1991, The British Journal of Radiology, 64, 510-515
Assessment of the psychological impact of a breast screening programme By A. R. Bull, MRCGP, M F C M and * M . J . Campbell, PhD Yorkshire Regional Health Authority, Park Parade, Harrogate, HG1 5AH, UK and 'Department of Medical Statistics and Computing, Southampton General Hospital, Southampton SO9 4XY, UK
{Received January 1990 and in revised form May 1990) Keywords: Psychological effects, Mammographic screening
Abstract. In order to assess the psychological effect of mammographic screening, questionnaires (which included psychometric tests) were sent to 750 women at invitation and, 6 weeks after screening, to 420 women normal after the first mammograph, to 240 women normal after special assessment, and to 68 women normal after open biopsy. Increasing degree of the investigation was associated both with increasing frequency of breast self examination (10% were practising breast self-examination at least once a week before screening compared with 24% for women after special assessment and 35% of women who had had an open biopsy (p < 0.001)), and with greater confidence that any malignancy in the breast would have been found. Psychometric scores showed no increase of general levels of anxiety or depression in the screened groups. For anxiety, percentages abnormal were 5, 4, 2 and 6 for the four groups, respectively, and for depression the percentages abnormal were 5, 4, 4 and 6, respectively; 10% of screened women were more anxious about having breast cancer as a result of the screening. At least 10% of women proceeding to open biopsy of benign lesions require professional counselling and support. Psychological ill effects were not detected by the psychometric test among women who did not proceed to this final investigation. Behavioural changes did suggest a raised awareness or fear of potential cancer among the screened population.
Several authors have suggested that screening for breast cancer could raise the level of cancerophobia and general anxiety in the screened population (Gravelle etal, 1982; Maguire, 1983; Roebuck, 1986). In 1983, Maguire claimed that the introduction of a breast cancer screening programme in Manchester had led to an increased number of women presenting with cancerophobia, who "are extremely anxious, constantly fearful about cancer of the breast, and frequently attend their doctor's surgery or hospital clinics with alleged lumps" (Maguire, 1983). He recommended that screening programmes should try to determine the extent of associated psychological complications. In 1988, Salisbury and Southampton Health Districts introduced a mammography screening programme, according to the recommendations of the Forrest report
(Forrest, 1986). We decided to assess the psychological effects on well women of participation in the screening process. Methods
Four stages of the screening programme were identified as Stage A: the invitation to attend; Stage B: routine mammography; stage C: attendance at a special clinic for investigation of mammographic abnormality by ultrasound, further radiography, or fine-needle cytology; and Stage D: attendance for surgical biopsy of a probably malignant lesion. Women at Stage A were selected by taking the first 125 women listed for each of the first six general practices involved in the programme (Group A). The first 420 of these women who were found to be normal at
Table I. Age distribution by group Group A
C
B
D
Age (years)
n
(%)
n
(%)
n
(%)
n
(%)
50-54 55-59 60-64 65-70 Unknown Total
122 154 185 40 40 541
(22.6) (28.5) (34.2) (7.4) (7.4)
76 113 105 26 11 331
(22.9) (34.1) (31.7) (7.9) (3.3)
66 54 54 15 15 204
(32.3) (26.5) (26.5) (7.4) (7.4)
10 18 16 4 1 49
(20.4) (36.7) (32.7) (8.2) (2.0)
510
The British Journal of Radiology, June 1991
Psychological impact of mammographic screening Table II. Frequency of breast self-examination compared with before screening frequency Group B
C
Also in A
More often As often Less often No response Total
D
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
35 148 34 10 227
(15) (65) (15) (4)
15 71 12 6 104
(14) (68) (12) (6)
75 101 24 4
(37) (49) (12) (2)
22 20 5 2
(45) (41) (10) (4)
204
Stage B were identified as the second study group (Group B). The first 240 of all women found to be normal at Stage C, and the first 68 of all women normal at Stage D, formed Groups C and D, respectively, regardless of whether they had been part of Group A.
