Antibiotic prescribing and non-prescribing for possible urinary tract infection in nursing home residents

BACKGROUND AND PROBLEM DEFINITION. Diagnostic uncertainty in nursing homes (NHs) drives physicians to prescribe antibiotics (AB) for suspected urinary tract infections (UTI), whereas part of these prescriptions are not required. Diagnosing UTI is challenging in this setting. In current practice, clinicians frequently base their diagnosis and subsequent AB prescriptions on nonspecific signs and symptoms (S&S), i.e. S&S that are not related to the urinary tract. This is due to the widely shared view that diseases often present atypically in older patients and that self-report of complaints is often not reliable in NH residents due to frequent cognitive impairments. However, it has been described in the literature, that AB are often not required for nonspecific signs and symptoms (S&S) ascribed to a possible UTI, either because their mild nature allows for self-limitation of the infection, or because they are not related to a UTI at all, but this diagnosis is made erroneously in the presence of concomitant bacteriuria. Asymptomatic bacteriuria (ABU) is highly prevalent in the nursing home setting and a gold standard to distinguish ABU from a ‘true’ UTI is lacking. Hence, evidence is lacking on which S&S exactly allow for withholding AB. Prior to the current research proposal, a Delphi panel with (inter)national experts on UTI, will reach consensus on S&S attributed to a UTI that should not (yet) be treated with AB, but justify a ‘watchful waiting’ approach. Based on the outcomes, the Delphi panel will develop a UTI treatment algorithm to support clinical decision making on S&S that justify AB prescription and S&S that justify a ‘watchful waiting’ approach in cases of suspected UTI in NH residents. This algorithm will be available before start of the research proposed here. OBJECTIVE: We hypothesize that use of this UTI treatment algorithm based on the Delphi study will provide clearer guidance to clinicians and will lead to better AB stewardship and reduced AB use in NHs. The proposed study aims to implement and evaluate this algorithm. If our hypothesis proves correct, implementation of the UTI treatment algorithm may have substantial benefits for both individual patients (better management, less exposure to side-effects of AB and drug interactions) and public health (less AB resistance development). The findings will also contribute to a much needed revision of the current guideline on UTI in NH residents of the Dutch Association of Elderly Care Physicians and Social Geriatricians (Verenso). STUDY DESIGN: cluster randomized trial in 2 x 5 university affiliated nursing homes of approximately 150 resident places. In addition a process analysis will be conducted to determine the validity of the intervention and to gain insight into relevant facilitators and barriers for future implementation of the UTI treatment algorithm. STUDY POPULATION: NH residents with ‘suspected UTI’ PRIMARY OUTCOME: AB use for ‘suspected UTI’ SECONDARY OUTCOMES: course of symptoms, alternative diagnosis at index consultation, changes in treatment < 3 weeks, hospitalization, mortality, (other) complications, total AB use at NH level. DATA ANALYSIS: For the primary outcome a three level logistic regression model will be used in analysis. For the secondary outcomes multilevel regression will also be used to compare between groups. A second-order penalized quasi-likelihood estimation procedure will be applied, using MLwiN. Both quantitative and qualitative data analysis will be performed for the process evaluation. TIME SCHEDULE: 48 months