Abstract
The aim of this study was to determine the proportion of near-miss dispensing errors in hospital pharmacies in Japan. A prospective multi-center observational study was conducted between December 2018 and March 2019. The primary objective was to determine the proportion of near-miss dispensing errors in hospital pharmacy departments. The secondary objective was to determine the predictive factors for near-miss dispensing errors using multiple logistic regression analysis. The study was approved by the ethical committee at The Institute of Medical Sciences, University of Tokyo, Japan. A multi-center prospective observational study was conducted in 20 hospitals comprising 8862 beds. Across the 20 hospitals, we assessed data from 553 pharmacists and 53039 prescriptions. A near-miss dispensing error proportion of 0.87% (n = 461) was observed in the study. We found predictive factors for dispensing errors in day-time shifts: a higher number of drugs in a prescription, higher number of quantified drugs, such as liquid or powder formula, in a prescription, and higher number of topical agents in a prescription; but we did not observe for career experience level for clinical pharmacists. For night-time and weekend shifts, we observed a negative correlation of near-miss dispensing errors with clinical pharmacist experience level. We found an overall incidence of near-miss dispensing errors of 0.87%. Predictive factors for errors in night-time and weekend shifts was inexperienced pharmacists. We recommended that pharmacy managers should consider education or improved work flow to avoid near-miss dispensing errors by younger pharmacists, especially those working night or weekend shifts.
INTRODUCTION
Medication errors are a global safety concern. It has been reported that a total of 153502 errors in hospital settings in Japan in January to March 2021, approximately 30.8%, were medication related errors.1) Medication related errors frequently occur on clinical wards when patients or medical staff handle or prepare medicines. However, most hospital medical staff assume medicines supplied from the pharmacy are correct.
James reported dispensing error in a systematic review of data from the U.K. and U.S. That reported unpreventable dispensing errors for 0.008–0.02% and preventable one for 0.11–2.7% in U.K. That also reported in U.S. for 0.06–18% and 0.75%, respectively.2) Heinrich’s Theory of Accident estimates that each serious injury has 300 minor incidents preceding it. It is therefore important to learn from medication incidents and errors. In addition, it is often possible to predict future errors by analyzing near-miss data. However, information on near-miss dispensing errors is lacking.
The proportion of near-miss errors may be decreased over time for reasons, such as 1) using accurate dispensing support technologies, such as barcode-checking, robotic-dispensing systems, etc.3–8) and 2) developing a strong medication safety culture. Collection of data, such as those related to incidents and medication errors, calculating the error rates, and analyzing for types of error, reasons for errors, time and dispensing processes, is not difficult, because hospitals and pharmacies continuously maintain data on medication errors that affected or potentially affected patients’ safety via incident reports.9–12) However, current information on near-miss errors is often insufficient because such errors are usually corrected and eliminated from any records through the routine work process, such as the final stage counter-checking.2,13–17) Therefore, proper collection of near-miss error data requires a well-designed study. It was on this basis that we planned our multi-center prospective observational study of hospital pharmacy dispensing errors in Japan.
MATERIALS AND METHODS
Study DesignA multi-center prospective observational study was conducted in 20 hospitals, which comprised 8862 beds across 8 university hospitals, 8 acute phase hospitals, and 4 chronic phase hospitals in Tokyo, Japan (Supplementary Table 1). The study protocol was developed by a working group during a one-day meeting of the study investigators. As this observational study required no patient information, institutional review board approval was not required under Japanese regulations; however, in consideration of the social risk of this research, we sought and received approval for the study protocol from the institutional review board of The Institute of Medical Science Hospital, The University of Tokyo.
Data Collection and Study ProcedureA one-week data collection period was conducted in each hospital pharmacy department between December 2018 and March 2019. The period of April was avoided to ensure no data quality issues, due to newly qualified pharmacists often starting employment in the month of April in Japan. Study investigators were accepted after completing a one-day training session explaining the data collection methods according to the study protocol. At each hospital pharmacy department, investigators collected daily data for the consecutive 7 d from Monday to Sunday, avoiding public holiday weeks. Near-miss dispensing error information was collected in each prescription-based record at the pharmacy. Prescriptions for medicines administered by injection were excluded in our study. Data monitoring was conducted at the data center within 7 d after the data collection in each hospital to confirm data reliability. Institutional information for each hospital was collected after the study period was complete.