49
The numbers were chosen to give a 95% power at the 5% significance level of detecting an increase in the prevalence of depression from 10 to 20% comparing Group A with Group D. A short questionnaire was posted to all women in
Table Ilia. Frequency of breast self-examination by group Group
A
B
Not in B
Never < Once/month Once/month Once/week > Once/week No response Total
C
Also in B
Also in A
Not in A
D
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
56 155 69 25 6 1 311
(18) (50) (22) (8) (2) (0)
63 94 54 15 3 0 229
(28) (41) (24) (7) (1) —
68 44 85 24 5 3 229
(30) (19) (37) (11) (2) (1)
22 23 47 10 0 0 102
(22) (23) (46) (10) — —
24 34 97 41 8 0 204
(12) (17) (48) (20) (4) —
1 1 18 12 5 0 49
(14) (14) (37) (25) (10) —
Table IHb. Frequency of breast self-examination before and after screening After screening
Before screening Never < Once/month Once/month Once/week > Once/week Total
Never
< Once/ month
Once/ month
Once/ week
> Once/ week
Total
48 19 0 1 0 68
8 34 2 0 1 45
5 34 44 1 0 84
1 5 6 10 2 24
0 0 2 3 0 5
62 92 54 15 3 226
= 15.2, d.f. = ! , / » < 0.001. Vol. 64, No. 762
511
A. R. Bull and M. J. Campbell Table IVa. Psychometric score (depression) by group
Group
Not in B
Normal Borderline Abnormal Total
D
C
B
A Also in B
Also in A
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
232 52 26 310
(75) (17) (8)
186 26 10 222
(83) (12) (4)
187 23 12 222
(84) (10) (5)
95 7 2 104
(91) (7) (2)
168 25 9 202
(83) (12) (4)
43 3 3 49
(88) (6) (6)
X2= 11.2, d.f. = 6, /> = 0.083.
Table IVb. Psychometric score (depression) before and after screening
was borne in mind, and a paired comparison of these women who appeared in both Groups A and B was made using a paired Mest or a Wilcoxon rank sum test.
After screening
Before screening Normal Borderline Abnormal Total
Normal Borderline
Abnormal
Total
176 11 0 187
2 4 6 12
186 26 10
8 11 4 23
222
McNemar x2 = 0, d.f. = 1, p = 1.0. Group A with their letter of invitation to attend for screening. A similar questionnaire was posted to all women of Groups B, C and D, 6 weeks after a normal result was declared. No reminders were sent. The questionnaire asked for demographic information, family history of breast cancer, habits of selfexamination and impressions of the screening programme. It incorporated the Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983), which has been validated against other such instruments (Aylard et al, 1987) and was designed for use in both outpatient and community settings (R. P. Snaith, personal communication). The scale comprises two sets of seven questions, one testing for anxiety, the other for depression, none of which refer to somatic symptoms. The maximum (worst) score for each set is 21. Statistics Group B will contain a group of women who completed a questionnaire before screening in Group A, and a group of women who failed to complete the questionnaire before screening. Thus, Groups A and B will have some women in common and a comparison of Groups A, B, C and D by an analysis of variance or non-parametric Kruskal-Wallis test is not strictly valid since the observations were not all independent. This 512
Results Five-hundred and forty-one (72%) replies were received to 750 questionnaires sent to Group A, 331 (79%) to 417 sent to Group B, 204 (85%) to 240 sent to Group C, and 49 (72%) to 68 sent to Group D. Not every questionnaire was entirely complete. Of the 331 in Group B only 229 (69%) had completed a questionnaire before screening. Table I compares the age distribution in each of the four groups and shows that it was similar, apart from an increase in 50-54 year olds in Group C, which is consistent with the low specificity of single-view mammography in younger age groups (Maguire, 1983). Table II shows the response to the question "Compared with what you did before you were screened, do you examine your breasts for lumps". Ranking the three categories "less often", "as often" and "more often" by 1, 2, 3, respectively, we find mean ranks of 2.0, 2.3 and 2.4 for Groups B, C and D, respectively, differences that are most unlikely to have arisen by chance (Kruskal-Wallis %2 = 28.6, d.f. = 2, p < 0.001), suggesting that women in Groups C and D now examine their breasts more often. Table Ilia quantifies the frequency of breast selfexamination for all four groups. Within Groups B, C and D, there was a significant correlation (Spearman's r = 0.24, p < 0.001) between increasing intensity of the investigation and increasing frequency of subsequent breast self-examination. For the 226 women who answered the question in both Groups A and B (Table Illb), 64 had increased breast self-examination and 26 had decreased (%2 = 15.2, d.f. = 1, p < 0.001). The mean scores (range) for the set of psychometric questions that tested for depression were, for Groups A, B, C and D, respectively, 5.0 (0-19), 4.23 (0-15), 4.25 (0-16) and 3.82 (0-18). There was a significant difference between the four groups (Kruskal-Wallis X2 = 19.0, d.f. = 3, p = 0.0003). However, a multiplecomparison test indicated that this was entirely due to The British Journal of Radiology, June 1991
Psychological impact of mammographic screening Table Va. Psychometric score (anxiety) by group Group A
B
Not in B
Normal Borderline Abnormal Total
C
Also in B
Also in A
Not in A
D
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
253 40 20 313
(81) (13) (6)
186 27 6 219
(85) (12) (3)
184 25 10 219
(84) (11) (5)
91 10 2 103
(88) (10) (2)
174
(86) (12) (2)
42
(86) (8) (6)
24 4 202
4 3 49
X2 = 5.09, d.f. = 6, p = 0.053.
