237
Mental Health SUICIDE PREVENTION BY THE SAMARITANS A Controlled Study of Effectiveness C. JENNINGS
B. M. BARRACLOUGH M.R.C.
Clinical Psychiatry Unit, Graylingwell Hospital, Chichester, West Sussex PO19 4PQ
J. R. Moss* Department of Community Health, University of Nottingham To test the hypothesis that Samaritan services prevent suicide, the trends in suicide-rates of county boroughs with Samaritan services have been compared with trends in matched control boroughs without Samaritan services. The suicide-rate for both groups of boroughs has decreased progressively and to the same extent. This finding is not consistent with suggestions that Samaritan services prevent suicide.
Summary
INTRODUCTION
BETWEEN 1963 and 1970 the suicide-rate in England fell by a third, from 12 to 8 per 100000, a change of suicide incidence unique among European countries over this period. To identify the cause of this remarkable decline would be a major step towards a rational, planned programme of preventing suicide. Plausible1 explanations have been advanced-safer domestic gas,’ improved hospital care of the poisoned, better treatment by general practitioners of their depressed patients, a change for the better in the psychiatric services both in hospitals and in the community,2 and the rapid spread of the Samaritan movement.3 There is evidence to support each of these explanations, but insufficient to convince. This paper reconsiders the effectiveness of the Samaritans as a suicide-prevention service. The Samaritans Incorporated is an independent national organisation which aims to provide a confidential non-directive listening and "befriending" service to troubled people. By appropriate national and local advertising the despairing and suicidal are invited to visit or telephone their local Samaritan branch for help. Branches are staffed by volunteers carefully selected and then trained for the job. The growth of the Samaritan organisation has been phenomenal-from the foundation branch in St. Stephen Walbrook, London, where the service began in 1953, to 165 branches and one million calls in 1975.4 There is a sound prima facie case that Samaritan methods of coping with troubled people may prevent suicide. Potential suicides are attracted, for Samaritan clients have a high suicide-rate; the growth of the organisation parallels the decline in the suicide-rate; the Samaritans, it is claimed, are unique to Britain3 and the decline in the suicide-rate certainly is. A scientific case has also been made.6 Bagley saw that a controlled study of Samaritan effectiveness was pos,
*Present address: Health Research auon, Birmingham.
Unit, British Rubber Manufacturers Associ-
sible by comparing suicide-rate trends in towns with Samaritan branches and control towns without branches. Towns with Samaritan services should have less suicide than control towns. The heart of the matter is the criterion for control town selection. Bagley adopted two quite different criteria. The first was a measure of the "ecological" similarity of towns, the second, predictors of the future suicide-rate. The ecological approach used the results of a study7 which summarised in a principal component analysis the variation of 57 demographic, health, social, and economic variables among 157 large towns in England and Wales. Using loadings on the first two of the principal components as the criterion of ecological similarity, he chose as a control the closest non-Samaritan county borough to each of the 15county boroughs with Samaritan branches established by mid-1963. As predictors of the suicide-rate, the second approach used the percentage of the population aged 65 or over, the number of females per thousand males, and a socialclass index.7 The non-Samaritan county borough with the least average distance of rank from each county borough with a Samaritan branch was chosen as a control. Bagley’s result was impressive. The 15 Samaritan boroughs showed a fall in their average suicide-rate of 6%, while control boroughs showed rises-20% when the ecological approach for choosing control towns was used and 7% when suicide-rate predictors were used. The difference in average suicide-rates, 26% and 13%, was statistically significant for both approaches, a finding in favour of the notion that Samaritans prevent suicide. That the Samaritans are responsible for the fall in the national suicide-rate, thus saving thousands of lives, is widely believed.38Bagley’s findings are the most convincing evidence; and his findings, if true, have profound implications throughout the Western world for the planning of suicide-prevention services. In retesting Bagley’s hypothesis that Samaritan services in county boroughs are associated with a reduction in suicide, we have improved on his method for selecting control towns and have looked at a larger number of
boroughs. METHOD
We used four methods of selecting controls, two based on the ecological approach, and two on predictors of the suicide rate. In method i we found the non-Samaritan county borough most nearly resembling each of Bagley’s original 15 Samaritan boroughs, using the loadings from four components as coordinates instead of the two which he used. Matchings based on four components should produce pairs of boroughs more like one another than matchings based on two. In method n we paired each county borough with the one closest to it, again using loadings on the four components, and then excluded pairs if (a) neither borough had a Samaritan branch, or (b) either had a branch opened within three years of the other. This procedure yielded 23pairs. In methods in and iv we used predictors of the suicide-rate to select control boroughs. In method m we employed as a matching variable the suicide-rate before the opening of the Samaritan branch, assuming that by choosing pairs of towns with similar suicide-rates we would be controlling for those factors which caused that suicide-rate. The suicide-rates of towns and districts in England and Wales are highly correlated over short periods.9 The non-Samaritan country boroughs with the pre-opening suicide-rate closest to each of the first 33 Samaritan boroughs were chosen as controls.
