| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From Duke University Medical Center (F.J.K., C.L.E., D.S.) and Duke University (F.J.K., A.L.H.B., A.K.A.,), Durham, North Carolina; Wayne State University (M.A.L., D.J.M.), Detroit, Michigan; University of North Carolina (J.W.C.), Chapel Hill, North Carolina; University of Lousiville School of Medicine (J.L.S.), Louisville, Kentucky; and Ohio University (L.F.), Athens, Ohio.
Address reprint requests to: Francis J. Keefe, PhD, Duke University Medical Center, Box 3159, Durham, NC 27710. Email: keefe003{at}mc.duke.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Each issue of Psychosomatic Medicine from 1939 to 1999 was reviewed to identify papers that dealt with pain, painful medical conditions, or pain management. A total of 150 papers were identified and grouped into seven categories: 1) case studies; 2) studies of personality traits and other individual differences; 3) psychophysiological studies of pain; 4) studies using pain induction techniques; 5) studies examining the relation of relation of race, ethnicity, and culture to pain; 6) studies of pain unique to women; and 7) studies examining treatments for pain.
RESULTS: A substantial number of studies on pain and painful conditions were published in the Journal in the 1940s and 1950s, and that number has almost doubled in the most recent full decade of the Journal. Within the pain area, however, the topics of interest to psychosomatic researchers have been, and continue to be, quite diverse. Although publications on certain methods or topics (eg, psychodynamic case studies, physiological correlates of pain) have decreased over time, publications on other topics (eg, personality traits and individual differences) have remained relatively constant, and publications on still other topics (eg, studies using pain induction techniques; studies of race, ethnicity, and culture; women’s pain; and treatment studies) have flourished recently.
CONCLUSIONS: Considered overall, the results of our review suggest that the face of pain research published in PM has changed considerably in the past 60 years. Given the ongoing commitment of psychosomatic researchers to this area, we expect this evolution to continue in the years to come.
Key Words: pain, • pain assessment, • pain management, • chronic pain.
Abbreviations: ACTH = adrenocorticotropic hormone;; CBT = cognitive-behavioral therapy;; EEG = electroencephalography;; EMG = electromyography;; MDQ = Menstrual Distress Questionnaire;; MMPI = Minnesota Multiphasic Personality Inventory;; PM = Psychosomatic Medicine;; PMSS = premenstrual symptoms and syndromes;; RCT = randomized, controlled trial;; TMD = temporomandibular disorder.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
The purpose of this article is to review of how psychosomatic research on pain has evolved over the past 60 years. We sought to evaluate how the topics studied by psychosomatic researchers interested in pain have changed, how the research methods used have evolved, and how the conceptualization of the causes and consequences of pain have developed. An overarching goal was to determine how these topics, methods, and conceptualizations have influenced contemporary psychosomatic research on pain assessment and treatment. To accomplish these goals, we systematically reviewed each issue of the field’s leading journal, Psychosomatic Medicine (PM), from its inception in 1939 through the end of the year 1999. Although we acknowledge that there is the possibility of historical bias in focusing on this one journal, we believe that PM is widely considered to be the premier journal in the field of psychosomatic research and that the articles published reflect the major trends in psychosomatic medicine.
A total of 150 papers was identified in PM from 1939 to 1999 that dealt with pain, chronic pain syndromes, medical conditions or diseases for which pain is a primary symptom (eg, inflammatory bowel disease, rheumatoid arthritis), or the treatment of pain. We reviewed these 150 papers and sought to categorize them into mutually exclusive categories. Because we were not aware of any systems that could guide our classification process, we chose to have the members of our research team independently sort the articles into categories based on their own knowledge of the pain literature and research methods, and then we compared and discussed each person’s categorical sorting for similarities. We reached a consensus on seven mutually exclusive categories: 1) case studies; 2) studies of personality traits and other individual differences; 3) psychophysiological studies of pain; 4) studies using pain induction techniques; 5) studies examining the relation of race, ethnicity, and culture to pain; 6) studies of pain unique to women; and 7) studies examining treatments for pain. Figure 1 displays the total number of articles published in PM in each of these categories for each decade from the 1940s to 1990s.1 Variations in the number of studies within each category are evident over time, and we will comment on these variations throughout this review.
|
|
CASE STUDIES RELATED TO PAIN |
|---|
|
TOPABSTRACTINTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
The earliest years of PM saw the publication of numerous case studies related to painful medical conditions. We identified 27 case studies (2–28). As can be seen in Figure 1, 16 case studies were published in the 1940s, 6 in the 1950s, and 5 in the 1960s, but no case studies related to pain have been published in PM since then. Case studies of single individuals were common, but many reports covered a series of patients with the same disorder. With the exception of case studies on phantom limb pain and headache (10, 15, 19), most reports focused on diseases or conditions for which pain was a component of the primary medical concern. The most common presenting condition was ulcerative colitis, followed by other gastrointestinal conditions, especially peptic ulcer, and then other painful disorders such as rheumatoid arthritis and Raynaud’s syndrome.
The typical case study described a patient whose disease or pain was not well managed medically and who was referred for psychiatric treatment. Most case reports provided extensive reviews of the psychotherapeutic treatment, sometimes session by session, with notations on how the patient’s medical condition or pain responded to specific psychological or emotional events that were examined during therapy or that occurred in the patient’s life. A few papers provided only a general overview of a pain condition and then used brief case excerpts to reinforce their points. As a rule these case studies argued that emotional events and conflicts stimulated the pain or prompted disease onset or exacerbation. We noted a tendency, however, for later case studies to present their case material in less detail, to emphasize the observational nature of the report, and to be more tentative in their conclusions.
Traditional psychoanalysis provided the material for several case reports. For example, Karush and Daniels (20) reported on the psychoanalytic treatment of two women with ulcerative colitis; one was treated for 350 hours, and the other was treated for 400 hours. In contrast other case reports used modified forms of psychodynamic therapy that were described as brief (9 months or less), face-to-face, or directive. In reports on larger series of cases, "biographical anamnesis" was sometimes used. This was described as a guided interview over several sessions to obtain an extensive history and was performed by an encouraging and supportive therapist.