Table Vb. Psychometric score (anxiety) before and after screening
respectively; those who were less anxious scored 85%, 10% and 5% for normal, borderline and abnormal, respectively; and those who were unchanged scored After screening 89%, 8% and 3% for normal, borderline and abnormal, respectively. For anxiety, more anxious scored 65%, Normal Borderline Abnormal Total 26% and 8%, less anxious scored 84%, 10% and 5% and unchanged scored 90%, 7% and 3% for normal, Before screening borderline and abnormal, respectively. There is a highly Normal 173 12 1 186 significant difference in the scores for both anxiety Borderline 11 12 4 27 (p < 0.001) and depression (p < 0.01). Abnormal 0 1 5 6 Table VI shows an increasing confidence in the sensiTotal 184 25 10 219 tivity of the screening process with degree of investigation (x2 = 48.3, d.f. = 3, p < 0.001), with 96% of McNemar x2 = 0.5, d.f. = !,/» = 0.8. those in Group D confident that the screening programme was very likely to or would always detect a malignancy, compared with 81% of Group A. A total of 21 women were Jess confident after screening than Group A scoring higher than the others, and a compari- before and 49 more confident (Table VIb). When asked, son of Groups B, C and D was not significant "Do you think that a woman has a good chance of (Kruskal-Wallis * 2 =1.9, d.f. = 2, p = 0.4). For the being cured from breast cancer, 97.8% of all women second set, which tested for anxiety, the scores were, replied "only if the cancer is caught early enough". respectively, 4.97 (0-20), 4.43 (0-17), 4.32 (0-15) and There was no difference between the groups. Only two 4.27 (0-14). Again there was a significant difference women (in Group A) replied 'never'. Remarks made in the space on the questionnaire for between the groups (Kruskal-Wallis %2 = 10.7, d.f. = 3, p = 0.014) that was entirely due to Group A scoring open comment were categorized by ARB. A total of 349 higher than the others, and a comparison of Groups B, (85.7%) of Groups B, C and D said that the programme C and D was not significant (Kruskal-Wallis %2 = 0.7, was very good or of clear personal benefit. Six per cent d.f. = 2, p = 0.7). of women had found it a threatening or frightening When the psychometric scores were grouped into experience, and 4% remarked that it had been physiranges of normal (0-7), borderline (8-10) and abnormal cally uncomfortable but made no other comment. ( > 10), there was no significant difference between the groups (Tables IVa and Va). In Table IVb, 14 people Discussion The greater the exposure to screening procedures, the became more depressed after screening and 15 less so. In Table Vb, 15 people became more anxious and 12 less more likely were women subsequently to have increased so. When asked "has the screening left you more the frequency of breast self-examination for lumps anxious about having breast cancer, less anxious or (Table II). This could be a beneficial effect of the prounchanged", of all responders in Groups B, C and D, 57 gramme provided that such an increase did not denote (10%) said that screening had left them feeling more obsessional behaviour. Table Ilia suggests that for Group B and C, breast anxious about having breast cancer, 232 (41.2%) were less anxious, and 273 (48.5%) were unchanged. For self-examination increased in the short-term from never depression, those who were more anxious scored 70%, or less than once a month to once a month or once a 28% and 2% for normal, borderline and abnormal, week. Table Illb suggests it is particularly women going Vol. 64, No. 762
513
A. R. Bull and M. J. Campbell Table Via. "If a woman has breast cancer, how likely is the breast screening programme to find it?" Group A Not in B
Always Very likely Quite likely Unlikely Missing Total
D
C
B Also in B
Also in A
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
27 233 48 1 2 311
(9) (75) (15) (0) (1)
31 147 47 0 4 229
(14) (64) (21) — (2)
36 170 21 — 2 229
(16) (74) (9)
16 74 7 — 5 102
(16) (73) (7)
48 137 15 0 4 204
(24) (67) (V)
15 32 2 0 0 49
(31) (65) (4)
(1)
(5)
(2)
Pooling "unlikely" with "quite likely" and omitting "missing", x2 = 48.3, d.f. = 6, p < 0.001.
Table VIb. Change in opinion before and after screening After screening Always Very likely
Quite likely
Total
15
1 5 15 21
31 147 47
Before screening Always Very likely Quite likely
17 4
Total
36
15 125 28 168
225
McNemar £ = 10.4, d.f. = I, p < 0.001.
