Method. NeuRA Infectious agents April 2016

Transcription

1 Introduction Exposure to infection in utero is often cited as a risk factor for schizophrenia. Much focus is given to the influenza virus, despite studies yielding inconsistent findings. This topic summarises the available evidence for the risk of developing schizophrenia following exposure to influenza and other infectious agents, both before and after birth; as well as the prevalence of markers of earlier infection (antibodies) in adults with schizophrenia. The physiological mechanisms of any association of these infectious agents with schizophrenia are largely unclear. The Herpesviridae are a family of viruses which cause latent, recurring, and sometimes lifelong infections. These include Herpes simplex virus (HHV1 & 2) which causes oral and/or genital herpes; the Varicella Zoster Virus (VZV,HHV3) which causes chicken pox, shingles and rarely, encephalitis; the Epstein-Barr Virus (EBV, HHV4) and Cytomegalovirus (CMV, HHV5) which cause neurological complications; and the Herpes lymphotropic virus (HHV6), which causes roseola (skin rash and fever). Borna Disease Virus (BDV) is the key causative component of Borna disease, a neurological syndrome primarily affecting animals (particularly horses, cattle, sheep, dogs and cats). However, human infection with BDV has been linked to some psychiatric illnesses through its neurological interactions. Human Endogenous Retroviruses (HERVs) are fragments of ancient viral infections that became embedded within the germ cells (sperm and eggs), and are passed on to subsequent generations, making up a large proportion of the human genome. HERVs are proposed to have involvement in some autoimmune diseases. The Human T-lymphotropic virus Type I (HTLV- 1) is a human retrovirus that integrates into immune cells and is associated with an increased risk of developing cancers such as adult T-cell leukemia, myeloma, and lymphoma. The Chlamydiaceae family of bacteria can cause a range of infections in humans, including chlamydia and trachoma (Chlyamydia trachomysis) and pneumonia (Chlamydophila pneumonia, Chlamydophila psittaci). Toxoplasma gondii is a parasitic protozoa, hosted by domestic cats and other warmblooded animals including humans. Toxoplasma gondii infection is usually of minor consequence to an adult but can have serious implications for a foetus. Method We have included only systematic reviews (systematic literature search, detailed methodology with inclusion/exclusion criteria) published in full text, in English, from the year 2000 that report results separately for people with a diagnosis of a schizophrenia spectrum disorder. Due to the high volume of systematic reviews we have now limited inclusion to systematic meta-analyses. Where no systematic meta-analysis exists for a topic, systematic reviews without meta-analysis are included for that topic. Reviews were identified by searching the databases MEDLINE, EMBASE, CINAHL, Current Contents, PsycINFO and the Cochrane library. Hand searching reference lists of identified reviews was also conducted. When multiple copies of reviews were found, only the most recent version was included. Review reporting assessment was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) NeuRA April 2016 Page 1

2 checklist that describes a preferred way to present a meta-analysis 1. Reviews with less than 50% of items checked have been excluded from the library. The PRISMA flow diagram is a suggested way of providing information about studies included and excluded with reasons for exclusion. Where no flow diagram has been presented by individual reviews, but identified studies have been described in the text, reviews have been checked for this item. Note that early reviews may have been guided by less stringent reporting checklists than the PRISMA, and that some reviews may have been limited by journal guidelines. Evidence was graded using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) Working Group approach where high quality evidence such as that gained from randomised controlled trials (RCTs) may be downgraded to moderate or low if review and study quality is limited, if there is inconsistency in results, indirect comparisons, imprecise or sparse data and high probability of reporting bias. It may also be downgraded if risks associated with the intervention or other matter under review are high. Conversely, low quality evidence such as that gained from observational studies may be upgraded if effect sizes are large, there is a dose dependent response or if results are reasonably consistent, precise and direct with low associated risks (see end of table for an explanation of these terms) 2. The resulting table represents an objective summary of the available evidence, although the conclusions are solely the opinion of staff of NeuRA (Neuroscience Research Australia). Results We found nine systematic reviews that met our inclusion criteria Moderate quality evidence suggests a medium sized increase in the reporting of childhood central nervous system viral infections in adults with schizophrenia compared with controls. They may also have higher levels of markers for Human Herpesvirus-2 (small effect); Borna Disease Virus (small to medium effect); Human Endogenous Retroviruses (HERV-W: large effect); Chlamydophila pneumoniae and Chlamydophila psittaci (large effects); and Toxoplasma gondii (small to medium effect). Moderate to low quality evidence indicates that prior to onset of schizophrenia, people may have a small increase in antibodies to Toxoplasma gondii, while those with recent onset schizophrenia may have a medium sized increase. Moderate to low quality evidence suggests a medium size increased risk of schizophrenia following exposure to maternal infections in utero, particularly upper respiratory tract, genital or reproductive infections, also herpes simplex virus, and inflammatory cytokines TNF-a and IL-8, but not influenza. Moderate to low quality evidence shows people with schizophrenia have higher Toxoplasma gondii antibody levels than people with a mood disorder. NeuRA April 2016 Page 2