The typical case report highlighted certain observations and offered extensive psychodynamic formulations. For example, Daniels (2) reported on a 32-year-old woman with ulcerative colitis who had extreme difficulty extricating herself physically and emotionally from her mother and who denied her own pregnancy—delivering her baby into the toilet, where it drowned. Daniels proposed that repressed hostility, particularly toward the mother, was a key conflict in ulcerative colitis. Johnson et al. (12) reported on a series of 33 patients with rheumatoid arthritis who were seen in psychotherapy or anamnestic interviews. They observed emotional inhibition, a self-sacrificing nature toward others, and unconscious rebellion and resentment against men, leading to their proposal that women with rheumatoid arthritis have chronic inhibited hostility and aggression as a result of masochistic dependence. Brenner et al. (15) described 10 patients with chronic headaches and concluded that suppression of anger and resentment were core issues for these patients. Finally, although some authors suggested that the specific disease of the patient had a symbolic meaning associated with the patient’s conflict, others disagreed. Saul (4) described a competitive and ambitious young man who developed psychogenic back pain due to muscular contractions during sleep, and Saul (29) concluded that the "pain has no direct psychological meaning. . .but would be only the end result of a process initiated by the emotional tensions."
Although most case reports argued that developmental factors caused the pain or disorder, external events were noted as interacting with psychodynamic factors or even having relatively independent effects on pain. Several case studies focused on the effects of World War II. For example, Groen (11) described a patient whose symptoms of ulcerative colitis increased during bombing raids and disappeared when occupying forces arrested a hated family member.
Comment
Case reports from patients in psychotherapy provided the initial method of studying the intricate and complex connections between a patient’s history, emotional processes, and pain. These reports provided a rich bed of clinical observations that buttressed earlier psychoanalytic conceptualizations and generated new psychosomatic formulations. However, these case studies suffered from well-known limitations, including selection biases in the patients examined, selective reporting of observations, and unstructured assessments. These studies also struggled with whether to consider self-reports as reliable; for example, Saul (29) criticized Harris’s (7) attempt to use a patient’s self-report of anger because "there is a question [as to] the importance of consciousness of symptoms." Case studies dealing with painful conditions disappeared completely from PM nearly four decades ago. This shift probably reflected the growing concerns about the limitations of case studies, which also were voiced in the broader fields of psychology and psychiatry, as well a shift in editorial policy favoring studies with more rigorous research methods. As shown in our next section, coincident with the decline in case studies was the rise in publications of group studies of personality factors that were assessed using more rigorous measures and standardized protocols.
Yet paradoxically at least two key features of the case study approach have found new life in contemporary research. First, the case study’s comprehensive and integrative examination of a person in their life context is now reflected in the increased interest in qualitative research. Second, the clinical formulations presented in the early case studies typically focused on the complex, often unconscious emotional regulation processes giving rise to pain. As noted below, subsequent years witnessed a shift away from the study of these complex emotional regulation processes in favor of traits, specific negative emotions, and conscious coping strategies. Yet as attested to by the recent interest in topics such as alexithymia, repression, denial, and disclosure, the field has renewed interest in the processes by which people recognize, express, and regulate emotions, and in people’s varying levels of awareness or consciousness of these processes. Thus, the case study approach had unique strengths that continue to influence us today.
PERSONALITY AND INDIVIDUAL DIFFERENCE STUDIES OF PAIN
Another major theme in pain research is that differences among individuals in their personalities, psychiatric status, emotional functioning, or coping strategies influence pain. These individual difference variables have usually been conceptualized as relatively stable or trait-like variables. As Figure 1 shows, we identified 43 articles on individual differences and pain in PM, with studies published in every decade to the present (30–72).
The vast majority of these studies used one of two types of correlational design. The first is a group comparative design in which a group of patients with a given pain condition or a painful medical disorder was compared cross-sectionally to one or more control groups on some individual difference variable. Control groups have varied as a function of the type of variables that researchers sought to control, but control subjects typically have been healthy people, psychiatric patients without pain, or patients with other pain conditions. More recent studies, however, have used a different design. These studies have attempted to account for variability within a single sample rather than comparing groups of pain patients with control groups. These latter studies, which have become more common in recent years, relate the individual difference variable to pain and functioning, usually in a cross-sectional design, in which the individual difference, pain, and adjustment measures were administered simultaneously. A few studies have used longitudinal designs to test how a baseline measure predicted outcomes assessed at some later time.
The types of individual difference variables that have been studied and the methods used to assess these variables have evolved dramatically over the decades. The articles published in the journal, as well as those in the field at large, have progressed through four types of assessment methods: projective tests, objective inferential personality tests, face-valid scales of specific traits or states, and interviews. Each method has been used to target different types of variables.
Early studies made use of projective testing almost exclusively. These papers appeared from 1945 (30) until 1960 (36). All of these studies examined the classic psychosomatic illnesses (rheumatoid arthritis, migraine headaches, peptic ulcers, or duodenal ulcers), and the studies were conducted primarily to test theories of conflict specificity or specific personality constellations in different disorders. The traits, defenses, or conflicts that researchers assessed included oral strivings and dependency, repression of hostile feelings, and transformation of affect into muscle tension or body image. These processes were conceptualized as primarily or fully unconscious; thus, researchers used projective assessment methods because of their presumed ability to access unconscious processes. The Rorschach Inkblot Test has been the most commonly used projective test in these studies (30, 33–36), but this instrument was sometimes supplemented with other projective tests, including human figure drawings (32, 34), the Thematic Apperception Test (32, 36), or handwriting analysis (31).
A good example of this type of research is a study by Ross and McNaughton (30), who compared Rorschach test results of patients with migraine headaches with those of various control groups (ie, neurotics, patients with other conditions, patients with organic brain disease). Rorschach examinations were conducted and scored in a standardized fashion, and cognitive and behavioral indices of rigidity were more common in the migraine patients. The authors then made inferences about these indices regarding presumed perfectionism, conventionality, intolerance, and difficulty in sexual adjustment contributing to migraines.
In 1965 the journal published a study in which an objective, inferential personality instrument, the Minnesota Multiphasic Personality Inventory (MMPI), was used to study patients with persistent pain (38). Moos and Solomon used the MMPI to compare women with rheumatoid arthritis to their nonarthritic sisters. Other than several studies by Moos and Solomon, we found only one other study published in PM using the MMPI (47) on patients with persistent pain and no use of any other broadband inferential objective personality test. Although there were relatively few studies of the MMPI published in PM, an extensive literature on the use of the MMPI to assess personality characteristics of persons having persistent pain emerged in the 1960s to 1980s (42, 73).
