You are here
An early intervention for psychosis and its effect on criminal accusations and suicidal behaviour using a matched-cohort design
Schizophrenia Research, Volume 176, Issue 2-3, October 2016, Pages 307 - 311
Early interventions for psychosis have been shown to reduce psychotic symptoms and hospital use for first-episode patients, but the effect on suicidal and criminal behaviour has not been reliably determined. This study aimed to examine whether an early intervention for psychosis program (EPPIS) reduced criminal behaviour, suicide attempts, and hospital-based service use.
The study utilized administrative data to match clients of EPPIS to historical controls. Regression was used to determine the effect of treatment by EPPIS on inpatient use, emergency department use, suicide attempts/deaths, and criminal accusations.
A sample of 244 patients was matched to 449 controls. EPPIS patients had lower odds of being accused of a crime both during and after treatment. Suicidal behavior was less frequent among patients, both during treatment (p < 0.0001) and after (HR = 0.39; 95% CI: 0.17 to 0.94). During treatment there were more emergency department visits for the patients (RR = 2.54; 95% CI: 1.56 to 4.58), but no difference in inpatient usage compared to controls. Post-treatment, both emergency department and inpatient usage were higher among patients.
EPPIS patients had reduced suicide attempts and criminal accusations. Increased emergency department use could indicate that encouraging treatment during a crisis may increase service use, while reducing suicidal and criminal behaviour.
Keywords: Psychosis, Suicide, Emergency medicine, Health service use, Psychiatry, Crime.
Psychotic disorders are some of the most disabling mental health conditions (Casey et al, 2011 and Saha et al, 2007). These disorders can severely disrupt a person's ability to function effectively in their social and work lives (Casey et al, 2011 and Saha et al, 2007). Recently, there has been a move towards early intervention for those with psychotic symptoms with the aim of improving the care for these individuals, and also preventing disruptions to their developmental trajectories (Birchwood et al., 1998). Another goal of early intervention is to reduce the need for inpatient treatment in this population (Cocchi et al., 2011). There is evidence that early interventions for psychosis are more effective than treatment-as-usual when it comes to inpatient hospital use (McGorry et al, 1996, Chen et al, 2011, Fowler et al, 2009, Dodgson et al, 2008, Petrakis et al, 2012, Randall et al, 2015, and Cullberg et al, 2002). The effectiveness of these programs at reducing suicidal behaviours and has not been firmly established, and research is hindered by the rarity of suicide events requiring larger samples and follow up period in order to be analyzed The effect on criminal justice involvement has only been examined once, with no significant effect found (Stevens et al., 2013).
The Early Psychosis Prevention and Intervention Service (EPPIS) was recently implemented in Winnipeg, Canada. This program targets those experiencing a first episode of psychosis using a treatment modality similar to previous early intervention services (Boden et al, 2010, Bertelsen et al, 2008, and McGorry et al, 1996), including assertive case management, group and individual psychotherapy, increased focus on psychosocial treatment factors, antipsychotic medications, and family inclusion in treatment process. The main goal was to examine the impact on suicidal behaviour and justice system involvement. We examined the effect of the program on being accused of a crime, or being the victim of a crime, due to the association among psychotic disorders and violence (Douglas et al, 2009 and Large and Nielssen, 2011), and the high rate of victimization in first-episode patients (Fisher et al., 2015). The secondary objective was to determine whether EPPIS was effective at reducing inpatient and emergency department (ED) service use for patients with a first episode of psychosis, as expected based on previous research.
