Primary care clinicians’ perceptions about antibiotic prescribing for acute bronchitis: a qualitative study

Thirteen primary care clinicians – 12 medical doctors and 1 nurse practitioner – were recruited from 3 Brigham and Women’s Primary Care Practice-Based Research Network-clinics in Boston. The clinics consisted of 1 hospital-based practice and 2 urban community health centers with approximately 175 clinicians which all serve a socioeconomically, racially, and ethnically diverse patient population. Clinicians were recruited via emails sent by study staff to practice medical directors of the 3 clinics who, in turn, forwarded the request to primary care clinicians. If clinicians expressed interest in participating, study staff contacted the clinicians and obtained consent to participate. Interviews took place during March 2011. Clinicians were internal medicine physicians or internal medicine nurse practitioners, 69% female, with a mean age of 43, and an average of 14 years of experience. Clinicians received US$20 for their participation. The Brigham and Women’s Human Research Committee approved the study protocol.

The interview guide was developed as part of an iterative process between our team and a consulting Ph.D.-level, qualitative researcher with extensive training and 12 years of research experience in mixed methods design, field methods, and analysis. We used conceptual models of guideline adherence [12] and appropriate antibiotic prescribing [13] to develop the interview guide, ensuring that a range of concepts would be included. The guide focused on the following goals and domains (Appendix 1):

Interviews were conducted by a medical anthropologist with a Masters Degree in Applied Socio-Cultural Anthropology and 17 years experience in conducting qualitative research interviews. Interviews were completed over the phone and lasted between 30 and 60 minutes. The interviewer obtained IRB-approved verbal informed consent to conduct and record the interview as well as use the content of the interview for analysis and publication. The interviewer generally followed the interview guide, asked open-ended questions, but also improvised questions to elicit additional responses or clarify prior responses. If the participant did not offer solutions for domain 3, the interviewer mentioned specific potential interventions included in the interview guide. We conducted 13 interviews to collect general assessments of attitudes regarding acute bronchitis, reasons for antibiotic prescribing, and solutions to reduce antibiotic prescribing for acute bronchitis. Preliminary analysis of the interviews suggested we were approaching thematic saturation after 12 interviews.

Each interview was recorded, transcribed verbatim, and analyzed according to a conventional comprehensive qualitative analysis method [14]-[17]. We used a two-stage coding process: structural coding (Level 1) and thematic coding (Level 2). Structural coding followed the structure of the interview guide: every question received a structural code that was applied to the appropriate text. Thematic coding was based on themes that arose from the structural coding, and was applied in a second-pass analysis. Thematic analysis was inductive and followed the structure of the interview. The thematic content analysis was at the question level, such that themes emerged from the questions. We resolved differences in interpretation through discussion. We used NVivo (Version 8, QSR International) to code, categorize, search, retrieve, attach analytical memos and create conceptual relationship networks in our textual data that had been taxonomically coded. Once the 2-stage coding process was completed and reviewed by the entire research team, we generated a comprehensive thematic analysis summary report, including exemplary quotes. The analysis was performed and the summary report was written and delivered by the same experienced qualitative researcher that completed the interviews.

In this manuscript, we organized themes by decreasing number of participants who discussed each theme. While this provides quantitative impression of the frequency with which participants discussed each theme, we do not mean to imply that themes with a higher number of discussants are necessarily more important.

The consistent theme that emerged regarding guidelines and antibiotic prescribing for acute bronchitis was that all participants agreed with guidelines stating that antibiotics are not indicated for acute bronchitis and clinicians felt like antibiotics for acute bronchitis were overprescribed by other clinicians, but not by themselves. One participant stated the guidelines were,

Well established guidelines allowed clinicians to avoid prescribing antibiotics by having an authoritative norm that states antibiotics are not necessary. Despite the availability of guidelines, clinicians felt that other clinicians overused antibiotics.

Prior qualitative studies of clinicians’ attitudes towards antibiotic prescribing for respiratory infections were mainly done in Europe or in the United States over 10 years ago. Two recent systematic reviews identified 45 individual studies that examined clinicians’ views of antibiotic prescribing for respiratory infections [10],[11]. Only 3 reviewed studies examined prescribing for respiratory infections in the United States in the past decade [18]-[20]. Mangione-Smith and colleagues found that physicians incorrectly perceive patient questioning about whether or not they need antibiotics as indicating a desire for antibiotics, when in fact questioning was not associated with patient desire for antibiotics [18]. Ong and colleagues found that physicians correctly identified only 27% of patients who wanted antibiotics, but that physician perception of patient desire for antibiotics was the main driver of antibiotic prescribing [19]. Hart and colleagues found that physicians try to balance their own individual “best practice” with perceived patient satisfaction by using education, negotiation, holding-firm, or giving-in [20].