In PM both projective and broadband inferential objective trait measures were replaced by face-valid questionnaires and checklists about specific domains of functioning. In 1970 the Cornell Medical Index was used to assess various demographic, psychosocial, and medical characteristics of patients with inflammatory bowel disease (42, 74), and a mood adjective checklist was used to study correlates of pain in patients with rheumatoid arthritis (43). The late 1970s witnessed an explosion of research using such instruments as the Beck Depression Inventory (47), Spielberger State-Trait Anxiety Inventory (47), and the Profile of Mood States (43), as well as instruments to assess hostility, depression, neuroticism, illness behavior, relationship impairment, counterdependency, and coping strategies. Unlike broadband personality trait tests, which are designed to compare a respondent to normative data, these instruments were typically used not to make normative conclusions but to present relationships between higher or lower scores on these scales and pain-related variables.
For example, Gross (51) administered the Coping Strategies Questionnaire (75) to 50 low back pain patients with lumbar disk disease before undergoing a surgical laminectomy and found that coping factors predicted the degree of pain relief reported after the surgery. Kerns and his colleagues (76) examined coping profiles from the Multidimensional Pain Inventory on 322 patients with temporomandibular disorder (TMD) and found that the development of chronic TMD was predicted by coping style classification.
The shift toward self-reports of individual difference variables was also reflected in the fourth major assessment method that we identified, the interview. Although the interview was also used by some of the early psychosomatic researchers who used projective tests, by the mid-1960s researchers publishing in PM began to rely more on interview data. In 1965 Moos and Solomon (37) used a semistructured interview to gather information on key personality features in both rheumatoid arthritis patients and their closest age- and sex-matched healthy siblings. Although no statistical analyses were conducted on the interview data, the authors noted striking differences in the self-descriptions and childhood experiences of the patients. In 1973 Cohen and Lazarus (45) used an interview to identify the typical mode of coping used by 61 patients to deal with the stress of an upcoming elective surgery. Based on their responses to the interview, patients were classified into one of three types of coping groups, and these groups were compared on pain and adjustment after surgery. A recent study by Adler et al. (53) used an interview to test Engel’s hypothesis about psychogenic pain and the pain prone patient (77). This study compared open-ended interview data about historical variables, medical history, relationships, abuse experiences, and parental health from groups of patients with psychogenic pain, organic pain, psychogenic nonpainful bodily symptoms, or nonpainful organic disease.
From the mid to late 1990s, structured interviews were used in a number of studies of chronic pain. These interview studies have focused on assessing two domains quite relevant to personality and emotion. Structured psychiatric interviews, such as the Diagnostic Interview Schedule, have been used to ascertain the presence of formal psychiatric diagnoses in chronic pain patients and comparison groups (55, 63, 65, 72). The other primary use of interviews has been to assess abuse, victimization, and family problems in people with pain (53, 61, 63, 66). It is noteworthy that interviews, although substantially less efficient than questionnaires as research tools, are being used to assess two domains for which self-reports are considered less reliable, psychiatric diagnosis and histories of abuse and victimization.
Comment
The field of psychological assessment witnessed substantial conflict between advocates of projective and objective personality testing in the 1960s and 1970s. The outcome of this struggle was a great decrease in projective testing and rise in objective personality assessment. This was likely due to the perceived advantages of objective personality measures, including ease of scoring, greater reliability, the availability of normative data, and the need to make less extreme personality inferences from the obtained responses. Consistent with this larger trend, the use of projective techniques fell into disfavor in PM as well, and since 1960 no studies of pain published in PM have used projective testing to assess personality. Interestingly, however, although the MMPI went on to become widely used in pain research (42, 73), studies using this and other objective personality tests in people with persistent pain never gained a substantial foothold in PM.
In contrast there has been a substantial increase in the use of face-valid self-report measures and interviews to assess specific psychiatric disorders, developmental experiences, and coping strategies. The use of specific questionnaires to assess patients’ conscious self-perceptions continues apace and stands in stark contrast to the goals of earlier personality researchers, who used projective techniques because they believed that self-reports were unreliable and incapable of assessing the key personality dynamics. This evolution is perhaps most clearly seen in the transition from the projective assessment of unconscious psychodynamic defenses in the early years to the substantial recent interest in the self-reported assessment of coping strategies, which are conceptualized as conscious efforts to cope with identified stressors.
These articles have tended to interpret personality as causing or exacerbating the pain rather than, for example, being a consequence of the pain. Yet a major limitation of these studies is that they are correlational and primarily cross-sectional, rendering it impossible to establish causality. There is accumulating evidence that persistent pain substantially alters cognitive and emotional functioning. Although the articles in this journal have tended to advocate the view that psychological factors cause or influence pain, we noted a tendency for more recent research to be more cautious in interpretations of causality. Although still infrequent, we also noticed a trend for more longitudinal studies, which offer better tests of causality than cross-sectional designs. In addition, early research in this category sought to identify specific personality types among people with particular pain conditions, but these efforts did not meet with much success as evidence for substantial personality heterogeneity within disorders was found. This led to a substantial reduction in between-group designs—comparing pain patients with control subjects—and an increase in within-group correlational designs.
There has been some continuity in this literature, however. Emotional processes and disorders remain of interest, and researchers continue to explore the possibility that such processes contribute to pain. It is noteworthy, and perhaps ironic, that the recent surge of interest in early experiences that predate pain onset, particularly sexual and physical abuse and premorbid depression, returns the field to ideas that interested early researchers: that conflicts during development alter emotional functioning and may lead to chronic pain.
PSYCHOPHYSIOLOGICAL STUDIES OF PAIN
Another set of studies also examined correlates of pain, in this case psychophysiological and other biological correlates of naturally occurring pain or painful diseases. We found a total of 20 physiological studies of pain in PM, and as shown in Figure 1, most of these studies were published before the 1960s (78–97). These studies typically examined patients with some chronically painful condition and presented some type of mental or emotional stressor or stimulus to the person while physiological measures were recorded. In some studies control groups were also assessed. A wide variety of physiological measures were used in these studies, including electroencephalography (EEG), electromyography (EMG), electrodermal activity, pupillometry, cardiovascular activity, radiography, and assessment of electrolytes and other fluids.
