A systematic review of factors prolonging or reducing the duration of untreated psychosis for people with psychosis in low-and middle-income countries

Aim: This review aims to identify factors that may prolong or reduce the duration of untreated psychosis for people with psychosis in low-and middle-income countries. Methods: Electronic searches of six databases were conducted, to find studies from low-and middle-income countries on people with psychotic disorders provided they statistically measured an association between factors that may prolong or reduce the duration of untreated psychosis. Studies were critically appraised and a narrative synthesis exploring differences between and within studies is presented. A socio-ecological model is used to convey the main findings. Results: Thirty studies of 16 473 participants in total were included in this review. Taken together participants were 51.5% male and 48.5% female. Various factors potentially associated with longer duration of untreated psychosis for people with psychosis in low-and middle-income countries were found. Examples of these factors are an insidious mode of onset, greater family stigma and low social class. Other factors, such as marital status, educational level, diagnostic type, predominant symptoms and employment status, yielded inconsistent results. Conclusions: The methodological quality of the included studies limits the conclusions of this review. The results indicate an urgent need for further high-quality research in these countries. The socio-ecological model is a helpful framework for clinicians, scholars, and decision-makers to conceptualize factors that may affect the duration of untreated psychosis, highlight gaps in the literature as well as reflect on potential prevention strategies that may ultimately support early intervention services for people with psychosis in developing countries.

causes of disability globally and may lead to premature death (Correll et al., 2018).These have a relatively high prevalence and a lifetime risk of 3%-3.5% (Malla & McGorry, 2019).The onset of psychotic symptoms may create marked distress in affected individuals and families (NICE, 2014).The personal, social, and financial repercussions of psychotic disorders make them a major public health issue (Malla & McGorry, 2019).
In recent decades, in response to often suboptimal clinical outcomes of people with first-episode psychosis (FEP), early intervention services (EIS) have been developed (Bird et al., 2010).EIS have spread across high-income countries (HICs) (Bird et al., 2010) and recently gained attention in low-and middle-income countries (LMICs) (Singh & Javed, 2020).There is mounting evidence that this form of secondary prevention may positively impact employment outcomes, global functioning, and quality of life as well as reduce psychotic symptoms, long-term disability, and comorbidities (Mwesiga et al., 2021).High-quality evidence has demonstrated that their effectiveness far exceeds usual care (Malla & McGorry, 2019).EIS include recommended treatments, for example, initiation of low-dose antipsychotic medication combined with psychoeducation (WHO, 2016).
Reducing the duration of untreated psychosis (DUP) is also a fundamental objective of EIS (NICE, 2014), with the critical period hypothesis suggesting that a patient's prognosis is determined during the early stages of psychosis-meaning that the longer the DUP, the worse the patient's clinical outcomes (Fabri Cabral & Chaves, 2009).
DUP has similarly received increasing attention in research and clinical settings since the 1980s (Devi Thakoor et al., 2016).Although there are disagreements as to the exact definition of DUP, the current literature broadly defines it as the period between the appearance of psychotic symptoms to the start of initiating antipsychotic medication (Kaminga, Dai, et al., 2019).Delayed help-seeking is a common occurrence for people experiencing psychotic symptoms and this has led to DUP being reported in terms of months and even years (Sharifi et al., 2009).There is considerable variation in reported lengths of DUP depending on the population studied (Sadeghieh Ahari et al., 2013).
Research has yet to identify the distinct mechanisms underlying an association between DUP and clinical outcomes (Fresan et al., 2020).A range of factors have been postulated to affect the DUP (e.g., sociocultural factors, Qiu et al., 2017), as a potentially modifiable prognostic factor.Possible factors influencing the DUP have been identified (Fresan et al., 2020), although results from different studies are often conflicting (Okasha et al., 2016).Given that treatment delay appears to have negative implications for people with psychosis (Chen et al., 2019), understanding the potential contributing factors to DUP could facilitate the identification, planning and implementation of targeted efforts in hopes of reducing DUP, improving outcomes of people with psychosis and alleviating suffering (Kaminga, Dai, et al., 2019).This is a particularly pressing matter in LMICs where approximately 80% of all FEP occur (Singh & Javed, 2020).Early intervention for people with psychosis in developing countries is high on the agenda of the World Psychiatric Association (Vaitheswaran et al., 2021).Despite unique challenges, such as limited funding and scarcity of specialized mental health workers (Singh & Javed, 2020), EIS have been implemented across LMICs (Corrêa-Oliveira et al., 2021).However, benefits of EIS may not be fully exploited if little attention is given to reducing DUP (Dama et al., 2019).It therefore seems beneficial to work on both EIS and DUP simultaneously (Dama et al., 2019), especially since research has highlighted a substantially longer average DUP in LMICs (Maric et al., 2016) and DUP has been associated with an unfavourable prognosis for people with psychosis in LMICs too (Burns & Kirkbride, 2012).
The purpose of this review is to systematically assess studies to answer the research question: What are the factors prolonging or reducing the duration of untreated psychosis for people with psychosis in LMICs?
A Socio-Ecological Model (SEM) was chosen to help conceptualize and illustrate the diverse factors potentially affecting DUP, and their overlapping influence on people with psychosis in LMICs (McCloskey et al., 2011, pp. 20-23).SEM was originally proposed in the 1980s as a theory-based framework for understanding the many interactive forces affecting health (Kilanowski, 2017).Since then, it has been used across disciplines as a framework for prevention (Centers for Disease Control and Prevention, 2022).

