Associations between psychosis and visual acuity impairment: A systematic review and meta‐analysis

Abstract Objectives Several theories propose that visual acuity impairment is associated with psychosis. Visual impairment could lead to psychosis or the converse, or they may share underlying pathology or risk factors. In the first evidence synthesis in this area for over 25 years, we collated studies measuring the association between visual acuity impairment and psychosis. Methods We searched the MEDLINE, EMBASE, PsycINFO, and Web of Science databases for studies published from 1992 to 2020, using the Newcastle Ottawa Scale to assess risk of bias. We narratively synthesized findings and meta‐analyzed sufficiently homogenous results. Results We included 40 papers, which reported on 31 studies. Evidence from seven cohort studies was inconsistent, which precluded meta‐analysis of this study design. These contradictory results also made it difficult to draw conclusions regarding a temporal association. We found evidence for an association from eight cross‐sectional studies treating visual acuity impairment as the exposure and psychosis as the outcome [pooled odds ratio (OR) =1.76, 95% confidence interval (CI): 1.34–2.31], and four with the reverse exposure and outcome (OR: 1.85, 95% CI: 1.17–2.92). Seven case–control studies with mixed findings were found, but only two primarily addressed our research question, and these findings were mixed. Conclusions Although evidence supports a cross‐sectional association between visual acuity impairment and psychosis, further research is needed to clarify the temporal direction, given the mixed findings in cohort studies. Understanding the association may give insights into prevention strategies for people at risk of visual acuity impairment and psychosis.


| INTRODUCTION
There is a growing body of research exploring how visual acuity impairment and psychosis interrelate. Psychosis is a broad term encompassing illnesses such as schizophrenia and their associated symptoms, which include delusions, hallucinations, and thought disorder. In this review, we define visual acuity impairment as reduced measured acuity or the subjective experience of not seeing clearly.
An association between psychosis and impaired visual acuity has been demonstrated cross-sectionally and longitudinally, including very large population-based datasets. 1,2 It extends to both psychotic disorders and psychotic symptoms. 1,3 The relationship may be bidirectional. Psychotic disorders might lead to visual acuity impairment through reducing an individual's ability to seek help to maintain good vision, for example by attending optician appointments. 3 Both psychotic illnesses and antipsychotic medications also increase risks of developing diabetes and cardiovascular disease, which can damage retinal vasculature and reduce eyesight. 4,5 Antipsychotic medications may directly cause blurred vision as an anticholinergic side effect, 6 and symptoms such as disorganization could increase tolerance for blur. 7 Nevertheless, studies showing that childhood ocular pathology is associated with increased risk of schizophrenia in adulthood imply that psychotic illnesses leading to reduced visual acuity cannot fully account for the association. 8,9 It is also speculated that aberrant visual input might be a potential cause of psychosis. 10,11 Visual hallucinations have been instigated through blindfolding healthy individuals. 12 This phenomenon is well-recognized in Charles-Bonnet syndrome, where loss of vision leads to visual hallucinations without other psychotic symptoms. 10 The "Protection Against Schizophrenia" model posits that the effects of aberrant visual input on cortical NMDA receptors might cause psychosis, with absent and perfect vision being protective. 10,11 This model is based on an observed absence of any case reports of a congenitally cortically blind person developing schizophrenia, despite shared risk factors for these conditions. 13 The mechanisms by which congenital or early blindness may be protective are not clear, but it is suggested that enhanced perceptual processing, greater working memory capacity, and prevention of abnormal visual input conferred by early blindness may play a role. 13 Indeed, Silverstein et al. have described how many cognitive functions typically impaired in schizophrenia are in fact enhanced in congenitally blind individuals. 13 A further possibility linking visual acuity impairment and psychosis is that common neuropathological processes cause both. 14,15 Retinal thinning has been associated with schizophrenia and neurodegenerative diseases such as Alzheimer's disease and may reflect a change in central nervous system function. 16 Further, the offspring of people with psychotic illnesses, who have a higher risk of developing psychosis themselves, have been found to have altered retinal function. 17,18 Visual acuity impairment and psychotic illnesses could also share risk factors, for example, vitamin D deficiency. 19,20 The most recent literature review regarding sensory impairment and psychosis in older adults was conducted in 1993. 21 It reported that evidence of association between visual acuity impairment and psychosis was inconsistent and highlighted methodological flaws in the existing research. Limitations included lack of appropriate control groups and unreliable measurement of visual acuity impairment. 21

| Aims of the study
We aimed to update and expand on this review, by conducting the first systematic review for over 25 years to explore the association between visual acuity impairment and psychosis across all age-groups. We aimed to determine the strength of evidence for an association, and the direction of this relationship, to inform future research, and specifically work exploring the etiology of psychosis.

| METHODS
This review was registered on PROSPERO (CRD42019129214). 22 We used the OVID interface to search the databases MEDLINE, EMBASE, and PsycINFO on August 18, 2020, limiting studies to human subjects and English language.

Summations
• An association between psychosis and visual acuity impairment is well-evidenced in crosssectional research, but findings from longitudinal studies give inconsistent results. • Future research should focus on establishing the temporality of this association and elucidating the underlying mechanisms.

Limitations
• This review was limited to English language studies. • The meta-analyses from this review are based on observational studies, and results must therefore be interpreted with caution. • We did not identify any studies investigating the longitudinal association between psychosis as exposure and visual impairment as outcome.
The databases Open Grey 23 and Web of Science were also searched on August 19 and September 10, 2020, respectively. We combined search terms encompassing visual acuity impairment with terms related to psychosis. These are described in Supplement Material S1. Inclusion criteria were as follows: 1. Quantitative studies of any design that compared psychotic symptoms or illnesses as the outcome in people with visual acuity impairment relative to people without, or 2. Quantitative studies of any design that compared visual acuity impairment as the outcome in people with psychotic symptoms or illnesses relative to people without.
We included research studies published from January 1, 1992, onwards where: • Visual acuity was defined as either measured visual acuity or self-reported visual clarity • Psychosis was defined as either reporting psychotic symptoms, or diagnosis of psychotic disorder whether selfreported or determined by psychiatric interview or from medical records.
Subjective impairment has previously been reported to be a suitable proxy for objective impairment. 24 Exclusion criteria were as follows: • Studies with fewer than 30 exposed participants, due to limited validity relating to low power to detect an association. 25 • Studies reporting a measure of color blindness or visual processing but without a measure of visual acuity or selfreported visual clarity. • Studies that excluded participants with visual acuity worse than 20/20 on Snellen chart or equivalent. • Studies that only measured visual hallucinations and no other psychotic symptoms. This criterion was added to avoid overestimating any effect due to studies focused on Charles Bonnet syndrome, which differs from psychotic disorders in that the hallucinations are an isolated symptom and full insight is retained.
We exported search results into EndNote 26 and then Covidence 27 to facilitate co-screening and record keeping by separate reviewers. Duplicates were removed automatically using inbuilt duplicate detection software in these packages.
The first author (NS) screened the titles and abstracts of potential studies to determine inclusion, with a 10% random sample of records independently screened by ME. Eligibility of studies was subsequently confirmed by NS, and ME independently checked the full text of >20% of all retrieved articles. Disagreement was resolved through discussion and consensus between NS and ME, and, if necessary, CC or JH.
Risk of bias was assessed using the Newcastle Ottawa Scale (NOS) for cross-sectional, case-control, or cohort studies by NS and ME independently. 28,29 Where assessing the association between visual acuity impairment and psychosis was not the primary aim of the study, the risk-of-bias score related to the quality of the study regarding measurement of this specific association. The authors agreed, regarding the interpretation of the NOS for cross-sectional studies, that we did not require assessment of outcome to be blinded, if it was objective; studies to include a power calculation if the sample size was >1000, nor for studies to have established comparability between respondents and non-respondents if the response rate was >90%. We defined a quality score of 7+ as indicating a low risk of bias, consistent with published systematic reviews. 30,31 We deemed studies with a score of <4 to have a high risk of bias. Disagreements regarding NOS score were resolved through discussion.
We prioritized studies with a low risk of bias in the narrative review. Where multiple relevant results were reported, we reported odds or hazard ratios with the most robust level of adjustment in the Forest plots. [32][33][34] We report distance visual acuity impairment where both near and distance visual acuity impairments were reported on separately, for comparability with other studies and consistency with criteria for certifiable visual impairment. 35 Similarly, we chose schizophrenia when multiple psychotic disorder diagnoses were assessed, for comparability. We reported findings according to study type. Where there were enough studies rated as at low risk of bias, we compared studies of older and younger adults to account for the burden of psychosis in older adults potentially having different etiology, such as neurodegenerative disease. 36 For the cross-sectional studies, we also compared studies that reported only on psychotic symptoms and those that included psychotic diagnoses, and studies that used objective or subjective measures of visual impairment.
We summarized the level of evidence using the Evidence Based Medicine Consult guidelines, where.
• A = consistent evidence from randomized-controlled trials • B = consistent evidence from observational studies • C = extrapolations from observational studies at higher risk of bias • D = troublingly inconsistent evidence from studies at any level. 37,38 We specified in advance that we would conduct metaanalysis if three or more studies with low risk of bias could be combined. 22 We used random-effects meta-analysis to account for differences between study designs. This type of analysis includes a measure of estimated between-study heterogeneity in the weighting, to avoid giving an overly precise | 9 SHOHAM et Al. estimate in the presence of heterogeneity. 39 If a compatible effect estimate was not reported but could be calculated from raw data, we did this. We combined cross-sectional studies that reported an odds ratio. We treated studies that used visual acuity impairment as exposure and psychosis as outcome and the converse separately, since these odds ratios are not theoretically interchangeable when adjusted. We used fully adjusted odds ratios where possible, due to evidence and guidelines, suggesting that this is likely to obtain the leastbiased pooled estimate. [32][33][34] We also separately combined unadjusted odds ratios where these were provided or could be calculated, as a sensitivity analysis to reduce heterogeneity. We reported the I 2 statistic to describe the proportion of variation in results caused by heterogeneity. We classed 25% as low heterogeneity, 60% as moderate, and 75% as high. 40 Data were analyzed in STATA version 16. 41 3 | RESULTS

| Search results
NS screened all 5700 titles and abstracts for inclusion, and ME co-screened 570 (10%) (Figure 1). Cohen's kappa coefficient for inter-rater reliability was >0.8, with agreement for >99% abstracts. 280 full texts were screened, of which ME co-screened 65 (23%), giving Cohen's kappa of 0.63, with 88% agreement. The reasons for exclusion of full texts are shown in the PRISMA diagram ( Figure 1). ME also checked data extraction from four (10%) of studies, with complete agreement. Forty papers that reported on 31 studies were finally included in the review. Emailing four experts in the field did not identify any additional studies.

| Studies classed as at low risk of bias
The 5/7 cohort studies rated as having a low risk of bias 2,8,9,53,54 recruited three distinct populations: young male military conscripts, 2,53 older people, 54 and children. 8,9 Two very large studies explored whether, in young male military conscripts, refractive error predicted future diagnosis of psychotic illness. 2,53 Both measured visual acuity with Snellen charts and used linked hospital records to determine subsequent psychotic illness diagnosis status, but they found opposing results. A Swedish study of >1 million young men 2 found that worse visual acuity increased the risk of psychotic illness, while an Israeli study 53 of >650,000 young men found that it reduced the risk of schizophrenia. We noted some key differences between these studies. The Israeli study focused exclusively on schizophrenia and assessed corrected visual acuity. 53 It did not state length of follow-up or describe the measure of visual acuity impairment in detail, and reported a lower prevalence of myopia than another study using the same data. 68 The Swedish study assessed multiple measurements of the exposure including uncorrected visual acuity and additionally tested non-affective psychotic disorder as an outcome. 2 It included a sensitivity analysis excluding participants who developed psychosis within five years of the exposure measurement to ensure prevalent psychosis was not driving the findings, which were robust to this. 2 A study of older adults used Swedish national registry data from >3 million people aged 60 in 1980 and investigated whether visual acuity impairment predicted diagnosis of very late-onset schizophrenia-like psychosis (VLOSLP) up to 31 years later. 54 Contrary to the authors' hypothesis, visual acuity impairment predicted a significantly lesser likelihood of being diagnosed with VLOSLP. The authors comment that this finding was unexpected and suggest that using registerbased diagnoses may have led to artificial evidence of negative association as people with psychotic illness can be less able to access health care and are therefore less likely to have visual acuity impairment recorded. 54 They may also be more likely to be subject to "diagnostic overshadowing", where physical complaints are wrongly attributed to psychiatric illness. 54 Further, survivor bias is possible, as participants who received a diagnosis of psychosis earlier in life were not included. 54 Two smaller studies (n = 242 and n = 110) examined children including offspring of parents with psychotic illness and matched comparators. 8,9 Both found that ophthalmic problems in childhood were associated with a future diagnosis of schizophrenia spectrum disorder. These studies measured visual acuity at ages 4 9 and 11-13 8 and followed children up for 18 and 20 years, respectively.
There were only two cohort studies in similar populations that reported results compatible with combination in metaanalysis, 2,53 and as these results were contradictory, it was not appropriate to combine them.

