Errors Associated With Medication Administration by a Nurse During Hospitalisation: A Prospective Observational Multicentric Study

Abstract

ABSTRACT

Aims

To explore all medication administration errors (MAEs) throughout the entire process of medication administration by nurses in the inpatient setting, to describe their prevalence, and to analyse associated factors, including deviation from the good practice standards.

Background

Worldwide, MAEs are very common and regarded as a serious risk factor to inpatient safety. Nurses assume an essential role in the hospital setting during the administration of medications.

Design

The prospective observational study was carried out in accordance with the STROBE guidance.

Methods

This study was conducted in four regional hospitals from June to August 2021. MAEs were collected when nurses administered medications to the adult inpatients during the morning, noon, and evening medication rounds at the internal, surgical, and follow‐up care departments in each hospital over three consecutive days. Direct observation by the multidisciplinary team was employed. MAEs were classified as major MAEs (from the potentially most serious and common to all drug forms), specific MAEs (specific to a drug form), and procedural MAEs (e.g., patient identification, hygiene standards, or generic drug substitution). Predictors of either major MAE or specific MAE frequency were analysed using the generalised linear model and the decision tree model.

Results

Overall, 58 nurses administering medication to 331 inpatients at 12 departments were observed. In total, 6356 medication administrations were observed, of which 461 comprised major MAEs, 1497 specific MAEs, and 12,045 procedural MAEs. The predictors of the occurrence of major MAEs and specific MAEs were the specific hospital, the nurse's length of practice (less than 2 years), and two procedural MAEs (the unclear prescription and the wrong strength).

Conclusions

Non‐adherence to the standard processes in healthcare facilities for prescribing and administering drugs increased the prevalence of severe MAEs. Determinants of MAE occurrence such as incorrect prescriptions or limited experience of nurses should be considered.

Implication for the Profession and Patient Care

The identified determinants of MAE should be considered by hospital stakeholders in their support programs to reduce the level of burden for nurses during medication administration.

Patient or Public Contribution

Neither patients nor public was not involved in the design, data collection, or dissemination plans of this study. The researchers observed nurse care delivery at medical departments acting as passive participants.

Introduction

One of the inherent and common risks in hospital settings is medication errors, which can lead to inappropriate use of medication, causing harm, additional costs, and even death in inpatients. Therefore, the problem of medication errors, potentially occurring in all medication-related activities (prescription, transcription, dispensation, preparation, and administration), ought to be thoroughly assessed as a determinant of patient safety in the healthcare environment (Azar et al. 2023; Ciapponi et al. 2021; Moyen, Camiré, and Stelfox 2008). This issue was emerged by the WHO global initiative from 2017, which aims for a 50% reduction in medication-related harm (Donaldson et al. 2017).

Background

Errors made during medication administration are known as medication administration errors (MAE). Although MAEs are observed the most frequently (above 50% on average), they are still difficult to measure, and therefore not sufficient attention is paid to this topic (Westbrook et al. 2020). Administering medication is a complex process influenced by various factors within the healthcare system organisation. Nurses assume an essential role in hospital settings since the administration of medications is one of their main tasks and responsibilities (Jessurun et al. 2023). The overall prevalence of MAE is high across different health systems. According to two systematic reviews, the median prevalence of MAE per administered drug reached 10.5% (Berdot et al. 2013) and 8.0% (Keers et al. 2013b), respectively. Although the majority of observed MAEs had only minor clinical impact on inpatients, the studies based on direct observation still showed a high prevalence of potentially harmful MAEs, reaching almost 10% (Berdot et al. 2021; Cottney and Innes 2015; Härkänen et al. 2015; Jessurun et al. 2023).

The types and definitions of MAEs vary between studies. Usually, MAEs tend to comprise dose omission, administration of a wrong drug, wrong drug form, wrong dose, unauthorised drug, deteriorated drug, administration of a drug to a wrong patient, as well as wrong drug handling, wrong administration technique, or the wrong route of administration. The wrong time of drug administration is regarded as a less serious MAE by some authors. This error is usually, due to its high prevalence, not included in the observation or excluded from the subsequent analysis (Berdot et al. 2021; Jessurun et al. 2023; Owens et al. 2020). Other MAEs, such as improper or unauthorised drug form modification (especially splitting and crushing), are rarely studied or reported in detail (Fodil et al. 2017; Haw, Stubbs, and Dickens 2007; Westbrook et al. 2024). On the other hand, non-adherence to good practice standards during the process of drug administration is observed in some studies, yet the significance of these errors remains largely unclear (Rishoej et al. 2018). Still, some evidence suggests that they may contribute to MAEs (Westbrook et al. 2011).