from less than once a month to once a month. Such behaviour is normal and acceptable, although perhaps paradoxical, since having been declared clear by the screening process, many women might be expected to have seen no further need for any self-examination. Indeed, increasing exposure to the screening process did convince many women that if a cancer were present it would have been found (Table VI). The increased frequency of breast self-examination suggests that screening did raise awareness of potential malignancy in the breast for Groups B and C, although not to a pathological degree. Whether or not these changes in frequency of examination persist can only be assessed by long-term follow-up. In Group D, however, there was a significant increase (to over 10%) in women examining themselves for lumps more often than once a week (Table III). Since Group D is expected to comprise 1% of women screened (Forrest, 1986), the number of women who appear to have been precipitated into excessive selfexamination is 0.1% of those screened, or 0.07% of those invited to screen. These figures suggest that at least 10% of women proceeding to open biopsy of 514
benign lesions may require professional counselling or support. The questionnaires were completed by women who had received a normal result, since the study was designed specifically to assess the psychological cost to those women without breast pathology. It is very likely that a greater degree of anxiety occurs during the period before the result is known. For this reason, the Southampton and Salisbury screening programme gives great importance to the full and rapid provision of information and quick progression from one stage to the next, which will have contributed significantly to containing the anxieties of the clients (Fentiman, 1988). An interval of 6 weeks from last examination to questionnaire was chosen to detect prolonged psychological effect in normal women. No reminders were sent to nonresponders because to do so would have included them at a different interval from that of the majority of the study population. A response rate of over 75% is acceptable and was similar for each of the six general practices used. It is possible that those whose anxiety had been raised by screening were less likely to respond than those who were not affected, although the response rate in each of the four study groups was similar suggesting that anxiety was not an important factor of non-response. Whether or not this is the case requires further study. An increased consciousness of, and examination for, lumps in the breast may lead to an increase, even if by only a small percentage, in presentations to the local general practitioner. The level of such presentations would be worth monitoring. The majority of women commented favourably on the programme. It is perhaps not surprising that, with exposure to the more thorough investigation of successive stages, there was increasing confidence in the ability of the programme to detect any malignancy (Table VI), although this is at odds with the increased frequency of self-examination for lumps among women in Groups C and D (Table III), a paradox that could be explained by The British Journal of Radiology, June 1991
Psychological impact of mammographic screening
a feeling of reassurance that no cancer already existed, but increased concern about the possibility of developing a breast lump. The mean anxiety and depression scores on the HAD scale were not raised following screening—indeed, among the screened groups they were reduced, although when grouped into ranges of normal, borderline and abnormal there was no difference between the groups. Nevertheless, 10% of those who had attended reported that the programme had made them more anxious, a figure similar to that of 8% at 6 months reported in Edinburgh (Dean et al, 1986). The HAD scores showed that this 10% showed greater levels of anxiety and depression than the remainder. The fact that the HAD scale did not indicate an overall increase in anxiety or depression may be due to their occurrence at a subclinical level. The studies that have validated the HAD scale have done so on its ability to detect clinically significant anxiety and depression (Zigmond & Snaith, 1983; Aylard et al, 1987). The frequency of breast self-examination following screening shows that such levels of anxiety were provoked only in a small minority of those women who proceeded to open biopsy. Both the HAD scale and habits of selfexamination show that the overall prevalence of clinically significant emotional disturbance was not increased by screening, which contradicts the allegations of Maguire (1983). Nevertheless, there does appear to have been an increase in awareness of, or anxiety about, breast cancer among the screened population, albeit at a subclinical level, reflected in an increase in self-examination (which may in itself be seen as a beneficial consequence) and by self-reported anxiety in 10% of women.
Vol. 64, No. 762
We conclude that the psychological effects of breast cancer screening, though detectable, are small in the general population, but of note in those who proceed to open biopsy of a benign lesion.
Acknowledgment This study could not have been done without the support of the staff in the Breast Screening Unit office under Miss Sue Terry. We also thank Miss Jo Scott for typing and printing the questionnaires and Miss Alison Lee for typing the manuscript.
References AYLARD, P., GOODING, J., MCKENNA, P. & SNAITH, R., 1987. A
validation study of three anxiety and depression self assessment scales. Journal of Psychosomatic Research, 31, 261-268. DEAN, C ,
ROBERTS, M., FRENCH, K. & ROBINSON, S.,
1986.
Psychiatric morbidity after screening for breast cancer. Journal of Epidemiology and Community Health, 40, 71-75. FENTIMAN, I., 1988. Pensive women, painful vigils: consequences of delay in assessment of mammographic abnormalities. Lancet, i, 1041-1042. FORREST, P., 1986. Breast Cancer Screening, (HMSO, London). GRAVELLE, H., SIMPSON, P. & CHAMBERLAIN, J., 1982. Breast
cancer screening and health service costs. Journal of Health Economics, 1, 185-207. MAGUIRE, G., 1983. Possible psychiatric complications of screening for breast cancer. British Journal of Radiology, 56, 284 (Abstr.). ROEBUCK, E., 1986. Mammography and screening for breast cancer. British Medical Journal, 292, 223-226. ZIGMOND, A. & SNAITH, R., 1983. The Hospital Anxiety and
Depression 361-370.