| Screening and selection process
All retrieved search results were imported into EndNote.Titles and abstracts were independently screened for relevance.All potentially eligible papers were fully reviewed and assessed against the predetermined eligibility criteria developed using the PICOSS method (Boland et al., 2017, p.51) and outlined in Table 1.Only studies written in English were accepted and grey literature data were excluded.
Backward and forward citation searching were performed on included studies using Google Scholar (Boland et al., 2017, pp. 71-72).Studies that did not meet the inclusion criteria were excluded.Borderline cases were discussed and resolved through consensus with LB.

| Data extraction
Data summary tables were created to summarize descriptive and analytic data from the studies.The study design was determined for several studies using the Centre for Evidence-Based Medicine Study Designs resource (www.cebm.ox.ac.uk, n.d.).Age is reported verbatim.
Country income level was defined using World Bank Data (2022).

| Quality assessment
Both the AXIS tool for cross-sectional studies and the SIGN methodology checklist for case-control studies were used as they share comparable scales assessing quality of reporting, study design quality and risk of bias (Downes et al., 2016;SIGN, 2021).The AXIS tool is a concise tool of 20 questions developed by medical experts (Downes et al., 2016).The SIGN methodology checklist for case-control studies is a 14-question checklist developed by the SIGN executive in collaboration with healthcare professionals, individuals, and patient organisations (SIGN, 2021).
An overall category assessment for each was defined as excellent, good, fair, or poor.'Excellent' was assigned when all studies fulfilled all criteria for a category, 'good' when all studies met >50% of criteria for a category, 'fair' when the majority of studies fulfilled >50% of the criteria for a category and 'poor' when no study met 50% of the criteria for a category.

| Synthesis and analysis methods
All studies underwent data synthesis, and no data were suitable for a meta-analysis given the heterogeneity of the studies (Boland et al., 2017, p.142-145).A narrative synthesis summarising and exploring key findings between and within studies is presented.This review adheres to the PRISMA checklist guideline (Page et al., 2021) and was informed by the Synthesis Without Meta-Analysis guideline (Campbell et al., 2020).Main findings are presented using SEM levels (Centers for Disease Control and Prevention, 2022; Figure 1).

| Study selection
The initial search taken together yielded 432 articles.After removing duplicates, the title and abstract of 287 articles were reviewed.Of the 23 reports subsequently sought for retrieval, two reports were not accessed due to the inability to contact the authors.A remaining 14 articles were ultimately included.Backward and forward citation searching of these articles yielded 17 additional articles that were assessed for eligibility, bringing the total number of included articles included to 30.For details, please see Figure 2. Almost all studies (n = 27) were from middle-income countries.

| Study characteristics
Sample size ranged from 37 (Ananthi et al., 2017) to 7252 participants T A B L E 1 Key eligibility criteria, in terms of the PICOS framework.(Chen et al., 2019), with a median sample size of 107.5.Of the 16 473 participants, 51.5% were male and 48.5% were female.Please see

| Factors and instruments
DUP was inconsistently defined, although all studies defined the onset of psychosis as the start of DUP more than half of the studies (n = 18) did not provide details on this.The rest incorporated further information in their definition, on the type of psychotic symptoms, duration of symptoms, or who observed or reported these symptoms.

