Alcohol-related fatal-and-injury traffic crashes in Türkiye (2015–2024): temporal trends and an ecological evaluation of alcohol control policies

Abstract

Background

Alcohol-impaired driving is a well-established and preventable cause of road traffic deaths and injuries. This study examined national time trends in the relative burden of alcohol-related fatal-and-injury road traffic crashes in Türkiye between 2015 and 2024 and interpreted the findings within an ecological policy and enforcement context.

Methods

We conducted a national ecological time-trend analysis using annual aggregate road traffic crash statistics from the Turkish Statistical Institute for 2015–2024. The outcome was the annual count of alcohol-related crashes resulting in death or injury. To estimate changes in the relative burden over time, the natural logarithm of the annual total number of fatal-and-injury crashes was included as an offset. Temporal trends were first assessed using Poisson regression; after overdispersion was identified, estimates were re-evaluated using negative binomial regression. Results are reported as incidence rate ratios (IRR) with 95% confidence intervals (CI).

Results

The proportion of alcohol-related fatal-and-injury crashes among all fatal-and-injury crashes declined from 2.13% in 2015 to 0.72% in 2024. In the Poisson model, calendar year was negatively associated with the relative burden (IRR 0.883; 95% CI 0.880–0.887; p < 0.001), although model fit indicated overdispersion. In the negative binomial model, the direction of the association remained negative but was not statistically significant (IRR 0.883; 95% CI 0.709–1.097; p = 0.260). Sensitivity analyses excluding 2020 alone and excluding both 2019 and 2020 yielded nearly identical estimates, indicating that the negative but non-significant trend was not materially driven by potentially atypical pandemic-period observations.

Conclusions

Descriptive data showed a marked decline in the relative burden of alcohol-related fatal-and-injury road traffic crashes in Türkiye from 2015 to 2024. However, negative binomial regression did not confirm a statistically significant temporal trend, indicating that model-based evidence should be interpreted cautiously. Interpretation should consider the ecological nature of the data and the possibility of time-varying enforcement intensity, detection, and reporting practices.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27792-5.

Introduction

Alcohol use is a long-established and strongly evidence-based risk factor for road traffic harm. Taken together, experimental, epidemiological, and toxicological studies clearly demonstrate the adverse effects of alcohol on driving performance [1–4]. Alcohol impairs attention, decision-making, reaction time, and other cognitive functions, and many of these effects have been shown to begin even at blood alcohol concentration (BAC) levels below legal limits [1, 2]. As BAC increases, crash risk rises in a consistent dose–response manner, and this relationship becomes more pronounced for fatal crashes [1, 3–5].

This strong association between alcohol and road traffic crashes has made alcohol control policies a core component of road safety strategies. In many countries, legal BAC limits, penal sanctions, and enforcement mechanisms aim to reduce drink-driving. However, the real-world effects of policy interventions are not always uniform. Studies evaluating reductions in legal BAC limits have reported heterogeneous findings: while some analyses found significant decreases in crashes and traffic-related health outcomes, others did not demonstrate statistically significant effects [6–8]. Evaluations of Utah’s reduction of the BAC limit from 0.08 g/dL to 0.05 g/dL reported decreases in total crashes, but suggested that the effect was more evident for property-damage-only crashes [7]. Similarly, the impacts of stricter drink-driving laws have not always been consistent across settings [8].

Enforcement mechanisms play a decisive role in the effectiveness of alcohol policies. Random breath testing (RBT) and related road policing strategies not only increase detection of violations but also strengthen deterrence by shaping drivers’ perceived likelihood of being caught [9, 10]. Evidence from the pandemic period further indicates that, despite substantial reductions in traffic volume, risky driving behaviours did not disappear and some behaviours may have increased in relative terms [10–12].

Methodological choice is also critical in studies examining crash patterns and policy effects. Road traffic crash data are inherently count data, and Poisson regression models are therefore widely used. However, when overdispersion is present—i.e., the variance exceeds the mean—standard error estimates from Poisson models may be affected [13]. For this reason, negative binomial models have been recommended as a more appropriate approach for analysing crash frequency and rate outcomes [13, 14].