The first psychophysiological study appeared in 1939, the inaugural year of the journal. Mittleman and Wolff (78) studied eight patients with Raynaud’s disease, a disorder of decreased blood flow and thermal dysregulation in the fingers and/or toes that leads to pain in these extremities. The authors described how various emotional stressors delivered in the laboratory led to dramatic decreases in finger temperature.
Studies of the psychophysiological correlates of headache were the most common type of this study that we identified, including five original articles and one review. Simons and Wolff (81) used the EMG to study mechanisms of posttraumatic headache, and they found evidence of elevated muscle activity in the majority of headache sufferers. Arellano et al. (84) studied EEG responses during mental stressors (mental arithmetic and a memory task) and found that people with migraine headaches had different EEG patterns than people with simple headaches and normal control subjects. Schottstaedt (88) explored whether variations in plasma fluid balance preceded vascular headache and played a role in headache etiology. Schottstaedt failed to identify a reliable pattern of electrolyte (potassium, sodium, creatinine) excretion before, during, or after headaches. Finally, two studies tested whether headaches after lumbar puncture were due to vascular "leakage" or to psychological factors such as expectations (92). Redlich et al. (80) found that headaches were much more common when lumbar puncture was conducted with a 22-gauge needle than with a 16-gauge needle, but patients’ emotional stability and intelligence were not highly correlated with the occurrence of post—lumbar puncture headaches. Kaplan (92) revisited this issue 21 years later with improved methodology. He conducted a double-blind study and found that there was no difference in the incidence of headache between diagnostic and sham lumbar punctures.
Also in these early years, psychophysiological studies of patients with rheumatoid arthritis or ulcers were fairly common. Gottschalk et al. (82) tested Alexander’s hypothesis that certain psychological conflicts result in increased muscle tension among people with rheumatoid arthritis. Jungmann and Venning (86) used radiology to examine peristalsis and other stomach changes among patients with peptic ulcer while they experienced the stress related to having an injection. Fisher and Cleveland (90) tested the hypothesis that people with "exterior" symptoms, such as found in rheumatoid arthritis, are more physiologically reactive on the body’s exterior than are people with "internal" symptoms, such as duodenal ulcers, who are more reactive on the body’s interior. Patients with either of these diseases were exposed to a sudden loud noise (metal bar dropped on the floor behind them) while their galvanic skin response (an exterior response) and heart rate (an internal response) were monitored. They reported data consistent with the theory. However, a study 4 years later by Williams and Krasnoff (91) on the same patient groups was unable to replicate these earlier findings.
Beginning in 1975 Moldofsky and colleagues published three studies investigating the relationship between sleep architecture and fibrositis syndrome (93, 95, 96), which is characterized by widespread musculoskeletal pain, localized tenderness, and chronic fatigue. In the first study they compared the patients with fibrositis and normal control subjects and found the anomalous presence of alpha rhythms in non—rapid eye movement (delta) sleep EEGs (93). Their second study examined a broader array of possible etiologic factors, including altered sleep physiology, altered brain serotonin metabolism, environmental and emotional factors, and inflamed joints (95). In the third study (96) they improved on their control group by comparing the sleep of women with fibrositis to that of women with postaccident pain. Using this better control group, they found no group differences in sleep architecture, leading them to conclude that sleep changes are not necessarily specific to fibrositis and that emotional stress may trigger the onset of symptoms.
Although there has been a decrease in the number of publications dealing with psychophysiology and pain in PM during the past two decades, the few studies that have been published have used more ecologically valid stimuli or stressors. In 1981 Block (94) sought to understand the impact of the interpersonal environment on the behavior of patients with chronic pain. Block examined skin conductance and heart rate responses of spouses of hospitalized chronic pain patients while the spouse watched videotapes of pain behavior emitted by the patients. Hatch et al. (97) sought to test the theory that neck and frontal EMG is elevated among patients with tension headaches. They studied physiological responding to stressors in the natural environment by using ambulatory recording for several days on subjects with tension headaches and control subjects without headaches.
Comment
There are several noteworthy observations on the evolution of studies of psychophysiology and pain. First, many of these studies made distinctive and clear tests of specific hypothesized mechanisms, and the results either supported or, often, refuted the hypothesis. This stands in contrast to the case studies, which were purely descriptive and were often presented to support a theory, and the personality and individual difference studies, which seemed to only support rather than refute theory. In this set of psychophysiological studies, we also see for the first time clear attempts at building on prior studies by attempting to replicate or refute earlier work, such as the studies of lumbar puncture—induced headaches (80, 92), exterior vs. internal symptoms in rheumatoid arthritis and ulcer (90, 91), and the series of studies of fibrositis and sleep (93, 95, 96). The attempts at replication typically used much better methodology, such as blinded assessments or more appropriate control groups.
There has also been evolution in the types of stressors or stimuli used. Early psychophysiological studies attempted to use emotional stressors or conflict situations that were appropriate to the disorder or specific patient being studied. This theoretically and ecologically valid approach, however, was very difficult to operationalize and control, and subsequent studies shifted to more easily operationalized and structured stressors, such as noise, mental arithmetic, and injection threat, which unfortunately were less relevant to the emotional processes of interest. It is noteworthy that recent years have witnessed a return toward more ecologically valid stressors, such as interactions with spouses and ambulatory recording in the natural environment.
Studies in this category were published primarily from the 1940s to 1960s and have been largely absent from PM in the last several decades. This trend reflects a pattern seen in pain research in general, where attempts to demonstrate unique or specific psychophysiological correlates of disorders such as headache, rheumatoid arthritis, or ulcer generally have not been fruitful. Many of the studies in this area have had negative findings or results that were moderated by individual differences among the patients. Interestingly there has been an upsurge of neuroscience research on pain, and most of it shows that a matrix of multiple brain sites, rather than a single brain site, are involved in pain perception (98, 99). Also, as seen in this set of studies, a wide variety of physiological measures have been used, and their links to psychological processes are complex and multidetermined. The recognition of this complexity of psychophysiological measures has led to a decrease in their use in many fields.
