Failure to identify a male winter-born schizophrenia subgroup clinically

Eur Psychiatry 1996;11:34-39 Q Elsevier, Paris

34

Original article

Failure to identify a male winter-born schizophrenia subgroup clinically J Modestin I, 0 Wtirmle 2, R Ammann 2 I Psychiatrische Universitiitsklinik 2 Psychiatrische Universitiitsklinik (Received

Ziirich. Lenggstrasse 31. CH 8029 Ziirich 8; Bern, Bolligenstr II I, CH 3072 Bern, Switzerland

30 November

1994; accepted

29 August

1995)

Summary - The distribution of 282 Research Diagnostic Criteria (RDC) and of 224 DSM-III-R schizophrenic patients, respectively, by month of their birth was studied. The winter-spring birth rate excess was confirmed with a maximum from January to March. Winterborn and summer-born schizophrenics were compared. No convincing differences were found with regard to a larger set of demographic, psychosocial and clinical variables in univariate comparisons, nor did we succeed in separating and identifying a special winter-born schizophrenia subgroup using the method of cluster analysis. Negative results were obtained in spite of the fact that many of the included variables reflected the course of the illness and the degree of chronicity. Either the set of the variables we used has not been ‘correct’ enough or the separation of a winter-born subgroup is not clinically feasible. schizophrenia

/ birth

distribution

/ clinical

subgroups

/ duster

INTRODUCTION

There is undoubtedly a winter-spring schizophrenia birth rate excess in comparison with the expected figures from the general population (Dalen, 1975; Bradbury and Miller, 1985; Watson, 1990; Fossey and Shapiro, 1992). Nevertheless, the efforts to characterize the winter-born schizophrenia subgroup with the help of demographic, social and clinical variables have so far not been successful. The winter-born rate was reported as being higher in male (Dalen, 1975) and paranoid male schizophrenics (Hsieh et al, 1987), but also in female (Parker and Neilson, 1976), urban-born female (O’Callaghan et al, 1995) and nonparanoid female schizophrenics (Nasrallah and McCalleyWhitters, 1984). Reviewed by Torrey and Torrey (1980), there is no evidence of any important sex differences in the seasonality of schizophrenic births. On the one hand, more winter-born patients were found among schizophrenics with shorter duration of hospitalization, the latter implying a good prognosis (Dalen, 1975; Pulver et al, 1983; Rodrigo et al, 1991). However, the corresponding data are very heterogeneous, the hospitalization

analysis

length of three years (Dalen, 1975) already indicates chronic&y and the difference between 67 and 80 days (Rodrigo et al, 1991) seems to be clinically meaningless. On the other hand, winter birth excess was found among schizophrenics who had never been married (Watson et al, 1984) and in chronic schizophrenia with negative syndrome (Opler et al, 1984), both presumably indicating a severe form of illness. Finally, H%fner et al (1987) found identical seasonal patterns of births in acute and chronic schizophrenia and in paranoid and other subgroups. The excess of winter-born cases is possibly higher in early onset (Pulver et al, 1983) and in early onset female (Pulver et al, 1981) schizophrenia, in systematic schizophrenias (according to Leonards’ classification) which also have a low genetic loading (Franzek and Beckmann, 1992), and in nonfamilial/sporadic cases (O’Callaghan et al, 1991). It was claimed that winter birth excess occurred only in schizophrenic patients of the lowest socioeconomic class (Barry and Barry, 1964; Gallagher et al, 1984); however, Odegard (1974) stated that it is not a lower class trait. In these studies only relatively small samples were often used, the number of assessed variables was mostly limited and, as some authors explicitly

