Journalof Psychosomatic Research,Vol. 22, pp. 183 to 191. 0 PergamonPressLtd. 1978. Printedin Great Britain.
PATTERNS IN THE RELATIONSHIP AND PSYCHIATRIC SYMPTOMS IGOR GRANT,*’
OF LIFE EVENTS OVER TIME JOEL YAGER~
Abstract-A prospective study was conducted to investigate the relationship of life-event changes to changes in psychiatric symptoms over time. Eighty-nine male psychiatric outpatients and a comparison group of 107 hospital and university employees completed a Schedule of Recent Events (SRE) and a Symptom Checklist (XL) every two months for 18 months. Symptoms were significantly (p < 0.009) correlated with life events at each of the nine measurement periods (r ranged from 0.38 to 0.58). When event and symptom scores for each subject over time were graphed, 61.7% of the cases could be classified into one of three event-symptom relationship typologies. A concordant relationship (events and symptoms covaried together) appeared in 34.7%; an unresponsive relationship (major changes in events associated with no symptom change) was seen 21.4%; and an antecedent-consequent relationship (build-up in events followed by build-up in symptoms) was identified in 5.4%; 38.3% of the cases could not be classified. Although the study supports previous findings that when large groups are studied, statistically significant events-symptoms correlations can be demonstrated, it also shows that such group data may be obscuring important individual differences. The partitioning of groups according to eventssymtpom typologies may help identify people whom it would be fruitful to study in greater depth from an adaptational viewpoint. In this way we may come closer to understanding why life events affect the health of some people and not others. IN
PREVIOUS publications, we reported that life changes were strongly related to symptom changes among patients and non-patients, that psychiatric patients experienced more events of all kinds, especially undesirable, negative happenings than non-patients, that patients perceived virtually all events as more stressful and as requiring more readjustment than did non-patients, and that undesirable events showed strong correlations with symptoms whereas desirable events showed no relationship whatever [l-2]. Additionally, we were able to report that age, social position, and marital status contributed little to the correlation of life events and symptoms We have now completed 18 months of prospective study relating life events with psychiatric disturbance. In this paper we report on the relationship between life events and psychiatric symptomatology over time and describe patterns we have identified in life event-symptom relationships.
METHOD Eighty-nine psychiatric outpatients recruited from a V.A. Mental Health Clinic and 107 nonpatients recruited from among employees at the V.A. Hospital and nearby University of California at San Diego, have now completed the first 18 months of a three-year study. Every two months, ‘Psychiatry and Research Services, V.A. Hospital, San Diego and Department of Psychiatry, University of California at San Diego School of Medicine. aResearch Service, V.A. Hospital, San Diego, California. *Psychiatry Service, Brentwood V.A. Hospital, Los Angeles, and Department of Psychiatry, University of California at Los Angeles School of Medicine. ‘Psychology Service, V.A. Hospital, San Diego, and Department of Psychiatry, University of California at San Diego School of Medicine. *Reprint requests: Dr. Igor Grant, Psychiatry Service (116), V.A. Hospital, 3350 La Jolla Village Drive, San Diego, CA 92161, U.S.A. 183
IGOR GRANT, HERVEY L. SWEETWOOD, JOEL YAGER
subjects complete a Schedule of Recent Events (SRE)  and a 67-item Symptom Checklist @CL), especially designed for this project. The SCL, previously described, requires subjects to self-report the occurrence in the past two months of a large number of somatic, cognitive, affective, drug use and other behavioral symptoms on a five-point frequency scale ranging from “not at all” to “every day” [2, 4, 51. Although non-patients are younger, slightly higher in social class, and less frequently divorced than patients (Table 1), our initial findings have suggested that none of these three variables contributes in any significant way to the correlation of life events and symptoms . TABLE 1.-DEMOGRAPHIC
Characteristic Age (p < 0.001) Social class (p = 0.001) 1 II II1 IV V Marital status (p < 0.02) Never married Married Divorced, widowed or separated
Non-patients n= 107 SD. d 31.9 & 9.35 n % 6 5.6 11 10.3 61 57.0 22 20.6 7 6.5
Patients n = 89 P 44.6
S.D. & 12.6
6 23 37 18
5.6 6.7 25.8 41.6 20.2
0, 36/p 54.2
% 24.7 46.1
p = significance of group differences for characteristic
RESULTS Figure 1 illustrates symptom reports by patients and non-patients during the ll-month follow-up period. As can be seen, patients are consistently more symptomatic than non-patients (p < 0.001). Levels of symptoms over time tend to remain constant for the groups as a whole, except that patient self-report of symptoms declines in the first follow-up period. Figure 2 illustrates that patients also report far more life changes over time than do non-patients (p < 0.001). Again, whereas non-patient event levels tend to be stable, those of patients tend to decline early in the follow-up and to subRAWSCL SCORE
I 2o t
FOLLOW-UP INTERVALS IN MONTHS
sympom scores for patients and non-patients.