| Moderate and higher risk-of-bias studies
The 2/7 cohort studies classed as having moderate risk of bias reported data on older adults and found that visual acuity impairment was associated with subsequent psychotic symptoms. 44,48

| Case-control studies
There were seven case-control studies, reporting data on 723 people aged over 18 21,55,[63][64][65][66]69 (Table 2). These all compared people with a diagnosed psychotic illness to controls without psychosis. Two investigated differences in visual acuity impairment as a primary aim, 21,62 with the remainder assessing acuity purely to establish comparability between groups in visual processing or retinal thickness experiments. 55,[63][64][65][66]70 All case-control studies measured objective visual acuity impairment using Snellen, LogMAR, Freiburg, or similar charts. The Freiburg test is a computerized visual acuity test. 71 Five studies allowed some degree of correction of impairment using aids during the measurement, 21,55,63,64,66,72,73 and two did not state whether correction was employed. 65,69 The number of participants in each study ranged from 60 to 130.
Of note, four studies excluded participants with myopia above a certain level, 55,63-65 one broadly matched groups by visual acuity (but still tested for a difference), 66 and one selected controls from an eye clinic. 62 This likely limited the ability of case-control studies to detect an association.

| Studies classed as at low risk of bias
Only one study was rated as having low risk of bias. 55 This study reported on a group of 30 adults with schizophrenia attending a secondary care center and 30 matched controls. 55 It excluded people with myopia requiring lenses greater than 2.0 diopters (classed as mild myopia 74 ). There was a lower mean visual acuity score in the schizophrenia group compared with the control group, but this was not strongly supported by statistical testing (p = 0.068).

| Studies classed as at moderate or higher risk of bias
Three of the six case-control studies at moderate risk of bias found evidence of lower visual acuity in the groups with psychotic illness for at least some types of vision, 21,65,72 while the remaining three found no difference between the groups. 62,63,66 One study found that the lower visual acuity applied to people with established schizophrenia, but not people with first-episode psychosis. 64

| Objective measures of visual impairment
Four of the seven studies including younger adults used objective measures of visual acuity impairment, either LogMAR test score, Kay's pictures, or a diagnosis of blindness or low vision in clinical records. 3,58-60 They include the only study of younger adults, which did not find evidence of association at the p < 0.05 level. 58 This took place in a psychiatric facility and was relatively smaller than other studies (n = 356). The point estimate for the odds ratio for visual acuity impairment in schizophrenia relative to other diagnoses was still suggestive of an association. 58

| Subjective measures of visual impairment
Three studies including younger adults reported the association between self-reported sight difficulty and psychosis. 1,56,57 All found evidence of association, including one international study with over 2 million participants. 1 Adjusted odds ratio point estimates ranged from 1.64 to 2.16.

| Objective measures of visual impairment
Only 1/8 studies of older adults used an objective measure of visual acuity impairment. 52 This measured (either) T A B L E 1 Cohort studies reporting on psychosis in visual impairment.

Study and country Population
Sample size (exposed participants) Exposure

| Studies at moderate or higher risk of bias
Three out of four studies at moderate risk of bias found evidence of association between visual acuity impairment and psychosis: two in older adults 42,43 and one in adults. 67 The other study found an association between visual acuity impairment and paranoid ideation, but not delusions or hallucinations. 45

| Meta-analysis
We combined results for the twelve cross-sectional studies with low risk of bias that reported an odds ratio or allowed one to be calculated, dividing these according to whether they treated visual acuity impairment (n = 8) or psychosis (n = 4) as the exposure ( Figure 3). 1,3,44,46,48,50,51,[56][57][58][59][60] We included two separate samples reported in one study. 48 We used fully adjusted odds ratios where possible. 32 (Figure 3). There were too few studies to test subgroups when the exposure was psychosis. There was no strong evidence of publication bias from a Funnel plot or Egger's test (p = 0.386 where visual acuity impairment was the exposure, and p = 0.593 where psychosis was the exposure) ( Figure S1).

