The sample was obtained from a questionnaire implemented between 02 November 2021 and 18 February 2022 in users from primary healthcare units during one of the pandemic peaks (see Supplementary Graph SM1). These 147 health units are located in the Centre Region of Portugal. The response rate was 99.7%, with 7,126 valid answers. The questionnaire aimed to monitor people’s satisfaction towards the services provided by primary healthcare units. It is based on the measurement instrument EuroPeP [19, 20] and validated for Portugal.

Additionally, some other questions were asked to gather information for scientific purposes. For instance, it was asked about unmet healthcare needs and taking drugs following the format used in other international surveys, such as the European Health Survey and the Survey of Health Ageing and Retirement in Europe (SHARE). The Regional Health Authority Directive Board and Ethics Committee approved questionnaire’s implementation.

We started to perform a descriptive analysis of the data. Then we estimated a logistic regression for all the sample and separately for women and men because empirical evidence supports sex differences concerning healthcare use [21–23]. We also performed i) the VIF test to check multicollinearity across independent variables, ii) the linktest for testing specification errors, and iii) the Hosmer-Lemeshow test for model goodness-of-fit. Results are reported using odds-ratio (OR), 95% confidence intervals (CI) and the correspondent p-values. We computed predictive probabilities and margin effects for the results concerning the new effects in this analysis, specifically those resulting from taking prescribed drugs and attending a remote medical appointment.

All the analyses presented in this work were performed in STATA 15.

i. The probability that a person takes OTC is about 19.6% (p-value < 0.001), given that all the independent variables are set at their mean value.

So, the marginal effect of taking prescribed drugs (iv) is larger than that of attending a remote appointment (v).

The key findings of this work point, firstly, people already taking prescribed medicines or attending remote medical appointments are more likely to take non-prescribed drugs. Secondly, unmet healthcare needs seem to be a driver for taking non-prescribed drugs. Finally, when stratifying the sample by sex, some statistically significant results for women are not equally found for men.

This work’s first major significant finding is that people who attended a remote medical appointment or were taking prescribed drugs presented a higher likelihood of taking non-prescribed drugs. And the increased probability of taking an OTC drug is slightly higher for people already taking prescribed medications than for those who attended a remote appointment. These findings are new evidence of how people behave under a constraint situation of a pandemic.

It is worth, though, analysing if this behaviour continues or will change in the future. In the public health crisis scenario of COVID-19, people were being constrained to stay at home or to work under strict conditions. Without the human contact of health professionals and given prescribed drugs needed (and prescriptions were often obtained from an online or digital prescription), the pharmacist was the closest human contact people could afford, which could motivate people to take OTC drugs [10, 24].

The second major finding concerns to the role of unmet healthcare needs as a driver for taking OTC drugs; this result was also found before the pandemic in Portugal [7]. The results previously found showed that people were more likely to take non-prescribed drugs if they could not access healthcare for financial constraints and excess distance. However, people on waiting lists were less likely to choose OTC.

The evidence obtained in this research presents a slightly different pattern. People who need healthcare but cannot get an appointment, being on a waiting list, and people who have an appointment postponed, forced to be on hold, are motivated to take OTC. So, during the pandemic, when healthcare units struggled to deal with the public health crisis and the access to healthcare was difficult, people seemed to be substituting doctors for pharmacists and choosing to take non-prescribed drugs. Although this is a reasonable choice during the pandemic, the learning process may result in a permanent substitution between doctors and pharmacists. This scenario calls for future research to understand the dynamics of people choosing to take OTC drugs. On the other hand, another concern for this situation is that the absence of healthcare and the masking of symptoms with these drugs may delay a diagnosis or worsens health conditions, which could be prevented. This is one of the major problems raised with the pandemic and the incapacity of response of the health system to attend to all needs at that time.

