Relationship Between Visual Acuity, Colour Vision, Contrast Sensitivity and Stereopsis, and Road Traffic Accidents: A Systematic Review and Meta-Analysis

Abstract

The aim of this study was to evaluate the relationship between visual functions and road traffic accidents (RTAs) by meta-analysis of observational studies. The analysis included all drivers of motor vehicles, regardless of age, and those using private or public transport. Self-reported visual outcomes were excluded. The risk of RTA in patients with reduced visual acuity was observed in commercial drivers in cross-sectional studies (PR 1.54, 95% CI 1.26–1.88), but not in private drivers in cohort (RR 1.04, 95% CI 0.74–1.46) or case–control studies (OR 1.04, 95% CI 0.78–1.40). A non-statistically significant association between colour vision defects and RTA was observed in cross-sectional studies (PR 1.50, 95% CI 0.91–2.45). No evidence was found for an increased risk of accidents in people with reduced stereopsis. In older adults with abnormal contrast sensitivity, a weak risk of RTA was observed in cohort studies. Evidence from low-quality cross-sectional studies suggests an increased risk of RTAs among commercial drivers with reduced visual acuity. The few case–control and cohort studies identified did not show an association between accident occurrence and visual function. Attention needs to be paid to this issue to facilitate the conduct of high-quality research that can support the development of road safety policies.

1. Introduction

The World Health Organization (WHO) reports that the annual average number of deaths from road traffic accidents (RTAs) was 1.19 million in 2021. Furthermore, RTAs were the primary cause of mortality among children and young adults aged 5–29 years, with 92% of fatalities occurring in low- and middle-income countries. This underscores the critical need to prioritise road safety as a public health concern [1].

In 2020, the United Nations General Assembly adopted resolution 74/299, thereby proclaiming the Decade of Action for Road Safety 2021–2030. This initiative was established with the objective of reducing road traffic deaths and injuries, which are recognised as posing a significant threat to the sustainable development of countries worldwide. Furthermore, it is acknowledged that road transport is a complex system, necessitating the harmonious interaction between people, vehicles, and road infrastructure to ensure a high level of safety [2].

Driving is a complex task involving motor, visual, auditory, and executive functions that are essential for human information processing and decision making. It has been estimated that 90% of this information comes from vision and is directly related to reaction time and the response to the perception of the different traffic situations faced by the driver [3]. There are significant differences in visual standards for driving between countries, including the type of tests required to obtain a driving licence and the frequency of re-assessment. Visual acuity is the most commonly used measure, but when used alone it can be a poor predictor and miss the impact of other factors [4].

In order to improve road safety, it is necessary to be aware of all the preventable risk factors for RTAs, which differ according to the age of the drivers [5]. The relationship between risk factors and vision has been studied in older adults with eye diseases such as glaucoma and cataracts, but other types of visual changes such as stereopsis, contrast sensitivity and colour vision abnormalities can affect healthy young people.

The visual system is made up of the refractive, sensory, anatomophysiological and oculomotor subsystems that interact to respond to stimuli while driving. Its functioning has therefore been the subject of worldwide research, to understand the relationship between vision and driving, and is the basis of the regulatory standards for obtaining a driving licence in all countries. A limitation of the existing literature is the use of small sample sizes in many of the studies, which may have an impact on the reliability of the findings. This makes it challenging to ascertain the likely magnitude of the effect of visual impairment on accidents, which is crucial for decision making by road safety regulators.

Conversely, it is widely recognised that, in addition to age, a number of chronic medical conditions can contribute to the deterioration of vision and, consequently, to a reduction in the ability to react effectively while driving [6]. One narrative systematic review has analysed the impact of some visual disorders and impairments on the risk of motor vehicle crashes (MVCs) [7]. Other systematic reviews with meta-analyses have shown positive associations between visual impairment and the risk of being involved in an MVC in low- and middle-income countries, with a 46% higher risk of crashing among those with poor visual acuity and a 36% higher risk among those with impaired colour vision [8]. With limited evidence based on a small number of studies, a direct association between impaired contrast sensitivity (40% higher risk), visual acuity (21% higher risk) and MVCs has also been suggested [9].

Although there appears to be sufficient evidence, the present review focuses mainly on studies conducted in healthy individuals, i.e., subjects who have not been selected for having a systemic or ocular disease, regardless of age. Paradoxically, two-thirds of traffic deaths occur in people of working age (18–59 years) [1], the age group with the highest concentration of healthy people and the largest number of people on the road every day.

This manuscript is structured as follows: Section 1 introduces the problem of road traffic accidents and the importance of studying aspects of visual function; Section 2 explicitly describes the methodology used to develop the study; Section 3 presents the qualitative and quantitative results obtained from the information collected; and Section 4 discusses the findings and the limitations of the review.

Therefore, the aim of this systematic review was to evaluate the relationship between visual functions and road traffic accidents (RTAs) through a meta-analysis of observational studies.

2. Materials and Methods

2.1. Systematic Literature Search

The electronic search was conducted in January 2022 in the databases Medline (PubMed), Embase, Cochrane Central Register of Controlled Trials (CENTRAL) (London, UK) and Cochrane Database of Systematic Reviews (Ovid) (Philadelphia, PA, USA), LILACS, APA PsycNET, CINAHL, and TRID (the TRIS and ITRD database) and in grey literature databases (Opengrey (Nancy, France), EThOS (London, UK), and TDX (Barcelona, Spain)).

The strategy used in PubMed ((Drivers OR Automobile Driving OR Motorcycles OR Motor Vehicles) AND (Vision Disorders OR visual function OR Vision, Ocular OR Night Vision OR Refractive Errors OR Visual Acuity OR Color Vision OR stereopsis OR Depth Perception OR Contrast Sensitivity OR visual abilities) AND (Accidents, Traffic OR motor vehicle collision OR motor vehicle crash)) was adapted to the other indices according to the respective thesauri. All results were exported to the collaborative web application Rayyan [10] for duplicate detection and removal and title and abstract reading.

In addition, the bibliographic references of all the selected articles were checked.

2.2. Criteria for Study Selection

Observational studies that had evaluated the association between a visual function (visual acuity, stereopsis, colour vision, or contrast sensitivity) and the occurrence of RTAs in drivers of all ages, in private or public transport, and in any type of motor vehicle were included. Visual functions could have been assessed with any type of test or be expressed on any measurement scale, but those studies based on self-reported outcomes were excluded from this meta-analysis.

The outcome of interest was RTAs reported through police records or self-report, regardless of whether the driver was at fault or whether the event resulted in injury.

