stressors that increase susceptibility to the common cold in healthy adults. Health Psychol 1998;17:214–23.[CrossRef][Medline]
  7. Cohen F, Kemeny ME, Kearney KA, Zegans LS, Neuhaus JM, Conant MA. Persistent stress as a predictor of genital herpes recurrence. Arch Intern Med 1999;159:2430–6.[Abstract/Free Full Text]
  8. McEwen BS, Seeman T. Protective and damaging effects of mediators of stress: elaborating and testing the concepts of allostasis and allostatic load. Ann NY Acad Sci 1999;896:30–47.[CrossRef][Medline]
  9. Coe CL. Concepts and models of immunological change during prolonged stress. In: Schedlowski M, Tewes U, editors. Psychoneuroimmunology: an interdisciplinary introduction. New York: Kluwer Academic/Plenum Publishers; 1999.
  10. Whiteside TL, Herberman RB. Role of human natural killer cells in health and disease. Clin Diagn Lab Immunol 1994;1:125–33.[Medline]
  11. Bartolomucci A. Resource loss and stress-related disease: is there a link? Med Sci Monit 2005;11:RA147–54.[Medline]
  12. Stefanski V, Engler H. Social stress, dominance and blood cellular immunity. J Neuroimmunol 1999;94:144–52.[CrossRef][Medline]
  13. Segerstrom SC, Miller GE. Psychological stress and the human immune system: a meta-analytic study of 30 years of inquiry. Psychol Bull 2004;130:601–30.[CrossRef][Medline]
  14. Ben-Eliyahu S, Shakhar G, Page GG, Stefanski V, Shakhar K. Suppression of NK cell activity and of resistance to metastasis by stress: a role for adrenal catecholamines and beta-adrenoceptors. Neuroimmunomodulation 2000;8:154–64.[CrossRef][Medline]
  15. Irwin M, Daniels M, Weiner H. Immune and neuroendocrine changes during bereavement. Psychiatr Clin North Am 1987;10:449–65.[Medline]
  16. Dekaris D, Sabioncello A, Mazuran R, Rabatic S, Svoboda-Beusan I, Racunica NL, Tomasic J. Multiple changes of immunologic parameters in prisoners of war. Assessments after release from a camp in Manjaca, Bosnia. JAMA 1993;270:595–99.[Abstract/Free Full Text]
  17. Ironson G, Wynings C, Schneiderman N, Baum A, Rodriguez M, Greenwood D, Benight C, Antoni M, LaPerriere A, Huang HS, Klimas N, Fletcher, MA. Posttraumatic stress symptoms, intrusive thoughts, loss, and immune function after Hurricane Andrew. Psychom Med 1997;59:128–41.
  18. Cohen F, Kearney KA, Zegans LS, Kemeny ME, Neuhaus JM, Stites DP. Differential immune system changes with acute and persistent stress for optimists vs pessimists. Brain Behav Immun 1999;13:155–74.[CrossRef][Medline]
  19. Esterling BA, Kiecolt-Glaser JK, Bodnar JC, Glaser R. Chronic stress, social support, and persistent alterations in the natural killer cell response to cytokines in older adults. Health Psychol 1994;13:291–8.[CrossRef][Medline]
  20. Irwin M, Brown M, Patterson T, Hauger R, Mascovich A, Grant I. Neuropeptide Y and natural killer cell activity: findings in depression and Alzheimer caregiver stress. FASEB J 1991;5:3100–7.[Abstract]
  21. Irwin M, Hauger R, Patterson TL, Semple S, Ziegler M, Grant I. Alzheimer caregiver stress: basal natural killer cell activity, pituitary-adrenal cortical function, and sympathetic tone. Ann Behav Med 1997;19:83–90.[Medline]
  22. Bartley M. Unemployment and ill health: understanding the relationship. J Epidemiol Community Health 1994;48:333–7.[Abstract/Free Full Text]
  23. Beale N, Nethercott S. The health of industrial employees four years after compulsory redundancy. J R Coll Gen Pract 1987;37:390–4.[Medline]
  24. Cook DG, Cummins RO, Bartley MJ, Shaper AG. Health of unemployed middle-aged men in Great Britain. Lancet 1982;1:1290–4.[Medline]
  25. D'Arcy C, Siddique CM. Unemployment and health: an analysis of "Canada health survey" data. J Health Serv 1985;15:609–35.
  26. Dirksen ME. Unemployment: more than the loss of a job. AAOHN J 1994;42:468–76.[Medline]
  27. Leeflang RLI, Klein-Hesselink, DJ Spruit, IP. Health effects of unemployment-II. Men and women. Soc Sci Med 1992;34:351–63.[CrossRef][Medline]
  28. Brown DW, Balluz LS, Ford ES, Giles WH, Strine TW, Moriarty DG, Croft JB, Mokdad AH. Associations between short- and long-term unemployment and frequent mental distress among a national sample of men and women. J Occup Environ Med 2003;45:1159–66.[CrossRef][Medline]
  29. Keefe V, Reid P, Ormsby C, Robson B, Purdie G, Baxter J, Ngati Kahungunu Iwi Incorporated. Serious health events following involuntary job loss in New Zealand meat processing workers. Int J Epidemiol 2002;31:1155–61.[Abstract/Free Full Text]
  30. Kasl SV, Cobb S, Brooks GW. Changes in serum uric acid and cholesterol levels in men undergoing job loss. JAMA 1968;11:1500–07.
