To the authors´ knowledge, this study is the first of this scale to identify SM prevalence and associated factors in the general EU population. SM prevalence varies significantly across scientific literature based on population characteristics and methodological differences. Nonetheless, the estimated prevalence found in our study, 34.3% during a two-week sample window, is comparable with previous studies featuring similar characteristics. A systematic review performed during the COVID-19 pandemic discovered that European SM prevalence was the lowest of all geographical regions at 40.8% [37], potentially attributable to greater access to prescribed medications [38]. In Germany, one 2015 study reported a seven-day OTC consumption prevalence of 46.3% [39], while a 2017 study reporting on the second wave of the EHIS found a two-week self-medication prevalence of 42.1% [40]. Similarly, a study performed in Paris, France in 2018 placed SM prevalence during the past four weeks at 53.5% [41]. In our study, prevalence varied substantially between countries.

Along the same lines, our multivariable analysis concludes that health clusters are the strongest determinant of SM in the EU. We discover that residents of Eastern Countries B are four times as likely to self-medicate (AOR = 4.00; 95%CI = 3.81–4.21) with respect to those of Southern countries. Previous research working with data from the first iteration of the EHIS found similar results among the elderly, noting the highest SM prevalence in Poland, part of Eastern Countries B cluster, and the lowest in Spain, a member of the Southern countries cluster [42]. Additional research also shows that Poland and Czechia, members of the Eastern Countries B block, display extremely high rates of SM at 59% and 85% respectively [5, 43]. This result emphasizes the substantial influence of localized EU health systems on population health behaviors. Specifically, health service provision, generation of health resources, and health financing clearly have a substantial impact on SM, even surpassing the impact of traditional socioeconomic, demographic, health, or lifestyle variables. As such, it highlights the need to reconsider traditional epidemiological determinants and underscores the importance of health systems in shaping our health practices. It is also worth noting that across the board, women are more significantly influenced to self-medicate by their country´s health system than men, providing yet another example of how the gender influences health.

In addition, our study identifies young adults (ages 25–44) as the most likely to self-medicate (AOR = 1.21; 95%CI = 1.12–1.31). These findings are congruent with prior research that found that SM is highest at younger ages and trends downward as age increases [4, 40, 44, 45]. Younger people tend to seek less medical attention, face less comorbidities, and perceive their health as better than those who are older [46, 47], which may make SM an appealing option versus prescription medications. A cross-sectional study in the United Kingdom highlighted that younger age is associated with increased risk for misuse of non-prescription medications, which also could contribute to SM [48]. Similarly, an Australian study by Vong et al. also found that younger individuals are less likely to follow directions for non-prescription medicines [6]. Coupled with the fact that younger people tend to care less about and seek less health information [49], these findings highlight the importance of finding new ways to educate young people about safe SM.

Consistent with findings from previous studies [5, 16, 39, 40, 45, 50, 51], our study shows that women are more likely to self-medicate than men (AOR = 1.74; 95%CI = 1.68–1.81). This finding is consistent across the EU, though the difference between men and women varies across countries. Research has cited barriers to health care, such as unmet needs stemming from long wait times [52] or socioeconomic disparities related to occupation, income, and education [53] as explanations for women´s greater propensity self-medicate. However, while these could be contributing factors, we find that the relationship between sex and SM continues even in multivariable models separating these variables, indicating that these are not the primary driving forces of this inequality. Instead, we find more convincing explanations surrounding women´s greater inclination toward self-care [42, 49]. A Finnish study by Ek on gender differences in health information highlighted that women are more engaged and informed health decision-makers, more frequently seeking health information from lay sources and displaying interest in health repercussions [49]. In the same vein, gender bias in the treatment of conditions and prescription of medicines is well-documented [54], and women could turn to SM to ease the burden of certain ailments. Men, on the other hand, are often encouraged to “tough it out” and avoid medication [55]. This contrast underscores the multidimensional nature of gender roles and the ways in which they manifest themselves in health habits. Regardless, in the variables collected in this study, the factors influencing SM in men and women are highly comparable.

