Hospital Anxiety and Depression Scale (HAD): some psychometric data for a Swedish sample

Transcription

1 Acta Psychiatr Scand 1997: 96: Printed in UK - all rights reserved Hospital Anxiety and Depression Scale (HAD): some psychometric data for a Swedish sample Copyright 0 Munksgaard I997 ACTA PSYCH I ATRI CA SCANDINAVICA ISSN X Lisspers J, Nygren A, Soderman E. Hospital Anxiety and Depression Scale (HAD): some psychometric data for a Swedish sample. Acta Psychiatr Scand 1997: 96: Munksgaard The Hospital Anxiety and Depression Scale (HAD) was evaluated in a Swedish population sample. The purpose of the study was to compare the HAD with the Beck Depression Inventory (BDI) and Spielberger s State Trait Anxiety Inventory (STAI). A secondary aim was to examine the factor structure of the HAD. The results indicated that the factor structure was quite strong, consistently showing two factors in the whole sample as well as in different subsamples. The correlations between the total HAD scale and BDI and STAI, respectively, were stronger than those obtained using the different subscales of the HAD (the anxiety and depression subscales). As expected, there was also a stronger correlation between the HAD and the non-physical items of the BDI. It was somewhat surprising that the factor analyses were consistently extracting two factors, depression and anxiety, while on the other hand both BDI and STAI tended to correlate more strongly with the total HAD score than with the specific depression and anxiety HAD subscales. Nevertheless, the HAD appeared to be (as was indeed originally intended) a useful clinical indicator of the possibility of depression and clinical anxiety. J. Lisspers, A. Nygren, E. Soderman3 MidSweden University at Ostersund and Institute for Future Studies, Stockholm, Section of Personal Injury Prevention, Karolinska Institute, Stockholm and Institute for Future Studies, Stockholm and %partment of Clinical Neuroscience, Karolinska Institute. Stockholm and MidSweden University at Ostersund. Ostersund. Sweden Key words: anxiety; depression; factor analysis Eva Soderman. Department of Human Resources, Management and Environment, MidSweden University at Ostersund. S Ostersund. Sweden Accepted for publication January 11, 1997 Introduction The Hospital Anxiety and Depression scale (HAD) was developed by Zigmond & Snaith (1). It is a brief questionnaire (containing 14 items), and was originally designed to detect emotional disturbances in non-psychiatric patients treated at hospital clinics. The scale measures both anxiety and depression on two separate subscales, each containing 7 items. The HAD has been used to detect anxiety and depression in psychiatric patients (2-5), as well as in medical patients (6-13). It is of great importance to diagnose anxiety and depression in medical patients, and the HAD selfassessment scale is intended to help clinicians to obtain an early indication of possible psychiatric disorder or symptoms which might interfere with the treatment. The HAD has been reported to be an effective screening tool in medical settings (6, 14). Items referring to physical symptoms (e.g. headaches or dizziness) have been eliminated, since such reactions may also be attributed to the medical disease and treatment itself. Somatic symptoms of depression have been indicated as confounding factors in patients with medical illnesses (15, 16). Most of the items of the HAD concerning depression measure cognitive and emotional aspects, such as anhedonia - which is both a core component of depression and one that is unlikely to be affected by the presence of physical illness (18). The anxiety items are also focused on the cognitive and emotional aspects of anxiety. The HAD scale appears to be a reliable and valid method for measuring emotional distress in medical patients (1, 3). Several studies have compared the HAD with the DSM-I11 criteria for anxiety and depression (18), using the Structured Clinical Interview for DSM-I11 (SCID). HAD has shown a high level of sensitivity and specificity in a study of patients in a genito-urinary clinic (14). Case definition by a score of 8 on either the anxiety or depression subscales produced optimal results, giving sensitivities of 82% and 70% and specificities of 94% and 68% for depressive and anxiety 281