Study ObjectivesThe primary objective was to find the proportion of near-miss dispensing errors in a hospital pharmacy setting. The secondary objectives were (1) to find the types of near-miss dispensing errors, and errors through the recording of environmental factors, and (2) to find any predictive factors for near-miss dispensing errors by comparing with control prescriptions.
Definition of Dispensing ErrorData for near-miss dispensing errors was collected by each investigator after reporting from the double-checking pharmacists on duty. Briefly, when a pharmacist finished dispensing, a second pharmacist checked the prescription for its content, quantity, and other pharmaceutical requirements according to the patient’s conditions to determine possible mistakes. If an error was deemed to have occurred, it was registered in handwriting on a near-miss information reporting form. The causes of these dispensing errors were confirmed by the pharmacist making the near-miss error.
A near-miss error was defined as that found by the second pharmacist, while an incident was defined as an error found either before or after the administration of the prescribed medication to a patient by medical staff or patient. Several previous reports have used as a defined denominator, the number of items, but in our study we defined the denominator as the number of prescriptions. The number of items on a prescription was included in our statistical analysis to analyze predictive factors for near-miss dispensing errors.
Data SourceFor each near-miss or incident related prescription, two control prescriptions were also collected randomly in each hospital in each study day for each shift. Data were collected as to whether the patient was an in- or out-patient, time of occurrence of the error, years of experience of the pharmacist, and the primary error type (such as incorrect drug, incorrect number of drugs, correct drug but incorrect amount, incorrect weighing of powder, use of wrong patients’ label, incorrect identification of patient, forgotten or omitted questioning of the prescribing doctor, forgotten or omitted departmental rule for dispensing). Environmental reasons for the errors as given by the dispensing pharmacist were recorded; these included mix-ups due to similar shaped, look alike or sound alike drugs, a complex procedure for dispensing, and errors due to the drug’s storage position or place. We also recorded the number of drugs in the prescription. Furthermore, we collected this information from prescriptions without error as our control data.
To calculate the proportion of near-miss dispensing errors, the denominator was all prescriptions in each hospital in each study day.
Statistical AnalysisMultivariate analysis was conducted to estimate the risk for error in hospital pharmacy practice. Twice as many prescriptions without error, as compared to the total number of prescriptions with near-miss dispensing errors, were randomly collected in each hospital for each day. Prescriptions with and without near-miss errors were compared using logistic regression analysis. Association with potential predictors for near-miss dispensing errors was built using simultaneous variates for the number of drugs in a prescription, the number of quantified drugs in a prescription, the number of topical agents in a prescription, and career experience for the clinical pharmacist in years. In the analysis, we excluded “forgotten or omission of department original rule for dispensing” because of (1) department original rule did not affect to patients safety, (2) department original rule is difficult as per of generalizability. Interaction was observed between day-time working and night-time or weekend shifts, so we stratified the analysis for 1) day-time working and 2) night-time and weekend shifts. We followed standard methods to estimate sample size for multiple logistic regression, requiring at least ten outcomes for each independent variable to avoid any overfitting.
Data were expressed as medians with ranges, or means ± standard deviations. We used JMP 15® (SAS Institute Inc, Cary, NC, U.S.A.) data analysis software.
RESULTS
Overview of Total Near-Miss Dispensing Errors in the Clinical SettingA total of 825 near-miss dispensing errors was observed in the study of 20 hospitals with 553 pharmacists (Fig. 1, Table 1). We analyzed 2466 prescriptions in total, comprising 1642 control prescriptions from the planned 1650 control prescriptions, and 825 near-miss dispensing error prescriptions, plus 1 prescription lacking sufficient data. From these prescriptions, we excluded 363 prescriptions for the reason “forgotten or omission of department original rule for dispensing” because of (1) department original rule did not affect to patients safety, (2) department original rule is difficult as per of generalizability. Finally, therefore, 461 near-miss dispensing error prescriptions and 1642 control prescriptions were examined in this study. Denominator for study 1 week was totally 53039 prescriptions. The daily total number of prescriptions ranged from 8691 to 9426 on weekdays, 5825 on Saturdays and 1993 on Sundays in the study days (Table 1). We fulfilled the primary objective for our study by calculating dispensing errors at 0.87% of all prescriptions. The incidence for near-miss dispensing errors was observed from 0.87% to 1.06% on weekdays, 0.55% on Saturdays and 0.25% on Sundays (Table 1). Incident was observed in 18 out of 461 near-miss dispensing errors in our study. Near-miss dispensing error was 1.1 (0.1–5.0) % in our study hospitals.