Publications in English or Spanish were included. No date restrictions were imposed.

The initial selection was made by independent reading of titles and abstracts (DG and FR) and disagreements were resolved by consensus.

2.3. Data Extraction and Quality Assessment of Studies

Information was extracted independently by two of the authors (DG and FR) using Microsoft Excel with the following data: author, title, year, country, study design, number of participants, age, sex, time of follow-up (cohorts), type of transport and vehicle, definition and outcome of each visual function assessed, definition and frequency of RTAs, measure of association (odds ratio [OR], risk ratio [RR], prevalence ratio [PR] and hazard ratio [HR]) or data for its calculation, and adjustment variables used.

For visual acuity, colour vision, and contrast sensitivity outcomes, preference was given to data from both eyes or, alternatively, from the eye with better vision.

If the information required by the study protocol was not described in the report analysed, previous reports of the same study were consulted.

Publications with data from the same study were excluded to ensure that only a single report was included.

Study quality was assessed using the National Heart, Lung, and Blood Institute tool, which assigns a quality rating of good, fair, or poor. The tool includes 14 criteria for observational cohort and cross-sectional studies and 12 criteria for case–control studies [11]. For cohort and cross-sectional studies, a score of 12–14 was considered good quality, 9–11 as fair quality, and ≤8 as poor quality. For case–control studies, a score of 11–12 was considered good quality, 8–10 as fair quality and ≤7 as poor quality. Each study was assessed by two authors (DG and FR), and conflicting assessments were adjudicated by a third author (EI).

2.4. Statistical Analysis

Estimates of association (OR, RR, PR, and HR) and their 95% confidence intervals (CIs) were combined using random-effect models (DerSimonian–Laird method). Statistical heterogeneity was assessed using the I² test, which was considered high when >75% [12]. Publication bias was estimated using Egger’s regression method for subgroups with at least 10 studies.

The meta-analysis was performed separately for each visual function and by type of driver (private or commercial), as commercial drivers have much more driving exposure than non-commercial drivers. The analysis of studies with motorcyclists was also performed separately.

All statistical analyses were performed using Stata, v. 12 (StataCorp, College Station, TX, USA).

3. Results

3.1. Results of the Search

The initial search yielded 2398 results, and after excluding duplicates and checking compliance with the selection criteria, 39 articles were selected (Figure 1). Of the studies included in the qualitative synthesis, 35 were combined into different meta-analyses based on the visual function analysed, the type of observational study, and the measure of association reported. Five of the selected studies were not finally included in the quantitative synthesis, either because they did not report data for exposed and injured subjects in their results [13,14,15], or because they did not report data for unexposed subjects [16], or because the authors reported visual acuity results as mean and standard deviation rather than as categorical data comparable to the other studies [17].

Figure 1. PRISMA flow diagram of the systematic literature review search process.

3.2. Characteristics of the Included Studies

Nine cohort studies published between 1998 and 2018 were included, all of which were conducted in high-income countries (six in the USA) (Table 1). In addition, four case–control studies published between 1987 and 1998 were included, three of which were conducted in high-income countries (two in the USA) (Table 2). Of the 26 cross-sectional studies published between 1985 and 2022, 23 were from low- and middle-income countries (including 13 from Nigeria and 3 from India) (Table 3).

Table 1. Characteristics of the cohort studies included in the qualitative review.