PAIN INDUCTION STUDIES
Although studies examining how psychological traits or physiological responses relate to chronic pain have been intriguing, these studies have been limited by the fact that they are unable to control or manipulate pain. A number of more recent studies published in PM have examined how psychosocial factors influence responses to experimentally induced pain in the laboratory. We identified 18 studies of this type, and as can be seen in Figure 1, all were published from the late 1960s to the present (100–117). These studies examined sensitivity to experimentally induced pain, typically among either healthy people or at least pain-free individuals, to better understand which psychological factors influence pain. Studies in this group used a wide variety of pain stimuli, including needle prick, saline injections, thermal stimulation, electrical stimulation, cold pressor tests, ischemic techniques, and mechanical pressure. A few studies, not included in our tally, used pain stimuli as generic stressors to understand functioning in other nonpain conditions, such as neurotic disorders or hypertension.
Six of the pain induction studies published in PM manipulated the presence or absence of some mental or physical state using a randomized experimental design and then examined responses to induced pain. The methods of three of these studies will be presented as examples. Feather et al. (103) used sensory decision theory methods to assess sensitivity and pain report criterion in response to thermal pain induction among healthy adults after they received either a placebo pill (described as a "potent, fast-acting analgesic" medication) or no pill, using a within-subject design. Knox et al. (106) evaluated the effects of acupuncture and expectancy on experimentally induced cold pressor pain using a factorial design. Acupuncture vs. placebo acupuncture was crossed with high, low, and no expectancy conditions in a small sample of healthy people. Finally, Zillman et al. (114) induced positive, neutral, or negative moods by having participants view mood-altering videotapes and then examined pain sensitivity to pressure and the cold pressor test. All three of these studies found evidence for the power of placebo, expectations, or positive moods on pain reports. Interestingly the Feather et al. study (103) found that although placebo did not influence sensitivity to pain (ie, the ability to discriminate between high- and low-intensity noxious stimuli), it did have a significant effect on pain reporting (higher pain report criterion.)
Another set of studies were correlational rather than experimental, examining relationships between various demographic, biological, and psychological factors and responses to induced pain. These factors included body image and anxiety (100), repressive coping/self-deception (110), eating disorders and endogenous opioids (111), depression or panic disorder (116), localized vs. generalized hyperalgesia among patients with headache vs. fibromyalgia (117), and autonomic reactivity among children with recurrent abdominal pain or healthy control subjects (101). Two examples will illustrate these studies. Rubin et al. (101) studied 13 children with recurrent abdominal pain and 12 healthy children, all of whom were exposed to cold pressor pain, and the two groups were compared on autonomic arousal as indexed by pupillary reaction and recovery times. Lautenbacher et al. (116) studied both naturalistic pain reports and responses to experimental pain (heat, cold, and pressure) in people diagnosed with depression or panic disorder and normal control subjects. It is noteworthy that the latter study concluded that clinical pain complaints and experimentally induced pain seem to be influenced by different processes.
A review article published in PM on the use of signal detection theory and methodology for pain research (105) was followed by several studies using this approach. Yang et al. (109) used signal detection methods to assess responses to thermal pain stimulation in patients with chronic low back pain and healthy control subjects. Malow and Dougher (107) used this methodology to evaluate the effects of a transcutaneous nerve stimulator on responses to mechanical pain stimulation.
Finally, two recent studies by Fillingim, Maixner, and colleagues (113, 115) examined the relationship between blood pressure and pain sensitivity. The first study (113) examined whether sex differences in blood pressure may explain commonly reported sex differences in sensitivity to laboratory pain. Male and female normotensive participants had their blood pressure assessed and were exposed to thermal pain and ischemic pain inductions, and the relationship between blood pressure and pain reports were examined for each sex. In a second study, Maixner et al. (115) extended their exploration of links between blood pressure and pain sensitivity in a study of female patients with TMDs and matched control subjects.
Comment
The publication of research on experimentally induced pain has become much more frequent since the 1960s. We suspect that this stems in part from the interpretive complexity of correlational studies, particularly when such studies are conducted on people who have had persistent pain. It is noteworthy that a substantial proportion of these studies used a randomized experimental design, in which the psychosocial variable of interest (eg, expectancy, mood, acupuncture) was manipulated to determine its effects on induced pain. A number of experimental studies also used signal detection theory methods, which enable one to carefully assess both sensitivity to noxious stimuli and pain report criteria. Overall, experimental studies can provide some of the most robust evidence of the causal role played by psychosocial factors in influencing pain.
However, despite the greater methodological control provided by pain induction methods, these studies have limitations. Most clinicians and researchers are interested in acute clinical pain or persistent pain, and the generalizability from induced pain studies to either acute clinical pain or persistent pain is limited. For example, subjects in induced pain studies are aware of the source of the acute pain stimulus and know that they can stop the stimulus at any time. This is typically not the case for a person who must endure an acutely painful clinical procedure (eg, an invasive medical procedure), and it is certainly not true for people with persistent pain. Future research is needed to determine the degree to which the results of laboratory-induced pain studies can be generalized to people experiencing pain outside the laboratory setting.
STUDIES OF RACE, ETHNICITY, CULTURE, AND PAIN
The introduction of the gate-control theory of pain in the 1960s (118) ushered in an era where the conceptualization of pain was broadened to include social and cultural influences. In addition, the increasing diversity of our society and concomitant interest in understanding the health and functioning of all people spurred research on social and demographic correlates of pain, particularly with respect to race, ethnicity, and culture. In this section we review studies published in PM that examined differences among racial or ethnic groups on pain reports or behaviors. We found seven studies dealing with this topic (119–125). As Figure 1 shows, all of these papers were published in the past three decades, with four of the seven studies published after 1990.
The first article in PM dealing with ethnicity and pain was published by Woodrow et al. (119) in 1972. These authors examined pain tolerance to mechanical pressure among 40,000 members of a managed care program, and they compared people of Asian descent to blacks and to whites. One finding that is relevant to our discussion is that observed racial group differences were generally smaller than age and sex differences. Three years later, Weisenberg et al. (120) reported on the attitudes and pain reports of black, white, and Puerto Rican patients at an emergency dental clinic. Streltzer and Wade (121) related ethnicity to the treatment of postoperative pain by reviewing medical charts of all patients undergoing elective cholecystectomy in two Hawaiian hospitals. They classified patients as Caucasian, Chinese, Japanese, Filipino, Hawaiian, or other, and then compared groups on the amount of parenteral narcotics received. Of relevance to our review is that although ethnic factors were important in univariate analyses, multivariate analyses revealed that they accounted for only a small proportion of the overall variance in narcotics received, particularly after controlling for the effects of age, sex, and surgical and hospital factors.