| Main findings
Findings are presented using four SEM levels: individual, relationship, community and societal.Please refer to Table 3 for more details.

| Individual level
Many investigated factors generated conflicting results.Two studies (Kini et al., 2015;Okasha et al., 2016) found that a lower age was associated with shorter DUP while others (n = 8) did not find an association between the two.Two studies reported an association between male gender and longer DUP (Chen et al., 2019;Fresan et al., 2020) while one study showed the opposite (Chee et al., 2010) and it was a non-significant result in nearly half of the studies (n = 13).Two studies from a lower-middle income country found a link between living in a rural area and longer DUP (Nallapanemi et al., 2015;Okasha et al., 2016) while other studies (with the same income level) did not (Ananthi et al., 2017;Kini et al., 2015;Sharifi et al., 2009).
Ethnicity was statistically associated with DUP in two different studies.A shorter DUP was associated with indigenous ethnicity compared with Malay, Chinese and Indian ethnicities (Chee et al., 2010) and black ethnicity compared with non-black ethnicity (Burns & Kirkbride, 2012).
Two interconnected factors were measured heterogeneously across studies with inconsistent results.Patients with a lower educational level had longer DUP (Chee et al., 2010;Chen et al., 2019;Kaminga, Dai, et al., 2019;Kini et al., 2015;Okasha et al., 2016) although this was non-significant in 10 studies.A relationship was found between employment and shorter DUP (Fresan et al., 2020;Okasha et al., 2016;Qiu et al., 2017), while seven found no association and another study showed the inverse relationship (Myaba et al., 2021).
Four studies examined substance use in relation to DUP using semi-structured questionnaires (Nallapanemi et (Davis et al., 2016;Paruk et al., 2015).None found an association with drugs such as alcohol, cannabis or nicotine and DUP.However, one found a relationship between lifetime use of amphetamines or Mandrax and longer DUP (Davis et al., 2016).
Numerous investigations examined the potential influence of clinical factors on DUP.For instance, middle eastern and Asian studies demonstrated an association between an insidious mode of onset of psychosis and longer DUP (Kini et al., 2015;Mishra et al., 2021;Okasha et al., 2016;Qiu et al., 2017;Sharifi et al., 2009).Higher age at illness onset was associated with shorter DUP in numerous studies (Burns & Kirkbride, 2012;Chen et al., 2019;Kaminga, Myaba, et al., 2019;Nallapanemi et al., 2015;Paruk et al., 2015) while another study indicated the opposite (Okasha et al., 2016).Additionally, a study with a national sample size of 5745 found an association with medical comorbidities and longer DUP (Chee et al., 2010) whereas another study with a small sample size of 60 reported the reverse conclusion (Mishra et al., 2021).
Other researchers explored associations between diagnosis type or symptoms and DUP.Seven studies that investigated predominant symptoms in relation to DUP used the Positive and Negative Syndrome Scale, finding mixed results.Four showed that a predominance of negative symptoms was associated with longer DUP (Effiong & Albert, 2016;Fresan et al., 2020;Odinka et al., 2014;Okasha et al., 2016) while another study found the opposite (Kini et al., 2015) and two did not find an association (Paruk et al., 2015;Sharifi et al., 2009).Another study reported that patients having auditory hallucinations compared to other positive symptoms had longer DUP (Sadeghieh Ahari et al., 2013).Additionally, a longer DUP was associated with a diagnosis of schizophrenia in two Malawian studies (Kaminga, Dai, et al., 2019;Myaba et al., 2021) and the diagnosis of delusional disorder in an Egyptian study (Okasha et al., 2016).Studies conducted in India and China found no relationship between DUP and diagnosis types (Devi Thakoor et al., 2016;Nallapanemi et al., 2015).
Several studies supported an association between psychological factors and DUP.A study from Serbia demonstrated that shorter DUP was associated with a greater openness to experience, or the presence of a stressor at illness onset (Maric et al., 2016).Chinese patients with the perception that an intimate relationship might be broken due to mental illness had longer DUP (Devi Thakoor et al., 2016).Other factors predictive of a longer DUP were having public self-consciousness (Myaba et al., 2021) and poorer social adjustment (Fresan et al., 2020).Researchers from Malawi found a link between poor insight level and longer DUP (Kaminga, Myaba, et al., 2019) while this was inconclusive in another study from the same country (Myaba et al., 2021).
Help-seeking behaviour had mixed results.Findings from four studies demonstrated an association between seeking a first contact with non-medical professionals with longer DUP (Adeosun et al., 2013;Effiong & Albert, 2016;Gupta et al., 2021;Tomita et al., 2015) whereas other studies found no association between types of first contact and DUP (Kini et al.,

| Relationship level
Multiple studies revealed a relationship between family-related factors and DUP.For instance, higher family stigma (Devi Thakoor et al., 2016;Mishra et al., 2021), lower family resiliency (Mo'tamedi et al., 2014) and lower family awareness of mental illness (Devi Thakoor et al., 2016) were associated with longer DUP.Nonetheless, a family member's education or occupation did not yield significant results (Qiu et al., 2017).A study from South America reported an association between living with a relative and shorter DUP (Oliveira et al., 2010) while four studies from Asia did not establish that association (Nallapaneni, Lanka and Paritala, 2015;Qiu et al., 2017;Takizawa et al., 2020;Mishra et al., 2021).Furthermore, having no available caregiver (Okasha et al., 2016) and having at least one parent deceased (Kaminga, Myaba, et al., 2019) were statistically associated with longer DUP.
There were mixed results in two Asian studies with markedly different numbers of participants (45 vs. 7252) with regard to marital status.Kini et al. (2015), found a link between married status and shorter DUP, while Chen et al. (2019) showed evidence to the contrary.In contrast, nine studies reported no association (Ananthi et al., 2017;Kaminga, Dai, et al., 2019;Mishra et al., 2021;Myaba et al., 2021;Nallapanemi et al., 2015;Naqvi et al., 2009;Oliveira et al., 2010;Sharifi et al., 2009;Takizawa et al., 2020).Furthermore, one study found no association between having a stable partner and DUP (Qiu et al., 2017) while another study demonstrated that widowed status was associated with a longer DUP (Chen et al., 2019).

| Community level
A study from South Africa with participants with a mean age of 25.8 years old established an association between an earlier meeting with a traditional healer and longer DUP (Burns et al., 2010) although another study from the same country with participants with a mean age of 15.9 years old did not (Paruk et al., 2015).Multiple studies investigated the type of person leading to contact in relation to DUP.Kaminga, Myaba, et al. (2019) reported a link between a longer DUP and getting a referral from a community-based volunteer or a traditional healer.Alternatively, police involvement in the pathway to care yielded divergent results.One study reported that when police officers played a role in the pathway to care, patients had shorter DUP (Burns & Kirkbride, 2012) although this was not statistically significant in other studies (Chee et al., 2010;Nallapanemi et al., 2015).The type of care setting at first contact was not significant in two studies (Chee et al., 2010;Takizawa et al., 2020).Alternatively, Burns and Kirkbride (2012) demonstrated an association between a greater level of community participation and neighbourhood connectedness, and a longer DUP.Another potential predictor of a longer DUP is higher level of social quality of life (Myaba et al., 2021).

| Societal level
Two related factors were positively associated with a longer DUP, namely lower social attainment (Mishra et al., 2021) and low social class (Okasha et al., 2016).Low family income, which was inconsistently measured across studies, was linked to a longer DUP in Devi Thakoor et al. (2016) although no association was found in 25% of the studies (n = 8).Lastly, a greater distance from a house to the hospital was more likely to prolong DUP in Takizawa et al. (2020).