Table 1. Frequency for Near-Miss Dispensing Errors in a Total of 20 Hospitals
| The number of dispensing errors | The number of total prescriptions | Incidence (%) |
|---|
| Monday | 84 | 9146 | 0.92 |
| Tuesday | 82 | 8691 | 0.94 |
| Wednesday | 82 | 9426 | 0.87 |
| Thursday | 96 | 9039 | 1.06 |
| Friday | 80 | 8919 | 0.90 |
| Saturday | 32 | 5825 | 0.55 |
| Sunday | 5 | 1993 | 0.25 |
| Total | 461 | 53039 | 0.87 |
Of the total of 461 near-miss dispensing errors, the most frequent occurrence for 221 (47.9%) was “incorrect number of drugs” (Table 2). The next most frequent errors for 108 (23.4%) were “forgotten or missed inquiry to the prescribing doctor,” “incorrect drug” for 61 (13.2%), “use of wrong patients’ label” for 32 (6.9%), and “incorrect weighing of powder or liquid” for 22 (4.8%) (Table 2).
Table 2. Near-Miss Dispensing Error Categories
| Near-miss error type | The number of errors | Incidence (%) |
|---|
| Incorrect number of drugs | 221 | 0.42 |
| Forgotten or omitted questioning of the prescribing doctor | 108 | 0.20 |
| Incorrect drug | 61 | 0.12 |
| Use of wrong patients’ label | 32 | 0.06 |
| Incorrect weighing of powder or liquid | 22 | 0.04 |
| Correct drug but incorrect amount | 13 | 0.02 |
| Incorrect identification of patient | 4 | 0.01 |
| Total | 461 | 0.87 |
Of the environmental reasons for near-miss dispensing errors for the 463 errors (where multiple reasons were permitted), the most frequently observed were “complexing procedure for dispensing,” for 40 (8.6%), and “drug’s storage position or place” for 36 (7.8%), “sound-alike” for 30 (6.5%), and “look-alike” or miss-leads by “mix-ups due to similar shaped” that contains complexing number of tablets in a sheet, for 14 and 26 (3.0 and 5.6%), respectively. We noted that 317 (68.5%) of the errors did not show any drug related reasons such as busy, tired, depending physical or mental condition, etc. (Table 3).
Table 3. Near-Miss Dispensing Error Categories
| Environmental reasons for near-miss error | The number of reasons | Incidence (%) |
|---|
| No-reasons for drug-related problems | 317 | 0.60 |
| Complexing procedure for dispensing | 40 | 0.08 |
| Errors due to the drug’s storage position or place | 36 | 0.07 |
| Sound alike | 30 | 0.06 |
| Mix-ups due to similar shaped | 26 | 0.05 |
| Look alike | 14 | 0.03 |
| Total | 463 | 0.87 |
Including duplicate answer.
To determine the predictive factors for near-miss dispensing errors, logistic regression analysis was conducted for the near-miss error prescriptions and control prescriptions (Table 4). In the analysis of these data we found the most important predictors for near-miss errors during day-time shifts were: the number of drugs in a prescription (+1 drug, 1.14 [1.10–1.18], p < 0.0001), the number of quantified drugs, such as liquid or powder formula, in a prescription (+1 drug, 1.24 [1.05–1.48], p = 0.0132) and the number of topical agents in a prescription (+1 drug, 1.14 [1.02–1.26], p = 0.0177) (Table 4). In contrast, for night-time and weekend shifts, the predictive factor for dispensing near-miss errors was career experience of the clinical pharmacist (+1 year, 0.90 [0.81–1.00], p = 0.0464) and the number of drugs in a prescription (+1 drug, 1.52 [1.20–1.92], p = 0.0005).