In 1990 Zatzick and Dimsdale (123) published a meta-analysis of 13 studies that examined the effects of culture on laboratory-induced pain. In addition to concluding that there was insufficient evidence of a difference between cultural groups on the neurophysiologic detection of pain or responses to pain, these authors noted two major limitations of the available studies. First, these studies lacked standard definitions of culture, race, and ethnicity; and second, they tended to ignore the tremendous diversity within the groups examined.
Studies published in the 1990s on ethnicity or culture and pain seem to have been influenced by these concerns and recommendations. Moore (122) conducted an ethnographic exploration of pain coping perceptions among 54 patients and 31 dentists of Chinese, Anglo-American, or Scandinavian descent. All participants were interviewed, and a systematic content analysis was conducted to identify common themes. The ethnic populations were described as being of "first generation ethnicity, living in ethnic communities in the greater Seattle area." This paper attempted to relate pain coping responses to the degree of Western or Eastern acculturation. Ng et al. (124) examined the effects of ethnicity on the use of postoperative analgesia among African American, Hispanic, and white patients. They found some ethnic group differences in analgesia administration, and these differences remained significant even after controlling for age, gender, insurance status, and the number of diagnoses. Finally, Edwards and Fillingim (125) exposed African American and white college students to thermal pain stimuli and then compared these groups on various pain report indices. Of note, this paper recognized the complexity of race issues, defined quite carefully the racial groups, and eliminated subjects whose self-identified race was equivocal.
Comment
Over the past three decades, particularly in the 1990s, there has been increasing interest among researchers publishing in PM in the relationship of race, ethnicity, and culture to pain. The increase in publications probably reflects the growing awareness on the part of researchers and funding agencies of the need to consider ethnic diversity in health and health care. It may also reflect the journal’s editorial approach, as the editor during the last decade has been particularly interested in issues related to ethnicity.
There are some limitations or areas of concern in these studies. First, as noted by Zatzick and Dimsdale (123), there continues to be confusion among researchers in the use of the terms race, ethnicity, and culture. Researchers tend to use these terms interchangeably, seemingly without realizing the differences and without clarifying the terms in the Methods section or discussing their conceptualization or limitations. This was especially clear in the studies published before 1990, although recent studies suggest some improvement. Traditionally race has been applied as a loosely defined term that classifies people based on their ancestry and distinguishing biological characteristics. Ethnicity, a much broader concept, is normally defined as a group of people who have similar beliefs, laws, customs, histories, and language, and are most often of the same race. Culture is an even broader term, defined as a stable carrier of distinctive structures and patterns that influence perspectives, cognitions, and health. Of these terms, culture was most frequently presented as the topic for study, but in actuality, was rarely or never studied. Researchers should strive for conceptual clarity in the use of these terms because the meaning of findings hinges on the consistent and accurate use of them.
Second, the implications of finding racial or ethnic group differences are not clear. What do these differences mean? Are the observed differences due to underlying biological differences? Do they reflect socioeconomic differences? Are they due to differential acculturation regarding the meaning of pain or how pain is communicated to others? Does the context, such as the setting of the study, the attitudes of treating clinicians, or the race of the experimenter, interact with the subject’s race to influence pain responding? It is noteworthy that the studies examined here usually found some differences among ethnic and cultural groups in pain behaviors, but often these differences were smaller than the effects attributable to other factors. Thus, many questions need to be addressed so that we can determine whether pain research should continue to compare race/ethnicity categories or instead study other processes that are associated with these categories.
STUDIES OF PAIN UNIQUE TO WOMEN
In addition to research on ethnic and cultural diversity, there also has been substantial interest in pain conditions that affect only women. Although women report more intense pain than men in response to medical conditions and aversive medical procedures (126), and also report more frequent and intense headaches and musculoskeletal pain than men (127), there are at least two pain conditions that only women encounter, labor/childbirth and premenstrual pain. We identified 16 studies published in PM that focused on these two conditions in women (128–143). As can be seen in Figure 1, the number of these studies published in PM has increased substantially over the past two decades.
Four of the 16 studies dealt with issues of labor and childbirth. Two studies dealing with natural childbirth were published in 1952. The first study used a 6-month follow-up questionnaire to retrospectively assess recollection of labor in series of women who had either received (N = 381) or did not receive (N = 389) training in natural childbirth (129). The authors found little evidence of benefits of natural childbirth. However, they acknowledged that the questionnaire methods had limitations and that women may not have accurately reported their experiences to the hospital personnel collecting the data. The second study used a combination of interview and observation methods to understand how natural childbirth related to women’s memory of labor and delivery (128). Each participant’s narrative of the childbirth process was then compared with staff ratings of the woman’s pain to judge the patient’s ability to accurately remember the event. The next study on labor and childbirth was not published until 1974. Davenport-Slack and Boylan (131) used a comprehensive set of interview, observation, and questionnaire methods to examine psychological correlates of childbirth pain, including attitudes toward childbirth, training for childbirth, menstrual pain, and sexual desire, among 75 women. A review paper published in PM in 1980 (133) noted the need for more rigorous treatment studies of natural childbirth featuring random assignment to treatment and credible control conditions to determine the potential benefits of natural childbirth. Such controlled studies were subsequently published in the pain literature, and their results provide mixed evidence with regard to the efficacy of natural childbirth (144–146).
Twelve studies in the journal have examined pain in the context of premenstrual symptoms and syndromes (PMSS). Several studies specifically examined pain as a major symptom of PMSS. One such study was a publication by Moos (130) that presented the Menstrual Distress Questionnaire (MDQ) and its psychometric properties. The MDQ includes eight scales, one of which was specifically designed to measure painful symptoms of PMSS, such as headache, cramps, and backache. Results of this study provided strong support for the reliability and validity of the MDQ, and this measure subsequently has become widely used by individuals conducting research on PMSS. The MDQ has been used in a number of PMSS studies published in PM, including studies examining the relationship between personality factors and PMSS (132) and the relationship between affective symptoms and PMSS (135).
Five studies published in PM examined the relationship of mood to PMSS symptoms (136, 139–142). Two studies focused specifically on the issue of variability in PMSS symptoms. Marbach et al. (142) examined seasonal variability of PMSS but found no evidence that the premenstrual symptoms of general discomfort and edema increased during the dark months of the year. Fillingim et al. (143) reported that pain sensitivity varies throughout the menstrual cycle, with greater variability of ischemic pain sensitivity compared with thermal pain sensitivity.