| RESULTS OF QUALITY ASSESSMENT
The overall quality of reporting of the cross-sectional studies was good.Each met more than 50% of the AXIS tool criteria for the quality of reporting, namely five of the seven criteria or more, except one study (Okasha et al., 2016) which met three criteria.Overall, studies adequately described their methods, results, and limitations.Nevertheless, 40% of studies (n = 12) did not define the target population, only the study population, and 20% of studies (n = 6) did not clearly indicate their study objective.The case-control study (Ayres et al., 2007) met three of the five questions of the SIGN methodology because it did not report the participation rate as well as the confidence intervals.
The overall study design quality of the studies was fair.Most of the cross-sectional studies met more than 50% of the criteria in the AXIS tool for the study design quality.Nearly all studies (n = 23) met four of the seven criteria, but the remaining studies did not (Ananthi et al., 2017;Devi Thakoor et al., 2016;Naqvi et al., 2009;Paruk et al., 2015;Sadeghieh Ahari et al., 2013).The study design quality was undermined by several factors.More than 85% of studies (n = 25) did not provide a sample size justification and did not select a sample frame similar to the target population.Additionally, almost half of the studies (n = 12) did not meet the criterion of justified conclusions and discussion because they did not discuss relevant information such as confounders and non-significant results.The casecontrol study (Ayres et al., 2007) did not meet two of the six criteria due to an inability to confidently assess whether an association between the exposure and the outcome was real and whether the study results were directly applicable to the targeted patient group.
The risk of bias results of the cross-sectional studies was poor.All of these met less than 50% of the AXIS tool risk of bias criteria.
Almost 80% of the studies (n = 23) did not mention information about response rate and non-responders and thus did not pass three of the seven criteria from the outset.Furthermore, more than 25% of the studies (n = 8) did not demonstrate a selection process ensuring representativeness of the reference population.More than half of the studies (n = 18) used a combination of validated and non-validated instruments to measure their risk factors, which was not sufficient.
Similarly, most studies (n = 23) did not use a validated measure for DUP, while n = 5 did.The risk of bias management for the casecontrol study (Ayres et al., 2007) was excellent as it met all three criteria for that category in the SIGN methodology checklist.Results of the study quality assessment are provided in Supporting Information, Appendix B.

| Interpretation of the results
The main finding is that various factors may be associated with DUP for people with psychosis in LMICs.Twenty-three factors were associated with a longer DUP, 17 had inconsistent results, and 15 were non-significant.Certain results are consistent with earlier studies.For instance, numerous studies from HICs also support that an insidious mode of onset is related to longer DUP (Broussard et al., 2013;Chen et al., 2019;Compton et al., 2008;Hui et al., 2013;Kalla et al., 2002).
Moreover, results from the case-control study by Ayres et al. (2007) are aligned with a recent systematic review that reported no overall association between DUP and cognition (Allott et al., 2017).Although this review found mixed results regarding gender, 80% of the studies that investigated that factor (n = 13) did not find an association with DUP, consistent with Cascio et al. (2012).Boonstra et al. (2012) found a link between negative symptoms and a longer DUP which is aligned with the results of 60% of the studies that investigated it in this review (n = 7).A systematic review on ethnicity and DUP highlighted that three studies found an association between black ethnicity and shorter DUP (Anderson et al., 2013) which echoes the finding of an included study (Burns & Kirkbride, 2012).
On another note, we found different directions of association and potential determinants of DUP across continents and LMICs.These conflicting results likely reflect the limitations of evidence, but may also reflect the influence of culture on potential factors related to DUP.This hypothesis was highlighted in Devi Thakoor et al. (2016) who compared two LMICs using the same methods and obtained divergent factors for each country.Another study that compared DUP correlates of two HICs also reported this (Kalla et al., 2002) although research on this is scarce.Thus, more investigations comparing potential factors influencing DUP between cultures is needed.These studies could take the form of comparison between LMICS, between HICS and LMICs or between ethnic subgroups.This may help to better understand the reasons for the variations between studies.
Another key finding is that the methodological quality of the studies substantially limits the conclusions and recommendations of this review.Almost all 23 factors (n = 20) associated with DUP were supported by a single study, suggesting that they were insufficiently studied across LMICs.There were exceptions however, namely mode of onset, family stigma and ethnicity were supported by five or fewer studies.In comparison, factors with contradictory results were supported by a significantly higher number of studies.Examples of these factors are educational level (n = 15) and first contact with nonmedical providers (n = 7).Almost half of the studies (n = 13) had small sample sizes, which may have led to type 2 errors (Downes et al., 2016).Many studies did not control for confounding factors which most likely skewed the result of their associations (Downes et al., 2016).Furthermore, our sample was almost entirely comprised of cross-sectional studies, which makes it impossible to determine causal relationships (Setia, 2016).
There was a low level of representativeness and therefore generalizability due to the use of convenience sampling, heterogenous illness characteristics, and concerns of non-response bias (Downes et al., 2016).To add to this, these studies were conducted within the formal healthcare systems of the varying nations, meaning that harder-to-reach patients or those using informal health services may not have been included, likely representing a significant number of people in LMICs (Kumah, 2022).Moreover, the use of non-validated instruments to measure factors and DUP compromised the validity, reliability, and comparability of the results (Downes et al., 2016).The reliance on retrospective data collection and self-reports of DUP introduced a risk of recall and desirability biases (Devi Thakoor et al., 2016).These limitations should inform future actions.Funding, development, and replication of higher-quality studies across LMICs is strongly recommended to provide relevant conclusions, policy recommendations and to help understand inconsistencies between studies.
To achieve this, the use of reliable and validated instruments to measure factors and the determination of a consensus for the definition of DUP is imperative.
A secondary finding of this review is that a SEM is a relevant framework for healthcare professionals, researchers, and decision- Three selected results encompassing different levels of the SEM will be highlighted as they embody important aspects of LMICs and could be potential targets to DUP reduction if more results are compiled over time.The first example is taken from Davis et al. (2016) who demonstrated an association between a longer DUP and the use of amphetamines or Mandrax.Substance use disorders are prevalent public health problems in LMICs (Sarkar et al., 2021) and are projected to increase in coming years (Nadkarni et al., 2022).Given that substance use and psychosis are common co-occurring disorders (Davis et al., 2016), brief psychosocial interventions to target substance use in LMICs (WHO, 2016) may be useful potential strategies for DUP reduction.The second example comes from two studies that found a link between family stigma and longer DUP (Devi Thakoor et al., 2016;Mishra et al., 2021).This is consistent with previous evidence that stigma towards people with mental illnesses may delay mental health treatment (Wainberg et al., 2017).Stigma is an important issue in the global mental health community as it is a significant contributing factor to lack of access to mental health services (Wainberg et al., 2017).The development of promising methods such as contact-based interventions and educational campaigns (Patel et al., 2014) may be interesting future avenues to reduce DUP for people with psychosis in LMICs.The last example comes from Kaminga et al. (2019) who reported a relationship between referral from community-based volunteers or traditional healers and longer DUP.This reflects an important reality and challenge of LMICs which is the presence of medical pluralism (Orr & Bindi, 2017).Many people in LMICs will seek out traditional healers for their mental health needs (Orr & Bindi, 2017).A collaborative model with traditional healers may be an interesting solution to improve access (Gureje et al., 2020) for people with psychosis in LMICs.
At present, too little robust data on these and other potential factors influencing DUP are available, as this review indicates.As the understanding of the factors and their relationship to each other becomes clearer, it may be possible to develop and implement targeted interventions.It is unlikely that a single intervention would be effective in reducing DUP in LMICs (Malla, 2022).The SEM would allow for a synergistic and multisectoral approach to this endeavour.
Future concerted and sustained efforts to reduce DUP at all levels of SEM may in turn improve EIS, clinical outcomes, and overall wellbeing of people with psychosis in LMICs.

