health data available through the Western Australian Data Linkage System(WADLS) (Holman et al., 1999).Demographics, hospitalizations for intestinal infections (gastroenteritis),respiratory infections before the age of three, which is the early psychologicaldevelopment stage (Hales et al., 2010), and diagnosis of schizophrenia in out-patient or inpatient psychiatric service before the end of follow-up (25–29 yr)were identiﬁed for the birth cohort from three databases linked by the WADLS.Data sources included: (i) Midwives Notiﬁcations Data, (ii) mental health servicerecords from the WA Mental Health Ambulatory Data System; (iii) WA HospitalMorbidity Data System; (iv) Birth Registry data; and (v) Mortality Data. The WAHospital Morbidity Data System included admission records (separation records)from all WA public and private acute hospitals (Holman et al., 1999). Thereliability and quality of these linked data have been conﬁrmed previously(Holman et al., 1999). The mental health service records and the WA HospitalMorbidity Data System included all WA public and private psychiatric hospitalsand all attendances at public psychiatric outpatient clinics (Holman et al., 1999).In order to control for potential confounding effects due to abnormal prenataldevelopments, newborns with birth weight less than 2500 g, an Apgar score at ﬁveminutes of less than seven, or with a father or mother younger than 18 at birthwere excluded from this study.
2.2. Exposure and outcome
In the Hospital Morbidity Data, diagnosis was coded using ICD-9 for the period1979–1999 which covered the early psychological development stage of the cohort.ICD-9 codes 001-139, 460-466 and 480-488, were used to identify all infections,intestinal and acute respiratory infections requiring hospitalizations. Diagnoses of schizophrenia were coded with either ICD-9 codes or ICD-10 codes in both databasesfor the observation period (Holman et al., 1999), and thus ICD-9 codes (295) andICD-10 codes (F20) were used to identify diagnosis of schizophrenia.
2.3. Statistical analysis
In this study, we investigated the effect of hospitalizations for all types of infections combined, intestinal infections and acute respiratory infections occur-ring before the age of three. Exposures to hospitalization for treatment of infections were grouped into three categories: (1) no hospitalization, (2) onehospitalization and (3) two or more hospitalizations. Cumulative incidence ratesof schizophrenia until the end of follow-up (25–29 yr) were calculated for eachcategory of exposure. Variations in the risk of schizophrenia across differentfrequencies of intestinal and respiratory infections were also described withKaplan–Meier failure graphs.In the multivariate analysis we ﬁrst assessed the association between allinfections occurring between 0 and 3 yr of age and risk of schizophrenia, and thenwe assessed the relation for two common groups of infection: intestinal infections(001-009) and acute respiratory infections (460-466, 480-488). Multivariatelogistic regression was used to control for a number of potential confounders(covariates) including: ages of mother and father at delivery; Apgar score at ﬁveminutes; whether the mother previously had any negative birth outcomes(miscarriage, death at birth or still birth); calendar year of birth; and socio-economic status. Socio-economic status was determined by the Australia Bureauof Statistics’ (ABS) Index of Relative Socio-Economic Disadvantage (IRSED, ageneral measure of economic disadvantage of residential families in an area) forresidential location (Local Government Area, LGA) at birth (McLennan, 1998).
Ethical approval for this study was obtained from the Department of Health,GovernmentofWesternAustralia,HumanResearchEthicsCommittee(DOHWAHREC).
The univariate analysis and stratiﬁed analysis included a totalnumber of 51,464 participants. Data from 45,291 participants wereused in multivariate analysis. Around 12% of the sample wasexcluded in the multivariate analysis due to missing values incontrol variables. Overall 538 cases were observed among the51,464 participants during the follow-up period. The cumulativeincidencewas10.45 per1000(95%conﬁdenceinterval: 9.61–11.38).There were 43,950 participants who did not have a hospitalizationfor infection disease before the age of three. In this group, 426schizophrenia cases were observed. Seven thousand ﬁve hundredand fourteen participants had one or more hospitalization forinfectiondiseasebeforetheageofthree,andamongthose,112caseswere observed. The accumulative incidence ratio of exposed-to-unexposed was 1.53 (95% CI 1.23–1.89). Hospitalizations for intest-inal infections and acute respiratory infections accounted for about90%ofallinfectionsrequiringhospitalizationbeforetheageofthree.In the stratiﬁed analysis, the cumulative risk of schizophrenia beforethe age of 25–29yr was signiﬁcantly higher among those with twoor more hospitalizations for all types of infections, intestinal infec-tionsoracuterespiratory infectionsbeforetheageof three(Table1).Fig. 1indicates that from the age of 18yr the cumulative risk of schizophrenia was highest among participants who had two ormore hospitalizations for intestinal and/or acute respiratory infec-tions during early psychological development, whereas it was low-est for those without. The trend across the three groups was testedwith Log-rank test. The
value for the log-rank test result was
0.001, indicating a signiﬁcant increase in the cumulative risk of schizophrenia across the three exposure levels. As shown inTables 2 and 3,after controlling for potential confounding factors,the risk of schizophrenia was signiﬁcantly higher for those with twoor more hospitalizations; however, an elevated but non-signiﬁcantodds ratio was observed for those with one hospitalization. Thecumulative risk shown inTable 1andFig. 1was not strongly
inﬂuenced by the potential confounding factors tested (Table 3).
Male participants with two or more hospitalizations for infec-tions before the age of three had an 80% higher risk of schizophreniaafter controlling for potential confounding factors. These ﬁndings
Cumulative incidence rate by hospitalizations of infections.Frequency of hospitalizations forinfectionsCases Total numberof participantsCumulativeincidencerateConﬁdenceinterval
valueAll types0 426 43,950 9.69 8.81 10.661 63 5259 11.98 9.36 15.33 0.1232
49 2255 21.73 16.42 28.75 0.001Intestinal infections0 489 48,819 10.02 9.17 10.941 32 2088 15.33 10.84 21.67 0.0272
17 557 30.52 18.97 49.10 0.001Acute respiratory infections0 462 46,521 9.93 9.07 10.881 51 3706 13.76 10.46 18.11 0.0332
25 1237 20.21 13.66 29.91 0.017
01%0.5%1.5%2%1511 16 21 31ageHospitalisation (0)Hospitalisation (>1)Hospitalisation (1)
Cumulative risk of schizophrenia (
axis) by frequency of hospitalizationsfor intestinal and acute respiratory infections in early life (
W. Liang, T. Chikritzhs / Psychiatry Research