3 Arias I, Sorlozano A, Villegas E, de Dios Luna J, McKenney K, Cervilla J, Gutierrez B, Gutierrez J associated with schizophrenia: A meta-analysis Schizophrenia Research 2012; 136(1-3): View review abstract online Comparison Rates of infection markers in people with schizophrenia vs. controls. Moderate quality evidence (direct, large samples, consistent, imprecise) suggests people with schizophrenia may have higher levels of markers for Human Herpesvirus-2 (small effect); Borna Disease Virus (small to medium effect); Human Endogenous Retroviruses (HERV-W: large effect); Chlamydophila pneumoniae and Chlamydophila psittaci (large effects); and Toxoplasma gondii (small to medium effect). Herpesviridae family Human Herpesvirus-1 No differences were reported between people with schizophrenia and controls; 11 studies (N = 664), OR 1.37, 95%CI 0.78 to 2.39, p = 0.273, Q 11.69, p = 0.306, I % Meta-regression suggested no differences in results due to detection technique Human Herpesvirus-2 Significant, small effect of an increase in markers in people with schizophrenia vs. controls; 6 studies (N = 2288), OR 1.34, 95%CI 1.09 to 1.70, p = 0.05, Q 4.27, p = 0.511, I 2 not reported Note: The largest included study in this analysis was the only study showing a positive association and utilised blood samples from newly born babies ( < 1 week after birth), so antibodies came from the mother Varicella Zoster Virus No differences were reported between people with schizophrenia and controls; 4 studies (N = 138), OR 1.17, 95%CI 0.16 to 8.58, p = 0.877, I 2, p not reported Epstein-Barr Virus No differences were reported between people with schizophrenia and controls; 5 studies (N = 258), OR 1.67, 95%CI 0.57 to 4.94, p = 0.352, Q 4.29, p = 0.306, I % NeuRA April 2016 Page 3

4 Cytomegalovirus No differences were reported between people with schizophrenia and controls; 15 studies (N = 929), OR 0.86, 95%CI 0.54 to 1.38, p = 0.544, Q 12.81, p = 0.541, I 2 not reported Meta-regression suggested no differences in results due to detection technique Human Herpesvirus-6 No differences were reported between people with schizophrenia and controls; 3 studies (N = 113), OR 0.34, 95%CI 0.49 to 2.42, p = 0.283, Q, I 2, Q-test p not reported Borna Disease Virus Significant, small to medium sized effect of an increase in markers in people with schizophrenia vs. controls; 23 studies (N = 3853), OR 2.03, 95%CI 1.35 to 3.06, p < 0.01, Q 32.70, p = 0.086, I % Meta-regression suggested no differences in results due to detection technique Human Endogenous Retroviruses No differences were reported in HERV between people with schizophrenia and controls; 4 studies (N = 212), OR 3.66, CI95% 0.79 to 16.95, p = Significant, large effect of an increase in HERV-W markers in people with schizophrenia vs. controls; 5 studies (N = 519), OR 19.31, 95%CI 6.74 to 55.29, p < 0.001, Q 5.42, p = 0.247, I % Human T-cell Lymphotropic Virus No differences were reported between people with schizophrenia and controls; 2 studies (N = 302), OR 0.58, 95%CI 0.20 to 1.62, p =0.297 Chlamydiaceae family Chlamydia trachomatis No differences were reported between people with schizophrenia and controls; 2 studies (N = 422), OR 4.39, 95%CI 0.03 to , p = Chlamydophila pneumonia Significant, large effect of an increase in markers in people with schizophrenia vs. controls; 2 studies (N = 422), OR 6.34, 95%CI 2.83 to 14.19, p < 0.001, Q, I 2, Q-test p not reported Chlamydophila psittaci NeuRA April 2016 Page 4

5 Significant, large effect of an increase in markers in people with schizophrenia vs. controls; 2 studies (N = 422), OR 29.05, 95%CI 8.91 to 94.70, p < 0.001, Q, I 2, Q-test p not reported Toxoplasma gondii Significant, small to medium sized effect of an increase in markers in people with schizophrenia vs. controls; 8 studies (N = 2381), OR 2.70, 95%CI 1.34 to 4.42, p = 0.005, Q 33.89, p < 0.001, I 2 = 79.3% Authors state this heterogeneity was explained by detection technique, with results from DNA in brain biopsies (OR %CI 0.03 to 97.01) being significantly different (p <0.001) from results from serum antibodies (OR 2.74, 95%CI 1.33 to 5.62) Influenza virus No differences were reported between people with schizophrenia and controls Other infectious agents Individual studies assessed: parvovirus B19, parvovirus AAV-2, John Cunningham (JC) virus, BK virus, HERV-K115, Human immunodeficiency virus (HIV) and Toxocara seroprevalence Only Toxocara seroprevalence showed a large effect of increased markers in schizophrenia; OR (N = 198), 95%CI 9.71 to , p < Consistency in results Precision in results Directness of results Consistent where reported. T.gondii heterogeneity explained by detection technique Imprecise Direct Fekadu A, Shibre T, Cleare A Toxoplasmosis as a cause for behavioural disorder overview of evidence and mechanisms Folia Parasitologica 2010; 57(2): View review abstract online Comparison 1 Toxoplasma gondii antibodies in people with schizophrenia vs. controls. NeuRA April 2016 Page 5