2.1. Data source
This study utilized information from the PATHS Data Resource which was derived from administrative databases stored in the Population Health Research Data Repository at the Manitoba Centre for Health Policy (Nickel et al., 2014). At the core of the PATHS Data Resource is the Manitoba Health Insurance Registry which includes individual-level information for Manitobans covered under the province's universal health care insurance plan. Through this registry, we have data on over 99% of individuals residing in Manitoba (Nickel et al, 2014, Roos and Nicol, 1999, and Roos et al, 2015). The administrative databases comprising the PATHS Resource contain individual-level data on all contacts with services paid for and/or provided by the Manitoba Government. Health data used for this study include physician claims for services rendered, providing counts and diagnostic information for outpatient contacts; hospital discharge abstracts; and emergency department system information. Inpatient discharge data in the repository are coded using the International Classification of Diseases (ICD) version 10-CA (Canadian version) after April 1, 2004 and ICD-9-CM prior to this date. Physician claim data for outpatient contacts were coded using ICD-9-CM throughout the period of this study. Prescription drug use for all medications dispensed within the community was obtained from the Drug Program Information Network (DPIN). Drug use is classified using Generic Drug names or Anatomical Therapeutic Chemical (ATC) classification codes. Vital statistics data were used to identify deaths classified as suicide among the study population. Justice data from the Manitoba Department of Justice were used to measure individuals' contact with the justice system. Data were acquired from the Manitoba Adolescent Treatment Centre (MATC) to identify individuals treated by EPPIS between 2003 and 2012.
2.2. Study design
In order to be included in the treatment group, patients needed to have 3 or more recorded treatment sessions in the EPPIS program. Due to an inability to find a comparable group of concurrent controls, a historical control cohort was used. Patients with similar patterns of psychosis diagnosis to EPPIS patients in the years immediately prior to the start of the program (1998–2001) were identified and included in the initial control sample. Controls were not obtained from 2002 to ensure a clear separation between control and client population in terms of their probability of being enrolled in EPPIS. Potential confounders were identified using a high-dimensional propensity score (Schneeweiss et al., 2009) model SAS® software macro, which selected variables that were correlated to both the treatment and the outcome (hospitalization). We adjusted for confounders included in the propensity score model using inverse probability of treatment weights. The sample was trimmed asymmetrically at the 2.5th and 97.5th percentiles of the propensity score (Stürmer et al., 2010). Trimming was necessary to ensure that EPPIS patients and controls were similar enough to make direct comparisons. Cases and controls were assessed during the treatment program (up to 24 months) and after they were discharged from the program. There was no limit to the length of follow-up, and therefore the controls had longer lengths of follow-up available for analysis. Standardized differences were assessed for the identified confounders to check whether the adjustment was effective at reducing differences between the two groups.
A total of eight outcomes were assessed in this study. The occurrence of suicidal behaviour was examined. This outcome was derived using ICD-10-CA, ICD-9-CM, and treatment tariff codes (details on the specific codes are available in the MCHP concept dictionary (Manitoba Centre for Health Policy, 2013). Suicidal events identified in this study contained suicide attempts identified in the hospital data and deaths attributed to suicide in vital statistics data. Two outcomes were the occurrence of criminal accusations (i.e., a patient being charged with a crime, but not necessarily prosecuted or convicted), and criminal victimization (as assessed in justice department reports). Three outcomes involved the use of inpatient services derived from the hospital discharge abstract data: the number of inpatient episodes; the number of bed days used; and whether a patient was ever hospitalized. There were two outcomes based on ED use: number of ED visits; and whether patients had ever visited the ED.
2.4. Statistical analysis
Regression was used to derive rate ratios (RR; negative binomial regression), odds ratios (OR; logistic regression), and hazards ratios (HR; Cox proportional hazards regression) and their 95% confidence intervals (95% CI). The occurrence of rate outcomes (e.g., ED episodes) was analysed using RRs. The odds of ever having an event occur were assessed using ORs. Analyses were conducted for the period of treatment and during post-treatment. For the post-treatment analysis, ORs were determined for a two-year period following discharge from the program. RRs and HRs were calculated for the post-treatment phase without limiting the follow-up period, because these methods are not affected by differential length of follow-up. Age was included as a covariate in all of the regression models. The analysis for this paper was generated using SAS software, Version 9.4 of the SAS System for Windows (SAS Institute Inc, 2011).
Ethics approval was obtained from the University of Manitoba's Health Research Ethics Board (HREB #: H2011:294) and the Government of Manitoba's Health Information Privacy Commission (HIPC #- 2013/2014 – 22).
