Eligibility criteria

We sought to identify all studies:

• Published between 1 January 1950 and 31 August 2021

• Reporting the incidence of psychotic disorders in any study, wholly or partially conducted in Ireland, or which provided sufficient information to derive incidence rates

• Including participants from the general adult population aged 16–64 years old

• Including any non-organic psychotic disorder according to a validated classification system (i.e. Diagnostic and Statistical Manual [DSM] (American Psychiatric Association 2013) or International Statistical Classification of Diseases and Related Health Problems [ICD] (World Health Organization 2004).)

• Published in English

Studies were excluded if they focused solely on:

• Organic psychotic disorders, or those with clear evidence of an intellectual disability

• Non-population-based samples or specialist groups (e.g. perinatal mental health, prison, military or other specialist populations)

Information sources and search strategy

We searched four electronic databases: OVID Medline, Embase, PsycInfo and Web of Science. We also screened citation lists of included publications, and contacted authors of relevant papers to identify potentially missed studies or omitted data relevant to our review.

We developed our search strategy (see Supplemental Methods) in conjunction with a UCL librarian, adapted from a previously validated search strategy for incidence studies (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019). Our search strategy included terms for (i) psychotic disorders, (ii) incidence / first episode psychosis [FEP], and (iii) a location-based term for studies restricted to the Republic of Ireland. Our initial search was conducted between March 2021 and August 2021 by one researcher (RPJ).

Study selection

After excluding duplicates, citation titles were screened independently by two authors (RPJ, JBK). Citations were rated as either excluded, insufficient information or included; discrepancies were resolved by consensus, with only citations flagged as included or as insufficient information progressing to abstract review. The same process was repeated at abstract review stage. One author (RPJ) reviewed the full text of remaining citations for eligibility, with uncertainties resolved by consensus with the senior author (JBK).

Data extraction

Data extraction was conducted by a single author (RPJ) into a prespecified spreadsheet, with uncertainties discussed and resolved with the senior author (JBK), who also reviewed the spreadsheet for accuracy. We extracted citation-level (e.g. setting, publication year, diagnostic outcomes, study quality), rate-level (e.g. number of cases, denominator data, rates and standard errors), individual-level (e.g. age, sex, ethnicity or socioeconomic status) and area-level (e.g. deprivation) data, as provided in each citation (see Supplemental Methods; Supplemental Table 2). Data were extracted from each citation according to the original metric reported in each study (i.e. the age groups reported in each study, or quintiles of deprivation).

Diagnostic outcomes

We extracted rate-level data from all identified citations according to the diagnostic codes used in each original citation. We then used algorithms previously developed for use in other national and international systematic reviews of psychotic disorders (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019; Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012) to create broad diagnostic outcomes of interest (Supplemental Table 3) following the International Classification of Diseases, 10^th edition [ICD-10]. We included the following outcomes: all psychotic disorders (ICD-10 F10-33), non-affective psychotic disorders (F20-29), schizophrenia (F20), affective psychotic disorders (F30.2, 31.2, 32.3, 33.3), bipolar disorder (F30.2, 31.2), depression with psychotic features (F32.3, 33.3) and substance-induced psychoses (F1x.5). We did not distinguish between different types of rates in our study (i.e. ‘first contact’, ‘first episode’, ‘first admission’ etc) because they typically estimate a similar latent construct of ‘treated’ incidence, with nomenclature reflecting the mode of care delivery at the time (for example, studies of ‘first admission’ tend to be older, based on case ascertainment when treatment was almost exclusively based on inpatient care). Exceptions exist, including the CAMFEPS epidemiological study, which employed population-based case finding, such that rates may be closer to the true incidence in the population, which we captured as part of our assessment of methodological quality.