6 Moderate to low quality evidence (direct, large sample, unable to assess consistency or precision) suggests people with schizophrenia may have higher levels of antibodies for Toxoplasma gondii in serum and cerebrospinal fluid compared with controls. Prevalence of Toxoplasma gondii Significant large effect of increased prevalence of antibodies for t.gondii in people with schizophrenia compared to healthy controls; 7 studies (6 case-controls, 1 review of 15 serological surveys in China), N = 7930 (4925 cases, 3005 controls) Seroprevalence ranged from 26 to 97% in people with schizophrenia vs. 8.9 to 52.9% in controls OR range 2.29 to 6.88, 95%CI range 1.17 to Significant medium effect of increased magnitude of antibodies (immunoglobulin G, IgG) for t.gondii in people with schizophrenia; 2 case-control studies, N = 1549 (399 schizophrenia, 1150 healthy controls) 1 study reported data: HR 1.24, 95%CI 1.12 to case-control study reported that patients with recent onset schizophrenia who were not taking medication had significantly raised levels of serum and cerebrospinal fluid IgG levels compared to controls, and patients taking medication showed lower levels but this did not differ significantly from controls. Note review authors also state that studies reported an association between higher t.gondii IgG antibody levels and more severe positive symptoms as well as a significant correlation between the use of antipsychotics with immunomodulatory effect and lower antibody titre (serointensity). Note - 1 study reported significantly higher antibody concentration among people with schizophrenia older than 45 compared to those under 45 years and a significant decrease in antibody concentration among people with schizophrenia who are treated with antipsychotic. Consistency in results Precision in results Directness of results Comparison 2 No measures of consistency reported appears consistent Unable to assess (CI range across studies) appears imprecise Direct Toxoplasma gondii antibodies in people with schizophrenia vs. people with mood disorders. Moderate to low quality evidence (large samples, imprecise, NeuRA April 2016 Page 6

7 unable to assess consistency, direct) suggests people with schizophrenia may show increased levels of Toxoplasma gondii compared to people with mood disorders. Prevalence of Toxoplasma gondii 2 out of 3 studies found a significant increase in antibody levels in people with schizophrenia compared to people with a mood disorder Significant large effect of increased prevalence of antibodies for t.gondiii in people with schizophrenia compared to people with depression; 1 case-control study, N = 200 (100 schizophrenia, 50 depression, 50 controls) Seroprevalence in schizophrenia = 66%; in depression = 24%, OR 6.15, 95%CI 2.85 to Significant medium effect of increased magnitude of antibodies IgG for t.gondii in people with schizophrenia compared to bipolar and other mood disorders; 1 case-control study, N = 413 (186 schizophrenia, 258 bipolar disorder and other mood disorders (31 both diagnosis)) plus 2 controls for each case OR 1.79, 95%CI 1.01 to study reported no increase in antibodies in patients with schizophrenia or any other psychiatric morbidity; 1 cross-sectional study, N = 452 farmers (assessed with CIS-R) Consistency in results Precision in results Directness of results Comparison 3 No measures of consistency reported appears inconsistent Imprecise Direct Risk of schizophrenia following exposure to toxoplasma gondii antibodies during gestation. Low quality evidence (large sample, indirect, unable to assess consistency, imprecise) is unable to determine any association between exposure during pregnancy to toxoplasma gondii antibodies and possible later risk for schizophrenia or executive functioning in adulthood. Antibody levels NeuRA April 2016 Page 7

8 Exposure to maternal IgG antibodies during gestation may have had a significant large effect on later development of schizophrenia; 1 nested case-control study (schizophrenia spectrum disorders), N = 186 (63 cases with exposure from pregnancy, 123 controls) OR 2.61, 95%CI 1.00 to 6.82 Cognitive ability No difference in executive functioning among those with gestational exposure compared to those without gestational exposure; 1 case-control study (schizophrenia and schizophrenia spectrum), N = 50 (26 exposed to Toxoplasma gondii during gestation, 24 unexposed controls) no data reported, although the results were not significant, review authors report greater impaired executive functioning among those with gestational exposure compared to those not exposed. Consistency in results Precision in results Directness of results N/A one study per outcome Imprecise for antibody level, not reported for cognitive ability Indirect exposure inferred through maternal infection Khandaker G, Zimbron J, Dalman C, Lewis G, Jones P Childhood infection and adult schizophrenia: A meta-analysis of population-based studies Schizophrenia Research 2012; 139(1-3): View review abstract online Comparison Central nervous system (CNS) infections in childhood in adults with schizophrenia or non-affective psychosis vs. general population. Moderate quality evidence (large sample, consistent, imprecise, direct) suggests a medium size effect of increased risk of childhood CNS viral infections in people with schizophrenia compared with the general population. CNS infections NeuRA April 2016 Page 8