Comment
Our review identified a small, but growing literature on pain unique to women in PM. For the most part, these studies have been primarily descriptive and correlational in nature and, with one exception (147), have not used controlled experimental designs to examine psychotherapeutic and pharmacological interventions for treating or managing pain in women. The paper by Moos (130) was unique in our review of pain research in the journal because it was the only one in PM devoted solely to the development of an instrument for pain assessment. This publication led to a spate of studies in the journal using the same instrument, which allows for contiguity in the literature.
STUDIES OF INTERVENTIONS FOR PAIN
It is difficult to conceive of a more important clinical problem in psychosomatic medicine than the effective treatment of pain. Clinicians in psychosomatic medicine are frequently asked to provide consultation or treatment for patients having persistent pain due to arthritis, gastrointestinal disorders, cancer, headache, or musculoskeletal disorders. Typically the patients referred for treatment have long histories of pain and multiple failures in responding to conventional medical or surgical interventions. In general the treatment approaches used by these clinicians have focused on identifying and then altering psychological and social factors that can influence pain and disability. In this section we review studies published in PM that focused specifically on interventions to reduce pain, whether these were review articles, correlational studies, or randomized, controlled trials (RCTs). (Case studies in which the treatment context provided information on the role of psychological variables in pain were reviewed earlier.) Our review identified 19 articles in this category (148–164). As can be seen in Figure 1, the number of these studies has been increasing over the past three decades.
Chronologically these 19 studies can be divided into two major epochs. Articles published before early 1973 were not RCTs, but many of the articles published from late 1973 to the present were RCTs. The first article on pain treatment that was published in PM appeared in 1953 (148). This study reported on symptomatic improvement and emotional changes in patients with rheumatoid arthritis who were treated with adrenocorticotropic hormone (ACTH) therapy. This was a partially controlled outcome study in which a nonrandomized subsample was given placebo treatment before crossing over into ACTH therapy. The study is noticeably different from subsequent pain intervention studies published in PM in two regards. First, the generally positive outcomes were analyzed descriptively rather than statistically; and second, the manner in which patients were assessed strongly reflected the extant dominant psychoanalytic model (eg, "arthritic symptoms were an essentially hysterical solution to unacceptable sexual urges").
Three other early articles dealt with miscellaneous pain intervention topics. In 1962 the journal published a review of controlled and noncontrolled outcome and experimental studies of hypnosis treatment for various pain conditions (149). The second article reported a cross-sectional study of nursing staff and patient variables that influence in-patient pain treatment (150). Also published during this period was a positive report of an uncontrolled pilot study of a treatment combining autogenic training and biofeedback training for migraine and tension headaches (165).
The second epoch of treatment-related studies, 1973 through 1999, is marked by the relatively frequent appearance of RCTs (N = 9) or reviews of RCTs (N = 2). Ten of these studies dealt with psychologically based therapies, and one was a study of pharmacotherapy. Increasing methodological sophistication is evident in these studies, marked by desirable design features including measures of patients’ treatment expectancies, the use of treatment manuals to standardize treatment delivery, the use of attention control comparison groups, and the comparison of alternate forms of active treatment (eg, various biofeedback methods). All of the psychological treatments fit under the general rubric of cognitive-behavioral therapy (CBT), and all studies of such treatments concluded that the respective approaches were generally efficacious for the targeted conditions when compared with no therapy or attention control conditions.
The first RCT of a pain intervention appeared in the journal in 1973 (166). This investigation compared the effects of EMG biofeedback, pseudo-biofeedback, and no treatment control on tension headaches. Patients kept daily records of headache activity and medication usage, from which weekly mean scores were derived. Another representative intervention study is a report published by Marcus et al. (147) of a 4-session CBT treatment protocol that combined progressive relaxation, skin-warming biofeedback, educational information, and physical therapy for headaches during pregnancy. In this article the results of two studies were reported. The first study used an uncontrolled pre-post design, and the second study randomly assigned 25 pregnant women with headaches to either the CBT intervention or an attention control condition (headache education and skin-cooling biofeedback) and verified that there were no between-group differences in patients’ treatment expectancies.
There were several non-RCT articles published during this epoch, and they varied greatly in focus. One paper attempted to address the widespread undertreatment of cancer pain by providing clinicians with a review of pharmacological and psychological treatment options (155). Another reported on psychological changes, such as decreased depression and anxiety, reported by headache patients across three CBT studies (159).
Comment
The studies covered in this treatment section reflect two important trends that have developed over the past six decades. The first trend was the emergence of the biopsychosocial model of pain. The treatment of pain was long guided by the biomedical model, which viewed pain as a simple sensory event directly related to underlying disease or tissue damage. Hence, treatment was directed toward correcting the organic dysfunction or pathology. However, as seen in our review of studies published in PM, pain has come to be viewed as a complex phenomenon—one not exclusively determined by tissue damage or organic dysfunction but also strongly influenced by a wide array of factors such as attentional focus, mood, meaning of the situation, prior learning history, cultural background, environmental contingencies, and social support. With acceptance of a biopsychosocial model of pain, pain treatment specialists began developing and testing a variety of biopsychosocial interventions aimed at addressing such factors.
The second important trend was the increasing emphasis on methodologically rigorous treatment outcome studies that empirically test the efficacy of psychosocial interventions. This trend was reflected in a dramatic decline in the publication of descriptive or uncontrolled intervention studies, which were largely supplanted over time by RCTs. These RCTs generally have provided strong evidence for the efficacy of biopsychosocial pain management.
It is noteworthy, however, that PM, long known for its emphasis on psychodynamic models or the role of emotions in disease pathology and treatment, has published pain intervention studies that have used only cognitive-behavioral interventions, such as biofeedback, relaxation training, problem solving, and coping skills training.
MAJOR THEMES AND FUTURE DIRECTIONS
Research on pain has been appearing in issues of PM since its inception. As shown in Figure 1, a substantial number of studies on pain and painful conditions were published in the journal in the 1940s and 1950s, and that number has almost doubled in the most recent full decade of the journal. Although part of this increase is due, no doubt, to the increased numbers of articles published yearly in the journal, it seems that research on pain has gradually become more prevalent. Within the pain area, however, the topics of interest to psychosomatic researchers have been, and continue to be, quite diverse. Although publications on certain methods or topics (eg, psychodynamic case studies, physiological correlates of pain) have decreased over time, publications on other topics (eg, personality traits and individual differences) have remained relatively constant, and publications on still other topics (eg, studies using pain induction techniques; studies of race, ethnicity, and culture; women’s pain; and treatment studies) have flourished recently.
