Cross-sectional study using a self-administered questionnaire

A total of 256 participants completed the questionnaire (response rate 82%). Theoretical knowledge was good (mean score of 6 ± 1.3 on 7 questions) in contrast to poor awareness (< 33%) of local AMR rates of key-pathogens. Participants strongly agreed that AMR is a problem worldwide (70%) and in Peru (65%), but less in their own practice (22%). AM overuse was perceived both for the community (96%) and the hospital settings (90%). Patients' pressure to prescribing AMs was considered as contributing to AM overuse in the community (72%) more than in the hospital setting (50%). Confidence among AM prescribing was higher among attending physicians (82%) compared to residents (30%, p < 0.001%). Sources of information considered as very useful/useful included pocket-based AM prescribing guidelines (69%) and internet sources (62%). Fifty seven percent of participants regarded AMs in their hospitals to be of poor quality. Participants requested more AM prescribing educational programs (96%) and local AM guidelines (92%).

A self-administered questionnaire was distributed in both hospitals among residents (i.e. physicians in training) and attending physicians (i.e. staff physicians after completion of training and specialization). Medical doctors from psychiatry, radiology, ophthalmology and anaesthesiology were not included as they do not routinely prescribe AMs. Questionnaires were distributed on site during working hours and participants were asked to respond immediately. There was no incentive for subjects to participate and no reminders were supplied. The questionnaire content was based on a previous survey described in the U.S. and adapted to the Peruvian system [9]. Prior to release, it was reviewed by a team of six Peruvian infectious diseases physicians to assess the relevance and wording of the questions as well as accuracy of the translation into Spanish. The 38-item questionnaire addressed the professional profile of the participants and frequency of AM prescription (5 questions), their awareness about the current scope of AMR (6 questions), sources of information and continuing education about AMs (2 questions), confidence and seeking inputs (5 questions), factors influencing decisions around AM prescription (5 questions) and the acceptability and appropriateness of potential interventions (6 questions) (Additional file 1). Questions used a 4 or 5-point Likert scale (which included answers ranging from "strongly agree" to "strongly disagree", from "very useful" to "not useful at all" and from "always" to "never"). The survey also included seven questions that assessed basic knowledge about the clinical indications, spectrum, administration and pharmacology of AMs. Three case-based questions addressed the choice of AMs for treating acute diarrhoea, an upper respiratory tract infection and sepsis in a patient with impaired renal function; one question addressed safety of AMs during pregnancy, and three questions addressed the spectrum of AMs and their ability to cross the blood-brain barrier. Finally, in order to evaluate physician awareness about AMR rates within local hospitals, participants were asked to estimate the proportion of Klebsiella pneumoniae resistance to cephalosporins and Pseudomonas aeruginosa resistance to ciprofloxacin (answer options "20% or less", "20%-50%", "more than 50%" or "don't know"). The true rate was obtained from a surveillance study on AMR in Lima hospitals in 2008.

The study was approved by the Institutional Review Board from Universidad Peruana Cayetano Heredia, Lima, Peru, and by the Ethical Committees in each hospital. Based on the anonymous nature of the collected data, informed consent form was not taken.

The average score to the questions regarding knowledge of AMs was 6 out of 7(SD ± 1.3). For the case-based questions about acute diarrhoea and upper respiratory tract infection, the vast majority of participants agreed that there was no need to start an AM (238, 93% and 194, 76%, respectively). The knowledge about the need to reduce the dose of AM in a patient with severe renal impairment was assessed by presenting a sepsis case where ceftriaxone and gentamicin were prescribed. About three quarters (n = 194, 76%) correctly identified that AMs would need to be reduced in this case. Furthermore, nearly all participants (n = 250, 99%) correctly replied that metronidazole has activity against anaerobes and 213 (83%) participants correctly answered that methicillin resistant Staphylococcus aureus (MRSA) is not susceptible to cephalosporins, the remaining participants (n = 41, 16%) incorrectly responded that it is susceptible to cefalotine, cefuroxime or ceftriaxone. The majority (n = 237, 93%) of participants agreed that amoxicillin is safe during the first three-month period of pregnancy whereas 17 (7%) incorrectly answered that ciprofloxacin or gentamicin are safe. A total of 180 (70%) participants correctly answered that ceftriaxone is the most effective drug crossing the blood-brain barrier where as 62 (24%) and 10 (4%) of participants incorrectly chose vancomycin and clindamycin above ceftriaxone. With regard to the estimate about local AMR rates, it was striking that only 51 (20%) of participants correctly estimated that > 50% of K. pneumoniae isolates are resistant to cephalosporins, whilst half, 129 (50%) answered that the resistance rate was 20%-50% and 47 (18%) answered 'don't know'. In response to the question about resistance rates of P. aeruginosa to ciprofloxacin, 82 (32%) of participants gave correct estimates (i.e. 20-50%), 118 (46%) answered that the rate was higher than 50% and 39 (15%) answered 'don't know'.

