Journal of Affective Elsevier Biomedical
Disorders, 4 Press
(1982)365-37 1
365
The NIMH Diagnostic Interview Schedule Modified to Record Current Mental Status Michael R. Von Korff and James C. Anthony Department
of Mental Hygiene, Johns Hopkins School of Hygiene and Public Health, 615 N. Wolfe Street, Baltimore, MD 21205 (U.S.A.) (Received 25 January, 1982) (Accepted 22 April, 1982)
Summary The National Institute of Mental Health Diagnostic Interview Schedule (DIS) was modified to record detailed information on current mental status in addition to the lifetime symptom history. Use of the modified DIS in a field survey indicates that information on current symptoms is required to distinguish persons who meet all DSM-III criteria for Major Depressive Episode at or around the time of interview from former cases who fail to meet all criteria at interview. Thus, the unmodified DIS may overestimate the l-month period prevalence rates for Major Depressive Episode, by counting symptomatic former cases as having the disorder at or around the time of the interview. An analysis of symptom count data also suggests that the unmodified DIS count of lifetime depressive symptoms is not a good measure of current symptom status.
Introduction The National Institute of Mental Health Diagnostic Interview Schedule (DIS) was developed by Lee Robins and her associates for use in community surveys and other research concerning mental disorders (Robins et al. 1981). The DIS looks promising for improved large scale mental disorder surveys. It covers the DSM-III criteria for a selected set of important mental disorder categories and provides information pertinent to DSM-III diagnoses. Error is controlled through standardized wording and sequence of questions. The interview can be administered by trained lay interviewers.
This work was supported
by NIMH
Grant
MH-33870.
366
One notable limitation of the DIS is that it does not assess how recently individual reported symptoms have occurred; the DIS elicits lifetime symptom reports. This limitation is of concern because detailed information on current mental status is essential for many of the analytic objectives of surveys of the distribution of mental disorders and other research concerning the manifestations, causes, and effects of mental disorders. This methodological report discusses some of the advantages of collecting information on current symptoms, and describes a simple change used to elicit and record this information when using the DIS. It also presents data from a field survey that used the modified DIS, and gives some evidence that such information has practical, as well as conceptual importance.
Importance of Information on Current Mental Status Gruenberg (1963), Goldberg (1972), Wing et al. (1974) Williams et al. (1980) and others have offered reasons for drawing a distinction between recent and past symptoms of mental illness. In the context of epidemiological studies of mental disorders, this distinction permits one to identify current mental disorder cases who meet all diagnostic criteria at or around the time of interview, distinguishing them from past cases with limited or no residual symptoms. The distinction also permits study of temporal sequence and clustering of symptoms, a task which has special importance in psychiatric research, including studies of course, treatment response, classification, and diagnostic criteria. Moreover, if one records recent symptoms, it is possible to index recent distress, as in a count of current symptoms. Compared with a count of symptoms experienced over an entire lifetime without regard to recency, a count of current symptoms can be more readily interpreted as a predictor or correlate of concurrent and subsequent events and behavior, such as the use of drugs or health services. The unmodified DIS does not differentiate past and recent symptoms, except to date the most recent symptom or spell, and for three disorders, to date the worst episode of the disorder ever experienced. As a result, the diagnostic algorithms may group current cases meeting all diagnostic criteria at the time of interview with past cases which have only limited residual symptoms or emotional distress unrelated to the prior disorder. Furthermore, the unmodified DIS provides no count of recent symptoms. With these considerations in mind, we made a change in the DIS to obtain information on the recency of occurrence of individual symptoms. This modified DIS is being used in the field investigations of the Eastern Baltimore Mental Health Survey (EBMHS) of the NIMH Epidemiologic Catchment Area (ECA) Program. In the EBMHS to date, a probability sample of more than 3000 household residents has been interviewed with the modified DIS. More than 800 subjects, selected as a probability sample of the household interviewees, have also been carefully examined by one of a team of psychiatrists with no knowledge of the DIS results. In addition, the modified DIS is being used at the Duke and UCLA ECA program sites, and in other NIMH sponsored research.
367
The Modification to the DIS The DIS ‘Probe Flow Chart’ is used when interviewees report particular symptoms (Robins et al. 1981). The Probe Flow Chart specifies a branching sequence of follow-up questions to assess whether a symptom meets pre-specified criteria for severity, and whether it can be attributed to ‘physical illness, injury, alcohol, or drugs’. Our basic innovation was to add one question to the Probe Flow Chart: ‘When was the last time you had [INSERT SYMPTOM] (when it was not due to a
542.
Was
there
ever
a period
of
weeks when your interest in was a lot less than usual? VOLUNTEERS
NO INTEREST
several sex -
EVER = 96
PieSENT No . . . . . . . . . . . . Blw
wit
PRESENT 01
. . . . . . . . . . .
02 03
. . . . . . . Med exp Both . . . . . . . . . .