In Türkiye, alcohol control is implemented within a multi-component policy framework that includes constitutional obligations to protect health, regulations on alcohol sales/service and advertising/promotion, and drink-driving sanctions embedded in traffic legislation. In the present study, this regulatory structure is considered as an ecological policy context for interpreting temporal trend findings [15–17]. Research directly examining alcohol and road traffic outcomes in Türkiye remains limited. Nazif-Munoz et al. reported that the 2013 alcohol-related traffic law (Law No. 6487) was associated with reductions in traffic fatality rates; however, that study focused on the relationship between a specific legal intervention and overall traffic fatalities rather than on long-term national trends in officially recorded alcohol-related fatal-and-injury crashes [15]. Similarly, Demir compared alcohol-related death rates in Türkiye with those of countries included in EUROSTAT data and described a decline in alcohol-related mortality over time, but did not examine road traffic crash outcomes or the relative burden of alcohol-related crashes within all fatal-and-injury crashes [18]. Therefore, the available evidence from Türkiye is informative but fragmented, as it is oriented either toward policy evaluation or alcohol-related mortality rather than direct analysis of national crash trends. The relative share of alcohol-related crashes among all road traffic crashes is therefore an important public health indicator.

Accordingly, this study aims to examine temporal trends in alcohol-related fatal-and-injury road traffic crashes in Türkiye between 2015 and 2024 using Turkish Statistical Institute data and to interpret the findings ecologically in the context of alcohol control policies and enforcement practices [19].

Methods

Data source and variables

This study is a national ecological time-trend analysis examining temporal patterns in alcohol-related fatal-and-injury road traffic crashes in Türkiye. Data were obtained from the official annual Road Traffic Accident Statistics published by the Turkish Statistical Institute (TurkStat), and annual aggregate data for the period 2015–2024 were analysed. The study period was restricted to 2015–2024 because TurkStat’s open-access annual reports have presented numerical data on alcohol-at-fault coding in a consistent format starting from 2015. According to TurkStat metadata, these statistics cover road traffic accidents involving death or injury detected by the police and gendarmerie.

The dependent variable was the annual number of officially recorded alcohol-at-fault fatal-and-injury road traffic crashes recorded in a given year. As the denominator for comparative purposes, we used the total annual number of fatal-and-injury road traffic crashes. The primary outcome was defined as the proportion of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury road traffic crashes:

Proportion = (Alcohol-related fatal-and-injury road traffic crashes) / (Total fatal-and-injury road traffic crashes).

Data were analysed at the national aggregate level rather than the individual level. In the open-access annual tables used in this study, alcohol-related crashes were identified through the official alcohol-at-fault coding reported in the dataset. Because the public dataset does not provide individual-level blood alcohol concentration values, year-specific testing coverage, or detailed field testing procedures, the findings should be interpreted as trends in officially recorded alcohol-at-fault crashes rather than direct estimates of driver-level alcohol exposure. Calendar year was included in the model as a continuous variable representing the temporal trend.

Ecological definition of the policy context

The legal and implementation framework for alcohol control was not included in the regression models as covariates for a causal policy-effect analysis; instead, it was defined as an ecological policy context to support interpretation of the temporal trend findings. Accordingly, the regulatory environment was summarised based on: (i) Law No. 4250 on spirits and alcoholic beverages; (ii) the drink-driving prohibition and related sanctions/enforcement practices under Article 48 of Highway Traffic Law No. 2918; and (iii) secondary legislation regulating alcohol sales and service. To reflect an in-period regulatory update, the regulation amendment dated 20 September 2023 was also included in this contextual framework (Table 1). The pre-2015 framework was not treated as an analytical baseline, but rather as the pre-existing regulatory context already in force at the start of the study period. This summary was used solely to provide interpretive context for the trend analysis and was not intended to test the causal effects of individual regulations.

Table 1.