To understand the entire context of a prevalence of MAEs, numerous factors should be considered. These factors can be categorised into five main areas: 1. organisational factors (unclear prescriptions, communication, training, lack of guidelines for medication administration, technologies, working environment); 2. nurse-related stress (day or night shifts, day of the week, workload, interruptions, non-standard administration processes); 3. nurse characteristics (knowledge, education, experience, age); 4. patient characteristics; and 5. drug characteristics (route of administration, Anatomical Therapeutic Chemical (ATC) classification) (Al Tehewy et al. 2016; Berdot et al. 2012; Blignaut et al. 2017; Cottney 2014; Feleke, Mulatu, and Yesmaw 2015; Flynn et al. 2012; Frith et al. 2012; Härkänen et al. 2015; Haw, Stubbs, and Dickens 2007; Hung et al. 2015; Hwang et al. 2016; Jessurun et al. 2023; Mekonen, Gebrie, and Jemberie 2020; Nguyen et al. 2014; Raja, Ali, and Sherali 2019; Tissot et al. 2003; Westbrook et al. 2010; Wondmieneh et al. 2020).

Still, evidence of MAEs in Czech hospitals is limited, while no direct observational studies have been published to date.

Aim

The aim of this study was to explore MAEs throughout the entire process of medication administration by nurses in the inpatient setting, to describe their prevalence and to analyse associated factors, including deviation from the good practice standards.

Methods

This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies (von Elm et al. 2008).

Design

The prospective observational study was conducted in four hospitals in one region of the Czech Republic. Data was collected in a three-month research period, from June to August 2021. The observation included the entire process of medication administration by a nurse to inpatients during the morning, noon, and evening medication rounds at the internal, surgical, and follow-up care departments in each hospital over three consecutive days (36 days of observation in total).

Settings

The study was conducted in three medium-sized hospitals (over 300 beds per hospital) and one large university hospital. In all hospitals, basic electronic medical records (EMRs) were used without any electronic support systems. Medications were prescribed by the physician using free-text entries, while the prescription habits might differ in various departments of the same hospital. For medication administration, a paper-based medication administration record (MAR) was printed daily from the EMR, except in one department, where a handwritten MAR was used. Nurses only recorded the administration of the medication. No further details of the administration process (e.g., substitution, splitting, or crushing) were documented in the MAR.

All hospitals used a department-based medication storage system administered by nurses. They were also ordering medications from the hospital pharmacy. However, pharmacists were not involved in the process otherwise. According to hospital policy, during the medication rounds, nurses had to use a medication trolley to administer medications at the bedside of each inpatient.

In the Czech Republic, nurses with all levels of education are authorised to administer drugs to hospitalised patients. Only nurses with secondary education levels cannot administer intravenous drugs. However, this route of administration was not reflected in our study.

Samples

The enrollment of inpatients and nurses was not randomised. However, it was unaffected by the observers. All inpatients present and nurses in service at the department during the observation period were eligible to participate.

For statistical design, the G*Power 3.1.9.7 (Faul et al. 2007, 2009) software was used. According to the systematic reviews (Berdot et al. 2013; Keers et al. 2013b), a total MAE rate of 10% was estimated. Assuming that two MAE types (major and specific MAE), not necessarily occurring equally, were defined and focusing on the event per variable ≥ 10 (Peduzzi et al. 1996) for each MAE, with 30 variables (27 predictors +3 within-subject levels), a total of 5782 medication administrations were designed to be obtained in our study based on three days of observation per department. Finally, after completing 108 medication rounds, data on 6356 medication administrations were collected.

All medication administrations of all hospitalised patients were included in our study except for the administration of individually prepared drugs (compounding); food supplements and infusions; administration outside of observed medication rounds; administration that was not possible to observe (e.g., disagreement of the patient, refusal to enter isolation, ethical reasons).

Ethical Considerations

The study was approved by the Commission for Ethics in Research of the Faculty of Pharmacy in Hradec Králové, Charles University (on 26.6.2019) and in accordance with the Declaration of Helsinki. Participation of the nurses as study subjects was voluntary. They were informed in advance through verbal communication; therefore, they could refuse and be replaced during the observation. According to the legal authority, no additional patient consent was required for this study setting, while patient refusal of observation was fully respected. The observing team did not interfere in the medication administration processes, except if a situation with a potential risk to the inpatient was witnessed. In such cases, according to the ethics committee, the administration process had to be discreetly halted just before medication reached the inpatient.