A total of 284 cases were identified in the EPPIS program records. The initial control group consisted of 1036 individuals. Asymmetric trimming of the study cohort resulted in 244 cases and 449 controls (Table 1). In addition to sex and income quintile, the propensity score algorithm detected and adjusted for numerous diagnoses identified as potential confounders:
- personality disorders (ICD-9: 301),
- drug abuse (ICD-9: 305),
- depressive disorder (ICD-9: 311),
- refraction/accommodation disorder (ICD-9: 679),
- bronchitis (ICD-9: 490),
- soft tissue and skin disorders (ICD-9: 709, 729),
- hyperkinetic syndrome (ICD-9: 314),
- chronic nasopharyngitis (ICD-9: 472),
- macrolides and lincosamides (ATC: J01F),
- opioids (ATC: N02A),
- anti-depressants (ATC: N06A),
- ectoparasiticides (ATC: P03A),
- anticholinergic agents (ATC: N04A),
- psychostimulants and nootropics (ATC: N06B),
- viral vaccines (ATC: J07B). These covariates were identified by the SAS® macro as being variables that were related to both program involvement and hospital outcome. Notably, a prior history of suicidal behaviour was not selected as a possible confounder. These diagnoses are comorbid symptoms/conditions for psychosis. Refraction/accommodation disorder and ectoparasiticides may be diagnoses that reflect hallucinations or delusions assessed by a clinician before they are severe enough to become obvious as mental issues rather than physical issues. After weighting, the standardized differences were all under 0.10, with the exception of macrolides and lincosamides (0.101). This indicates that the control group is sufficiently similar to the EPPIS patients based on the measured covariates. Mean length of treatment was 510 days (median = 467). Available follow-up period post-treatment was longer for the control group (mean of 10.2 years versus 3.1 years). Unadjusted counts and rates for the outcomes are presented in Table 2. In general, the use of services was higher in the treatment group. However, there were no suicide attempts during treatment within the treatment group, and few attempts during the post-treatment phase (counts have been censored due to privacy concerns). During treatment 10.1% of the treatment group was accused of a crime and 3.2% were victims of a crime. Since there were no recorded suicidal events (deaths or attempts) in the EPPIS group during treatment, a regression analysis was not possible for this outcome. Pearson χ2 was used instead and demonstrated a significant difference in favour of the intervention (p < 0.0001). Through the entire post-treatment period, the occurrence of suicidal behaviour was significantly lower in the EPPIS group (HR = 0.39, 95% CI: 0.17 to 0.94). During post-treatment there were no suicide deaths in the intervention group, and fewer than 5 suicide deaths (specific number censored due to privacy concerns) among the controls. The odds of being accused of a crime were significantly lower in the EPPIS group both during treatment (OR = 0.48, 95% CI: 0.34 to 0.68) and during the 2-year post-treatment period (OR = 0.62, 95% CI: 0.47 to 0.82). The rate of accusations was also lower in the EPPIS group, but the difference was not significant (RR = 0.64, 95% CI: 0.35 to 1.16), likely due to insufficient power for this analysis. Results for being victimized were non-significant. Results of the regression analysis of inpatient use are presented in Table 3. During treatment there were no significant differences between the two groups in number of hospital episodes, number of hospital days, and the occurrence of any hospitalization. After treatment the EPPIS group had significantly higher rates of hospital episodes and bed days used, as well as a higher hazard of ever being hospitalized. Results of the ED regression analysis are presented in Table 4. Both during and after treatment, members of the EPPIS group had more ED visits (during: RR = 2.68, 95% CI: 1.56 to 4.58; after: RR = 1.76, 95% CI: 1.19 to 2.60). The odds of ever visiting the ED were also significantly higher in the EPPIS group (during: OR = 4.01, 95% CI: 3.14 to 5.11; after: OR = 1.98, 95% CI: 1.58 to 2.47).
EPPIS and control group characteristics.
a Quintile not found/determined.
