Rief, Winfried PhD; Martin, Alexandra PhD; Klaiberg, Antje PhD; Brähler, Elmar PhD
Illness behavior describes the various aspects of how patients cope with their illness. The term illness behavior has originally been introduced from Mechanic (1) who also emphasized socioeconomic issues. Illness behavior covers features such as health care use, urging doctors to do investigations, taking medication, being disabled at work, avoidance of physical activity, and expression of symptoms to family members and significant others. Pilowsky (2,3) introduced the term “abnormal illness behavior” to summarize behavioral aspects that might contribute to the maintenance of the disorder. Interestingly, illness behavior shows only moderate associations with illness severity. This implies that people with the same illness show very different illness behavior, and it highlights the fact that individual and social factors determine a major part of illness behavior.
Anxiety and depression are important features associated with illness behavior. Not only in itself, but also in combination with medical complaints, depression and anxiety are associated with higher health care costs (4–8). Another important determinant of illness behavior is somatic complaints. The most frequent somatic complaints are rarely explained by organic diseases but have to be considered as “unexplained” or “somatoform” symptoms (9). Patients with somatization syndrome (multiple somatic complaints without a medical condition explaining the symptoms) constitute a major part of patients with extraordinary health care use (10,11). Patients with somatic complaints in combination with organic illness attributions tend to increase health care use (12,13). As somatic complaints are an extremely frequent phenomenon (14), the health care relevance of this syndrome is substantial.
To understand illness behavior, it should be kept in mind that not only health care use but also other aspects are relevant features contributing to the ways people cope with disorders. A recently published analysis of illness behavior (15) demonstrated that different aspects of illness behavior share only medium interrelationships around 0.31. In that study, we developed a scale assessing the need for the verification of diagnosis, the expression of symptoms, the need for medication and treatment, aspects of illness consequences such as sick leave from work, reduction of social activities, and physical deconditioning, as well as the factor “body scanning” (attention focusing to bodily processes). The medium intercorrelation of these aspects highlights the fact that specific features of illness behavior can be very individual, and illness behavior is not a homogenous construct. Analysis of the association of psychopathological features with these different aspects of illness behavior is lacking.
Another shortcoming of the studies investigating associations between mental disorders and illness behavior so far is the focus on specific treatment settings. Most studies have investigated patients at specific institutions or primary care offices. This implies selection biases of the patient samples that could influence the results. Moreover, only patients already showing health care use per definition are included in these studies. Therefore, we wanted to investigate associations between illness behavior with features of mental disorders in a representative sample of the general population.
A representative sample of the general population of Germany was selected with the assistance of a demography consulting company (USUMA, Berlin, Germany). The area of Germany was separated into 306 sample areas representing the different structures of the country. After selection of a sample area, households of the respective area were selected per chance; finally members of this household fulfilling the inclusion criteria were again selected by chance. Inclusion criteria were German as a native language and age above 13. The sample was aimed to be representative in terms of age, gender, and education. A first attempt was made for 3855 addresses following a random-root procedure. All subjects were visited by an interviewer, informed about the investigation, and self-rating questionnaires were presented. The assistant waited until participants answered all questionnaires and offered help if patients didn’t understand the meaning of questions, although all subjects were able to read. From the 3855 selected addresses, 3776 were valid. If not at home, a maximum of three attempts was made to contact the selected person. A total of 2507 people between the ages of 14 and 92 years agreed to participate (participation rate, 66.4% of valid addresses). Mean age was 48.8 (SD 17.9); 55% were female, 52% were married, 9% divorced, 13% widowed; and 51% had higher education.
To assess mental disorders, we used the complete German version of the Patient Health Questionnaire (PHQ) (16). The PHQ has been developed from the interview guideline used in the PRIME-MD studies (17). As an instrument for psychiatric case definition, it demonstrated good validity and reliability (18) and showed the best characteristics in comparison with the Hospital Anxiety and Depression Scale HADS or the World Health Organization Well-Being Index (WBI-5). Its sensitivity was 98% and specificity 80% compared with structured clinical interviews (19,20). Nine items of the PHQ assess depressive symptoms; the diagnosis of major depression is indicated if 5 or more symptoms are answered as “more than half of the day” including at least item 1 and item 2 (item 1: little interest or pleasure in doing things; item 2: feeling down, depressed, and hopelessness). If less than 5 items are answered with “more than half of the day,” but at least 2 depression items including item 1 or 2, criteria for a depressive syndrome are fulfilled. Thirteen items ask for common somatic symptoms. If at least 3 of these items are answered with “strong disabling,” criteria of a somatic syndrome are fulfilled. The criterion of ≥3 symptoms has been confirmed in a study by Löwe et al. (21). As these items cover somatic symptoms that only rarely have clear organic explanations (e.g., dizziness, fatigue, back pain, sleeping troubles), we will also use the term somatization. Further validation data for the 15 somatic symptom items were analyzed in a data set of more than 6000 patients (22). Five items assess aspects of panic disorder; these items are dichotomous and all have to be fulfilled for the definition of panic. Again the case definition of panic disorder according to the PHQ has been validated (23). Normative data for PHQ depression and panic items are published elsewhere (24).
