Effect of C reactive protein point-of-care testing on antibiotic prescribing for lower respiratory tract infections in nursing home residents: cluster randomised controlled trial

Abstract Objective To evaluate whether C reactive protein point-of-care testing (CRP POCT) safely reduces antibiotic prescribing for lower respiratory tract infections in nursing home residents. Design Pragmatic, cluster randomised controlled trial. Setting The UPCARE study included 11 nursing home organisations in the Netherlands. Participants 84 physicians from 11 nursing home organisations included 241 participants with suspected lower respiratory tract infections from September 2018 to the end of March 2020. Interventions Nursing homes allocated to the intervention group had access to CRP POCT. The control group provided usual care without CRP POCT for patients with suspected lower respiratory tract infections. Main outcome measures The primary outcome measure was antibiotic prescribing at initial consultation. Secondary outcome measures were full recovery at three weeks, changes in antibiotic management and additional diagnostics during follow-up at one week and three weeks, and hospital admission and all cause mortality at any point (initial consultation, one week, or three weeks). Results Antibiotics were prescribed at initial consultation for 84 (53.5%) patients in the intervention group and 65 (82.3%) in the control group. Patients in the intervention group had 4.93 higher odds (95% confidence interval 1.91 to 12.73) of not being prescribed antibiotics at initial consultation compared with the control group, irrespective of treating physician and baseline characteristics. The between group difference in antibiotic prescribing at any point from initial consultation to follow-up was 23.6%. Differences in secondary outcomes between the intervention and control groups were 4.4% in full recovery rates at three weeks (86.4% v 90.8%), 2.2% in all cause mortality rates (3.5% v 1.3%), and 0.7% in hospital admission rates (7.2% v 6.5%). The odds of full recovery at three weeks, and the odds of mortality and hospital admission at any point did not significantly differ between groups. Conclusions CRP POCT for suspected lower respiratory tract infection safely reduced antibiotic prescribing compared with usual care in nursing home residents. The findings suggest that implementing CRP POCT in nursing homes might contribute to reduced antibiotic use in this setting and help to combat antibiotic resistance. Trial registration Netherlands Trial Register NL5054


Supplementary Appendix
. Patient status and management. P2 Table S2. Antibiotic treatment at initial consultation and changes (start, switch, cessation and prolongation) during follow-up.
P3 Table S3. Type of additional diagnostics performed at initial consultation. P4

Supplementary data
Sampling quality P9  Table S2. Antibiotic treatment at initial consultation and changes (i.e., start, switch, cessation and prolongation) during follow-up.

Methods
This supplementary data regarding sampling quality was assessed in a process evaluation, alongside assessment of intervention quality and implementation knowledge of CRP POCT in the UPCARE trial. Data sources that were used included: logs of communication, nursing home intake forms (i.e. information on location and ward size), anonymous overview provided by the software provider of the case report forms (i.e. the number of patients opting out and excluded per organization), trial data (i.e. illness severity and inclusion moment), and a process evaluation questionnaire (i.e. information on enrollment). The main focus of the questionnaire was intervention quality, therefore, the questionnaire was only sent to physicians and nurses in the intervention group. Figure S3 shows that the research team invited 27 NH organizations, of which ten agreed to participate at the initial invitation. One NH agreed at a later time and started April 2019. Reasons for not participating were: not fitting inclusion criteria (five), organizational issues (seven), no time because of participation in another study (three), misfit with local research priorities (one), negative experience with CRP POCT in the organization (one), not willing to take the risk of being randomized to the control group (one). Participating NHs were located in different geographical regions in the Netherlands. Figure S4 shows, amongst others, the average ward sizes of the intervention and control group after randomization. The randomization resulted in similar distribution of psychogeriatric ward size per organization between the intervention and control group. The intervention group organizations had on average fewer somatic residents and more geriatric rehabilitation patients. Figure S4 also shows the flow chart of inclusion, with opt-out and exclusion rates, and ward capacity. After receiving the information letter, 7% intervention group and 11% control group NH residents or their legal representatives chose to opt-out. Of eligible participants, 5% (intervention group) and 7% (control group) were excluded because of our exclusion criteria, which was lower than the anticipated 10%. The final number of participants was 79 in the control group and 162 in the intervention group.

Informed consent, participation rate and reach
The between-group difference in ward type of participants ranged between 2-8% for the different types of wards. For instance, the proportion of psychogeriatric residents among all participants in the intervention group was 35% compared to 29% in the control group. Most participants were moderately ill in both groups (78% resp. 75%). The majority of physicians who included participants indicated in the questionnaire (response rate: 33%) that they included 25-50% of potential cases. Reasons that were mentioned for being unable to include patients were difficulty in obtaining informed consent, a high workload, and high turnover rates at geriatric rehabilitation wards. Participant enrollment most often took place during office hours (87%, range 63-100%), 9% (0-29%) during weekends, and 5% (0-12%) during nights.