Author, Country and Year	Setting, Sample, Age and Sex	Study Period/Time of Follow-Up	Visual Acuity *	Colour Vision *	Contrast Sensitivity *	Stereopsis *	Accidents **	Association with RTA	Adjustment Variables	Quality Rating (NIH)
Owsley et al., USA 1998 [18]	Community-based sample of older drivers n = 294 55 to 87 years old Male 53.7%	3 years of follow-up 1990–1993	ETDRS Binocular ≥20/40 87.4% <20/40 12.6%	-	Pelli–Robson Binocular log > 1.5 83% log ≤ 1.5 17%	TNO stereo test <500″ 68.7% ≥500″ 31.3%	State-recorded. At least 1 crash in the 3-year follow-up period 19.1% The estimated annual crash rate was 74 per 1000 person-years of driving and 7.1 per million person-miles of travel	VA RR 1.45 (95% CI 0.58–3.64) CS RR 0.87 (95% CI 0.35–2.17) StereoacuityRR 0.76 (95% CI 0.36–2.74)	None	Good
Sims et al., USA 2000 [19]	Community-based sample of older drivers n = 174 55 to 87 years old Male 52.6%	1991–1996	ETDRS Binocular <20/40 6.9%	-	Pelli–Robson Binocular log < 1.8 9.1%	-	State-recorded. A single crash, 80.3%. The remaining drivers having up to four events. Incidence of 10 crashes per million person miles travelled	VA RR 1.07 (95% CI 0.26–4.47) CS < 1.8 RR 1.49 (95% CI 0.34–6.57)	None	Good
Keeffe et al., Australia 2002 [20]	Population based study: The Melbourne Visual Impairment Project (VIP) n = 1771 40–101 years old Female 55%	Between 4 and 7 years (mean 4.5 (SD 0.64), median 4 years)	logMAR ≥6/12 97.4% <6/12 2.6%	-	-	-	Self-reported accident record in the intervening 5 years since the baseline study 19.1%	VI RR 0.91 (95% CI 0.48–1.71)	None	Fair
Margolis et al., USA 2002 [21]	The Study of Osteoporotic Fractures n = 1416 65–84 years old Female 100%	Baseline exam (1986–1988) Follow-up 5.7 years avg.	Bailey and Lovie Impaired acuity VA ≤ 20/40 4.7% LogMAR mean 0.08 (SD 0.12)	-	Vistech Vision Contrast Test System 6500 Low spatial frequencies Mean 73.7 (SD 33.3) High spatial frequencies Mean 33.5 (SD 27.0)	Howard Dolman Apparatus Mean 2.02 (SD 2.44)	Police-reported. Rate of crashes 0.07 per driver per year	VA 20/40 HR 1.14 (95% CI 0.73–1.80) LogMAR HR 1.08 (95% CI 0.97–1.19) CS Low spatial frequencies HR 0.99 (95% CI 0.89–1.10) High spatial frequencies HR 0.94 (95% CI 0.85–1.04) Distance depth perception HR 1.01 (95% CI 0.92–1.11)	Age and miles driven per week	Good
Rubin et al., USA 2007 [22]	Salisbury Eye Evaluation (SEE) Study (older participants who reported driving more than 500 miles during the prior year) n = 1801 65–84 years old Female 50.3%	1993–1995 14 to 1567 days (median, 1087; interquartile range = 933–1323)	ETDRS Binocular Better than 20/25 (<0.1) 81.9% 20/25–20/40 (0.1–0.3) 14.8% Worse than 20/40 (>0.3) 3.2%	-	Pelli–Robson Better eye 1.65 (>36) 50.9% 1.35–1.65 (30–36) 46.0% 1.35 (<30) 3.1%	Randot Circles test 13% failed at 457 arc sec (stereodeficient) Passed 82.1% Failed 17.9%	State records 6.7%	VA HR 1.17 (95% CI 0.78–1.70) CS <1.7 HR 0.75 (95% CI 0.49–1.21) ≥1.7 HR 1.25 (95% CI 0.44–5.65) Stereopsis worse 457″ HR 1.44 (95% CI 0.88–2.26)	Miles driven, age, race, sex, Mini-Mental State Examination score, education, comorbidities, living alone and depression	Good
Emerson et al., USA 2012 [23]	General community older drivers without neurological diseases n = 100 65 to 89 years old Male 51%	Mean total follow-up time 6.40 years (range: 0.46–9.03)	ETDRS Mean −0.05 (SD 0.11) Median −0.06 Range −0.26–0.24	-	Pelli–Robson Both eyes Mean 1.80 (SD 0.15) Median 1.80 Range 1.35–1.95	-	Police crash reports in which the participant was the driver, categorised as “at-fault”, “not at-fault” or “not enough Information”. Crash rate at 1 year Cumulative incidence 7.1% (95% CI 3.4–14.3); at 2 years, 16.4% (95% CI 10.4–25.3]; at 4 years 28.6% (95% CI 20.5–38.9), and at 6 years 32.4% (95% CI 23.8–43.0)	VA HR 0.95 (95% CI 0.64–1.40) CS HR 1.19 (95% CI 0.82–1.71)	Age, gender, education, and baseline mileage driven per week	Fair
Green et al., USA 2013 [24]	Population-based sample n = 1998 70 to 99 years old Male 56.5%	MVCs report for the 5 years before study enrollment	Electronic Visual Acuity test Binocular <20/40 8.1%	-	Pelli–Robson Binocular log < 1.5 6.7%	-	Police-reported. 34% At fault 15.4% 5.6 per 1,000,000 person-miles for persons without impairment 6.4 per 1,000,000 person-miles for persons with VA impairment 8.3 per 1,000,000 person-miles for persons with CS impairment	VA RR 1.04 (95% CI 0.74–1.48) At fault RR 1.08 (95% CI 0.66–1.76) CS RR 1.42 (95% CI 1.00–2.02) At fault RR 1.52 (95% CI 0.93–2.49)	Age, race, sex, number of medical conditions, Mini-Mental State Examination, Motor-Free Visual Perception Test, and Trail-Making Test Part B.	Good
Woolnouph et al., Canada, Australia and New Zealand 2013 [17]	The Candrive II/Ozcandrive cohort n = 1230 70 to 94 years old Male 64.2%	2 years prior to the baseline assessment	Snellen Binocular No collision Mean 14.0 (SD 5.1) Collision Mean 14.8 (SD 4.7)	-	-	-	State records 5.1% involved in an MVC	VA p > 0.01	None	Fair
Chen et al., Taiwan 2018 [25]	Motorcycle riders n = 256 ≥60 years old Male 51.2%	1-year study period 2013–2014	Rosenbaum cards Abnormal <20/50 31.3%	-	-	-	Self-reported 33.7%	Abnormal VA Without a crash 32.2% With a crash 29.1% p = 0.622	None	Fair

CS: contrast sensitivity; ETDRS: Early Treatment Diabetic Retinopathy Study chart; HR: hazard ratio; MVC: motor vehicle collision; NIH: National Institute of Health—USA (quality assessment tool for cohort and cross-sectional studies); RR: relative risk; RTA: road traffic accident; SD: standard deviation; VA: visual acuity. * The test used, cut-off point, and binocular results or, if not available, those of the eye with better vision are reported. ** The source of the accident report, the definition of accidents, and the incidence are stated.

Table 2. Characteristics of the case–control studies included in the qualitative review.

Author, Country and Year	Setting, Number of Cases and Controls, Age and Sex	Case Definition	Visual Acuity *	Colour Vision *	Contrast Sensitivity *	Stereopsis *	Association with RTA	Adjustment Variables	Quality Rating (NIH)
Humphriss South Africa 1987 [26]	Construction company 196 cases/170 controls No gender or age information provided	Company records	Orthorater vision screener Binocular Cases 12.7% fail Controls 5.9% fail	Orthorater vision screener	-	Orthorater vision screener	VA OR 2.34 (95% CI 1.09–5.02) ** Stereopsis OR 0.94 (95% CI 0.32–2.76) ** Colour vison differences no significant	None	Fair
Gresset et al., Canada 1994 [27]	Population-based 1400 cases/2636 controls 70 years old Male 100%	Insurance company	Cases: 6/12 or 6/15 8% Controls: 6/12 or 6/15 7.7%	-	-	Stereoacuity threshold greater than 200 sec arc Cases 0.4% Controls 0.4%	VA lower than 6/12 and binocularity OR 0.97 (95% CI 0.68–1.38) VA lower than 6/12 and lack of binocularity OR 1.23 (95% CI 0.88–1.72)	Traffic convictions, mileage, time spent and frequency of driving during rush hours	Good
McCloskey et al., USA 1994 [28]	Population-based 235 cases/448 controls ≥65 years old Male 49.9%	Drivers treated for injuries sustained in police-reported collisions	Aided VA 20/15 or 20/20 Cases 37.2% Controls 30.8% 20/25 or 20/30 Cases 46.1% Controls 50.8% 20/40 Cases 10% Controls 11.3% 20/50 or 20/60 Cases 3.9% Controls 5.9% 20/70 or greater Cases 2.8% Controls 1.1%	-	-	-	Aided VA 20/25 or 20/30 RR 0.7 (95% CI 0.5–1.1) 20/40 RR 0.6 (95% CI 0.3–1.2) 20/50 or 20/60 RR 0.3 (95% CI 0.1–0.9) ≥20/70 RR 4.3 (95% CI 0.5–40.3)	Controls matched to cases by age, sex, and country of residence	Fair
Owsley et al., USA 1998 [29]	Population-based 179 cases/115 controls 55 to 87 years old Male 54%	Drivers who had incurred at least one vehicle crash between 1985 and 1990 resulting in an injury to anyone in the involved vehicles according to the accident report. The other case group (n = 101) consisted of drivers involved in non-injurious crashes	ETDRS Binocular Injurious crashes ≤20/40 85.9% >20/40 14.1% Non-injurious crashes ≤20/40 85.6% >20/40 14.4% Non crash ≤20/40 90.4% >20/40 9.6%	-	Pelli–Robson Binocular Injurious crashes log >1.5 79.5% log ≤1.5 20.5% Non-injurious crashes log >1.5 83.8% log ≤1.5 16.2% Non-crash log >1.5 77.4% log ≤1.5 22.6%	TNO stereo test Injurious crashes <500 arcseconds 59% ≥500 arcseconds 41% Non-injurious crashes <500 arcseconds 71.8% ≥500 arcseconds 28.2% Non-crash <500 arcseconds 75.7% ≥500 arcseconds 24.3%	Injurious crashes AV worse than 20/40 OR 1.6 (95% CI 0.6–3.8) SC ≤ 1.5 OR 0.9 (95% CI 0.4–1.8) Stereoacuity ≥500 arcseconds OR 2.2 (95% CI 1.1–4.1) Non-injurious crashes AV worse than 20/40 OR 1.6 (95% CI 0.7–3.6) SC ≤1.5 OR 0.7 (95% CI 0.3–1.3) Stereoacuity ≥500 arcseconds OR 1.2 (95% CI 0.7–2.3)	None	Good