Ecological policy context used for interpreting temporal trend analysis (Türkiye, 2015–2024)

Period/Year	Regulatory/Policy Framework	Methodological Use in This Study
Pre-2015 framework already in force at the start of the study period	Core restrictions under Article 6 of Law No. 4250 on alcohol sales/service/advertising (e.g., prohibition of sales/service to persons < 18 years, prohibition of sales via press/broadcast/mail, retail sales ban between 22:00–06:00, restrictions on sales in specific locations)	Considered the pre-existing regulatory context already in force at the start of the study period when interpreting 2015–2024 trends
2015–2024 (continuity)	Article 48 of Highway Traffic Law No. 2918 and traffic enforcement practices (alcohol checks, administrative sanctions, license suspension, vehicle impoundment/removal from traffic)	Used as the ecological context defining the enforcement component of drink-driving control
2020	Amendments to the sanction framework of Law No. 4250 under Law No. 7255 (especially additional administrative sanctions for violations of selected Article 6 provisions and license cancellation in repeated violations)	Included as an example of a regulatory update within the study period
2023	Regulation amendment dated 20.09.2023 (Official Gazette No. 32315) updating procedures and principles on the sale/service of tobacco products and alcoholic beverages (e.g., definitions, implementation provisions, licensing/application procedures)	Included as an in-period secondary legislation update within the ecological policy context

This table summarizes the ecological policy context used to support interpretation of temporal trend findings. Regulatory and enforcement components were not included in regression models as independent variables/covariates; therefore, the table does not provide causal evidence and is intended only as interpretive context. The pre-2015 framework is presented as the regulatory context already in force at the start of the study period, not as an analytical baseline with observed pre-period outcome data

Statistical analysis

All analyses were performed using IBM SPSS Statistics for Windows, Version 25.0 (IBM Corp., Armonk, NY, USA). Temporal trends were initially evaluated using Poisson regression with a log link function. The dependent variable was the annual count of alcohol-related fatal-and-injury road traffic crashes, and the natural logarithm of the annual total number of fatal-and-injury road traffic crashes was included as an offset. This modelling strategy was used to estimate changes over time in the relative burden (i.e., the proportion-based trend) of alcohol-related crashes among all fatal-and-injury crashes. Accordingly, the denominator was selected to address the study question as a within-crash-burden comparison rather than as a population-based or fleet-based incidence measure; alternative denominators such as the number of licensed drivers or registered vehicles would reflect a different analytical framework.

The model was specified as:

where Y_i denotes the number of alcohol-related fatal-and-injury road traffic crashes in year i, Year_i is the continuous temporal trend variable, and the offset term represents the logarithm of the total annual number of fatal-and-injury road traffic crashes. Calendar year was entered as the sole explanatory variable because the primary aim was to describe the overall national temporal trend in the relative burden of officially recorded alcohol-related crashes during 2015–2024, rather than to estimate the causal effect of a single policy intervention. For this reason, the legal and enforcement framework was treated as ecological interpretive context rather than modelled as an interrupted intervention series.

Model fit was assessed using the deviance/df and Pearson χ²/df statistics; values substantially greater than 1 were interpreted as evidence of overdispersion. When overdispersion was identified, analyses were repeated using negative binomial regression. Results are presented as incidence rate ratios (IRR), obtained by exponentiating regression coefficients, with 95% confidence intervals (CI). Statistical significance was set at (p < 0.05). Data from 2019 to 2020 were retained in the primary analysis because they belonged to the prespecified national study period and reflected real-world conditions that may have influenced crash patterns. However, these years were interpreted cautiously in view of possible pandemic-related changes in mobility, enforcement visibility, and reporting.

Results

Examination of the annual distribution of fatal-and-injury road traffic crashes and alcohol-related fatal-and-injury road traffic crashes during the study period showed a descriptive decrease in the relative proportion of alcohol-related crashes among all fatal-and-injury crashes (Table 2; Fig. 1). Over the 10-year period, the proportion declined gradually from 2.13% in 2015 to 0.72% in 2024, indicating a descriptive reduction in the relative burden of alcohol-related fatal-and-injury road traffic crashes within the overall fatal-and-injury crash burden (Table 2).

Table 2.

Annual distribution of the proportion of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury road traffic crashes (Türkiye, 2015–2024)

Year	Total fatal-and-injury road traffic crashes (n)	Alcohol-related fatal-and-injury road traffic crashes (n)	Proportion (%)
2015	183,011	3,900	2.13
2016	185,128	3,544	1.91
2017	182,669	3,193	1.75
2018	186,532	2,875	1.54
2019	174,896	2,334	1.33
2020	150,275	1,869	1.24
2021	187,963	2,099	1.12
2022	197,261	1,802	0.91
2023	235,071	1,721	0.73
2024	266,854	1,934	0.72

Fig. 1.