Not included in the systematic reviews, was one newer qualitative study by Ackerman and colleagues [21]. Ackerman and colleagues, as part of a randomized controlled trial of print and electronic-delivered decision support to encourage judicious antibiotic prescribing, found improved awareness of antibiotic resistance, but clinicians still cited patient expectations, time pressure, and diagnostic uncertainty as barriers to judicious antibiotic prescribing [22].

Some of the many qualitative studies from Europe highlight similar themes from our study and other studies regarding antibiotic prescribing for acute cough. These studies describe non-medical reasons for antibiotic prescribing such as the physician having previously missed a diagnosis of pneumonia in a different patient, patient expectations, and maintaining patient satisfaction [23]-[26]. European studies describe clinical reasons for prescribing antibiotics such as diagnostic uncertainty, fear of making an error of omission (i.e., not prescribing an antibiotic when it was indicated), and reliance on certain clinical factors like lung findings, fever, shortness of breath, sputum production, or crackles [26],[27]. One multinational European study by Brookes-Howell and colleagues highlighted international variations in guidelines, systems to reduce patients’ expectation for antibiotics, and availability of antibiotics without a prescription [25]. Recently, a study by Mustafa and colleagues found that Welsh primary care physicians, rather than directly asking about patient desire for antibiotics, preferred to use a “running commentary” during the physical examination – reviewing findings and implications with the patient as they are examined to reassure, share information, and strategically educate – to set patient expectations that an antibiotic prescription was not necessary without appearing curt or dismissive [18],[28],[29].

Our qualitative study has several limitations. First, clinicians were academically-affiliated and those who agreed to participate may have a particular interest in judicious antibiotic prescribing or be less likely to prescribe antibiotics. Second, participants may have been constrained by the content of the interview guide or given socially desirable answers. However, to avoid topic constraint or social desirability, the interviewer asked open-ended questions and encouraged participants to speak broadly. Third, our sample size was relatively small. However, the low number of emergent themes suggests we approached thematic saturation. Fourth, many of the themes discussed were about participants’ views on other clinicians’ attitudes and behavior, which may or may not reflect other clinicians’ actual attitudes and behavior.

If the patient has an acute cough, is not immunosuppressed, does not have a concomitant alternative diagnosis, has normal vital signs and a normal lung exam, antibiotics are not indicated [1]. Despite clinicians’ perception that patients are only interested in antibiotics, other studies have shown that clinicians overestimate this desire and poorly predict which patients want antibiotics [19],[30],[31]. Clinicians should be educated that antibiotic prescribing is, at most, marginally associated with patient satisfaction [19],[31]-[33]. Future guidelines should address the role of explicit and perceived patient demand in antibiotic prescribing.

Ongoing education and patient-directed materials such as handouts, posters, and over-the-counter prescription pads may further decrease perceived patient demand for antibiotics [34]. Outside of clinic visits, as suggested by our study, improved quality reports – although prior studies about monitoring and feedback are conflicting [35]-[37] – and pre-visit interventions should continue to be evaluated to decrease inappropriate antibiotic prescribing for acute bronchitis [38]. Accurate point of care tests, which were not addressed in our study, might be acceptable to clinicians and patients [39],[40].

Clinicians should feel secure in their decision that antibiotics are not indicated, should not assume patients want antibiotics, and consider using the “running commentary” method to educate and reassure as they do the physical examination [28],[29],[41]. Additionally, as our participants pointed out and as previously described, clinicians might consider using a term other than “acute bronchitis”, like a “chest cold”, which is accurate, but less associated with patient expectation for antibiotics [42],[43]. Rather than focusing on antibiotics, clinicians should focus on symptomatic treatment and educating patients at the time of the visit about realistic expectations for the duration of cough, which lasts, on average, three weeks [18],[44].

Questions for Clinicians’ Understanding of Antibiotic Treatment for Acute Cough in Primary Care