  31. Kasl SV, Cobb S. Blood pressure changes in men undergoing job loss: a preliminary report. Psychosom Med 1970;32:19–38.[Abstract/Free Full Text]
  32. Brenner MH, Mooney A. Unemployment and health in the context of economic change. Soc Sci Med 1983;17:1125–38.[CrossRef][Medline]
  33. Gallo WT, Bradley EH, Falba TA, Dubin JA, Cramer LD, Bogardus ST Jr, Kasl SV. Involuntary job loss as a risk factor for subsequent myocardial infarction and stroke: findings from the health and retirement survey. Am J Ind Med 2004;45:408–16.[CrossRef][Medline]
  34. Brenner MH. Economic changes and heart disease mortality. Am J Public Health 1971;61:606–11.[Free Full Text]
  35. Gerdtham UG, Johannesson M. A note on the effect of unemployment on mortality. J Health Econ 2003;22:505–18.[CrossRef][Medline]
  36. Herberman N. NK cells and other natural effector cells. New York: Academic Press; 1982.
  37. Pollock RE, Lotzova E, Stanford SD. Mechanism of surgical stress impairment of human perioperative natural killer cell cytotoxicity. Arch Surg 1991;126:338–42.[Abstract/Free Full Text]
  38. Lew F, Tsang P, Solomon S, Selikoff IJ, Bekesi JG. Natural killer cell function and modulation by {alpha}IFN and IL2 in AIDS patients and prodromal subjects. J Clin Lab Immunol 1984;14:115–21.[Medline]
  39. Dixon WJ, Brown MB, Engleman L, Hill MA, Jennrich RI. BMDP statistical software. Berkeley: University of California Press; 1988.
  40. Schwartz JE, Stone AA. Strategies for analyzing ecological momentary assessment data. Health Psychol 1998;17:6–16.[CrossRef][Medline]
  41. Polakoff PL. Unemployment affects workers' health; stress targets emotional, physical being. Occup Health Saf 1987;56:34.[Medline]
  42. Lutgendorf SK, Vitaliano PP, Tripp-Reimer T, Harvey JH, Lubaroff DM. Sense of coherence moderates the relationship between life stress and natural killer cell activity in healthy older adults. Psych Aging 1999;14:552–63.[CrossRef]
  43. Gennaro S, Fehder W, Nuamah IF, Campbell DE, Douglas SD. Caregiving to very low birthweight infants: a model of stress and immune response. Brain Behav Immun 1997;11:201–15.[CrossRef][Medline]
  44. Frattaroli J. Experimental disclosure and its moderators: a meta-analysis. Psychol Bull 2006;132:823–65.[CrossRef][Medline]
  45. Esterling BA, Kiecolt-Glaser JK, Glaser R. Psychosocial modulation of cytokine-induced natural killer cell activity in older adults. Psychosom Med 1996;58:264–72.[Abstract/Free Full Text]
  46. Bodnar JC, Kiecolt-Glaser JK. Caregiver depression after bereavement: chronic stress isn't over when it's over. Psych Aging 1994;9:372–80.[CrossRef]
  47. Bartley M, Sacker A, Clarke P. Employment status, employment conditions, and limiting illness: prospective evidence from the British household panel survey 1991–2001. J Epidemiol Community Health 2004;58:501–6.[Abstract/Free Full Text]
  48. Grayson JP. Reported illness after a CGE closure. Can J Public Health 1989;80:16–9.[Medline]
  49. Matoba T, Ishitake T, Noguchi R. A 2-year follow-up survey of health and life style in Japanese unemployed persons. Int Arch Occup Environ Health 2003;76:302–8.[Medline]
  50. Waldron I, Herold J, Dunn D, Staum R. Reciprocal effects of health and labor force participation among women: evidence from two longitudinal studies. J Occup Med 1982;24:126–32.[Medline]
  51. Gan X, Zhang L, Solomon GF, Bonavida B. Mechanism of norepinephrine-mediated inhibition of human NK cytotoxic functions: inhibition of cytokine secretion, target binding, and programming for cytotoxicity. Brain Behav Immun 2002;16:227–46.[CrossRef][Medline]
  52. Hurwitz BE, Brownley KA, Motivala SJ, Milanovich JR, Kibler JL, Fillion L, LeBlanc WG, Kumar M, Klimas NG, Fletcher MA, Schneiderman N. Sympathoimmune anomalies underlying the response to stressful challenge in human immunodeficiency virus spectrum disease. Psychosom Med 2005;67:798–806.[Abstract/Free Full Text]
  53. Pike JL, Smith TL, Hauger RL, Nicassio PM, Patterson TL, McClintick J, Costlow C, Irwin, MR. Chronic life stress alters sympathetic, neuroendocrine, and immune responsivity to an acute psychological stressor in humans. Psychosom Med 1997;59:445–57.
  54. Jiang XH, Guo SY, Xu S, Yin QZ, Ohshita Y, Naitoh M, Horibe Y, Hisamitsu T. Sympathetic nervous system mediates cold stress-induced suppression of natural killer cytotoxicity in rats. Neurosci Lett 2004;357:1–4.[CrossRef][Medline]




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 Google Scholar
Google Scholar
Right arrow Articles by Cohen, F.
Right arrow Articles by Stites, D. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Cohen, F.
Right arrow Articles by Stites, D. P.
Related Collections
Right arrow Immunology
Right arrow Stress and Coping

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS