Introduction A prospective cohort study was set up to investigate a possible association between antibiotic prescribing and antibiotic resistance of Escherichia coli urinary tract infection in the community. A direct link between prescribing and resistance was analysed by searching the records for previous prescription of antibiotics for patients with a suspected urinary tract infection. Participating practices were requested to send a sample from all patients presenting with symptoms of urinary tract infection. Upon receipt of the sample in the laboratory, a letter explaining the study, an opt out form and a freepost envelope were sent to all adult patients. A website with additional information and including an “opt-out” button was set up for the study.

Results A total of 1362 urine samples were submitted by the 22 participating practices representing 1178 adult patients. The 22 practices send in between 15 and 115 samples. In total, 193 patients actively responded to the letter: 142 opt-outs by letter, 15 through the website, 2 who opted out by phone and 12 who sent the letter back without indication, making a total of 171 patients or 14.5% opt-out, 22 patients (1.9%) explicitly opted in. Opt out percentages varied from 6.1% to 25.5% by practice. We received 2 expressions of concerns over the opt-out method. A response to the concerns was communicated through their GP practice and no further concerns were expressed. The total group consisted of 941 women (79.9%) and 237 men (20.1%). Their mean age was 50.9 (SD 20.8) and median age was 47. Patients who opted out were slightly older (50.4 vs 52.8) and the percentage of females was slightly higher (79.5 vs 83%) but these differences were not found to be significant. Patients who opted out through the website were significantly younger than those who used the letter (non-parametric, 53.5 vs 38.7). The number of patients with a positive urine sample (identified organism) was 395 (34.2%). Patients with a positive sample were not more likely to opt-out compared to those with a negative sample.

Conclusions Overall the opt-out method was well received and participation in the study reached 86.5%. The low number of complaints indicates that this is a generally acceptable method of patient recruitment. The 13.5% opt out shows that it effectively empowers patients to decline participation. The high comparability of the patients opting out with the rest of the patients is reassuring for extrapolation of the results of the study.