Thirdly, as found in previous empirical studies [25–27], women or people with higher education are more likely to take OTC medicines. Despite the absence of information concerning people’s income, the level of education may be a good proxy for the available income so that higher education can be associated with higher income. People with higher income are also more likely to buy non-prescribed drugs. People with higher education levels are more endowed with personal resources (cognitive, informative and relational) that may empower people in their health [28–32].

On the other hand, people who are either employed or unemployed are less likely to take OTC drugs. Given the pandemic context and lockdown measures, it is reasonable to expect that people, no matter their labour status, may adopt identical behaviours concerning self-medication. However, the reasons supporting these behaviours may be different. For example, while employed people, who work at home or in essential services, fear to mask COVID-19 illness; unemployed people may be financially constrained to afford to buy non-prescribed drugs.

Finally, we refer to the sex differences found in our estimations [33]. Men are more influenced by predisposing factors. While education positively contributes to OTC drug consumption, single, divorced or widowed men are less likely to buy those medicines compared to married men. Women seem less associated with these factors, while education and divorce are the only motivating factors to take OTC. Conversely, to men, women are more prone to be influenced by enabling factors. While employed and unemployed women take OTC; the same association cannot be found for men. One possible explanation for this evidence comes from the learning experience related to accessing and using pharmacy services that women have from buying contraceptive pills [7, 34].

The major strength of this work is the additional contribution given to previous research exploring the drivers of OTC drug buying in European countries.

Some limitations, though, are to be addressed. First, the questionnaire asks if people took drugs that a physician did not prescribe. This question may have a biased interpretation. People may consider taking medications from the bathroom cabinet that were prescribed long before or medicines prescribed for a family member or a friend as possible answers.

Second, the questionnaire question focuses on taking OTC drugs 2 weeks before. Given that questionnaires were applied during the winter, it could be that people were more likely to take flu-related OTC drugs. However, if it were implemented during the spring or summer, people would be taking allergy-related drugs. In this way, we do not expect a strong seasonal bias in collected data.

Third, data were collected inside primary healthcare units in the Centre Region of Portugal. It may be argued that this sample may not represent the Portuguese population. Nevertheless, the main differences between those two geographic spaces are in the ageing index and the unemployment rate. While Portugal has an ageing index of 184.6, in Centre Region it is equal to 228.6, and the unemployment rate is slightly lower in Centre Region (6.6% in Portugal and 5.8% in Centre Region) [35]. However, age was not found to be a driver of OTC, and both labour conditions of employment and unemployment were found to be statistically significant, so the results found for Centre Region may not be so different from those that would be found for Portugal. In the Supplementary Table SM1, we present the different descriptive statistics across the previous study [7] and Census 2021 for Centre Region and Portugal.

Finally, the analysis performed in this work cannot be interpreted as causal relationships. The analysis provides correlations and predicted probabilities related to taking OTC by people. For a causal analysis, different data and methods would be needed.

The policy implications that arise from our results are mainly two-fold. Firstly, the concern for the substitution that people seem to be making between the absence of access to healthcare and taking non-prescribed drugs for self-medicating. This substitutive decision raises health risks resulting from wrong decisions on self-care for what seems a minor illness.

Secondly, the complementarity between remote appointments, specially those for requesting prescriptions or having a medical consultation and taking OTC drugs. While a certain value is added from this sort of non-presential appointment, there is an issue concerning human relationships. Further research needs to be followed to understand to what extent the absence of direct or close human contact may contribute to increased health empowerment perception and overconfidence by people, which may not be advantageous for a timely correct diagnosis.

Introducing remote medical appointments seems to be a new driver for people taking non-prescribed drugs. Further studies need to be conducted to monitor this behaviour and understand the pharmacist’s role in this choice. Additionally, the prevalence of unmet healthcare needs may increase the likelihood of people choosing OTC drugs, which may not be cost-effective from a societal and individual perspective. Therefore, one future policy measure that significantly impacts OTC drugs is the reduction of unmet healthcare.