A unifying conceptual notion that has guided psychosomatic research on pain is the importance of individual differences in understanding pain. The theoretical models used to understand these differences, however, have changed considerably over the years. Early studies were guided almost exclusively by psychodynamic models that emphasized the role of unconscious processes, internal conflicts, and personality constellations as causes of, or contributors to, pain. More recent studies have adopted a much broader biopsychosocial model that recognizes that pain is a complex phenomenon that is influenced by, and in turn influences, biological, psychological, and social factors.
As we have noted, there are a number of trends in the methodology used to study pain. One trend is the move away from less rigorous case studies to more methodologically sound experimental designs. This change was precipitated by concerns about selection bias, clinical bias, and lack of external validity inherent in the case study method. A second trend is that projective testing has been replaced by more standardized, self-report measures. With this change has come an improvement in both the reliability and validity of measures. Early studies relied on a fairly restricted set of dependent variables, whereas more recent studies have used multiple dependent variables. Interestingly, the interview remains one of the most popular methods of assessment in the pain studies we reviewed. Although interview methods used in early studies were primarily clinical and of unknown reliability and validity, the most recent studies using interviews employ structured, reliable, and valid interview protocols. As a result of many of these changes, data analysis has moved from a descriptive approach that attempts to identify themes and underlying conflicts in case study material to more powerful inferential statistics. These newer, statistical approaches to data analysis not only provide the opportunity to examine relationships between variables but also to directly test the effects of experimental manipulations. One of the most important methodological trends has been the increasing publication of RCTs testing clinical interventions to reduce pain and disability. Interestingly, literature reviews also have moved from qualitative analyses of selected, published studies to comprehensive quantitative analyses of published and unpublished studies conducted using meta-analysis.
The populations studied in these research investigations also have changed over the years. Early studies focused primarily on pain in conditions or diseases considered to be psychogenic in nature. More recently, studies have focused on a much wider range of disease-related and non—disease-related pain conditions. Also, recent research has attempted to study how psychosocial factors contribute to pain by inducing acute pain in otherwise pain-free individuals. Recent studies also have been more consistent in their inclusion of populations that vary in terms of gender, race, and other demographic factors.
It should be remembered, however, that publications in this journal dealing with pain are not necessarily reflective of pain research in the larger field. This is particularly true for research in the last two decades. The development of specialty pain journals (eg, Pain, The Clinical Journal of Pain, The European Journal of Pain) as well as the increasing tendency for medical journals dealing with painful conditions (eg, Arthritis and Rheumatism, Gastroenterology, Headache) to publish psychosocial research have siphoned away psychosocial pain research that might otherwise have been published in PM. In addition, there are inherent limitations to making conclusions about trends when examining only one journal. Unique editorial policies or preferences may bias the types of research published. For example, we have noted that relatively few studies using objective personality trait measures such as the MMPI have been published in PM compared with the broader literature. We suspect that studies of coping and other cognitive-behavioral constructs (eg, self-efficacy) are also underrepresented in the journal compared with the larger literature, whereas psychodynamically oriented case studies are probably overrepresented.
In the past decade or so, there has been a surge of interest in the role of emotional factors in health, and this interest has been found in pain research as well (167). Because PM has been a leading outlet for research on how emotional processes influence health, we believe that the journal is well situated as an outlet for investigators participating in this renewed interest in pain and emotion.
There are several important directions for future pain research. First, the majority of papers we reviewed were correlational or descriptive in nature. Not only do these studies render determinations of causality impossible, they usually fail to address the multiple contributory factors that influence pain. Experimental studies, both laboratory-based studies of mechanisms and clinic-based studies of intervention, are increasingly being used because they overcome such problems. Along these lines, researchers need to be careful to control for variables that can confound the relationships they seek to understand (eg, the relation of stress to pain). In particular, it is important to use methodological or statistical methods to control for the influence of background factors (age, sex, race) and disease factors (eg, disease severity, stage) that can influence pain.
One of the most interesting directions for future research is to go beyond documenting the fact that differences occur on the basis of gender, race, or culture to understanding the underlying mechanisms responsible for these differences. Recent research on gender and pain, for example, suggests that neurological factors such as the propensity for descending control over pain (168) and psychological factors such as coping (169) may be important in explaining gender differences.
Future studies also need to take full advantage of newly developed technologies for studying pain. Noninvasive methods are now available to study biological responses likely to be related to pain (99). These include methods for assessing changes in gene expression that occur during exposure to pain (170), reflex responses related to central modulation of pain (168), and imaging techniques for studying brain areas involved in pain perception. Personal data assistants are now widely available and can be programmed to gather data in the natural environment (171). Taken together, these new methods are potentially powerful tools for analyzing how variables such as stress and coping affect pain and disability, but to date none of these studies have been published in PM.
Perhaps the most important goal for future research is to improve our understanding of how to best treat pain. First, there is a need for outcome studies that test the efficacy of protocols that combine psychologically based therapies with biomedical treatments. Second, research needs to examine the cost-effectiveness of interventions and different methods of treatment delivery (eg, minimal therapist contact, home-based intervention vs. traditional clinic-based approach). Third, treatment studies need to take into account the social context of pain treatment by involving spouses, family members, and caregivers in treatment interventions. Fourth, patients appear to vary in their responses to pain treatments, and more investigation is needed to identify predictors of treatment response so as to permit the tailoring of treatment to the needs of individual patients. Finally, studies are needed to compare more thoroughly studied cognitive-behavioral interventions with novel interventions such as emotional disclosure. Disclosure-related interventions were described in case studies published in PM as early as 1939, and interest in these interventions has recently peaked with studies showing that they are effective in reducing disease activity and psychological distress in patients suffering from rheumatoid arthritis (172, 173).
Considered overall, the results of our review suggest that the face of pain research published in PM has changed considerably in the past 60 years. Given the ongoing commitment of psychosomatic researchers to this area, we expect this evolution to continue in the years to come.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication April 5, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONCASE STUDIES RELATED TO...NOTESACKNOWLEDGMENTSREFERENCES |
|---|
