A stepped wedge trial design [27
] was conducted on six medical units at two hospitals in Calgary, Alberta, Canada. The intervention was sequentially rolled out (i.e. in-serviced) on a new unit every two weeks for a total of 12 weeks of study. All six units received the intervention, although the order in which units received the intervention was staggered. Until the intervention was rolled out on a unit, it acted as a control unit for purposes of analysis. Due to practical and scheduling issues within these active clinical units we were unable to randomize the units in terms of rollout order. However, all the units involved were medical units with similar staffing structures and all had previously shown a desire to work with the geriatric team. The intervention involved three components:
1. An electronic nurse-initiated order set (e.g. delirium and fall risk screening; regular reorientation, ambulation and toileting protocols; feeding assistance; bowel routines; falls prevention strategies; non-pharmacological sleep routines; pain monitoring; encouragement of independence in activities of daily living) orderable by nurses from within a menu of common order sets for medical patients (►),
Electronic nurse-initiated order set
2. educational in-servicing (e.g. education about relevant age related changes and evidence-informed strategies to provide optimal care to the older in-patient), and
3. a binder of geriatric resource materials (e.g. cognitive and depression screening; behavioural mapping; non-pharmacologic delirium prevention strategies; falls screening and prevention; normal age related changes).
The electronic nurse-initiated geriatric order set was developed by a multi-disciplinary team using evidence obtained from published clinical trials, systematic reviews and practice guidelines [3
]. The order set incorporates the current best evidence aimed at addressing delirium, falls, continence promotion, and optimal nutrition and hydration. Issues of workflow impact and sustainability were considered during development. The order set was made available within the hospitals’ electronic medical record, after it was vetted by frontline nurses, nurse educators and managers, local professional practice leaders, nursing council and clinical informaticians. The research team and local geriatric clinical nurse specialists developed standardized educational materials containing information that supported the electronic order set and the overall care of the older medical inpatient. The brief in-servicing (i.e. 15 minutes) was conducted on a new unit at the start of every 2-week study period. It was offered multiple times on each unit in order to help ensure the majority of unit nurses had the opportunity to attend, and it was kept brief in order to allow nurses to attend during a work shift. At the start of the 12-week rollout, the order set was available to all users of the electronic medical record but the unit nurses were not informed about its presence until the educational in-servicing.
We used a mixed methods approach to analysis. Data was collected from the electronic medical record of all patients 65 years of age or older who were residing on the study units on the days of data collection (ranging from 17–44 patients per unit per data collection period), which were at baseline and then once at the end of every 2-week study period over the 12-weeks of study. The primary quantitative outcome was the rate of use of the order set, as determined by the number of patients 65 years or older with the order set on their electronic medical record compared to the total number of patients 65 years or older on the units at the time of data collection. This process outcome was chosen in order to help determine how effectively the evidence-based strategies were incorporated into routine clinical care.
Secondary outcomes included the number of falls as documented on incidence reports, the average number of days in hospital, and the total number of consults ordered for each of orthopedics, geriatrics, psychiatry and physiotherapy. Given that incidence reports do not identify the patient, and therefore several fall reports could be generated by one person who falls frequently within an assessment period, the outcome of falls was considered as either no incidence reports versus one or more incidence reports within a given assessment period. Data was extracted from the electronic medical record using data queries based on unit and age. The unit of analysis was the medical unit. Given data was collected at the unit level and included no individual identifying data, and the fact that the intervention was considered usual best evidence clinical care the Ethics Board waived the need to obtain individual patient consent. The data were first explored with descriptive statistics including means and medians for continuous variables and frequencies for categorical data.
Repeated measures models were used to assess a time by intervention effect on outcomes using transformations as needed to meet model assumptions. Upon finding no evidence of a time effect, straightforward linear regression and logistic models were used to test for the effect of the intervention, while adjusting for number of patients and testing covariate effects of average age, average length of stay, percentage of female patients and unit [31
The qualitative analysis included interviews that were conducted approximately six months after the completion of the 12 week roll out. Nursing staff from four of the six medical units agreed to participate in interviews. Interviews were conducted on each unit during a time approved by the unit nurse manager (i.e. when the manager and/or educator were available on the unit to be interviewed). Unit nurses were recruited with the assistance of the unit nurse managers and using snowball sampling, although the numbers were limited by nurse availability on these busy medical units. Frontline nurses, unit nurse managers and educators were interviewed in order to explore issues around the implementation of the KT intervention. Interviews were delayed until six months after roll out in order to help identify potential issues with sustainability of the intervention. Participation was considered as implied consent. An experienced interviewer, after providing participants with information sheets outlining the purpose of the study, conducted semi-structured interviews using a standardized list of questions. Domains of inquiry included impact on workload, perceived impact on patient care and barriers and facilitators to use. Responses were transcribed from audio recordings and merged with the notes taken in situ. No personal identifying data was collected. Two investigators independently analyzed the data using a content analysis approach. Common themes were identified, agreed upon and categorized. This study received ethical approval from the University of Calgary Conjoint Health Research Ethics Board.