Journal of Affective Disorders, 1 (1979) 3-8 @ Elsevier/North-Holland Biomedical Press
SHOULD ‘NON-FEIGHNER AFFECTIVE DISORDER?
WILLIAM Department Road, Iowa
CORYELL
and MING
3
SCHIZOPHRENIA’
BE CLASSIFIED
WITH
T. TSUANG
of Psychiatry, The University City, Iowa 52242 (U.S.A.)
of Iowa
College
of Medicine,
500 Newton
SUMMARY Narrow definitions of schizophrenia increase homogeneity at the expense of leaving unclassified many patients with schizophrenic symptoms. Family history and follow-up studies indicate that many such patients ought to be classified with those having affective disorders. This study determines morbid risks for affective disorder and schizophrenia in first degree relatives of patients with chart but not research diagnoses of schizophrenia. Comparisons with morbid risk figures for relatives of individuals satisfying research criteria for depression, mania or schizophrenia indicate that the ‘non-Feighner schizophrenia’ group is probably too heterogenous to be classified entirely as affective disorder or as schizophrenia.
INTRODUCTION
Broad definitions of schizophrenia tend to encompass many cases with features traditionally considered atypical of schizophrenia (Taylor and Abrams 1975a). These features include affective symptoms, acute onset of illness, a prior history of remission and the presence of confusion or perplexity. Cases with schizophrenic symptoms have a better prognosis if one or more of these features are present (Johanson 1958; Vaillant 1962; Stephens et al. 1966). A variety of studies have also shown many such cases to have family histories with high rates of affective disorder (Clayton et al. 1968; Fowler et al. 1972; Taylor et al. 1974). In order to increase homogeneity in groups designated schizophrenic, a narrow definition proposed by Feighner et al. (1972) has incorporated these atypical features as exclusion criteria. These criteria, then, define poor prog-
This work was supported in part by NIMH grant No. MH31673. Dr. Coryell is an Assistant Professor of Psychiatry and Dr. Tsuang is a Professor chiatry. Reprint requests to: Dr. Coryell, Department of Psychiatry, The University College of Medicine, 500 Newton Road, Iowa City, IA 52242, U.S.A.
of Psyof
Iowa
4
nosis subgroups from larger groups designated schizophrenic by broader definitions (Winokur and Tsuang 1975). In clinical usage, however, the Feighner criteria leave many cases unclassified (Taylor and Abrams 1975b). How should such excluded cases be classified? Since studies have shown strong associations between ‘good prognosis schizophrenia’ and affective disorder and since the Feighner criteria for schizophrenia were designed to exclude cases without a poor prognosis, then many cases so excluded probably represent variants of affective disorder. Croups of such excluded cases may or may not be homogenous enough for routine classification with the affective disorders, however. The present study uses family history data to address the issue of homogeneity in a large group defined by a failure to meet Feighner criteria despite a chart diagnosis of schizophrenia. METHODS
The authors of an earlier paper reviewed a consecutive series of inpatient charts from the University of Iowa Psychopathic Hospital dating from 1934 to 1944 to distill groups meeting research criteria for primary affective disorder and schizophrenia (Morrison et al, 1972). Two hundred charts met criteria for schizophrenia of Feighner modified by Morrison et al. (1972). This left 315 charts which did not, despite a discharge diagnosis of schizophrenia. Of this group of 315, five were recently reclassified leaving 310 here designated as ‘non-Feighner schizophrenics’. During the decades between 1930 and 1950 social workers and physicians routinely obtained extensive family history information for each admission to the University of Iowa Psychopathic Hospital. This information was type written under separate headings for each first degree relative and often extended for several pages. In an earlier paper specially designed criteria were applied to the family history information of the above mentioned probands meeting criteria for affective disorder or schizophrenia. The diagnosis of schizophrenia in a relative required an illness (1) beginning before the age of 40 years, (2) leading to chronic hospitalization or chronic social incapacity, (3) involving sustained bizarre behavior, delusional thinking or considerable seclusiveness and (4) lacking evidence of marked affective symptoms. If a relative suffered from an illness with