Discussion

In this large, multisite cohort of individuals with FEP receiving CSC, suicidal ideation and attempts declined substantially over 2 years, with approximately 76% lower odds of ideation and 85% lower odds of attempts by 24 months, confirming previous evidence of reductions in suicidality during early intervention programmes. ^{Reference Tahmazov, Bosse, Glemain, Nabbe, Guillou and Blachier5,Reference Phalen, Jones, Davis, Sarpal, Dickerson and Vatza7} The ethnic diversity of sample is an important strength, and the higher proportion of male participants is consistent with the sex distribution typically observed in FEP cohorts. ^{Reference Phalen, Jones, Davis, Sarpal, Dickerson and Vatza7} At admission, 34% reported suicidal ideation in the preceding 6 months, underscoring high baseline risk because ideation may predict later suicidal behaviour. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, ⊘hlenschlæger and Quach8} Suicide attempts were reported by 7% of participants, a rate somewhat lower than in previous cohorts, possibly reflecting differences in definitions and assessment windows. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, ⊘hlenschlæger and Quach8,Reference Addington, Williams, Young and Addington9}

Younger age was associated with increased risk, consistent with previous work showing elevated suicide risk early in the course of psychotic illness. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, ⊘hlenschlæger and Quach8,Reference Nordentoft, Laursen, Agerbo, Qin, Høyer and Mortensen10} Being female was associated with higher odds of suicidal ideation and number of suicide attempts, consistent with previous evidence. ^{Reference Austad, Joa, Johannessen and Larsen11} White and non-Hispanic participants had higher odds of suicidal ideation, although these differences are not well characterised in FEP and may also reflect differences in symptom reporting, cultural expression of distress, pathways to care or unmeasured factors such as patient–provider ethnic concordance. ^{Reference Oluwoye, Davis, Kuhney and Anglin12}

Longer DUP was associated with higher odds of ideation and attempts, consistent with previous evidence and a recent systematic review identifying longer DUP and depressive symptoms as key risk factors for suicidal behaviour in early psychosis. ^{Reference Pompili, Serafini, Innamorati, Lester, Shrivastava and Girardi1,Reference Ricci, Sarni, De Berardis, Fraccalini, Martinotti and Maina13} This finding highlights the key role of early detection in supporting people with psychosis and strengthening pathways that reduce delays to CSC entry. Previous psychiatric hospitalisation was also associated with increased risk, probably reflecting greater illness severity, consistent with elevated risk during and following in-patient treatment. ^{Reference Shrivastava, Berlemont, Campbell, Johnston, De Sousa and Shah14}

Six participants died by suicide during follow-up (0.2%), a rate lower than in some previous FEP cohorts (e.g. 2.9% over 2 years), ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, ⊘hlenschlæger and Quach8} although lower rates have also been reported (e.g. 0.9% over about 2 years). ^{Reference Robinson, Cotton, Conus, Schimmelmann, McGorry and Lambert3,Reference Addington, Williams, Young and Addington9} Direct comparisons are limited by differences in ascertainment and follow-up, because deaths were identified through discharge records and a substantial proportion of EPINET discharges reflected loss of contact. ^{Reference Phalen, Smith, Jones, Reznik, Marti and Cosgrove15} All deaths occurred among males, consistent with broader evidence that males are more likely to die by suicide even though females more often present with suicidal behaviour. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, ⊘hlenschlæger and Quach8} These deaths underscore the need for ongoing suicide risk monitoring during early psychosis treatment.

This study has several limitations. As an observational cohort without a comparison group, reductions in suicidality cannot be causally attributed to CSC. Depression was not included in adjusted models because of measurement variability and missingness, despite its relevance to suicidality. ^{Reference Ricci, Sarni, De Berardis, Fraccalini, Martinotti and Maina13} Lifetime history and both method and lethality of suicide attempts were unavailable, limiting interpretation of gender differences. Suicide-related outcomes were assessed using clinician-rated CAB items derived from routine clinical workflows rather than from standardised, suicide-specific instruments, and DUP was based on clinician estimates of psychosis onset using available patient, family and record information. These measures may have introduced variability across sites and over time. Data on CSC team diversity was unavailable, limiting interpretation of ethnic differences in suicidal ideation. Follow-up declined substantially after 2 years, and participants lost to follow-up had higher baseline suicidal ideation than completers, suggesting possible selective attrition. Although missing data remain a challenge in LHS, these real-world settings provide unique opportunities for ongoing quality improvement. ^{Reference Srihari and Keshavan6} More systematic use of evidence-based suicide risk assessment tools in early psychosis programmes may improve both clinical care and data quality in LHS.

In summary, suicide-related outcomes declined substantially during the first 2 years of specialty care in this large, multisite FEP cohort. However, individuals with longer DUP and previous hospitalisation remained at elevated risk. These findings underscore the importance of early detection, sustained suicide risk monitoring and the use of evidence-based approaches to suicide risk assessment and intervention in early psychosis treatment.