2 Lisspers et al. disorder, respectively. Hamer et al. studied deliberate self-harm (DSH) patients and reported a sensitivity of 88% and a specificity of 78% using the cut-off score of 8 on the depression subscale (19). Wilkinson & Barczak (20), again using a cut-off score of 8, reported a specificity of 86% and a sensitivity of 90% on the full-scale HAD in a study of patients attending their general practitioner. Razavi et al. (6) have reported a lower sensitivity and specificity on each subscale, especially the anxiety subscale, compared to the fullscale HAD. To our knowledge, normal population data on the HAD scale have so far only been published in one study (12), in which a relatively small sample was used. The aim of the present study was to study the HAD scale in a larger sample of the Swedish general population, and to test the validity of the HAD by comparing it with other generally accepted measures of depression (the BDI) and anxiety (the STAI). A secondary aim was to analyse the factor structure of the HAD in a Swedish population sample using the Swedish version of the HAD. Material and methods Subjects and procedure A set of questionnaires measuring anxiety and depression was administered to 1300 randomly selected subjects between 30 and 59 years of age in the county of Jamtland, Sweden. The subsamples consisted of equal percentages of men and women in three different age groups, namely 30-39,40-49 and years. The questionnaires were mailed to the subjects in February 1993, and 1 week later a reminder was mailed to all subjects. The questionnaires were returned anonymously by 642 participants. In total, 18 questionnaires were incompletely filled out and were therefore excluded, leaving a final sample of 624 subjects included in the study. score, and by summing the ratings for the 7 items of each subscale to yield separate scores for anxiety and depression. The Beck Depression Inventory (BDI) is one of the most widely used psychiatric rating scales for depression. The original version of the BDI was used, which is a 21-item self-reported inventory designed to measure the severity of depression. Subjects respond to the scale by rating each symptom item with a score ranging from 0 (absent) to 3 (severe or persistent presence of the symptom). The BDI is scored by summing the ratings for the 21 items to yield a total score that can range from 0 to 63. In different studies the scale has been divided into cognitive/affective and somatic subscales. Two different sets of such subscales analysed in an earlier study (23) were also used here. Spielberger s State and Trait Anxiety Inventory (STAI) is a well-known questionnaire designed to evaluate anxiety. The original version (Form X) was used. The STAI consists of two self-report rating scales (State Anxiety and Trait Anxiety), each of which contains 20 items on a 4-point Likert scale. The State Anxiety scale (STAI-S) asks respondents to describe how they feel right now, at this moment and the Trait Anxiety scale (STAI-T) asks them to state how they generally feel. The STAI-S and STAI-T are each scored by summing the ratings for the 20 items to yield a total score that can range from 20 to 80. Results The demographic data (sex, age group and place of residence) for the participants are listed in Table 1. Chi-square tests indicated that there were no significant differences between the age groups or urban and rural locations in terms of gender distribution. Zigmond & Snaith (1) have suggested two cutoff scores for detecting anxiety and depression, Table 1. Demographic data for the study participants Instruments The set of questionnaires contained the Swedish versions of the Hospital Anxiety and Depression Scale (HAD), the Beck Depression Inventory (BDI) (21) and Spielberger s State Trait Anxiety Inventory (STAI) (22). The Hospital Anxiety and Depression Scale is a self-report rating scale designed to measure both anxiety and depression. It consists of two subscales, each containing seven items on a 4-point Likert scale (ranging from 0-3). The HAD is scored by summing the ratings for the 14 items to yield a total 282 Sex Female Male Age group (years) Place of residence Urban Rural Mean age, 44 years. range, years. n %