| Summary of evidence
We found grade D (troublingly inconsistent) evidence for an association between objectively measured visual acuity impairment as an exposure and schizophrenia as an outcome in longitudinal studies. We found no longitudinal studies that F I G U R E 2 Cohort Studies Reporting on risk of Psychotic Illness in Visual Impairment. AHR, adjusted hazard ratio; 95% CI, 95% confidence interval. investigated the converse relationship (psychosis as an exposure and visual acuity impairment as an outcome).
We also found grade D (troublingly inconsistent) evidence for an association between visual acuity impairment and psychosis from case-control studies.
We found grade B (consistent evidence from observational studies) evidence, however, for a cross-sectional association between visual acuity impairment, whether measured objectively or subjectively, and psychosis. This applied to both younger and older adults, and to psychotic diagnosis and symptoms.
There is some suggestion from two studies that the association may be larger for schizophrenia than for other psychotic illness diagnosis. 2,64 Arguably, this could support the hypothesis that visual impairment is a consequence rather than a risk factor for psychosis, because schizophrenia is typically associated with poorer functioning than other diagnoses and therefore more likely to impair eye care. 75 Alternatively, it could support a hypothesis that visual impairment is specifically a risk factor for schizophrenia. 13 The positive associations between visual impairment and symptoms were greater for hallucinations than delusions in three studies, which separated these out, 42,45,51 raising the possibility that visual hallucinations partially drove the associations seen. We were, however, unable to separate these from other types of hallucination.
For a longitudinal relationship between visual acuity impairment as exposure and psychosis as outcome, evidence was conflicting in adult populations. 2 Both (small, cohort) studies of children found evidence of association 8,9 ; suggesting that ophthalmic problems during a critical developmental phase may be implicated on the causal pathway to a future diagnosis of psychosis.
The discrepancy in findings between cross-sectional and longitudinal studies warrants further attention. It could suggest that psychosis leads to visual impairment, rather than the converse. No studies have yet tested the longitudinal association between psychosis as exposure and visual acuity impairment as outcome. It might also be that psychosis and visual impairment co-occur and are not causally associated.
Perhaps visual acuity impairment reflects brain aging, 76 and its impairment reflects faulty neural processing as much as faulty refraction. Nevertheless, given that the most robustly conducted longitudinal study found a temporal association, which was strongest for corrected acuity, 2 the hypothesis that visual acuity impairment could be an etiological factor contributing to the development of psychosis remains plausible. The pooled cross-sectional odds ratios are larger than those for some established risk factors for schizophrenia, for example, obstetric complications and birth seasonality 77,78 ; but smaller than or similar to those for others, such as having an affected parent or childhood trauma. 78,79 Regardless of whether a causal relationship exists between visual impairment and psychosis, the evidence to date suggests that clinicians caring for people with psychotic illnesses should be alert to the increased chance that their patients will have impaired visual acuity. Facilitation of optical testing could improve eye care for this group. 3 Wider uptake might also mean that complications of comorbidities associated with psychotic illnesses such as diabetes are detected earlier, preventing sight loss. 4,5 Similarly, clinicians caring for people with visual impairment should be aware of the potential for mental illness, so that patients can be signposted to appropriate support when needed.

| Strengths and limitations
Our systematic review is the first to bring together studies of visual acuity impairment and psychosis across the lifespan. We searched multiple databases and incorporated a wide variety of designs. We included studies where the visual acuity impairment-psychosis relationship was not the primary focus of the study.
Limitations exist in the included studies. Clearly, randomized-controlled trials are not possible in this area, so all studies were observational. The strength of our findings is inevitably dependent on the methodology of the included studies. While we found several large, cohort studies at low risk of bias, their findings were conflicting. This may be due to measurement differences, differences in severity of visual acuity impairment, or different lengths of follow-ups. Limitations make it difficult to draw conclusions from the case-control studies. Consequently, we concluded that the evidence from study designs that might allow interpretation of the direction of association between visual acuity impairment and psychosis was inconsistent and that it is not possible to surmise from this review whether a potential causal relationship exists.
There was statistical evidence of high heterogeneity between studies, except for four studies investigating a crosssectional association between visual acuity impairment as exposure and psychosis as outcome in adults of any age. [56][57][58]60 The high heterogeneity is to some extent expected T A B L E 2 Case-control studies reporting odds of visual impairment in psychosis. Patients with a DSM-IV diagnosis of schizophrenia recruited from outpatient and long term assisted living settings Diagnosis was confirmed by their treating clinician, chart review, and SCID. People with best corrected visual acuity <0.8 decimal were excluded.

