Designing a continuous data-driven feedback and learning initiative to improve electronic prescribing: an interdisciplinary quality improvement study

Introduction

The WHO Global Patient Safety Challenge aims to reduce severe avoidable medication-related harm by 50% by 2023[1]. Research suggests that providing timely, trusted feedback that incorporates relevant action can improve practice. However, a key barrier is lack of prescribing error data. Hospital electronic prescribing (EP) data may help address this gap.

Aims

To explore approaches for continuously monitoring medication safety signals using existing or new EP data, and to deliver personalised prescribing feedback and learning to improve patient safety.

Methods

We conducted a feasibility study (November 2019 - February 2020) on a 28-bed adult gastroenterology. This ward was chosen because of a high prescribing error rate. All foundation year 1 and 2 doctors, and pharmacists on the ward, participated in the study. The study team comprised pharmacists, doctors, quality improvement experts and clinical analysts, and used a quality improvement approach to design and test (i) methods for extracting electronic data to calculate prescribing accuracy rates, (ii) ways to refine a paper-prototype of an electronic pharmacists’ interventions form, (iii) potential digital medication safety indicators, and (iv) approaches for feedback for doctors to augment existing verbal feedback from pharmacists. Data were documented in accordance with local information governance and analysed using Excel. Acceptability and usability was assessed through verbal feedback from participants during weekly huddles. Outcome measures: feasibility of using EP to determine prescribing accuracy, user acceptability and usability of data collection, feedback and learning by pharmacists and doctors. We also measured changes in prescribing accuracy rate, pharmacists’ interventions, and quality of prescribing for targeted problematic medications.

Results

Extracting EP data required multiple data linkages to be configured and validated, and not all required data were available. Potential digital medication safety indicators: utility of the reason code ‘prescribed in error’ and actions by pharmacists to modify medications were limited by underuse and lack of data granularity. After testing different ways to extract relevant EP data, we eventually used a combination of EP and manual retrospective review of electronic patient records to determine prescribing accuracy rates. An intervention form was redesigned to tally interventions and capture details for contextual learning for email feedback to doctors and weekly prescribing improvement huddles. Doctors reported emails as timely and helpful for gaining new prescribing- and system-related knowledge. Pharmacists reported intervention data as providing invaluable evidence to drive improvement.

Statistical process control charts showed no special cause variation around a mean prescription accuracy rate of 98% for inpatient orders, and 87% for discharge orders. By contrast, pharmacists recorded a mean of 10 interventions/day with 7 special cause variation (above upper control limit of 19) in the first two months. Omission of venous thromboembolism prophylaxis was identified as a priority medication issue. Specific prescriber- and system-based improvements were suggested (Jan 2020), some implemented (Feb 2020) and others fed back to the thrombosis committee (Feb 2020).

Conclusion

Harnessing the potential of EP data to improve medication safety requires the workforce to have a deeper understanding of the EP data structure and processes. Using a quality improvement approach, we developed a feedback and learning model that is acceptable and useful to pharmacists and doctors. Further research should explore adapting the approach to other clinical areas.

Reference

1. Sheikh, A., Dhingra-Kumar, N., Kelley, E., Kieny, M. and Donaldson, L. The Third Global Patient Safety Challenge: Tackling Medication-Related Harm. Bulletin of the World Health Organization. World Health Organisation. 2017;95:546-546A.

Relationship of current and past smoking to mortality and morbidity in patients with left ventricular dysfunction

OBJECTIVES

The aim of this study was to evaluate the impact of smoking in patients with left ventricular dysfunction.

BACKGROUND

The impact of smoking in patients with left ventricular dysfunction has not been well-studied.

METHODS

We compared the incidence of death, hospitalization due to heart failure and myocardial infarction (MI) in current smokers to ex-smokers of < or =2 years and ex-smokers of >2 years duration to never-smokers among participants of the Study Of Left Ventricular Dysfunction (SOLVD) Prevention and Intervention trials. Participants all had left ventricular ejection fraction (LVEF) <35% and follow-up was over a mean of 41 months.

RESULTS

Complete smoking status and outcome data were available in 6,704 subjects. There were 1,562 current smokers, 1,317 ex-smokers of < or =2 years, 2,354 ex-smokers of >2 years and 1,471 never-smokers. After adjusting for baseline differences of age, LVEF, race and etiology of heart failure, current smoking was associated with a significantly increased all-cause mortality (relative risk [RR]: 1.41, 95% confidence interval [CI]: 1.25 to 1.58, p < 0.001) compared with ex-smokers and never-smokers. The incidence of death or recurrent congestive heart failure requiring hospitalization or MI was significantly greater (RR: 1.39, 95% CI: 1.26 to 1.52, p < 0.001) in current smokers compared with ex-smokers and never-smokers. There were no significant differences in the number of deaths or hospitalizations due to heart failure between ex-smokers and never-smokers. This effect was consistent across both the SOLVD Prevention and Treatment trials.

CONCLUSIONS

Current smoking is a powerful independent predictor of morbidity (recurrent heart failure and MI) and mortality in patients with left ventricular dysfunction. Quitting smoking appears to have a substantial and early effect (within two years) on decreasing morbidity and mortality in patients with left ventricular dysfunction, which is at least as large as proven drug treatments recommended in patients with left ventricular dysfunction.