Outcome counts and rates.
|Hospital episodes||Count||Crude rate||Count||Crude rate|
|Hospital days||Count||Crude rate||Count||Crude rate|
|2 year post treatment||73||33.5||108||24.2|
|ER Visits||Count||Crude rate||Count||Crude rate|
|Ever in ER||N||%||N||%|
|2 year post treatment||123||56.4||158||36.8|
|Ever attempted suicide||N||%||N||%|
|2 year post treatment||C||1.4||8||1.8|
|Accused rate||Count||Crude rate||Count||Crude rate|
|2 year post treatment||36||16.5||91||20.4|
|2 year post treatment||12||5.5||24||5.4|
C indicates value is censored due to small cell count.
ER: emergency room.
Effect of EPPIS program on inpatient service use.
|Average treatment effect (ATE)||Program effect||P value|
|During treatment||[RR. (95% CI)]||1.37 (0.78 to 2.41)||0.27|
|Post treatment||[RR. (95% CI)]||2.30 (1.49 to 3.54)||0.0002|
|During treatment||[RR. (95% CI)]||1.73 (0.57 to 5.22)||0.33|
|Post treatment||[RR. (95% CI)]||2.70 (1.35 to 5.40)||0.005|
|During treatment||[OR. (95% CI)]||1.25 (0.99 to 1.58)||0.06|
|Post treatment||[HR. (95% CI)]||1.58 (1.33 to 1.88)||< 0.0001|
|2-year post treatment||[OR. (95% CI)]||1.45 (1.14 to 1.84)||0.003|
RR: rate ratios, OR: odds ratios, HR: hazards ratios, CI: confidence interval.
All results adjusted for age.
All results weighted to adjust for identified confounders.
Bold indicates a p value of less than 0.05.
Effect of EPPIS program on emergency department service use.
Average treatment effect (ATE)
|Emergency department visits||Ever had emergency department visit|
|Program effect||P value||Program effect||P value|
|During treatment||[RR. (95% CI)]||2.68 (1.56 to 4.58)||0.0003||[OR. (95% CI)]||4.01 (3.14 to 5.11)||< 0.0001|
|Post treatment||[RR. (95% CI)]||1.76 (1.19 to 2.60)||0.004||[HR. (95% CI)]||2.34 (2.02 to 2.70)||< 0.0001|
|2-year post treatment||[OR. (95% CI)]||1.98 (1.58 to 2.47)||< 0.0001|
RR: rate ratios, OR: odds ratios, HR: hazard ratios, CI:
All results adjusted for age.
All results weighted to adjust for identified confounders.
This analysis shows that the early intervention had an impact on three important areas of behaviour in people with psychotic disorders. First, it is effective at reducing the occurrence of suicidal behaviours. To the best of our knowledge, this study is the first to show a reduction in suicidal behaviour both during and after the study period due to an early intervention. Secondly, it reduced criminal accusations in the clients. Thirdly, and counter to our expectations, the program increased the use of both inpatient and emergency department services.
Although previous early interventions have been found to be effective at reducing suicide deaths, there is only mixed evidence for a positive effect on suicidal behaviours during the treatment period, and there has not been strong evidence of a protective effect on suicidal behaviour once the treatment period ends (Goldberg et al, 2006, Harris et al, 2008, Agius et al, 2007, Cullberg et al, 2002, Bertelsen et al, 2008, Chen et al, 2011, and Grawe et al, 2006). A study on early detection programs found evidence that they reduce suicidality at the start of treatment (Melle et al., 2006). Harris et al. (2008) found a significant reduction in deaths from suicide after controlling for covariates in a proportional hazards regression analysis. This study did not analyse the effect on non-fatal suicidal behaviour. Chen et al. (2011) also found a significant reduction in deaths, with their intervention group having approximately 1/3rd the incidence of suicide deaths compared to their control group. They did not find a significant reduction for suicide attempts, despite these being more common and providing more statistical power. Studies by Agius, Graw, Cullberg, Goldberg, and Bertelsen failed to find a significant effect on the occurrence of suicide attempts (Goldberg et al, 2006, Agius et al, 2007, Cullberg et al, 2002, Bertelsen et al, 2008, and Grawe et al, 2006). Cullberg et al., 2002 found no effect on the occurrence of suicide deaths. Recent research has suggested that depressive symptoms, cannabis use, being male, living in an urban centre, poor adjustment, high anxiety and unusual thought content are specific markers of risk among first-episode patients (Ayesa-Arriola et al., 2015). Further research attempt to improve early psychosis interventions by targeting individuals with these markers. Extending the treatment period has been shown to improve functional remission and may also be beneficial at reducing suicidal behaviours (Chang et al., 2016).
