Can Stress Cause Disease? Revisiting the Tuberculosis Research of Thomas Holmes, 1949-1961

The History of Psychosomatics

The concept that a patient's emotions and social environment may influence health and disease dates back to ancient Greece [7]. However, interest in the mind-body connection declined after 1880 with the development of the germ theory. In the case of tuberculosis, for example, Robert Koch's 1882 discovery of Mycobacterium tuberculosis seemed to prove that the disease was infectious. The introduction of skin testing in 1908, however, showed that many more persons were infected with the bacillus than actually had the disease [8]. The relevant question thus became the following: Why did certain persons infected with M. tuberculosis develop disease, whereas others did not?

Because the germ theory could not explain the occurrence of disease, old psychosocial concerns resurfaced. However, it was not until the 1920s and 1930s that a new discipline, psychosomatic medicine, generated interest in researching the relation between the mind and body [9-11]. Early investigators studied psychosomatics from different perspectives. For example, Franz Alexander [12] championed a “psychoanalytical” approach that posited connections between specific personality types and diseases such as hypertension and ulcerative colitis [13]. Proponents of the psychophysiologic school, building on Walter Cannon's concept of homeostasis [14], examined physiologic responses to both physical and psychosocial stressors. Finally, the psychobiologists, epitomized by Adolf Meyer and Helen Flanders Dunbar, favored a more holistic approach that viewed disease as the result of a complex combination of biological and psychological forces. Meyer depicted this interaction in his famous life chart, which associated disease with important emotional events in a patient's life [9, 15-17].

World War II provided an important impetus for psychosomatic work. Reports of the abominable conditions in concentration camps had roused great interest about the interaction between stress and disease [18]. Tuberculosis was also in the news at this time. In 1946, Selman Waksman had discovered streptomycin, the first of several antibiotics that successfully treated the disease [19]. Whereas the potential “quick fix” of antibiotics encouraged many physicians to emphasize a strictly biological model of tuberculosis, a few researchers struggled to reintroduce psychosomatic issues.

Thomas Holmes

Included among these few researchers was Thomas Holmes, a North Carolinian who completed medical school and a medical residency at Cornell-New York Hospital. After his residency, he remained at Cornell, beginning a fellowship in psychosomatic medicine in 1947. Holmes' interest in psychosomatics stemmed from his relationship with Harold Wolff, a professor of neurology at Cornell best known for his elucidation of the cause of migraine headaches. Wolff's broader research interest was in psychophysiology: studying how the body, through the mediation of the neural system, responded to stressful situations. Wolff defined stress as including both personal events, such as divorce or retirement, and ecologic or environmental factors, such as the onset of war or residence in a poor neighborhood. The body's responses to stress were protective and adaptive, Wolff concluded [20], although they could produce disease if exposed to excess.

Wolff became Holmes' mentor on a project that eventually became a book titled The Nose [21]. This study showed the way in which stressful emotional events caused predictable physiologic responses in the noses of the study participants. For example, anxiety resulting from sexual problems or an officious mother-in-law led to increases in inflammatory cells in nasal secretions and tissue edema.

Despite their collaboration, the personalities of Wolff and Holmes could not have been more different. Wolff was renowned for his punctiliousness, appearing promptly for rounds in a newly starched white coat with a flower in his lapel. Students feared the intense Wolff, and it was said that at least one student each year fainted while presenting to him (Interview with Emanuel Wolinsky, 13 May 1994). Holmes, in contrast, was relaxed, a gentle and humorous man whose great loves were teaching and mentoring.

After concluding his fellowship in 1949, Holmes joined the faculty of the University of Washington Medical School. Although trained in internal medicine, Holmes was based in the Department of Psychiatry in Seattle. Shortly after his arrival, he became the staff psychiatrist at the local public tuberculosis sanatorium, known as Firland. It was at Firland that Holmes did several studies that examined the interaction between stress and tuberculosis.

Research in Psychosomatics

The issue of stress was closely related to the concept of resistance to tuberculosis that Rene Dubos and many other commentators had long discussed [22]. Preventing tuberculosis, Dubos argued, required not only preventing the spread of the bacillus but also increasing the ability of humans to resist the infection. One way to accomplish this was to reduce the “stresses and strains of life” [23]. It was the interaction among stress, resistance, and tuberculosis that Holmes addressed in his first project, a study of urinary 17-ketosteroid levels.

