Christchurch Health And Development Study Research Articles

The role of social support in reducing the long-term burden of cumulative childhood adversity on adulthood internalising disorder.

Previous research indicates that social support is protective for the mental health outcomes of exposure to childhood adversity. However, the impact of social support as a protective factor following exposure to cumulative childhood adversity is understudied with prospective longitudinal data. The aim of this present study was to examine how social support mediates the impact of cumulative exposure to childhood adversity on internalising disorder in adulthood. The Christchurch Health and Development Study (CHDS) is a general population birth cohort, born in 1977 and representative of Christchurch, New Zealand at the time of the cohort members' birth. The present study used a generalised estimating equations (GEE) framework to analyse direct associations between a cumulative measure of childhood adversity (CA) and internalising disorders (major depression, and any anxiety disorder), and indirect associations through social support. Results indicated a dose-dependent relationship between increased exposure to CA and worsened odds of a diagnosis for major depression and any anxiety disorder, respectively. There was also a significant mediating effect of social support on the direct associations between CA and both major depression (OR (95%CI) =0 .98 (0.97, 0.99), p < 001) and any anxiety disorder (OR (95%CI) = .98 (0.97, 0.99), p < 001). The findings indicate that social support reduces the impact of childhood adversity on adult mental health, and is therefore a target for future work examining potential interventions following CA.

The roles of extraversion and neuroticism in the relationship between childhood adversity and adolescent substance misuse.

Although the influences of extraversion and neuroticism on the relationship between childhood adversity and substance misuse have been considered in adults, they are not yet clear and have not been examined among adolescent samples. This study sought to investigate the relationship between childhood adversity and adolescent substance misuse, alongside the influences of extraversion and neuroticism, using data from a longitudinal birth cohort study. Statistical analyses were performed on a longitudinal dataset provided by the Christchurch Health and Development Study (CHDS). After controlling for confounds, childhood adversity was a significant predictor of substance use disorder symptoms in adolescence. Moderation analyses showed that experiences of childhood adversity were most strongly associated with adolescent substance use disorder symptoms at higher levels of extraversion and at higher levels of neuroticism. Among adolescents who have experienced childhood adversity, extraversion and neuroticism may be risk factors for substance use disorder. By encouraging professionals to target adolescent substance misuse intervention and prevention approaches towards those who have experienced childhood adversity and are high in extraversion and/or neuroticism, these findings may help to diminish the prevalence of adolescent substance misuse and improve the health of adolescents.

Life course development following childhood adversity: methods and findings from the Christchurch Health and Development Study.

To review research developments on childhood adversity in the Christchurch Health and Development Study (CHDS) since 2001. Narrative overview of the published work of the CHDS since 2001 in the context of research methods. The CHDS research has continued to evolve as the cohort has aged. A clear focus has remained on the measurement of outcomes associated with psychosocial risk factors over the life course. This focus has allowed the CHDS to examine the linkages between exposure to adversity in childhood and later mental health, psychosocial and occupational outcomes across the life span to middle adulthood. The CHDS has many strengths, including prospective measurement of a broad and deep set of constructs, the use of multiple informants for data, and a range of statistical approaches suited to repeated measures longitudinal data. The CHDS has pioneered new approaches to the study of human development over the lifespan, which has been instrumental in investigating childhood adversity. The CHDS continues to provide unique information from a population cohort that has been studied for more than four decades. Future research will include examination of factors that mitigate the effects of childhood adversity and enhance resilience.

Change in the food environment and measured adiposity in adulthood in the Christchurch Health and development birth cohort, Aotearoa, New Zealand: A birth cohort study.

This study investigated associations between change in the food environment and change in measured body mass index (BMI) and waist circumference (WC) in the Christchurch Health and Development Study (CHDS) birth cohort. Our findings suggest that cohort members who experienced the greatest proportional change towards better access to fast food outlets had the slightly larger increases in BMI and WC. Contrastingly, cohort members who experienced the greatest proportional change towards shorter distance and better access to supermarkets had slightly smaller increases in BMI and WC. Our findings may help explain the changes in BMI and WC at a population level.