common affective symptoms from which he recovered without significant personality defect, with or without psychiatric hospitalization, he was considered to have an affective disorder (Winokur et al. 1972). One of the present authors (WC) applied these same criteria to the charts of the ‘non-Feighner schizophrenics’. The present study uses the Weinberg shorter method to correct for age distribution in the calculation of morbidity risks for schizophrenia and affective disorder. This method deducts from the total number of a class of relatives all those who have not entered the age of risk and half the number who are between the upper and lower limits of the age of risk. These limits are 15 and 60 years of age for affective disorder and 15 and 40 years of age for
5
schizophrenia. Among the first degree relatives there were few children who had entered the age of risk for either schizophrenia or affective disorder. For similicity, ‘relatives’ will refer here only to parents and siblings. The statistical method used for comparing morbid risk of family members of the various proband groups was a comparison of two proportions. A P value of less than 0.05 was considered statistically significant. RESULTS
Affective disorder was significantly more frequent in the relatives of nonFeighner schizophrenics than in those of Feighner schizophrenics as shown in Table 1. Affective disorder in the relatives of non-Feighner schizophrenics was significantly less frequent than in the relatives of depressives. was significantly Table 2, on the other hand, shows that schizophrenia more frequent among relatives of non-Feighner schizophrenics than among those of the depressives. Morbid risk for schizophrenia in relatives of the non-Feighner schizophrenics was not significantly different than that for relatives of the Feighner schizophrenics. Relatives of manic probands were somewhat more likely than those of ‘non-Feighner schizophrenics’ and less likely than those of depressive probands to have affective disorder. The inverse was true regarding morbid risk for schizophrenia among relatives of manic probands. None of these differences were significant, however.
TABLE
1
MORBIDITY RISKS FOR AFFECTIVE OF DIFFERENT PROBAND GROUPS
DISORDER
Number cases of affective disorder
of
IN FIRST
Number of relatives
DEGREE
BZ
RELATIVES
MR (%)
Schizophrenic
probands
32
1017
584
5.48 a
(N = 200) ‘Non-Feighner
schizophrenic’
76
1692
885
8.59
probands (N = 310) Manic probands
37
519
328
11.28
(N = 100) Depressive probands
123
1413
859
14.32
b
(N = 225) Primary affective
160
1932
1187
13.48
b
probands a Sig. diff. b Sig. diff.
disorder
(N = 325) from MR for relatives from MR for relatives
of non-Feighner of non-Feighner
schizophrenics, schizophrenics,
P < 0.05. P < 0.01.
6 TABLE
2
MORBIDITY RISKS PROBAND GROUPS
FOR
SCHIZOPHRENIA
Schizophrenic probands (N = 200) ‘Non-Feighner schizophrenic’ probands (N = 310) Manic probands (N = 100) Depressive probands (N = 225) Primary affective disorder probands (N = 325) a Sig. diff. from MR for relatives
IN 1” RELATIVES
OF DIFFERENT
Number of cases of schizophrenia
Number relatives
13
1017
617
2.11
25
1692
1112
2.25
5
519
434
1.15
5
1413
1228
0.41 a
10
1932
1662
0.60 a
of non-Feighner
schizophrenics,
of
BZ
MR (%)
P < 0.01.
DISCUSSION
Tsuang et al. (1976) presented family history data for 85 of the present group of 310 probands in an earlier report. At index admission these cases had had affective symptoms and/or a history of remission. Parents and siblings of these 85 probands resembled those of the primary affective disorder group in terms of morbid risk for affective disorder. In contrast, patterns of illness in the relatives of the larger group of 310 are dissimilar to those of relatives of both the schizophrenic and primary affective disorder groups. In terms of percent recovered at follow-up also, the above group of 85 is closer to the primary affective disorder cohort then the larger group of 310. Morrison et al. (1973) while blind to research diagnosis evaluated follow-up material routinely obtained after index admission. He found that 22% of the group failing to meet Feighner criteria for schizophrenia had recovered during the follow-up period averaging 3.2 years. The proportions recovered in the unipolar and bipolar affective disorder cohorts were 54% and 59%, respectively. Tsuang et al. reported the recovery rate for the subgroup of 85 to be 44%. These family history and follow-up findings suggest that, though subgroups of cases excluded by narrow definition of schizophrenia may be appropriately grouped with the affective disorders, the excluded group as a whole is too heterogeneous to be so classified. Affective disorder in this study was found less frequently among family members of manics than among those of depressives while the reverse WAS true for morbid risks of schizophrenia. These findings are quite different from those of Perris (1966) who found more affective disorder and less schizophrenia among family members of bipolar probands than among fam-