scale. Acta
Psychiatrica
Scandinavica,
67,
515
Assessment of the psychological impact of a breast screening programme By A. R. Bull, MRCGP, M F C M and * M . J . Campbell, PhD Yorkshire Regional Health Authority, Park Parade, Harrogate, HG1 5AH, UK and 'Department of Medical Statistics and Computing, Southampton General Hospital, Southampton SO9 4XY, UK
{Received January 1990 and in revised form May 1990) Keywords: Psychological effects, Mammographic screening
Abstract. In order to assess the psychological effect of mammographic screening, questionnaires (which included psychometric tests) were sent to 750 women at invitation and, 6 weeks after screening, to 420 women normal after the first mammograph, to 240 women normal after special assessment, and to 68 women normal after open biopsy. Increasing degree of the investigation was associated both with increasing frequency of breast self examination (10% were practising breast self-examination at least once a week before screening compared with 24% for women after special assessment and 35% of women who had had an open biopsy (p < 0.001)), and with greater confidence that any malignancy in the breast would have been found. Psychometric scores showed no increase of general levels of anxiety or depression in the screened groups. For anxiety, percentages abnormal were 5, 4, 2 and 6 for the four groups, respectively, and for depression the percentages abnormal were 5, 4, 4 and 6, respectively; 10% of screened women were more anxious about having breast cancer as a result of the screening. At least 10% of women proceeding to open biopsy of benign lesions require professional counselling and support. Psychological ill effects were not detected by the psychometric test among women who did not proceed to this final investigation. Behavioural changes did suggest a raised awareness or fear of potential cancer among the screened population.
Several authors have suggested that screening for breast cancer could raise the level of cancerophobia and general anxiety in the screened population (Gravelle etal, 1982; Maguire, 1983; Roebuck, 1986). In 1983, Maguire claimed that the introduction of a breast cancer screening programme in Manchester had led to an increased number of women presenting with cancerophobia, who "are extremely anxious, constantly fearful about cancer of the breast, and frequently attend their doctor's surgery or hospital clinics with alleged lumps" (Maguire, 1983). He recommended that screening programmes should try to determine the extent of associated psychological complications. In 1988, Salisbury and Southampton Health Districts introduced a mammography screening programme, according to the recommendations of the Forrest report
(Forrest, 1986). We decided to assess the psychological effects on well women of participation in the screening process. Methods
Four stages of the screening programme were identified as Stage A: the invitation to attend; Stage B: routine mammography; stage C: attendance at a special clinic for investigation of mammographic abnormality by ultrasound, further radiography, or fine-needle cytology; and Stage D: attendance for surgical biopsy of a probably malignant lesion. Women at Stage A were selected by taking the first 125 women listed for each of the first six general practices involved in the programme (Group A). The first 420 of these women who were found to be normal at
Table I. Age distribution by group Group A
C
B
D
Age (years)
n
(%)
n
(%)
n
(%)
n
(%)
50-54 55-59 60-64 65-70 Unknown Total
122 154 185 40 40 541
(22.6) (28.5) (34.2) (7.4) (7.4)
76 113 105 26 11 331
(22.9) (34.1) (31.7) (7.9) (3.3)
66 54 54 15 15 204
(32.3) (26.5) (26.5) (7.4) (7.4)
10 18 16 4 1 49
(20.4) (36.7) (32.7) (8.2) (2.0)
510
The British Journal of Radiology, June 1991
Psychological impact of mammographic screening Table II. Frequency of breast self-examination compared with before screening frequency Group B
C
Also in A
More often As often Less often No response Total
D
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
35 148 34 10 227
(15) (65) (15) (4)
15 71 12 6 104
(14) (68) (12) (6)
75 101 24 4
(37) (49) (12) (2)
22 20 5 2
(45) (41) (10) (4)
204
Stage B were identified as the second study group (Group B). The first 240 of all women found to be normal at Stage C, and the first 68 of all women normal at Stage D, formed Groups C and D, respectively, regardless of whether they had been part of Group A.