Validation of a Bayesian learning model to predict the risk for cannabis use disorder

Background:Cannabis use disorder (CUD) is a growing public health problem. Early identification of adolescents and young adults at risk of developing CUD in the future may help stem this trend. A logistic regression model fitted using a Bayesian learning approach was developed recently to predict the risk of future CUD based on seven risk factors in adolescence and youth. A nationally representative longitudinal dataset, Add Health was used to train the model (henceforth referred as Add Health model). Methods:We validated the Add Health model on two cohorts, namely, Michigan Longitudinal Study (MLS) and Christchurch Health and Development Study (CHDS) using longitudinal data from participants until they were approximately 30 years old (to be consistent with the training data from Add Health). If a participant was diagnosed with CUD at any age during this period, they were considered a case. We calculated the area under the curve (AUC) and the ratio of expected and observed number of cases (E/O). We also explored recalibrating the model to account for differences in population prevalence. Results:The cohort sizes used for validation were 424 (53 cases) for MLS and 637 (105 cases) for CHDS. AUCs for the two cohorts were 0.66 (MLS) and 0.73 (CHDS) and the corresponding E/O ratios (after recalibration) were 0.995 and 0.999. Conclusion:The external validation of the Add Health model on two different cohorts lends confidence to the model’s ability to identify adolescent or young adult cannabis users at high risk of developing CUD in later life.

Parenting and Home Environment in Childhood and Adolescence and Alcohol Use Disorder in Adulthood

PurposeParenting is a modifiable factor affecting the development of alcohol use disorder (AUD); however, the persistence of this effect into adulthood remains poorly understood. This study aimed to explore the longitudinal relationship between positive parenting and AUD in adulthood. MethodsData were gathered from the Christchurch Health and Development Study (CHDS), a birth cohort of 1,265 children born in Christchurch (New Zealand) in mid-1977. Positive parenting was quantified to age 16, and included the extent to which cohort members self-reported: high scores on measures of maternal and paternal care; low scores on a measure of maternal and paternal overprotection; high scores on a measure of parental attachment; low scores on a measure of parental intimate partner violence; and occasional or no use of physical punishment. Outcome measures were AUD incidence and symptoms at ages 15–35, with potential confounding factors and time-dynamic covariates included. ResultsThere was a significant association between positive parenting and AUD outcomes, with higher levels of positive parenting associated with a lower incidence of AUD and AUD symptoms. Controlling for confounding factors reduced the association between positive parenting and AUD outcomes, but they remained statistically significant. Adjustment for mental health, life stress, and employment reduced the magnitude of the association between positive parenting and alcohol outcomes to statistical nonsignificance. ConclusionsParenting factors in childhood and adolescence are linked to AUD outcomes in adulthood, as well as mental health, substance use, and life stress. Investment in positive parenting in adolescence may reduce AUD and associated harms in adulthood.

SA53 - METHYLOME IMPACTS OF CANNABIS USE IN THE CHRISTCHURCH HEALTH AND DEVELOPMENT STUDY

Background Cannabis has been very much in the news of late, with controversy and debate surrounding legalisation and medicinal uses purposes. However, there is also strong evidence for negative consequences of prolonged cannabis use. Significant contributions in this area have been made by the Christchurch Health and Development Study (CHDS), a longitudinal study of a birth cohort of 1265 children born in 1977 in Christchurch (New Zealand), who have been studied on 23 occasions from birth to the age of 35 (n=987). Methods It is now clear that many environmental factors, including drugs, impact on the pattern of epigenetic marks throughout the genome. We sought to explore the impacts of heavy cannabis use on the methylome in participants of the CHDS. We selected DNA (purified from peripheral blood samples) of 96 CHDS subjects including 49 heavy long-term cannabis smokers (25 of whom also smoked tobacco), and 47 non-smoking, non-cannabis using controls. Epigenome wide analysis of methylation was carried out using Illumina 850 K EPIC arrays. Results Our key observations were that for the cannabis plus tobacco smokers compared with non-smoking controls, the most differentially hypermethylated sites were in the AHRR and F2RL3 genes (adjusted P 4.4×10^-8, 4.9×10^-6 respectively), replicating multiple previous studies on methylomic effects of tobacco. For those sites (~15000) significantly showing greater than 5% differential methylation after multiple comparison adjustment, 99% were hypermethylated. Comparing cannabis only users with non-smoking controls, for those sites (~4500) significantly showing greater than 5% differential methylation after multiple comparison adjustment, 99% were hypomethylated. The differentially methylated sites which were hypermethylated in tobacco smokers but hypomethylated in cannabis only users were most strongly represented in PI3K-Akt and oxytocin KEGG pathways (FDR corrected p values Discussion These data suggest that cannabis has quite different methylome effects to tobacco smoking, and that there may be interaction effects which modify the methylome impacts when both cannabis and tobacco are smoked. We are seeking to validate a proportion of these methylation changes using bisulfite-amplicon sequencing in a wider selection of samples from the CHDS and other cohorts. These analyses may help to identify epigenome impacts of cannabis use, and point to gene regulatory processes that underlie both potentially positive as well as adverse consequences of its use.