9 3 studies: 1035 cases and over 1.2 million controls Significant small effect suggesting increase in the risk of any childhood CNS infections in adults with schizophrenia, compared to controls; Any CNS infections: RR 1.80, 95%CI 1.04 to 3.11, p = 0.03, Q not reported, p = 0.10; I 2 55% Significant medium effect suggesting increase in the risk of childhood viral CNS infection in adults with schizophrenia, compared to controls; Viral infections (2 studies): RR 2.12, 95%CI 1.17 to 3.84, p = 0.01, Q not reported, p = 0.07; I 2 70% No difference in the risk of childhood bacterial infection in adults with schizophrenia, compared to controls; Bacterial infections (2 studies): RR 0.72, 95%CI 0.14 to 3.54, p = 0.68, Q, p-value, I 2 not reported Note: all above effect estimates were not statistically significant when a random effects model was applied, although reported effect sizes were slightly larger 1 additional study reported increased cumulative incidence of CBV-5 meningitis in schizophrenia (12.5%, 95%CI 4.90% to 29.90%) 1 study reported increased risk of tuberculosis in schizophrenia (RR 15.00, 95%CI 2.00 to ) 1 study reported increased risk of cytomegalovirus (RR 16.6, 95%CI 4.30 to 65.10), and mumps (RR 2.7, 95%CI 1.20 to 6.20) in non-affective psychosis Consistency in results Precision in results Directness of results Consistent where applicable Imprecise Direct Khandaker G, Zimbron J, Dalman C, Lewis G, Jones P Prenatal maternal infection, neurodevelopment and adult schizophrenia: a systematic review of population based studies Psychological Medicine 2013; 43(2): View review abstract online Comparison Exposure to prenatal maternal infections in adults with schizophrenia vs. population controls. Moderate to low quality evidence (direct, large samples, unable to assess consistency or precision) suggests a medium size NeuRA April 2016 Page 9

10 increased risk of schizophrenia following exposure to maternal infections in utero, particularly upper respiratory tract, genital or reproductive infections, also herpes simplex virus, toxoplasma Gondii, and inflammatory cytokines TNF-a and IL-8. Prenatal maternal infections Serum assays Herpes simplex virus (HSV) 3/5 studies (N = 1200) reported a significant increased risk of schizophrenia (~ 50%) after exposure to HSV-2 infection during pregnancy. 1 study (N = 110) reported a non-significant, 30% increase in risk of schizophrenia spectrum disorder following HSV-2 exposure, and 1 study (N = 648) reported no increase in risk of schizophrenia following HSV-1 or HSV-2 exposure. Toxoplasma Gondii 2 studies (N = 771) reported significant increased risk of schizophrenia (ORs 1.79 and 2.61) after exposure to elevated levels of maternal IgG antibodies to Toxoplasma gondii. One study found no association between Toxoplasma gondii exposure and non-affective psychosis but was limited by a very small sample of psychosis (N = 40). Influenza 2 studies (N = 372) reported increased risk of schizophrenia after exposure to influenza, but neither study reported significant differences compared to controls. Inflammatory cytokines 1 study (N = 71) reported significantly higher tumour necrosis factor (TNF)-α levels in mothers of offspring with schizophrenia. Adult offspring of women with both third trimester infection and elevated TNF-α levels were eight times more likely to develop psychosis. 1 study (N = 177) reported a twofold increase in interleukin (IL)-8 levels during mid- to late pregnancy in mothers of offspring with schizophrenia. Clinically diagnosed infections General infections 1 study (N = ) reported maternal hospitalisation for any infection during second trimester of pregnancy was associated with increased risk of schizophrenia in offspring (RR 1.72, 95% CI 1.15 to 2.46). 1 study (N = 7941) reported bacterial infection during the first trimester was reported to be associated with a twofold increased risk of schizophrenia at follow-up at both age 34 and 47 years. Risk was also increased, but not statistically significant for second trimester exposure. No significant increase in risk after prenatal exposure to viral infections. NeuRA April 2016 Page 10