Winter-born schizophreniasubgroup admitted (Kendell and Kemp, 1987; Rodrigo et al, 1991) the data sets were incomplete and the reliability of the data including diagnosis unknown. Therefore, the present study was undertaken. Its differentiating characteristics are the inclusion of a more comprehensive set of variables, the definition of cases using modem diagnostic criteria and the high reliability of all ratings including diagnosis. Also, evaluating the results, an attempt was undertaken to go beyond univariate comparisons. METHODS The study was performed in the frame of a larger-scale investigation of male inpatients. Subjects were all male schizophrenics hospitalized at least once between 19851987 at the Psychiatric University Hospital of Bern. The Hospital provides a complete inpatient care for all ca 380’000 residents of the catchment area which includes a rural as well as a population of the city of Bern. A total of 1,590 males were treated in the hospital in the three year study period. Those born between 1909 to 1969 who had received the clinical diagnosis of schizophrenia/schizophrenia related disorders (369 patients), alcohol/drug abuse/dependency (624 patients) or affective disorder (338 patients) were re-diagnosed with the help of the research diagnostic criteria (RDC; Spitzer et al, 1978) using the method of retrospective evaluation of clinical charts. RDC schizophrenia was diagnosed in 292 patients; 246 (84%) of them had received clinical diagnosis of schizophrenia previously. Ten patients dropped out (lack of data, subsequent revision of diagnosis), 282 RDC schizophrenic males were included in this study. A total of 224 of them fulfilled DSM-III-R criteria for schizophrenia. Among RDC schizophrenics there were 28 (lo%), first admission, and among DSM-III-R schizophrenics only 16 (7%). Hospital clinical records of all 282 patients were thoroughly scrutinized and relevant data were extracted. The following demographic and psychosocial variables were evaluated: age, marital status, foreign born status, place of residence, educational level, social class of patient and the family of origin, intergenerational social mobility, living and vocational situation at the time of index-admission, measures of guardianship. The psychiatric variables exam&d included the following: diagnosis (besides RDC, DSM-III-R criteria for schizophrenia were applied), age at first psychiatric therapy, age at first psychiatric hospitalization, duration of psychiatric illness, number and duration of psychiatric hospitalizations, total time spent in psychiatric hospitals; broken home, childhood abuse, aggressive behavior, previous suicide attempts, alcohol/drug abuse; psychiatric disorders leading to hospitalizations and alcoholism in the first degree relatives. All these variables were either primarily clearly determinable

35

(such as age) or were deflned and operationalized as precisely as possible (eg, the patients were judged as coming from a broken home if at least one parent or parent surrogate was lacking for at least two years before reaching 18 years of age). The occupational level allowing the social class designation was classified with the help of the scale proposed by Moore and Kleining (1960) and modified by Dilling and Weyerer (1978). Furthermore, age at first offense, full account of criminal record in general and of different types of offenses in particular, number of court appearances and number and kind of sentences were included, the data having been provided by the Swiss Central Criminal Record Department and the Division of Penal Control of the Canton of Bern. Our clinical charts were generally of acceptable quality and included in many cases psychiatric expert opinions and clinical records from other psychiatric institutions. Therefore, the complete set of data was available in the majority of probands; social class of the family of origin could not be identified in 32 patients, educational level in four patients, work situation before indexadmission in three patients and living situation at indexadmission in one patient. A total of 66 clinical charts (every 20th of 1,331 including 15 of 282, all of which were studied in the same way) were evaluated independently by two other investigators in order to assess the reliability of the ratings. Scorings of the three investigators were compared with each other; four variables of low reliability were excluded from the analysis. The average correlation coefficients for all continuous variables included in the analysis and enumerated above were between .90 and 1.00, the average weighted kappas for all categorical variables between .68 and 1.00. lu the next step, all 282 RDC schizophrenic males as well as those 224 fulfilling the DSM-III-R criteria for schizophrenia were distributed by their month of birth. Birth distribution of the patients was compared with what was expected calculated on the basis of birth distribution of the male general population of the Canton of Bern born in the years 1969-1992 (older data were not available) using the one-sample x2 test. Then, the groups of winter-born (November to April, months with higher than expected birth frequencies) and summer-born (May to October, months with lower than expected birth hequencies) DSM-III-R schizophrenic males were compared with each other in the univariate comparisons with regard to all variables numerated above. The more narrow (DSM-III-R) schizophrenia definition was chosen for this comparison. The two-sample ~2 test and MannWhitney U test were used. Those variables (indicated in table I) where there was at least some difference in the distribution between winter-born and summer-born schizophrenics (not necessarily reaching a level of statistical significance) and those potentially of importance

