Neurosis in Soldiers - A Follow-up Study

With Aubrey Lewis

The Lancet, April 25, 1942, pp. 496-499


   In neurosis centres where most of the patients are soldiers, the success of treatment must be judged by the proportion of those treated who are able to resume and continue at military duties. It is misleading to judge it by the proportion of those treated who, by the time they leave hospital, are either free of symptoms or merely thought likely to be able to carry out their military duties. Neurotic illness is so dependent on circum­stances, and especially on the stresses of the soldier's life, from some of which he is free while in hospital, that his condition after some weeks in this relatively sheltered place is not a sure indication of his fitness. Conclusions about the success of treatment must therefore be based on condition 3 months, 6 months or more after discharge. As with the assessment of treatment so with the assess­ment of prognosis. It is not sufficient to weigh the likely prognostic factors in men who are apparently well on discharge from hospital against those in men who are invalided from the Army by a medical board because the hospital doctors and others think them unsuitable that would only show the factors determining doctors' opinions about probable fitness for military duty. It is necessary to compare groups in whom a trial of fitness after treatment has been made in the Army and the outcome ascertained.

   In each of the two hospitals in which we work two groups were compared: one made up of men who on inquiry some months after discharge from hospital were doing full duty in the Army efficiently and willingly; and the other of men who sooner or later after return to duty had been discharged from the Army under para. 390 (XVI) K.R. 1940 as ceasing to fulfil Army physical requirements. The groups consisted in the one hospital of 100 men each, and in the other hospital of 50 men each; in all, therefore, 150 men on full duty were compared with 150 men who had been discharged. The samples were selected by taking all successive names, up to the required number, in the lists of men on whom the follow‑up questionnaire had been answered by the commanding officer of the man's unit. The records of these men had been compiled by many psychiatrists in the two hospitals; the extraction of data from the records was done by one person in each hospital. It was therefore unlikely that in the group as a whole individual bias could have greatly influenced what observations were made and recorded and unlikely that the extraction of the relevant data from the records was done incon­sistently. The conditions were thereby laid down for a statistically unobjectionable controlled experiment, such as is but seldom possible in clinical medicine.

   We took, first, 35 attributes which from experience and common belief might be thought of prognostic signific­ance in cases such as these. It soon became evident that some of them could not profitably be included; a reduced list of 20 attributes was finally taken, and the presence or absence of each attribute in the individual patients determined (table I). Combining our figures. the significant differences between the "full duty” sample and the "discharged" are in respect of these attributes: positive family history; unsatisfactory work record; psychopathic traits of personality; symptoms of the present illness before enlistment; previous dislike of Army service, keenness to get back to Army duty; exposure to Dunkirk or other severe stress of bombard­ment and active service; querulous hypochondriacal complaints; amnesic phase; surly paranoid attitude.


   In every case the difference is to the disadvantage of the discharged group, keenness to get back to the Army and exposure to severe stress counting as favourable attri­butes, the rest as unfavourable ones. In the case of all these attributes the difference between the two groups is more than twice its standard error, the actual ratio being given in the last column of the table; but a glance down the table shows that there are consistent differences between the, two groups in respect of all the other attri­butes, though they are not large enough to have, individually, statistical significance. In a few cases the difference is trifling‑for instance, there is little difference shown in the average time spent in hospital, and in the frequency of anxiety symptoms or of an adequate precipitating cause (a favourable feature). In 2 cases the difference between the two groups is not quite what one might expect. The presence of depressive symptoms is shown to be favourable rather than other­wise ; and there is a surprisingly small difference between the two groups in the frequency of the grosser degrees of stupidity as shown by school record. One of us who first employed a less strict criterion (standard 6 or lower) obtained a larger difference between the two groups but it was thought better to maintain the general rigour of definition throughout the table. A " positive family history," for instance, means that at least one person among the patient's parents or siblings had an actual psychiatric illness for which he has been treated, or had shown a sufficiently gross degree of psychiatric abnor­mality to lead to social consequences. The appearance of any attribute in the table depends on its having been reported in the notes : it is therefore likely that the figures, say, for" dislike of Army service," "keenness to return to the Army" or "querulous hypochondriasis" indicate only those in whom these features were prominent. In one of our hospitals only half the material could be utilised for the recording of the first two of these attributes‑i.e., only those cases admitted after a time when the recording of these factors had shown itself to be important.


    Two different collections of statistical material cannot be combined for analysis unless it can be shown that there are not significant differences between the one collection and the other. Therefore our two collections were compared with each other (table II). It was found that the difference between the two hospitals was more than twice its standard error in respect of the following attributes : unsatisfactory work records, symptoms before enlistment, adequate external cause (difference significant in both full duty and discharged groups); and previous psychopathic traits, dislike of the Army, querulous hypochondriasis and fugue (difference significant in the discharged group only). It is probable that these differences are due to systematic differences between the two hospitals in the attention paid to these attributes, or the degree to which they had to be present before they were recorded, or possibly to a difference in the nature of the clinical material. It would therefore be unsafe to combine the two sets of figures for the purpose of getting any absolute pro­portions. This procedure would in any case be of little value because one has no wish to know the absolute frequency of, say, a positive family history in a special group of neurotics who were returned to duty, though its frequency in the general run of military neurotic patients would admittedly be of interest. Furthermore our method of recording the presence or absence of an attribute did not provide for any variation in its intensity and such findings as "poor work recordmight have been much more widely assessed by other investigators.


