DESCRIPTION OF STUDY
From 1972 to 1978 a large, multidisciplinary team consisting of over 50 individuals collaborated in a prospective longitudinal study of health change in air traffic controllers (ATC).1'2 The principal investigators, Robert Rose, David Jenkins and Michael Hurst, conducted the study at Boston University School of Medicine, using a representative sample of 416 air traffic controllers who worked in New York and New England regions. Controllers were drawn from larger and smaller facilities including airports and the large air route traffic control centers that are responsible for many hundreds of square miles of air space.
There were numerous goals of the study, including the assessment of physical and psychiatric health problems among the controllers and a search for factors which could be implicated as increasing the risk for developing new health problems in this group. Although the study systematically assessed both the prevalence and incidence of many physical problems, including the very significant frequency of hypertension among controllers, this article will be restricted to discussion of the extent of psychopathology along with those variables which were found to be predictors of psychopathology.
A stratified random sample of 416 controllers was selected from a total eligible pool of 685 men. Each man was followed for 36 months. At the time of entry, every man came in for a day-long series of examinations, including a psychiatric interview, a physical examination and review of systems, and filled out a questionnaire about work, work attitudes, job morale, etc. Individuals returned to Boston University for a total of five visits at approximately nine-month intervals. In between visits individuals filled out a monthly health diary and a Zung Mood and Anxiety Checklist.' Individuals were also studied at work, where blood pressure was measured every 20 minutes or blood drawn for endocrine analyses. At the same time, the amount of work they were doing was noted along with the rating of their behavioral arousal.
PSYCHIATRIC STATUS SCHEDULE
The assessment of psychopathology was made using the Psychiatric Status Schedule ( PSS).4 This was selected because comprehensive data were available on psychiatric patients and various control groups, along with this instrument's accepted reliability and validity/''1 The PSS has a total of 32 1 symptoms which are recorded as present or absent. The interview is conducted with the use of primary questions to elicit information about symptoms, but also provides follow-up questions in the event that the first answers are insufficient to judge the presence or absence of a particular symptom. We were able to achieve very high inter-rater reliability among the interviewers using the PSS.
FREQUENCY OF OCCURRENCE OF SIGNIFICANT SYMPTOMATOLOGY IN THOSE STUDIED ALL FIVE TIMES BY PSS
Three major summary scales of the PSS were of greatest interest to us. These were: subjective distress, which essentially measures various symptoms of depression and anxiety; impulse control disturbance, which assesses impulsive behavior (eg, illicit drug use, difficulties with the law, frequent fighting); and alcohol abuse.
Our first method of examining the data compared three groups of men. The members of the first group were without any positive scores during the course of the study (asymptomatic). The second were those who had problems on the first exam at entry and continued with problems during the course of study (prevalence cases). The third were those men who developed new problems during the course of study which were not apparent at intake (incidence cases). We also compared men with respect to severity of symptoms. We contrasted those men who had the most frequent problems (severe) with those who had only infrequent problems (mild cases) versus those who remained asymptomatic.
The use of the PSS provided a quantitative continuum for each scale. In order for a scale score to be considered positive or symptomatic, it had to exceed a cut-off score. These were usually defined as greater than one standard deviation from the urban community sample. These values were very close to those found in pre-treatment psychiatric outpatients or former psychiatric inpatients. Thus, no man was considered as having significant subjective distress, impulse control problems or alcohol abuse unless his score was close to values seen in psychiatric outpatients prior to treatment. We might be accused of underestimating the frequency of significant psychopathology as individuals could have had elevated scores which were not considered "a case" because it was not sufficiently high to meet the cut-off criterion.
The frequency of occurrence of significant symptomatology is given in Table I . It is notable that approximately 13% of the men had significant depression/ anxiety as measured by subjective distress scale at least once during the three years they were followed. Five percent of the men had significant scores on at least two of the five exams. It must be remembered that these findings are not among a patient population, but in men who were, by definition, employed in full-time jobs.
It is apparent from Table 1 that the most prevalent problems for the controllers were difficulties in impulse control. This scale has seven items which assess antisocial impulses, 20 items on drug abuse and six items on reported overt anger. The most frequent positive items on this scale among the controllers were repeated minor illegal acts, visible rages or fits of anger, heated arguments or impulses to commit illegal acts. It is notable that 15% of this population had significant scores on impulse control disturbance at least two or more times.