Study and country Cases Controls
Control status was determined using the psychotic screening SCID. Potential control subjects were excluded if any of their first-degree relatives had a history of psychotic illness. Patients aged 18-60 and diagnosed with schizophrenia or firstepisode psychosis, referred to the study by mental health inpatient staff. All patients were receiving antipsychotic medication. Diagnoses were confirmed by SCID. Groups were matched on visual acuity, but between-group acuity was still tested due to small differences.
Healthy controls without diagnosable lifetime psychiatric conditions (confirmed using SCID); no use of psychotropic medication over the preceding 6 months, and no first-degree relatives with psychotic illness.

Risk-of-bias rating (for outcome of interest)
60 ( The groups did not differ in acuity using ANOVA. Medium

T A B L E 3
Cross-sectional studies reporting on association between psychosis and visual impairment in people of any age.

Study and country Sample
Year of data collection Sample size (number of exposed participants) Exposure Healthcare records: diagnosis of schizophrenia (F20, ICD10) made by a mental health specialist in a public mental health resource.

Outcome Factors adjusted for
Results (for exposed relative to unexposed participants)

T A B L E 4
Cross-sectional studies reporting on association between visual impairment and psychosis in older adults.

Study and country Sample
Year of data collection Sample size (number of exposed participants) Exposure A key limitation of the cross-sectional studies was that most used subjective measures of visual acuity such as asking participants whether they had eyesight difficulties, and the majority also relied on self-reported psychotic symptoms. No cross-sectional studies report adjusting for antipsychotic medication use, which may affect associations, in primary analyses.
A limitation of the two studies of children is that visual acuity was combined with other measures to make an overall visual examination score, and not reported on independently, so we cannot draw conclusions as to whether it was impaired visual acuity that drove the association with subsequent psychotic illness. greatest risk factor; the association may therefore have been obscured by studies that did not employ optimal correction. 2 There were also limitations of the review methodology. It is possible that studies reporting no association may have been missed due to not being indexed in databases, although the funnel plot and Egger's test did not show evidence of publication bias. We did not collate studies that used objective measures of visual processing, such as visual evoked potentials or electroretinograms, although this has been done previously. 16 We were unable to search for studies that were not written in or translated into English. Some search terms, for example, 'blind', were not included due to having alternative meanings (such as allocation concealment) that would bring up a very large number of irrelevant studies. Nevertheless, handsearching reference lists and contacting experts aimed to ensure that any relevant studies not captured in the original search

Outcome
Factors adjusted for Results (for exposed relative to unexposed participants)

Risk-ofbias rating
Proxy-reported hallucinations ascertained by asking "Does he or she ever see or hear things that are not there?" Medium were still detected. We were unable to co-screen all abstracts and studies, but the high concordance between reviewers in the subset co-screened suggests that doing so would have had minimal impact on the results of the review. We attempted to contact authors for additional details regarding some studies and frequently did not receive replies. There is some evidence that inter-rater reliability on the Newcastle-Ottawa Scale (NOS) is lower than ideal, but we aimed to mitigate this by discussing the NOS for all studies. 80 We decided not to include studies that excluded participants with vision above the 20/20 level, but in doing so could have missed a small number of studies that measured group differences within the "normal" range. 81 Finally, meta-analysis of observational data should be interpreted with caution, given the tendency for variation in study designs and unmeasured confounding. 82

| CONCLUSIONS
Overall, current evidence supports the existence of a crosssectional association between visual acuity impairment and psychosis across age-groups. Since the largest and highest quality cohort studies have found contradictory evidence regarding the longitudinal association between visual acuity impairment and subsequent psychotic illness; however, the issue is far from resolved. The possibility of psychosis leading to subsequent visual acuity impairment has not been explored longitudinally. Future research should focus on exploring potential bidirectional longitudinal associations using objective measures of visual acuity impairment and psychosis, and in different degrees of visual acuity impairment.

ACKNOWLEDGMENTS
This work was funded by a National Institute of Health Research (NIHR) Fellowship NIHR300703 (Dr Shoham). It was also supported by Grant 211085/Z/18/Z from the Wellcome Trust (Dr Hayes) and the University College London Hospitals National Institute for Health Research Biomedical Research Centre. We would like to thank the four people with lived experience of visual impairment, psychosis, or caring for someone with these conditions, who advised us on this review. We would also like to thank the two reviewers, whose contribution greatly improved this paper. Any errors that remain are ours.

DECLARATION OF INTEREST
Natalie Shoham plans to continue to research the association between visual impairment and psychosis for her NIHRfunded PhD.

PEER REVIEW
The peer review history for this article is available at https:// publo ns.com/publo n/10.1111/acps.13330.

DATA AVAILABILITY STATEMENT
No new data generated.