11 Respiratory infection 2 studies (N = 16964) reported that maternal upper respiratory tract infection during the second trimester was reported to be associated with a two to threefold increased risk of schizophrenia in adult offspring by age 47 years. Genital, reproductive and urinary infection 3/4 studies reported increased risk of schizophrenia following exposure to infections, mostly during the early stages of gestation. 1 study (N = 7794) reported that maternal genital or reproductive infection during the peri-conceptional period (30 days before and after the last menstrual period) was associated with a fivefold increased risk schizophrenia in adult offspring, with no increase in risk during pregnancy. 1 study (N = ) reported that hospital admission for genital infection during pregnancy was associated with a twofold increased risk of schizophrenia in adult offspring. 1 study (N = 7941) reported that maternal gonococcal infection during pregnancy was associated with a threefold increased risk of schizophrenia by age 47 years. 1 study (N = 23404) reported no association between maternal hospitalization with pyelonephritis during pregnancy and risk of schizophrenia in the offspring. Interaction between prenatal maternal infection and other risk factors 1 study (N = 23404) reported a fivefold increased risk of schizophrenia in offspring exposed to maternal pyelonephritis during pregnancy who also have a family history of schizophrenia. 1 study (N = 744) reported that mother s sexual behaviour prior to and during pregnancy was relevant to HSV-2 seropositivity and subsequent increased risk of schizophrenia. 6 studies (N = 532) reported that exposure to maternal infection or increased inflammatory cytokines during pregnancy were associated with significant deficits in executive functioning, childhood and adult verbal IQ, and greater IQ decline during the pre-morbid period and increased ventricular volume, increased length of the cavum septum pallucidum, and reduced cortical volume in offspring who developed schizophrenia. Consistency in results Precision in results Directness of results Unable to assess Unable to assess Direct Monroe JM, Buckley PF, Miller BJ Meta-Analysis of Anti-Toxoplasma gondii IgM Antibodies in Acute NeuRA April 2016 Page 11

12 Psychosis Schizophrenia Bulletin 2015; 41(4): View review abstract online Comparison Toxoplasma gondii antibodies in people with acute psychosis vs. controls. Moderate to low quality evidence (large samples, some inconsistencies, imprecise, indirect) suggests a medium sized increased risk of toxoplasma gondii IgM antibodies in chronic patients with acute psychosis compared with controls. Toxoplasma gondii IgM antibodies Significant, medium sized increased risk of positive T. gondii IgM antibodies in people with acute psychosis compared with controls, with no differences for first-episode patients; Any psychosis: 16 studies, N = 4060, OR 1.68, 95%CI 1.23 to 2.27, p = 0.001, I %, p = After removing two studies with the highest seroprevalence of positive IgM antibodies in patients and controls, the association was stronger and the heterogeneity was no longer significant; OR 2.81, 95%CI 1.80 to 4.37, p <.001, I 2 = 30.5%, p = Meta-regression analyses showed increased risk in studies from Asia, Europe, and South America (regions with a lower prevalence of T. gondii IgM positive controls), than in studies from Africa and the Middle East. There were no differences according to age, sex, or publication year. Relapsed, chronic schizophrenia: 10 studies, OR 2.54, 95%CI 1.63 to 3.96, p <.001, I %, p = 0.05 After excluding one study, the association was large and the heterogeneity was not significant; OR 6.89, 95%CI 3.10 to 15.34, p < 0.001, I2 = 24.8%, p = First-episode psychosis: 4 studies, OR 1.47, 95%CI 0.84 to 2.58, p = 0.181, I 2 5.9%, p = Authors report no evidence of publication bias. Consistency in results Precision in results Directness of results Consistent (with studies removed) Imprecise Direct NeuRA April 2016 Page 12

13 Selten JP, Frissen A, Lensvelt-Mulders G, Morgan VA Schizophrenia and 1957 Pandemic of Influenza: Meta-analysis Schizophrenia Bulletin 2010; 36(2): View review abstract online Comparison 1 The number of cases of schizophrenia among people born after the 1957 influenza pandemic compared to the number of cases of schizophrenia among people born in surrounding years (between 1 and 6 years on either side of 1957). Studies considered infants born in the first month after the peak as having been exposed during the ninth month of pregnancy and classified the other months accordingly to assess first, second and third trimester exposure. Moderate to low quality evidence (large samples, some inconsistencies, precise, indirect) suggests no association between exposure to influenza during pregnancy and risk of schizophrenia in the offspring. Studies conducted in the United States, Europe, and Australia where 1957 pandemic came in one wave and where approximately 50% of the population reported having had influenza No significant differences in the number of cases of schizophrenia between people potentially exposed to influenza prenatally and those potentially not exposed; 6 studies, N = unclear, OR 0.99, 95%CI 0.96 to 1.03, p > 0.05, Q 15.69, p < 0.05, I 2 55% For males only: OR 0.95, 95%CI 0.87 to 1.04, p > 0.05, Q 11.66, p < 0.05, I 2 57% For females only: OR 0.96, 95%CI 0.85 to 1.09, p > 0.05, Q 3.60, p > 0.05 No significant differences in the number of cases of schizophrenia between people potentially exposed to influenza prenatally during any trimester and those potentially not exposed; First trimester: OR 0.91, 95%CI 0.85 to 0.98, p > 0.05, Q 10.93, p > 0.05 For males only: OR 0.88, 95%CI 0.75 to 1.04, p > 0.05, Q 1.37, p > 0.05 For females only: OR 0.80, 95%CI 0.63 to 1.00, p > 0.05, Q 3.16, p > 0.05 Second trimester: OR 1.00, 95%CI 0.93 to 1.07, p > 0.05, Q 3.93, p > 0.05 For males only: OR 0.99, 95%CI 0.85 to 1.16, p > 0.05, Q 11.60, p < 0.05, I 2 57% NeuRA April 2016 Page 13