3 Hospital Anxiety and Depression Scale namely 8 for doubtful cases and 11 for cases. The prevalence of doubtful cases of anxiety in this sample was 12% and that of depression was 9%. The prevalence of cases of anxiety was 8% and that of depression was 6%. The response rate was quite low (48%), so the prevalence data for anxiety and depression must be interpreted with caution with regard to generalizing to the actual prevalence of anxiety and depression in the population. The mean score on the total HAD scale was 8.53 (range 0-42), on the anxiety subscale (HAD-A) was 4.55 (range 0-21) and on the depression subscale (HAD-D) was 3.98 (range 0-21). Cronbach s alpha was 0.84 for HAD-A, 0.82 for HAD-D and 0.90 for the total HAD scale. The corresponding alpha values for BDI, STAI-S and STAI-T were 0.87, 0.95 and 0.91, respectively. A two-way ANOVA was performed, which showed no significant differences in either gender or age groups with regard to total HAD, HAD-A or HAD-D score. Furthermore, no significant differences were found between the sexes within each age group. An unpaired t-test was also performed, which showed no significant difference between the groups with urban or rural place of residence on either scale. The mean values, standard deviations and distributions for each gender and age group are listed in Table 2. Comparisons with the results obtained by Sullivan et al. (12) revealed significant differences on both the HAD-A and the HAD-D scales. The men in our sample scored significantly above these comparison values on both subscales. Women, on the other hand, had significantly lower mean scores on the anxiety subscale (see Table 3). A number of Pearson s product-moment correlations were calculated in order to evaluate the relationships between the different anxiety and depression scales and subscales. As expected, a strong correlation was found between the total HAD scale and the anxiety and depression subscales (0.92 and 0.90, respectively). A strong correlation was also, as expected, found between total HAD and BDI scores (0.73), and between HAD-D and BDI scores (0.71). The correlation between total HAD and STAI-T was 0.71, and that between total HAD and STAI-S was Similar to the relationship between HAD-D and BDI, the correlations between total HAD and STAI-S/STAI-T (0.68 and 0.71, respectively) were somewhat stronger than those between the anxiety subscale (HAD-A) and STAI-S/STAI-T (0.64 and 0.66, respectively). The correlations between HAD-A and STAI-T (0.66) and between HAD-D and STAI-T (0.64) were almost identical. In order to determine whether the relationship between HAD and different components of the BDI made any difference, the BDI was divided into two alternative sets of non-physical and physical subscales (24). As can be seen in Table 4, the correlations were stronger between HAD and the non-physical subscales of the BDI than between HAD and the physical subscales. Table 2. Mean scores, standard deviations (SO) and distributions on the HAD scale Total HAD score HAD-A score HAD-D score Total Total Total Total sample Mean so 25th percentile 50th percentile 75th percentile Range Female subjects Mean so 25th percentile 50th percentile 75th percentile Range Male subjects Mean Sd 25th percentile 50th percentile 75th percentile Range

4 Lisspers et al. Table 3. Mean scores compared to reference values Population Reference sample value (12) t P-value HAD-A Female subjects <005 Male subjects <001 HAD-D Female subjects NS Male subjects <00001 a NS. non-significant. Table 4. Correlations between HAD and BDI non-physical and physical subscales BDI Total HAD HAD-A HAD-D Non-physical subscale 16 items (1-14, 18 and 19) items (1-13) Physical subscale 5 items ( and 21) items (14-21) were loading on the appropriate factor. Factor 1 Anxiety (items 1, 3,5, 9, 11 and 13), and factor 2. Depression (items 2, 4, 6, 7, 8, 10, 12 and 14), accounted for 42.2% and 10.0%, respectively, of the total variance. Further analyses of the male and female subgroups showed similar results. In the male subgroup, item 7 was again loading on the wrong factor. Furthermore, in the male subgroup, item 10 ( I have lost interest in my appearance ) was not loading on either the anxiety factor or the depression factor. In the female subgroup, item 7 had almost