Table 4. Predictive Factors for Near-Miss Dispensing Error
| Near-miss error (n = 435) | Control (n = 1545) | Adjusted odds ratio (95% CI) | p-Value |
|---|
| Day-time working |
| The number of drugs in a prescription, median (range) | 3 (1–22) | 2 (1–18) | 1.14 (1.10–1.18) | <0.0001 |
| The number of quantified drugs in a prescription, median (range) | 0 (0–10) | 0 (0–5) | 1.24 (1.05–1.48) | 0.0132 |
| The number of topical agents in a prescription, median (range) | 0 (0–6) | 0 (0–9) | 1.14 (1.02–1.26) | 0.0177 |
| Career for clinical pharmacist (year), median (range) | 7 (0–45) | 8 (0–45) | 1.01 (1.00–1.02) | 0.1308 |
| Near-miss error (n = 26) | Control (n = 97) | Adjusted odds ratio (95% CI) | p-Value |
|---|
| Night-time and day-time in weekend |
| The number of drugs in a prescription, median (range) | 2.5 (1–10) | 1 (1–8) | 1.52 (1.20–1.92) | 0.0005 |
| The number of quantified drugs in a prescription, median (range) | 0 (0–5) | 0 (0–4) | 1.02 (0.53–1.98) | 0.9437 |
| The number of topical agents in a prescription, median (range) | 0 (0–1) | 0 (0–3) | 0.98 (0.40–2.43) | 0.9701 |
| Career for clinical pharmacist (year), median (range) | 2 (0–13) | 2 (0–36) | 0.90 (0.81–1.00) | 0.0464 |
DISCUSSION
In this study, we first determined that the proportion of near-miss dispensing errors in Japan is approximately 0.87% of all hospital prescriptions, as prospectively assessed from 53039 prescriptions across 20 hospitals.
The main predictive factor for near-miss dispensing errors was a higher number of drugs in a prescription. We found that at night time or weekends, the less experienced clinical pharmacists are higher risk for near-miss dispensing error in our study.
In some countries the responsibility for pharmacy work, such as prescription checks and dispensing, is divided between the pharmacist and pharmacy technicians. In Japan, dispensing is mainly carried out by pharmacists. To calculate the total amount of each prescription drug or one-dose packaging, a unique formula is usually used to quantify liquid or powder formulations for geriatric or pediatric patients based on their body weight or body mass index, because of their difficulty of swallowing tablets. In addition, most pharmacies in Japan have approximately 2000 drugs in stock.
In our study we observed dispensing near-miss errors for 0.87% of all prescriptions written across 20 hospitals (Table 1). In previous reports the range for dispensing error was calculated by different numerators and denominators, our data was comparable to that of previous multiple center retrospective observational studies in Japan which reported errors in the range of 0.6 to 1.3%.2,18) Our well-designed data will provide a proportion for benchmarks for near-miss dispensing errors at hospital pharmacy settings, that will expect for available for brush up for dispensing at hospital pharmacy.
Near-miss dispensing errors were recorded to be “incorrect number of drugs” for 221 (0.42%) of the total 53039 prescriptions (Table 2). Most of these errors were near-miss dispensing errors in sheet shape countable drugs according to the prescription. In Japan, there is no form of dispensing where medicines are provided to patients in bottles as they are in Europe and the U.S., and some patients are provided with one-dose package dispensers. These measured drug doses need to be dispensed according to the prescription and require some calculation by the pharmacist. In addition, sometimes tablets come in sheets of 7, 10, 14, 21, 30, etc., as determined by the pharmaceutical company. However, these various shapes lead to dispensing errors because of the requirement for complexing calculations such as sheet shape containing 4, 5, 6, 7, 10, 14, 21, 30-tablets in Japan according to the prescription.