04 05
No Interest
96
Drugs/ale
. . .
-
2 uks . . . . . 1 mo . . . . . . 6 mcs . . . . .
06 07 08
l
-
1 yr
. . . . . .
09
l
l+
yrs
. . . . . .
10
2 wks . . . . . 1 m . . . . . .
06 07
a
RF . . . . . . . . . . . . 97
543.
Has two
there weeks
worthless,
ever been a period of or mre when you felt sinful
OF guilty?
DK . . . .._.._.._
98
No ............ RF ............ DK ............
01
-
97 98
Has
there
ever
two
weeks
OF mere
been
a when
period you
of had
No ............ ..... Drugs/ale M?d exp .......... Both RF ............ DK ............
.......
than
Fig.
1. The
is
normal
for
you?
DIS Probe Flow Chart and response coding categories.
01 03 04 05 97 98
6 ams 1 yr . . . . . . l+ yrs . . . . . .
09 10
-
06
l l
l+
lo9’
l
lD9’
. . . . . 68
-
544.
l
2 1 6 1
wks . . . . . mo ...... mos ..... yr ...... yrs
......
07 08 09 10
368
physical condition or to using medication, drugs or alcohol)?‘. The phrase in parentheses is read only when the DIS specifies its applicability for the symptom reported. The response to this probe is assigned to one of 5 coding categories (within the past 2 weeks, month, 6 months, 1 year, or more than 1 year ago). The section on drug abuse does not use the probe flow chart, so we inserted simple questions to secure recency information. The sections on anorexia nervosa, anti-social personality, tobacco dependence, gambling and psycho-sexual disorders were not modified. Lee Robins examined the modified DIS and reported that its diagnostic results should be comparable to those of the DIS. The DIS-computerized diagnostic algorithms have been restated for the modified DIS without changing their logic or specifications.
Methods The modified DIS was employed during EBMHS field work which began in January 1981. A probability sample of 3817 eastern Baltimore household addresses was identified for the survey. The sampling plan called for selecting all household members aged 65 years and older, as well as 1 household member aged 18 through 64 years, who was selected at random using a Kish selection table. Sampled respondents were asked to participate in a 90-min interview which included the modified DIS. Household rosters were obtained from 3618 of the households in the sample for a screener completion rate of 95%. From the households providing roster information, 4238 persons were selected for interview. Of these persons, 3481 completed interviews for a completion rate among identified respondents of 82%. The overall completion rate of 78% is estimated by the product of the screener and respondent completion rates.
Results and Discussion Comparison of the DIS Diagnosis of current Major Depressive Episode and current Depressive Symptoms DSM-III criteria for Major Depressive Episode (296.2x, 296.3x) require the presence of clinically significant dysphoria with symptoms from at least 4 of 8 groupings of depressive symptoms experienced during the episode (American Psychiatric Association, 1980). The DIS follows these rules. If subjects’ responses to the DIS questions suggest that they have ever met the DSM-III criteria for Major Depressive Episode, the DIS diagnostic algorithm specifies whether the Major Depressive Episode is present at the time of the interview by the response to this pair of questions: ‘Are you in one of these spells of feeling low and disinterested and having some of these other problems now?’ and ‘When did your last spell like that end?‘. Subjects who say they have had such a spell in the month prior to interview are classified as cases of Major Depressive Episode occurring in the month prior to
369 TABLE
1
NUMBER OF DISTINCT DEPRESSIVE SYMPTOMS OCCURRING IN THE MONTH PRIOR TO INTERVIEW FOR PERSONS CLASSIFIED BY THE DIS ALGORITHM AS HAVING A MAJOR DEPRESSIVE EPISODE IN THE MONTH PRIOR TO INTERVIEW Cases identified in the Eastern Baltimore Mental Health Survey. Number of depressive symptoms in month
Number of cases
Cumulative percent of cases
0
4 5 9 8 13 8 7 2
7.0 15.8 31.6 45.6 68.4 82.5 94.7 96.5 100.0
57
100.0
1 2 3 4 5 6 7 8 Total
interview,
1
provided they have met the other DIS requirements for the disorder at some time in their lives. This is true even though the current spell may not meet all of the DSM-III criteria for Major Depressive Episode. In the EBMHS household interviews, the DIS algorithms classified 57 persons as having a Major Depressive Episode in the month prior to interview. Table 1 displays the number of distinct depressive symptoms reported by these 57 DIS-nominated cases for the same period of time. While all of these cases reported a spell of feeling depressed with 4 or more distinct depressive symptoms at some time in their lives, only 31 of the 57 (54%) had 4 or more distinct depressive symptoms present in the month prior to interview. If the l-month period prevalence for a disorder is to represent the number of cases who meet diagnostic criteria during the month in relation to population size, then these findings suggest the possibility that the unmodified DIS may appreciably overestimate the l-month period prevalence of Major Depressive Episode. It would do so by failing to segregate past cases with limited or no residual symptoms from cases who met criteria during the month prior to interview. It is conceivable that a similar problem of period prevalence overestimation may occur whenever the period of interest is less than a person’s lifetime, though the problem may become less acute as the duration of the period is increased. However, with the modified DIS, and with DIS algorithms modified to take the symptom recency information into account, this problem of overestimation can be readily avoided. Before turning from this issue, we should note that the unmodified DIS is not unique with respect to unclear distinctions between former cases, past cases with
