Journal of Psychosomatic Research 59 (2005) 181 184

Use of depression rating scales in chronic fatigue syndrome

M. Henderson*, C. Tannock
Academic Department of Psychological Medicine, GKT School of Medicine and Institute of Psychiatry, Weston Education Centre, Cutcombe Road, London SE5 9RJ, United Kingdom Received 11 November 2003; accepted 20 April 2004

Abstract Objective: The aim of this study was to examine the performance of three commonly used depression rating scales in a hospital sample of patients with chronic fatigue syndrome (CFS). Methods: Sixty-one patients with CDC criteria for CFS completed the General Health Questionnaire (GHQ), the Hamilton Depression Scale (HAM-D) and the depression subscale of the Hospital Anxiety and Depression Scale (HADS-D). Current psychiatric status was assessed using the Structured Clinical Interview for DSM-III-R. Disorders: Patient version (SCID-P). Receiver operating curves were drawn for each of the depression rating scales. Results: Thirty-one percent of the patients were depressed according to the SCID-P. Using the standard cut-offs, both GHQ and HAM-D overestimated the number of depressed patients, whilst the HADS-D underestimated the number. The receiver operating curves suggest that the optimum cut-offs for GHQ, HAM-D and HADS-D in this population are 7/8, 13/14 and 8/9, respectively. Conclusions: Standard cutoffs may not be appropriate when using depression rating scales in CFS patients in a tertiary care setting. D 2005 Elsevier Inc. All rights reserved.

Keywords: Chronic fatigue syndrome; Depression; General Health Questionnaire; Hamilton Depression Scale; Hospital Anxiety and Depression Scale

Introduction Chronic fatigue syndrome (CFS) is a common disorder characterised by disabling medically unexplained fatigue lasting greater than 6 months and a number of both physical and psychological symptoms [1]. The role of psychiatry in CFS has caused much controversy; nonetheless, psychiatric diagnoses are common. Estimates of the prevalence of depression and anxiety, the most common psychiatric diagnoses, vary from 15% to 75% [2 6]. The recognition of psychiatric conditions, in particular, depression, in this group is important. Most patients with CFS will see either primary or secondary care physicians with little specialist training in psychiatric assessmentpatients in a CFS clinic may have a major depressive disorder (MDD) in the absence of operationally defined CFS, or comorbid with CFS. MDD in CFS is a marker for more persistent symptoms and failure to return to work [7].
* Corresponding author. Tel.: +44 20 7848 5278; fax: +44 20 7848 5408. E-mail address: m.henderson@iop.kcl.ac.uk (M. Henderson). 0022-3999/04/$ see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2004.04.374

Only three studies have examined the performance of depression rating scales in CFS. Buchwald et al. [2] compared the 28-item General Health Questionnaire (GHQ-28) with a structured clinical interview. High levels of psychiatric morbidity were found using standard cut-offs, but alternative cut-offs were not examined. Farmer et al. [8] assessed patients using the Beck Depression Inventory (BDI; [9]) and the 60-item GHQ (GHQ-60; [10]). These were compared with the gold standard of the Schedule for the Clinical Assessment of Neuropsychiatry (SCAN; [11]). Both the GHQ-60 and the BDI performed badly, although the study was limited by the fact that the rating scales and the gold standard were performed on different occasions. A more recent study [12] examined the Hospital Anxiety and Depression Scale (HADS-D; [13]) and the Medical Outcomes Study (MOS)mental health scale [14] in comparison with the gold standard of the Revised Clinical Interview Schedule (CIS-R; [15]). The MOS could not be recommended, as it yielded too many false positives, but the authors found that, if the cut-off for the depression scale of the HADS was lowered to 9/10, it was a valid screening tool in CFS patients.

182

M. Henderson, C. Tannock / Journal of Psychosomatic Research 59 (2005) 181 184

Table 1 Standard and alternative cut-offs for depression rating scales TEST SCID-P HAM-D GHQ-12 HADS-D Area under curve 0.988 0.972 0.957 Cut-off 11/12 (standard) 13/14 (best) 3/4 (standard) 7/8 (best) 10/11 (standard) 8/9 (best) Depression cases 19 35 24 32 19 5 12 (31%) (57%) (39%) (52%) (31%) (8%) (20%) Sensitivity (95% confidence interval) Gold standard 1.0 (0.831.0) 0.89 (0.69 0.97) 0.89 (0.69 0.97) 0.79 (0.57 0.91) 0.26 (0.12 0.49) 0.53 (0.32 0.73) Specificity (95% confidence interval) 0.62 (0.47 0.75) 0.83 (0.69 0.92) 0.64 (0.49 0.77) 0.9 (0.78 0.96) 1.0 (0.921.0) 0.95 (0.84 0.99) PPV 0.54 0.71 0.53 0.79 1 0.83 NPV 1 0.95 0.93 0.9 0.72 0.82

PPV positive predictive value. NPV negative predictive value.