14 For females only: OR 1.07, 95%CI 0.87 to 1.30, p > 0.05, Q 9.51, p > 0.05 Third trimester: OR 1.05, 95%CI 0.98 to 1.12, p > 0.05, Q 16.85, p < 0.05, I 2 58% For males only: OR 0.99, 95%CI 0.84 to 1.16, p > 0.05, Q 10.61, p > 0.05 For females only: OR 1.04, 95%CI 0.85 to 1.28, p > 0.05, Q 7.59, p > 0.05 Studies conducted in Japan where 1957 pandemic came in two waves and where approximately 50% of the population reported having had influenza No significant differences in the number of cases of schizophrenia between people in Japan potentially exposed to influenza prenatally and people in Japan potentially not exposed; 3 studies, N = unclear, RR 0.98 (CI not reported) For Japanese men: RR 0.92, 95%CI 0.81 to 1.04, p > 0.05 For Japanese women: RR 1.08, 95%CI 0.92 to 1.26, p > 0.05 Consistency in results Precision in results Directness of results Comparison 2 Some moderate inconsistencies Precise Indirect measure of exposure Studies examining the risk of schizophrenia among people born to mothers who were pregnant during the pandemic and reported having had influenza compared to mothers who were pregnant during the pandemic and reported not having had influenza. Moderate quality evidence (large sample, consistent, imprecise, direct) suggests no association between exposure to the 1957 influenza pandemic during pregnancy and risk of schizophrenia in the offspring. All areas No differences in the number of cases of schizophrenia between people whose mothers reported having had influenza during the pandemic and mothers who reported not having had influenza during the pandemic; 2 studies, N = 16,529, RR 1.20, 95%CI 0.59 to 2.42, p > 0.05 Consistency in results Precision in results Directness of results Consistent Imprecise Direct measure of exposure NeuRA April 2016 Page 14

15 Sutterland AL, Fond G, Kuin A, Koeter MWJ, Lutter R, van Gool T, Yolken R, Szoke A, Leboyer M, de Haan L Beyond the association. Toxoplasma gondii in schizophrenia, bipolar disorder, and addiction: systematic review and meta-analysis Acta Psychiatrica Scandinavica 2015; 132: View review abstract online Comparison Toxoplasma gondii antibodies in people with schizophrenia vs. controls. Moderate to low quality evidence (large samples, inconsistent, imprecise, direct) indicates that people prior to onset of schizophrenia, or those with chronic schizophrenia may have a small increase in antibodies to Toxoplasma gondii, while people with recent onset schizophrenia may have a medium size increase. Toxoplasma gondii antibodies Significant, small effect of increased Toxoplasma gondii antibodies in people with schizophrenia vs. controls; 42 studies, N ~ 60,000, OR 1.81, 95%CI 1.51 to 2.16, p < , I 2 82% Results adjusted for publication bias: OR %CI 1.21 to 1.70 Significant, medium effect of increased Toxoplasma gondii antibodies in people with recent onset schizophrenia vs. controls; 10 studies, N not reported, OR 2.18, 95%CI 1.58 to 3.01, p < , I 2 43% Significant, small effect of increased Toxoplasma gondii antibodies in people with chronic schizophrenia vs. controls; 28 studies, N not reported, OR 1.88, 95%CI 1.46 to 2.42, p < , I 2 83% Significant, small effect of increased Toxoplasma gondii antibodies prior to onset of schizophrenia vs. controls; 7 studies, N ~ 49,000, OR 1.30, 95%CI 1.05 to 1.61, p = 0.01, I 2 42% The differences in magnitude of these ORs was significant (Q B p = 0.01) NeuRA April 2016 Page 15

16 Subgroup analysis of serointensity revealed a between significant group effect (Q B = 11.2, p = 0.001), with unpublished studies showing no effect (OR 0.79) and published studies a significant effect (OR 2.17). A significant between group effect was found depending on region of study (Q B = 17.3, p = 0.004), with a higher ORs in Africa, South America, the Middle East and Asia and more modest ORs in Europe and North America. No differences in results according to study quality, study mean age, and sex. Consistency in results Precision in results Directness of results Inconsistent Imprecise Direct measure of Toxoplasma gondii antibodies Torrey EF, Bartko JJ, Lun ZR, Yolken RH Antibodies to Toxoplasma gondii in patients with schizophrenia: a metaanalysis Schizophrenia Bulletin 2007; 33(3): View review abstract online Recent update: Torrey EF, Bartko JJ, Lun ZR, Yolken RH Toxoplasma gondii and Other Risk Factors for Schizophrenia: An Update Schizophrenia Bulletin 2012; 38(3): View review abstract online Comparison Assessment of results from serological assay that measure Toxoplasma gondii antibodies in people with schizophrenia and in controls. Moderate to low quality evidence (large samples, unable to assess consistency, imprecise, direct) indicates that people with chronic or first episode schizophrenia may have increased antibodies to Toxoplasma gondii, depending on the assay used. NeuRA April 2016 Page 16