7
ily members of depressives. Differences in selection criteria may explain some/of the incongruity. Perris required that bipolar probands have a prior history of both depressive and manic episodes with intervening periods of clear remission. Since the Feighner criteria have no such requirement, it is possible that a proportion of the mania cohort was misclassified. Indeed Winokur found that 6 of the 66 manic probands (9%) with available followup information greater than one year later received from other clinicians a diagnosis of schizophrenia (Winokur 1974). The rate of clinical diagnostic revision in the depressive cohort was much lower (3%). Also, a research design relying upon chart review of family histories may result in misclassification of bipolar illness in first degree relatives of schizophrenia. Individuals with rapidly cycling bipolar illness might well have been described by family members as chronically strange and would have been at high risk for long term hospitalization in the 1930’s and 1940’s. Family history studies such as these lack the power to suggest specific changes in boundaries of diagnostic concepts. An extensive follow-up family study of the ‘non-Feighner schizophrenics’ is underway. The resulting data in conjunction with careful assessment of clinical features recorded at index admission will be used to suggest refinements of the existing research criteria. Hopefully, the results will move the controversies surrounding the terms ‘schizoaffective’ and ‘atypical schizophrenia’ closer to resolution. ACKNOWLEDGEMENT
The authors are grateful tistical analysis.
to Jerome
Fleming, M.S., for his assistance
in sta-
REFERENCES Clayton, P.J., Rodin, L. and Winokur, G. Family history studies, Part 3 (Schizoaffective disorder, clinical and genetic factors including a one to two year follow-up), Compreh. Psychiat., 9 (1968) 31-49. Feighner, J.P., Robins, E., Guze, S.B., et al., Diagnostic criteria for use in psychiatric research, Arch. gen. Psychiat., 26 (1972) 57-63. Fowler, R.C., McCabe, M.S., Cadoret, R.J. and Winokur, G., The validity of good prognosis schizophrenia, Arch. gen. Psychiat., 26 (1972) 182-185. Johanson, E., Chapter II, Presentation of the material. Acta psychiat. neurol., 33 (Suppl. 125) (1958) 23-47. Morrison, J., Clancy, J., Crowe, R. and Winokur, G., The Iowa 500, Part 1 (Diagnostic validity in mania, depression, and schizophrenia), Arch. gen. Psychiat., 27 (1972) 457-461. Morrison, J., Winokur, G., Crowe, R. and Clancy, J., The Iowa 500 - The first follow-up, Arch. gen. Psychiat., 29 (1973) 678-682. Perris, C., A study of bipolar (manic-depressive) and unipolar recurrent depressive psychosis, Part 1 (Genetic investigation), Acta psychiat. stand., Suppl. 194 (1966) 15-44. Stephens, J.H., Astrup, C. and Mangrum, J.C., Prognostic factors in recovered and deteriorated schizophrenics, Amer. J. Psychiat., 122 (1966) 1116-1121.
8 Taylor, M.A. and Abrams, R., Manic-depressive illness and good prognosis schizophrenia, Amer. J. Psychiat., 132 (1975a) 741-742. Taylor, M.A. and Abrams, R., A critique of the St. Louis psychiatric research criteria for schizophrenia, Amer. J. Psychiat., 132 (1975b) 1276-1280. Taylor, M.A., Gaztanaga, P. and Abrams, R., Manic-depressive illness and acute schizophrenia - A clinical, family history, and treatment-response study, Amer. J. Psychiat., 131 (1974) 678-681. Tsuang, M.T., Dempsey, M. and Rauscher, F., A study of ‘atypical schizophrenia’, Arch. gen. Psychiat., 33 (1976) 1157-l 160. Vaillant, G.E., The prediction of recovery in schizophrenia, J. nerv. ment. Dis., 135 (1962) 534-543. Winokur, G., Diagnostic stability over time in schizophrenia, mania and depression, New Engl. J. Med., 290 (1974) 1027. Winokur, G. and Tsuang, M.T., A clinical and family history comparison of good outcome and poor outcome schizophrenia, Neuropsychobiology, 1 (1975) 59-64. Winokur, G., Morrison, J., Clancy, J. and Crowe, R., The Iowa 500, Part 2 (A blind family history comparison of mania, depression, and schizophrenia), Arch. gen. Psychiat., 27 (1972) 462-464.