CS: contrast sensitivity; ETDRS: Early Treatment Diabetic Retinopathy Study chart; NIH: National Institute of Health—USA (quality assessment tool for case–control studies); OR: odds ratio; RR: relative risk; RTA: road traffic accidents; VA: visual acuity. * The test used, cut-off point, and binocular results or, if not available, those of the eye with better vision are reported. ** Odds ratios were calculated when not reported.

Table 3. Characteristics of the cross-sectional studies included in the qualitative review.

Author, Country and Year	Sample, Age and Sex	Study Population	Visual Acuity *	Colour Vision *	Contrast Sensitivity *	Stereopsis *	Accidents **	Association with RTA ***	Adjustment Variables	Quality Rating (NIH)
Davison England 1985 [30]	n = 1000 16 to 79 years old >55 years = 173 Male 83.7%	Private drivers	Keystone Telebinocular Binocular ≥ 6/7.5 93.6% 6/7.5 o 6/6.5 4.3% <6/7.5 2.1%	Keystone Telebinocular ≥1 colour plates misread 4.4% (5.3% of males)	-	Keystone Telebinocular Unsatisfactory (score < 4) 8.4%	Self-reported 13.4% within the previous 3 years	≥55 years and binocular VA 6/9 or worst PR 7.68 (95% CI 2.47–23.88)	None	Poor
Ivers et al., Australia 1999 [31]	The Blue Mountains Eye Study n = 2379 >49 years old	Private drivers	Vectorvision CSV-1000 chart. Best eye < 20/40–20/60 4.3% <20/60 1.2%	-	Vectorvision CSV-1000 chart. Best eye 3 CPD 14.8% 6 CPD 15.9% 12 CPD 21.4% 18 CPD 25.3%	-	Self-reported 5.8% in the last 12 months	VI < 20/40–20/60 PR 1.3 (95% CI 0.6–2.8) VI < 20/60 PR 1.2 (95% CI 0.3–5.0)	Age, sex, past and current use of benzodiazepines, phenothiazines, and antidepressants, self-reported history of stroke, arthritis, angina, heart attack, hypertension, and diabetes, health status and hearing impairment	Poor
Mwangi Kenia 2001 [32]	n = 539 21 to 65 years old Male 99.6%	Public service vehicle drivers (van, minibus and bus)	Snellen Better eye ≥6/9 87.4% <6/9 12.6%	Ishihara Deficient 1.7%	-	-	Self-reported Included both minor and major accidents. 40.5% in the last 3 years	VI PR 1.17 (95% CI 0.94–1.46) Colour vision PR 1.38 (95% CI 0.76–2.50)	None	Poor
Abebe et al., Ethiopia 2002 [33]	n = 1879 Mean 33.5 (SD 10.2) years old Male 93.7%	Private, taxi, state or institutional drivers	-	Ishihara Abnormal 4.5% (Deuteranopes 65.9%, protanopes 34.1%)	-	-	Self-reported Any degree of damage inflicted on human beings or material damage caused by a motor vehicle: 21.2% in the past 3 years of active driving	Colour vision PR 1.54 (95% CI 1.11–2.13)	None	Fair
Oladehinde et al., Nigeria 2007 [34]	n = 215 21 to 75 years old Male 100%	Commercial drivers	Snellen Better eye Normal 96.7% <6/18–6/60 3.3%	Ishihara The only driver with colour vision abnormality was involved in RTA	-	TNO stereo test Does not report results	Self-reported History of RTA 26.5%	VI PR 3.50 (95% CI 2.38–5.14) Stereopsis PR 0.67 (95% CI 0.19–2.41)	None	Poor
Bekibele et al., Nigeria 2007 [13]	n = 99 38 to 60 years old Male 100%	Drivers from a college of medicine and a hospital	VI WHO definition3.3%	Ishihara Does not report results	-	-	Self-reported Mild (with no serious bodily injuries or damage to the vehicle), moderately serious or very serious (with major injuries/mortality or damage to the vehicle) 16,2% since they commenced driving. 4% in the last 5 years	VI < 6/18 OR 1.5 (95% CI 0.5–4.6)	None	Poor
Adekoya et al., Nigeria 2009 [35]	n = 399 Mean 44.7 (SD 10.1) years old Male 100%	Public transport, commercial intercity vehicle	Snellen Better eye VI WHO definition 2.8% FRSC 11.5%	Ishihara Abnormal colour vision: the inability to correctly identify 1 or more pseudoisochromatic plates 4.3%	-	-	Self-reported History of RTA 20.8%	VI < 6/9 RP 1.07 (95% CI 0.52–2.24)	None	Poor
Ovenseri-Ogomo et al., Ghana 2011 [16]	n = 206 18 to 68 years old Male 100%	Commercial drivers	Snellen ≥6/9 in the better seeing eye 87.9% <6/18–6/60 6.8%	Ishihara Abnormal 3.4%	-	-	Self-reported History of RTA Does not report prevalence data	VI was not significantly associated with a positive history of RTA (p = 0.050) None of those who had colour vision defect had a history of RTA	None	Poor
Isawumi et al., Nigeria 2011 [36]	n = 99 22 to 66 years old Male 100%	Commercial motor vehicle drivers (buses and taxis)	Snellen WHO classification Either eye ≥ 6/18 93.9% <6/18 6.1%	Ishihara Abnormal 6.1% (95%CI, 0.2–8.0) The abnormalities were distributed as deutan 1.0%, protan 4.1% and both 1.0%	-	-	Self-reported History of RTA 20.2%	VI PR 4.41 (95% CI 2.24, 8.69) Colour vision PR 1.72 (95% CI 0.52–5.75)	None	Poor
Onabolu et al., Nigeria 2012 [37]	n = 524 19 to 66 years Male 100%	Commercial drivers	Snellen Worse eye ≥ 6/12 88.4% <6/12 11.6%	-	-	-	Self-reported Previous RTA 18.1%	VI PR 0.51 (95% CI 0.23–1.12)	None	Poor
Korir Kenya 2012 [38]	n = 186 18 to 52 years Male 100%	Motorcycle riders	Lea test Better eye logMAR 0.2 and better 97.3% Worse 0.2 2.7%	Ishihara Abnormal 2.2% Missing 1.6%	-	Titmus 40 61.8% 50 to 400 32.3% 3000 to 0 4.3% Missing 1.6%	Self-reported Major (requiring medical attention at a medical facility) and minor (treated at site of the accident) and whether they were at fault 55.4%	VA PR 1.46 (95% CI 0.93–2.31) Colour vision PR 0.90 (95% CI 0.34–2.43) Stereopsis ≤ 3000 PR 1.14 (95% CI 0.66–1.98)	None	Poor
Omolase et al., Nigeria 2012 [39]	n = 90 22 to 70 years old Male 100%	Commercial intercity drivers	Snellen Better eye 6/5–6/18 97.8% 6/18–6/60 2.2%	Ishihara Normal 100%	-	-	Self-reported History of RTA 20%	VI PR 2.59 (95% CI 0.61–11.04)	None	Poor
Abraham et al., Nigeria 2013 [40]	n = 291 Mean 41.5 (SD 11.1) years old Does not report sex	Commercial drivers, active for at least three months	Snellen Better eye ≥ 6/9 93.1% ≤6/12 6.9%	Ishihara Abnormal 2.2%	-	-	Self-reported Included major and minor accidents. A major accident is one resulting in serious damage of the vehicle and/or causing any form of injury or death to person(s) or livestock. 24.7%	VI PR 1.63 (95% CI 0.55–4.87) None of the colour defective drivers was involved in RTA	None	Poor