The alcohol abuse scale is composed of 16 symptoms. The most frequent ones exhibited by the controllers were: I) admitting that "one keeps drinking even though one wants to stop." 2) admitting to a drinking problem. 3) having three or more drinks alone, and 4) claiming one can't remember what happened the night before or admits to regular intoxication. Approximately 20% of the men had a significantly elevated score on alcohol abuse at least once and 1 1% had significant scores two or more times. The controllers who had positive scores on alcohol abuse would qualify for the diagnosis of alcohol abuse in DSM-HI (305.0). However, no controllers who were currently working would qualify for alcohol dependence (alcoholism. 303.9). Although individuals were clearly at risk of becoming alcohol dependent, it should be remembered that all controllers must have an annual exam from a physician in order to continue controlling traffic. Individuals so clearly affected with alcoholism would be disqualified and not have remained as subjects throughout our study. Throughout the study, individuals who were found to have significant psychopathology were at much greater risk for being disqualified. This will be presented later in greater detail.
One major focus was to compare the 1 17 men who developed problems during the course of the study with those who remained asymptomatic throughout the study. We also compared those who entered with problems with the other two groups. As our major interest was to search for predictors of psychopathology and not just correlates, the focus was on what distinguished those who later became symptomatic after intake from those who remained asymptomatic.
ZUNG ANXIETY AND DEPRESSION SCALES
The controllers were most cooperative in returning the Zung forms. Over 90% of the men completed most of the forms so that the level of depression and anxiety in the interval between visits could be analyzed. The checklist was scored according to the standard method.
Symptoms of depression were much more common than anxiety symptoms. The data were converted to yearly frequency of positive depression scores.
Approximately 70% of the controllers were asymptomatic throughout the year. Twenty percent of the men had more than one but less than three months per year with symptomatic responses. Seven percent of the men had three to nine months per year in which they had symptoms (intermittent group). The chronic group composed of approximately 4%i of the men had nine to 12 months in which they were symptomatic. It is from the intermittent and chronic groups that almost all of the cases with positive subjective distress scores were found, and there was a good correspondence between the Zung data and that obtained from the interview; the only exception was that individuals, as might be expected, tended to overrate their symptoms on the Zung checklist when compared with the data obtained from the PSS interview.
JOB DESCRIPTION INVENTORY: COMPARISON OF ATOS WITH OTHER WORKERS
We assessed a large number of psychosocial variables in the controllers looking for what might function to predispose individuals to psychopathology. The first set of variables were clustered in the ATC questionnaire, designed and pre-tested by our group to look specifically at differences among controllers with respect to a number of job attitudes, usually related to controlling traffic. These included attitudes about work, ways of attempting to cope with what was perceived as stressful about work, magnitude of investment in ATC work, etc. We also assessed job and work morale as measured by standard instruments, such as the Job Description Inventory (JDl)4 or the Leadership Behavior Description Questionnaire (LBDQ)"1 as well as specific questions regarding their perception of management practices. We also utilized peer ratings of technical competence or ease of working with one another. We used three scales to measure life change events: the Holmes and Rahe scale; the Paykel-Uhlenhuth scale; and the Review of Life Events (ROLE), our expanded list of 104 life events.' Mz We employed several standard psychological questionnaires such as the California Personality Inventory and Jenkins Activity Scale (measures Type A behavior). All these variables were collected at intake and prior to the time at which we collected the psychopathology data from the PSS and Zung. Consequently these can be used as potential predictor variables.
Those individuals who developed new problems in psychopathology (incidence cases) were younger and had less experience than the asymptomatic controllers. They had significantly lower work satisfaction as well as lower co-worker satisfaction. The issue of worker satisfaction was a major problem for the controllers. As a group, as shown in Table 2, they were significantly less satisfied with work itself when income or educational level was taken into account. These findings are consistent with the controller's tendency to be excessively critical not only of their supervisors and the agency itself, but also of one another.
Individuals who developed problems were further distinguished from those who remained asymptomatic in several ways. They had higher subjective costs in that they stated that work interfered with their relations with others, their peace of mind, etc. They reported using physical activity less to cope with work pressures, but a higher frequency Of using alcohol to unwind or relax after work. They also reported significantly lower marital support. At the same time, those controllers who developed significant psychopathology were highly invested in ATC work. They were more likely to state that they disliked having to restrict aircraft prior to having them enter their sector (ie, they wanted to handle as many planes as possible). They also felt that it was important for them to try and fill pilot requests even when the request would cause them extra work. However, they were less successful in their ability to unwind or to discharge tensions built up at work. Consequently, they were more likely to answer yes to the following types of questions: "After I have left work I continue thinking about all the possible conflicts (between planes) and work them through again in my mind;" "I stay in 'high gear' and have trouble relaxing once I leave work;" "In describing me, my friends would say that I eat, drink and think ATC work." Thus, these men were invested in their work, unable to remove themselves from it when they went home, felt that they had less support from their wives and were more liable to drink with their buddies after work, but at the same time they reported lower work and coworker satisfaction. They were younger and had less experience, and certainly could be described as having been less mellowed by their experience as a controller. In addition, these men also reported significantly higher levels of life change events in the six months prior to their entering the study. They experienced significantly more events and reported more distress in their attempts fo adjust to these events.
