Effect of providing feedback and prescribing education on prescription writing: An intervention study.

BACKGROUND/
OBJECTIVE: Accurate medication prescribing important to avoid errors and ensure best possible outcomes. This is a report of assessment of the impact of providing feedback and educational intervention on prescribing error types and rates in routine practice.
METHODS: Doctors' prescriptions from selected wards in two tertiary hospitals in central Nigeria were prospectively reviewed for a 6-month period and assessed for errors; grouped into six categories. Intervention was by providing feedback and educational outreach on the specialty/departmental level at one hospital while the other acted as the control. Chi-squared statistics was used to compare prescribing characteristics pre- and post-intervention.
RESULTS: At baseline, error rate was higher at the control site. At the intervention site, statistically significant reductions were obtained for errors involving omission of route of administration (P < 0.001), under dose (P = 0.012), dose adjustment in renal impairment (P = 0.019), ambiguous orders (P < 0.001) and drug/drug interaction (P < 0.001) post intervention though there was no change in mean error rate post intervention (P = 0.984). Though House Officers and Registrars wrote most prescriptions, highest reduction in prescribing error rates post intervention was by the registrars (0.93% to 0.29%, P < 0.001).
CONCLUSION: Writing prescriptions that lacked essential details was common. Intervention resulted in modest changes. Routinely providing feedback and continuing prescriber education will likely sustain error reduction.

Introduction

Accurate medication prescribing is an important process in ensuring the best possible outcomes in the treatment and management of diseases. A prescription should clearly communicate with a pharmacist or dispenser what therapy a particular patient is to get, how much of a specific medicine should be taken, how often and for how long. It should also clearly identify the prescriber, be signed in ink and be dated.[1] The illegibility of the prescription or the omission of any of these details could result in misinterpretation, medication errors and adverse drug events.[2]

Studies examining the quality of prescribing by Nigerian medical practitioners have reported a need to improve prescribing quality.[234] Writing of prescriptions that lacked appropriate details and other errors were common though may not always result in actual adverse outcomes for patients.[567] However, when prescribers involved in potentially serious prescribing errors were interviewed, most stated that they were not aware of having made any errors in the past.[8] Such a lack of awareness may be because the prescribing, dispensing, monitoring and administration of medicines involve other persons of different professions in the continuum of patient care. While this can be regarded as a safety feature in increasing the chance of an error being identified by someone other than the original prescriber, a consequence is that once an error is identified and resolved, giving a feedback to the prescriber takes a lower priority. Hence, prescribers rarely have the opportunity to learn from their prescribing lapses.

Provision of audit and feedback about prescribing has been found to be useful in influencing health professional behavior and performance.[91011] In the audit and feedback process, an individual’s professional practice or performance is measured and then compared to professional standards or targets. The result is then fed back to the individual either verbally, in writing or both. It is often used together with other interventions such as reminders, academic detailing or educational outreaches. In a review of studies that evaluated educational outreach visits (EOV), some researchers[1213] reported that outreach visits consistently provided small changes in prescribing which might be potentially important when hundreds of patients are affected. They concluded that multifaceted interventions incorporating EOV were generally effective in improving appropriate care and prescribing when compared to no intervention, but more mixed effects were observed for educational outreach as a single intervention.

An initial study assessed prescriptions and explored the causes behind lapses in prescriptions written by junior doctors.[8] This study evaluated the impact of an intervention on reducing prescribing errors. The objective was to determine the effectiveness of an educational outreach incorporating audit and feedback in improving prescription writing among doctors. It also highlights the applicability of a multifaceted intervention in altering some types of prescribing errors.

Subjects and Methods

Study setting

The study took place during a 15-month period (July 2011–September 2012) at two tertiary hospitals in central Nigeria. The hospitals, apart from providing referral services to the numerous primary and secondary health care facilities within the Federal Capital Territory of Nigeria and beyond, also serve as institutions for training of health personnel offering undergraduate, graduate, postgraduate and residency programs.