Achigbu et al., Nigeria 2013 [41]	n = 615 Mean 38.1 (SD 10.3) years old Does not report sex	Commercial motorcyclists	Snellen Better eye ≥6/12 98.9% <6/18–3/60 1.1%	Ishihara Abnormal 2.6%	-	-	Self-reported History of RTA 57.7%	VI PR 1.49 (95% CI 1.10–2.04) Colour vision PR 1.65 (95% CI 1.42–1.91)	None	Poor
Pepple et al., Nigeria 2014 [42]	n = 400 25 to 62 years old Male 100%	Commercial drivers	Snellen WHO definition Good vision 98.2% Impaired vision 1.8%	Ishihara Abnormal 4.5%	-	-	Self-reported Major accident (damage to the vehicle beyond repairs or death to person(s) or livestock) and minor accident (damage to the vehicle but without loss of lives) 45.5%	VI PR 0.62 (95% CI 0.19–2.02)	None	Poor
Chidi-Egboka et al., Nigeria 2015 [43]	n = 328 20 to 70 years old Male 100%	Commercial drivers who drive four wheel vehicles and convey passengers both intra and intercity	Snellen FRSC standard for commercial driving (better or equal to 6/9 in the better eye and 6/24 in the second eye) Adequate 90.9% Inadequate 9.1%	Ishihara Abnormal 10.4%	-	-	Self-reported RTA in the last 5 years 19.5%	VA PR 1.42 (95% CI 0.75–2.69) Colour vision PR 0.99 (95% CI 0.47–2.11)	None	Poor
Okafor et al., Nigeria 2015 [44]	n = 407 Mean 43.4 (SD 10.8) years old No gender information	Intercity commercial minibus drivers	Snellen Better eye 6/5–6/9 90.2% ≤6/12 9.8%	-	-	-	Self-reported History of RTA 15.2% in the past 5 years	VA PR 0.96 (95% CI 0.51–1.79)	None	Poor
Boadi-Kusi et al., Ghana 2016 [45]	n = 520 Mean 39.2 (SD 11) years old Male 100%	Commercial vehicle drivers	≥0.2 logMAR 97.5% <0.2 logMAR 2.5%	Hardy– Rand–Rittler Abnormal 7.1% Protan 45.9%, deutan 35.1%, tritan 18.9%	-	Randot Stereoacuity chart Abnormal 15.4%	Self-reported History of RTA 22.5%	Colour vision PR 1.63 (95% CI 1.02–2.61) Stereopsis PR 1.13 (95% CI 0.75–1.72)	None	Fair
Kshatri et al., India 2016 [46]	n = 195 22 to 65 years old Male 100%	Inter-state commercial drivers (truck drivers)	Snellen Better eye 6/5–6/18 94.4% 6/18–6/60 5.6%	Ishihara Normal 100%	-	-	Self-reported History of RTA 19%	VA PR 0.96 (95% CI 0.26–3.47)	None	Poor
Verma India 2016 [47]	n = 387 20 to 60 years old Mostly male	Drivers of private and public transport	Keystone DVS-GT Deluxe Vision Screener—Model # 1158WE Both eyes Acceptable: 6/6–6/12 90% Unacceptable: more than 6/12 10%	Keystone DVS-GT Deluxe Vision Screener—Model # 1158WE Acceptable: If subject identifies all three colours 86% Unacceptable: If the subject does not identify colours correctly 14%	Keystone DVS-GT Deluxe Vision Screener—Model # 1158WE Acceptable: Subject able to read within 40% contrast 96% Inacceptable: Subject unable to read within 40% contrast 4%	Keystone DVS-GT Deluxe Vision Screener—Model # 1158WE Acceptable: If both near and far signs are observed correctly 71% Inacceptable: The subject fails to identify the position of close and far away signs 29%	Organisational databases or self-reported crash history 67.4%	VA PR 1.47 (95% CI 1.32–1.64) Colour vision PR 0.86 (95% CI 0.68–1.08) CS PR was not calculated due to error in the data. Depth perception PR 1.7 (95% CI 1.52–1.90)	None	Poor
Vision Impact Institute India 2017 [48]	n = 627 20 to 70 years old Male 99.7%	Commercial drivers	Keystone View Vision Screeners test Binocular Unacceptable ≤ 20/50 16% Marginal 20/40 15% Acceptable ≥20/30 69%	Keystone View Vision Screeners test Red/green Unacceptable: None or 1 correct 19% Marginal: 2 correct 20% Acceptable: 3 correct 61% Blue/violet Unacceptable: None or 1 correct 23% Marginal: 2 correct 24% Acceptable: 3 correct 53%	Keystone View Vision Screeners test Unacceptable ≥ 60%: 4% Marginal 50%: 2% Acceptable ≤ 40%: 94%	Keystone View Vision Screeners test Unacceptable 32″ 68% Marginal 45″ 22% Acceptable 74 to 592″ 9%	Self-reported Crash history 4.3%	VA marginal/unacceptable PR 3.25 (95% CI 1.54–6.87) Does not report colour vision data Stereopsis marginal/unacceptable PR 1.98 (95% CI 0.27–14.43)	None	Poor
Okamura Japan 2019 [14]	n = 458 ≥40 years old Male 78.2%	General drivers	Log MAR Better eye < 0.7 Police-registered MVCs (with MVC 0%, without MVC 0.5%) Self-reported at-fault MVC (With MVC 0%, without MVC 0.5%) Police-registered MVCs: With MVC −0.03 [SD 0.08], without MVC −0.06 [SD 0.10] Self-reported at-fault MVC: With MVC −0.04 [SD 0.08], without MVC −0.06 [SD 0.10]	-	-	-	Police-registered at-fault MVCs, many of which resulted in injuries: 5% Self-reported in the previous five years, by level of severity (resulting in injury or in property damage only) and level of at-faultiness (not at fault, partially at fault, and at fault) Self-reported at fault 7%	Police-registered or self-reported: VA in better eye <0.7 p = 1.00 Police registered. VA in better eye (logMAR) p = 0.204 Self-reported VA in better eye (logMAR) p = 0.417	None	Poor
Ojabo et al., Nigeria 2020 [15]	n = 122 18 to 56 years old Male 100%	Commercial vehicle drivers	Snellen Both eyes ≥6/12 98.4% <6/12 1.6%	-	-	-	Self-reported Involvement in RTA in the past 10 years 44.3%	The two drivers with impaired visual acuity have never had RTA	None	Poor
Okafor et al., Nigeria 2020 [49]	n = 315 21 to 70 years old Male 98.7%	Commercial long-distance drivers	Snellen Normal (6/6–6/18) 90.5% Visual impairment (<6/18–6/60) 8.9% Severe visual impairment (<60/60–3/ 60) 0.6%	-	-	-	Self-reported History of RTA 47.9%	VI PR 1.37 (95% CI 1.01–1.85)	None	Poor
Ahmed et al., Bangladesh 2022 [50]	n = 691 18 to 84 years old Male 100%	Bus drivers	Snellen Better eye ≤6/9 18%	-	-	-	Self-reported RTA while driving a bus or minibus 9%	VA 6/7.5 in better eye PR 1.36 (95% CI 0.78–2.39)	None	Poor
Vofo et al., Cameroon 2022 [51]	n = 207 22 to 72 years old Male 100%	Commercial vehicle drivers (bus, taxi, truck)	Snellen Better eye ≥6/9 85.9% <6/9 ≥ 6/12 3.9% <6/12 ≥ 6/18 7.2% <6/18 ≥ 6/24 2.9% <6/24 ≥ 6/36 0.5%	-	-	-	Self-reported RTA over the previous 10 years 59.9%	VA < 0.5 PR 1.71 (95% CI 1.46–2.01)	None	Poor

