Measurement of psycho-social stress in relationship to an illness (a controlled study of 100 cases of malignancy).
It is well documented that stress (of the nature of a single life event or a series of crises) can cause psychiatric and physical illnesses,, and that stressors may have an important part to play in influencing the onset and propagation of malignancy. However, the entire chain of events i.e., psychic stress-hormonal imbalance-cardiovascular and coagulation responses-effect on tumour cell dissemination and implantation-has not yet been demonstrated to go on to completion. This paper presents a systematic examination of the probable association between malignancy and psychosocial stress.
This study was conducted in the outpatient services of the Department of Psychiatry, King Edward VII Memorial Hospital and the Tata Memorial Cancer Hospital, Bombay.
One hundred patients with an histologically established diagnosis of malignancy were serially included in the study. The control group consisted of 100 employees of various cadres of the Tata Textile Mills, Bombay. The index and control groups were matched for age (in decades), marital, and socioeconomic status. They were not matched for sex as the majority of mill workers were males.
The patients and controls were interviewed by a clinician and a psychiatric social worker, using a prepared proforma. Three methods were used for measuring life stress. The first two were the Social Readjustment Rating Scale (SRRS) by Holmes and Rahe and the Life Experience Inventory (LEI) by Cochrane and Robertson. These scales consisted of check list of life events. The subject had to indicate the event or events that had occurred and the duration of time during which this particular life change was operating. A table was prepared listing the number of individuals in the index and control groups who admitted experiencing the different items in- the check-list and the two groups were then compared by using the test of difference in proportions for each item.
Each event was assigned a value expressed as "Life Change Units" (LCU). This provided a numerical index of the degree of stress generated by the occurrence of such an: event for an average person. The subject's total score on the SRRS or the LEI is the sum of the LCUs which he has reported as experienced., The third scale was developed by us from the earlier check-lists. It recorded those life events which were present and were then quantified for severity on a seven point :unidirectional scale (0-6), which gave the. 'Scaled Life Change Unit' values (SLCU). This quantification was done on the basis of information supplied by the patient who had experienced the stress. Thus the total LCU scores were derived from LCU values which had been arbitrarily assigned while the total SLCU scores were derived on the basis of the experience of change perceived by each individual. Therefore, every subject finally had three different score values indicating his/her psychosocial stress viz: total SRRS, LEI and SLCU scores.
The index and the control groups were not matched for sex. This could have introduced a bias as three events could happen only to females; e.g. 'unwanted pregnancy', 'miscarriage', 'abortion' and two events could occur exclusively to males ('wife begins or stops work', and 'pregnancy of wife'). Therefore, 10 females from the control group were matched for age with 10 females from the index group; similarly, 50 males from the index group were matched with 50 males from the control group and the data was analysed for mean SLCU scores.
An yearwise distribution of total SLCU scores for each subject was obtained for three years preceeding the illness. These scores were arranged in all possible combinations according to whether they were equal during all three years or were altered during the first or third year as compared to the second year.
[Table - 1] shows the demographic characteristics of the index and control groups and the diagnostic category and the duration of the illness in the index group.
[Figure 1], [Figure 2] and [Figure 3] show the frequency distributions of the total scores for the three different indices. The scores for the index group followed a normal distribution, while the scores for the control group followed a negative exponential distribution. The total scores of the two groups differed and the median scores for the control group were lower than those for the index group for all the three indices.
[Table - 2] shows that SLCU score values were significantly higher in the index group in males as well as in females indicating that sex would not introduce a bias in data. The subsequent analysis, was therefore, done on the original sample.
[Table - 3] compares the total SLCU scores and the SLCU scores in four areas of life change in the index and the control groups. The difference between the two groups was statistically significant indicating-greater psychosocial stress in the index group.
On comparing the two groups who admitted experiencing the different items in the checklist, it was revealed that the index and the control groups differed statistically (p < 0.01) only in case of two items (income and change in sleeping habits). "Getting in debts" was present in a large number of individuals in both group (33 in the control group and 35 in the index group); similarly 'sex difficulties' were present in 43 individuals in the control group and 34 in the index group.
The 95th percentile point for total SLCU score was 35.5 in the control group; for LCU score (on SRRS) it was 340 and for LCU score (on LEI) it was 345. The total scores of the index group (n = 100) on each scale viz. SRRS, SLCU and LEI were compared among themselves. The rank correlation coefficients were as follows (1) SLCU: SRRS 0.57; (2) LEI: SLCU 0.81 (3) LEI: SRRS 0.64. These values indicate that all the three scales are comparable for measurement of psychosocial stress.