Time trend in the proportion of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury road traffic crashes (Türkiye, 2015–2024)

In the Poisson regression analysis, calendar year was significantly and negatively associated with the relative burden (proportion-based trend) of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury crashes (Wald χ² = 3280.838, p < 0.001). The temporal trend coefficient (B = − 0.124) was consistent with a decreasing pattern in the relative frequency of alcohol-related crashes. Based on the exponentiated coefficient, each one-year increase was associated with an approximately 11.7% decrease in the relative burden of alcohol-related fatal-and-injury road traffic crashes (IRR = 0.883; 95% CI: 0.880–0.887) (Table 3). However, deviance/df and Pearson χ²/df values substantially greater than 1 indicated overdispersion in the data. Therefore, the analyses were repeated using negative binomial regression.

Table 3.

Temporal trend in alcohol-related fatal-and-injury road traffic crashes: Poisson regression (Türkiye, 2015–2024)

Variable	B	SE	Wald	p-value	IRR (Exp(B))	95% CI (IRR)
Year	−0.124	0.0022	3280.838	< 0.001	0.883	0.880–0.887

In the negative binomial regression analysis, the association between calendar year and the relative burden of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury crashes was not statistically significant (Wald χ² = 1.269, p = 0.260). The trend coefficient remained negative (B = − 0.125) but did not reach statistical significance. Although the IRR was below 1, the 95% CI included 1, indicating that the magnitude of the association was not estimated with statistical precision (IRR = 0.883; 95% CI: 0.709–1.097) (Table 4).

Table 4.

Temporal trend in alcohol-related fatal-and-injury road traffic crashes: negative binomial regression (Türkiye, 2015–2024)

Variable	B	SE	Wald	p-value	IRR (Exp(B))	95% CI (IRR)
Year	−0.125	0.1113	1.269	0.260	0.883	0.709–1.097

Sensitivity analyses were performed to assess whether potentially atypical pandemic-period observations materially influenced the findings. Excluding 2020 alone did not meaningfully alter the results (B = -0.126, SE = 0.1111, Wald χ² = 1.279, p = 0.258; IRR = 0.882, 95% CI: 0.709–1.096). Similarly, excluding both 2019 and 2020 yielded nearly identical estimates (B = -0.126, SE = 0.1113, Wald χ² = 1.276, p = 0.259; IRR = 0.882, 95% CI: 0.709–1.096). These findings indicate that the direction and magnitude of the temporal trend were robust to exclusion of potentially atypical years (Supplementary Table S1).

Discussion

In this study, we observed a descriptive decline in the relative share of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury road traffic crashes in Türkiye over 2015–2024. Because the total number of fatal-and-injury road traffic crashes increased over the study period, the estimated trend should be interpreted as reflecting a change in the relative share of alcohol-related crashes within the overall fatal-and-injury crash burden, rather than as a population-based incidence trend. This descriptive pattern was consistent in direction with the regression estimates, although the negative binomial model did not demonstrate a statistically significant temporal association. This descriptively identified pattern is consistent with the experimental and observational literature demonstrating alcohol’s impairing effects on driving performance. Experimental driving simulation and eye-tracking studies indicate that alcohol disrupts attentional allocation, visual scanning strategies, and hazard perception, and that drivers’ capacity to adopt compensatory behaviours under alcohol influence remains limited [2]. In this framework, increased crash risk is not only attributable to impaired motor coordination, but also to deterioration in cognitive control and decision-making processes.

Epidemiological and forensic–toxicological evidence further supports these mechanisms. Studies examining the prevalence of alcohol and other psychoactive substances among drivers involved in fatal crashes consistently emphasise alcohol’s role, particularly in fatal collisions. Data from Norway suggest that alcohol positivity remains an important risk component in fatal crashes [4], while studies from Ontario have reported concurrent presence of alcohol and other substances among fatally injured drivers [3]. Accordingly, interpretation of the proportional decline observed in our study should not be confined to the biological effects of alcohol alone, but should be situated within broader contextual dynamics that shape drink-driving behaviour.