Ketosteroids were of interest because Hans Selye [24] had just shown how various stressful stimuli produced a standard response in the adrenal glands: the production of glucocorticoid hormone. Because the hormone was excreted as 17-ketosteroids, measuring the urinary ketosteroid level enabled researchers to estimate how much stress someone had experienced.

In studying 109 sanatorium residents, Holmes and his colleagues found that persons with normal or high 17-ketosteroid levels had minimal tuberculosis and that their disease tended to be resolving (Figure 1). In contrast, patients with low 17-ketosteroid excretion tended to have far advanced tuberculosis and to be deteriorating clinically. Holmes and his colleagues also examined the emotional state of these patients and found that those with high 17-ketosteroid levels tended to be anxious, whereas those with low levels were depressed. This latter association made sense given that stress and anxiety increased steroid production [25].

Holmes and his coworkers cautiously drew two major conclusions. Because persons with high 17-ketosteroid levels were improving clinically, the researchers suggested that this state was “consistent with resistance to tuberculosis.” Holmes and colleagues also inferred that emotional state influenced adrenal function and thus the course of tuberculosis [25].

The study was flawed in several ways. Like most clinical research done during this era, its uncontrolled and unblinded protocol raised the possibility of many types of bias. Beyond this, however, was an inherent contradiction. Selye [24] had argued that the stress-induced increase in glucocorticoid levels interfered with the body's ability to fight infection, thereby promoting the spread of disease. In Holmes and colleagues' study, however, the patients with increased stress and steroid production were the ones who were recovering. Despite this inconsistency, the study was published in 1954 in the American Review of Tuberculosis.

The research had an additional flaw that would continue to plague the work of Holmes and others in psychosomatics [26]: Even if stress was acknowledged to be associated with activation or worsening of tuberculosis, how could one prove that it caused the deterioration? Certainly one might expect that worsening tuberculosis itself would cause greater stress. The psychosomatics literature often acknowledged this “chicken and egg” problem, and Holmes' later research would try to address this problem by using controls and prospective methods.

In his next series of studies, Holmes worked with Norman Hawkins, a doctoral student in sociology. Using Harold Wolff's broad notion of stress, Holmes and Hawkins examined the ecologic characteristics of the 481 Seattle residents in whom tuberculosis was diagnosed in 1952. The researchers found that higher rates of tuberculosis and more severe disease predominated in parts of Seattle that had high percentages of persons who were not white, had low incomes, and had “skid row”-type lifestyles. In other words, tuberculosis preferentially affected persons in the lower economic and social classes [27, 28].

However, Hawkins was not simply interested in confirming the well-known fact that tuberculosis was a disease of the poor. Rather, he wanted to understand why only certain persons with tuberculous infection ultimately developed disease. The answer, Hawkins postulated, stemmed from the emotional and psychological behavior that resulted from such persons' social relationships and interactions [29].

Hawkins first tested this theory by interviewing 100 patients admitted to Firland, paying particular attention to the following seven issues: marital break by death, divorce, or separation; irregular sleep; financial hardship; job dissatisfaction; drug dependence; alcoholism; and disability. Hawkins found that many patients reported these problems. For example, 71% had experienced financial hardship and 52%, job dissatisfaction; 31% met criteria for alcoholism [30].

These personal difficulties, Hawkins noted, clustered in the 2 years preceding the patient's admission to Firland. This temporal association, which recalled the life chart of Adolf Meyer, suggested that the “apparently high frequency of disorganizing variables” may have played an etiologic role in activating the dormant tuberculosis. It was this hypothesis that Hawkins subsequently explored after adding a control group.

In this study, later published in the American Review of Tuberculosis and Pulmonary Diseases [28], Hawkins and Holmes compared 20 Firland employees who had become tuberculous between 1949 and 1954 with 20 nontuberculous sanatorium employees matched for age, sex, race, income, duration of employment, and skin test status at the time they began working at the sanatorium. The “previous occurrence of psychosocial stresses” was measured by the Schedule of Recent Experience, an instrument devised by Hawkins. Examining the same type of variables as before, Hawkins and Holmes found that the tuberculous employees reported increasing numbers of “disturbing occurrences” during the 2 years preceding their illnesses (Figure 2). Moreover, for each stressor, the persons with tuberculosis were more likely than the controls to have experienced such increases. In several of the categories, the results were statistically significant Table 1 [28, 31].