49
The numbers were chosen to give a 95% power at the 5% significance level of detecting an increase in the prevalence of depression from 10 to 20% comparing Group A with Group D. A short questionnaire was posted to all women in
Table Ilia. Frequency of breast self-examination by group Group
A
B
Not in B
Never < Once/month Once/month Once/week > Once/week No response Total
C
Also in B
Also in A
Not in A
D
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
56 155 69 25 6 1 311
(18) (50) (22) (8) (2) (0)
63 94 54 15 3 0 229
(28) (41) (24) (7) (1) —
68 44 85 24 5 3 229
(30) (19) (37) (11) (2) (1)
22 23 47 10 0 0 102
(22) (23) (46) (10) — —
24 34 97 41 8 0 204
(12) (17) (48) (20) (4) —
1 1 18 12 5 0 49
(14) (14) (37) (25) (10) —
Table IHb. Frequency of breast self-examination before and after screening After screening
Before screening Never < Once/month Once/month Once/week > Once/week Total
Never
< Once/ month
Once/ month
Once/ week
> Once/ week
Total
48 19 0 1 0 68
8 34 2 0 1 45
5 34 44 1 0 84
1 5 6 10 2 24
0 0 2 3 0 5
62 92 54 15 3 226
= 15.2, d.f. = ! , / » < 0.001. Vol. 64, No. 762
511
A. R. Bull and M. J. Campbell Table IVa. Psychometric score (depression) by group
Group
Not in B
Normal Borderline Abnormal Total
D
C
B
A Also in B
Also in A
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
232 52 26 310
(75) (17) (8)
186 26 10 222
(83) (12) (4)
187 23 12 222
(84) (10) (5)
95 7 2 104
(91) (7) (2)
168 25 9 202
(83) (12) (4)
43 3 3 49
(88) (6) (6)
X2= 11.2, d.f. = 6, /> = 0.083.
Table IVb. Psychometric score (depression) before and after screening
was borne in mind, and a paired comparison of these women who appeared in both Groups A and B was made using a paired Mest or a Wilcoxon rank sum test.
After screening
Before screening Normal Borderline Abnormal Total
Normal Borderline
Abnormal
Total
176 11 0 187
2 4 6 12
186 26 10
8 11 4 23
222
McNemar x2 = 0, d.f. = 1, p = 1.0. Group A with their letter of invitation to attend for screening. A similar questionnaire was posted to all women of Groups B, C and D, 6 weeks after a normal result was declared. No reminders were sent. The questionnaire asked for demographic information, family history of breast cancer, habits of selfexamination and impressions of the screening programme. It incorporated the Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983), which has been validated against other such instruments (Aylard et al, 1987) and was designed for use in both outpatient and community settings (R. P. Snaith, personal communication). The scale comprises two sets of seven questions, one testing for anxiety, the other for depression, none of which refer to somatic symptoms. The maximum (worst) score for each set is 21. Statistics Group B will contain a group of women who completed a questionnaire before screening in Group A, and a group of women who failed to complete the questionnaire before screening. Thus, Groups A and B will have some women in common and a comparison of Groups A, B, C and D by an analysis of variance or non-parametric Kruskal-Wallis test is not strictly valid since the observations were not all independent. This 512
Results Five-hundred and forty-one (72%) replies were received to 750 questionnaires sent to Group A, 331 (79%) to 417 sent to Group B, 204 (85%) to 240 sent to Group C, and 49 (72%) to 68 sent to Group D. Not every questionnaire was entirely complete. Of the 331 in Group B only 229 (69%) had completed a questionnaire before screening. Table I compares the age distribution in each of the four groups and shows that it was similar, apart from an increase in 50-54 year olds in Group C, which is consistent with the low specificity of single-view mammography in younger age groups (Maguire, 1983). Table II shows the response to the question "Compared with what you did before you were screened, do you examine your breasts for lumps". Ranking the three categories "less often", "as often" and "more often" by 1, 2, 3, respectively, we find mean ranks of 2.0, 2.3 and 2.4 for Groups B, C and D, respectively, differences that are most unlikely to have arisen by chance (Kruskal-Wallis %2 = 28.6, d.f. = 2, p < 0.001), suggesting that women in Groups C and D now examine their breasts more often. Table Ilia quantifies the frequency of breast selfexamination for all four groups. Within Groups B, C and D, there was a significant correlation (Spearman's r = 0.24, p < 0.001) between increasing intensity of the investigation and increasing frequency of subsequent breast self-examination. For the 226 women who answered the question in both Groups A and B (Table Illb), 64 had increased breast self-examination and 26 had decreased (%2 = 15.2, d.f. = 1, p < 0.001). The mean scores (range) for the set of psychometric questions that tested for depression were, for Groups A, B, C and D, respectively, 5.0 (0-19), 4.23 (0-15), 4.25 (0-16) and 3.82 (0-18). There was a significant difference between the four groups (Kruskal-Wallis X2 = 19.0, d.f. = 3, p = 0.0003). However, a multiplecomparison test indicated that this was entirely due to The British Journal of Radiology, June 1991
Psychological impact of mammographic screening Table Va. Psychometric score (anxiety) by group Group A
B
Not in B
Normal Borderline Abnormal Total
C
Also in B
Also in A
Not in A
D
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
253 40 20 313
(81) (13) (6)
186 27 6 219
(85) (12) (3)
184 25 10 219
(84) (11) (5)
91 10 2 103
(88) (10) (2)
174
(86) (12) (2)
42
(86) (8) (6)
24 4 202
4 3 49
X2 = 5.09, d.f. = 6, p = 0.053.