SHOULD ‘NON-FEIGHNER AFFECTIVE DISORDER?
WILLIAM Department Road, Iowa
CORYELL
and MING
3
SCHIZOPHRENIA’
BE CLASSIFIED
WITH
T. TSUANG
of Psychiatry, The University City, Iowa 52242 (U.S.A.)
of Iowa
College
of Medicine,
500 Newton
SUMMARY Narrow definitions of schizophrenia increase homogeneity at the expense of leaving unclassified many patients with schizophrenic symptoms. Family history and follow-up studies indicate that many such patients ought to be classified with those having affective disorders. This study determines morbid risks for affective disorder and schizophrenia in first degree relatives of patients with chart but not research diagnoses of schizophrenia. Comparisons with morbid risk figures for relatives of individuals satisfying research criteria for depression, mania or schizophrenia indicate that the ‘non-Feighner schizophrenia’ group is probably too heterogenous to be classified entirely as affective disorder or as schizophrenia.
INTRODUCTION
Broad definitions of schizophrenia tend to encompass many cases with features traditionally considered atypical of schizophrenia (Taylor and Abrams 1975a). These features include affective symptoms, acute onset of illness, a prior history of remission and the presence of confusion or perplexity. Cases with schizophrenic symptoms have a better prognosis if one or more of these features are present (Johanson 1958; Vaillant 1962; Stephens et al. 1966). A variety of studies have also shown many such cases to have family histories with high rates of affective disorder (Clayton et al. 1968; Fowler et al. 1972; Taylor et al. 1974). In order to increase homogeneity in groups designated schizophrenic, a narrow definition proposed by Feighner et al. (1972) has incorporated these atypical features as exclusion criteria. These criteria, then, define poor prog-
This work was supported in part by NIMH grant No. MH31673. Dr. Coryell is an Assistant Professor of Psychiatry and Dr. Tsuang is a Professor chiatry. Reprint requests to: Dr. Coryell, Department of Psychiatry, The University College of Medicine, 500 Newton Road, Iowa City, IA 52242, U.S.A.
of Psyof
Iowa
4
nosis subgroups from larger groups designated schizophrenic by broader definitions (Winokur and Tsuang 1975). In clinical usage, however, the Feighner criteria leave many cases unclassified (Taylor and Abrams 1975b). How should such excluded cases be classified? Since studies have shown strong associations between ‘good prognosis schizophrenia’ and affective disorder and since the Feighner criteria for schizophrenia were designed to exclude cases without a poor prognosis, then many cases so excluded probably represent variants of affective disorder. Croups of such excluded cases may or may not be homogenous enough for routine classification with the affective disorders, however. The present study uses family history data to address the issue of homogeneity in a large group defined by a failure to meet Feighner criteria despite a chart diagnosis of schizophrenia. METHODS
The authors of an earlier paper reviewed a consecutive series of inpatient charts from the University of Iowa Psychopathic Hospital dating from 1934 to 1944 to distill groups meeting research criteria for primary affective disorder and schizophrenia (Morrison et al, 1972). Two hundred charts met criteria for schizophrenia of Feighner modified by Morrison et al. (1972). This left 315 charts which did not, despite a discharge diagnosis of schizophrenia. Of this group of 315, five were recently reclassified leaving 310 here designated as ‘non-Feighner schizophrenics’. During the decades between 1930 and 1950 social workers and physicians routinely obtained extensive family history information for each admission to the University of Iowa Psychopathic Hospital. This information was type written under separate headings for each first degree relative and often extended for several pages. In an earlier paper specially designed criteria were applied to the family history information of the above mentioned probands meeting criteria for affective disorder or schizophrenia. The diagnosis of schizophrenia in a relative required an illness (1) beginning before the age of 40 years, (2) leading to chronic hospitalization or chronic social incapacity, (3) involving sustained bizarre behavior, delusional thinking or considerable seclusiveness and (4) lacking evidence of marked affective symptoms. If a relative suffered from an illness with common affective symptoms from which he recovered without