Hawkins and Holmes also investigated the participants' psychiatric composition by using the Cornell Medical Index. This index measured “level of personal integration” by asking questions about medical problems and issues such as anxiety, anger, and tension. The authors found that 13 of the 20 employees with tuberculosis were “pathologically disturbed” compared with only 5 of the employees without tuberculosis; this finding was also statistically significant [28].

Given the high levels of personality disturbances and the clustering of stressful events among the tuberculous employees, Hawkins and Holmes had seemingly shown that psychosomatic factors played an important role in the genesis of tuberculosis. When they published their findings in 1957, however, their conclusion was noticeably tentative. It was “tenable,” they wrote [28], to postulate that psychosocial crisis was one of the precipitant causes of the development of tuberculosis.

Indeed, Hawkins and Holmes were well aware of the methodologic flaws in their work. The study was retrospective and thus relied on the ability of the participants to recall previous stressful events. Related to this was the old chicken and egg question. In other words, was it possible that employees who had become tuberculous had experienced more stressful events because they were developing the disease? Or had the disease process or the “shock of diagnosis” caused them to give the “immediately preceding experiences a heightened emotional tone” [28]?

Holmes' next study attempted to address some of these issues by using a partially prospective design. The project was conducted by James Hart, a medical student being supervised by Holmes. Hart looked at sanatorium residents who had “thrown a positive”: that is, those who had redeveloped positive sputum after having tested negative for at least 3 months. The study, done in 1957 and 1958, involved 21 such patients and 24 controls whose sputum had remained negative [32].

Because patients and controls had been well matched for both the degree of their tuberculosis and the therapy they had received, Hart concluded that medical factors could not explain the differing outcomes. He thus proceeded to study the psychosocial histories of the patients and controls using the Cornell Medical Index and other instruments. Hart concluded that persons who had developed disease “[w]hen faced with emotional problems and stresses during hospitalization … had fewer resources with which to handle problems” [32]. This paucity of resources stemmed from long histories of unstable lifestyles, emotional and economic problems, and lack of close interpersonal relationships. Because these persons were less equipped to deal with stress, Hart asserted, they had developed tuberculosis.

This work had the same potentially confounding factors as Hawkins' studies; Hart therefore also added a prospective component. He took some of his data and created a new instrument that assigned point totals for various emotional disturbances. In September 1957, he analyzed 10 of the original controls and assigned them scores that reflected their baseline psychological difficulties. Hart predicted that the 2 patients with particularly high scores would “throw a positive” by September 1958, whereas the other 8 would not. Hart's prediction came true, as did a later prediction that none of the remaining 14 controls would have sputum conversion [32].

The final major study conducted by Holmes, titled “Experimental Study of Prognosis,” was published in 1961 [33]. Holmes and colleagues used the Berle Index, an instrument that identified psychological and social factors characteristic of recovering patients. A high Berle score predicted recovery.

Prospectively studying 41 randomly selected patients, Holmes once again achieved expected results. When 26 patients who had achieved normal or high Berle scores (a score of 30 to 58) were located 5 years after testing, none had been classified as a treatment failure. In contrast, 5 of the 15 patients with low Berle scores (a score of 16 to 28) had become treatment failures. The researchers carefully acknowledged that factors other than psychosocial adjustment might have contributed to their findings, and they noted that treatment failure was also significantly associated with male sex, old age, nontuberculous illness, and hospital discharge against medical advice. However, unlike most persons doing epidemiologic research, Holmes was not particularly concerned with teasing out these variables. The study, the authors concluded, had shown the “interrelationship of the natural history of disease and the physiologic, psychologic and sociologic disciplines” [33]. Although few would have disagreed with this statement, it was much less evident that Holmes' decade of work had proven that stressful events caused tuberculosis.

Contextualizing Holmes' Research

How did the “scientific” medical community respond to Holmes' psychosomatic theories? Although his peers did not necessarily agree with all of his conclusions, Holmes apparently did succeed in convincing them that his work was important. For example, in addition to his Air Force funding, he received grants from Washington State and the National Institute of Mental Health. Either Holmes or his students gave presentations at three annual meetings of the National Tuberculosis Association between 1954 and 1959. In addition, as noted earlier, the American Review of Tuberculosis published two of his studies.