Methodological quality and risk of bias of included studies

To assess risk of bias, we identified seven relevant epidemiological criteria (Supplemental Table 4, a–g) indicative of study quality for estimating incidence rates in psychiatric epidemiology, based on previous reviews (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019; Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012). These items included reporting of a defined catchment area, accurate denominator estimation, population-based case finding, standardised research diagnoses, defined inclusion criteria, and conduct of a ‘leakage’ study to identify cases potentially missed during the initial ascertainment period (see Supplemental Table 4 for more details). We also included two additional criteria (h–i) regarding the reporting of sufficient data to derive both incidence rates and associated statistical uncertainty (h), and whether crude and standardised rates were reported (i). In a small deviation from our protocol, we retrospectively dropped criterion (e) because our review did not identify any study that reported incidence by ethnicity or country-of-birth, the primary purpose for which this criterion has been used previously (Kirkbride et al., Reference Kirkbride, Fearon, Morgan, Dazzan, Morgan, Tarrant, Lloyd, Holloway, Hutchinson, Leff, Mallett, Harrison, Murray and Jones2006). Thus, each citation was scored on our risk of bias assessment from 0 to 8 (‘a–i’, except ‘e’).

Data synthesis

We adopted a narrative synthesis to describe variance in incidence of psychotic disorders in Ireland for each diagnostic group (see ‘Diagnostic outcomes’, above), where possible assessing variance by sociodemographic and socioenvironmental characteristics. In instances where several citations provided overlapping data from the same setting (for example, for a series of studies in the Cavan-Monaghan region), we reported data from the most relevant (i.e. principal) citation(s) for a given outcome. Incidence rates were reported per 100,000 person-years as our measure of effect.

Where at least five incidence rates (with corresponding standard errors) were available for a given analysis, we also conducted a random effects meta-analysis to formally pool the data, and estimated between-study heterogeneity in reported rates. Heterogeneity was estimated using the Q-test, and quantified using the I²-statistic (Higgins and Thompson Reference Higgins and Thompson2002). Where possible, we also conducted meta-regression to inspect whether reported rates varied by risk of bias (study quality). We also assessed identified citations for the presence of small study effects, the most common of which is publication bias, via visual inspection of funnel plots and a formal Egger’s test of publication bias where we had sufficient data to pool estimates. All analyses were conducted in Stata, version 17.

Study selection

After removing duplicate records, our initial search identified 1,975 citations (Fig. 1) of which 21 met our inclusion criteria (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Keatinge Reference Keatinge1986; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2012; Keatinge Reference Keatinge1987; Daly et al., Reference Daly, Webb and Kaliszer1995; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Waddington et al., Reference Waddington, Baldwin, Morgan, Quinn, Scully and Kinsella2004; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Keatinge Reference Keatinge1988; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014). Via author contacting, we identified a further two citations which met inclusion criteria (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021). Of the 23 citations included in this review (Table 1), only six unique citations contained poolable incidence rate data for meta-analyses (see below) as the remainder of citations did not provide a corresponding estimate of standard error alongside reported rates.

Figure 1. PRISMA flow diagram of search strategy.

Table 1. Overview of included citations on the incidence of psychotic disorders in Ireland

Legend: N: number of cases; NR: Not reported; ∼derived; Pop.: population.

Location: C-M: Cavan-Monaghan; DSC: Dublin South Central; D: Dublin; C&W: Clare & Wexford; M&W: Monaghan & Waterford; SD&C-M: South Dublin & Cavan-Monaghan, SD&W: South Dublin & Wicklow; CWR: Carlow, Westmeath, Roscommon.

Study: CAMFEPS: Cavan-Monaghan First Episode Psychosis Study; COPE: Carepath for Overcoming Psychosis Early; WHO: World Health Organization 10-country / DOSMED study; K-T: Keatinge thesis; K-O: Keatinge other; DETECT: Dublin East Treatment and Early Care Team; 3CS: Three-county Study.

Diagnostic confirmation method: SCID: Structured Clinical Interview for DSM Disorders; Clinical: Clinical diagnosis; PSE: Present State Examination; PPHS: Personal & Psychiatric History Schedule; SCL: Syndrome Check List; AS: Aetiology Schedule.

Diagnostic outcomes included: FEP: All first episode psychotic disorders; NAP: Non-affective psychosis; AP: Affective psychosis; SZ: Schizophrenia; BD/M: Bipolar disorder/Mania; PD: Psychotic Depression; SIP: Substance-induced psychotic disorders; O: Other psychotic disorders.

# Conference abstract.

* likely, but not reported. Based on other citations from same group/setting, or other deduction.

Study characteristics

Citations were published between 1986 (Keatinge Reference Keatinge1986) and 2021 (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021) and contained incidence data on psychotic disorders collected between 1974 (Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987) and 2016 (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021). Sample sizes ranged from just 30 new cases of bipolar disorder or mania with psychotic features in Dublin South Central (Daly et al., Reference Daly, Webb and Kaliszer1995) between 1975 and 1981, to 593 new cases of any FEP in the Dublin East Treatment and Early Care Team [DETECT] EIP service in South Dublin and Wicklow between 2006 and 2014 (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021).

Twelve citations (52.2%) reported incidence rates of psychotic disorders in the Cavan-Monaghan First Episode Psychosis Study [CAMFEPS] (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2012; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Waddington et al., Reference Waddington, Baldwin, Morgan, Quinn, Scully and Kinsella2004; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), with a further citation reporting inception rates of FEP in the COPE EIP service in the same catchment area following the end of the CAMFEPS study in 2011 (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021). Six citations provided incidence data from studies partially set in Dublin (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Daly et al., Reference Daly, Webb and Kaliszer1995; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992), including the World Health Organization [WHO] Ten-country study (Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) and three citations from the DETECT EIP service (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014). Other locations included Clare and Wexford (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Keatinge Reference Keatinge1986; Keatinge Reference Keatinge1987; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014), Waterford (Keatinge Reference Keatinge1988) and Carlow, Westmeath and Roscommon (Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987), as part of the ‘The Three-County Study’.

All included citations used either DSM-IV (n = 17; 73.9%) or ICD criteria (n = 6; 26.1%). The most common method of diagnostic assessment reported was the Structured Clinical Interview for DSM Disorders [SCID] (n = 10; 43.5%). The most common diagnostic outcomes were ‘all FEP disorders’ (n = 18; 77.3%) and ‘affective psychotic disorders’ (n = 18; 78.3%), followed by ‘non-affective psychotic disorders’ (n = 12; 52.2%) and ‘bipolar disorder with psychotic features’ (n = 12; 52.2%) (Table 1; Supplemental Figure 1). Most citations included in this review adopted an age range from 15 years and older (Table 1) with no upper cut-off (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Keatinge Reference Keatinge1986; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Keatinge Reference Keatinge1987; Daly et al., Reference Daly, Webb and Kaliszer1995; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Waddington et al., Reference Waddington, Baldwin, Morgan, Quinn, Scully and Kinsella2004; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Keatinge1988; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014). Four citations (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987) specified an upper cut-off of 64 or 65 years old, while the WHO 10-Country study was restricted to 15–54 years (Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992). One citation did not report an age range (Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010).

Risk of bias within studies

Study quality varied from one (Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003) to seven (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) out of eight, with a median of five (interquartile range [IQR]: 4–6; see Supplemental Table 5). Only one criterion was met by all studies: ‘a. Defined catchment area’, while the criterion met least often (N = 6; 26.1%) was ‘i. reporting crude and standardised rates’.

Variation in incidence rates by individual characteristics

All first episode psychotic disorders

We identified fifteen citations which estimated the overall crude incidence of all first episode psychotic disorders in Ireland (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Waddington et al., Reference Waddington, Baldwin, Morgan, Quinn, Scully and Kinsella2004; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), conducted between 1978 (Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) and 2016 (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021), and published between 1992 (Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) and 2021 (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021). Of these, 11 citations (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) provided overlapping data with four principal citations (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992); nine (Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) contained overlapping data from the CAMFEPS study (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021), and two (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014) provided overlapping data from the DETECT EIP service (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021). The four principal citations (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) provided unique information on incidence rates from 1187 FEP cases in total; crude incidence rates were similar across three studies (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992), varying from 22.0 (Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) (95%CI: 17.3–28.0) to 25.6 (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021) cases per 100,000 person-years (95%CI: 23.6–27.8), but was higher (34.1 per 100,000 person-years; 95%CI: 31.0–37.5) in one study from Cavan and Monaghan (Fig. 2a) (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021). Study quality was high (rated 6 or 7 out of 8; Table 1).

Figure 2. Crude incidence rate per 100,000 person-years of selected psychotic outcomes in the Republic of Ireland, overall and by sex. Legend: 100 kpy: 100,000 person-years. (A) All psychotic disorders; (B) All psychotic disorders, by sex; (C) Non-affective psychotic disorders; (D) Non-affective psychotic disorders, by sex.

Five principal citations (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) provided FEP incidence data separately for men and women, with evidence of higher rates for men than women (Fig. 2b). For example, Nkire et al. (Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021) reported the incidence rate ratio [IRR] for all clinically relevant psychotic disorders to be 1.3 times higher in men than women (95%CI: 1.1–1.6), while Fayyaz et al. (Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021) reported a greater crude IRR of 2.1 (95%CI: 1.4–3.1). Study quality was high (at least 6 out of 8; Table 1).

We failed to identify any citation which had published incidence rates by age group, although Omer et al. (Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) reported incidence rate ratios by age group in Cavan and Monaghan, with highest rates for men aged 15–34, although no statistically significantly differences by age for women were observed. Further data on median age-at-onset in these studies is given in the Supplemental Results section.

One citation reported FEP incidence rates by migrant status and region-of-birth (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021), from the DETECT EIP service in South Dublin and Wicklow. Although no overall difference in FEP incidence was observed between migrant and Irish-born groups (IRR: 1.02; 95%CI: 0.83–1.24), subgroup analyses found rates were 1.83 (95%CI: 1.11–3.02) times higher for people born in Africa than Ireland.

Schizophrenia

We identified 16 citations (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Scully et al., Reference Scully, Quinn, Morgan, Kinsella, O’Callaghan, Owens and Waddington2002; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Keatinge Reference Keatinge1986; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Keatinge Reference Keatinge1987; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Keatinge Reference Keatinge1988l; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992), including seven principal citations (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Keatinge Reference Keatinge1986; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Keatinge Reference Keatinge1988; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) on the overall incidence of schizophrenia. We rated study quality as high (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992) or medium (Keatinge Reference Keatinge1986; Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014; Keatinge Reference Keatinge1988; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987). These seven citations provided nine overall incidence rates of schizophrenia (with data from the Monaghan and Waterford (Keatinge Reference Keatinge1988), and Clare and Wexford (Keatinge Reference Keatinge1986) citations disaggregated by county). Incidence rates ranged from 6.4 (95%CI: 5.2–8.0) in Cavan-Monaghan (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021) to 57.0 (95%CI: 41.5–77.8) per 100,000 person-years in Waterford (Keatinge Reference Keatinge1988), representing an almost 9-fold variation.

We estimated the pooled incidence of schizophrenia as 20.0 per 100,000 person-years (95%CI: 10.6–37.5; Fig. 3), despite very high heterogeneity between estimates (I²: 97.6%), from eight estimates of incidence provided by six of the seven citations identified here (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Keatinge Reference Keatinge1986; Keatinge Reference Keatinge1988; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992); one citation could not be included as no standard error was derivable (Lyne et al., Reference Lyne, Renwick, O’Donoghue, Madigan, Kinsella and Turner2014). In a random effects meta-regression, higher study quality was associated with lower incidence of schizophrenia (p < 0.001; Fig. 4). We found no evidence of ’small study effects’ (Supplemental Figure 3; Egger’s test p = 0.77).

Figure 3. Forest plot of the incidence of schizophrenia in the Republic of Ireland, 1974–2016. Legend: 100 kpy: 100,000 person-years. The citation from Keatinge which reported data from ‘Likely Monaghan’ and ‘Likely Waterford’ is included despite the author not explicitly reporting the catchment areas of this study, only that the study took place in two Irish counties. However in that publication, the author does acknowledge the two hospitals involved in the data collection for this study, which can be traced back to the counties of Monaghan and Waterford, respectively.

Figure 4. Correlation between study quality and (log) incidence of schizophrenia in the Republic of Ireland, 1974–2016. Legend: 100 kpy: 100,000 person-years. The figure indicates strong, statistically significant (p = 0.04) negative correlation between (higher) study quality (x-axis) and (lower) (log) incidence of schizophrenia.

Five principal citations reported incidence rates of schizophrenia separately for men and women (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992), with higher point estimates consistently found for men (Supplemental Figure 4); two recent citations from separate samples in Cavan and Monaghan, including the CAMFEPS study (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021) (IRR: 3.1; 95%CI: 1.9–5.2) and COPE EIP service (IRR: 3.3; 95%CI: 1.6–7.0) reported rates that were three times higher for men than women.

Although three principal citations reported the number of new cases of schizophrenia by age group (Keatinge Reference Keatinge1986; Keatinge Reference Keatinge1988; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987), only one provided accompanying estimates of incidence (Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987), and this was underpowered to detect trends by age (Supplemental Figure 2b).

Variation in incidence rates by area-level characteristics

We identified four principal citations (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Keatinge Reference Keatinge1988; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) and one overlapping citation (Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2012) which reported the incidence of any psychotic disorder outcome by area-level characteristics, including socioeconomic deprivation (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), social fragmentation (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), population density (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016), or rural-urban classification (Keatinge Reference Keatinge1988; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014). Three of the four studies, all rated as high quality (6 or 7 out of 8) were published since 2010 (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), with one older, lower quality (4 out of 8) study published in 1988 (Keatinge Reference Keatinge1988); this study reported overlapping first admission rates of schizophrenia in rural Monaghan (51.0 per 100,000 person-years; 95%CI: 39.1–66.6) and more urban Waterford (57.0; 95%CI: 41.7–78.0).

Kelly et al. (Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010) reported first admission rates of various psychotic outcomes between 1995 and 2001 in a rural setting (Cavan-Monaghan) and an urban setting (South Dublin), and found higher crude rates of schizophrenia in men (IRR: 1.92; 95%CI: 1.52–2.44) and women (IRR: 1.34; 95%CI: 1.00–1.80) living in urban areas. By contrast, crude rates of affective psychotic disorders appeared lower in the urban setting (IRR_men: 0.48; 95%CI: 0.34–0.67) and women (IRR_women: 0.60; 95%CI: 0.43–0.83).

In Cavan and Monaghan, Omer et al. (Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) found sex-specific effects of place on the incidence of all psychotic disorders, such that women living in the most urban (IRR: 1.34; 95%CI: 1.03–1.73) and most social fragmented (IRR: 1.72; 95%CI: 1.33–2.24) parts of the region had higher rates. By contrast, these effects were not observed for men, but their rates were higher in the most deprived part of the region (IRR: 1.58; 95%CI: 1.25–2.01). Following multivariable adjustment only material deprivation remained associated with increased rates of psychotic disorder (IRR per standard deviation increase in deprivation: 1.13; 95%CI: 1.03–1.24) in this sample. This finding accords with further evidence from the DETECT service, which reported that age-standardised incidence rates of FEP were 3.43 times higher in the most-versus-least deprived quartile (95%CI: 1.24–7.75). Partial evidence (third vs first quartile) of increased rates in more socially fragmented areas was also observed (IRR: 2.40; 95%CI: 1.05–5.51) (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016). No statistically significant variation by population density or social capital was observed, although we noted a trend-level association in incidence between quartiles with the highest and lowest levels of social capital (IRR: 1.43; 95%CI: 0.99–2.06). Similarly to Omer et al. (Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014), findings from DETECT indicated a stronger effect for deprivation on FEP rates in men, and a stronger effect for social capital on FEP rates in women.