significant personality defect, with or without psychiatric hospitalization, he was considered to have an affective disorder (Winokur et al. 1972). One of the present authors (WC) applied these same criteria to the charts of the ‘non-Feighner schizophrenics’. The present study uses the Weinberg shorter method to correct for age distribution in the calculation of morbidity risks for schizophrenia and affective disorder. This method deducts from the total number of a class of relatives all those who have not entered the age of risk and half the number who are between the upper and lower limits of the age of risk. These limits are 15 and 60 years of age for affective disorder and 15 and 40 years of age for
5
schizophrenia. Among the first degree relatives there were few children who had entered the age of risk for either schizophrenia or affective disorder. For similicity, ‘relatives’ will refer here only to parents and siblings. The statistical method used for comparing morbid risk of family members of the various proband groups was a comparison of two proportions. A P value of less than 0.05 was considered statistically significant. RESULTS
Affective disorder was significantly more frequent in the relatives of nonFeighner schizophrenics than in those of Feighner schizophrenics as shown in Table 1. Affective disorder in the relatives of non-Feighner schizophrenics was significantly less frequent than in the relatives of depressives. was significantly Table 2, on the other hand, shows that schizophrenia more frequent among relatives of non-Feighner schizophrenics than among those of the depressives. Morbid risk for schizophrenia in relatives of the non-Feighner schizophrenics was not significantly different than that for relatives of the Feighner schizophrenics. Relatives of manic probands were somewhat more likely than those of ‘non-Feighner schizophrenics’ and less likely than those of depressive probands to have affective disorder. The inverse was true regarding morbid risk for schizophrenia among relatives of manic probands. None of these differences were significant, however.
TABLE
1
MORBIDITY RISKS FOR AFFECTIVE OF DIFFERENT PROBAND GROUPS
DISORDER
Number cases of affective disorder
of
IN FIRST
Number of relatives
DEGREE
BZ
RELATIVES
MR (%)
Schizophrenic
probands
32
1017
584
5.48 a
(N = 200) ‘Non-Feighner
schizophrenic’
76
1692
885
8.59
probands (N = 310) Manic probands
37
519
328
11.28
(N = 100) Depressive probands
123
1413
859
14.32
b
(N = 225) Primary affective
160
1932
1187
13.48
b
probands a Sig. diff. b Sig. diff.
disorder
(N = 325) from MR for relatives from MR for relatives
of non-Feighner of non-Feighner
schizophrenics, schizophrenics,
P < 0.05. P < 0.01.
6 TABLE
2
MORBIDITY RISKS PROBAND GROUPS
FOR
SCHIZOPHRENIA
Schizophrenic probands (N = 200) ‘Non-Feighner schizophrenic’ probands (N = 310) Manic probands (N = 100) Depressive probands (N = 225) Primary affective disorder probands (N = 325) a Sig. diff. from MR for relatives
IN 1” RELATIVES
OF DIFFERENT
Number of cases of schizophrenia
Number relatives
13
1017
617
2.11
25
1692
1112
2.25
5
519
434
1.15
5
1413
1228
0.41 a
10
1932
1662
0.60 a
of non-Feighner
schizophrenics,
of
BZ
MR (%)
P < 0.01.
DISCUSSION
Tsuang et al. (1976) presented family history data for 85 of the present group of 310 probands in an earlier report. At index admission these cases had had affective symptoms and/or a history of remission. Parents and siblings of these 85 probands resembled those of the primary affective disorder group in terms of morbid risk for affective disorder. In contrast, patterns of illness in the relatives of the larger group of 310 are dissimilar to those of relatives of both the schizophrenic and primary affective disorder groups. In terms of percent recovered at follow-up also, the above group of 85 is closer to the primary affective disorder cohort then the larger group of 310. Morrison et al. (1973) while blind to research diagnosis evaluated follow-up material routinely obtained after index admission. He found that 22% of the group failing to meet Feighner criteria for schizophrenia had recovered during the follow-up period averaging 3.2 years. The proportions recovered in the unipolar and bipolar affective disorder cohorts were 54% and 59%, respectively. Tsuang et al. reported the recovery rate for the subgroup of 85 to be 44%. These family history and follow-up findings suggest that, though subgroups of cases excluded by narrow definition of schizophrenia may be appropriately grouped with the affective disorders, the excluded group as a whole is too heterogeneous to be so classified. Affective disorder in this study was found less frequently among family members of manics than among those of depressives while the reverse WAS true for morbid risks of schizophrenia. These findings are quite different from those of Perris (1966) who found more affective disorder and less schizophrenia among family members of bipolar probands than among fam-