Patterns in the relationship
of life events and psychiatric symptoms over time
sequently stabilize. Tt is not clear at present whether declines in patient reports of symptoms and events represent true decreases in such happenings, or simply regression toward the mean.
w------..~ NON-PATIENTS -'_ 4
FOLLOW-UP INTERVALS IN MONTHS
scores for patients and non-patients.
Events and symptoms are significantly (p < 0.009) correlated at each follow-up period. These correlations range from 0.38 at the 14-month follow-up to 0.58 at the two-month follow-up. The typical correlation is in the order of 0.45. When undesirable events only are correlated with symptoms, all event-symptom correlations are significant (p < 0.001) at all time periods, their magnitude ranging from 0.39 to 0.60, with the modal correlation being 0.55. This confirms our earlier observation that undesirable events are the most powerful contributors to variability in symptoms. In order to examine more closely possible temporal relationships between occurrence of events and reporting of symptoms, we plotted event-symptom charts for each subject. Analysis of these eventsymptom charts has revealed several discrete patterns of event-symptom relationships. Types of temporal relationships Type 1. Concordant
between events and symptoms
For a large proportion of our subjects (34.7 %) events and symptoms tend to increase or decrease together during the same two-month time period. Figure 3a illustrates a typical “concordant” pattern. An event-symptom chart was rated as concordant if events and symptoms changed in the same direction in at least 5 of the 8 time intervals scored. Change had to be more than f 0.2 S.D. in T score to qualify as a real change in that variable. The presence of concordance can be confirmed mathematically by replotting symptoms and events. This procedure generates a scattergram for each subject where each point represents a given time period. The magnitude of the resultant Pearson product moment correlation can then serve as a measure of concordance (synchrony over time) of events and symptoms for that particular subject. Figure 3b replots the event-symptom information contained in Fig. 3a. The high product-moment correlation (r = 0 796) confirms that for this subject events and symptoms do, indeed, change together over time. In addition to illustrating concordance, a chart can also be described in terms of variability of symptoms or events over time and intensity of symptoms or events over time. Thus, Fig. 3a shows that both symptoms and events are highly variable over time and that both show no net decline in overall intensity for this particular subject. In other cases we have observed rather low variability and low but consistent intensity of both symptoms and events over time, while still others illustrate a curvilinear pattern in intensity over time. Type 2. Event unresponsive relationship, 21.4% of our subjects showed little symptom variation in spite of large changes in reported events. Figure 4a illustrates the event-symptom chart of one such subject. A scattergram of events vs symptoms with time as the unit of analysis yields a scatter of points (r = 0.031) confirming that there is no relationship between events and symptoms over time (Fig. 4b). relationship.
HERVEYL. SWEETWOOD, JOEL YACER
and MARVINS. GFRST
i’,CctlaogeICOre I LCUi
score (SCL 1
chart showing concordant
(type I) pattern
. . .
I = 796
l__..l_.__ml_. 40 50
.~ STANDARD SCORES
from Fig. 3a
Type 3. Antecedertt-consequent relationship. Figure 5a illustrates a symptom--event profile in which symptoms and events are apparently unrelated to each other. Yet, closer inspection shows that in 5 of 8 intervals, changes in symptoms were preceded in the previous two months by changes in events in the same direction. If we replot the same information with event scores shifted two months forward m time, the resultant pattern looks “concordant”. In Fig. 5b clear circles represent a replot of the Information contained in Fig. 5a and dark circles represent a scattergram developed from a hypothetical “phase shifted” chart. By shifting events forward two months in time we are able to convert an indifferent correlation of events and symptoms (r := 0.315) to a highly significant one (r = 0.99). Many prevailing notions of event-symptom causality in psychiatry led us to expect that a sizeable proportion of our subjects would illustrate such build-ups of life events prior to increases in symptoms.
Patterns in the relationship
of life events and psychiatric symptoms over time
--- Pr)chlnlrlc symptom ICOre (SCLI
30 ;20 1, 2
chart showing unresponsive (type ?) pattern.