In terms of personality characteristics, the incidence cases were much more liable to state that they had a decreased sense of well-being, as measured by the California Personality Inventory, and reported a lower sense of responsibility in that they were much less likely to endorse items that one might label as "respect for authority." The low score for controllers in this responsibility scale does not indicate so much a lack of responsibility or socialization as it does a low regard for rules and regulations and a tendency to complete tasks their own way in order to get them done. This is consistent with controllers oft-stated self-image that "we make the system work." Controllers who had more psychiatric problems were even more likely to have lower scores than the somewhat lower group scores shown by the controller work force. In addition, controllers who developed significant problems were more likely to be Type A in their behavior compared with the asymptomatic group.
We wanted to know which were the most important discriminators among those who were asymptomatic, those who developed problems, and those who entered symptomatic and, therefore, subjected these individual predictors to a multivariate discriminative analysis. Table 3 shows the five variables that contributed most to discriminating between these three groups. In general those with the most abnormal scores were those who entered with significant problems, the prevalence cases. The incidence cases usually had the next level of significantly positive scores compared with controls. It is notable that the incidence and prevalence cases did describe themselves as not free from worries, but burdened and ill-tempered as reflected by lower scores on the sense of well-being scale of the CPI. They were unable to get their jobs off their minds which is an essential issue rated by the tension discharge rate. This, of course, was compounded by the fact that, as we have learned, they were highly invested. They had low work satisfaction and tended to drink more.
Very similar findings emerged when we analyzed the data from the Zung Depression scores. Those who had more frequent and more intense depression as self-rated on the Zung Depression Inventory were also the ones who had reported themselves as low on a sense of well-being on the CPl. They also reported more frequent life changes and low co-worker satisfaction. Thus when one looks at the potential predictors of high Zung Depression scores, there are very similar variables to what we found when we clustered the controllers by means of their scores obtained from the PSS interview.
THE MOST SIGNIFICANT PREDICTORS OF PSYCHOPATHOLOGY (DISCRIMINATIVE ANALYSIS)
Our final method of looking at psychopathology was to group men not just with respect to the presence or absence of new psychopathology, but in terms of severity. We clustered men into three groups: asymptomatic, mild, and severe. The mild group had a positive score on subjective distress, impulse control disturbance or alcohol abuse only once during the five interviews, while the severe group had positive scores on more than one occasion. Fifty-one out of 302 men, all of whom had all five PSS interviews, or about 17% of the men fell into this severe category. Eighty percent of the severe group had impulse control problems, 65% had alcohol abuse problems and 40% had subjective distress problems. These men could all be considered impaired psychiatrically, even though they were at work throughout the study.
When we analyzed which variables were the best discriminators of the asymptomatic, mild, and severe groups, we found that they were quite similar to those that differentiated the incidence cases from the asymptomatic group. Our severe group had lower self-control on the CPl, more frequent marital dissatisfaction, as well as lower marital resources, lower satisfaction with training, used drinking more to cope with the perceived tensions of the job and reported a diminished ability to bounce back following an absence from work to their previous level of functioning.
CONSEQUENCES OF PSYCHOPATHOLOGY
Another question that may emerge relates to the significance of these findings. Did our detection of psychopathology among the controllers have any consequences? The answer, as one might imagine, is yes. We kept track of the men that were disqualified during the course of the study and asked whether they were overrepresented by those men in whom we had detected any psychopathology. This may be looked at in two ways. There were 20 men who were medically disqualified by the FAA during the course of the study. About half of them were disqualified for a medical diagnosis, half for a psychiatric one. Of these men, 19 were found to have significant psychopathology. Nine were in our prevalence case group and ten in the incidence case group. We also analyzed the data from the converse perspective. There were 135 asymptomatic controls, only one of which (0.7%) was medically disqualified. Among the 99 prevalence cases, nine or 9.1% were disqualified and among the 117 incidence cases, ten were disqualified giving a rate of 8.5%. Thus, it is clear that determination of significant psychopathology carried with it a significantly increased risk for being medically disqualified. SUMMARY
There was a significant amount of psychopathology detected among a representative group of air traffic controllers from the New York and New England regions. These men, who had an average of 11 years on the job, all were screened independently by the FAA on a yearly basis, which is required for controllers to continue to work. Approximately 15% of the men had significant psychopathology on a level comparable to pre-treatment psychiatric outpatients. They certainly would be candidates for psychiatric intervention. A larger number, perhaps another 15% to 20%, had milder symptomatology which caused them distress and could represent the group in which symptoms would be most likely to intensify over time. In any event, having significant psychiatric symptoms placed them at a significantly increased risk of being medically disqualified.