238 SUICIDE-RATE CHANGES IN SAMARITAN AND CONTROL BOROUGHS
be impossible which.
to
decide without labels which
curve
is
DISCUSSION
By none of the four methods was it possible to demonthat boroughs with a Samaritan branch show a
strate
greater reduction in suicide-rates than do control bor-
oughs. Although a positive result is usually more convincing than a negative one, there are two reasons for believing our methods to be superior to Bagley’s and
Method iv took the proportion of single-person households the matching variable. We used this variable as a suiciderate predictor because three-quarters of the variation in male suicide-rates and half the variation in female suicide-rates between local-authority areas can be explained by the proportion of single-person households.lO The proportion living alone has been known for many years to correlate highly with the suicide-rate.11 Each borough was paired with the borough ranked next to it on the percentage of people living alone. Pairs were excluded if (a) neither borough had a Samaritan branch, or (b) either had a branch opened within three years of the other. This procedure yielded 35 pairs of Samaritan and control boras
oughs. The average suicide-rate for the three-years after the branch opened was compared with the three-year average rate for the period before it opened. If a branch opened in the second half of the year it was counted as having opened in the subsequent year. The percentage change in the suicide-rate between the two periods was found for each Samaritan borough and its control borough for the same period. The Wilcoxon matched pairs signed ranks test, one-tailed, 12 was applied to test the hypothesis that Samaritan boroughs would show a significantly greater fall, or smaller rise, in the suicide-rate than the controls. We used the Wilcoxon distributed normally.
test
because the data
were not
RESULTS
The four Wilcoxon tests revealed no significant differthe 5% level between the Samaritan and the control boroughs (see accompanying table). An alternative analysis is shown in the figure, where are plotted the mean suicide-rates for the six years before and after the opening of a Samaritan branch, the Samaritan and control boroughs being chosen by method III. Knowing the hypothesis under test, it would ences at
that our result is therefore correct. Firstly, our methods of selecting control towns are better than Bagley’s. In methods I and II the four components used account for 60% of the variance, the two components 43%. Thus our pairs of towns resemble each other, more than do Bagley’s, on those variables that the components represent. In methods III and iv both matching variables accounted for 45% of the variance in post-opening suicide-rates of control towns. Bagley’s suicide-prediction variables accounted for 35%. Secondly, large numbers are more convincing than small numbers; methods II, III, and iv employed samples of 23, 33, and 35 pairs of boroughs as opposed to the 15 available to Bagley in 1964. A search of reports published since 1965 revealed two other studies which compared suicide-rate trends in areas where there were suicide-prevention centres with other areas, matched on relevant variables, where no such service existed.13,14 Neither study showed that the suicide-rate had been reduced in areas where these centres operated. Examination in Edinburgh of the short-term effect of the BBC TV series "The Befrienders" showed that, although more people contacted the