identical loadings (0.43 and 0.40) on the two factors (Table 5). To test the stability of the factor structure, the sample was randomly divided into two similar subgroups, with 312 subjects in each. In sample 1 the result was similar to that for the total sample. All items except for item 7 were loading on the right factor. In sample 2, however, item 7 was for the first time loading on the appropriate factor, but item 14 ( I can enjoy a good book or TV programme) was not loading on either the anxiety or the depression factor. Factor analysis A principal-component analysis (oblique rotation) was performed in order to test the validity of the original subscales (1). The analysis of the total sample yielded two factors (eigenvalue >1.0). These two factors corresponded fairly closely to the original anxiety and depression subscales. All but one item (item 7, I can sit at ease and feel relaxed ) Discussion According to the present study, the Swedish version of the HAD seems to be bidimensional as in the original formulation by Zigmond & Snaith (l), and as has also been found in factor analyses by Moorey et al. (8) and Sigurdardottir et al. (13). The factor structure that emerged in the present study is consistent with the results obtained by Moorey et al. (8). All but one item (item 7) and, in the male Table 5. Factor analyses for HAD: total sample and male and female subgroups Total sample Male subjects Female subjects Factor 1 Factor 2 Factor 1 Factor 2 Factor 1 Factor 2 Original subscale (anxiety) [depression) [anxiety) (depression) (anxiety) (depression) 1. Anxiety 2. Depression 3. Anxiety 4. Depression 5. Anxiety 6. Depression 7. Anxiety 8. Depression 9. Anxiety 10. Depression 11. Anxiety 12. Depression 13. Anxiety 14. Depression Eigenvalue Total variance 1%) Results for a cuf-off of

5 Hospital Anxiety and Depression Scale subsample, two items (items 7 and 10) loaded on the appropriate factor. Moreover, the internal consistency was found to be high for both the anxiety and the depression subscales. The stability of the factor structure appears to be quite high, since analyses performed on different subsamples extracted two fairly comparable factors. These results are consistent with the findings of Moorey et al. (S), who also reported two factors, both in subsamples of men and women and in two split-halves samples. However, the results conflict with those from a study by Brandberg et al. (ll), who found three and even four factors in different Swedish samples of patients with malignant melanoma. Those authors suggested that the two-factor structure in the Swedish version of the HAD could be expected only for rather ill groups of patients, and that the healthier the patients, the more factors would emerge. However, in the present study on a Swedish population sample the two-factor structure did emerge almost perfectly. With regard to both depression and anxiety, there was a tendency for the entire HAD score to correlate more strongly with the respective comparison measure (BDI and STAI) than the more specific subscales (HAD-D and HAD-A). These results are consistent with the findings of Malasi et al. (4) Silwerstone (5) and Razavi et al. (6), and cast some doubt on the validity of the subscales. Furthermore, the relationship between the depression subscale (HAD-D) and the STAI-T was almost identical to the relationship between the HAD-A and the STAI-T. However, as expected, the correlation between the HAD and the nonphysical subscale of the BDI was stronger than that between the HAD and the physical subscale of the BDI. A weakness of the present study was that the response rate was quite low (only 48%). This suggests a need for caution when interpreting the prevalence data for depression and anxiety in the general population. However, it might not have the same impact on analyses of relationships between the different scales and factors. It seems somewhat surprising that the factor analyses are consistently extracting two factors, anxiety and depression yet that, on the other hand, the relationships between the BDI and the HAD depression subscale and the STAI and the HAD anxiety subscale are weaker than those obtained using the total HAD scale. It will be important to study in more detail the validity of the HAD, BDI and