The Scale for the Assessment of Illness Behavior (SAIB) (15) has 25 items assessing aspects of “verification of diagnosis” (5 items; e.g., “I have a specialist confirm a diagnosis given by my general practitioner”), “expression of symptoms” (6 items; e.g., “I often try to explain my current state of health to other people”), “medication/treatment” (5 items; e.g., “I always have the most important medicines at home”), “consequences of illness” (5 items; e.g., “I’m not able to concentrate on my work when suffering from physical complaints”), and “body scanning” (4 items; e.g., “I pay a lot of attention to the different processes going on within my body”). In an earlier study, the results of the SAIB were validated with doctor ratings using an inpatient sample (15). It could be confirmed that “SAIB diagnosis verification” is associated with doctor ratings on “Patient seeks medical information,” “SAIB medication/treatment” correlates with doctor ratings on “uncritical attitudes toward medication,” “SAIB illness consequences” is associated with doctors rating on “avoidance of physical demands” and “complaining,” whereas “SAIB scanning” has associations with doctor ratings on “Patient focuses attention on symptoms.” Items of the SAIB are categorized from 1 (“completely true”) to 4 (“not at all”). Two of the 25 items are inversely coded. For all dimensions, higher scores indicate lower illness behavior.
Subjects were also asked to rate how frequently they visited general practitioners, medical specialists, psychiatrists, psychotherapists/psychologists during the last 12 months. Moreover, they had to indicate how many days they were on sickness leave and how many days of inpatient treatment they had during the last 12 months.
As a preanalysis, associations between illness behavior (SAIB) and health care use were computed to determine whether these constructs are sufficiently independent. The first step of the major analyses was to compare the five groups on aspects of illness behavior using one-way analysis of variances and subsequent pair-wise comparisons. As the PHQ also allows the constitution of continuous variables for depression (sum of depression items 1–9) and for somatic complaints (sum of somatic items 1–13), the impact of somatization and depression on aspects of illness behavior can also be analyzed using linear regression analysis. As some aspects of illness behavior might be associated with age and sex, linear regression analysis on the criterion “illness behavior” were computed with predictor variables age, sex, somatic syndrome, and depression. Both analysis procedures (group comparison as well as the regression approach) were repeated for health care use variables, such as the number of doctor visits in primary care. To get a quick overview on health care use, doctor visits were also dichotomized for the major categories general practitioners (GPs), medical specialists, psychiatrists, and psychologist/psychotherapist into the two categories “visited yes/no during the last 12 months.” For these variables, χ2 tests were used.
Before analyzing group-specific data, we wanted to examine whether illness behavior (as assessed with the SAIB) and health care use cover identical information. Therefore, we computed regression analyses with the 5 SAIB scores as predictor variables and the health care use variables as criterion. All corrected R2 were smaller than 0.10 indicating that the SAIB variables and health care use have some associations but do not overlap substantially. Some significant associations between illness behavior (SAIB) and health care use were the following: doctor visits (nonpsychiatrically specialized) were best predicted from SAIB “need for medication/treatment” (β = −0.23; p < .001); inpatient treatment days and working disability days were best predicted from SAIB “illness consequences” (β = −0.08; p < .01 and β = −0.07; p < .01, respectively). To summarize, these results consider illness behavior and health care use as different (although in some parts associated) constructs.
Aspects of Illness Behavior (SAIB) in Depression, Somatic Syndrome, and Panic Disorder
According to the case definition of the PHQ, 77 people fulfilled the criteria of a depressive syndrome (without panic or somatic syndrome), 24 fulfilled the criteria of major depression (without somatic or panic syndrome), 102 fulfilled the criteria of somatic syndrome (including 15 subjects with major depression), and 30 fulfilled the criteria of panic (8 with co-morbid major depression).