As part of a larger study, we examined the properties of three different commonly used depression rating scales in comparison with a structured clinical interview in a population of CFS patients.

Method The recruitment and selection of patients is described in more detail elsewhere [16,17]. Patients were recruited as part of a larger study from those attending a teaching hospital department of either immunology or psychiatry for the assessment of their chronic fatigue. All patients met CDC criteria for CFS [1], hence, recognised medical causes for their fatigue had been excluded. Axis I disorders, including depression, were ascertained by use of the Structured Clinical Interview for DSM-III-R patient edition (SCID-P; [18]). This has similar reliability to other structured interviews [19] and is widely used as a dGold StandardT when investigating the use of screening instruments, not the least in patients with physical symptoms [20,21]. All participants were seen by one of the authors

(CT), and informed consent obtained. Patients were assessed using the Hamilton Depression Rating Scale (HAM-D; [22]) and completed the 12-item version of the GHQ (GHQ-12; [10]) and the depression scale of the HADS-D [13]. The study was approved by the Local Research Ethics Committee. Statistical analysis was performed using the Statistical Package for the Social Sciences SPSS 10.0. Receiver operating characteristics (ROC) curves were drawn, by plotting the sensitivity against 1 specificity for each score on each scale.

Results Sixty-one patients were studied. The mean age was 41 years (S.D. =11.5 years); 43 (70%) were female, 29 (48%) married or cohabiting, and 57 (93%) were white. The mean illness duration was 7.8 years (95% CI = 5.8 9.8); 43 (70%) were members of a self-help group, and 21 (34%) reported a past psychiatric history. Using the SCID-P, 19 (31%) were categorised as suffering from MDD.

Fig. 1. ROC curves for depression rating scales.

M. Henderson, C. Tannock / Journal of Psychosomatic Research 59 (2005) 181 184

183

Table 1 shows the performance of each of the depression rating scales in this group, using the standard cut-off for case/noncase. Using the SCID-P as the gold standard, ROC curves were plotted for each scale and are shown in Fig. 1. For each scale, points of maximum change were noted, and the performance of each scale was then assessed using this as a new cut-off. These are also shown in Table 1. The proportion of CFS cases diagnosed as depressed by the SCID-P is in keeping with much of the literature on depression in CFS. The GHQ and the HAM-D both scored many more cases of depression than were diagnosed using the SCID-P, whilst the HADS-D scored many fewer. It is possible that the GHQ and the HAM-D attributed somatic symptoms to depression whilst the HADS-D, specifically designed for the physically ill, scored in the opposite way. In previous ROC analyses, it has been suggested that an area under the curve (AUC) of greater than 0.8 is a good indicator that a screening tool is valid and efficient in a particular population. All the scales studied here scored in excess of this suggesting that, if the alternative cut-offs are used, they are valid tools in this population. From these data, it is suggested that in a CFS population, a more valid cut-off for the GHQ is 7/8, the HADS-D is 8/9 and the HAM-D is 13/14.

patients responses to the depression questionnaires may have been affected by having answered questions in the earlier structured interview. Our sample was drawn from both psychiatric and tertiary medical settings. Such patients are not necessarily representative of the wider CFS population. Although we think it unlikely, our population may have high levels of depression due to bias on the part of referrers opting to refer to a dpsychiatricT CFS service. A similar study in primary care would almost certainly produce different results. Another arm of the study involved MRI scans of the patient, and it is possible that this part of the study attracted a skewed population of patients. Our numbers are smaller than those in Morrisss group; our ability to select alternative cut-off points would undoubtedly have been improved by larger numbers.