A host of psychosocial predictors were found to significantly discrimínate those who developed problems as well as those who had more severe problems. There were similar predictors discriminating the new cases from asymptomatic controllers as well as the more severe cases from less severe. These included diminished ability for self-control and a diminished sense of responsibility (less apt to follow the rules), more Type A behavior, high investment in the job with a corresponding diminished ability to unwind after work, lower marital support, low co-worker satisfaction, as well as more frequent use of alcohol.
The risk for psychopathology consists of many different factors, incorporating various aspects of the controllers' psychosocial sphere. They relate to an intensification of some personality traits, such as diminished self-control or impaired ability to follow the rules, which was, in some part, encouraged in the peculiar work environment of the controller. As was often found clinically, most of those who got in trouble or became symptomatic, did care, were interested in their work (at least early on), but had less ability to unwind, to cope and to discharge built-up tensions. We also found the controllers had lower social supports, perceived their wives as less sympathetic and were much more critical of co-workers. They also resorted to heavy drinking, usually with some friends after work, and not unexpectedly, got into more fights and experienced more intense and frequent feelings of depression, whether or not these feelings progressed to a level consistent with depressive illness.
Many of these factors which are predictive of future problems appear susceptible to some form of preventive intervention at both the individual and the workplace levels. Some factors appear to have grown out of specific job interactions associated with ATC work, others are perhaps products of high levels of responsibility, often requiring rapid, time-pressured decisions. However, increased support of co-workers or supervisors could potentially alleviate some of the pressures or stressors associated with the job of controlling traffic.
Despite the fact that some of the predictors of psychopathology among these men appear specific or more unique to the job, many do not. Growing alienation, increased use of alcohol, and poor marital relations appear to be those which might be predisposed to greater problems in any group. Thus, these findings document many of our clinical assumptions, and support our attempts to intervene clinically before patients who present with such symptoms get significantly worse.
1. Rose RM. Jenkins CD, Hurst MW: Air Traffic Controller Health Change Studv, Report to PAA. Galveston, Texas. The University of Texas Medical Branch Press, 1978.
2. Rose RM. Jenkins CD. Hursi MW: Health change, in air traffic controllers: A prospective study. I. Background and description. Psychosomatic Medicine 1978; 40: 143-166,
3. Zung WWK: The differentiation of anxiety and dépressive disorders: A biometrie approach. Psychosomatic 1971; 12:380-384.
4. Spitzer RL, Endicou J. Fïeiss JL. et al: Psychiatric status schedule. A technique for evaluating psychopathology and impairment m role function trig. Arch Gen Psychiatry 1970; 33:41-55.
5. Luborsky L: Assessment of outcome of psychotherapy by independent clinical evaluators: Review of the most highly recommended research measures, in Waskow JE, Parloff MB (eds): Psychotherapy Change Measures. US Dept Health. Education, and Welfare publicatidn No. (ADM) 74-170. US Government Printing Office, 1972, pp 233-243.
6. Strupp HH: Review of the psychiatric status schedule: Subject forms in Buros O (ed): The Seventh Mental Measurements Handbook, ed 2. Highland Park. New Jersey. Gryphon Press, 1972. vol 1, pp 309-317.
7. Spitzer RL. Endicou J. Cohen J, et al: The psychiatric status schedule for epidemiological research. Archives General Psychiatry J980; 37: 1 193-1200.
8. Rose RM: Detection of psychopathology in air traffic controllers. Panel on screening normals for psychiatric vulnerability. Chicago, American P&ychiauie Association. 1979.
9. Smith PC. Kendall LM. Hulin CL: The Measurement of Satisfaction in Work and Retirement. Chicago. Rarid McNally & Co, 1969.
10. Stodgill RM. Coons AE: .U'ader Behavior: Its Description and Measurement. Columbus. Ohio, Ohio S tate University Bureau of Business Research. 1957.
11. Hurst MW: Lite changes and psychiatric symptom development: Issues of convent, scoring und clustering, in Raneit JE, Rose RM. Carmen GL(cds): Stress and Mental Disorder. New York. Raven Press. 1979.
12. Hurst MW. Jenkins CD. Rose RM: The assessment of life change stress: A comparative and methodological inquiry. Psychosomatic Medicine 1978; 40:127-143.
FREQUENCY OF OCCURRENCE OF SIGNIFICANT SYMPTOMATOLOGY IN THOSE STUDIED ALL FIVE TIMES BY PSS
JOB DESCRIPTION INVENTORY: COMPARISON OF ATOS WITH OTHER WORKERS
THE MOST SIGNIFICANT PREDICTORS OF PSYCHOPATHOLOGY (DISCRIMINATIVE ANALYSIS)