Table Vb. Psychometric score (anxiety) before and after screening
respectively; those who were less anxious scored 85%, 10% and 5% for normal, borderline and abnormal, respectively; and those who were unchanged scored After screening 89%, 8% and 3% for normal, borderline and abnormal, respectively. For anxiety, more anxious scored 65%, Normal Borderline Abnormal Total 26% and 8%, less anxious scored 84%, 10% and 5% and unchanged scored 90%, 7% and 3% for normal, Before screening borderline and abnormal, respectively. There is a highly Normal 173 12 1 186 significant difference in the scores for both anxiety Borderline 11 12 4 27 (p < 0.001) and depression (p < 0.01). Abnormal 0 1 5 6 Table VI shows an increasing confidence in the sensiTotal 184 25 10 219 tivity of the screening process with degree of investigation (x2 = 48.3, d.f. = 3, p < 0.001), with 96% of McNemar x2 = 0.5, d.f. = !,/» = 0.8. those in Group D confident that the screening programme was very likely to or would always detect a malignancy, compared with 81% of Group A. A total of 21 women were Jess confident after screening than Group A scoring higher than the others, and a compari- before and 49 more confident (Table VIb). When asked, son of Groups B, C and D was not significant "Do you think that a woman has a good chance of (Kruskal-Wallis * 2 =1.9, d.f. = 2, p = 0.4). For the being cured from breast cancer, 97.8% of all women second set, which tested for anxiety, the scores were, replied "only if the cancer is caught early enough". respectively, 4.97 (0-20), 4.43 (0-17), 4.32 (0-15) and There was no difference between the groups. Only two 4.27 (0-14). Again there was a significant difference women (in Group A) replied 'never'. Remarks made in the space on the questionnaire for between the groups (Kruskal-Wallis %2 = 10.7, d.f. = 3, p = 0.014) that was entirely due to Group A scoring open comment were categorized by ARB. A total of 349 higher than the others, and a comparison of Groups B, (85.7%) of Groups B, C and D said that the programme C and D was not significant (Kruskal-Wallis %2 = 0.7, was very good or of clear personal benefit. Six per cent d.f. = 2, p = 0.7). of women had found it a threatening or frightening When the psychometric scores were grouped into experience, and 4% remarked that it had been physiranges of normal (0-7), borderline (8-10) and abnormal cally uncomfortable but made no other comment. ( > 10), there was no significant difference between the groups (Tables IVa and Va). In Table IVb, 14 people Discussion The greater the exposure to screening procedures, the became more depressed after screening and 15 less so. In Table Vb, 15 people became more anxious and 12 less more likely were women subsequently to have increased so. When asked "has the screening left you more the frequency of breast self-examination for lumps anxious about having breast cancer, less anxious or (Table II). This could be a beneficial effect of the prounchanged", of all responders in Groups B, C and D, 57 gramme provided that such an increase did not denote (10%) said that screening had left them feeling more obsessional behaviour. Table Ilia suggests that for Group B and C, breast anxious about having breast cancer, 232 (41.2%) were less anxious, and 273 (48.5%) were unchanged. For self-examination increased in the short-term from never depression, those who were more anxious scored 70%, or less than once a month to once a month or once a 28% and 2% for normal, borderline and abnormal, week. Table Illb suggests it is particularly women going Vol. 64, No. 762
513
A. R. Bull and M. J. Campbell Table Via. "If a woman has breast cancer, how likely is the breast screening programme to find it?" Group A Not in B
Always Very likely Quite likely Unlikely Missing Total
D
C
B Also in B
Also in A
Not in A
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
n
(%)
27 233 48 1 2 311
(9) (75) (15) (0) (1)
31 147 47 0 4 229
(14) (64) (21) — (2)
36 170 21 — 2 229
(16) (74) (9)
16 74 7 — 5 102
(16) (73) (7)
48 137 15 0 4 204
(24) (67) (V)
15 32 2 0 0 49
(31) (65) (4)
(1)
(5)
(2)
Pooling "unlikely" with "quite likely" and omitting "missing", x2 = 48.3, d.f. = 6, p < 0.001.
Table VIb. Change in opinion before and after screening After screening Always Very likely
Quite likely
Total
15
1 5 15 21
31 147 47
Before screening Always Very likely Quite likely
17 4
Total
36
15 125 28 168
225
McNemar £ = 10.4, d.f. = I, p < 0.001.