Ethnic Differences in Adolescent Mental Health Problems: Examining Early Risk Factors and Deviant Peer Affiliation

Using life-course longitudinal data from the Christchurch Health and Development Study (CHDS), we examined ethnic differences in rates of psychiatric disorder among New Zealand adolescents, comparing New Zealand Māori to their European peers. The CHDS includes a large birth cohort of New Zealand children who have been regularly assessed throughout childhood, adolescence, and adulthood. Ethnicity (New Zealand Māori vs. non-Māori) was assessed at age 14 years. Internalizing and externalizing disorders were assessed at age 15, 16, and 18 years. A diverse range of risk factors were included to test if differences in mental health difficulties were explained by greater exposure to early life-course challenges. We found significant differences in rates of mental disorder during adolescence between New Zealand Māori and non-Māori, but these differences were largely explained by early developmental challenges and adolescent peer affiliations. Differences across the two ethnic groups in rates of internalizing disorders were explained by the increased exposure amongst Māori to socioeconomic disadvantage during childhood, while differences in externalizing disorders were explained by greater exposure amongst Māori to childhood family adversity and deviant peer affiliation during adolescence. The findings point to the significant influence of cumulative early life-course risk factors in accounting for the ethnic differences between Māori and non-Māori in the development of adolescent internalizing and externalizing disorders.

Exposure to parental separation in childhood and later parenting quality as an adult: evidence from a 30-year longitudinal study.

Previous research has documented that exposure to parental separation/divorce during childhood can be associated with long-term consequences into adulthood. This study sought to extend this literature by examining associations between childhood exposure to parental separation/divorce and later parenting behavior as an adult in a New Zealand birth cohort. Data were drawn from the Christchurch Health and Development Study (CHDS), a longitudinal study of a birth cohort of 1,265 children born in 1977 in Christchurch, New Zealand. Information about exposure to parental separation and divorce was gathered annually from birth to 15years. At the 30-year follow-up, all cohort members who had become parents (biological or nonbiological) were assessed on several parenting dimensions (sensitivity, warmth, overreactivity, inconsistency, quality of child management, and physical punishment). The analyses showed that exposure to more frequent parental separation in childhood and adolescence was associated with lower levels of parental sensitivity and warmth, greater overreactivity, and an increased use of physical punishment as a parent, after controlling for a wide range of family socioeconomic and psychosocial factors, and individual child characteristics. The findings suggest that as exposure to parental separation increases, so does the likelihood of experiencing multiple developmental challenges in childhood and adolescence. As an adult, these life-course experiences can have small but significant associations with the quality of parenting behavior.

Life-stress and reactivity by gender in a longitudinal birth cohort at 30 and 35years.

Previous literature has shown gender differences in reactivity to stressful life events. However, it is unclear whether gender differences in stress reactivity are consistent across a series of life event domains among longitudinal adult sample populations. Data were gathered from the Christchurch Health and Development Study (CHDS). The CHDS is a longitudinal birth cohort of 1265 children born in 1977 in Christchurch, New Zealand. Cohort members were questioned on their experience of, and distress from, a series of life event domains (interpersonal problems; victimization; illness/death; pregnancy/parenthood; employment/finance problems) spanning two age-periods 25-30years (data collected in 2007) and 30-35years (data collected in 2012). The data were pooled across observations and analyzed using population-averaged repeated-measures regression methods. Overall, men and women reported experiencing similar numbers of life events for each domain. However, men reported more victimization and more employment/financial problems; women reported more illness/death events. Women reported experiencing more distress per life event for the domains of interpersonal problems, illness/death and pregnancy/parenthood. Men and women reported similar distress per life event for the victimization and employment/finance domains. The results were robust to control for: child and adolescent factors (childhood abuse exposure; adolescent personality; mental health) and adult factors (mental health; self-esteem). These findings are consistent with a growing body of evidence indicating that some life events including interpersonal problems, illness/death and pregnancy/parenthood may be intrinsically more distressing for women. Detection of life event distress is important to aid in the prevention of mental/physical health problems.

Adiposity and psychosocial outcomes at ages 30 and 35.