There is evidence suggesting people with psychosis are more likely to engage in violent behaviour or to be victimized themselves (Fisher et al, 2015, Large and Nielssen, 2011, Douglas et al, 2009, and Maniglio, 2009). This is the first analysis showing an early intervention program with a positive effect on justice system outcomes. These improvements have long-term implications given the possible negative consequences of obtaining a criminal record. Because of the potential seriousness of the criminal justice outcomes, these findings indicate strongly that early intervention programs can have meaningful and positive impacts on psychosis patients. Another recent study examined the effect of an early intervention study on criminal convictions and found no significant reduction in those receiving early intervention versus standard care (Stevens et al., 2013). It is not clear why these studies found contradicting results. It is possible that the lower occurrence of convictions, compared to accusations, reduced the power of the other study. However, the rates in that study were nearly identical between groups.
Prior studies in this area suggest that early intervention programs similar to this program reduced the use of hospital services, although not every study showed a statistically significant effect (Randall et al., 2015). However, our findings can be interpreted as an improvement in overall treatment for psychosis patients. The program encourages patients to seek treatment when they require it, and also teaches family members about psychosis. Part of this education is recognizing when the disorder may be worsening, and connecting patients and their families to available resources, including seeking hospital treatment when necessary. The program also aims to inform families about involuntary treatment options, which may be necessary for patients experiencing rapid deterioration of function and insight. These factors may work together to cause patients in the program to seek more treatment at hospitals, and are not necessarily an indication that they are performing more poorly than the controls. The increase in service use could thus represent previously unmet treatment needs. These results could indicate that the program is ineffective at treating these individuals and causes them to use more hospital-based services. However, this explanation is not consistent with the other positive results.
4.1. Strengths and limitations
There are several key strengths to this study. We conducted a population-based analysis that adjusted for a wide range of potential confounding variables to reduce the likelihood of bias. We also used objective outcome measures not subject to attrition bias. The use of administrative data ensures that all hospital use for these individuals is recorded and available for the analysis. It is unlikely that performance bias occurred with this study design given that there were no significant, non-program-related changes between the historical control period and the program time period. There are also several limitations to the study. There is no information available about the severity of symptoms for each individual. The results could be confounded by the severity of the disorder between the groups. Historical controls were used because appropriate concurrent matching controls were unavailable, because the program was effective at enrolling its target population into the program. Controls were identified using a pattern of psychotic disorder diagnosis, but it is still possible that some differences between the groups remain with respect to the nature of their symptoms. Misclassification could occur though misdiagnosis or a coding error, but neither would systematically bias the results, since their occurrence should be similar between the two groups. The sample had a high percentage of male patients and it is unclear why. This could limit the generalizability of the sample, perhaps applying specifically to younger patients who are more likely male. The suicidal behaviour outcome does not include non-hospital treated cases. It should identify most of the serious suicidal incidents involving inpatient treatment and/or incidents resulting in death. It is unlikely that missed events will bias the results. The justice data only includes instances where victims and/or perpetrators were identified by the justice system.
This study presents several novel findings with respect to the impact of an early intervention program for psychosis patients on suicidal and criminal behaviour. Although the early intervention program resulted in increased hospital service use, the other positive outcomes of this program suggest that the treatment is effective, and that the increased service use may represent previously unmet need. Perhaps early intervention programs should place less emphasis on reducing hospital service use, instead aiming to deliver the best level of care in any setting by responding to the fluctuations in the acuity of psychosis disorders. Proper emergency department assessment and hospitalization of acute episodes could be responsible for the positive effects on suicide and justice outcomes demonstrated in this study. Future research is needed on assessing which factors involved in these programs are responsible for these effects.