Well-respected physicians took Holmes' research seriously. For example, the noted tuberculosis specialist Walsh McDermott maintained an avid interest in Holmes' work. Kerr White, later of the Rockefeller Foundation, termed Holmes' presentation at the 1955 National Tuberculosis Association meeting “magnificent” [34]. In a recent interview, White recalled Holmes as a “serious fellow doing serious work” (20 February 1995). Yet, if many commentators included “emotional strain” as a contributing cause of tuberculosis, others questioned both Holmes' methods and his basic claim that emotions and disease were connected.

Holmes' work received a similarly mixed reception at Firland. As mentioned above, the 1950s were a triumphant era in tuberculosis therapy. Not only did the new medications successfully treat most cases of the disease, but surgeons had begun to use lung resection as an auxiliary measure [35]. The Firland staff shared in the excitement over these advances.

In this environment, Holmes' emphasis on psychosocial mechanisms of disease recalled speculations from a more ignorant, less “scientific” past. Although his gregarious personality won him many friends among the Firland staff, it was not uncommon to hear snickering when Holmes made presentations at the weekly case conferences held to make decisions about patient management (Interview with Edith Heinemann, 13 November 1994). As one commentator later said, Holmes' work was “the biggest bunch of baloney I ever heard” [36].

For many staff members, however, Holmes' presentations struck an important chord, reintroducing the actual patient into the case conferences. The medical history and chest radiograph were of less interest to Holmes than was a consideration of patients in the context of their larger environment. The following discussion of a 29-year-old black woman was typical:

“At this point in her life, the man she later married returned from war service to the small town in Louisiana, and after a month of superficial acquaintanceship, they were married. The marital adjustment was always poor … She spoke of herself as ”not a hotblooded woman“ who preferred church activities. Her husband preferred sports and parties. She resented the fact that he was not a good provider and stated, ‘My father built a good home for my mother. Here we are packed in; I need my own home.’ The husband chose Seattle as a place to live, and in 1946 they moved here … The patient always resented the separation from her family and stated, ‘I always keep the fare home available’” [33].

Holmes then explained why this particular woman had become tuberculous: “It was in this setting of unfulfilled dependency needs, and an increasingly strained marital adjustment in a new and unsympathetic environment, that the patient developed pulmonary tuberculosis” [33]. This analysis recalled the holistic approach to psychosomatics that placed disease in the context of a patient's personal history.

Epistemologic Issues

Did Thomas Holmes actually possess extreme beliefs about the epistemology of tuberculosis, believing that the disease was caused by stressful events in the setting of preexisting psychosocial disturbances, or did he merely think that such factors were among many that contributed to the development of tuberculosis?

Some of Holmes' statements suggest that his theories of disease were radical. Among his more incendiary assertions were that congestive heart failure had nothing to do with the heart, that antibiotic therapy had not affected the incidence of tuberculosis, and that germs did not exist (37, 38; Interview with Archibald Ruprecht, 13 November 1994). At first glance, such pointedly outrageous claims about the origin of disease call Holmes' credibility into question. Had he simply chosen to pursue research in psychosomatics because it provided the best forum for a born showman to express controversial ideas? This does not appear to be the case. Holmes was a notorious provocateur, willing to use such broadsides to challenge his audiences. In more sober moments, he readily acknowledged that infection with the tubercle bacillus was a necessary prerequisite for tuberculosis. As Holmes himself suggested, his more outrageous statements were designed to encourage colleagues and students to “realize that there is more than one point of view in the science of medicine” [39]. In other words, understanding tuberculosis, even in the antibiotic era, entailed knowledge of psychological and social issues.

Yet, Holmes did not simply wish to broaden the perspective on tuberculosis. Both his research and his presentations at Firland sought to challenge the standard model of the disease as a straightforward infection. Holmes truly conceptualized tuberculosis as a psychosomatic disease: Although infection with the bacillus was necessary, tuberculosis could not be understood without recognition of the etiologic role played by underlying personalities and stressful situations. As Norman Hawkins [40] noted, his work with Holmes “raise[d] a serious question whether contagion is even an important issue.”

Although Holmes had thus developed a sophisticated model of tuberculosis as a psychosomatic disease, he had little to say about its prevention or treatment. This omission is perhaps not surprising if the deep roots of the psychosocial problems associated with poverty are considered. Of course, one way to alleviate stress among the poor would have been to provide them with improved jobs, housing and nutrition. Yet, as with most persons involved with the antituberculosis movement, Holmes did not campaign for social reform [41]. His preventive and therapeutic suggestions generally emphasized the need to educate persons about their “fears, anxieties, and misperceptions,” enabling them to anticipate stressful events and thus adjust better to them [32]. The best way to uncover such information was to encourage patients and caregivers to openly discuss such issues [42]. This emphasis on examining and modifying individual habits—as opposed to altering the living environment—reflected both the triumph of the “new public health” movement and the growing influence of Freudian psychotherapeutics after World War II [43, 44].

