The primary outcome measure was the incidence of AEs (sum of the incidence of pressure ulcers, urinary tract infections and falls).
A pressure ulcer is an area of localised damage to the skin and underlying tissue caused by a combination of pressure and shear. Pressure ulcers are classified in four grades according to the guidelines [19, 21, 40]. Pressure ulcers were considered present if a patient developed a PU grade 2 or worse. If a patient had a PU grade two or worse at the first visit, that PU lesion was excluded from the registration of PUs until the PU healed. Patients with an already present PU grade two or worse were only registered if they developed additional PU lesions.
A urinary tract infection is bacteriuria with clinical symptoms as: frequent urinating, pain while urinating, abdominal pain, fever and delirium, urinary incontinence [18, 24]. During this study we defined a urinary tract infection as present if it was diagnosed by a medical doctor. Patients were excluded from the registration of urinary tract infection for a period of three weeks if they had a urinary tract infection until the infection was cured.
A fall is an unexpected event in which the participant comes to rest on the ground, floor, or lower level [20, 41]. In this study the falls were measured by examining the patient files, assuming that if a patient fell it was reported in his or her file.
The secondary outcome measures were 1) the percentage of patients that received preventive care and 2) the knowledge of nurses regarding the three topics.
Prevention is important in patients at risk for one of the AEs. Preventive measurements were considered present when the care was performed according to the guideline.
The risk of pressure ulcers was measured with the PrePURSE  and the Braden scale  in hospitals and nursing homes, respectively. Next preventive care was measured: position while lying or sitting; if patients' heels were lifted; use of pressure-reducing material or alternating pressure material in bed or chair; presence of a repositioning scheme.
Hospital patients were at risk for a urinary tract infection if they had at least one of the next four risk factors [18, 23]: 1) a urinary catheter in situ or the week before, 2) incontinence of faeces, 3) urinary retention or 4) a urinary tract infection in the last two years. According to the guideline, all nursing home patients were considered at risk for a urinary tract infection . Next preventive care was measured: personal hygiene, frequent toilet visits, unnecessary indwelling catheter and unobstructed urine flow.
To identify hospital patients at risk for falls we used the STRATIFY . According to the guideline all nursing home patients were considered at risk for falls, except those who were totally immobile . Next preventive care was measured: if the file had a written multidisciplinary plan with multi-factorial preventive interventions; a periodic evaluation of the multidisciplinary plan; a periodic evaluation of the multi-factorial risk factors for falls.
The knowledge of nurses about risk assessment and effective preventive care was measured using a written knowledge test. Each topic had twenty questions, on which nurses could answer 'correct', 'not correct', or 'do not know'.
The knowledge test was developed from questionnaires  (knowledge test used in an implementation study of a pressure ulcer guideline in the Netherlands (Schoonhoven, L. 2004) and geriatric educational material of the prevention of falls, 2007) and student tests of the HAN University of Applied Sciences on the three topics. The face validity was tested by sending the questionnaire to the members of the research group (LS, JAJM, RTCMK and TvA), and the expert on each topic. Finally, nurses in hospitals and nursing homes were asked to pretest the questionnaire.
Power calculation was based on the primary outcome, with a two-sided alpha of 0.05 and 80% power for the analysis of both the hospital and the nursing homes data.
As randomisation was on ward level, a ward was considered to be a cluster. To account for these clusters an intra class correlation coefficient of 0.01 was used in the calculation.
In hospitals, the incidence of pressure ulcers (10%) will be the highest contributor to our combined AE measure. The incidence of urinary tract infection and falls in the same patients is unknown. Therefore we assumed that the count of these three AEs will be 12% (an additional 1% for falls and 1% for urinary tract infections). We aimed to achieve a reduction of 50% as studies on the prevention of pressure ulcers have shown this is attainable [45, 46]. To detect a decrease in AEs (from 12% – 6%) 1250 patients had to be included in each hospital group.
In the nursing homes, the incidence of falls will be the highest (60%). We assume that the additional contribution of pressure ulcers and urinary tract infection to AEs will be negligible. We aimed to achieve a reduction of 60% as a study on the prevention of falls showed this was attainable . Therefore this study wanted to achieve a reduction of AEs from 60 – 36%. To detect this decrease in the nursing homes, 100 patients had to be included in each group.
The results will be analysed separately for hospitals and nursing homes, as patient characteristics, length of stay and nurse characteristics differ between hospitals and nursing homes.
The difference in incidence of AEs between the intervention and the control group during the follow up period will be analysed using a random effects Poisson regression analysis, including the following covariates: ward (random effect), institution and the baseline results of the ward.
The secondary outcomes will be evaluated in a similar way, using linear and logistic random effect models.