17 Differences between the number of people with schizophrenia with positive antibody test compared to the number of controls with positive antibody tests 2012 analysis Significantly higher odds of having a positive antibody test (ELIZA, CF, IHA or dye test) in people with schizophrenia compared to controls; 38 observational studies, N = 14,773, OR 2.73, 95%CI 2.21 to 3.38, p < , Q test and I 2 not reported 2007 analysis: Subgroup analysis to investigate differences between people with chronic and first episode schizophrenia No differences between chronic and first episode schizophrenia; Chronic patients vs controls; 16 observational studies, N = 8918, OR 2.79 First episode schizophrenia patients vs controls; 7 observational studies, N = 1999, OR 2.54 Subgroup analysis to investigate differences between studies using different serological tests Significant differences between results of studies using different serological tests; CF; 3 observational studies, N = 1414, OR 1.38 ELISA; 14 observational studies, N = 6087, OR 2.61 Dye test, 3 observational studies, N = 2460, OR 2.54 IHA; 3 observational studies, N = 958, OR 8.27 X , p < Subgroup analysis to investigate differences between published and unpublished data Significant differences between results of published and unpublished data studies with a higher OR are more likely to have been published; Published; 17 observational studies, N = 9563, OR 2.97 Unpublished; 6 observational studies, N = 1356, OR 2.16 X 2 4.8, p < 0.03 Consistency in results Precision in results Directness of results Inconsistent Imprecise Direct measure of Toxoplasma gondii antibodies NeuRA April 2016 Page 17

18 Explanation of acronyms CF = complement fixation, CI = Confidence Interval, CIS-R = Clinical Interview Schedule-Revised, d = Cohen s d and g = Hedges g = standardized mean differences (see below for interpretation of effect sizes), ELIZA = enzyme-linked immunosorbent assay, HR = hazard ratio, I² = the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance), IgG = immunoglobulin G, IHA = immune hemagglutination, N = number of participants, OR = odds ratio, p = probability of obtaining that result (p < 0.05 generally regarded as significant), Q = Q statistic (chi-square) for the test of heterogeneity, X 2 = chi-squared test NeuRA April 2016 Page 18

19 Explanation of technical terms * Bias has the potential to affect reviews of both RCT and observational studies. Forms of bias include; reporting bias selective reporting of results; publication bias - trials that are not formally published tend to show less effect than published trials, further if there are statistically significant differences between groups in a trial, these trial results tend to get published before those of trials without significant differences; language bias only including English language reports; funding bias - source of funding for the primary research with selective reporting of results within primary studies; outcome variable selection bias; database bias - including reports from some databases and not others; citation bias - preferential citation of authors. Trials can also be subject to bias when evaluators are not blind to treatment condition and selection bias of participants if trial samples are small 12. Different effect measures are reported by different reviews. Prevalence refers to how many existing cases there are at a particular point in time. Incidence refers to how many new cases there are per population in a specified time period. Incidence is usually reported as the number of new cases per 100,000 people per year. Alternatively some studies present the number of new cases that have accumulated over several years against a person-years denominator. This denominator is the sum of individual units of time that the persons in the population are at risk of becoming a case. It takes into account the size of the underlying population sample and its age structure over the duration of observation. Reliability and validity refers to how accurate the instrument is. Sensitivity is the proportion of actual positives that are correctly identified (100% sensitivity = correct identification of all actual positives) and specificity is the proportion of negatives that are correctly identified (100% specificity = not identifying anyone as positive if they are truly not). Weighted mean difference scores refer to mean differences between treatment and comparison groups after treatment (or occasionally pre to post treatment) and in a randomised trial there is an assumption that both groups are comparable on this measure prior to treatment. Standardised mean differences are divided by the pooled standard deviation (or the standard deviation of one group when groups are homogenous) that allows results from different scales to be combined and compared. Each study s mean difference is then given a weighting depending on the size of the sample and the variability in the data. 0.2 represents a small effect, 0.5 a medium effect, and 0.8 and over represents a large treatment effect 12. Odds ratio (OR) or relative risk (RR) refers to the probability of a reduction (< 1) or an increase (> 1) in a particular outcome in a treatment group, or a group exposed to a risk factor, relative to the comparison group. For example, a RR of 0.75 translates to a reduction in risk of an outcome of 25% relative to those not receiving the treatment or not exposed to the risk factor. Conversely, an RR of 1.25 translates to an increased risk of 25% relative to those not receiving treatment or not having been exposed to a risk factor. An RR or OR of 1.00 means there is no difference between groups. A medium effect is considered if RR > 2 or < 0.5 and a large effect if RR > 5 or < lnor stands for logarithmic OR where a lnor of 0 shows no difference between groups. Hazard ratios NeuRA April 2016 Page 19