* The test used, cut-off point, and binocular results or, if not available, those of the eye with better vision are reported. ** The source of the accident report, the definition of accidents, and the prevalence are stated. *** Prevalence ratios were calculated when not reported. FRSC: Federal Road Safety Commission; CS: contrast sensitivity; MVC: motor vehicle collision; NIH: National Institute of Health—USA (quality assessment tool for cohort and cross-sectional studies); PR: prevalence ratio; RTA: road traffic accidents; SD: standard deviation; VA: visual acuity; VI: visual impairment; WHO: World Health Organization.

Of the 39 studies retrieved for the qualitative review, 2 (5.1%) assessed the four visual functions of interest for this study, 8 (20.5%) assessed three functions, 17 (43.6%) assessed two functions and 12 (30.8%) assessed only one visual function.

Three of the included articles—one case–control study [29] and two cohort studies [18,19]—corresponded to the same study and were included because they reported data on different visual functions or were collected at different times.

Of the cross-sectional studies, 57.7% included only men, 88.5% were from low- and middle-income countries, 73.1% were from African countries, and 88.5% included public transport drivers. Of the cohort and case–control studies, 92.3% were conducted in high-income countries and in older adult populations. One case–control study [21] included only women. The occurrence of RTAs was self-reported in 96.2% of the cross-sectional studies and 22.2% of the cohort studies. Two cross-sectional studies [38,41] and one cohort study [25] evaluated motorcyclists.

The occurrence of RTAs was self-reported in 92.3% of the cross-sectional studies and in 22.2% of the cohort studies; none of the case–control studies used surveys to measure the outcomes. The definition of RTA was reported in 26.9% of the cross-sectional studies, 75% of the case–control studies and 88.9% of the cohort studies. The frequency of accidents was heterogeneously reported, with prevalence values ranging from 4% in the last 5 years [13] to 67.4% in an undefined period [47]. The cumulative incidence ranged from 5.1% in 2 years [17] to 33.7% in 1 year [25] (Table 1, Table 2 and Table 3).

Regarding the assessment of methodological quality by study type, 55.6% and 44.4% of the cohort studies were good and fair, respectively. The least met criteria were not assessing exposure more than once (66.7%) or not assessing different levels of exposure or treating it as a continuous variable (44.4%). For case–control studies, 50% were of good quality and 50% were of fair quality; the least common criterion was the ability to show that exposure occurred before the outcome (50%). For the cross-sectional studies, 92.3% were of poor quality and 7.7% were of fair quality; all of them assessed self-reported outcomes, and 96.2% did not adjust for other variables.