STANDARD SCORES 7c
LIFE CHANGE SCORE
of unresponsive relationship
from Fig. 4a
However, with the methods we used, only a very small proportion of our population (5.4%) can be categorised clearly into the antecedent-consequent temporal pattern. Type 4. Uncertain relationship. In 38.3 % of the cases, the two investigators who studied the eventsymptom charts could not agree that any definite relationship existed between events and symptoms over time. In many cases events and symptoms had a clearly random relationship. In other cases there was such low variability both in events and symptoms that it was impossible to determine whether a relationship existed. A small number of charts showed variability in events over time, but extremely high symptom levels at all reporting periods. Since these relationships were seen mostly
IGOR GRANT, HERVEY L. SWEETWOOD, JOEL YAGER and
MARVIN S. GERST
(type 3) pattern.
STANDARD SCORES 6o
ID 879 r
Q PEARSON'S I - 315 (BEFORE SHIFTI PEARSON'S i = 997 [AFTER SHIFT1
oO. .. 0
among patients, it is possible that their distress levels were chronically high enough so that changes in events could have little additional impact. Finally, a few patients and non-patients showed shifts from what appeared to be an antecedent-consequent pattern in the first few reporting periods to a concordant pattern in the last half of the follow-up. In these cases events and subsequent symptoms tended to be of greater magnitude early in the study and became less intense toward the end of the follow-up. It is possible that such cases represent persons who ordinarily have basic type I profiles. but then change into type 3 response patterns when an unusually large number of events bombards them.
Patterns in the relationship
of life events and psychiatric symptoms over time
In this report we have presented the results from the first 18 months of a prospective investigation into the relationship of life events and psychiatric symptoms over time in a group of 89 psychiatric outpatients and 107 non-patients. When we performed measures of association between events and symptoms for all of our subjects grouped together and for psychiatric patients and non-patients separately, we found moderate correlations (in the order of r = 0.45) at each of nine 2-month time intervals. In this sense, the present results support our own previous findings  as well as those of Myers [6, 71, Uhlenhuth and Paykel [8, 91 and Brown ef al. [lo, 113, namely, that if one carefully inventories the occurrence of events and symptoms during a defined period of time for a large group of people, then meaningful correlations between events and psychiatric symptoms emerge. Generally lost in the reports of such investigations (and our own previous report is no exception) is the tremendous variability in the occurrence of events, in the experiencing of symptoms, and in the temporal relationship (or lack thereof) between such events and symptoms. Our approach to the study of individual variation has involved plotting life eventpsychiatric symptom life charts. Although we are only half way through a three-year prospective study, our findings at this point are provocative. It appears that 62% of our 196 event-symptom charts can be classified into one of three temporal typologies. We have called “concordant” those patterns where the events and symptoms wax and wane together over time. We have termed patterns “antecedent-consequent” when an upsurge of events regularly precedes symptom build-up. And we have termed “event unresponsive” those charts which showed large changes in events, but little or no changes in symptoms. Fully 32% of event-symptom charts cannot be classified at present. This does not mean that the relationship of events and symptoms for all of these subjects is random, although that certainly appears to be the case for a good proportion of them. As our study proceeds over 36 months, it may be possible to decrease the number of “uncertain” charts by identifying patterns with more complex relationships whose nature can be appreciated only after studying much longer time periods than the 18 months available to us so far. If our typologies do turn out to be stable over time, then our results, at first glance, raise interesting questions about a cherished notion held by many life events researchers and dynamically oriented clincians-that life change leads to health and symptom change. At most, in only about 40% of 196 cases (those typed as “concordant” and as “antecedent-consequent”) could life changes be said possibly to precede changes in symptoms; and, moreover, in only the 5.6% “antecedent-consequent” cases is this demonstrated with certainty. With regard to the “concordant” cases, it is possible that clues to more precise symptom-event relationships are hidden within the twomonth period. While such concordant occurrences of life events and symptoms may merely represent differing ways of labeling changes in adaptation (i.e. both life events and symptoms changes being consequences of a common as yet unidentified set of antecedents), it is also possible that the two-month time intervals used in this study are too long to permit a sufficiently detailed analysis of the actual sequencing of events and symptoms. The latter possibility is suggested by many studies that report briefer periods of time (e.g. in the order of days to several weeks) between personally significant negative life events and the onset of acute psychiatric [lo-131 or medical [14, 151disturbances.
IGOR GRANT, HERVEYL. SWEETWOOD, JOEL YAGERand MARVINS.