Samaritan branch, the number of suicides was not affected, nor was the admission-rate for self-poisoning.1S Despite the continued rapid growth of the Samaritan organisation and in the number of clients contacting it, the suicide-rate has not dropped as steeply since 1970. Suicide rates per 100 000 from 1970 are 81, 8.1, 7.7, 7-8, and for 1974, 7.9. Over the same period numbers of Samaritan branches in the British Isles grew from 115 to 154 and new Samaritan clients per year increased from 70 000 to 190 000.4 Although we do not deny the value ox the Samaritans in relieving distress and providing human contact and comfort, our conclusions must cast doubt on the scientific case for attributing the fall in the suicide-rate to the work of the Samaritans. Changes in the suicide-rate of the present order have been observed before--decreases during both World Wars and an increase at the time of the economic depression in the 1930s. The complex social and economic influences invoked to explain those changes may again be at work. The Samaritans Incorporated kindly provided the opening dates of the Samaritan branches in county boroughs. The calculations of the four-component coordinates were done by Dr J. C. G. Pearson of the Department of Community Health, University of Nottingham. Dr A. Levey suggested the figure. REFERENCES
Suicide-rate trends for Samaritan and control by method III.
boroughs,
chosen
1. Hassall, C., Trethowan, W. H. Br. med. J. 1972, i, 717. 2. Barraclough, B. M. Soc. Sci. Med. 1972, 6, 661. 3. Fox, R. R. Soc. Hlth J. 1975, 95, 9. 4. Yorke, M. The Samaritan (special issue), 1977, p. 1. 5. Barraclough, B. M., Shea, M. Lancet, 1970, ii, 868. 6. Bagley, C. Soc. Sci. Med. 1968, 2, 1. 7. Moser, C. A., Scott, W. British Towns. Edinburgh, 1961.
239
Making
Health and Social Security. This dramatic reduction has allowed us to take down 38 beds and donate 21 to the geriatric services. I am firmly in favour of making the best use of what we have, rather than continually asking for more; in fact the administration promptly cut our establishment rather than allowing us to redeploy some of the 14 nurses saved into, say,
Do
INTEGRATION OF PSYCHIATRIC ASSESSMENT FOR THE ELDERLY
day care.
IAN THOMSON
ACCUMULATION OF ELDERLY PATIENTS
Whitecroft Hospital, Newport, Isle of Wight
At the end of the survey our 63 beds contained 20 graduates who had spent, on average, 21 years in the hospital, 30 patients with dementia, and 6 with other diagnoses; 7 beds were empty. In planning services it is important to know how rapidly new long-stay patients will accumulate. At the end of the three years we had accumulated 21 patients, all but 2 of whom were suffering from senile dementia. 6 had first been admitted to the acute ward and 15 to the psychogeriatric ward.
"More people are surviving into late life, a period when mental disothers are common ... The disorders of late life share a nasty habit of recurring, proving resistant to treatment or pursuing a progressive course to an early death. Thus they threaten our current assets."D. J. JOLLEY, Br.,3‘. Psychiat. News and Notes, 1976, 11, 11. now form a seventh of the fifth of all psychiatric admissions. the beds in many mental hospitals, and will occupy a larger proportion in coming years. In our small psychiatric service for the elderly we have considerably reduced the number of beds by re-integrating the assessment of psychogeriatric patients with the admission and assessment of younger patients in the acute ward.