370
limited residual symptoms, and current cases. The DSM-III does not give very much attention to this problem. Moreover, we recognize that research aims may necessitate variation in these distinctions: in a study of a community’s treatment needs, a case definition based upon past, current and residual symptoms may be appropriate. Comparison of lifetime and current symptom counts Most measures of depressed mood in current use elicit a tally of reports of symptoms present at the time of interview or during a week or two prior to interview (e.g., CES-D, SCL-90). The unmodified DIS yields a similar tally for the respondent’s lifetime, but not for any shorter period other than the ‘worst spell’ of feeling depressed, which is dated only in terms of subject’s age at that spell. This would not be important if there were a high correlation between the lifetime tally and the tally for a more recent period. Unfortunately, this does not seem to be the case. For the 3323 respondents who completed the DIS section on depressive symptoms, the Pearson correlation coefficient of the count of depressive symptoms present in the month prior to interview provided by the modified DIS and the lifetime symptom tally was only 0.66. Table 2 shows that the poor correlation is due in part to the large numbers of persons who report fewer than 3 depressive symptoms in the month prior to interview while reporting more depressive symptoms on a lifetime basis. There were 183 persons who had a lifetime symptom count of 4 or more, but who reported no depressive symptoms in the month prior to interview.
TABLE
2
LIFETIME DEPRESSIVE SYMPTOM COUNT THE MONTH PRIOR TO INTERVIEW 158 cases missing data were persons unable physical illness or cognitive impairment. Lifetime
One-month symptom count
symptom
BY THE COUNT
to complete
OF DEPRESSIVE
a full DIS interview,
usually
SYMPTOMS
because
of severe
count
0
1
2
3
0 1 2 3 4 5-7 8-10 11+
1462
493 162
224 129 63
111 53 36 21
48 35 26 27 I
Total
1462
655
416
221
143
4
5-l
IN
8-10
ll+
Total
82 36 39 29 33 39
39 11 10 15 9 18 10
14 4 2 2 4 9 15 6
2473 430 176 94 53 66 25 6
258
112
56
3 323
371
Conclusions
and Suggested Future Work
Retaining its strengths, we have modified the DIS to provide a record of a respondent’s current mental status and the recency of individual symptoms. We have used EBMHS data to suggest that the unmodified DIS may overestimate the prevalence of current cases of Major Depressive Episode unless recency of individual symptoms is taken into account and unless there is attention to the distinction between subjects who do and do not meet diagnostic criteria during the period of interest. In addition, these data suggest limited utility of the unmodified DIS’ lifetime symptom count as an index of current mental status. In future work, the problems we have illustrated with respect to Major Depressive Episode will be investigated for the DIS sections on Manic Episode, Alcohol and Drug Abuse or Dependence, and the Schizophrenic Disorders. We expect the modified DIS to make the DIS data record more useful through (1) a better ability to assess when diagnostic criteria have been met, (2) an improved view of the course of disorders and transitions from one mental state to another, (3) an improved capacity to evaluate and improve diagnostic criteria, and (4) the provision of current symptom counts. Future DIS users should consider added questions on dates of first and last symptom experiences, as well as symptom severity, which could also improve the utility of the DIS data record.
Acknowledgements Suggestions from Morton Kramer, Sam Shapiro, Helen Orvaschel, Marshal Folstein and Jay Burke are gratefully acknowledged. Field work for the Eastern Baltimore Mental Health Survey was conducted by Survey Research Associates of Baltimore, MD.
References American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 3rd edition, Washington, DC, 1980. Goldberg, D.P., The Detection of Psychiatric Illness by Questionnaire, Cambridge University Press, London, 1972. Gruenberg, E.M., A review of mental health in the metropolis - The Midtown Manhattan Study, Milbank Memorial Fund Quarterly, 16 (1963) 77-93. Robins, L.N., Helzer, J.E., Croughan, J. and Ratcliff, K.S.. National Institute of Mental Health Diagnostic Interview Schedule, Arch. Gen. Psychiat., 38 (1981) 381-389. Williams, P., Tarnopolsky, A. and Hand, D., Case definition and case identification in psychiatric epidemiology - Review and assessment, Psychol. Med., 10 (1980) 101-I 14. Wing, J.K., Cooper, J.E. and Sartorius, N., Measurement and Classification of Psychiatric Symptoms, Cambridge University Press, London, 1974.