The MOWER (middle of the week everyone gets a re-chart) pilot study: reducing in-hospital charting error with a multi-intervention

Medication charting error can result in prescribing errors which are potentially harmful, can lengthen hospital stay, and increase the financial burden to families and society [1, 2]. Prescription errors are estimated to affect 50% of hospital admissions [3], and the majority are most likely not reported [4]. The fast-growing numbers of multi-morbid, elderly patients are most at risk, as they are often prone to polypharmacy and have a higher incidence of dementia, therefore may be less aware if they are receiving the wrong medication [5, 6].

The causes behind charting and prescribing errors are multifactorial but have similar aspects from setting to setting, including heavy work load, lack of proper communication, and a hospital culture that views prescribing as a low-risk, unimportant chore [1, 7, 8]. In addition, junior doctors are often given the task of writing the medication charts despite being more likely to make prescribing errors due to insufficient knowledge and training in prescribing [1, 9]. Junior doctors report insecurity in how to prescribe [10], reluctance to oppose senior colleagues regarding medication decisions [1] and a need for more extensive pharmacology training in medical school [11, 12].

Interventions with multiple approaches such as chart training and physical reminders have been found more successful than those focusing on a single factor [13, 14]. Single factor interventions that have shown success when combined with education interventions include: audit and feedback [14], manual reminders such as posters with checklists for correct prescribing [14, 15] and involving pharmacists in medication charting and medical reconciliation [16].

In the current study, we implemented a multi-intervention strategy incorporating previously proven methods such as training of junior doctors [17], manual visual reminders [14], and a novel method involving weekday re-charting by the junior doctor under supervision of the registrar - the ‘MOWER’ (Middle Of the Week Everyone Gets a Re-chart) intervention. The aim of the multi-intervention was to raise awareness of the importance of charting practices and medication errors. The specific goal of the MOWER was to improve re-charting which can be prone to transcription error [18] by scheduling it for a weekday, therefore ensuring the primary team did the re-charting whilst engaging more experienced doctors to guide and educate the junior doctors. The idea of re-charting is consistent with proven audit and feedback interventions [14], as the medication chart is being checked and the junior doctor is receiving immediate feedback and instruction.

The main objective of our study was to determine whether a multifactorial intervention focusing on weekday team re-charting reduced charting error.

Of the 420 charts audited at baseline, 23% (n = 97) were taken on the weekend and 77% (n = 323) during the week. Due to re-charting on weekdays, charting on the weekend reduced to 17% during the intervention (n = 27). The proportion of re-charts was constant throughout the study at 20% of all charts. However, as expected due to mandatory re-charting on a weekday, the number of re-charts that naturally fell on a weekend decreased from 28% (n = 24) of all re-charts to 6% (n = 2) during the intervention.

At baseline, there was an average charting requirement error rate of 4.6 ± SD 1.3 per chart, this equates to a 58% error rate per chart for the 8 charting requirements (Table 2). The multi-intervention significantly reduced the overall error rate by 26%, 1.2 (95%CI 0.9, 1.5) errors per chart during the intervention. Similarly, the weekday chart errors during the intervention reduced by 27% per chart (1.2 (95%CI 0.9, 1.5) errors, p < 0.001; Table 2). Weekend chart error did not significantly reduce. The re-charts made the most improvement of any sub-group during the intervention with a 50% reduction from 4.4 ± 1.4 charting requirement errors to 2.2 ± 1.7 per chart. This was more than double the reduction seen in the primary charts, which showed a 20% reduction, or 0.9 (95%CI 0.6, 1.2) errors per chart (Table 2).

Of the eight charting requirements, the requirement with the highest compliance at baseline was ‘Dating and Signing’ of all prescriptions in the medication charts with error rate of 3.1% of all charts (Table 3 and Fig. 1). The requirement with the lowest compliance at baseline was writing the ‘Patient’s Surname’ below the patient label on the medication chart with an error rate of 92.9% of all charts. The charting requirements with the lowest compliance at baseline were also the requirements which had the most significant reduction during the intervention; ‘Indication for PRN medications’ (32.6% improvement); ‘Patient Surname’ (21.8% improvement); ‘Indication for some regular medications’ (19.2% improvement); ‘Generic name for regular medications’ (17.7% improvement) (Table 3).

This pilot study detected a 58% error rate per medication chart for the 8 targeted charting requirements. The multi- interventions approach, including MOWER, visual reminders and training of junior doctors, resulted in a 26% reduction during the intervention. The greatest error rate reduction of 50% was seen in the midweek re-charts. Most improvement was made for those charting requirements that were initially poorly attended to.