In contrast, “incorrect drug” errors were observed in 61 cases (0.12% of total prescription) (Table 2). These errors are a cause for considerable concern with the potential for life-threatening problems. Recently, information technology tools, such as bar-codes or picture verification systems, have been applied for one dose packages.8,19) In our study, dispensing supporting machines such as bar-codes readers or picture verification system in countable drugs is equipped in 8 hospital, bar-codes readers in quantified powder drugs is equipped in 16 hospitals, and bar-codes readers in quantified liquid drugs is equipped in 18 hospitals (Supplementary Table 1). In the near future, it is likely that the cost-benefit of these technologies will be acceptable for every hospital or institution.
For 108 of the near-miss errors (0.20% of total prescriptions) the reason given was “forgot to query the doctor about the prescription” (Table 2). This error was essentially an inappropriate dispensing process due to lack of consideration of the patients’ physical condition or some other concern in the prescription. This shows a need to better educate pharmacists in the pharmacology and physiology of various diseases.
For 40 of the near-miss errors (0.08% of total prescriptions), a drug related reason for the error, being “complexing procedure for dispensing,” was reported (Table 3). Recently, several complex drugs (including opioids and other drugs) were released from pharmaceutical companies which need pharmacists to be registered with the pharmaceutical company before being allowed to dispense them, and requires special records to be kept. This adds critical complexity to the dispensing procedures. Pharmacists and pharmaceutical companies should discuss the dispensing procedure for these drugs before they are released.
Our secondary finding that there is a positive correlation for near-miss errors and the number of drugs in a prescription (Table 4) was naturally understandable. To avoid or reduce these errors, the initiating pharmacy could supply targeted technology to the hospital pharmacy department.
The career experience of a clinical pharmacist was not related to the error for day-time shifts, but there was a negative correlation in night weekend shifts, showing the incidence for near-miss dispensing errors out-of-hours is likely to be higher with young pharmacists. In generally, senior pharmacists require not only dispensing but also management or education to young pharmacists in day-time working shift. In addition, senior pharmacists have higher knowledge and stable handling if something happen even in night working-time working solely. One possible reason, senior pharmacist requires no management or education work in night-time working shift, so, therefore, risk for dispensing error was observed in young pharmacist in night-time shift in our study. At night or on weekends, most pharmacies only employ one pharmacist. Error reducing technologies may be valuable in these situations.
Limitation for the StudyThis study has study design-dependent limitations. Data may not be complete, especially for night shifts. For day shifts, the study investigator or other trained pharmacists checked the work of the pharmacist on duty. However, during night or weekend shifts, most hospitals conduct one-pharmacist operations, therefore leading to the possibility that some errors were hidden. In addition, we could not assess the conditions including sick or mental conditions, that may affect quality of dispensing working. This may partially affect the results for this study.
CONCLUSION
This study was the first multi-center prospective research survey for near-miss dispensing errors in Japan. Our result for the incidence for near-miss dispensing errors was 0.87% of all prescriptions. Especially in night-time working, young pharmacists need to be careful for dispensing error.
Acknowledgments
We thank to Yoshihiro Itou (Department of Pharmacy, Eisei Hospital, Medical Corporation Eiseikai Association), Yusuke Hiruma (Department of Pharmacy, Minamino Hospital), Yasutaka Okuno (Department of Pharmacy, Yurin Hospital, Social Welfare Corporation Tokyo-Yurinkai) and Yoshirou Okabe (Department of Pharmacy, Tobu Chiiki Hospital, Tokyo Metropolotan Health and Medical Treatment Corporation) to kindly collect prescription data.
Author Contributions
All authors satisfied the ICMJE recommends 4 criteria.
Conflict of Interest
The authors declare no conflict of interest.
Supplementary Materials
This article contains supplementary materials.
REFERENCES
- 1) “65th Report for Project to Collect Medical Near-Miss/Adverse Event Information (from January 2021 to March 2021),” Japan Council for Quolity Health Care: ‹https://www.med-safe.jp/pdf/report_65.pdf›, accessed 21 December, 2021.
- 2) James KL, Barlow D, McArtney R, Hiom S, Roberts D, Whittlesea C. Incidence, type and causes of dispensing errors: a review of the literature. Int. J. Pharm. Pract., 17, 9–30 (2010).