7
ily members of depressives. Differences in selection criteria may explain some/of the incongruity. Perris required that bipolar probands have a prior history of both depressive and manic episodes with intervening periods of clear remission. Since the Feighner criteria have no such requirement, it is possible that a proportion of the mania cohort was misclassified. Indeed Winokur found that 6 of the 66 manic probands (9%) with available followup information greater than one year later received from other clinicians a diagnosis of schizophrenia (Winokur 1974). The rate of clinical diagnostic revision in the depressive cohort was much lower (3%). Also, a research design relying upon chart review of family histories may result in misclassification of bipolar illness in first degree relatives of schizophrenia. Individuals with rapidly cycling bipolar illness might well have been described by family members as chronically strange and would have been at high risk for long term hospitalization in the 1930’s and 1940’s. Family history studies such as these lack the power to suggest specific changes in boundaries of diagnostic concepts. An extensive follow-up family study of the ‘non-Feighner schizophrenics’ is underway. The resulting data in conjunction with careful assessment of clinical features recorded at index admission will be used to suggest refinements of the existing research criteria. Hopefully, the results will move the controversies surrounding the terms ‘schizoaffective’ and ‘atypical schizophrenia’ closer to resolution. ACKNOWLEDGEMENT
The authors are grateful tistical analysis.
to Jerome
Fleming, M.S., for his assistance
in sta-
REFERENCES Clayton, P.J., Rodin, L. and Winokur, G. Family history studies, Part 3 (Schizoaffective disorder, clinical and genetic factors including a one to two year follow-up), Compreh. Psychiat., 9 (1968) 31-49. Feighner, J.P., Robins, E., Guze, S.B., et al., Diagnostic criteria for use in psychiatric research, Arch. gen. Psychiat., 26 (1972) 57-63. Fowler, R.C., McCabe, M.S., Cadoret, R.J. and Winokur, G., The validity of good prognosis schizophrenia, Arch. gen. Psychiat., 26 (1972) 182-185. Johanson, E., Chapter II, Presentation of the material. Acta psychiat. neurol., 33 (Suppl. 125) (1958) 23-47. Morrison, J., Clancy, J., Crowe, R. and Winokur, G., The Iowa 500, Part 1 (Diagnostic validity in mania, depression, and schizophrenia), Arch. gen. Psychiat., 27 (1972) 457-461. Morrison, J., Winokur, G., Crowe, R. and Clancy, J., The Iowa 500 - The first follow-up, Arch. gen. Psychiat., 29 (1973) 678-682. Perris, C., A study of bipolar (manic-depressive) and unipolar recurrent depressive psychosis, Part 1 (Genetic investigation), Acta psychiat. stand., Suppl. 194 (1966) 15-44. Stephens, J.H., Astrup, C. and Mangrum, J.C., Prognostic factors in recovered and deteriorated schizophrenics, Amer. J. Psychiat., 122 (1966) 1116-1121.
8 Taylor, M.A. and Abrams, R., Manic-depressive illness and good prognosis schizophrenia, Amer. J. Psychiat., 132 (1975a) 741-742. Taylor, M.A. and Abrams, R., A critique of the St. Louis psychiatric research criteria for schizophrenia, Amer. J. Psychiat., 132 (1975b) 1276-1280. Taylor, M.A., Gaztanaga, P. and Abrams, R., Manic-depressive illness and acute schizophrenia - A clinical, family history, and treatment-response study, Amer. J. Psychiat., 131 (1974) 678-681. Tsuang, M.T., Dempsey, M. and Rauscher, F., A study of ‘atypical schizophrenia’, Arch. gen. Psychiat., 33 (1976) 1157-l 160. Vaillant, G.E., The prediction of recovery in schizophrenia, J. nerv. ment. Dis., 135 (1962) 534-543. Winokur, G., Diagnostic stability over time in schizophrenia, mania and depression, New Engl. J. Med., 290 (1974) 1027. Winokur, G. and Tsuang, M.T., A clinical and family history comparison of good outcome and poor outcome schizophrenia, Neuropsychobiology, 1 (1975) 59-64. Winokur, G., Morrison, J., Clancy, J. and Crowe, R., The Iowa 500, Part 2 (A blind family history comparison of mania, depression, and schizophrenia), Arch. gen. Psychiat., 27 (1972) 462-464.
