Setting
This single center, pragmatic, pilot study was conducted at the Department of General Internal Medicine, Bern University Hospital, Switzerland, a 950-bed tertiary care center. The department runs a total of 97 beds on six wards. Patients were enrolled from 17.01.2019 to 16.04.2019, the time period where influenza cases exceeded the epidemic threshold (68/100′000 patients/week) on a national level [13].
Introduction of DroPS
DroPS was introduced on four wards (totaling 73 beds, including 5 single rooms, 24 twin rooms, 2 four-bed and 2 six-bed rooms). Two wards (totaling 24 beds, including 4 single rooms, 2 twin rooms, 4 four-bed rooms) were defined as “regular” wards, where the pathogen-based single room isolation strategy was maintained. Based on the total number of beds, this corresponds to a ratio of 3:1.
The allocation of patients to the wards was done by the hospital bed managers, driven by bed availability.
DroPS strategy
DroPS was introduced at the patient bed site based on clinical criteria. DroPS could be started by any HCW (including those in training) if a patient presented new respiratory symptoms (cough, rhinitis, sore throat) and/or was diagnosed with pneumonia, exacerbated COPD or acute bronchitis. The respiratory pathogen – if known—had no influence on the approach taken. DroPS included the following: signage of the patient bed; curtain drawn next to patient bed; distribution of surgical masks and bedside disinfectant to the patient including information about its use when leaving the bed site (Fig. 1); HCW wearing a surgical mask if patient contact < 1.5 m of distance; and enforcement of standard hygiene precautions. Immunocompromised, non-cooperative and non-compliant patients were excluded from DroPS, as well as patients with suspected bacterial meningitis, pertussis, rubella or mumps. A detailed description of all DroPS measures is provided in Additional file 1: Table S1. The introduction of DroPS was preceded by extensive training and instruction of the nursing teams and physicians.
Standard pathogen-based single room strategy
On the regular wards, based on the detected viral pathogen, respiratory precautions were either continued as droplet or contact isolation in a single room or transformed into standard hygiene precautions, according to the institutional policy (Additional file 1: Table S2).
The isolation precautions on both the DroPS and regular wards were stopped as soon as the patient was no longer symptomatic or a reduction of symptoms to the individual’s baseline occurred.
Data sampling
For each hospitalisation, we collected admission date, patient identifier, case identifier, date of birth, gender, ward, room number and respiratory isolation status once daily (at 12:00 p.m.). Swab results and hospitalisation duration was retrieved from the digital patient management system.
There was continuous risk assessment of DroPS through monitoring and daily analysis of every possible HARVI during the assessment period.
Definitions/outcomes
Inclusions and exclusions
For the descriptive outcomes we included the patient population considered at risk for HARVI, comprising all hospitalisations with no respiratory isolation precautions during the first two days after admission. Excluded were hospitalisations with a transfer from DroPS to regular wards and vice versa (i.e. cross-over hospitalisations).
Diagnostics
On a daily basis, patients on all wards were screened for the onset of new respiratory symptoms. If present, a nasopharyngeal swab and influenza/RSV molecular rapid test was performed (Cobas® LIAT®, Roche, Switzerland). If negative, and if the onset of symptoms was hospital-acquired, this test was followed by a nasopharyngeal swab for multiplex respiratory virus PCR, directed at adenovirus, rhinovirus, coronavirus (not MERS or SARS-CoV), human metapneumovirus and human parainfluenza virus (from ARGENE® Respiratory menu, Biomérieux, France) (summarized in Fig. 2).
Definition of hospital-acquired viral infection (HARVI)
We evaluated patients for HARVI who developed acute respiratory symptoms after two days (i.e. ≥ day 3) after admission to one of the DroPS or regular wards (see Additional file 1: Table S3), in line with the definition of nosocomial infections, which is characterised as infection occurring 48 h or later after hospital admission [14].
Primary and secondary exploratory outcome
The primary composite outcome was the rate of hospitalisations with microbiologically confirmed HARVI with influenza or RSV in the patient population at risk.
Secondary outcomes were the rate of hospitalisations with microbiologically confirmed HARVI due to any respiratory viruses; the rate of hospitalisations with clinical HARVI and the rate of hospitalisations with clinical HARVI, either microbiologically confirmed or with missing test, in the patient population at risk.
Adjustments for the following variables were planned: patient characteristics; isolation strategy; data from the previous four hospitalisation days of the patient: single vs. multi-bed room, isolation rate / ward, isolation rate / room, isolation type of the patient and influenza vaccination rate of HCW.
Sample size
Based on the previous year's hospitalisation data, the rate of HARVI with influenza and RSV during the influenza period was estimated to be 0.7% (data not shown). Assuming similar patient numbers to the previous year, we estimated to include approximately 750 patient hospitalisations in the DroPS and 300 in the regular group. With this number of expected patients, a statistical significance level of 5%, and 80% power, we estimated that we would be able to detect a non-inferiority margin of approximately 1.3%. This means we considered a HARVI rate up to (0.7 + 1.3 =) 2% (i.e. the upper 95% confidence interval below 1.3%) in the DroPS group to be similar to the rate of 0.7% in the regular group (calculated using function “TwoSampleProportion.NIS” in package “TrialSize” in R [15]).
Statistics
Categorial variables are presented as number (percentages) and for continuous variables as median (interquartile range; IQR). Differences between DroPS and regular groups were investigated using the chi-square test (or variants thereof) for categorical variables and the non-parametric Wilcoxon rank sum test for continuous variables.
The primary outcomes were the unadjusted rate of confirmed HARVI with influenza or RSV on the DroPS and regular wards in the patient population at risk. We further described the risk difference between the DroPS and regular wards for the outcomes of confirmed HARVI with influenza or RSV. Analysis of further outcomes followed the same approach.
A level of 5% was considered statistically significant throughout. All analyses were performed using R version 3.6.1 [16].