Conclusion

As the 1950s progressed, Holmes made two major adjustments. He stopped describing episodes such as divorce and loss of job as stressful and began to call them “life events.” In addition, Holmes, like many other researchers at this time, turned away from tuberculosis. Holmes' attempt to inject psychosomatic issues into tuberculosis had not occurred at a very propitious time, given that antibiotics were causing a major decline in the disease. His subsequent work, which focused on the relation between life events and disease in general, became the cornerstone of modern mind-body research (11; Interview with Stewart Wolf, 14 March 1995).

In 1967, Holmes and Richard Rahe transformed Hawkins' old Schedule of Recent Experience into the Social Readjustment Rating Scale, which assigned numeric values to life events to quantify the probability that someone would become ill. Death of a spouse, for example, was worth 100 points (life change units), whereas trouble with a supervisor was worth 23 points. Prospective studies subsequently showed that 80% of persons who had a score of more than 300 points developed a serious illness within 2 years compared with only 30% of those with a score of less than 150 points [42].

This attempt to quantify what was to many a common perception—that stressful events increased the likelihood of illness—met with a popular response in the lay press. The life events scale was discussed not only in Time, but also in the syndicated newspaper column “Dear Abby” and the book Future Shock [45-48]. Despite this recognition, Holmes' work had grown more distant from orthodox medical thought. In contrast to his tuberculosis studies, his research on life events appeared largely in the psychosomatics literature.

In the past 15 years, however, psychosomatic concerns have once again resurfaced. In an era characterized by increased interest in the metaphorical and narrative meanings of illness [49, 50] and by a renewed emphasis on primary care medicine and the care of the whole patient [51-53], major medical journals have expanded their coverage of the role of psychosocial factors in promoting disease. Recent articles in Lancet [54] and Annals of Internal Medicine [55], for example, have reported increased survival of patients with breast cancer and those with cardiac disease who had better psychosocial support. The New England Journal of Medicine has published articles linking stress with both heart disease and the common cold [56, 57].

Modern research on the mind-body connection has benefited from both increasing knowledge of how the immune system works and the ability to measure immune function. Although the term “psychosomatics” is still used, research that directly studies immunity generally carries the more “scientific” designation of psychoneuroimmunology or neuroendocrinology [4, 5]. Recent studies in these fields have suggested that increased psychological distress may be associated with both a decreased proliferation of lymphocytes and lower natural killer cell function. Ongoing work is investigating the relation between these in vitro immunologic findings and clinical disease [4, 5, 58-60]. With only a limited understanding of the immune system, Thomas Holmes posited these types of associations 40 years ago.

Yet, like Holmes' early work on tuberculosis, recent research has also come under fire. The Social Readjustment Rating Scale and its more sophisticated progeny have been criticized for several reasons, including their inexact terminology and their inability to consider the fact that everyone responds differently to stressful situations [60-64]. To address these concerns, researchers now use complex statistical methods such as multiple logistic regression to ascertain more precisely which emotional conditions are actually associated with the development of disease [65, 66]. In addition, modern approaches such as systems theory have shown how a unidirectional model of disease causation such as that proposed by Holmes is too simplistic; causality may be bidirectional or even cyclic [67].

Although these advances in statistical methods and theory improve the validity of psychoneuroimmunologic and life events research, they also may, paradoxically, threaten the ability of such research to shed light on the “whole” patient. For Thomas Holmes, disease did not simply result from the invasion of bacteria or viruses into the body but rather represented an “adaptation to crisis in one's cultural experience” [68]. As Holmes wrote of his patients with tuberculosis [69]:

“[T]hese studies revealed the tuberculous subjects to be sensitive, anxious, rigid, and emotionally labile. These patients, when compared with the cultural norms, were in many ways marginal people at the time of onset of tuberculosis. They started life with an unfavorable social status and grew up in an environment that was for them crippling. They were, in essence, strangers attempting to find a place for themselves in the contemporary American scene.”

The challenge for modern psychosomatic research is to produce statistically convincing studies while retaining Holmes' broad, humanistic outlook.