3.3. Visual Acuity

Only one study did not report visual acuity (VA) results [33]. Of those that did report VA, the most common methods used were Snellen optotypes (47.4%), ETDRS (15.4%), or other types of visual acuity testing (15.8%). In addition, 29.0% used data from both eyes and 36.8% used data from the eye with better vision. Each study reported different cut-off points for classifying subjects with reduced VA, the most common being <6/12 (42.1%), <6/18 (26.3%), and <6/9 (23.7%).

The meta-analysis included 18 cross-sectional studies (16 of commercial drivers and 2 of private drivers), 3 case–control studies (private drivers) and 5 cohort studies (private drivers). In the cohort of private drivers (RR 1.04, 95% CI 0.74–1.46; I² = 0.0%; n = 2172 and HR 1.08, 95% CI 0.85–1.36; I² = 0.0%; n = 3317) and case–control (OR 1.04, 95% CI 0.78–1.40; I² = 0.0%; n = 5013) studies, the risk of RTA was not increased when VA was reduced (Figure 2). Cross-sectional studies found a non-statistically significant higher risk was found for private drivers (PR 3.00, 95% CI 0.53–17.07; I² = 84.5%; n = 3379), but a statistically significant higher risk for commercial drivers (PR 1.54, 95% CI 1.26–1.88; I² = 72.0%; n = 5714), with significant heterogeneity and no evidence of publication bias (Egger’s test p = 0.709) (Figure 3).

Figure 2. Forest plot of risk of road traffic accident (police-reported) among private drivers with reduced visual acuity by type of observational study [19,21,22,23,24,27,28,29].

Figure 3. Forest plot of risk of road traffic accident (self-reported) among private and commercial drivers with reduced visual acuity in cross-sectional studies [30,31,32,34,35,36,37,39,40,42,43,44,46,47,48,49,50,51].

3.4. Colour Vision

Nineteen studies (48.7%) included colour vision testing; 94.7% were cross-sectional, and all of them assessed commercial drivers. The most commonly used test was the Ishihara test (73.7%), with the frequency of alterations ranging from 0% to 10.4%. Only 57.9% of the studies reported results on the risk of RTA in the presence of abnormal colour vision.

The meta-analysis was performed with eight cross-sectional studies and found no increased risk of RTA with abnormal colour vision (PR 1.50, 95% CI 0.91–2.45; I² = 92.0%; n = 4367); there was a considerable degree of heterogeneity (Figure 4).

Figure 4. Forest plot of the risk of a road traffic accident (self-reported) in commercial drivers with abnormal colour vision, according to cross-sectional studies [32,33,34,36,42,43,45,47].

3.5. Stereopsis

Twelve studies (30.8%) assessed stereopsis using various vision screeners (33.3%), the TNO (25%) and Randot (16.7%) tests.

The meta-analysis included with four cross-sectional studies, two case–control studies and two cohort studies. In the cross-sectional studies, no risk of RTA was found in commercial drivers where reduced stereopsis was evidenced (PR 1.40, 95% CI 0.99–1.99; I² = 44.8; n = 1749), with moderate heterogeneity. No risk was observed in case–control studies (OR 1.42, 95% CI 0.89–2.28; I² = 0.0%; n = 4330) or cohort studies (HR 1.12, 95% CI 0.82–1.52; I² = 52.2%; n = 3213) (Figure 5).

Figure 5. Forest plot of the risk of road traffic accidents in commercial and private drivers with reduced stereopsis, according to the type of observational study [21,22,27,29,34,38,45,47,48].

3.6. Contrast Sensitivity

Ten studies (25.6%) reported contrast sensitivity outcomes, of which 60% of these were cohort studies and 60% used the Pelli–Robson test.

Three cross-sectional studies reported contrast sensitivity data [14,31,48] but did not assess the association with accident rates. A small risk of RTA was seen in older adults with impaired contrast sensitivity and in cohort studies reporting the cumulative incidence (RR 1.42, 95% CI 1.01–2.00; I² = 0.0%; n = 2172), but not in those reporting HR (HR 1.01, 95% CI 0.91–1.11; I² = 0.0%; n = 3313) (Figure 6).

Figure 6. Forest plot of the risk of road traffic accidents (police-reported) in drivers with reduced contrast sensitivity, with cohort studies [19,21,22,23,24].

4. Discussion

This systematic review and meta-analysis identified the risk of RTA based on the status of several visual functions that are assessed in common driving tests, which is specifically related to ocular pathology. A small number of cohort or case–control studies were found and were conducted in high-income countries and in populations of older adults. Conversely, cross-sectional studies have been conducted in low- and middle-income countries, often involving public service drivers, and the subjects have been predominantly male. This finding reflects the interest in issues related to the licencing of commercial drivers, who have very high exposure to driving, often in heavy traffic conditions. The other group of interest has been older adults with physical and functional comorbidities. There has been one review of the impact of visual function in the general population [14,47] and only three of the studies [25,38,41] have assessed motorcyclists. Studies conducted in commercial drivers or older adults could not be extrapolated to other types of drivers.

An association between impaired VA and RTAs has been found in cross-sectional studies, but not in cohort and case–control studies. However, only one cross-sectional study performed a multivariate analysis on aspects related to the general health status of the drivers [31].

Only two of the studies [47,48] reported the outcomes for the four visual functions of interest for this review, probably due to the use of vision screeners for measure them. Some studies measured all four variables but did not report the results of all of them in the study [13,26,34]. Others only reported VA results [13,14,15,20,25,28,30,38,45,50,51,52], so aspects of the visual function that may be important during driving are missing. This systematic review found a significant but weak association between reduced VA and RTAs in commercial drivers, according to cross-sectional studies. This finding follows the same trend as the results of Piyasena et al. (RR 1.41, 95% CI 1.26–1.59), although they had different selection criteria and included the same subjects more than once in their meta-analysis in order to use different cut-off points [8]. Although the ability to resolve details is the mandatory test for obtaining a driving licence and it has face validity, research on driving safety has produced results with significant but weak or no significant associations, partly due to inadequate sample sizes. In addition, static VA, measured under high contrast and high lighting conditions, does not reflect the visual complexity required for driving; for example, it does not take into account peripheral vision, movement, or changes in contrast levels. Another explanation for the results obtained is that visually impaired drivers may voluntarily stop driving [6].