STAI, especially in patients diagnosed as having major depression, dysthymic disorder or clinical anxiety. None the less, the present study has provided sufficient positive results to indicate that the HAD is (as it was originally designed to be) useful as a brief clinical indicator of possible depression and clinical anxiety. Acknowledgements This research was supported by grants from AMF Sjukforsakring and from MidSweden University at Ostersund (research theme Human Resource Development ). The authors wish to thank I? Carlsvard, B. Lindroos and G. Stbhl, MidSweden University at Ostersund, for collection and preliminary analysis of these data. References 1. ZIGMOND AS, SNAITH RI? The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983: 67: SNAITH RP, TAYLOR CM. Rating scales for depression and anxiety: a current perspective. Br J Clin Pharmacology 1985: 19: 17s-20s. 3. AYLARD PR, GOODING JH, MCKENNA PJ, SNAITH RP. A validation study of three anxiety and depression selfassessment scales. J Psychosom Res 1987: 31: MALASI TH, MIRZA IA, EL-ISLA ME Validation of the Hospital Anxiety and Depression Scale in Arab patients. Acta Psychiatr Scand 1991: 84: SILVERSTONE PH. Poor efficacy of the Hospital Anxiety and Depression Scale in the diagnosis of major depressive disorder in both medical and psychiatric patients. J Psychosom Res 1994: 38: RAZAVI D, DELVAUX N, FARVACQUES C, ROBAYE E. Screening of adjustment disorders and major depressive disorders in cancer in-patients. Br J Psychiatry 1990: 156; THOMPSON DR, MEDDIS R. A prospective evaluation of inhospital counseling for first-time myocardial infarction men. J Psychosom Res 1990: 34: MOOREY S, GREER S, WATSON M et al. The factor structure and factor stability of the Hospital Anxiety and Depression Scale in patients with cancer. Br J Psychiatry 1991: 158: LUNDQVIST c, SJOSTEEN A, BLOMSTRAND c, LIND B, SULLIVAN M. Spinal cord injuries. Clinical, functional and emotional status. Spine 1991: 16: SHIELL J, SHIELL A. The prevalence of psychiatric morbidity on a coronary care ward. J Adv Nursing 1991: 16: BRANDBERG Y, BOLUND C, SIGURDARDOTTIR V, SJODEN PO, SULLIVAN M. Anxiety and depressive symptoms at different stages of malignant melanoma. Psycho-Oncology: 1992: 1: SULLIVAN M, KARLSSON J, SJOSTROM L et al. Swedish obese subjects (SOS)- an intervention study of obesity. Baseline evaluation of health and psychosocial functioning in the first 1743 subjects examined. Int J Obesity 1993: 17: SIGURDARDOTTIR V, BOLUND C, BRANDBERG Y, SULLIVAN M. The impact of generalized malignant melanoma on quality of life evaluated by the EORTC questionnaire technique. Qua1 Life Res 1993: 2: BARCZAK P, KANE N, ANDREWS S, CONGDON AM, CLAY JC, BETS T. Patterns of psychiatric morbidity in a genito-urinary clinic. A validation of the Hospital Anxiety Depression Scale (HAD). Br J Psychiatry 1988: 152: KARANCI NA. Patterns of depression in medical patients and their relationship with causal attributions for illness. Psychother Psychosom 1988: 50: WESLEY LAVONNE A, GATCHEL RJ, POLATIN PB, KINNEY RK, 285

6 Lisspers et al. MAYER TG. Differentiation between somatic and cognitive/ affective components in commonly used measurements of depression in patients with chronic low-back pain. Let s not mix apples and oranges. Spine 1991: 16: SNAITH RP: The concepts of mild depression. Br J Psychiatry 1987: 150: American Psychological Association. Diagnostic and statistical manual of mental disorders, 3rd edn. Washington, DC: American Psychological Association, HAMER D, SANJEEV D, BUTTERWORTH E, BARCZAK P: Using the Hospital Anxiety and Depression Scale to screen for psychiatric disorders in people presenting with deliberate self-harm. Br J Psychiatry 1991: 158: WILKINSON MJB, BARCZAK P: Psychiatric screening in general practice: comparison of the General Health Questionnaire and the Hospital Anxiety Depression Scale. J R Coll Gen Pract 1988: 38: BECK AT, WARD CH, MENDELSON M, MOCH J, ERBAUGH J. An inventory for measuring depression. Arch Gen Psychiatry 1961: 4: SPIELBERGER CD, GORSUCH RL, LUSHENE RE. Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists, SODERMAN E, LISSPERS J. Diagnosing depression in patients with physical diseases using the Beck Depression Inventory (BDI). Scand J Behav Ther (in press). 286