In a future study, we intend to study a group of subjects using an intensive weekby-week survey of life changes and symptoms. In this way, it should be possible to uncover situations where life events and symptoms follow each other in rapid succession and where the present design, with reports at two-month intervals, may obscure sequencing which truly exists. The “event unresponsive” profile is also of great interest. Assuming that this profile does not simply reflect some peculiarity in response bias to our symptom checklist by this subgroup of subjects, it is possible that these people may perceive and cope with their environment in ways which insulate them from developing a stress response syndrome. Our “event unresponsive” subjects might be similar to those individuals that Hinkle  has described as having “an almost sociopathic attitude”. These were persons who were rarely physically ill and seemed to be unusually imperturbable. Cobb  has described similar persons as employing the ego adaptive mechanisms of repression and projection. Working in a different context, Wolff et al. [IS] noted a relationship between the adequacy of psychological defense mechanisms and mean urinary 17 hydroxycorticosteroid excretion rates in parents of children with leukemia. In future explorations of event unresponsiveness we plan to study the cognitive style, coping strategies, ego adaptive mechanisms, and attribution of locus of control in persons illustrating such a pattern, to assess if and how differences in these functions help account for such individuals’ apparent imperturbability. Acknowledgements-This research was supported by the Medical Research Service of the Veterans Administration. The authors gratefully thank Alton Woo, Debi Taylor, and Dobie Higley for their efforts that led to the completion of this article. REFERENCES I. GRANT l., GERST M. and YAGERJ. Scaling of life events by psychiatric J. Psychosom. Res. 20, 141-149 (1976).
patients and normals.
2. GRANTI., GERSTM., YAGFRJ., et al. Undesirable life events and symptoms in psychiatric patients and non-patients. Unpublished. 3. RAHE R. H., MEYERM., SMITHM., ef ul. Social stress and illness onset. J. Psychosorn. Kes. 8, 35-44 (1964). 4. YAGER J., GERST M. and GRANT I. Computer generated progress reports using a patient completed symptom checklist. Ev&ulio>z 4, 78 (1977). H., KRIPKE F., GRANTI. ef al. Sleep disorder and psychobiological symptomatology 5. SWEETWOOD in male psychiatric outpatients and male non-patients. Psychosom. Med. 38, 373-378 (1976). 6. MYERSJ. K., LINDENTHALJ. J., PEPPERM. P. et al. Life events and mental status: A longitudinal study. J. Hlth Sot. Behav. 13, 398-406 (1972). I. MYERSJ. K., LINDENTHALJ. J. and PEPPERM. Social class, life events and psychiatric symptoms. In Stressful Life Events (Edited by DOHRENWENDB. S. and DOHRENWENDB. P.), pp. 191-205. John Wiley, New York (1974). 8. UHLENHUTHE. G. and PAYKELE. S. Symptom intensity and life events. Archs Gen. Psychiat. 28, 473477 (1973). 9. PAYKELE. S. Life stress and psychiatric disorder: Applications of the clinical approach. In Stressful Life Evenrs (Edited by DOHRENWFNDB. S. and DOHRENWENDB. P.), pp. 135-159. John Wiley, New York (1974). Some 10. BROWN G. W., SKLAIR F., HARRIS T. 0. et al. Life events and psychiatric disorders-l. methodological issues. Psychol. Med. 3, 74-87 (1973). Nature of 11. BRCWN G. W., HARRIST. 0. and PETO J. Life events and psychiatric disorders-II. causal link. Psychol. Med. 3, 159-176 (1973). 12. COOPERB. and SYLPHJ. Life events and the onset of neurotic illness: An investigation in general practice. Psychol. Med. 3, 421-435 (1973). 13. PAYKELE. S., PRUXIFFB. A. and MYERSJ. K. Suicide attempts and recent life events. Archs Gen. Psych&
Patterns 14. ENGEL G.
in the relationship
rapid death during psychological stress: Folklore or wisdom? Ann. (1971). CASTELNUOVO-TEDESCO P. Emotional antecedents of perforation of ulcers of the stomach and duodenum. Psychosom. Med. 24, 398-416 (1962). HINKLE L. E., JR. The effect of exposure to culture change, social change and changes in interpersonal relationships on health. In Stressful Lifi Events (Edited by DOHRENWEND B. S. and DOHRENWEND B. P.), pp. 9-44. John Wiley, New York (1974). COBB S. A model for life events and their consequences. In Stressful Life Events (Edited by DOHRENWEND B. S. and DOHRENWEND B. P.), pp. 151-156. John Wiley, New York (1974). WOLFF C. T., FRIEDMAN S. B., HOFER M. A. et al. Relationships between psychological defenses and mean urinary 17-hydrocorticosteroid excretion rates. Part I. A predictive study of parents of children with leukemia. Psychosom. Med. 26. 576-591 (1964).
Intern. Med. 74, 771-782
of life events and psychiatric