Conflict of interest
JRR contributed to study design, and draft preparation and revision. DC contributed to study design, and draft preparation and revision. MS contributed to study design, and draft preparation and revision. CT contributed to study design, draft preparation and revision, and performed the data analysis. JB contributed to study design, and draft preparation and revision. LK contributed to study design, and draft preparation and revision. EB contributed to study design, and draft preparation and revision. AK contributed to study design, and draft preparation and revision. NCN contributed to study design, and draft preparation and revision. JE contributed to study design, and draft preparation and revision. MB contributed to study design, and draft preparation and revision.
The authors acknowledge the Manitoba Centre for Health Policy (MCHP) for use of data contained in the Population Health Research Data Repository under project # 2012-010 (HIPC #2013/2014-22). The results and conclusions are those of the authors and no official endorsement by MCHP, Manitoba Health, Healthy Living and Seniors (MHHLS), or other data providers is intended or should be inferred. Data used in this study were provided by MHHLS; Vital Statistics; Manitoba Jobs and the Economy; and the Manitoba Adolescent Treatment Centre. Authors would also like to thank the PATHS Equity Team for its support. PATHS Equity Team members: Charles Burchill; Mariette Chartier; Malcolm Doupe; Greg Finlayson; Randall Fransoo; Chun Yan Goh; Milton Hu; Doug Jutte; Lisa Lix; Patricia J. Martens (deceased); Colleen Metge; Colette Raymond; Les Roos; Noralou Roos; Rob Santos; Joykrishna Sarkar; Randy Walld.
- Agius et al., 2007 M. Agius, S. Shah, R. Ramkisson, S. Murphy, R. Zaman. Three year outcomes of an early intervention for psychosis service as compared with treatment as usual for first psychotic episodes in a standard community mental health team - final results. Psychiatr. Danub.. 2007;19:130-138
- Ayesa-Arriola et al., 2015 R. Ayesa-Arriola, E.G. Alcaraz, B.V. Hernandez, R. Perez-Iglesias, J.D. Lopez Morinigo, R. Duta, A.S. David, R. Tabares-Seisdedos, B. Crespo-Facorro. Suicidal behaviour in first-episode non-affective psychosis: specific risk periods and stage-related factors. Eur. Neuropsychopharmacol.. 2015;25:2278-2288 Crossref
- Bertelsen et al., 2008 M. Bertelsen, P. Jeppesen, L. Petersen, A. Thorup, J. Ohlenschlaeger, Q.P. le, T.O. Christensen, G. Krarup, P. Jorgensen, M. Nordentoft. Five-year follow-up of a randomized multicenter trial of intensive early intervention vs standard treatment for patients with a first episode of psychotic illness: the OPUS trial. Arch. Gen. Psychiatry. 2008;65:762-771 Crossref
- Birchwood et al., 1998 M. Birchwood, P. Todd, C. Jackson. Early intervention in psychosis. The critical period hypothesis. Br. J. Psychiatry Suppl.. 1998;172:53-59
- Boden et al., 2010 R. Boden, J. Sundstrom, E. Lindstrom, I.M. Wieselgren, L. Lindstrom. Five-year outcome of first-episode psychosis before and after the implementation of a modified assertive community treatment programme. Soc. Psychiatry Psychiatr. Epidemiol.. 2010;45:665-674 Crossref
- Casey et al., 2011 D.A. Casey, M. Rodriguez, C. Northcott, G. Vickar, L. Shihabuddin. Schizophrenia: medical illness, mortality, and aging. Int. J. Psychiatry Med.. 2011;41:245-251 Crossref
- Chang et al., 2016 W.C. Chang, V.W. Kwong, G.H. Chan, O.T. Jim, E.S. Lau, C.L. Hui, S.K. Chan, E.H. Lee, E.Y. Chen. Prediction of functional remission in first-episode psychosis: 12-month follow-up of the randomized-controlled trial on extended early intervention in Hong Kong. Schizophr. Res.. 2016;173:79-83 Crossref