20 measure the effect of an explanatory variable on the hazard or risk of an event. Correlation coefficients (eg, r) indicate the strength of association or relationship between variables. They are an indication of prediction, but do not confirm causality due to possible and often unforseen confounding variables. An r of 0.10 represents a weak association, 0.25 a medium association and 0.40 and over represents a strong association. Unstandardised (b) regression coefficients indicate the average change in the dependent variable associated with a 1 unit change in the dependent variable, statistically controlling for the other independent variables. Standardised regression coefficients represent the change being in units of standard deviations to allow comparison across different scales. Imprecision refers to wide confidence intervals indicating a lack of confidence in the effect estimate. Based on GRADE recommendations, a result for continuous data (standardised mean differences, not weighted mean differences) is considered imprecise if the upper or lower confidence limit crosses an effect size of 0.5 in either direction, and for binary and correlation data, an effect size of GRADE also recommends downgrading the evidence when sample size is smaller than 300 (for binary data) and 400 (for continuous data), although for some topics, this criteria should be relaxed 14. Inconsistency refers to differing estimates of treatment effect across studies (i.e. heterogeneity or variability in results) that is not explained by subgroup analyses and therefore reduces confidence in the effect estimate. I² is the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance) - 0% to 40%: heterogeneity might not be important, 30% to 60%: may represent moderate heterogeneity, 50% to 90%: may represent substantial heterogeneity and 75% to 100%: considerable heterogeneity. I² can be calculated from Q (chi-square) for the test of heterogeneity with the following formula; Indirectness of comparison occurs when a comparison of intervention A versus B is not available but A was compared with C and B was compared with C that allows indirect comparisons of the magnitude of effect of A versus B. Indirectness of population, comparator and or outcome can also occur when the available evidence regarding a particular population, intervention, comparator, or outcome is not available so is inferred from available evidence. These inferred treatment effect sizes are of lower quality than those gained from head-to-head comparisons of A and B. NeuRA April 2016 Page 20

21 References 1. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMAGroup. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. British Medical Journal. 2009; 151(4): GRADEWorkingGroup. Grading quality of evidence and strength of recommendations. British Medical Journal. 2004; 328: Torrey EF, Bartko JJ, Yolken RH. Toxoplasma gondii and Other Risk Factors for Schizophrenia: An Update. Schizophrenia Bulletin. 2012; 38(3): Arias I, Sorlozano A, Villegas E, Luna JdD, McKenney K, Cervilla J, et al. associated with schizophrenia: A meta-analysis. Schizophrenia Research. 2012; 136(1-3): Selten J-P, Frissen A, Lensvelt-Mulders G, Morgan VA. Schizophrenia and 1957 Pandemic of Influenza: Meta-analysis. Schizophrenia Bulletin. 2010; 36(2): Torrey EF, Bartko JJ, Lun ZR, Yolken RH. Antibodies to Toxoplasma gondii in patients with schizophrenia: a meta-analysis. Schizophrenia Bulletin. 2007; 33(3): Monroe JM, Buckley PF, Miller BJ. Meta-Analysis of Anti-Toxoplasma gondii IgM Antibodies in Acute Psychosis. Schizophrenia Bulletin. 2015; 41(4): Fekadu A, Shibre T, Cleare AJ. Toxoplasmosis as a cause for behaviour disorders--overview of evidence and mechanisms. Folia Parasitologica. 2010; 57(2): Khandaker GM, Zimbron J, Dalman C, Lewis G, Jones PB. Childhood infection and adult schizophrenia: A meta-analysis of population-based studies. Schizophrenia Research Khandaker GM, Zimbron J, Lewis G, Jones PB. Prenatal maternal infection, neurodevelopment and adult schizophrenia: a systematic review of population based studies. Psychological Medicine. 2013; 43: Sutterland AL, Fond G, Kuin A, Koeter MWJ, Lutter R, van Gool T, et al. Beyond the association. Toxoplasma gondii in schizophrenia, bipolar disorder, and addiction: systematic review and metaanalysis. Acta Psychiatrica Scandinavica. 2015; 132(3): CochraneCollaboration. Cochrane Handbook for Systematic Reviews of Interventions. 2008: Accessed 24/06/ Rosenthal JA. Qualitative Descriptors of Strength of Association and Effect Size. Journal of Social Service Research. 1996; 21(4): GRADEpro. [Computer program]. Jan Brozek, Andrew Oxman, Holger Schünemann. Version 32 for Windows NeuRA April 2016 Page 21