    But we are not concerned here with absolute frequencies but with differences or relative frequencies, and for this purpose our two collections of material can be safely combined. When, of all persons showing a given attribute, the proportion having a bad subsequent record is compared with the proportion who subsequently did satisfactorily (table III), the differences between the two hospitals fade away. In no case do the proportions calculated in the one hospital and the other differ by an amount approaching twice its standard error. We can therefore say that in demonstrating the psychiatric differences between men with good and bad posthospital records the findings of our two hospitals confirm each other.


    The failure, already mentioned, to allow for different intensity of manifestation probably weakens our next step, which was to see whether the ominous attributes (those which were significantly more common in the men discharged from the Army) occurred together with such frequency in individuals that a mere counting of them might have prognostic value. In other words, could it be shown that the occurrence in any individual of, say, five or more ominous attributes indicated an almost certain discharge from the Army, or that the occurrence of not more than one of the attributes in any individual meant that he had a good chance of doing well? The weakness of this lies in the lumping together of attributes irre­spective of their degree of severity in the individual patients, and in the concentration on ominous attributes without paying attention to the auspicious attributes. We were aware of this omission, but considered it im­practicable to deal statistically with the large number of favourable factors, though of course these had played the chief part in bringing about the permanent or temporary return to military duties, on discharge from hospital, of almost all the patients studied. One of us first com­pared the number of men in each of his two groups of 50 who had one, two, or more, ominous attributes, with the following results:


   It is evident from this analysis that 72% of those who did well in the Army had not more than four attributes, and that 68% of those who did badly in the Army had five or more attributes. The result has a high degree of statistical significance (χ2 = 16 ; for one degree offreedom P <0‑01).

   One important purpose of such an analysis is to indi­cate the criteria on which unsuitability for the Army may be detected before actual breakdown has occurred. The list of attributes, however, included symptoms such as hypochondriasis which if persistent would be enough to make Army duties impossible. It was therefore neces­sary to make a reduced list eliminating from among the ominous attributes those which could be directly and solely decisive against doing well in the Army. These were active dislike of the Army and unwillingness to resume duty, hypochondriasis, amnesia and surly para­noid attitude. Taking in each hospital 50 of the men on full duty and 50 discharged and combining them, the numbers in the reduced list were as follows:

Here the difference in distribution of the two groups is not so striking, nor (lees it teach such a high degree of statistical significance, though still an adequate one (χ2 = 146 ; for 6 degrees of freedom P <0025). It will be seen that 74% of those on full duty have 3 or less of the ominous attributes listed in this table.

   Because of these findings, at one of the hospitals the records were re‑examined of those men who had been sent back to the Army in spite of their possessing six or more of the ominous attributes, and who had subse­quently been discharged from the Army. In 6 instances, it was found, the psychiatrist who had been treating the patient in the neurosis centre had advised his discharge from the Army, but the military board had not agreed with this opinion and had sent the man back in a lower category. In the other instances the psychiatrist had either recommended that the man should be tried in a lower category on home service or had been influenced by the man's keenness to return to military duty. In 7 of them the fact that the man had improved rapidly while in hospital and was almost free from symptoms had led to his return, although the prognosis for military duty could be recognised as decidedly poor. On the other hand the one patient with less than two ominous attri­butes who had been subsequently discharged was found to have hysterical conversion symptoms which in them­selves were sufficient to make military duty impossible.

*         *         *

   It is important to recognise that the patients studied had been discharged from hospital before the procedure was introduced whereby a neurotic soldier, previously engaged in duties for which he was unfitted, could be transferred to more suitable duties on the psychiatrist's recommendation. There are grounds for believing that the results obtained since that procedure has been used are more satisfactory, and that criteria of prognosis will have changed correspondingly. It is obvious that prognosis must be influenced by the conditions under which the soldier will be working when he resumes duty on leaving hospital.


    At two hospitals for the treatment of neurosis a group of soldiers who on return to the Army after treatment were still on full duty some months later were compared with a group of those who had subsequently to be discharged from the Army. The medical records of 300 soldiers (150 in each group) were investigated, and 10 features determined which occurred with significantly different frequency. There were consistent differences in the frequencies of other features, which were however, individually, not statistically significant.

   Comparison of the separate data of the two hospitals confirmed this.

   The traits which were found with significantly greater frequency in men who proved unsuitable for military duty were: a history of mental disturbance, including neurosis, in parents or siblings; unsatisfactory work record prior to enlistment; psychopathic traits of per­sonality; symptoms of the present illness before enlist­ment; resentment or strong dislike of Army life; reluctance to return to Army duties; onset of the illness without its having been precipitated by exposure to bombardment, continuous danger, and other stresses of active service; querulous liypochondriasis; fugue or amnesia; and surly or paranoid attitude.

   The fewer of these attributes a patient has the greater the likelihood of his remaining in the Army. The evidence however does not indicate that prognosis can be safely based on the number of these attributes a patient is found to possess. The intensity with which each such attribute has been exhibited, its nature, and the favourable attributes also discovered must be taken fully into account in each case.

   The findings do not pay regard to the varying con­ditions of Army life, to which soldiers return; these also are important in determining the course of the man's illness and successful adaptation to Army requirements. The groups investigated were dealt with before a new and valuable procedure had been introduced permitting a measure of adjustment to be made on the psychiatrist's recommendation.