[Figure 4] shows that nine possible combinations can be obtained from the scores distributed over 3 year period. The pattern that was followed by the maximum number of patients (51 patients) was that the scores during the third preceeding year > scores during the second preceeding year < scores during the first preceeding year.
The investigation of psychosocial stress in terms of Life Change Events is a difficult proposition, particularly when diseases with an insidious onset are under consideration, since the disease process may have been operating for a considerable length of time before the emergence of clear cut symptoms. Therefore, many researchers have chosen illnesses with an acute onset e.g., myocardial infarction for the study of life change events. However, even in these instances psychosocial stress may have been operating for some time prior, and the acute manifestations may only signify the culmination of ongoing psychological and social pathology. This is probably true of many psychosomatic and other chronic illnesses including malignancy,, unlike acute infections and traumatic conditions.
About 90% of the subjects in both groups belonged to social classes III, IV and V [Table - 1]. The majority of the mill workers from the control group had left their families in villages, as they could not afford to accommodate them in the city where they were working. This created constant problems of adjustment and readjustment between the family's place of residence and the individual's place of work.
A cross cultural study between the Americans and the Japanese showed that the Japanese rated items concerning 'family' as the highest, whilst the Americans rated items concerning 'finance' and 'personal life' as the highest. In Japanese culture, the family is regarded as a basic unit of society. This also holds true for the Indian culture. However, items concerning the area 'finance' and not 'family' were assigned maximum weightage by the index group in this study.
Analysis of life change events depending upon whether the experience of an event was pleasant or unpleasant would not have been meaningful, because it is the change which is important rather than its direction., Certain events may be under the control of the subject experiencing them while others may not be so. This will also affect the stress experienced by the person and the total scores.
The items which are incorporated in the SRRS and LEI scales have been generated either by normal volunteers or by diverse patient population. As the list of items used in these scales is not exhaustive and as the items are culture relevant, it would be meaningful to do a future study with the list of life change event items modified for Indian populations used by Singh et al but with the quantification method as described in the present study.
In the SRRS and LEI scales, the items have been assigned weightage in an arbitrary fashion e.g. the item 'marriage' has been assigned as LCU value of 50, and the item 'death of spouse' possible LCU value of 100. All other items have been assigned values between these two extremes. These scales give an idea about the severity of change experienced but not about the possible response of the subject to the change. The scales would thus measure 'stress' rather than 'strain' (perceived stress). It is hoped that SLCU measures 'strain' which may be more important than 'stress'. Each individual would experience a different degree of strain for a given stress.
Our cut-off points for SRRS (349) and LEI (245) are different from the originally proposed cut off points of 150 and 159.4 respectively,, probably because of the difference in life styles, family structures etc. The results of a preliminary analysis using the LEI which was used on Indian and Pakistani residents in Britain showed that the mean scores for these two groups closely approximated the mean scores in British subjects. This would possibly be due to their adopting the Western way of life and standard of living.
We felt that circular reasoning may have been responsible for high SLCU scores in study samples where the illness was of recent origin. This would yield inflated values for scores just prior to the onset of the illness. Therefore, information about life change events was elicited over a period of three years rather than during the year just preceeding the illness in the present study. If circular reasoning had been responsible, and life change events had occurred secondary to the illness, then the SLCU scores should have been highest only during the year preceeding the illness and not as reported in the present study. Differences between recall of recent events and of remote events was not likely to have affected the scores, for the same reason. The scores were high in the third year and in the year preceeding the illness in the index group. The trend was not in a progressively increasing direction from the third year through the second year and to the first year. [Fig. 4].
If the stressors as independent variables are operating in a random fashion over a period of three years, then an equal number of patients (n = 11) is expected to fall in each of the nine hypothetical categories, according to whether the total SLCU scores remain at the same level during all three years or show any alteration. The emerging pattern showed instead that stress was maximal during the third year preceeding the illness. It decreased during the second year, and again showed an increase during the first year preceeding the illness, which was significantly different than the other patterns [Figure 4].
The authors thank the Dean, King Edward VII Memorial Hospital and Seth G.S. Medical College, Bombay, and Dr. D. J. Jussawalla, Chief Surgeon, and Director, Tata Memorial Cancer Hospital, Bombay, for permission to conduct and publish this study.
Grateful acknowledgement is made to the senior surgical staff of Tata Memorial Cancer Hospital for permission to interview their patients, and to the management of Tata Textile Mills, Parel, Bombay, for permission to interview their employees.