- Chen et al., 2011 E.Y. Chen, J.Y. Tang, C.L. Hui, C.P. Chiu, M.M. Lam, C.W. Law, C.W. Yew, G.H. Wong, D.W. Chung, S. Tso, K.P. Chan, K.C. Yip, S.F. Hung, W.G. Honer. Three-year outcome of phase-specific early intervention for first-episode psychosis: a cohort study in Hong Kong. Early Interv. Psychiatry. 2011;5:315-323 Crossref
- Cocchi et al., 2011 A. Cocchi, V. Mapelli, A. Meneghelli, A. Preti. Cost-effectiveness of treating first-episode psychosis: five-year follow-up results from an Italian early intervention programme. Early Interv. Psychiatry. 2011;5:203-211
- Cullberg et al., 2002 J. Cullberg, S. Levander, R. Holmqvist, M. Mattsson, I.M. Wieselgren. One-year outcome in first episode psychosis patients in the Swedish parachute project. Acta Psychiatr. Scand.. 2002;106:276-285 Crossref
- Dodgson et al., 2008 G. Dodgson, K. Crebbin, C. Pickering, E. Mitford, A. Brabban, R. Paxton. Early intervention in psychosis service and psychiatric admissions. Psychiatr. Bull.. 2008;32:413-416 Crossref
- Douglas et al., 2009 K.S. Douglas, L.S. Guy, S.D. Hart. Psychosis as a risk factor for violence to others: a meta-analysis. Psychol. Bull.. 2009;135:679-706 Crossref
- Fisher et al., 2015 H.L. Fisher, A. Roberts, F. Day, N. Reynolds, E. Iacoponi, P.A. Garety, T.K. Craig, P. McGuire, L. Valmaggia, P. Power. Impact of crime victimization on initial presentation to an early intervention for psychosis service and 18-month outcomes. Early Interv. Psychiatry. 2015;
- Fowler et al., 2009 D. Fowler, J. Hodgekins, L. Howells, M. Millward, A. Ivins, G. Taylor, C. Hackmann, K. Hill, N. Bishop, I. Macmillan. Can targeted early intervention improve functional recovery in psychosis? A historical control evaluation of the effectiveness of different models of early intervention service provision in Norfolk 1998-2007. Early Interv. Psychiatry. 2009;3:282-288 Crossref
- Goldberg et al., 2006 K. Goldberg, R. Norman, J.S. Hoch, N. Schmitz, D. Windell, N. Brown, A. Malla. Impact of a specialized early intervention service for psychotic disorders on patient characteristics, service use, and hospital costs in a defined catchment area. Can. J. Psychiatr.. 2006;51:895-903 Crossref
- Grawe et al., 2006 R. Grawe, I. Falloon, J. Widen, E. Skogvoll. Two years of continued early treatment for recent-onset schizophrenia: a randomised controlled study. Acta Psychiatr. Scand.. 2006;114
- Harris et al., 2008 M.G. Harris, P.M. Burgess, D.C. Chant, J.E. Pirkis, P.D. McGorry. Impact of a specialized early psychosis treatment programme on suicide. Retrospective cohort study. Early Interv. Psychiatry. 2008;2:11-21 Crossref
- Large and Nielssen, 2011 M.M. Large, O. Nielssen. Violence in first-episode psychosis: a systematic review and meta-analysis. Schizophr. Res.. 2011;125:209-220 Crossref
- Maniglio, 2009 R. Maniglio. Severe mental illness and criminal victimization: a systematic review. Acta Psychiatr. Scand.. 2009;119:180-191 Crossref
- Manitoba Centre for Health Policy, 2013 Manitoba Centre for Health Policy. Concept: Suicide and Attempted Suicide (Intentional Self Inflicted Injury). (, 2013)
- McGorry et al., 1996 P.D. McGorry, J. Edwards, C. Mihalopoulos, S.M. Harrigan, H.J. Jackson. EPPIC: an evolving system of early detection and optimal management. Schizophr. Bull.. 1996;22:305-326 Crossref
- Melle et al., 2006 I. Melle, J.O. Johannesen, S. Friis, U. Haahr, I. Joa, T.K. Larsen, S. Opjordsmoen, B.R. Rund, E. Simonsen, P. Vaglum, T. McGlashan. Early detection of the first episode of schizophrenia and suicidal behavior. Am. J. Psychiatry. 2006;163(5):800-804
- Nickel et al., 2014 N.C. Nickel, D.G. Chateau, P.J. Martens, M.D. Brownell, A. Katz, E.M. Burland, R. Walld, M. Hu, C.R. Taylor, J. Sarkar, C.Y. Goh. Data resource profile: Pathways to Health and Social Equity for Children (PATHS Equity for Children). Int. J. Epidemiol.. 2014;43:1438-1449 Crossref