Principal findings

We identified 23 citations which met criteria for our systematic review of the incidence of psychotic disorders in Ireland, published between 1950 and 2021. Although our review highlights heterogeneity in estimates of various psychotic disorders in Ireland, some consistent trends emerged. For example, three out of four principal citations reported the overall incidence of psychotic disorders to be between 22 and 25 per 100,000 person-years in working age adults up to 65 years old, including in two recent studies based on incepted rates via EIP services in rural and more urban parts of Ireland (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021). Only one study, based on a population-based case finding approach over a 15-year period in people aged 15 years and older, reported higher rates in the rural Cavan-Monaghan region of 34.1 per 100,000 person-years (Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021). Nonetheless, this apparent variation falls within internationally reported estimates. For example, the lower estimates are consistent with pooled rates reported in the most recent systematic review of the international literature since 2002 (26.6 per 100,000 person-years) (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019). Meanwhile, the upper figure from Cavan and Monaghan is consistent with pooled estimates in England for working aged adults (16–64 years old; 31.7 per 100,000 person-years) (Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016).

In our review, we identified sufficient data to pool the overall incidence of schizophrenia in Ireland, estimated as 20.0 per 100,000 person-years (95%CI: 10.6–37.5; N = 8 settings). This is similar to the latest available estimate reported in the international systematic review (18.6; 95%CI: 15.1–22.9) (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021). Fewer studies in Ireland had published rates of affective psychotic disorders, but available data placed rates between 6-10 per 100,000 person-years. This range falls between a slightly lower figure reported in the latest international review (4.6; 95%CI: 3.1–6.8) (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021) and a higher figure reported from a review of English studies (12.4; 95%CI:9.0–17.1) (Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016). Data on the incidence of substance-induced psychoses in Ireland were sparse, but where available, were very low.

We found consistent evidence of higher rates of most psychotic outcomes in men versus women, although affective psychotic disorders occurred more equally, again consistent with the wider literature (Jongsma et al., Reference Jongsma, Turner, Kirkbride and Jones2019; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012). Very few studies in our review published incidence data by age, and no studies reported incidence by ethnicity, with only one recent study (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021) reporting rates by migrant status and region- and country-of-birth. Finally, limited available evidence in Ireland (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Keatinge Reference Keatinge1988; Kelly et al., Reference Kelly, O’Callaghan, Waddington, Feeney, Browne, Scully, Clarke, Quinn, McTigue, Morgan, Kinsella and Larkin2010; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) found support for an association between higher levels of deprivation and social fragmentation and greater psychosis rates, in studies which included both rural (Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2014) and more urban (O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016) catchment areas. There was no clear evidence that population density or urban-rural status was associated with incidence.

Meaning of the findings

Our systematic review is drawn from a longstanding tradition of high quality epidemiological studies of the incidence of psychotic disorders in Ireland since the late 1970s. Notably, over half of the identified citations are based on the population of Cavan and Monaghan, providing a rich characterisation of the incidence (and in other reports, prevalence (Youssef et al., Reference Youssef, Kinsella and Waddington1991; Scully et al., Reference Scully, Owens, Kinsella and Waddington2004)) of psychotic disorders in one rural part of Ireland. One possible reason for the higher incidence rates observed in this region than elsewhere could be that the CAMFEPS study used epidemiological principles and true community case finding to ascertain cases all cases occurring in the population, regardless of help-seeking. A further 26% of works were conducted in parts of Dublin, providing evidence from a contrasting, urban setting. These data have the potential to inform planning need for care in EIP services in Ireland, particularly with the availability of recent data on the incepted rate of psychotic disorders published from the COPE (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021) and DETECT (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016) EIP services operating in these catchments, respectively. Similar initiatives have provided evidence-based predictions of future need for EIP care in England (McDonald et al., Reference McDonald, Ding, Ker, Dliwayo, Osborn, Wohland, Coid, French, Jones, Baio and Kirkbride2021; Kirkbride et al., Reference Kirkbride, Jackson, Perez, Fowler, Winton, Coid, Murray and Jones2013), which are currently incorporated in official guidelines from the National Institute of Health and Clinical Excellence [NICE] to inform service planning and commissioning (NHS England, 2016; NHS England 2020).

Despite this, we observed less evidence about the incidence in other parts of Ireland, for some diagnostic outcomes including affective psychotic disorders and substance use disorders, and by some important characteristics (including age, ethnicity and migrant status). Future studies could address these gaps to build a more complete epidemiological understanding of psychosis in Ireland.