These two tertiary hospitals were chosen for the diversity of the patient population, the acuity and academic environment provided for training and molding of doctors and other medical/nonmedical personnel, the consistency of medication/pharmacy service and data availability.

Data collection

The study involved a prospective review of in-patient prescriptions generated from selected wards (medical and pediatric) in the study hospitals. On one day each week, for a period of six months, pharmacists, who had been previously trained with reference to the study requirements, assessed newly written prescriptions for prescribing errors that met the study definition.[5] All prescriptions were cross-checked by the lead investigator (a clinical pharmacist with over 10 years work experience in the study hospital). Prescriptions, using data collection forms, were assessed for prescriber’s rank, name and signature, and elements specifying a well-written prescription such as medication name, strength, dosage, frequency, route, duration, directions for use and additional labeling instructions; use of abbreviations and prescribing of drugs that could interact adversely. Access to patients’ medical notes and other records provided opportunity to evaluate pertinent information such as transcription errors and dosage adjustment in renal/liver impairment as a way of determining correctness of prescribed drug dosages.

After 6 months of initial data collection, the intervention was formulated based on the information derived.

Prescriber identification

Ranks of doctors who wrote prescriptions were determined by their names and signatures. It is usual practice for doctors to title their records in patients’ notes by such terms as House Officer ward round; Registrar ward round; Senior Registrar ward round etc. Sometimes questions were asked from other clinicians in the wards to ascertain a prescriber’s identity when this was unclear. Prescribers that were unidentifiable by signatures were grouped under “prescriber not identified.”

Intervention

The intervention applied involved educational outreach encompassing structured teaching and feedback sessions, undertaken once, with the prescribers in the concerned departments. The feedback report, conducted by the lead investigator, consisted of power-point graphical presentations of prescribing error summaries identified from the baseline survey. It included list of types and examples of errors, the medications involved and teaching on the principles of writing a proper prescription along with commentaries on responsibility and adequate supervision. This provided opportunity for interaction with the prescribers in small groups, at the departmental level. All cadres of prescribers were present for the feedback sessions viz., Consultants, Senior Registrars, Registrars, Medical Officers, House Officers. Other professionals such as Nurses and Pharmacists were also present. The same educational content was used in all cases and lasted between 30 and 40 minutes.

This was followed 10–12 weeks later by another 6 months survey. Prescribing error types, rates and prescribers involved before and after the intervention were compared and analyzed to determine the impact of the intervention. The intervention was applied to the prescribers in one hospital (intervention site) while the other hospital acted as control site. The ethics committee of both hospitals approved the study.

Data analysis

Prescriptions from both intervention and control sites were assessed after the intervention, as in the baseline survey. Data were analyzed using the Statistical Package for the Social Sciences (SPSS) software version 17 (SPSS Inc., Chicago, Ill, USA). The error rate was calculated by dividing the number of errors detected by the total number of prescribed items. Chi-squared statistics was used to compare prescribing characteristics pre- and post-intervention. Statistical significance was placed as a P < 0.05.

Results

Impact of the intervention on in-patient prescription error types and rates

At baseline, 1824 and 2065 in-patient prescription orders were evaluated at the intervention and control sites respectively. Baseline prescribing errors were determined and grouped into categories for the six pre-intervention months and compared with prescribing error rates post-intervention [Table 1]. Writing of prescriptions that lacked necessary details, such as omitting to state the duration of prescribed medicines, route, and frequency of administration and dosage issues, was observed at both sites. Other errors were the use of unsafe abbreviations and prescribing drugs that could adversely interact.