- Petrakis et al., 2012 M. Petrakis, S. Penno, J. Oxley, H. Bloom, D. Castle. Early psychosis treatment in an integrated model within an adult mental health service. Eur. Psychiatry. 2012;27:483-488 Crossref
- Randall et al., 2015 J.R. Randall, S. Vokey, H. Loewen, P.J. Martens, M. Brownell, A. Katz, N.C. Nickel, E. Burland, D. Chateau. A systematic review of the effect of early interventions for psychosis on the usage of inpatient services. Schizophr. Bull.. 2015;
- Roos and Nicol, 1999 L.L. Roos, J.P. Nicol. A research registry: uses, development, and accuracy. J. Clin. Epidemiol.. 1999;52:39-47 Crossref
- Roos et al., 2015 L.L. Roos, N. Nickel, P. Romano, P. Fergusson. Administrative Databases. Encyclopedia of Biostatistics (John Wiley & Sons, Ltd Wiley Stats, 2015)
- Saha et al., 2007 S. Saha, D. Chant, J. McGrath. A systematic review of mortality in schizophrenia: is the differential mortality gap worsening over time?. Arch. Gen. Psychiatry. 2007;64:1123-1131 Crossref
- SAS Institute Inc, 2011 SAS Institute Inc. SAS Software. (, 2011) (Cary, NC, USA)
- Schneeweiss et al., 2009 S. Schneeweiss, J.A. Rassen, R.J. Glynn, J. Avorn, H. Mogun, M.A. Brookhart. High-dimensional propensity score adjustment in studies of treatment effects using health care claims data. Epidemiology. 2009;20:512-522 Crossref
- Stevens et al., 2013 H. Stevens, E. Agerbo, K. Dean, P.B. Mortensen, M. Nordentoft. Reduction of crime in first-onset psychosis: a secondary analysis of the OPUS randomized trial. J. Clin. Psychiatry. 2013;74:e439-e444 Crossref
- Stürmer et al., 2010 T. Stürmer, K.J. Rothman, J. Avom, R.J. Glynn. Treatment effects in the presence of unmeasured confounding: dealing with observations in the tails of the propensity score distribution - a simulation study. Am. J. Epidemiol.. 2010;172:843-854
a Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, MB, Canada
b Manitoba Centre for Health Policy, Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, MB, Canada
c Department of Psychiatry, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada
⁎ Corresponding author at: Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, MB, Canada.
☆ Declaration of interest: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: support from CIHR and the Heart & Stroke Foundation of Canada for the submitted work; JRR reports support from the University of Manitoba's GETS program for the submitted work. No financial relationships with any organisations that might have an interest in the submitted work in the previous three years: MB reports the financial support of the Government of Manitoba through the Manitoba Centre for Health Policy Population-Based Child Health Research Award; JB reports support from a CIHR new investigator award, and a NARSAD Young Investigator Grant.
☆☆ Transparency declaration
☆☆ The lead author affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.
★ Funding: This work was supported by the Canadian Institutes of Health Research (CIHR: ROH - 115206) and the Heart & Stroke Foundation of Canada (PG-12-0534), under the program of research entitled “PATHS Equity for Children: a program of research into what works to reduce the gap for Manitoba's children.” Jason R Randall wishes to acknowledge funding from the University of Manitoba's Graduate Enhancement of Tri-Council Stipends program. Dr Brownell acknowledges the financial support of the Government of Manitoba through the Manitoba Centre for Health Policy Population-Based Child Health Research Award. Dr Bolton is supported by a Canadian Institutes of Health Research New Investigator Award (113589) and a Brain & Behavior Research Foundation NARSAD Young Investigator Grant. The funders had no role in the conduct of this study or the writing of the manuscript.
© 2016 Published by Elsevier B.V.