To examine associations between adiposity and adult psychosocial outcomes (depressive symptoms, life satisfaction, self-esteem, household income, personal income, savings/investments) in a New Zealand birth cohort, by gender. Adiposity was assessed using Body Mass Index scores classified on a 3-point scale of BMI: <25.0, overweight (25.0-29.9) or obese (≥30). Data were gathered via face-to-face and telephone interviews for the Christchurch Health and Development Study (CHDS), comprising a birth cohort of 1265 children born in Christchurch, New Zealand, in 1977. BMI and psychosocial outcome information was collected in 2007 (30 years; n = 977) and in 2012 (35 years; n = 923). Population-averaged regression modeling showed evidence of statistically significant (p < 0.05) associations between increasing adiposity and adverse psychosocial outcomes for females, but not for males. After adjustment for child and family background covariates the strength of the associations for females was reduced; with four associations (depressive symptoms, life satisfaction, equivalized household income and savings/investments) remaining statistically significant (p < 0.05). In contrast, for males there was a significant (p = 0.008) positive association between adiposity and higher personal net weekly income after covariate adjustment. The findings suggest evidence of gender differences in the associations between adiposity and psychosocial outcomes. For females, there were small but pervasive tendencies for increasing adiposity to be related to more adverse mental health, psychological well-being and economic outcomes; whereas for males adiposity was either unrelated to these outcomes, or in the case of personal income, associated with greater economic advantage. The implications of these findings are discussed.

Tobacco smoking and cannabis use in a longitudinal birth cohort: Evidence of reciprocal causal relationships

BackgroundThere is evidence of associations between tobacco and cannabis use that are consistent with both a classical stepping-stone scenario that posits the transition from tobacco use to cannabis use (‘gateway’ effect of tobacco) and with the reverse process leading from cannabis use to tobacco abuse (‘reverse gateway’ effect of cannabis). The evidence of direct causal relationships between the two disorders is still missing. MethodsWe analysed data from the Christchurch Health and Development Study (CHDS) longitudinal birth cohort using advanced statistical modelling to control for fixed sources of confounding and to explore causal pathways. The data were analysed using both: (a) conditional fixed effects logistic regression modelling; and (b) a systematic structural equation modelling approach previously developed to investigate psychiatric co-morbidities in the same cohort. ResultsWe found significant (p<0.05) associations between the extent of cannabis use and tobacco smoking and vice versa, after controlling for non-observed fixed confounding factors and for a number of time-dynamic covariate factors (major depression, alcohol use disorder, anxiety disorder, stressful life events, deviant peer affiliations). Furthermore, increasing levels of tobacco smoking were associated with increasing cannabis use (p=0.02) and vice versa (p<0.001) over time. ConclusionsOur results lend support to the notion of both of ‘gateway’ and ‘reverse gateway’ effects. That is, the association between tobacco and cannabis use arises from a reciprocal feedback loop involving simultaneous causation between tobacco use disorder and cannabis use disorder.

Personality characteristics in childhood and outcomes in adulthood: findings from a 30 year longitudinal study.

Personality has been associated with a variety of outcomes in adulthood. Most of the literature related to mental state disorder and personality is cross sectional. Data from more than 900 participants of the Christchurch Health and Development Study (CHDS) were examined. Extroversion and neuroticism were measured at 14 years old and social outcomes at age 30. The presence of mental state disorder between 18-30 years old was identified. Multiple potential confounders in childhood were included in the analysis. Neuroticism at fourteen was significantly correlated with multiple environmental exposures whereas extroversion had relatively few associations. Regression analysis found that neuroticism at 14 predicted depression, anxiety, suicidality and overall mental health problems at 30 as well as poor self-esteem but not relationship quality or wellbeing. Extroversion at 14 predicted alcohol and drug dependence and overall mental health problems, but also predicted improved social wellbeing, self-esteem and relationship quality at 30. In this analysis extroversion interacts with significantly fewer environmental factors than neuroticism in predicting adult outcomes. Neuroticism at 14 years predicts poorer mental health outcomes in adulthood. Extroversion in childhood may be a protective factor in the development of mental disorder other than alcohol use disorders. Extroverted adolescents have more positive social outcomes at 30 years.

No association between mean telomere length and life stress observed in a 30 year birth cohort.