We identified only one study, published in 2021, which presented data on the incidence of psychotic disorders in different groups by migrant status and region-of-birth (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021). While there was no overall evidence that migrants had elevated rates of FEP (including non-affective and affective psychotic disorders separately), there was some evidence of heterogeneity by country-of-birth, with the most robust finding that rates were approximately doubled in people born in Africa compared with the Irish-born population. This effect size is somewhat lower than reported for people of Black African or Black Caribbean heritage consistently reported in studies in England (Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016), including in rural populations (Kirkbride et al., Reference Kirkbride, Hameed, Ioannidis, Ankireddypalli, Crane and Nasir2017), where rates are typically reported to be between 4-6 times greater than the White or White British comparison group. Although several explanations for this discrepancy are possible, it is worth noting that the findings in Ireland (by region-of-birth) may not be directly comparable with findings elsewhere based on ethnicity. As the Irish population becomes increasingly ethnically diverse, it will be important to precisely delineate psychosis risk by ethnicity in the future, and data from the 2022 Census will provide information about the extent of ethnic diversity in the country (the proportion of the population identifying as White Irish has decreased from 87.6% in the 2006 Census (Central Statistics Office, 2006) to 82.2% in 2016 (Central Statistics Office 2021)). This important issue emphasises the need to understand the intersectional roles of ethnicity, migrancy and region-of-origin in the aetiopathology of psychotic disorders.

To some extent the findings of our review accord with the international literature on the incidence of psychotic disorders and socioenvironmental factors in Global North contexts. Both deprivation (Croudace et al., Reference Croudace, Kayne, Jones and Harrison2000; Richardson et al., Reference Richardson, Hameed, Perez, Jones and Kirkbride2018; Lewis et al., Reference Lewis, Dykxhoorn, Karlsson, Khandaker, Lewis, Dalman and Kirkbride2020; Kirkbride et al., Reference Kirkbride, Jones, Ullrich and Coid2014) and social fragmentation (Allardyce et al., Reference Allardyce, Gilmour, Atkinson, Rapson, Bishop and McCreadie2005) have been consistently associated with the incidence of (at least) non-affective psychotic disorders. Interestingly, the non-linear association between deprivation and psychosis risk observed in the DETECT catchment by O’Donoghue et al. (Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021) (with elevated risk confined to the most deprived group only) has been reported frequently elsewhere (Croudace et al., Reference Croudace, Kayne, Jones and Harrison2000; Richardson et al., Reference Richardson, Hameed, Perez, Jones and Kirkbride2018). This suggests that a threshold of deprivation may exist, after which presentations for new onset cases of psychosis begin to increase in rate. While population density has also been longitudinally and cross-sectionally associated with the incidence of psychotic disorders in some (Richardson et al., Reference Richardson, Hameed, Perez, Jones and Kirkbride2018; Lewis et al., Reference Lewis, Dykxhoorn, Karlsson, Khandaker, Lewis, Dalman and Kirkbride2020; Kirkbride et al., Reference Kirkbride, Jones, Ullrich and Coid2014), but not all (Kirkbride et al., Reference Kirkbride, Morgan, Fearon, Dazzan, Murray and Jones2007) studies outside of the Republic of Ireland, no consistent gradient emerged in the limited data identified on this issue in this review.

Limitations

To our knowledge this was the first systematic review to examine the incidence of psychotic disorders in the Republic of Ireland. We prospectively registered our systematic review methodology and conformed to the PRISMA guidelines. Nonetheless, we acknowledge some limitations of our review. We restricted data and papers to published material identified via four major scientific databases, which may have omitted grey or unpublished data. Nonetheless, we identified rare works including PhD theses on this topic (Keatinge Reference Keatinge1986). Together with the absence of evidence of small study effects, this suggests that the number of omitted citations from this review was small. We also contacted authors of some included works, which led to the identification and inclusion of two recently published papers not identified during our original search (O’Donoghue et al., Reference O’Donoghue, Lyne, Roche, Mifsud, Renwick, Behan and Clarke2021; Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021). We recommend caution in interpreting rates from the recent COPE EIP service (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021) alongside other rates reported in this study; findings from COPE are reported as ‘incepted’ rather than ‘incidence’ rates, given that (unlike the earlier CAMFEPS study in the same region) rates were based on the number of new cases taken on by the EIP team, which may not be the same as the total number of new cases arising in this population.