| Strengths and limitations of the review
This systematic review has several limitations.The sample overrepresents certain regions of the world while underrepresenting lowincome countries.The investigation of more specific world regions or income levels might have changed the findings.In the same vein, conclusions about potential determinants of DUP might have been more accurate had this review investigated a more specific study population (e.g.non-affective psychoses only) (Large et al., 2008).Moreover, this review included studies of lower methodological quality which may have reduced its own quality (Boland et al., 2017, p. 122).Due to resource and time constraints, only studies available in English and published in peer-reviewed journals were included, which exposed this review to language and publication bias (Boland et al., 2017, p. 53).Additionally, we may have unintentionally omitted relevant studies that could have altered the results.The reliability of the research process might have been affected by this review being conducted by a single individual although guidance was received at every step (Boland et al., 2017, p. 18).This review also has its share of strengths.An explicit methodology was followed, and a rigorous quality appraisal was performed (Owens, 2021).Additionally, the search of six different databases established an appropriate degree of comprehensiveness which is indicative of current evidence on this subject (Boland et al., 2017).

| CONCLUSION
In conclusion, this review found many potential determinants of DUP for people with psychosis in LMICs.These factors range from mode of Twenty-nine cross-sectional studies and one case-control study were included.Each study investigated one or more factors potentially influencing DUP.At least 60% of the studies (n = 19) took place in psychiatric hospitals or psychiatric units of general hospitals.Nearly half of the studies (n = 14) came from the Asian continent followed closely by the African continent.One study was conducted across two countries, China and Mauritius (Devi Thakoor et al., 2016).Please see Figure 3 (MapChart, 2022).

