The present study, covering five regions in Sweden and, to the best of our knowledge, the largest study hitherto performed, confirms physicians’ need for drug information. It appears that physicians use multiple sources in the search for several drug-related issues, including, for example, dosing, adverse drug reactions, drug choices, and drug interactions. Furthermore, credibility, easy online access, familiarity, update frequency, and knowledge of origin or author seem to be crucial for their choice of drug information source. In addition, we found several personal factors associated with physicians’ choice of information sources.

Most of the physicians in our study searched for information about drugs at least weekly. Since drug dosing errors [5] and adverse drug reactions [6, 7] both are frequently occurring in health care, the finding of pharmacotherapy including dosing as the most common drug-related issue for searching and adverse effects as the next most common may be reassuring.

There was a similar pattern for physicians in primary care as for physicians active in other specialties with respect to the most commonly used sources for drug information, i.e., Pharmaceutical Specialties Inc. in Sweden as well as national online knowledge compilations. Our result that colleagues are an important source of information is consistent with two prior studies, both published in 2013 [8, 9]. However, asking a colleague was considerably less common in a study from Ireland, published in 2001, where 7% of general practitioners (GPs) and 29% of hospital doctors consulted colleagues for prescribing information [4].

In our study, the most conspicuous difference in the usage of sources of information between physicians active in primary care and physicians active in other specialties was that the former used information from the regional drug and therapeutic committee, e.g., regional prescribing guidelines, to a greater extent and PubMed and webpages of specialist associations to a smaller extent. These results may not be surprising. Indeed, regional prescribing guidelines are intended for physicians in primary care. A previous questionnaire study among 603 Swedish physicians in 2008 in primary care revealed that 97% used these guidelines [10], and in general, specialist associations do not focus on primary care.

Notably, about one-third of all physicians frequently use the search engine Google when searching for drug information. This is consistent with results from other studies that report high use of non-authoritative sources, including Google [11,12,13]. A descriptive survey including 444 randomly selected physicians in Nigeria, published in 2011, reported Google being the most frequently used search engine (73%), and PubMed being the most used medical database (70%) [11]. In a study from 2015 among medical residents in New Jersey, 56% reported using Google on mobile devices. However, that study reported that Google was primarily used to identify foreign drugs and only by 2.7% for questions regarding dosing [12]. The use of non-authoritative online information sources such as Google implies that the prescriber needs to be able to differentiate between reliable and non-reliable sources. In this context, it should be noted that a minority of our respondents had a PhD degree, and furthermore, a majority had no research experience at all. Nevertheless, 88% stated that credibility was the most important factor when choosing sources of drug information. This may imply that they find it important to evaluate reliability of information sources. This is consistent with a systematic review where credibility was one of five listed factors that could have significant effects on physicians’ choices of information sources [1].

In our study, there was a difference in who provides doctors with oral drug information at work. In the previously mentioned Irish study, the GPs reported consulting the pharmaceutical industry to a greater extent (42%) than the hospital doctors (18%) before prescribing a new drug [4]. In another previous Swedish study from 2011, about 85% of the GPs responding to a questionnaire reported getting too much information from the pharmaceutical industry [14]. In contrast, physicians in primary care in our study stated to a greater extent that the information was provided by representatives employed in the regions, while physicians in specialist care stated that the information came to a greater extent from the pharmaceutical companies. This could have implications for prescribing since a previous study published in 2010 showed that the presence of drug information from the pharmaceutical industry was negatively associated with adherence to the prescribing objectives [15]. It was also notable that about one-quarter of our respondents stated that they do not receive any information at all about drugs at their workplace, since prescribing drugs is an important professional activity for physicians that requires continuous updating.

In a study from Ethiopia published in 2013, respondents from a specialized hospital were more likely to consult drug information sources compared with respondents from primary care [16]. In our study, the relationship was reversed. Of physicians in primary care, 96% reported searching for information about drugs at least every week, compared with 83% for physicians active in specialties others than primary health care. This difference between our study and the previous one may reflect the extensive availability of drug information in primary care in a developed country compared to a developing country. Indeed, most of our respondents from primary care had searched for information over the last 6 months for the majority of the predefined groups of drugs. This seems logical since physicians in primary care meet patients with varying diseases, including older people with multiple morbidities.

Among physicians in primary care, we found that professional status as a junior doctor, as well as younger age, was associated with more frequent use of drug information in general. This finding is in line with a study from 2012 in which 721 French GPs answered a questionnaire about using the Internet for clinical information [17]. In addition, professional status as a junior doctor and younger age were, in our study, also associated with advice taken from colleagues. Interestingly, female primary care physicians more often used colleagues as a source of information and also found sources recommended by colleagues more reliable. This result is consistent with a previous study from Denmark, published in 2016, where young and female GPs were more likely to seek information from colleagues [18]. Unexpectedly, male physicians in primary care sought a greater extent of information about drugs in pregnancy compared with their female counterparts. Finally, our results of more extensive use of PubMed for physicians in primary care with research experience seem reasonable.

A large number of respondents is a strength of this study. However, the low response rate is an apparent limitation, affecting the generalizability of the results. In general, however, the response rate tends to be low in questionnaire studies of physicians [19]. To maximize the response rate, we kept the number of questions limited, used predefined answers, and sent several reminders. On the other hand, predefined multiple-choice answers may be considered a limitation. Although they were phrased by knowledgeable researchers, it cannot be excluded that relevant aspects were left out. The free-text response options may, to some extent, compensate for this issue, but the threshold to respond to such questions is probably higher. Although performed in five Swedish regions, it may also be regarded as a limitation that our study was restricted to Sweden. Indeed, international comparisons could add valuable insights.