Table 1, A and B presents means and standard deviations of the five groups for the variables of illness behavior as well as for the numerical scores for somatization and depression. For all variables, the F scores indicate significant group differences. People with panic syndrome have increased illness behavior compared with controls on all variables of the SAIB, whereas people with depressive syndromes have increased illness behavior for SAIB 2 (expression of symptoms) and SAIB 4 (illness consequences). For subjects with somatic syndrome, significant differences to controls were found for all variables except SAIB 1 (verification of diagnosis). Significant group differences between clinical groups were found for the SAIB variable 3: the need for medication/treatment is higher in panic subjects than in subjects with depressive syndrome. Analysis of the variables somatization and depression of the PHQ confirm the validity of the PHQ.
Regression Coefficients for Different Aspects of Illness Behavior
Table 2 shows that increased age is associated with increased illness behavior. Gender only contributes significantly to SAIB 3 (medication/treatment) with higher scores for women than for men. Even after controlling for other variables, depression has a substantial impact on the SAIB 4 scale (consequences of illness), but also on SAIB 2 (expression of symptoms), and the SAIB total score. Somatization has significant β weights of above 0.10 with the SAIB 3 scale (medication/treatment) and the SAIB 5 scale (scanning), as well as the SAIB total score. In sum, regression analysis confirms that somatization and depression have different impacts on aspects of illness behavior.
Use of Health Care Services (Dichotomous Variable)
As an initial overview of health care use, doctor visits were treated as a dichotomous variable (yes/no during last 12 months). Table 3 reflects the percentages of the subgroup reporting doctor visits of different specialists. For all health care providers, health care use was highest for panic patients. Patients with major depression and depressive syndrome do not differ substantially, both showing increased rates for GP and medical specialist visits. Around 90% of people with depression do not report any psychiatrist visits. Psychiatrists are consulted more by patients with somatic syndrome and panic disorder than by patients with depression. The majority of people from all 4 clinical syndromes do not seek help in offices of psychiatrists, psychologists, or psychotherapists. Detailed results of health care use are presented in Table 4A and B.
Use of Health Care Services and Social Support
Doctor visits were added to make one variable indicating the sum of visits to general practitioner offices and medical specialist clinics, the other indicating visits in psychiatrist and psychotherapist offices (see last 2 rows of Table 4, A and B). Although one-way analysis of variances indicate substantial group differences for all aspects of health care use, again it is obvious that only a small number of people with mental problems seek help in psychiatrists’ or psychotherapists’ offices. Pair-wise comparisons with controls revealed significant differences only for people with panic syndrome. However, all clinical syndromes are associated with significantly increased health care use in primary care. People with depressive syndrome have 2.1-fold the number of doctor visits to GPs and medical specialists compared with controls, people with major depression 2.2-fold, people with somatoform syndrome 2.1-fold, and people with panic syndrome 2.9-fold. Patients with panic disorder have higher rates than depressives for GP- and medical specialist-visits, as well as for inpatient treatment days. Inpatient treatment days are also higher in patients with somatic syndrome than in depressives.
Regression Coefficients for Health Care Use
Somatization and depression as continuous variables have independent impacts on the number of doctor visits, as regression coefficients confirm (see Table 5). For somatization, however, this specific effect is only demonstrated for medical doctors. In contrast, depression is also associated with increased numbers of visits to psychiatric or psychologist offices as well as with an increased number of inpatient treatment days.
This study demonstrated that depression, somatic syndromes, and panic disorder each contribute independently to aspects of illness behavior. All these clinical groups differ from healthy controls. Patients with panic syndrome showed the most abnormal scores, especially on the SAIB-factor medication/treatment and the total score (although this effect is partly attributable to comorbidity in panic patients). Regression analysis revealed the importance of depression for the ratings of illness consequences and illness expression, whereas somatic syndromes have substantial impact on medication/treatment as well as on body scanning. This implies that the associations of depression and somatization with illness behavior are specific. As these results are from a representative sample, to our knowledge the independence of these factors has been demonstrated in a population without selection biases for the first time.
All mental disorders are associated with an increase of health care use. However, this is mainly attributable to general practitioner and medical specialists visits. Whereas the number of psychiatry and psychotherapy visits is also increased, the increase in visits to nonpsychosocial medical specialists is higher than the increase in psychosocial specialist visits in the clinical groups. The regression analysis reveals that somatization and depression have independent impacts on the number of doctor visits.
Specificity has to be discussed considering possible comorbidity in the clinical groups. The depression group did not have psychiatric comorbidity, and in the somatization group around 80% had neither depression nor panic according to our assessments. However, care must be taken in interpreting the results for the panic group, as only 11 of the 30 panic patients had “pure” panic. Focusing on the impact of the categories of somatization and depression on health care use, patients with somatic complaints consult medical specialists (organic specialization) more frequently, whereas major depressives show highest mean values for sick leave. Patients with somatic problems also show the highest abnormal rates for medication/treatment and for body scanning, whereas depressives show more pronounced values for illness consequences in the SAIB. Although these patterns can be found in the means of the groups, only in regression analyses are they statistically significant.