At the global level, alcohol continues to be a key behavioural risk factor for road safety outcomes. The World Health Organization reports that road traffic injuries cause approximately 1.19 million deaths each year and that alcohol involvement is among the major preventable risk factors [20]. In high-income countries, alcohol’s contribution to road traffic deaths is often reported in the range of 20–30%; however, these proportions vary substantially across countries according to legal BAC thresholds, enforcement intensity, and cultural norms [1, 20]. This perspective suggests that the trend observed in Türkiye represents not only a national statistical pattern but also a public health indicator that should be interpreted comparatively within international policy and enforcement regimes. Nevertheless, cross-country comparisons must be interpreted cautiously due to differences in definitions, recording practices, and reporting systems.

Population-based indicators for 2010–2022 show that the proportion of individuals aged ≥ 15 years reporting alcohol use in the past 12 months declined from 14.9% to 12.1% [21]. This decrease temporally overlaps with the decline in the proportion of alcohol-related fatal-and-injury crashes observed in our study and suggests that changes in alcohol use prevalence may be considered alongside road traffic risk indicators. Because these data were derived from a different measurement framework than the crash statistics analysed in this study, they are presented here as contextual information only (see Supplementary Fig. 1). Therefore, the parallel trends should not be interpreted as evidence of a causal relationship or as a direct explanatory factor for the observed crash trend; rather, they may reflect complementary changes in population-level exposure relevant to policy discussion.

From this perspective, it would be inappropriate to attribute the observed decline to a single cause. In Türkiye, alcohol control is implemented through a multi-component framework combining structural measures that regulate sales/service, advertising/promotion, and availability with sanctions embedded in traffic legislation [16]. International evidence similarly indicates that macro-level policies influencing alcohol availability and consumption patterns may be associated with road safety outcomes [1]. Even when such policies do not explicitly target drink-driving, they may exert indirect effects by reshaping the “opportunity structure” for high-risk behaviours. Time restrictions on retail sales, prohibitions on sales/service to persons under 18 years, advertising/promotion restrictions, and constraints on remote sales channels collectively influence not only physical availability but also environmental visibility and the likelihood of spontaneous access to alcohol [16]. Therefore, it is theoretically coherent that these regulations could indirectly affect drink-driving. In the Türkiye context, this framework should be considered together with the sales/service and advertising/promotion restrictions in force during the study period, traffic sanctions, and the regulatory updates implemented in 2020 and 2023.

At the same time, the potential effects of this framework may extend beyond the level of formal legal restriction and operate through normative processes that reshape alcohol’s public visibility and normalised contexts of consumption. Advertising/promotion bans and sales regimes may influence not only availability but also how alcohol is coded within public culture as a legitimate and routine practice. From this standpoint, the decline observed over 2015–2024 may also relate to a slower-acting but potentially more durable normative reframing process generated by regulatory measures [1, 16].

The link between normative/structural change and road safety becomes particularly salient in the context of social drinking practices and the night-time economy. Drink-driving is often more than an individual choice; it represents a contextual behavioural pattern shaped by spatial organisation, alcohol-serving environments, transport alternatives, and social interaction dynamics. The night-time economy literature emphasises that alcohol consumption is embedded within social practices and spatial organisation, and that serving environments, social norms, and contextual regulation can be determinant for risk behaviours [22]. Accordingly, the effects of alcohol policies may emerge not only through individual consumption decisions but also through reorganisation of drinking and post-drinking mobility contexts.

In this context, measures such as time restrictions on alcohol supply may shift the timing of drinking and post-drinking mobility patterns, thereby transforming the opportunity structure for risk. Critically, the key issue may not be elimination of risk but redistribution into different contexts. The literature has discussed that availability restrictions may, in some settings, shift drinking from public venues to private settings, potentially generating different risk patterns [1]. Therefore, the trend observed in Türkiye should be interpreted not as a linear “restriction → reduction” relationship, but within a framework of spatial–temporal redistribution of drinking and contextual transformation of risk behaviours.