J Modestin et al

36 Table

I. Variables potentially differentiating between winter-born and summer-born schizophrenics. Winter-born schizophrenics

Summer-born schizophrenics

137 (loo)

87 (100)

23 (4-5 1)

Age (years) at first psychiatric therapy median (range) Age (years) at first psychiatric hospitalization median (range) Index-hospitalization > 1 year duration Total time (months) spent in psychiatric hospitals median (range) Broken home Alcohol abuse Drug abuse Residence in the city of Bern Marital status: single Living situation at index-admission with parents/relatives alone/homeless Downward social mobility Receiver of disability pension Measures of guardianship Criminal record Traflic law violations Number of court appearances 0 1-2 >2 Number of suspended sentences 0 1 >l

pa*

P

24 (6-75)

1.75*

.080

24 (4-5 1)

25 (16-75)

1.65*

.098

43 (31) 20 (l-608)

21 (24) 22 (l-652)

1.04 0.25*

>.lO >.lO

29 (33) 38 (44 17 (20) 29 (33) 73 (84)

1.29 1.15 1.08 1.82 1.25

>.lO >.lO >.lO r.10 >.lO

22 (25) 40 (46) 8 (9) 18 (21) 29 (33) 33 (38) 24 (28)

5.71

92 (67) 21(15) 24 (18)

54 (62) 26 (30) 7 (8)

9.03

111 (81) 23 (17) 3 (2)

73 w 7 (8) 7 (8)

8.13

36 (26) 50 (36) 35 (26) 58 (42) 122 (89) 52 43 21 41 35 45 21

(38) (31) (15) (30) (26) (33) (15)

1.77 2.34 1.58 0.61 4.98

.058 W = 2)

>.lO >.lO >.lO >.lO .026 .Oll

bf = 2) .017

(df = 2)

Percentages in parentheses unless indicated differently.

(indicating, eg, more or less severe course of illness, contribution of environmental factors, or diiculties/disruption in psychosocial functioning), were further explored. If there is a schizophrenia subtype caused by seasonal

factors, it will be over-representedamong winter-born schizophrenics.However, as this subgroup may be relatively small,thereis a risk of averagingout potentiallysignificant differencesby simplycomparingwinter- andsummer-born patients using univariate tests.Therefore, a cluster analysiswas performedto find out whether the larger group of winter-born

schizophrenic

patients con-

sistedof at leasttwo subgroups,the one of them correspondingto summer-born patients,the otheronebeingdifferent.In clusteranalysis,similarcasesarepooledtogether to build clusterseach of them being as homogenousas possible;the individual clustersshoulddiffer from each other as muchaspossible.The procedureof the K-means cluster analysiswas selectedwhich requiresa specific numberof clusters.Three different analysesyielding 2, 3 and 4 clusterswere performed.As the formationof clusters starts with the matrix of proximities (similarities) between cases, all values of the selected variables were

stamlardized to a range of 0 to 1. As mentioned, the data of 224 DSM-III-R schizophrenic patients pertaining to the variables indicatedin tableI wereincluded.

RESULTS The distribution by month of birth of 282 RDC male schizophrenic patients as well as of those 224 fulfilling the DSM-III-R criteria for schizophrenia is presented in figure 1, the quarterly distribution of the births of these patients compared with normal population controls is given in table II. Compared with controls, a significant birth excess was found in winter half of the year (months November to April; ~2 = 5.68, 1 dJ P = .017) and in the first quarter of year (months January to March). Of 58 RDC schizophrenia positive but DSM-III-R schizophrenia negative patients 29 were born in winter and 29 in summer half of year. Winter-born and summer-born DSM-III-R schizophrenics were compared using univariate analyses with regard to the already described larger set of demographic, psychosocial, personal

Winter-born

AN

FEB

MAR

Fig 1. Distrubution DSM-III-R (lower month of bii.