Telomeres are specialised structures that cap the ends of chromosomes. They shorten with each cell division and have been proposed as a marker of cellular aging. Previous studies suggest that early life stressors increase the rate of telomere shortening with potential impact on disease states and mortality later in life. This study examined the associations between telomere length and exposure to a number of stressors that arise during development from the antenatal/perinatal period through to young adulthood. Participants were from the Christchurch Health and Development Study (CHDS), a New Zealand longitudinal birth cohort which has followed participants from birth until age 30. Telomere length was obtained on DNA from peripheral blood samples collected from consenting participants (n = 677) at age 28–30, using a quantitative PCR assay. These data were assessed for associations with 26 measures of life course adversity or stress which occurred prior to 25 years of age. No associations were found between telomere length measured at age 28–30 years and life course adversity or stress for specific measures and for the summary risk scores for each developmental domain. The correlations were very small ranging from −0.06 to 0.06 with a median of 0.01, and none were statistically significant. Our results in this well-studied birth cohort do not support prior reports of such associations, and underscore the need for more extensive replication of proposed links between stress and telomere biology in larger cohorts with appropriate phenotypic data.

Commentary: (How) does the experience of parental divorce in childhood contextualize adult development? Reflections on Fergusson et al. (2014)

Fergusson, McLeod, and John Horwood (2014) have raised many interesting issues in their article considering the effects of parental separation/divorce in childhood on later adult partnership outcomes. In particular the article has contributed to our knowledge of the intergenerational transmission of marital instability. In my Commentary article I have: (a) situated the Christchurch Health and Development Study (CHDS) findings within the substantial body of literature on the intergenerational transmission of marital instability, (b) placed these findings within their sociocultural context, and (c) considered their clinical implications. I then consider the important role that the CHDS study has played in revealing the relative contributions of childhood family structure, social context, and family processes to patterns of intimate relationship formation in adulthood.

Parental separation/divorce in childhood and partnership outcomes at age 30

Previous research has found that children exposed to separation/divorce may also experience relationship problems in adulthood. The aim of this investigation was to examine this issue in a birth cohort of over 900 New Zealand children studied to age 30. Data were gathered over the course of the Christchurch Health and Development Study (CHDS). The CHDS is a 30year longitudinal study of a birth cohort of 1265 children born in Christchurch (NZ) in 1977. The data collected included the following: (a) timing and number of parental separations and divorces from birth to 15years; (b) partnership outcomes (16-30years) of the number of cohabiting/marriage partnerships; positive partner relations; negative partner relations; partner adjustment/conduct problems; and interpartner violence victimization and perpetration; and (c) potential covariate factors. Study findings showed the presence of significant associations between childhood parental separations/divorces and number of cohabiting/marriage partnerships (16-30years) (p<.001), negative partner relations (p=.021), extent of partner adjustment/conduct problems (p<.001), and perpetration of interpartner violence (p=.018). Childhood parental separation/divorce explained less than 2.5% of the variance in partnership outcomes. These associations were explained statistically by a series of covariate factors associated with childhood parental separation/divorce including parental history of illicit drug use, childhood sexual abuse, childhood conduct problems (7-9years), interparental conflict and violence, childhood physical punishment/maltreatment, family socio-economic status at the child's birth, and parental history of criminality. Tests of gender interaction showed that the effect of childhood parental separations/divorces may be the same for males and females. Analysis of the number of childhood parental separations/divorces experienced into three age groups (birth to 5, 5-10years and 10-15years) yielded similar results. These findings suggest that the general associations between childhood parental separation/divorce and partner relationships in adulthood reflect the consequences of various contextual factors that are associated with childhood parental separation.

First positive reactions to cannabis constitute a priority risk factor for cannabis dependence

To assess the association between first reactions to cannabis and the risk of cannabis dependence. A cross-sectional population-based assessment in 2007. A campus in a French region (Champagne-Ardennes). A total of 1472 participants aged 18-21 years who reported at least one life-time cannabis consumption, of 3056 students who were screened initially [the Susceptibility Addiction Gene Environment (SAGE) study]. Positive and negative effects of first cannabis consumptions, present cannabis dependence and related risk factors were assessed through questionnaires. The effects of first cannabis consumptions were associated dose-dependently with cannabis dependence at age 18-21 years, both according to the transversal approach of the SAGE study and to the prospective cohort of the Christchurch Health and Development Study (CHDS) assessed at the age of 25 years. Participants of the SAGE study who reported five positive effects of their first cannabis consumption had odds of life-time cannabis dependence that were 28.7 (95% confidence interval: 14.6-56.5) higher than those who reported no positive effects. This association remains significant after controlling for potentially confounding factors, including individual and familial variables. This study suggests an association between positive reactions to first cannabis uses and risk of life-time cannabis dependence, this variable having a central role among, and through, other risk factors.