The structure of the night-time economy and transport options appear to be particularly influential in this process. In areas with concentrated entertainment venues, mobility options during late hours and the visibility of enforcement are among the main structural determinants of drink-driving behaviour. Where transport alternatives are strong, substitution away from risky behaviour may be more feasible, whereas in settings with limited alternatives, risk may be more resistant to change [1, 23]. Thus, sustainably reducing risky behaviour depends not only on legal measures and sanctions, but also on strengthening safe return options, particularly during night-time hours [23].

Intermediary actors who help manage risk socially are also an important part of this framework. In recent years, increasing attention has been paid to the potential role of service providers and professionals in alcohol-serving environments in harm reduction. Studies evaluating Responsible Beverage Service (RBS) interventions suggest that bar staff may indirectly influence risk behaviours by recognising risky consumption, limiting overservice, and implementing visible “refusal of service” practices [24]. However, the evidence base is heterogeneous and, due to methodological quality limitations, consistent effects on injury outcomes are limited [25]. Nonetheless, time-series analyses reporting reductions in alcohol-related road traffic crashes following policy-level interventions such as mandatory server training suggest that RBS may be more effective in certain contexts when integrated with enforcement and sanction architectures [26]. International experience further indicates that a “package approach” may be more realistic than relying on a single intervention. Models combining enforcement, media-supported visibility (publicised enforcement), serving-environment regulations, and local transport alternatives have been reported to produce more sustainable behavioural deterrence. This supports the importance, in the Türkiye context, of designing policy not only around penalty levels but also around implementation visibility, local enforcement capacity, and the social contexts in which behaviours occur [27].

Another core dimension where regulations intersect with road safety is enforcement and deterrence. When considered together with implementation intensity and visibility, legal limits and sanctions for drink-driving constitute a normative–legal framework shaping driver behaviour. Studies examining the effects of lowering BAC limits indicate that effect sizes are context-sensitive and may vary with enforcement intensity, enforcement strategies, and societal norms [1, 7, 16]. The European road safety literature similarly underscores the decisive role of highly visible and sustained enforcement strategies in reducing alcohol-related risks [1]. A common theme across international evidence is that enforcement is not merely a legal sanction instrument; it is also a mechanism of “high-visibility deterrence” that reshapes risk perception. Meta-analytic findings on sobriety checkpoints indicate that enforcement programmes are associated with meaningful reductions in alcohol-related crashes, and that the magnitude of effect likely varies according to programme intensity, continuity, and contextual implementation characteristics. Accordingly, the decline observed in Türkiye should be interpreted not only in terms of the existence of legislation but also in relation to the field visibility of legislation and the continuity of enforcement practices [28].

Comparative country data also align with this interpretation. For example, while the proportion of total traffic deaths attributable to driving under the influence in the United States has been reported to remain high in recent years with a relatively plateauing pattern, Türkiye showed a more pronounced declining trend during the study period [29]. Heterogeneity is also notable within Europe: countries with lower legal BAC limits (particularly ≤ 0.03 g/dL) and systematic roadside testing often report lower levels of alcohol-related road traffic burden, whereas settings with a legal BAC limit of 0.05 g/dL and more limited enforcement intensity may report higher proportions [1, 28]. Although such comparisons do not establish causality, they support interpreting Türkiye’s trend as a process shaped jointly by the BAC threshold, enforcement intensity, and societal norms. At the same time, it should be acknowledged that the explanatory value of these comparisons is limited and that policy effects are closely linked to country-specific implementation capacity.

The pandemic period offered an important natural experiment for examining these relationships. During the COVID-19 period, despite reduced traffic volume in many countries, the expected declines in risky driving behaviours and some severe outcome indicators were not consistently observed. A study based on Canadian data reported increases in certain subgroups in risky behaviours such as speeding, driving under the influence of alcohol, and driving under the influence of substances, and reported a modest increase in the behaviour of “driving when one might have been over the legal alcohol limit” [11]. These findings indicate that reduced exposure does not, by itself, guarantee reduced risk.