F
I G U R E 1 Socio-Ecological Model adapted by the Centers for Disease Control and Prevention.Source: Centers for Disease Control and Prevention, 2022.The Social-Ecological Model: A Framework for Prevention.Centres for Disease Control and Prevention.http://www.cdc.gov/violenceprevention/about/social-ecologicalmodel.htmlPRISMA 2020 flow diagram for new systemaƟc reviews which included searches of databases, registers and other sources From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al.The PRISMA 2020 statement: an updated guideline for reporting systematic reviews.BMJ 2021;372:n71.doi: 10.1136/bmj.n71.For more information, visit: http://www.prisma-statement.org/Records identified from: Medline (Ovid) (n = 26) EMBASE (Ovid) (n = 56U R E 2 PRISMA Flow Diagram.The end of DUP was defined by phrases of varying specificity.The most common definitions were 'until the first contact with mental health services' (n = 7) and 'until the start of antipsychotic medication' (n = 6).Another frequent definition was 'until the initiation of adequate treatment' (n = 4).The factors studied also varied with 55 distinct factors investigated.Several studies investigated sociodemographic factors (n = 29), clinical factors (n = 22) and pathway to care (n = 14).Other investigated factors include stigma (n = 3) and socioeconomic factors (n = 3).Examples of validated instruments used to assess these factors were the Positive and Negative Syndrome Scale (n = 14), Premorbid Adjustment Scale (n = 4), Calgary Depression Scale (n = 2) and the Global Assessment of Functioning (n = 2).Examples of instruments chosen to determine DUP were the Pathway Encounter Form (n = 5), Nottingham Onset Schedule (n = 2), Symptom Onset of Schizophrenia (n = 2) and Beiser Scale (n = 2).
149, 95% CI (1.411-3.274),p value <0.001 r = À0.074,p value = 0.278 r = 0.041, p value = 0.546 r = 0.004, p value = 0.959 r = 0.179, p value = 0.019 r = À0.075,p value = 0.271 r = 0.047, p value = 0.495 r = 0.006, p value = 0.930 r = 0, p value = 0.998 r = À0.012,p value = 0delusion and harm to self and others: p value = 0.03 p = 0.60 p = 0.23 p = 0.22 Mean difference of DUP between people with auditory hallucination, visual hallucination, persecutory delusion and harm to self and others: p value = 0.03 p = 0.60 p = 0.23 p = 0.22 makers to conceptualise factors, map findings, and highlight gaps in the literature.As an illustration, studies were represented at all levels of the SEM, although nearly all studies (n = 29) investigated individual-level factors and fewer studies explored other levels.It is worth mentioning that the factors at the societal levels could be debated, but including these four factors here appeared reasonable as they were all closely related to social constructs or healthcare infrastructure.Further research to uncover more potential determinants of DUP at these levels is essential.For instance, future research comparing how macro-level factors may influence DUP could enable the subsequent development and implementation of interventions to target societal factors that are known to perpetuate disparities in access to care for people living in poorer resource settings(Burns & Kirkbride, 2012).
onset and substance use to family stigma and causal attribution of mental illness.However, factors such as gender, marital status, educational level, medical comorbidities, or family history of mental illness led to conflicting results.The findings should be viewed with caution given the methods employed by the studies.This review draws attention to the urgent need for higher quality research on this subject in LMICs.The results of this review are a starting point, and the SEM is a useful framework to consider for healthcare professionals, researchers, and policymakers.It is the ethical duty of the global mental health community to identify factors that may influence DUP to subsequently implement targeted efforts to reduce it, which in turn could facilitate access to EIS and improve the prognosis of those affected by psychosis in LMICs.

Table 2
for more details and characteristics.Most researchers opted for either the Diagnostic and Statistical Manual of Mental Disorders (n = 17) published in 1994, or the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (n = 11) published in 2015.
Study, patient, and illness characteristics.
al., 2015; Takizawa F I G U R E 3 Countries featured in this review.Source: MapChart (2022).World Map: Simple.MapChart.Available at: https://www.mapchart.net/world.htmlTA B L E 2 T A B L E 2 (Continued) T A B L E 2 (Continued) T A B L E 2 (Continued) T A B L E 2 (Continued) IQR, interquartile range; NOS, not otherwise specified; ICD-10, International Classification of Diseases, 10th edition; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th edition; DSM-IV-TR, Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision.Main findings.