This study was a cluster randomised, factorial, controlled trial carried out in the city of Barcelona, Catalonia (NCT registry, ID: NCT03931577).

Patients who fulfilled the inclusion criteria were given written and verbal information about the study and were asked to provide written informed consent. Patients older than 18 years with a first consultation for acute cough (new cough or worsening of a previous cough) of up to 3 weeks’ duration as the predominant symptom, which the clinician believed to be an infectious LRTI, were recruited. Patients with a working diagnosis of a non-infective disorder, such as heart failure, pulmonary embolus, or oesophageal reflux, use of antibiotics in the previous two weeks, hospitalisation due to an acute LRTI, immunological deficiencies, inability to provide informed consent and/or unable to follow the study procedures were all excluded. Pneumonia and acute exacerbations of chronic obstructive pulmonary disease (COPD) were not deemed as exclusion criteria, as they were included in the definition of LRTI. The four groups allocated were (a) healthcare professional training in enhanced communication skills, (b) training in the use of a point-of-care CRP test testing, (c) combined training in CRP testing and enhanced communication skills, and (d) usual care. Central ethical approval has been obtained from the Research Ethics Committee IDIAP Jordi Gol (reference approval no. P18/227).

Eligible primary care centres were randomised into the four groups of 5 practices per arm, stratified by two variables: (a) socioeconomic level based on the 2015 socio-demographic index issued by the Catalan government20, and (b) baseline daily defined dose of the systemic antibiotic prescribing rate of the different participating primary care centres (J01 therapeutic subgroup of the Anatomical Therapeutic Chemical Classification System) corresponding to the year 2018. Randomisation of primary care centres was achieved by computer generation of random numbers. A total of 20 primary care health centres of the Barcelona region were included and at least 3 general practitioners and 3 nurses participated in each practice. Every director from each practice provided their approval to take part in the study before starting it.

The primary outcome measure was double: (a) antibiotic consumption within the first six weeks, as documented in the case report form and double-checked by on the electronic medical history, and (b) the variation in quality-adjusted life years, which was collected by means of the EQ-5D-5L questionnaire and the visual analogue scale within the two first weeks. Although most studies consider a timeframe of four weeks for the first co-primary variable, we decided to extend this time to up to six weeks as 21% of patients with LRTI still have cough after this period21. Moreover, we investigated the difference in the overall antibiotic prescribing rates (anatomical therapeutic chemical classification J01) in the 20 centres participating in the winter season when the clinical trial was performed (November 2019 to February 2020) compared to the same period one year earlier. We differentiated the common antibiotic families most prescribed for LRTIs: J01C (β-lactams), J01D (cephalosporins), J01F (macrolides) and J01M (quinolones). Secondary outcomes are described in the Supplementary Material.

Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at Fundació Institut Universitari per a la recerca a l’Atenció Primària de Salut Jordi Gol i Gurina (IDIAPJGol). REDCap is a secure, web-based software for research data capture, offering intuitive interfaces for data entry, audit trails for tracking, automated export, and data integration22,23. Only site investigator teams had data access and editing rights.

The interventions and study methods are described in detail elsewhere24. Once the centre was randomised, participants received the corresponding intervention training according to the assigned arm. A two-hour training workshop was conducted for both interventions before the trial commenced, followed by monthly internet-based training capsules tailored to each intervention. These capsules utilised clinical cases, medical literature, and reminders. Both intervention groups received both training programs.

The usual care group followed standard practice procedures. Training for enhanced communication skills emphasized information gathering on patient concerns, symptom exchange, disease understanding, antibiotics, and antimicrobial resistance. It included agreement on a management plan, safety measures, and ensuring patient comprehension. Clinicians were given interactive informative booklets to use during consultations, highlighting key points and offering them to patients. These patient booklets were developed based on findings from qualitative studies in LRTI patients before the trial25,26. Training on CRP rapid test usage included practical guidance on integrating test results during consultations in a two-hour session by the study team. Clinicians learned how to target testing in cases of clinical uncertainty, with an emphasis on ruling out serious infection for values below 20 mg/l. Further details are provided in the Supplementary Material.

Healthcare professionals were asked to recruit sequential eligible adults within regular consultation hours, starting on November 19, 2019. During the index consultation the participating professionals reported the duration of the illness, the severity of cough and other symptoms (rated 0, not problematic, 6, as bad as could be), clinical examination data, presence of addictions and comorbidities, quality of life data, initial treatment prescribed, diagnosis, whether CRP was tested or not, and whether the booklet was used or not. Apart from the point of care CRP tests, clinicians decided investigations and treatment according to their usual practice. Follow-up information about symptoms was reported by patients in self-completed diaries that were followed for two weeks. The diary also included several questions about secondary outcomes. Information on quality of life was collected by the healthcare professional at the days 15 and 45 in the case report form REDCap.

Our study required 2940 patients with LRTI infection to detect a reduction in antibiotic prescribing from 60% to 45% (power 80%, α 0.05, follow-up 90%) when adjusted for clustering at a practice level (intracluster coefficient 0.07) as mentioned in the protocol24.

Continuous variables were described as means and standard deviations and categorical variables were expressed as percentages and frequencies. The study of the distribution of homogeneity of the patients according to baseline characteristics among the four trial arms was performed using the chi-square test or Fisher’s exact test for categorical variables, analysis of variance test for continuous variables, and Kruskal–Wallis test for discrete variables or non-normal continuous variables. Analyses were performed by intention to treat. A logistic regression model was used to assess antibiotic consumption within the first six weeks. The results were expressed in terms of odds ratio (OR) and the usual care arm was defined as the reference group. In addition, these ORs were adjusted for gender, age, smoking status, comorbidities of the patients, and the clustering effect using multilevel logistic regression. Specifically, the estimation was performed with the use of the generalised estimation method specifying a logit link function, binomial distribution, and exchangeable structure correlation. In relation to the EQ-5D-5L questionnaire score and visual analogue scale score, the differences between the 15 day and basal values across the four arms were analysed using a linear model. The parameters were estimated using the generalised estimation method specifying an identity link function, normal distribution, and exchangeable structure correlation. The results were expressed in terms of means. Analysis was performed using R software for Windows version 4.2.2 (R project for statistical computing; Vienna, Austria).

Further information on research design is available in the Nature Research Reporting Summary linked to this article.