Applying “lean” principles to improve hospital chemotherapy services

The introduction of electronic prescribing to Musgrove Park Hospital oncology day clinic in 2009 coincided with reports of problems initially centred on chemotherapy production. These included treatment delays, missing drugs, and confusion surrounding pre-emptively prepared chemotherapy. Service and incident reviews revealed that production processes had grown in attempts to compensate for unrealistic service demands. Over 50 per cent of patients waited some hours for treatment. Production run times were highly variable, frequently overrunning.

A systematic evaluation of all processes relating to chemotherapy prescribing, production and administration was undertaken using “lean” change management techniques.

The aim was to remove wasteful, unnecessary, unsafe and unreliable activities.

Method

Since our work was a quality improvement project, ethics committee approval was not required. The five production staff and the directorate manager for haematology and oncology were briefed and recruitment to the project begun.

Baseline data for production run finishing times were collected and estimates of patient waiting times concluded that over half of patients waited up to three hours for therapy. This was based on the fact that 75 per cent of chemotherapy was administered on the day of preparation and most was scheduled to be given before lunchtime and 25 per cent of therapy was administered the day after its preparation.

A process map of the entire chemotherapy system was constructed. Wasteful activities were identified (eg, telephone/bleep interruptions, staff visits), together with informal process activities (“work rounds”). The project team consisted of three production staff and the chief pharmacist, as chairman. Each mapping session lasted 45 to 60 minutes. Process bottlenecks were identified and the purpose of each step in the process evaluated. The chairman also acted as project sponsor. When low value activity was identified it formed the basis for subsequent tests of change. 

Failure mode and effects analysis was applied to the process map but was found to be of limited value on this occasion.

Wasteful activities were identified including: unnecessary complexity; irrational prioritisation; treatment delays; duplication; bottlenecks and work rounds to exceptional, unscheduled activities; and unco-ordinated scheduling between pre-assessment clinics, production runs and therapy.

The project group developed potential solutions to overcome these deficiencies. Solutions were tested via PDSA (plan, do, study, act) rapid cycle testing. Successful tests of change were implemented. Throughout the project production run finishing times were recorded on run charts and analysed via statistical process control models. Instances of therapy delays were monitored daily.

A final chemotherapy production process was developed.

Results

Median production run times fell from 4.07h (sd=0.52h) to 2.82 hours (sd=0.44h) equating to a weekly saving of 31.25h of staff time per week. Treatment delays decreased from more than 50 per cent to under 5 per cent. Throughout the project “near miss” error rates remained unchanged and no dispensing errors occurred. No additional staff resources were required.

Discussion

Our paper has been written as a quality improvement project whose aim was to systematically improve an existing process. Data gathering, analysis and accounting for bias is significantly different from that in typical scientific research. For instance some baseline data required estimation. Although baseline data are required, copious quantities are not, and an initial lack of data should not deter from undertaking quality improvement work.

During tests, positive bias was encouraged. The Institute of Healthcare Improvement refers to this as “stacking the deck in your favour”, the rationale being that a test is more likely to work in favourable circumstances.

Removal of low value activity halved the number of process steps. Safety was addressed by not preparing drugs in anticipation of late prescription authorisation. Consequently drug waste due to late treatment cancellations and dose adjustments also fell. Production runs became more efficient and able to cope with unscheduled activities without interfering with subsequent work.

Changing treatment scheduling for most therapy markedly avoided delays in scheduled therapy and is as follows:

• Day 1 — required blood tests were performed during GP/oncology appointment
• Day 2 — medicines were prescribed at pre-assessment together with some blood tests 
• Day 3 — chemotherapy was prepared for the next day and mid-morning that day
• Day 4 — therapy was administered

Appropriate scheduling and a more efficient production process allowed chemotherapy to be made more flexibly and without affecting subsequent work. These changes were incorporated into a service level agreement. 

A second major change addressed the bottleneck in in-process checking whereby the checker was placed in the isolator room. This provided direct technician supervision for production assistants and reduced run time by a median of nearly 90 minutes. The situation was monitored thereafter by selected process metrics.

Conclusion

Systematic interventions designed to reduce wasteful activity, combined with co-ordinated activity rescheduling, reduced production run times, patient waiting times and staff costs.