APR

M*Y

JUN

JIJL

*LG

SW

OCT

of 282 RDC (whole columns) part of columns) schizophrenic

NO”

schizophrenia

DEC

and 224 men by

and psychiatric variables. With the exception of a different patients’ living situation at index admission and a few indices of criminal behavior, no statistically significant differences were found (table I); considering the number of almost 50 comparisons which had been performed even those differences which appeared statistically significant could have become so by chance. With the help of cluster analysis the whole sample was divided in 2, 3 or 4 clusters. Cluster membership of each patient and distribution of their births (winter/summer half of year on the one hand and quarterly distribution on the other hand) were used to build two-way tables. The results give no indication of a specific subgroup of winter-born schizophrenic patients, disregarding the way of ‘winter-born’ definition. In other words, no cluster could be identified which would consist predominantly of individuals born in the winter half or in the first quarter of the year. Figure 2 (mosaic plots graphically representing two-way tables) presents an example of two cluster solutions for both ‘winter-born’ definitions.

DISCUSSION We studied 224 male DSM-III-R schizophrenic (mostly rehospitalized) inpatients of different ages (range 19 to 78 years, average 41 years) who presented a broad range of degrees of chronicity and psychosocial adjustment difficulties. Compared with the male general population, a significant excess of 23% of births was found in the winter half of the year between November and April, the highest excess of 28% being registered in January through March. This is quite in correspondence with the pertinent literature (compare H&fner et al, 1987). The age incidence effect (Lewis, 1989) will

subgroup

37

hardly play a role in our probands regarding their higher average age and broad age range. Its only marginal influence was demonstrated studying similar samples (O’Callaghan et al, 1991). Unfortunately, we were not able to compare our probands with the controls born at the same time. However, there seems to be little fluctuation of birth distribution by month over time (no significant differences between the years 1969-1980 and 1981-1992) as there are no significant differences in birth distribution comparing the Canton of Bern and the whole of Switzerland (births 1979-1992). No winter-born excess was found in our more broadly defined RDC positive but DSM-III-R negative schizophrenic patients; nevertheless, the subgroup would have been too small for this kind of study. Using univariate comparisons we did not succeed in finding any convincing differences between winter-born and summer-born patients with regard to the common demographic, psychosocial and clinical variables, many of the latter reflecting the course of the illness and the degree of chronicity. There were some, even though not significant, trends towards a more serious course of illness in winter-born patients, however, even using a more sophisticated method of cluster analysis, we did not succeed in separating and identifying a special winter-born subgroup which would encompass the winter-born schizophrenic patients’ surplus. Regarding other attempts to differentiate the winter-born schizophrenics or identifying a winterborn subgroup with clinical methods, no convincing results have been reported at present, including the studies which specifically aimed at such a differentiation. Kendell and Kemp (1987) compared larger samples of winter-born and summer-born schizophrenics with regard to the 13 most important so&-demographic (eg, age, sex, marital status and social class) and clinical (eg, diagnosis, number of admissions, cumulative duration of all admissions) variables. No unequivocal differences were found. In the same attempt, Baron and Gruen (1988) found in a small sample of 88 RDC schizophrenics the season of birth to be unrelated to the sex, birth order, age at onset, clinical subtype (paranoid vs non-paranoid) and family history of schizophrenia. Thus, our results based on the study of more well-defined group with the help of a more extended set of clinical variables of a known reliability, and evaluated with more advanced statistical methods follow those of other investigations. What might be the reasons for our failure? It may still be that we did not use the most appropriate variables such as thorough account of psychopathological findings. The recent results by Fran-

J Modestin

38 Table II. Quarterly lation.

distribution

of the births of 282 RDC

and 224 DSM-III-R

RDC

schizophrenia

schizophrenic

April-June

January-March RDC schizophrenia Observed (0) Expected 03 o/E x 100% x2 P DSM-III-R schizophrenia Observed (0) Expected (E) o/E x 100% x2 P

et al patients

in comparison

July-September

with normal

October-December

91 72.7 125% 4.61 .032

58 13.6 79% 3.31 .069

60 71.1 84% 1.73 >.l

13 64.6 113% 1.09 s.1

74 57.8 128% 4.54 ,033

48 58.4 82% 1.85 >.l

45 56.4 80% 2.30 >.l

57 51.3 111% 0.63 s.1

x2 = 10.74 and P = .013 with 3 df. DSM-III-R

schizophrenia

25

pop-

x* = 9.32 and P = .025 with 3 &

25

0 NOV-APR Fig 2. Results of the cluster and cluster membership. m

JAN-MAR

MAY-QCT analysis (two cluster 1; 0

cluster solution). cluster 2.