A Queensland study more directly examining drink-driving behaviours during the pandemic similarly reported that when stationary roadside RBT operations were temporarily suspended, behaviours did not disappear and could be sustained in some groups or become the subject of increased intention [10]. The study highlighted that decreases in visible RBT may affect perceived deterrence. Therefore, when interpreting patterns around 2020 in our analysis, it is necessary to consider not only changes in traffic volume but also changes in enforcement intensity/visibility and behavioural adaptation processes [10, 11]. However, in sensitivity analyses excluding 2020 alone and excluding both 2019 and 2020, the estimated temporal trend remained virtually unchanged, suggesting that the negative but non-significant association was not materially driven by potentially atypical pandemic-period observations.

Finally, international evidence suggests that alcohol-related road traffic risk is often part of a broader cluster of risk behaviours rather than a single behaviour in isolation. Toxicology-based findings on concurrent use of psychoactive substances highlight the multidimensional nature of risk dynamics, increasing the importance of enforcement and prevention strategies that are sensitive not only to alcohol but also to polysubstance use patterns [3, 4].

Overall, while this study documents a decline in the relative share of alcohol-related fatal-and-injury road traffic crashes in Türkiye, it positions this trend within a multi-layered risk ecosystem rather than attributing it to a single policy instrument. Regulations, enforcement strategies, and social norms may influence driving behaviour through both external deterrence (perceived likelihood of detection and sanction threat) and internal regulation (transformation of normalised drinking contexts and the redrawing of boundaries of social acceptability). Accordingly, the central question is not only “Did policies work?”, but also “Which mechanisms were dominant in which contexts, and how can effects be sustained?” This framework supports evaluating alcohol-related road traffic risk through the interaction of legislation, availability, enforcement visibility, and social norms [1, 16].

From a policy-implications perspective, the findings suggest that structural and behavioural interventions are complementary in managing alcohol-related road traffic risk. When macro policies regulating alcohol availability are combined with highly visible and sustained enforcement strategies and safe mobility alternatives to risky driving (e.g., public transport, taxi/ride-share options), there is potential for stronger and more sustainable impact. International syntheses indicate that publicised enforcement contributes not only through sanctioning but also by strengthening public perceptions of the likelihood of detection—an important mechanism in policy impact [27]. As highlighted in the ERSO report, particularly in rural and access-constrained settings, the availability of alternative transport options may be decisive for risk reduction [1]. In addition, comparative policy experience suggests that lower legal BAC thresholds (especially ≤ 0.03 g/dL), when coupled with strong enforcement, may be associated with lower levels of alcohol-related road traffic harm. This point is important because while 0.05 g/dL is a widely used legal threshold, the adoption of thresholds such as ≤ 0.03 g/dL reflects a policy approach defining a lower tolerance for risk [1, 28]. Moreover, ignition interlock programmes have emerged as an effective tool for reducing recidivism and alcohol-related crash risk, particularly among convicted drivers [30, 31]. International evidence indicates that these programmes can substantially reduce repeat offending while the device is installed; therefore, feasibility in the Türkiye context could be considered in future policy discussions, taking into account legal infrastructure, enforcement capacity, and cost-effectiveness. Taken together, this approach frames road safety as a public health phenomenon shaped by the interaction of normative, environmental, and institutional components beyond individual behaviour.

Limitations

Because this study is based on national-level aggregate (ecological) TurkStat data, the findings cannot be interpreted at the individual level with respect to alcohol use, driver characteristics, or crash circumstances; therefore, there is a risk of ecological inference bias. In addition, the definition of “alcohol-related” crashes, the accuracy of recording, and classification procedures were outside the researchers’ control, and potential changes over time in measurement, reporting, and recording practices may have influenced the observed temporal pattern. Moreover, because year-specific alcohol testing coverage was not available in the open-access dataset, part of the observed temporal variation may reflect differences in detection and recording rather than true changes in alcohol-related crash occurrence.

The use of annual totals limited our ability to assess seasonality, regional heterogeneity, and short-term changes in policy or enforcement. The relatively short time series (2015–2024; 10 annual observations) also likely reduced statistical power and the stability of coefficient estimates in regression models. The need to move from Poisson to negative binomial regression due to overdispersion, and the resulting wider confidence intervals, made this limitation more apparent.