References
[1] Fukuda KF, Straus SE, Hickie IF, Sharpe MC, Dobbins JG, Komaroff A. The chronic fatigue syndrome: a comprehensive approach to its definition and study. International chronic fatigue syndrome study group. Ann Intern Med 1994;121:953 9. [2] Buchwald DF, Pearlman TF, Kith PF, Katon WF, Schmaling K. Screening for psychiatric disorders in chronic fatigue and chronic fatigue syndrome. J Psychosom Res 1997;42:87 94. [3] Farmer AF, Jones IF, Hillier JF, Llewelyn MF, Borysiewicz LF, Smith A. Neuraesthenia revisited: ICD-10 and DSM-III-R psychiatric syndromes in chronic fatigue patients and comparison subjects. Br J Psychiatry 1995;167:503 6. [4] Hickie IF, Lloyd AF, Wakefield DF, Parker G. The psychiatric status of patients with the chronic fatigue syndrome. Br J Psychiatry 1990; 156:534 40. [5] Taerk GS, Toner BB, Salit IE, Garfinkel PE, Ozersky S. Depression in patients with neuromyasthenia (benign myalgic encephalomyelitis). Int J Psychiatry Med 1987;17:49 56. [6] Wessely S, Chalder T, Hirsch S, Wallace P, Wright D. Psychological symptoms, somatic symptoms, and psychiatric disorder in chronic fatigue and chronic fatigue syndrome: a prospective study in the primary care setting. Am J Psychiatry 1996;153:1050 9. [7] Bombardier CH, Buchwald D. Outcome and prognosis of patients with chronic fatigue vs chronic fatigue syndrome. Arch Intern Med 1995;155:2105 10. [8] Farmer AF, Chubb HF, Jones IF, Hillier JF, Smith AF, Borysiewicz L. Screening for psychiatric morbidity in subjects presenting with chronic fatigue syndrome. Br J Psychiatry 1996;168:354 8. [9] Beck A. Depression: clinical, experimental, and theoretical aspects. New York7 Hoeber Medical Division, Harper and Row, 1967. [10] Goldberg D. The detection of psychiatric illness by questionnaire. London7 Oxford Univ. Press, 1972. [11] Wing J, Babor T, Burke J, Cooper J, Giel R, Jablenski A, Sartorius N. SCAN: schedules for clinical assessment in neuropsychiatry. Arch Gen Psychiatry 1990;47:589 93. [12] Morriss R, Wearden A. Screening instruments for psychiatric morbidity in chronic fatigue syndrome. J R Soc Med 1998;91: 365 8. [13] Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983;67:361 70. [14] Stewart AF, Hays RD, Ware JE. The MOS Short-Form General Health Survey. Reliability and validity in a patient population. Med Care 1988;26:724 35. [15] Lewis GF, Pelosi AJ, Araya RF, Dunn G. Measuring psychiatric disorder in the community: a standardized assessment for use by lay interviewers. Psychol Med 1992;22:465 86.

Discussion MDD was common in this tertiary care group of CFS patients; so common that clinical assessment of these patients should routinely include detailed evaluation of mood. In determining the most discriminating level of dcasenessT in this population, our findings suggest that the widely used cut-off scores are set too low, at least for GHQ12 and HAM-D. With the use of any clinical tool, there is always a playoff between sensitivity and specificity: Is it worse to miss a dcaseT or to overmanage a dnoncaseT? We would argue that the evidence of significantly worse outcomes for CFS patients who are depressed, in addition to the high levels of depression found, means that identifying cases of depression is important, although we accept that this argument is tempered by the limited evidence of effectiveness for interventions to treat depression in CFS. We are unable to directly compare our results with those of Farmer et al. [8] who used a different version of the GHQ. Our results on the HADS-D, however, are similar to those of Morriss and Wearden [12], who found the optimum cut-off in their study to be 9/10. We are not aware of studies assessing the performance of the HAM-D or the GHQ-12 in CFS. Our study does have a number of limitations. Whilst our study has the advantage over Farmers in that all measurements were done at the same sitting, it is possible that

184

M. Henderson, C. Tannock / Journal of Psychosomatic Research 59 (2005) 181 184 [20] Leentjens AF, Verhey FR, Luijckx GJ, Troost J. The validity of the Beck Depression Inventory as a screening and diagnostic instrument for depression in patients with Parkinsons disease. Mov Disord 2000;15:1221 4. [21] Sharpe MF, Strong VF, Allen KF, Rush RF, Postma KF, Tulloh AF, Maguire PF, House AF, Ramirez AF, Cull A. Major depression in outpatients attending a regional cancer centre: screening and unmet treatment needs. Br J Cancer 2004;90:314 20. [22] Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960;23:56 62.

[16] Costa DC, Tannock CF, Brostoff J. Brainstem perfusion is impaired in chronic fatigue syndrome. QJM 1995;88:767 73. [17] Greco AF, Tannock CF, Brostoff JF, Costa DC. Brain MR in chronic fatigue syndrome. Am J Neuroradiol 1997;18:1265 9. [18] Spitzer RL, Williams JBW, Gibbon M, First MB. Structured clinical interview for DSM-III-R patient edition (SCID-P, version 1.0). Washington7 American Psychiatric Press, 1990. [19] Williams JB, Gibbon MF, First MB, Spitzer RL, Davies MF, Borus JF, Howes MJ, Kane JF, Pope HG, Rounsaville B. The structured clinical interview for DSM-III-R (SCID): II. Multisite testretest reliability. Arch Gen Psychiatry 1992;49:630 6.