Distribution

zek and Beckmann (1992), eg, seem to be in accordance with such an assumption. Some authors claim the winter birth rate excess to be higher in schizophrenic patients with a low genetic risk (Kinney and Jacobsen, 1978; Shur, 1982; Franzek and Beckmann, 1992); conversely, it has been proposed that, in some fetuses at high genetic risk for schizophrenia, an exaggerated level of abortions or stillbirths occurs if additional environmental noxious agents are present resulting in a

of 224 DSM-III-R

APR-JUN schizophrenic

JUL-SEP

OCT-DEC

men by their birth

season

decrease in winter and spring births of schizophrenic patients with high familial loading (Beckmann and Franzek, 1992). The winter birth rate excess seems to parallel the winter above average prematurity and CNS malformations rates (Dalen, 1975). It is considered to be a result of harmful environmental effects (Bradbury and Miller, 1985) such as temperature extremes, nutritional deficits, obstetric complications or, in the first place, infectious agents or factors closely related to them

Winter-born

schizophrenia

(Watson et al, 1984, Takei et al, 1994). Correspondingly, electrodermal hyporesponsivity (Katsanis et al, 1992), increased incidence of ventricular enlargement (d’Amato et al, 1994), larger ventricle-brain-ratio and greater sulcal prominence ratings in CT scans were found among winter-born schizophrenics (Zipursky and Schulz, 1987; Degreef et al, 1988), even though not more neurological soft signs (di Michele et al, 1992). Thus, a proportion of winter-born schizophrenics may represent an ‘organic’ schizophrenia subgroup. To identify the ‘organic’ substrate, however, we quite obviously have to go beyond the level of the common clinical data. Our negative results indicate that the presumed ‘organic’ component, if really present in some cases, does not necessarily closely correlate with the course of illness and with the ensuing patients’ psychosocial adjustment. Thus, it is possible that the schizophrenic phenotype will be the same, disregarding the whole scale of different potential aetiological components. REFERENCES d’Amato T, Rochet T, Dal&y J, Chauchat JH, Martin JP, Marie-Cardine M. Seasonality of birth and ventricular enlargement in chronic schizophrenia. Psychiatry Res: Neureimaging 1994;55:65-73 American Psychiatric Association. Diagnosric and Starisrical Manual df Mental Disorders, third edition, revised. Washington DC: American Psychiatric Association, 1987 BaronM, Gruen R. Risk factors in schizophrenia. Season of bii and family history. Br JPsychiatry 1988;152:460-5 Barry H, Bany H jr. Season of bii in schizophrenics. Its relation to social class. Arch Gen Psychiatry 1964; 11:385-9 1 Beckmann H, Franzek E. Deficit of birth rates in winter and spring months in distinct subgroups of mainly genetically determined schizophrenia. Psychopathology 199225:57-&d Bradbury TN, Miller GA. Season of birth in schizophrenia: a review of evidence, methodology, and etiology. Psycho1 Bull 1985;98:569-94 Dalen P. Season of birth. A study of schizophrenia and other metial disorders. Amsterdam, Oxford: North-Holland. 1975 Degreef G, Mukherjee S, Bildek R, Schnur D. Season if birth and CT scan findings in schizophrenic patients. Biol Psychiatry 1988;24:4614 Dilling H, Weyerer S. Epidemiologic psychischer Sriirungen und psychiatrische Versorgung. Miinchen, Wien, Baltimore: Urban und Schwmenberg, 1978 di Michele V, de Cataldo S, Rossi A, Casacchia M. Minimal brain damage and seasonality of birth in schizophrenic disorders. Eur Psychiatry 1992;7:91-2 Fossey E, Shapiro CM. Seasonality in psychiatry - a review. Can J Psychiarry 1992;37:299-308 Franzek E, Beckmann H. Season-of-birth effect reveals the existence of etiologically different groups of schizophrenia. Eiol Psychiatry 1992;32:375-8 Gallagher BJ, Jones BJ, McFalls JA. The ‘winter phenomenon’ among schizophrenics: differences between blacks and whites. J Clin Psycho1 1984;49: 1151-9 Hgfner H, Haas S, Pfeifer-Kurda M, Eichhom S, Michitsuji S.