Cardiovascular and cancer mortality among Canadians of European, south Asian and Chinese origin from 1979 to 1993: an analysis of 1.2 million deaths

BACKGROUND

Cardiovascular disease and cancer are important health problems worldwide, yet our knowledge of these conditions is derived principally from populations of European descent. To investigate ethnic variations in major causes of death in Canada, the authors examined total and cause-specific mortality among European, south Asian, and Chinese Canadians.

METHODS

Canadians of European, south Asian and Chinese origin were identified in the Canadian Mortality Database by last name and country of birth and in the population census by self-reported ethnicity. Age-standardized death rates by cause, per 100,000 population, were calculated for ages 35 to 74 years from 1979 to 1993 and in 5-year intervals grouped around census years (1979/83, 1984/88 and 1989/93).

RESULTS

Rates of death from ischemic heart disease were highest among Canadians of south Asian origin (men 320.2, women 144.5) and European origin (men 319.6, women 109.9) and were markedly lower among Canadians of Chinese origin (men 107.0, women 40.0); the rates declined significantly in all 3 groups over the study period. Rates of death from cerebrovascular disease were relatively low and showed less ethnic variation (Canadian men of European, south Asian and Chinese origin 49.5, 47.0 and 45.8 respectively; Canadian women of European, south Asian and Chinese origin 34.8, 39.0 and 42.2 respectively) and declined similarly in all groups over time. Rates of death from cancer were highest among Canadians of European origin (men 343.6, women 236.2), intermediate among those of Chinese origin (men 258.1, women 161.6) and lowest among those of south Asian origin (men 122.3, women 131.3). Over time, cancer mortality increased in Canadians of European origin but remained constant or declined in those of south Asian and Chinese origin.

INTERPRETATION

Substantial differences exist in rates of death from ischemic heart disease and cancer among European, south Asian and Chinese Canadians.

Increased winter mortality from acute myocardial infarction and stroke: the effect of age

OBJECTIVES

We examined seasonal variations in mortality from acute myocardial infarction (AMI) and stroke by age using 300,000 deaths in the Canadian Mortality Database for the years 1980 to 1982 and 1990 to 1992.

BACKGROUND

The effect of age on environmental determinants of AMI and stroke is not well understood.

METHODS

Seasonal variations were analyzed by month and for the four seasons (winter beginning in December). A chi-square test was used to test for homogeneity at p < 0.01, and relative risk ratios (RRs) for high and low periods were determined in relation to the overall mean. For each of four age subgroups, the magnitude of the seasonal variation was reported as the difference in mortality between the highest and lowest frequency seasons.

RESULTS

By month, AMI deaths were highest in January (RR = 1.090) and lowest in September (RR = 0.904), a relative risk difference of 18.6%. The seasonal mortality variation in AMI deaths (winter vs. summer) increased with increasing age: 5.8% for <65, 8.3% for 65 to 74, 13.4% for 75 to 84 and 15.8% for >85 years (p < 0.005 for trend). Stroke mortality peaked in January (RR = 1.113) and had a trough in September (RR = 0.914), a relative risk difference of 19.9%. The seasonal variation in stroke mortality also increased with age. Seasonal variations were not seen in those aged <65 years, compared with 11.6% for 65 to 74, 15.2% for 75 to 84 and 19.3% for >85 years (p < 0.005 for trend).

CONCLUSIONS

The elderly demonstrate a greater winter increase in AMI and stroke mortality than younger individuals. An understanding of these seasonal patterns may provide novel avenues for research in cardiovascular disease prevention.

Classifying ethnicity utilizing the Canadian Mortality Data Base

The study of ethnic differences in disease is a methodological challenge as ethnicity is often not identified in existing datasets and surrogate measures need to be used. We have developed a novel methodology combining last name and country of birth to study mortality patterns of Canadians of South Asian (SA) and Chinese (CH) ethnic origin and have compared death rates among SA, CH, and White (WH) Canadians.

METHODS

SA and CH were identified in the Canadian Mortality Data Base (CMDB) using the last name and country of birth of the deceased. Records of people who had been born in countries with large South Asian and Chinese populations (e.g. India, Pakistan, China, Hong Kong) were selected and manually screened by last name. A name directory was then created of distinct South Asian and Chinese names and this directory was used to search all other records in the CMDB for SA and CH deaths. Where necessary, other identifying characteristics such as first name and parents' last name were also used. Population counts were obtained from the Census self-reported question on ethnicity for SA and CH. WH were identified as non-immigrant Canadians who were neither SA nor CH. The method of assigning ethnicity in the CMDB and Census were assessed for comparability and issues of validity and reliability were addressed.

RESULTS

Using this method, 10,989 SA and 21,548 CH deaths were identified. There was marked heterogeneity in birthplace, with only 56% of SA born in South Asia and only 74% of CH born in Greater China. Last names had high validity for self-reported ethnicity in a population sample of SA and were highly reproducible. Mortality rates varied dramatically between groups studied. SA and WH had high rates of ischemic heart disease while stroke mortality was similar among all three groups. Cancer death rates were high in CH and WH and much lower in SA.

CONCLUSION

Last names and country of birth can be used to determined ethnicity of SA and CH with validity and reliability, and leads to a more accurate classification than country of birth alone. The contrasting patterns observed in mortality from major causes of death suggest many interesting hypotheses for further study.