Study design
The study was set up as a before-after comparison consisting of three periods, before, during and after intervention. Ten Dutch hospitals scattered over the country participated. Hospitals involved in the Dutch surveillance programme of nosocomial infections (PREZIES) were approached with information about the project. Hospitals that responded positively were included. Out of the10 hospitals selected two were university hospitals located in Leiden and Amsterdam. The other hospitals were general acute care hospitals located in Veghel/Oss, Deventer, The Hague, Haarlem, Roermond, Rotterdam, Tiel and Tilburg.
The hospitals were randomized into group A and B. A piece of paper with the name of the hospital was put in an opaque envelope and sealed after which the envelopes were mixed. Someone not involved in the study was asked to put an A or B randomly on five envelopes, respectively. Group B acted as control for unintended changes in the outcome parameters during the intervention period in group A hospitals. Each hospital selected as many wards as needed so that the expected number of patients with a urethral catheter was approximately 20 per site visit. There was no restriction on the type of department. The only criterion for ward selection was the number of patients with catheters. Assuming a base line catheter prevalence of 30% and a minimal of four site visits per period per group of hospitals, and taking into account an expected maximal inter-hospital variation of 30%, the power to detect changes of 10% is expected to be 85%, to detect changes of 20% is 94% and to detect changes of 30% is 99%, when a number of 2000 patients per group is included. The study was assessed by the medical ethical committee of the Leiden University Medical Centre which concluded that the study focused on improvement of patient care and, therefore, did not require permission by the committee (CME 05/36).
Intervention strategy
After a three-month period of baseline measurements the five hospitals allocated to group A started preparations for the intervention period in month four of the study. First in each hospital a small intervention group consisting of infection control professionals, (head)nurses, incontinence nurses and doctors, prepared the intervention, and then introduced it in the wards. In month nine the intervention period was closed. The hospitals of group B started the preparations in month nine after eight months of baseline measurements and stopped the intervention activities in month 14.
The aim of the intervention groups was to develop a method of implementation that best suited the specific circumstances of the hospital. The intervention groups were supported by a document describing intervention methods, literature about reducing the use of urinary catheters and a summary of the results of semi-structured group interviews in three hospitals to detect barriers and facilitators to the implementation. In one session per hospital an implementation expert instructed the local intervention group.
Data collection
During a period of 17 months each hospital was visited 20 times to include patients and collect data. On a study day a research nurse visited the participating wards of a hospital and included all patients with a urinary catheter with the exception of those having catheters for urological interventions. For each patient age, sex, the day of insertion of the catheter, the indication for catheterization on the day of insertion and the study day, signs and symptoms of urinary tract infection, and antibiotic use were recorded. A urine sample was collected for laboratory tests. The numbers of patients admitted to a ward and of those using an alternative for bladder catheterization (e.g. diaper, condom catheter) were registered. The infection control professionals of the hospitals followed the patients until removal of the catheter or discharge from the ward to determine the number of catheterization days.
Outcome parameters
The effect of the intervention was monitored by measuring the prevalence of patients with a urethral catheter, patients with an alternative to urethral catheterization, and urinary tract infections in catheterized patients, the percentage of patients with a correct indication for urethral catheterization at the time of catheterization and upon inclusion in the study, and duration of catheterization from insertion to removal.
Bacteriuria was defined as ≥ 105 colony-forming units/ml in at least one of the three media of a uricult (Uricult Plus; Orion Diagnostica, Finland). A symptomatic urinary tract infection was defined as the presence of fever or pus draining along the outside of the catheter or pain in the bladder region and bacteriuria. Symptomatic infection was also diagnosed when two of the above-mentioned clinical signs were present and the leukocyte esterase or nitrite test (Combur 2 Test® strips, Roche Diagnostics) was positive. Asymptomatic urinary tract infection was defined as bacteriuria in the absence of any of the above-mentioned clinical signs [10].
Correct indication for urethral catheterization was defined according to generally accepted criteria [11, 12].
Cost evaluation
To calculate the costs of the implementation activities everyone involved recorded the time spent on the activities. Other costs such as travelling, meetings and material expenses were listed. Time was related to the gross salary of the different employees [13]. The price level of 2008 was applied.
The costs of catheterization and daily care for a patient with an indwelling catheter were calculated based on the costs of materials and the mean time spent by healthcare workers. Mean time was determined by asking 18 experienced nurses employed in 5 different hospitals to indicate the time needed for specific actions such as insertion and removal of the catheter, emptying and changing catheter bags, daily care of the catheter site and solving problems. Costs of materials were based on hospital purchase prices.
Statistical analysis
For the analysis the period before intervention (start of study until first intervention activities in the wards) was compared with the period after intervention (start of activities in the wards until the end of the study). Time series analyses were used to estimate trends in outcome measures during the periods. The effect of the intervention on catheter prevalence and catheter duration was quantified by regression modeling. To account for paired observations, i.e. before and after intervention per hospital, hospitals were put in the model as random intercept. Intervention, ward type and their interaction were put in the model as fixed effects. The prevalence was modeled by a logistic regression model [14], while the duration was modeled by an exponential regression model [15]. For the latter, it was assumed that the probability of catheter removal is constant over time. Censoring was allowed. The models were fit using the statistical software package R [16]
The number of catheter days eliminated by the intervention was calculated by subtracting the average number of catheter days before and after the intervention, respectively, per hospital. All these figures were added to calculate the total number of saved catheter days per 100 patients.