Nearly half (63/135, 47%) of residents revealed they were very confident about the optimal use of antimicrobials compared to 99/121 (82%) of attending physicians (p < 0.001). A total of 78 (31%) participants agreed that it is difficult for them to select the correct AM, this was recorded for 36% participants from the medical departments versus 20% from the surgical departments (p = 0.014). Moreover, it should be noted that almost a quarter of participants (n = 58, 23%), strongly agreed and agreed that prescribing AMs when they are not required does not cause any harm. With regard to seeking inputs, when participants were asked about the frequency of reviewing their decision to prescribe AMs with a senior colleague, 15% replied 'never' and 57% 'sometimes'; only 6% answered 'always'. More than half (74/135, 55%) of residents declared that they never or only sometimes reviewed their decision with a senior colleague compared to 89% (108/121) of attending physicians (p < 0.001). This was seen more frequently among participants from surgical departments compared to those from medical departments (80% versus 67%, p = 0.03). Among the 219 participants who declared to review their decision to prescribe AMs with a senior colleague at least sometimes, nearly three quarters (161, 74%) reported that senior colleagues sometimes recommended a different AM.

Overall, 88 (34%) participants declared that there had been no lectures about AM use as part of academic activities within their departments during the previous year, although there was a slight difference between the medical and surgical departments (29% versus 45%; p = 0.015). Likewise, 37% (95/256) of participants had not participated in a course on AM use during the previous year; the rate was 65% among residents versus 35% among attending physicians (p = 0.003). Regarding sources of information, two-thirds (173, 68%) of participants reported having readily available sources of information on AMs. The "Sanford Guide on Antimicrobial Therapy" was considered as a very useful source (n = 129, 50%), although preferentially among residents (n = 78, 58%) compared to attending physicians (n = 51, 42%, p = 0.013). Internet sources were considered as very useful or useful by nearly two-thirds (159, 62%) of participants. Thirty six (14%) participants did not consider the national guidelines useful and a quarter (65, 25%) noted that they were not familiar with these guidelines. Advice from colleagues of higher rank or same rank were considered useful or very useful in 98 (38%) and 71 (28%) of participants respectively.

Overall, the theoretical knowledge about AMs including indications, administration and side effects ranged from very good to excellent. Despite this apparently good score for these questions, it should be noted that a quarter of participants still considered that it was correct to use AMs for upper respiratory tract infections. This suggests that this issue should be targeted in future educational interventions. Furthermore, it is known that in practice AM use may not reflect these results, and this can be illustrated by a recent study: in a rural Peruvian village, 58% of children with acute upper respiratory symptoms or watery diarrhoea (for which AMs are not recommended) were given AMs when they went to see a doctor [12]. This contrasts with the survey's results, in which the majority of participants answered correctly that AMs are unnecessary for either of these conditions. Likewise, it should be noted that a quarter of participants agreed with the statement that unnecessary prescribing of AMs does not cause any harm. Although the participants' overall knowledge about AMs was appropriate, most of them incorrectly estimated the local resistance rates of two key-pathogens in the hospital setting, K. pneumoniae and P. aeruginosa. Similar findings have been described in other studies [7,8]: Pulcini et al. showed that only 16% of young doctors in a French hospital knew the actual proportion of community acquired-Escherichia coli resistant to fluoroquinolones. Local microbiology laboratories are encouraged to maintain a database about the levels of resistance of key pathogens and diffuse it to prescribers: when reinforced by the local antibiotic committee, information may orient prescribing doctors [13].

The awareness of AMR as a worldwide and national problem was very high among the participants. However, in contrast, AMR was much less recognised as a problem in participants' own practice. This trend has also been observed among physicians surveyed in the U.S. (Table ​(Table1)1) [9-11]. On the other hand, qualitative research among general practitioners in the U.S. showed that most of the physicians interviewed were aware that inappropriate use of AM in their own practice contributes to increasing AMR [14]. Interestingly, the majority of participants recognized excessive use of AMs as a factor contributing to AMR in the community, but only half did so for the hospital settings.

Compared to attending physicians, residents in training were less confident about AM prescribing. This correlates with the findings of Srinivasan et al.: in this study, senior residents were more confident about optimal use of antimicrobials compared with first year-residents [9]. Moreover, residents tended to seek advice from their senior colleagues when prescribing, irrespective of their specialization (department) or hospital affiliation, compared with attending physicians who have more years of clinical experience. However, more residents declared that they consulted internet-based sources rather than approaching senior colleagues for advice. It is surprising than more than 50% of residents declared that they did not consult senior colleagues considering that both institutions were teaching hospitals. Other sources of AM guidance are discussed below.

The present survey also revealed information about the sources of information for AM use. The popularity of the Sanford Guide illustrates the accessibility of pocket-based treatment guidelines. Internet sources were ranked as the second most useful source. In this scenario, distant learning technologies which have been used successfully in Peru for other disciplines [15,16] may have a place in promoting educational AM prescribing programs. The poor appreciation of and familiarity with the national guidelines among the participants is striking and contrasts with the seemingly large demand for local AM guidelines.