Although using the total number of fatal-and-injury road traffic crashes as an offset is appropriate for evaluating relative change over time, alternative exposure measures and structural determinants—such as traffic volume (vehicle-kilometres travelled), vehicle ownership, the number of drivers, population structure, and regional mobility patterns—could not be incorporated into the models. Similarly, enforcement intensity, the field visibility of alcohol control practices, the frequency of BAC testing, law-enforcement capacity, variation in sanction implementation, and pandemic-specific changes in mobility/enforcement could not be directly measured. Accordingly, the observed trends should not be interpreted as the causal effect of a specific regulation or a single policy intervention.

Finally, because data on psychoactive substances other than alcohol, polysubstance use, and driver-level toxicological/clinical information were not available, we could not assess the position of alcohol-related risk within broader substance-use patterns. Therefore, the findings should be interpreted within the context of Türkiye’s official recording system and the current structure of available data.

Conclusions and recommendations

This study showed a marked descriptive decline in the relative share of alcohol-related fatal-and-injury road traffic crashes among all fatal-and-injury road traffic crashes in Türkiye between 2015 and 2024. While this pattern is encouraging from a public health perspective, the presence of overdispersion and the resulting statistical uncertainty in model-based estimates indicate that a statistically significant temporal trend could not be confirmed in the negative binomial analysis and that the observed pattern should be interpreted cautiously. Rather, the findings suggest that alcohol-related road traffic risk is shaped not only by individual driver behaviour but also by interacting structural determinants, including alcohol control policies, enforcement practices, social norms, and transport conditions. The decline in population-level alcohol use prevalence observed during 2010–2022 provides complementary context for interpreting the observed trend.

From a policy standpoint, it is important to maintain macro-level regulations that influence alcohol availability and public visibility (e.g., sales hours, advertising/promotion restrictions, age limits) and to ensure consistent implementation. However, the decisive component for real-world impact is high-visibility, sustained, and time/place-targeted traffic enforcement. Reducing drink-driving requires attention not only to punitive sanctions but also to contextual determinants of behaviour; in particular, strengthening safe return options during night-time hours (e.g., night public transport, taxi/ride-share access, locally tailored mobility solutions) may offer realistic substitutes that reduce reliance on risky driving.

In addition, systematically engaging intermediary actors within alcohol-serving environments—such as service staff, venue managers, and local stakeholders—could provide a complementary public health strategy that reinforces regulation and enforcement through harm-reduction practices. Finally, the current legal BAC limit (0.05 g/dL) could be re-evaluated—together with strong enforcement and implementation capacity—in light of lower-risk thresholds adopted in some settings (e.g., ≤ 0.03 g/dL) as a potential option for further risk reduction. Overall, the observed decline may represent a meaningful potential public health gain; however, sustaining and consolidating this gain likely depends less on the mere presence of legislation than on a multi-component, context-sensitive road safety approach with strong implementation capacity that integrates behavioural and ecological dimensions.

Abbreviations

BAC

Blood alcohol concentration

RBT

Random breath testing

TurkStat

Turkish Statistical Institute

WHO

World Health Organization

RBS

Responsible Beverage Service

ERSO

European Road Safety Observatory

COVID-19

Coronavirus disease 2019

Authors’ contributions

ÖA: Conceptualization, Data curation, Investigation, Formal analysis, Visualization, Writing—original draft preparation, Writing—review & editing, and overseeing manuscript revision during the peer-review process. Oİ: Conceptualization, Data extraction, Data cleaning, Writing—original draft preparation, and participation in manuscript revision. OU: Investigation, Writing—review & editing, and participation in manuscript revision. SÇ: Investigation, Writing—review & editing, and participation in manuscript revision.FK: Investigation, Writing—review & editing, and participation in manuscript revision.All authors reviewed the manuscript and approved the final version.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The dataset analysed during the current study is publicly available from TurkStat Road Traffic Accident Statistics (press release No. 54056) and can be accessed at: [https://veriportali.tuik.gov.tr/tr/press/54056](https:/veriportali.tuik.gov.tr/tr/press/54056%20) (Accessed 2 Mar 2026).

Declarations

The manuscript submitted to this journal has not been previously published and is not under consideration for publication elsewhere.

Ethics approval and consent to participate

This study used only publicly available, anonymised secondary data (open-access statistics from the Turkish Statistical Institute [TurkStat]). No individual-level identifiable data were collected or analysed; therefore, ethics committee approval was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.