Table 1

Pre- and post-intervention error types and rates at the study sites

Type of errors n=Total number of orders with errors	Control				Intervention
	Error rate Pre-I (%) (n=270)	Error rate Post-I (%) (n=347)	Percentage difference	P	Error rate Pre-I (%) (n=105)	Error rate Post-I (%) (n=99)	Percentage difference	P
Incomplete
Omission of duration/stop date	8.86	11.05	−2.19	<0.001	1.97	3.26	−1.29	<0.001
Dose/frequency omitted*	0.77	0.11	0.66	<0.001	0.11	0.17	−0.06	0.222
Drug name omitted	0.10	0.00	0.10	0.002
Route omitted	0.05	0.11	−0.07	0.105	0.33	0.06	0.27	<0.001
Direction for use omitted**	0.05	0.15	−0.10	0.021	0.00	0.06	−0.06	0.016
Dosing
Dose inappropriate (underdose)	1.21	0.38	0.83	<0.001	0.44	0.23	0.21	0.012
Dose inappropriate (overdose)	0.44	0.26	0.17	0.041	0.27	0.35	−0.08	0.342
Dose adjustment in renal impairment	0.00	0.08	−0.08	0.006	0.05	0.00	0.05	0.019
Ambiguous
Ambiguous order***	0.10	0.04	0.06	0.107	0.22	0.00	0.22	<0.001
Interaction
Drug/drug interaction^	0.58	0.11	0.47	<0.001	1.21	0.35	0.86	<0.001
Abbreviation
Unsafe abbreviation	0.82	0.76	0.07	0.596	0.38	0.52	−0.14	0.142
Others
Indicated drug not given	0.00	0.04	−0.04	0.051
Wrong route	0.10	0.00	0.10	0.002	0.00	0.06	−0.06	0.016
Therapeutic duplication^^	0.00	0.04	−0.04	0.051	0.33	0.47	−0.14	0.126
Total	13.08	13.13	−0.05	0.912	5.76	5.76	−0.01	0.984

*Prescribing a medicine that can be obtained in more than one strength and omitting to state the needed dose. Example: Writing an order for amoxicillin capsules without giving the dose needed, **Prescribing an eye drop such as gtt. chloramphenicol b.d. without stating on which eye needed and/or frequency of use, ***Writing orders for medicines such as vitamins or haematinics without stating which is intended, ^Refers to the prescribing of two or more medicines whose interaction could be potentially serious and where combined administration should be avoided. Example: Clopidogrel and omeprazole, ^^Therapeutic duplication refers to prescribing of two medicines that belong to the same class and have same therapeutic effects. Example Omeprazole and rabeprazole (both are proton pump inhibitors)

Prescriptions assessed after the educational intervention from both intervention and control sites were 1718 and 2643 respectively. Overall, there was no change in prescribing error rates post-intervention (5.76% to 5.76%; P = 0.984) at the intervention site. However, statistically significant error reductions were obtained for errors involving omission of route of administration (P < 0.001), underdose (P = 0.012), dose adjustment in renal impairment (P = 0.019), ambiguous orders (P < 0.001) and drug/drug interaction (P < 0.001) while statistically significant increases was observed in prescriptions that omitted frequency for use (P = 0.222) and overdose (P = 0.342).

At the control site, baseline prescribing errors 270/2065 (13.08%), increased to 347/2643 (13.13%) post-intervention. This was not statistically significant (P = 0.912).

Impact of the intervention on prescriber category

At the intervention site, junior doctors (HOs and Registrar) were responsible for writing most of the prescriptions generated in the in-patient, pre-intervention as well as post-intervention [Table 2]. The prescriber category with the highest reduction in prescription error rates post-intervention was the registrars (0.93% to 0.29%, P < 0.001). Error rate by House Officers increased from 3.51% to 4.48%. The unidentified category of prescribers gave a reduction in error rate from 1.32% to 0.99%, and this was statistically significant (P = 0.032).

Table 2

Pre- and post-intervention prescriber category error rates at the study sites

Physician	Control				Intervention
	Error rate Pre-I (%) (n=270)	Error rate Post-I (%) (n=347)	Percentage difference	P	Error rate Pre-I (%) (n=105)	Error rate Post-I (%) (n=99)	Percentage difference	P
House officer	7.70	8.44	−0.74	0.066	3.51	4.48	−0.97	<0.001
Registrar	2.08	1.25	0.83	<0.001	0.93	0.29	0.64	<0.001
Senior registrar	0.39	0.49	−0.10	0.267	0.00	0.00	0.00
Consultant	0.24	0.00	0.24	<0.001	0.00	0.00	0.00
Prescriber not identified	2.57	2.72	−0.16	0.490	1.32	0.99	0.33	0.032
Total	13.08	13.13	−0.05	0.912	5.76	5.76	−0.01	0.984

Similarly, junior doctors wrote the majority of the in-patient prescriptions at the control site pre- and post-intervention respectively [Table 2]. Statistically significant reduction in error rates were observed for registrars and consultants (P < 0.001) only. Total error rate reduction in the study period was not statistically significant (P = 0.912).