We are unaware of any other studies that have used a re-charting approach. However, there are a number of studies that have used a multi-intervention study involving education alongside other interventions such as changes in protocols or feedback to trainees. On the whole, our reduction in error rate of 26% on all charts during the intervention is comparable to these multi-intervention studies [24]. Interestingly, the greatest error rate reduction was seen in the re-charts during the intervention, more than twice the reduction in error rate seen in primary charts, and above the reduction rate found in other studies. There are several possible reasons for these results, one being that it provided another opportunity for the junior doctor to be trained in charting requirements. This agrees with other research that has found that education in combination with audit and feedback is an effective intervention method [25‐27]. It could be viewed that in creating the re-charts with a senior and junior doctor team the junior doctor is auditing the primary chart and being given feedback from the senior doctor. A second reason for the improvement could be that by having a team of two looking at the chart this increased accountability and vigilance, which has also been observed in other studies studying team effect [28, 29].

During the intervention our study found that weekend charting practice did not improve, possibly because weekend doctors often cover more than one ward and the doctors completing charts in the medical ward on the weekends may not have been trained in or were made aware of the intervention. As far as we know this is the first study that has compared weekend and weekday charting. The results may indicate the difference in care that is provided on weekends as recognized in higher mortality rates of those admitted to hospitals on weekends [30, 31].

In regards to specific charting errors, we found that the most common type of errors occurring at least once on any given chart were: not providing indication for medications (73% of charts), not providing generic names (76% of charts) and a non-existent or illegible prescriber name (35% of charts) whereas dating and signing was correct in almost all cases. These rates are very similar to other Australian studies [12, 32] but maybe higher than rates in the UK where the majority of errors have been to do with the omission of medications at admission [9, 22, 26]. Lack of indication, using brand names and an inability to identify the prescriber are all errors that can lead to signficant problems particularly when rotations change and new medical personnel are using the charts and are provided with inadequate information. The multi-intervention reduced all of these errors ranging from 15% (prescriber name) to 33% (indications for as needed medications) reductions.

A strength of our study is that the multi-interventions complemented one another and were delivered in different modes (i.e. visual reminders, training, charting practice), so training was reinforced at different times. Multi-interventions with education as a component have been shown previously to be more effective than visual reminders or audit and feedback alone [24, 32, 33]. To facilitate effective implementation of the intervention on the ward, it was vital to have the support of all clinicians regarding the intervention and the overall aims. Key to our success with this was in maintaining a collective focus on enhancing charting practices overall on the ward, rather than a punitive approach of identifying individual prescribers and highlighting their errors. This approach however meant that we were not able to control for individual clinician effects in our analyses. In regards to limitations, although we were able to investigate a number of potential error types we were unable to assess other categories of clinical importance such as listing of adverse drug reactions and route of administration due to a lack of trained staff to interpret these aspects of the chart. However, we feel that a range of criteria was used that reflects overall charting error and the influence of multi-intervention, including both specific medication information such as indication, generic name and maximum dose, and administration information such as chart number and writing the name of the patient, all of which have potentially hazardous outcomes if not followed correctly. We also recognize that the recent introduction of electronic prescribing in this hospital, which occurred after this study was completed, may improve some of the charting error rate. However, it has been found that electronic prescribing is still prone to human error [34‐36], which therefore does not preclude the need for a primary care team including a registrar to regularly review inpatient charts. In addition, we were unable to explore prescriber demographics and level of experience in this study, along with patient characteristics. This could be informative for a further study but may affect whether prescribers would be willing to be involved in the intervention.

In regards to further testing and implementing this intervention on a wider scale, for example, in a larger hospital setting, there needs to be involvement and commitment by hospital management as the intervention requires time for staff training on charting practices as well as dedicated time once a week for senior and junior staff to re-chart. In addition, this process would have to be repeated regularly due to staff turnover. Once this is embedded into the organization as a standard practice though, it should happen more naturally, and the net benefit of this time investment would be improved charting and hence better patient care. We acknowledge that there are barriers to changes in practice as junior medical officers are time-pressured in any given health care setting, hence our intervention engaged their immediate seniors rather than relying on pharmacy staff to enforce change. Regardless of location, this is a tool to potentially better ensure each patient’s primary junior medical officer better understands their care by taking full responsibility for their patient’s medications.

In conclusion, medication chart errors continue to occur on an unacceptable scale. We have shown that a multi-intervention including education of junior doctors, visual reminders and midweek re-charting are effective in improving charting compliance and reducing the rate of charting errors. We advise that a larger study is now conducted using the same multi-intervention MOWER strategy in different ward settings to evaluate feasibility and sustainability of this intervention.