Henningsen et al. (25) analyzed whether functional somatic syndromes are fully dependent on depression or anxiety. In their meta-analytical review, they found that the three syndromes are related to one another, but there is also a relative independency between depression, anxiety, and somatic syndromes. This is in line with our results showing that depression, anxiety, and somatic syndromes have independent impacts on health behavior.
The association between psychiatric status and health care use has been confirmed in other studies (e.g., 5,12,26,27). With our representative sample of more than 2500 people, we replicated results from Hansen et al. (6) who assessed 157 patients with psychiatric diagnoses. They found adjusted odds ratios of 3.6 for the impact of mental disorders on being a high user of health services. Whereas they did not differentiate between anxiety and depression, the odds ratios for somatoform disorders were even higher in their study.
In this article, we focused on illness behavior and health care use. A next step would be to examine mediating factors between mental illnesses and illness behavior. It seems obvious that cognitive factors such as causal illness attributions, catastrophizing interpretation of symptoms (28), or “attention focusing” might play a major role in the association between psychiatric disorders and health care use. Cheng (4) emphasized that the perceptual style, coping behavior, and psychological symptoms might be pivotal features of health care use. In some cases, childhood sexual abuse might increase the risk for mental disorders and medical use (29). It is also evident that health care use is directly related to costs. Simon et al. (8) found that the costs for depressed primary care patients are nearly twice as high as the costs for other primary care patients (see also refs. 7,30). Therefore, the investigation of determinants of illness behavior is crucial especially in a period of limited financial resources.
Although the strength of our study may be the general population approach, we have also to consider some shortcomings. Cases were identified using a self-rating scale (PHQ). Whereas this seems currently to be the best self-rating scale for case definition, structured clinical interviews can still be considered to be the gold standard for classification. However, some of our results confirm the validity of this approach. Moreover, the prevalence rates of current mental disorders are in the same range as our data and confirm that the PHQ does not overestimate psychological symptoms. Furthermore, our results also emphasize the consideration of sub-threshold depressive syndromes. In fact, we did not find differences in illness behavior or health care use between patients with major depression and patients with depressive syndrome, although both of them differ from healthy controls.
Self-report data might especially be less valid for the report of health care use. Compared with interview data, it can be suggested that health care use is underreported in self-reports, as compared with structured interview data or nation-wide health use registers. Because psychiatric and psychotherapeutic treatment is still stigmatized, it might be that people underreport health care use particularly in these domains. Although this weakness has to be considered, we assume that our self-report data are a linear transformation of the real health-care use; thus the figures for health care use might be higher in reality than presented in this paper, but the found association (e.g., regression analysis) would be the same if subjects underestimate the health care use consistently.
A further weakness is the definition of the panic group, which was too small to exclude patients with comorbid problems. Therefore the data for this group should not be overgeneralized, but might be substantially influenced from other disorders.
Another shortcoming could be the use of parametrical statistical approaches, although some variables might not fulfill the distribution requirements (e.g., number of doctor visits). In big samples, however, violations of the normal distribution assumption led to less important distortions in data analysis than in small samples. Moreover, statistical approaches for parametrical data allow more sophisticated multivariate analysis. Future studies should not only consider these shortcomings but should also focus on mediating variables between mental disorders with illness behavior.
1. Mechanic D. Social psychological factors affecting the presentation of bodily complaints. New Engl J Med 1972;286:1132–9.
2. Pilowsky I. Abnormal illness behaviour. Br J Med Psychol 1969;42:347–51.
3. Pilowsky I. Abnormal illness behaviour. Chichester, Wiley & Sons; 1997.
4. Cheng C. Seeking medical consultation: perceptual and behavioral characteristics distinguishing consulters and nonconsulters with functional dyspepsia. Psychosom Med 2000;62:844–52.
5. Barsky AJ, Delamater BA, Orav JE. Panic disorder patients and their medical care. Psychosomatics 1999;40:50–6.
6. Hansen MS, Fink P, Frydenberg M, Oxhoj M-L. Use of health services, mental illness, and self-rated disability and health in medical inpatients. Psychosom Med 2002;64:668–75.
7. Simon GE, von Korff M. Recognition, management, and outcomes of depression in primary care. Arch Fam Med 1995;4:99–105.
