Reducing Expectations for Antibiotics in Patients With Upper Respiratory Tract Infections

A Primary Care Randomized Controlled Trial

Anna Ishani Perera, MBChB; Mark Greenslade Thomas, MD; Keith James Petrie, PhD; Janet Frater, MBChB; Daniel Dang, PhD; Kiralee Ruth Schache, MHP; Amelia Frances Akroyd, MHP; Stephen Robert Ritchie, PhD


Ann Fam Med. 2021;19(3):232-239. 

In This Article


The study was a randomized controlled trial that compared the effects of viewing 2 brief tablet-based educational presentations, designed to reduce expectations for antibiotics in patients with URTIs, with the effects of viewing a control presentation. Participants were recruited from 2 large urban family practices in Auckland, during the winter (July to September) of 2018. The study was conducted in accordance with the Declaration of Helsinki, and the study protocol was granted approval by the Health and Disability Ethics Committee of New Zealand (Ref 17/STH/144) and was registered with the Australia New Zealand Clinical Trials Registry (Trial ID 12617000744358).


Patients whose main reason for presentation was symptoms of URTI—cough, earache, sore throat, or nasal congestion (or some combination thereof)—were invited to participate. Based on our previous findings,[15] we estimated that a behavioral intervention would have a small to moderate effect (Cohen d = 0.35) on patients' expectations for antibiotics. Combined with a power of .80 and an α level of .05, a total sample of 231 was required to detect this effect. Participant attrition at the postconsultation follow-up was estimated to be about 10%; therefore, we aimed to recruit at least 255 total participants (ie, 85 in each group).

In the family practitioner waiting room, research assistants provided participants brief information that their practice was participating in a University of Auckland study. They were not told that the aim of the research was to measure the impact of the various presentations on patients' expectations to receive antibiotics. Adult patients (aged ≥18 years) presenting with symptoms of URTI, and parents of children aged 0 to 7 years presenting with symptoms of URTI, were eligible to participate. Potential participants were excluded if they were attending an appointment with a nurse (there were no nurse prescribers at either practice) or were unable to read, write, or speak English.


Prepresentation Questionnaire. After providing written informed consent, each participant completed a questionnaire. Data collected included participant age, sex, level of education, and self-reported ethnicity; their symptoms (any of cough, sore throat, nasal discharge, or earache); and whether the visit to the family practitioner was for themselves or for their child. Participants used a 7-point Likert scale to indicate their level of agreement with the following statements/questions: (1) "How bad is the cold/flu that you/your child has today?" (to measure perceived illness severity); (2) "How worried are you about your/your child's cold/flu?" (to measure concerns about the illness); (3) "I think antibiotics are a helpful treatment for cold/flu" (to measure belief in the efficacy of antibiotics for URTIs); and (4) "I wish to receive antibiotics for my/my child's cold/flu" (to measure expectation for antibiotics). Response options ranged from 1 (strongly disagree) to 7 (strongly agree).

Presentation. After completing the prepresentation questionnaire, participants were randomized to view 1 of 3 presentations on a tablet device. Participants were allocated 1:1:1 without adjustment for prior allocation within the tablet application. Each presentation contained 6 slides, and the entire presentation took approximately 1 minute to complete. One presentation focused on the futility of antibiotics for URTIs and contained information about alternative treatments (eg, nasal spray, throat lozenges). The second presentation focused on the potential adverse effects of antibiotics and also contained information about alternative treatments for URTIs. The third (control) presentation promoted healthy lifestyle choices but did not mention antibiotics or URTIs.

Postpresentation Questionnaire. Immediately after the presentation, participants again used a 7-point Likert scale to indicate their level of agreement with the 2 previous statements addressing belief in efficacy and expectations: (1) "I think antibiotics are a helpful treatment for cold/flu," and (2) "I wish to receive antibiotics for my/my child's cold/flu." Participants then proceeded to their family practitioner consultation. Although the physicians consulting at the practices were aware of the study, they were not informed by the research team whether individual patients had consented to participate or which presentation participating patients had viewed.

Postconsultation. Immediately after their visit, participants reported whether they had been prescribed an antibiotic and their level of satisfaction (on a scale of 0 to 5) with the consultation. To collect information on whether patients had a prescription dispensed and whether patients may have been prescribed antibiotics by another clinician, all medications dispensed during the 7 days after the study visit were obtained from the National Pharmaceutical Collection database using each participant's unique National Health Index number.


We performed analyses using SPSS 25 (IBM Corp). The primary outcome was the change in participants' expectations of receiving antibiotics for their URTI after viewing a presentation. Changes in each participant's expectations to receive an antibiotic and in beliefs in the efficacy of antibiotics for URTIs for each group were compared using the Kruskal-Wallis test. An estimate of the effect size was determined by calculating the Cohen d from the Kruskal-Wallis H statistic. Factors associated with continuing to expect antibiotics after viewing a presentation were identified using multivariate ordinal logistic regression analysis; Likert scores were included as categorical variables.

We used using binomial logistic regression analyses to identify the factors associated with receiving an antibiotic prescription and with having an antibiotic dispensed during the 7 days after the consultation. The multivariate models included factors significantly associated with antibiotic prescription or dispensing after univariate binomial regression analysis, and included the presentation viewed as a factor of interest.