Discussion

A previous work undertaken in the study hospital[8] reported that potential factors identified from interviews with prescribers as risks for error causation included a lack of awareness of making prescribing errors. Prescriptions were usually written quickly and hampered by interruptions, with junior doctors hardly having time to cross-check their prescriptions before returning to uncompleted tasks. Consequently, an audit and feedback process combined with educational outreach was employed as intervention to improve prescribing outcomes with the notion that this would result in a reduction of such errors.

At baseline, prescribing error rates at the intervention site was lower than at the control site. Despite this, there was a similarity in the trend of error types. Omitting to write a stop date for ordered medicines ranked highest followed by incidences where drugs that could interact adversely were administered concurrently. Inappropriate dosing errors of under/overdose and use of unsafe abbreviations in writing medication orders were also observed. The intervention was expected to result in improvement of quality of prescriptions for effective comparison.

Statistically significant improvement in the writing of prescriptions was observed for intervention site prescribers specifically in areas of writing route of administration of drugs ordered, writing non-ambiguous orders, checking for drug interactions and appropriate dosing of medicines [Table 1]. However, when compared to the results obtained from the control site, there appear to be some confounding factors affecting the interpretation. For instance, statistically significant reduction in prescribing error rates was observed for registrars in both the control and intervention hospitals.

House Officers and Registrars were responsible for writing the bulk of in-patient prescriptions as revealed in this study. The intervention had minimal impact on the prescribing error rates of House Officers, as a result of their high turn-over at ward levels due to their intra-specialty rotations, quarterly inter-specialty rotations and bi-annual turn-over in the hospital. As the study progressed, it was noted that House Officers who received the audit and feedback on their prescribing at a particular clinical specialty were afterward, rotated to other units outside the study wards making it unfeasible to follow-up on their subsequent prescribing while new House Officers reported at the study wards. The period of the study also coincided with the recruitment of new House Officers to the hospitals. This was a major drawback as it was not possible to control these established processes. However, for the Registrars who rotate less frequently than House Officers and thus are more “resident” in the wards, the intervention appeared to have had some positive impact.

Researchers[1415] have demonstrated the effectiveness of intervention methods incorporating audit and education when the prescribing environment such as the prescribers and wards were well-controlled. Franklin et al.[10] reported that U.K Clinicians found provision of feedback to be useful and acceptable.

Overall, there was a modest reduction of some prescribing error types. Results from other studies lend support to the notion that the effectiveness of audit and feedback in improving prescribing is enhanced when it is provided more than once, is delivered both in verbal and written formats and includes an action plan or explicit targets.[1116]

Limitation

This study appeared to have been affected by some confounding factors that have implications for the interpretation of the data. Some changes resulting from the intervention was obscured due to the difficulties in identifying some prescribers from their signatures. Also, was the inability to control the prescribing environment (pre/post-intervention) due to the high turn-over of junior doctors (House Officers) in the study wards. Peculiarities of the hospital dynamics may explain the similarity in the effects observed in prescribing error reduction by registrars in the control site despite no intervention.

The need for further research is evident to evaluate the effect of the educational intervention in more strictly controlled settings and to employ a combination of intervention methods. The sustainability of changes arising from such interventions in the short and long term should be assessed.

Conclusion

There is need for improvement in prescription writing by doctors as several required details are omitted. This study has tested the feasibility of providing feedback and educational outreach on prescribing errors identified in prescriptions to prescribers in their clinical specialties. The intervention resulted in statistically significant reduction in omission of some details required in prescriptions and improvements in prescription writing by a category of prescribers.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.