8. Simon G, Ormel J, VonKorff M, Barlow W. Health care costs associated with depressive and anxiety disorders in primary care. Am J Psychiatry 1995;152:352–7.
9. Kroenke K, Mangelsdorff D. Common symptoms in ambulatory care: incidence, evaluation, therapy and outcome. Am J Med 1989;86:262–6.
10. Fink P. The use of hospitalizations by persistent somatizing patients. Psychol Med 1992;22:173–80.
11. Fink P. Surgery and medical treatment in persistent somatizing patients. J Psychosom Res 1992;36:439–47.
12. Kolk AMM, Hanewald GJFP, Schagen S, van Wijk CMTG. Predicting medically unexplained physical symptoms and health care untilization. A symptom-perception approach. J Psychosom Res 2002;52:35–44.
13. Rief W, Nanke A, Bender A, Zech T, Emmerich J. Causal illness attributions in somatoform disorders: Associations with comorbidity and illness behaviour. J Psychosom Res 2004;57:367–71.
14. Rief W, Hessel A, Braehler E. Somatization symptoms and hypochondriacal features in the general population. Psychosom Med 2001;63:595–602.
15. Rief W, Ihle D, Pilger F. A new approach to assess illness behaviour. J Psychosom Res 2003;54:405–14.
16. Löwe B, Spitzer RL, Zipfel S, Herzog W. PHQ-D. Gesundheitsfragebogen fuer Patienten [German Version of the Patient Health Questionnaire]. Karlsruhe, Pfizer; 2002.
17. Spitzer RL, Williams JBW, Kroenke K, Linzer M, deGruy FV, Hahn SR, Brody D, Johnson JG. Utility of a new procedure for diagnosing mental disorders in primary care. The PRIME-MD 1000 study. JAMA 1994;272:1749–56.
18. Spitzer RL, Kroenke K, Williams JB. Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. JAMA 1999;282:1737–44.
19. Löwe B, Spitzer RL, Gräfe K, Kroenke K, Quenter A, Zipfel S, Buchholz C, Witte S, Herzog W. Comparative validity of three screening questionnaires for DSM-IV depressive disorders and physicians’ diagnoses. J Affect Disord 2004;78:131–40.
20. Henkel V, Mergl R, Kohnen R, Maier W, Möller H-J, Hegerl U. Identifying depression in primary care: a comparison of different methods in a prospective cohort study. BMJ 2003;326:200–1.
21. Löwe B, Graefe K, Kroenke K, Zipfel S, Quenter A, Wild B, Fiehn C, Herzog W. Predictors of psychiatric co-morbidity in medical outpatients. Psychosom Med 2003;65:764–70.
22. Kroenke K, Spitzer RL, Williams JBW. The PHQ-15: validity of a new measure for evaluating the severity of somatic symptoms. Psychosom Med 2002;64:258–66.
23. Löwe B, Graefe K, Zipfel S, Spitzer RL, Hermann-Lingen C, Witte S, Herzog W. Detecting panic disorder in medical and psychosomatic outpatients comparative validation of the hospital anxiety and depression scale, the patient health questionnaire, a screening question, and physicians’ diagnosis. J Psychosom Res 2003;55:515–9.
24. Rief W, Nanke A, Klaiberg A, Braehler E. Base rates for panic disorder and major depression according to the brief Patient Health Questionnaire (Brief PHQ): a population based study. J Affect Disord 2004;82:271–6.
25. Henningsen P, Zimmermann T, Sattel H. Medically unexplained physical symptoms, anxiety, and depression: a meta-analytical review. Psychosom Med 2003;65:528–33.
26. Williams ERL, Guthrie E, Mackway-Jones K, James M, Tomenson B, Eastham J, McNally D. Psychiatric status, somatisation, and health care utilization of frequent attenders at the emergency department. A comparison with routine attenders. J Psychosom Res 2001;50:161–7.
27. Jyväsjärvi S, Joukamaa M, Väisänen E, Larivaara P, Kivelä S-L, Keinänen-Kiukaanniemi S. Somatizing frequent attenders in primary health care. J Psychosom Res 2001;50:185–92.
28. Rief W, Hiller W, Margraf J. Cognitive aspects in hypochondriasis and the somatization syndrome. J Abnorm Psychol 1998;107:587–95.
29. Newman MG, Clayton L, Zuellig A, Cashman L, Arnow B, Dea R, Taylor CB. The relationship of childhood sexual abuse and depression with somatic symptoms and medical utilization. Psychol Med 2000;30:1063–77.
30. Simon GE, VonKorff M, Barlow W. Health care costs of primary care patients with recognized depression. Arch Gen Psychiatry 1995;52:850–6.