from less than once a month to once a month. Such behaviour is normal and acceptable, although perhaps paradoxical, since having been declared clear by the screening process, many women might be expected to have seen no further need for any self-examination. Indeed, increasing exposure to the screening process did convince many women that if a cancer were present it would have been found (Table VI). The increased frequency of breast self-examination suggests that screening did raise awareness of potential malignancy in the breast for Groups B and C, although not to a pathological degree. Whether or not these changes in frequency of examination persist can only be assessed by long-term follow-up. In Group D, however, there was a significant increase (to over 10%) in women examining themselves for lumps more often than once a week (Table III). Since Group D is expected to comprise 1% of women screened (Forrest, 1986), the number of women who appear to have been precipitated into excessive selfexamination is 0.1% of those screened, or 0.07% of those invited to screen. These figures suggest that at least 10% of women proceeding to open biopsy of 514
benign lesions may require professional counselling or support. The questionnaires were completed by women who had received a normal result, since the study was designed specifically to assess the psychological cost to those women without breast pathology. It is very likely that a greater degree of anxiety occurs during the period before the result is known. For this reason, the Southampton and Salisbury screening programme gives great importance to the full and rapid provision of information and quick progression from one stage to the next, which will have contributed significantly to containing the anxieties of the clients (Fentiman, 1988). An interval of 6 weeks from last examination to questionnaire was chosen to detect prolonged psychological effect in normal women. No reminders were sent to nonresponders because to do so would have included them at a different interval from that of the majority of the study population. A response rate of over 75% is acceptable and was similar for each of the six general practices used. It is possible that those whose anxiety had been raised by screening were less likely to respond than those who were not affected, although the response rate in each of the four study groups was similar suggesting that anxiety was not an important factor of non-response. Whether or not this is the case requires further study. An increased consciousness of, and examination for, lumps in the breast may lead to an increase, even if by only a small percentage, in presentations to the local general practitioner. The level of such presentations would be worth monitoring. The majority of women commented favourably on the programme. It is perhaps not surprising that, with exposure to the more thorough investigation of successive stages, there was increasing confidence in the ability of the programme to detect any malignancy (Table VI), although this is at odds with the increased frequency of self-examination for lumps among women in Groups C and D (Table III), a paradox that could be explained by The British Journal of Radiology, June 1991
Psychological impact of mammographic screening
a feeling of reassurance that no cancer already existed, but increased concern about the possibility of developing a breast lump. The mean anxiety and depression scores on the HAD scale were not raised following screening—indeed, among the screened groups they were reduced, although when grouped into ranges of normal, borderline and abnormal there was no difference between the groups. Nevertheless, 10% of those who had attended reported that the programme had made them more anxious, a figure similar to that of 8% at 6 months reported in Edinburgh (Dean et al, 1986). The HAD scores showed that this 10% showed greater levels of anxiety and depression than the remainder. The fact that the HAD scale did not indicate an overall increase in anxiety or depression may be due to their occurrence at a subclinical level. The studies that have validated the HAD scale have done so on its ability to detect clinically significant anxiety and depression (Zigmond & Snaith, 1983; Aylard et al, 1987). The frequency of breast self-examination following screening shows that such levels of anxiety were provoked only in a small minority of those women who proceeded to open biopsy. Both the HAD scale and habits of selfexamination show that the overall prevalence of clinically significant emotional disturbance was not increased by screening, which contradicts the allegations of Maguire (1983). Nevertheless, there does appear to have been an increase in awareness of, or anxiety about, breast cancer among the screened population, albeit at a subclinical level, reflected in an increase in self-examination (which may in itself be seen as a beneficial consequence) and by self-reported anxiety in 10% of women.
Vol. 64, No. 762
We conclude that the psychological effects of breast cancer screening, though detectable, are small in the general population, but of note in those who proceed to open biopsy of a benign lesion.
Acknowledgment This study could not have been done without the support of the staff in the Breast Screening Unit office under Miss Sue Terry. We also thank Miss Jo Scott for typing and printing the questionnaires and Miss Alison Lee for typing the manuscript.
References AYLARD, P., GOODING, J., MCKENNA, P. & SNAITH, R., 1987. A
validation study of three anxiety and depression self assessment scales. Journal of Psychosomatic Research, 31, 261-268. DEAN, C ,
ROBERTS, M., FRENCH, K. & ROBINSON, S.,
1986.
Psychiatric morbidity after screening for breast cancer. Journal of Epidemiology and Community Health, 40, 71-75. FENTIMAN, I., 1988. Pensive women, painful vigils: consequences of delay in assessment of mammographic abnormalities. Lancet, i, 1041-1042. FORREST, P., 1986. Breast Cancer Screening, (HMSO, London). GRAVELLE, H., SIMPSON, P. & CHAMBERLAIN, J., 1982. Breast
cancer screening and health service costs. Journal of Health Economics, 1, 185-207. MAGUIRE, G., 1983. Possible psychiatric complications of screening for breast cancer. British Journal of Radiology, 56, 284 (Abstr.). ROEBUCK, E., 1986. Mammography and screening for breast cancer. British Medical Journal, 292, 223-226. ZIGMOND, A. & SNAITH, R., 1983. The Hospital Anxiety and
Depression 361-370.
scale. Acta
Psychiatrica
Scandinavica,
67,
515