One major limitation of our review was the marked heterogeneity and low number of studies generally available for any single outcome reported here. Moreover, we note that higher quality studies tended to report lower incidence rates of schizophrenia. Thus, the heterogeneity identified in this review may be explained by several factors, including differing age ranges of the studies, differences in case finding procedures, or stochastic variation. To mitigate against these issues, we relied on a narrative synthesis rather than meta-analysis.

We also note that 19 of the 23 citations included in this review were based on studies which collected data prior to (and including) 2010. Only four citations have provided more recent data on the incidence of psychotic disorder in Ireland, with the latest based on data up until 2016 (Fayyaz et al., Reference Fayyaz, Nkire, Nwosu, Amjad, Kinsella, Gill, McDonough, Russell and Waddington2021). We also noted that better study quality was strongly correlated with reporting of lower rates of schizophrenia (Fig. 4). While we found no formal evidence of small study effects (including publication bias), the Egger’s test may be underpowered when few data points (N = 8) are available. Together, these findings call for ongoing, high quality epidemiological studies of the incidence of psychotic disorder in Ireland. Importantly, these studies should collect a core set of minimum data aligned to age, sex/gender, ethnicity, migration and deprivation/urbanisation to allow precise characterisation of need for care for psychosis across the country. Although our review focussed on the adult age range, 16–64 years old, the lowest age range for any study included in our review was 15 years old (Scully et al., Reference Scully, Owens, Kinsella and Waddington2004; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016; Nkire et al., Reference Nkire, Scully, Browne, Baldwin, Kingston, Owoeye, Kinsella, O’Callaghan, Russell and Waddington2021; Keatinge Reference Keatinge1986; Baldwin et al., Reference Baldwin, Browne, Scully, Quinn, Morgan, Kinsella, Owens, Russell, O’Callaghan and Waddington2005; Morgan et al., Reference Morgan, Scully, Quinn, Kinsella, Owens and Waddington2001; Omer et al., Reference Omer, Kirkbride, Pringle, Russell, O’Callaghan and Waddington2012; Keatinge Reference Keatinge1987; Daly et al., Reference Daly, Webb and Kaliszer1995; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella and O’Callaghan2002; Waddington et al., Reference Waddington, Baldwin, Morgan, Quinn, Scully and Kinsella2004; Kingston et al., Reference Kingston, Owoeye, Kinsella, Russell, O’Callaghan, Waddington and Ritsner2011; Baldwin et al., Reference Baldwin, Scully, Quinn, Morgan, Kinsella, Owens, O’Callaghan and Waddington2003; Owoeye et al., Reference Owoeye, Kingston, Scully, Baldwin, Browne, Kinsella, Russell, O’Callaghan and Waddington2013; Keatinge Reference Keatinge1988; Browne et al., Reference Browne, Scully, Quinn, Morgan, Baldwin, Kinsella, Owens, O’Callaghan and Waddington2005; Ninuallain et al., Reference Ninuallain, O’Hare and Walsh1987; Jablensky et al., Reference Jablensky, Sartorius, Ernberg, Anker, Korten and Cooper1992), highlighting the paucity of evidence on psychosis incidence during childhood and early adolescence. Finally, our quality criteria were designed by the authors, adapted from previously published systematic reviews (Kirkbride et al., Reference Kirkbride, Errazuriz, Croudace, Morgan, Jackson, Boydell, Murray and Jones2012; O’Donoghue et al., Reference O’Donoghue, Lyne, Renwick, Lane, Madigan, Staines, O’Callaghan and Clarke2016). They may not necessarily reflect all criteria that other stakeholders in the field consider vital or relevant, and future work could explore this issue.