All the studies that looked at colour vision were cross-sectional and were carried out in low- and middle-income countries. The most commonly used test for diagnosis was Ishihara, but most of the studies did not specify the type of abnormality found. The results obtained in this study do not indicate an increased risk of RTA in those with colour vision abnormalities. This is similar to the observations of Piyasena et al. (RR 1.36 (95% CI 1.01–1.82)), who provide more reliable results by including data from the same study twice to use different cut-offs [8]. In our study, if cross-sectional studies of motorcyclists were included, a statistically significant association would be found (RP 1.47 95% CI 1.03–2.10). Colour vision deficiency is a risk factor for RTA, particularly in the case of protanopia, as it can affect the ability to react to traffic lights and require the driver to adjust their driving accordingly [53]. Normal colour vision ensures that drivers can correctly identify the traffic devices and coloured signals on the road, in particular the rear lights [6].

No association was found between reduced stereopsis and accidents. This seems strange because stereopsis provides information about the distances between objects, the volume of objects, and the bumps or unevenness of the road and is an indicator of the state of binocularity. Other reviews have reported contradictory results in people with poor stereopsis and in monocular individuals, and clinical studies have also found no significant differences in accident rates according to stereopsis [54]. Therefore, as of now, there is currently insufficient evidence to conclude that reduced stereopsis is a hazard in normal everyday driving for non-commercial drivers [55].

Contrast sensitivity was only reported by studies from high-income countries, and a significant but weak association with RTAs was observed. A prospective study of older adults in the USA, which was not included in this review because it was based on naturalistic driving data, found no association between photopic contrast sensitivity and higher rates of motor vehicle accidents, although an association was found with mesopic contrast sensitivity. Assessment of this function is relevant, given the large variability in road lighting levels, both during the day and at night. Drivers with impaired photopic contrast sensitivity are more likely to avoid challenging driving situations and reduce driving exposure (i.e., mileage) and are at higher risk of retiring from driving [56].

The results for stereopsis and contrast sensitivity are based on a very small number of studies, which may affect their reliability, so caution should be exercised in generalising them.

Data from cross-sectional studies do not allow comparisons of crash risk based on driver age. Piyasena et al. made this comparison based on the mean age (<40 years and ≥40 years) but found no difference [8].

With age, functional changes occur in VA, contrast sensitivity, and depth perception as well as in refractive error, dark adaptation, and glare recovery [57]. The cohort studies included in this review only measured visual functions once at baseline, but these may have changed during follow-up, especially in older adults. The studies could not conclude whether the more severe the change, the more severe the RTAs.

This review was focused on the behaviour of vision in healthy subjects, both in terms of ocular conditions and motor and cognitive aspects. Visual field alterations occur in several eye diseases, such as glaucoma, age-related macular degeneration, retinitis pigmentosa, and optic pathway lesions, and there is ample evidence of their impact on visual function [8,9]. Therefore, the visual field was not included in the visual functions analysed.

This review excluded studies that only included patients with systemic diseases, such as diabetes, Parkinson’s disease, or rheumatoid arthritis, and eye diseases, such as glaucoma, age-related macular degeneration, or cataracts.

Several biases were present in the included studies, especially in the cross-sectional studies. The reporting of VA was very variable, with binocular results, or results from the better eye, or only from the right eye. In most cases, it was not clear whether the data were taken with or without optical correction, or whether the driver was wearing their glasses at the time of the crash. There was also no mention of the lighting conditions of the optotypes or the locations where the tests were carried out. In addition, the type of test used and the cut-off points were different in each study. The same was true for colour vision, stereopsis, and contrast sensitivity. Also, all of the studies included in the meta-analysis were considered to be of low quality, so there is a need for cross-sectional studies using very rigorous methods and with better and more complete descriptions in the reports. Cohort studies in which accidents are reported by the police are particularly needed.

Regarding the measurement of outcomes, most of the articles did not specify the definition of RTA used. Very few of them distinguished whether the driver was at fault [14,18,19,23,24,29] and whether the accident was serious or minor [27,38]. In addition, self-reporting is likely to have underestimated the frequency of RTAs, so that only less serious accidents in which the subjects survived were included. It was clear that not all countries had data from police accident records.

Despite the increase in the number of vehicles and RTAs worldwide, which has become a public health issue, few studies were found from Asia and none from Latin America. Therefore, the organisations or institutions concerned, such as the road safety agencies, should promote studies that include the visual function of drivers, in addition to other relevant variables involved in the act of driving.

RTAs can be caused by human, vehicle, and environmental factors, among others [40], but only seven studies in the review present multivariate analysis models, demonstrating the need for adequate control of confounding variables to obtain valid results. Not only VA but also contrast sensitivity, colour vision, stereopsis and type of refractive error need to be considered to fully reflect the complexity of the visual system. Studies need to include detailed and clear methodological aspects, such as the tests used, conditions of use and cut-off points, to provide a complete insight into the visual aspects of driving. In addition, studies should, as far as possible, be based on objective accident reports that allow identification of the type of accident (minor or serious, at fault or not) and the period analysed, even in cross-sectional studies.

Research conducted in low- and middle-income countries has provided information through descriptive studies that report crash risk based on VA or colour vision alone. Conversely, studies from high-income countries show the importance of analysing complex variables that include visual processing speed and divided attention [6]. Of relevance are the studies by the University of Alabama Department of Ophthalmology group, conducted between 1993 and 2000, with retrospective and prospective results from the same group of patients. Some of these were excluded from this review because they did not present their results using measures of association [58,59]. The appropriate methodology for developing research to support decision making in road safety and vision should be adapted to the social and economic context of countries, especially low- and middle-income countries, so that the results can be implemented by policy makers and accepted by the interested public. In this regard, cohort and case–control studies of motorcyclists, who are generally young and visually healthy, would be very useful; this is a growing population in middle- and low-income countries.

The trend in new research in developed countries on road safety and its relationship to vision is moving towards autonomous vehicles and artificial intelligence. In the developing world, however, the study of human visual function will continue to be fundamental to decision making on policies to improve road safety. In low- and middle-income countries, lack of access to eye care can lead to a higher number of crashes due to untreated problems. It is important that road safety research takes into account local characteristics in terms of risk factors such as the prevalence of visual impairment, access to eye care, and local beliefs. Road safety policies should recognise this reality and promote measures such as regular eye examinations for drivers, subsidies for spectacles and awareness campaigns on the importance of eye health.

In conclusion, according to cross-sectional studies, there is a higher risk of RTA in drivers with impaired VA than in those with normal vision. The few case–control and cohort studies do not show an association between the occurrence of accidents and these visual functions. It is necessary to draw attention to this issue so that quality research can be conducted to support the formulation of road safety policies related to the visual health of drivers. The studies analysed, most of which are cross-sectional, may not be sufficient to take preventive action.