NATIONALLY, the over-65s
population, but a They occupy half
THE PATIENTS
In the three years from 1973 to
1975,
we
admitted 177 psy-
chogeriatric patients. The diagnoses were: senile dementia (55.4%), manic depression (27-7%), schizophrenia (4-5%), and others (12.4%). 113 patients were admitted once, and 24 had a total of 64 admissions. 16 (21.3%) of the patients with senile dementia had repeat admissions, and only 8 (12.9%) of those with other diagnoses. At the end of the survey period, 59 (43%) of the patients remained or had died in hospital, and 78 (57%) had been discharged. We were surprised at how many apparently intractably handicapped patients improved and went home. It seems likely that the prospects of leaving hospital for even the most disabled group were much improved if they were admitted to the acute ward. During the first year of the survey patients with a diagnosis of senile dementia were admitted to whichever ward had a vacant bed: 16 (64%) of the 25 patients (6 had previously been in the hospital) admitted to the acute ward had left hospital at the end of the survey; only 4 (23-5%) of the 17 patients (7 readmissions) admitted to the psychogeriatric ward were discharged from hospital. BED PROVISION
Another way of looking at how changing the management of psychogeriatric patients affects results is by considering bed provision. In 1970 the catchment population was only 19 490, but by 1975 it had doubled to 39 333 (8700 over 65). Over this period our beds decreased from 122 to 63-i.e., from 6.35 beds per thousand total area population to 1-61 (from 28-45 to 7.24 per thousand population over 65). We now have 2.3 beds for patients with dementia and 2-3 for "graduates" (patients who have grown old in hospital) per thousand elderly population; that is slightly fewer than the 2.5-3 beds per thousand advocated for severe dementia by the Department of
8. Dominian, J. Depression. London, 1976. 9. Sainsbury, P. in Psychiatrie der Gegenwart. Berlin, 1975. 10. Ashford, J. R., Lawrence, P. A. Int. J. Epidem. 1976, 5, 133. 11. Sainsbury, P. Suicide in London. London, 1955. 12. Siegel, S. Nonparametric Statistics. New York, 1956. 13. Lester, D. Hlth Serv. Reps, 1974, 89, 37. 14. Bridge, T. P., Potkin, S. G., Zung, W. W. K., Soldo, B.
Dis. 1977, 164, 18. 15. Holding, T. A. Br. med. J. 1975, iii, 751.
J. J.
nerv. ment.
INTEGRATION AND SELECTION
Many excellent services are run by full-time psychogeriatricians, but the segregation inherent in such services is a disadvantage. In a mental illness unit in the Midlands I found to my surprise that they were mixing drug addicts and psychogeriatric and neurotic patients, to mutual advantage.’ We feel that the admission ward should attempt to resemble the community in its age and sex structure and aim to treat the elderly with other agegroups, including adolescents. We run the ward as a therapeutic community with daily discussion groups and emphasis on cooperative team work. The younger patients have a stimulating effect on the elderly, and soon learn to accept some responsibility for them and encourage them to participate in the daily activities. The elderly, with their more obvious illness, help the younger neurotics
to come to terms
with life and
count
their
blessings. It has been estimated2 that the consultant psychiatrist has one hour per patient per year; if he visits one ward and has one case-conference rather than two, his time will be more effectively used. These time factors apply equally to social and remedial staff. Skilled nursing is particularly important in psychogeriatric wards, but few services are able to staff all wards with equally skilled nurses.
Traditionally, an admission ward is an area where nursing and other staff have high expectations for patients and are much involved with them: "Psychogeriatric patients do better when they are put in a stimulating environment and are specifically stimulated by Staff’.3 Too often in psychogeriatric wards the staff are so busy with the general care of their handicapped patients that direct stimulation of patients tends to go by the board. A problem highlighted by our policy of psychogeriatric care is that activity in the psychogeriatric ward is reduced by the low turnover of patients and the absence of direct new admissions. There are, however, positive advantages in selecting and training staff for a more long-stay, caring function on this ward. It is our firm view that attempting to have, say, three assessment beds on a ward where most of the patients are long-stay, would not work. Assessment patients all too easily become long-stay patients. We believe that our emphasis is right, and is more in 1. McClure, J. L. Personal Communication. 2. National Association for Mental Health. Mind Report, no. 12. 3. Brook, P. Br. J. Psychiat., 1976, 127, 42.