Childhood exposure to sexual abuse and partnership outcomes at age 30

In this study, 30-year longitudinal data from the Christchurch Health and Development Study (CHDS) were used to examine the associations between childhood exposure to sexual abuse and intimate relationship outcomes at age 30. In addition, a broad range of early childhood and family confounding factors were tested, and the role of intervening factors from adolescence was explored. The investigation analyzed data from a birth cohort of over 900 New Zealand adults studied to the age of 30. At ages 18 and 21 cohort members reported on any exposure to sexual abuse prior to age 16. This information, along with prospective data gathered in childhood and adolescence, was used to predict partnership outcomes at age 30. After adjustment for early childhood and family factors, exposure to more severe forms of childhood sexual abuse (CSA) was associated with earlier and more frequent cohabitation, higher rates of perpetrated interpartner violence (IPV), and early parenthood, lower relationship satisfaction and investment. Several factors from adolescence partially or fully mediated these associations, notably a history of early consensual sexual intercourse, higher number of sexual partnerships, substance abuse problems, and self-esteem. After adjustment for intervening factors, exposure to CSA remained significantly associated with IPV. The findings support a causal chain process, whereby early childhood and family factors place some individuals at risk for CSA. The extent of CSA exposure is related to adolescent risk taking, which in turn leads to early and more frequent cohabitation, risk of IPV, and lower relationship satisfaction and investment.

Anxiety disorders and suicidal behaviours in adolescence and young adulthood: findings from a longitudinal study

The aim of this study was to estimate the extent to which anxiety disorders contribute to an increase in suicidal behaviour after controlling for both observed and non-observed sources of confounding. Data were collected from the Christchurch Health and Development Study (CHDS), a 25-year longitudinal study of over 1000 participants. Measures of anxiety disorders [phobia, generalized anxiety disorder (GAD), panic disorder], major depression (MD), substance use disorders, conduct/antisocial personality disorder, stressful life events, unemployment, and suicidal ideation/attempts for subjects aged 16-18, 18-21 and 21-25 years were used to fit random and fixed effects regression models of the associations between anxiety disorders and suicidal behaviours. Anxiety disorders were strongly associated with suicidal ideation/attempts. Any single anxiety disorder increased the odds of suicidal ideation by 7.96 times [95% confidence interval (CI) 5.69-11.13] and increased the rate of suicide attempts by 5.85 times (95% CI 3.66-9.32). Control for co-occurring mental disorders, non-observed fixed confounding factors and life stress reduced these associations [suicidal ideation odds ratio (OR) 2.80, 95% CI 1.71-4.58; suicide attempts incidence rate ratio (IRR) 1.90, 95% CI 1.07-3.39]. Rates of suicidal behaviour also increased with the number of anxiety disorders. Estimates of the population attributable risk suggested that anxiety disorders accounted for 7-10% of the suicidality in the cohort. Anxiety disorders may be a risk factor for suicidality, even after controlling for confounding, with risks increasing with multiple anxiety disorders. Management of anxiety disorders may be an important component in strategies to reduce population rates of suicide.

Tests of causal linkages between cannabis use and psychotic symptoms

To examine possible causal linkages between cannabis use and psychosis using data gathered over the course of a 25-year longitudinal study. A 25-year longitudinal study of the health, development and adjustment of a birth cohort of 1265 New Zealand children (635 males, 630 females). The Christchurch Health and Development Study, a general community sample. A total of 1055 participants from the Christchurch Health and Development Study (CHDS) cohort for whom data on cannabis use and psychotic symptoms were available on at least one occasion from 18, 21 and 25 years. As part of this study, data were gathered on frequency of cannabis use and psychotic symptoms at ages 18, 21 and 25 years. Regression models adjusting for observed and non-observed confounding suggested that daily users of cannabis had rates of psychotic symptoms that were between 1.6 and 1.8 times higher (P < 0.001) than non-users of cannabis. Structural equation modelling suggested that these associations reflected the effects of cannabis use on symptom levels rather than the effects of symptom levels on cannabis use. The results of the present study add to a growing body of evidence suggesting that regular cannabis use may increase risks of psychosis. The present study suggests that: (a) the association between cannabis use and psychotic symptoms is unlikely to be due to confounding factors; and (b) the direction of causality is from cannabis use to psychotic symptoms.