- 3) Momattin H, Arafa S, Momattin S, Rahal R, Waterson J. Robotic pharmacy implementation and outcomes in Saudi Arabia: a 21-month usability study. JMIR Hum Factors, 8, e28381 (2021).
- 4) Councill EEAW, Axtell NB, Truong T, Liang Y, Aposhian AL, Webber KGI, Zhu Y, Cong Y, Carson RH, Kelly RT. Adapting a low-cost and open-source commercial pipetting robot for nanoliter liquid handling. SLAS Technol., 26, 311–319 (2021).
- 5) Batson S, Herranz A, Rohrbach N, Canobbio M, Mitchell SA, Bonnabry P. Automation of in-hospital pharmacy dispensing: a systematic review. Eur. J. Hosp. Pharm., 28, 58–64 (2021).
- 6) Momo K, Shimano M, Kanezaki Y, Minagawa A, Takagi A, Seino T, Koinuma M. Assessment of “look-alike” packaging designs related to medication errors using information technology. Pharmazie, 74, 310–312 (2019).
- 7) Westbrook JI, Reckmann M, Li L, Runciman WB, Burke R, Lo C, Baysari MT, Braithwaite J, Day RO. Effects of two commercial electronic prescribing systems on prescribing error rates in hospital in-patients: a before and after study. PLOS Med., 9, e1001164 (2012).
- 8) Jessurun JG, Hunfeld NGM, Van Rosmalen J, Van Dijk M, Van Den Bemt PMLA. Effect of automated unit dose dispensing with barcode scanning on medication administration errors: an uncontrolled before-and-after study. Int. J. Qual. Health Care, 33, mzab142 (2021).
- 9) St. Clair B, Jorgensen M, Georgiou A. Incidence of adverse incidents in residential aged care. Aust. Health Rev., 46, 405–413 (2022).
- 10) Mulac A, Hagesaether E, Granas AG. Medication dose calculation errors and other numeracy mishaps in hospitals: analysis of the nature and enablers of incident reports. J. Adv. Nurs., 78, 224–238 (2022).
- 11) Alrowily A, Jalal Z, Abutaleb MH, Osman NA, Alammari M, Paudyal V. Medication errors associated with direct-acting oral anticoagulants: analysis of data from national pharmacovigilance and local incidents reporting databases. J. Pharm. Policy Pract., 14, 81 (2021).
- 12) Adie K, Fois RA, McLachlan AJ, Chen TF. Medication incident recovery and prevention utilising an Australian community pharmacy incident reporting system: the QUMwatch study. Eur. J. Clin. Pharmacol., 77, 1381–1395 (2021).
- 13) Rajah R, Hanif AA, Tan SSA, Lim PP, Karim SA, Othman E, Teoh TF. Contributing factors to outpatient pharmacy near miss errors: a Malaysian prospective multi-center study. Int. J. Clin. Pharm., 41, 237–243 (2019).
- 14) Beso A, Franklin BD, Barber N. The frequency and potential causes of dispensing errors in a hospital pharmacy. Pharm. World Sci., 27, 182–190 (2005).
- 15) Seki Y, Yamazaki Y. Effects of working conditions on intravenous medication errors in a Japanese hospital. J. Nurs. Manag., 14, 128–139 (2006).
- 16) Ashcroft DM, Quinlan P, Blenkinsopp A. Prospective study of the incidence, nature and causes of dispensing errors in community pharmacies. Pharmacoepidemiol. Drug Saf., 14, 327–332 (2005).
- 17) Tseng HY, Wen CF, Lee YL, Jeng KC, Chen PL. Dispensing errors from look-alike drug trade names. Eur. J. Hosp. Pharm., 25, 96–99 (2018).
- 18) Soeda S, Takayanagi R, Watanabe M, Yamada Y. Study of safety evaluation in dispensing of medicines—analysis of relationship between dispensing error rate, inspecting error rate, and malpractice rate—. Yakugaku Zasshi, 137, 589–593 (2017).
- 19) FUJIFILM® Corp. web site of “PROOFIT 1D”: ‹https://www.fujifilm.com/jp/ja/healthcare/pharmacy-and-health-support/pharmacy_assistance/proofit_1d›, accessed 21 November, 2021.