subgroup

39

Abnormal seasonality of schizophrenic births. A specific finding? Eur Arch Psychiatry Neural Sci 1987;236:333-42 Hsieh HH, Khan MH, Atwal SS, Cheng SC. Seasons of birth and subtypes of schizophrenia. Acta Psychiat Stand 1987; 751373-6 Katsanis J, Ficken J, Iacono WG, Beiser M. Season of birth and electrodermal activity in functional psychoses. Biol Psychiatry 1992;31:841-55 Kendell RE, Kemp IW. Winter-born vs summer-born schizophrenics. Br J Psychiatry 1987; 15 1:499-505 Kinney DK, Jacobsen B. Environmental factors in schizophrenia: new adoption study evidence. In: Wynne LC, Cromwell RL, Matthysse S, eds. The nature of schizophrenia: New approaches to research and treatment. New York: John Wiley & Sons, 1978;38-51 Lewis MS. Age incidence and schizophrenia: Part I. The season of birth controversy. Schizophr Bull 1989;15:59-73 Moore H, Kleining G. Das soziale Selbstbild der Gesellschaftsschichten in Deutschland. Kiilner Zeitschriftfir Soziologie und Sozialpsychialrie, 1960; 12:86-l 19 Nasrallah HA, McCalley-Whitters M. Seasonality of birth in subtypes of chronic schizophrenia. Actu Psychiatr Stand 1984;69:292-5 O’Callaghan E, Gibson T, Colohan HA et al. Season of birth in schizophrenia. Evidence for confinement of an excess of winter births to patients without a family history of mental disorder. Br J Psychiatry 1991;158:764-9 O’Callaghan E, Cotter D, Colgan K, L&in C, Walsh D, Waddington JL. Confinement of winter birth excess in schizophrenia to the urban-born and its gender specificity. Br J Psychiatry 1995;166:51-4 Odegard 0. Season of birth in the general population and in patients with mental disorder in Norway. Er J Psychiatry 1974;125:397405 Opler LA, Kay SR, Rosado V, Lindenmayer JP. Positive and negative syndromes in chronic schizophrenic inpatients. J Nerv Menr Dis 1984;172:3 17-25 Parker G, Neilson M. Mental disorder and season of birth - A southem hemisphere study. Br JPsychiazry 1976;129:355-61 Pulver AE, Sawyer JW, Childs B. ‘Ihe association between season of biih and the risk for schizophrenia. Am J Epidemiol 1981;114:73549 Pulver AE, Stewart W, Carpenter WT jr, Childs B. Risk factors in schizophrenia: season of birth in Maryland, USA. Br J Psychiatry 1983;143:389-96 Rodrigo G, Lusiardo M, Briggs G, Ulmer A. Differences between schizophrenics born in winter and summer. Acfa Psychiatr Stand 1991;84:320-2 Shur E. Season of birth in hiah and low genetic risk schizophrenics. Br J Psychiatry 1~82;140:41~5 Spitzer RL, Endicott J, Robins E. Research Diagnostic Criteria (RDC} for a selected group of functional disorders, 3rd ed. New York: New York State Psychiatric Institute, 1978 Takei N, Sham P, O’Callaghan E, Murray GK, Glover G, Murray RM. Prenatal exposure to influenza and the development of schizophrenia: is the effect confined to females? Am J Psychiatry 1994;151:117-9 Torrey EF, Torrey BB. Sex differences in the seasonality of schizophrenic biis (Letter to the editor). Br J Psychiatry 1980;137:101-2 Watson CG, Kucala T, ‘Iilleskjor C, Jacobs L. Schizophrenic birth seasonality in relation to the incidence of infectious diseases and temperature extremes. Arch Gen Psychiatry 1984;4 1:85-90 Watson CG. Schizophrenic birth seasonality and the age-incidence artifact. Schizophr Bull 1990; 165-10 Zipursky RB, Schulz SC. Seasonality of birth and CT findings in schizophrenia. Biol Psychiatry 1987;22: 1288-92