Our study also reveals that immigrants born in other EU states have an increased likelihood to self-medicate (AOR = 1.16; 95%CI = 1.04–1.30) compared to native-born individuals. These findings align with previous research conducted in Spain which also found immigrant status to be a factor associated with SM [45]. Immigrants may lack the same supportive familiar system or access to health care as native-born residents, making SM an appealing tool. Alternatively, those with the resources to move to other EU member states may be particularly equipped with greater self-care consciousness. Notably, this finding was not statistically significant in those born outside the EU, which may be attributable to a lack of familiarity with or access to EU health services.

In line with previous research [51, 56], our study shows that residing in cities (AOR = 1.14; 95%CI = 1.09–1.19) is associated with an increased likelihood of self-medicating in comparison with rural living. This result may be attributable to greater access in urban areas to pharmacies or other stores selling SM products, or greater specialist use in cities versus general practitioners in rural areas, wherein responsibility for smaller health issues falls more upon the individual [56]. Pollution or other consequences of urban living could also be contributing factors, as a study by Davies et al. discovered a positive relationship between OTC medication sales and environmental contaminants found in cities [57].

In our multivariable model that encompasses both sexes, individuals who have completed higher education are 83% more likely to engage in SM than those without any formal education (AOR = 1.83; CI95%=1.60–2.09). This positive relationship is incremental between all levels of SM and education, as is common in health indicators [47]. These findings align with previous research, as a number of studies have identified a positive correlation between these two variables across diverse demographic groups [4, 5, 16, 40, 42, 44, 45, 51, 58]. More educated individuals may self-medicate more given their generally greater health literacy [59] and may consequently engage in greater self-care. Education level is also positively associated with social networking, which could lead to medication sharing and lay recommendations [52]. Concerningly, however, in an antibiotic study, more appropriate consumption behaviors were found in subjects with lower education levels, despite the fact that they were less health literate, because those with higher education were more likely to question their doctor´s orders [60]. Similar findings were yielded by Vong et al. in Australia who discovered reduced likelihood amongst the most educated to follow directions on non-prescription medicines [6]. In other words, education may grant excessive confidence in patient decision-making with respect to medications. This phenomenon may be specific to developed countries or the sphere of the EU, where a higher baseline level of health literacy and health care access exists, as a systematic review of antibiotic consumption found that while worldwide, improved education level is associated with significantly lower odds of misuse, in Europe, the opposite is true [59].

With respect to income, our findings align with existing literature: while a relationship exists between greater income and SM (AOR = 1.14; 95%CI = 1.07–1.21), the robust association often cited is exclusive to bivariate analysis, falling apart in multivariable analysis. On its surface, this result may appear surprising, as increased income improves access to non-reimbursed medicine [41]. Instead, we find that this relationship is mediated by other variables, notably education, employment status (AOR = 1.24; 95%CI = 1.12–1.37) and barriers to health care, including unmet needs due to inability to afford medical examination or treatment (AOR = 1.27; 95%CI = 1.12–1.42) and waiting lists (AOR = 1.38; 95%CI = 1.23–1.55). The latter has been identified as a substantial and common barrier to health care utilization [52, 53] as greater income permits the affordance of superior private health care plans, bypassing the public system waiting lists [41]. Put shortly, income becomes a more relevant factor when it becomes a barrier to accessing medical care. Our findings also suggest that women may be more likely to self-medicate in response to waiting lists (AOR: 1.43 vs. 1.32) while men may be motivated to self-medicate due to inability to afford medical examination or treatment (AOR: 1.38 vs. 1.20). A possible explanation is that waiting lists may be a less traversable barrier than inability to afford medical examination or treatment.

In the same vein, our results reveal that visits to general practitioners and family doctors (AOR = 1.21; 95%CI = 1.15–1.26), as well as medical or surgical specialists (AOR = 1.21; 95%CI = 1.17–1.26), are positively associated with SM. These findings are likely attributable to greater prophylaxis among those who receive regular medical care. Though some previous studies have logically linked lack of doctor visits to SM [45], this relationship may only be applicable where SM serves as a substitute for treatment by a medical professional due to health care barriers. Our study also finds that presence of a long-standing health problem serves as a risk factor for SM (AOR = 1.39; 95%CI = 1.33–1.45). Findings on the effects of chronic conditions on SM are mixed, as some subjects may have regular medical treatment and rely on prescription medications to manage their ailments, while others engage in SM due to experience in dealing with their illnesses [3, 41, 42, 45, 55, 61, 62]. A study performed in the greater Paris area by Vanhaesebrouck et al. failed to find a relationship between the two variables due to this phenomenon [41]. A UK general population study revealed that use of NPM with a long-standing illness is predictive of NPM abuse [48], making this risk factor particularly significant as it may reflect inappropriate SM.

Concerning lifestyle habits, alcohol consumption (AOR = 1.23; 95%CI = 1.19–1.28), smoking (AOR = 1.05; 95%CI = 1.01–1.10), and vaping (AOR = 1.19; 95%CI = 1.06–1.32) display positive associations with SM. Previous studies have linked alcohol use [42, 45] and vaping [63] to SM. With respect to smoking, an association with SM has been reported in women (but not men) [45], in adolescents [64], and for OTC analgesic use [3], while this study highlights a broader connection. Both smoking and alcohol use have ironically been found to be positively associated with self-perceived health, as ill individuals may abstain from such practices [46]. The same could be applicable to SM, as those capable of engaging in smoking, vaping, and alcohol drinking may be healthier individuals. Alternatively, SM may be more frequent for these individuals in reaction to the negative health consequences of said habits. It is also possible that this relationship is explained by overlapping motivators between SM and these habits, which in and of themselves could be considered ways of self-medicating. Additionally, our study reveals that greater physical activity also emerges as a factor associated with SM (AOR = 1.27; 95%CI = 1.22–1.32). This may reflect also self-care habits, as the employed concept of SM encapsulates more than non-prescription medicine, including products like vitamins and supplements, which are often favored by those who are physically active. A study of French adults using the same definition of SM also found a positive association between SM and physical activity, though the results were non-significant [31].

This study is subject to a series of limitations. The first limitation stems from the cross-sectional nature of the data, which does not allow us to establish causality. Second, the EHIS did not collect data regarding consumption of specific medication classes, which may display differing relationships with the variables. Future studies including this component could help to disentangle the heterogeneity of SM. Along the same lines, the EHIS made reference to binary sex, which may inadequately capture the range of diversity and social, political, and economic forces expressed in gender [65]. Furthermore, our study features an uneven sample size between countries, with several countries with low SM prevalences having amongst the largest sample sizes, such as Spain and Italy, which could have led to an underestimation of SM prevalence in the EU. Additionally, social desirability bias also could have led to underreporting on SM. This is furthered by the fact that several differences existed between countries in data collection methods and sampling design, which could have altered results and reduce comparability between countries [66]. In the case of this study, individuals interviewed face-to-face or over the phone may have felt additional social pressure to underreport self-medicating versus those who filled out self-administered written or online surveys.

Moreover, data was collected non-simultaneously across the EU, as surveys were carried out over multiple years and seasons, and consequently, seasonal and annual variance of conditions that provoke SM could have been impactful. Specifically, lower temperatures are associated with greater OTC respiratory medication sales [67], and as a result, SM prevalence could be greater in colder months. It is worth noting that the vast majority of responses were collected in autumn. Along the same lines, three countries (Germany, Spain, and Malta) collected data after the beginning of the COVID-19 pandemic, which could alter SM prevalence in either direction [68]. An additional limitation is that the self-reported data contained in this study is subject to recall bias. The two-week window of the dependent variable question should have reduced some of the bidirectional effects of recall bias, but can complicate SM prevalence comparisons with other studies. A German study pointed out that the short reference period and self-assessment method used in the EHIS may lead to prevalence estimates that differ substantially from epidemiological studies [47]. Finally, the non-response rate, ranging from 12 to 78% based on country [68], may also be influential, as those who refused to participate could have shared insights into SM, even if the direction of this effect is indeterminable.

Nonetheless, the quality report of the EHIS 3 details that the data underwent validation, calibration, and non-response adjustments procedures to minimize the effect of all potential sources of sampling and non-sampling errors, resulting in dataset that is highly harmonized and allows for a high degree of comparability across EU member states [68]. Coupled with the robust weighted sample size of 255,758, the authors feel that none of the aforementioned limitations should dampen the strength or relevancy of the findings and their applicability to the EU general population.