The model for improvement is based on a trial-and-learning approach to improvement in which small, frequent cycles of change are used to turn ideas into action. This process allows ideas to be discarded if they are not effective, modified when the idea shows promise, and implemented when deemed successful. Each cycle of change has four steps and is commonly referred to as the PDSA cycle.3
PLAN – Design a change, list the actions needed to trial the change, and predict the outcome.
DO – Try the change on a small scale and document the results. Use measures identified above and document problems and unexpected observations.
STUDY – Analyze the data collected, compare to the predicted outcome, and summarize what was learned. Note that a substantial proportion of tests – 25 to 50 per cent – will result in no improvement but can still produce learning that can be applied to redesigning the change.
ACT – Take action based on what was learned. This could include discarding ineffective changes, modifying a change to address problems encountered, trying a successful change on a larger scale, or making a decision to implement a change that seems to be working.
The project team continued with data collection in the last two weeks of every month to assess the impact of the interventions. The letter to prescribers was followed one month later by the inservice education program and accompanying posters. Each month, feedback was provided to prescribers showing how their use of units in ordering medications compared to their anonymous colleagues, and a graph showing how the proportion of their total orders that used the full word units changed over time (run chart). After three months, a 75 per cent compliance rate with the use of units to order the target medications had been achieved; more than half of the prescribers were consistently using units for all their orders.
After four months, it was noted that three prescribers began to revert to their former practice of consistently using U instead of units. A targeted intervention was designed for this small group which consisted of personal contact by their chief of service to emphasize expectations of performance related to medication ordering and a phone call from the pharmacy to request a correction of any orders in which U was the dose designation. This was successful in changing the ordering practices of this group.
After five months, the use of U began to gradually increase again across the target units. The timeline of the initiative was extended to try some new strategies. It was decided to add to the order form a reminder to avoid using abbreviations. As an interim solution to test this change, a sticky label was affixed to the top of each new order sheet by the unit clerk. After two months, the use of units had climbed to 80 per cent of all applicable orders. However, this was not considered to be sufficient as patients were still at risk of medication errors from this practice. Finally a policy was implemented in which medication was not dispensed until the order was clarified. This reinforced to the pharmacy and nursing staff the importance of not making assumptions about a dose before dispensing or administering a high risk medication like insulin to a patient.
By the end of a nine month project period, 98 per cent of orders for the target medications from the medicine and surgery services used a dose designation of units. The remainder of the orders required follow-up by the pharmacy before the medication was dispensed.
The educational intervention took place over a two week period after which the project team resumed data collection on all insulin orders. After each four week period, bar graphs were created showing how the use of units during that period compared to the baseline period. Prescriber feedback was sent out showing each prescriber how their use of units was changing over time compared to that of their anonymous colleagues. Posters were placed in the pharmacy and in staff areas at the facility to show how the use of units by physicians and in transcribed orders was changing over time.
After four months, the nursing and pharmacy staff were consistently using units to transcribe orders for insulin, with a 97 per cent compliance rate. Staff responded quickly to personal requests, usually from their colleagues, to change an order they had transcribed incorrectly. There was a steady increase in the use of units by prescribers, with an overall 75 per cent compliance rate by the end of the fourth month. Compliance ranged from 100 per cent to zero – three prescribers continued to use U to prescribe insulin for all orders. The medical advisor contacted these individuals personally to discuss expectations for ordering insulin. Compliance issues were also noted when new physicians or locums providing vacation relief prescribed insulin. These prescribers were sent a letter from the medical advisor followed by a prescriber feedback letter after four weeks of data collection.
By the end of the six month period the aim of 85 per cent compliance with the insulin ordering policy had been achieved and the ordering practices of all physicians had improved. However the project team decided that this was not sufficient because patients were still at risk of medication errors from misinterpreted orders. It was decided to implement a policy in which insulin doses designated with U would not be dispensed until the order was clarified with the prescriber. This reinforced to the staff the importance of not making assumptions about a dose before administering a high risk medication like insulin to a patient.
Data collection continued for another two months at the end of which the compliance rate for prescribers was 98 per cent. Nursing and pharmacy staff consistently continued to use units when transcribing orders for insulin.