All data were stored anonymously by alphanumeric code in the database, while neither the patients nor the nurses could be identified retrospectively. Moreover, only data necessary for observation were obtained: inpatient code, year of birth, gender, inpatient room number, and all medications used, including strength, dosage form, and dosage schedule; nurse age, gender, education, and work experience in healthcare and in any given hospital department. The database management was supervised by a company specialising in protecting and storing sensitive data (Sprinx Systems, Czech Republic).

Definitions and Classification of MAE

In this study, MAE (regardless of their cause) was considered as any deviation from the prescriber's order, the Summary of Product Characteristics (SmPC), current guidelines, or the process standards of the given healthcare facilities. Consequently, a unique 3-group classification for MAEs was developed by the authors: major MAEs (in some studies only MAE), the potentially most serious and common to all drug forms; specific MAEs, specific to a drug form including drug administration and food timing; procedural MAEs, including adherence to inpatient identification, hygiene standards, performing a generic drug substitution, unclear prescription, and others. Procedural MAEs included situations that do not necessarily lead to MAEs but may increase the likelihood of major or specific MAEs which may occur as a deviation from the standard for medication administration. The definition and classification of each MAE are presented in Table 1.

TABLE 1 Definitions and classification of medication administration errors.

MAE type	Definition
Major MAE
Wrong drug	Different active substance administered.
Wrong dose	Other than the prescribed dose, or wrong/unapproved dose in case the dose was not prescribed.
Wrong time	Time deviation greater than 15 min.
Wrong inpatient	Administration to a different inpatient.
Wrong drug form	E.g., controlled-release form instead of regular-release form, etc.
Wrong way of use	E.g., wrong inhalation technique, swallowing chewable tablets, etc.
Unordered drug	Administration of extra dose or a drug not prescribed (excl. wrong drug MAE).
Drug omission	A prescribed drug was not administered within the medication round and non-administration was not documented properly. This error excludes all other MAEs.
Deteriorated drug	Administration of expired, or any other unusable drug.
Specific MAE
Inappropriate drug form modification	Splitting or crushing of solid drug form was not in accordance with the SmPC or factual databases, respectively.
Inappropriate food timing	According to SmPC or factual databases, respectively, proper food timing was not adhered to. Applicable for oral drug forms and insulins.
Inappropriate drink	According to SmPC or factual databases, respectively, a drink with interaction potential was used for oral drug administration.
S.c. application error	Error in a syringe angle during the application, aspiration, skin fold, skin disinfection, or wrong application (location or other).
Other specific MAE	Tablet parts were stored and/or later used, the suspension was not well shaken, and errors specific to ophthalmic, transdermal, or other injection drug forms.
Procedural MAE
Unclear prescription	The prescribed name, concentration/strength, or dose was ambiguous and was not verified before administration by the nurse.
Missing identification of an inpatient	The inpatient's identity was checked neither by asking for the name nor by identity bracelet check.
Missing hand disinfection	The nurse did not disinfect her hands before administering drugs to an inpatient, despite having touched a potentially impure surface. The correctness of the disinfection process was not assessed.
Inappropriate hygiene handling	Touching the tablet/capsule with bare hands, using a tablet/capsule which had fallen; using tools that were not clean or not cleaned before use.
Nurse interruption	Interruption during the preparation of an individual drug.
Generic substitution	Voluntary exchange of the drug aiming to use the same active substance from another producer without prescriber's permission.
Wrong strength	Use of different than prescribed strength of the drug.
Missing education of an inpatient	Inpatient was not educated before injectable drug administration.
Administration by a different nurse	Administering by another nurse, who had not actually prepared the drug.
Unchecked drug intake	The nurse did not check drug intake/usage by the inpatient.
Documentation error	No record of drug administration or no administration was created.
Trolley not under the supervision	The trolley with drugs was left unguarded or unlocked.
Missing opening date report	For multiple-dose forms only (e.g., syrups, drops, etc.)

Data Collection Process

Direct observation method was applied for data collection, which was realised within 3–6 multidisciplinary observation teams for each hospital (2–3 observers in a team, including at least one pharmacist and one nurse). To reduce observer heterogeneity, the six-step process was followed: 1. 2-h briefing prior to each data collection by the entire team; 2. use of a pre-printed structured data collection form including the patients' prescribed medication; 3. performing one test observation to become familiar with the environment of the given department 1 day before the actual observation; 4. consensual assessment of observed data within the team immediately after each observation; 5. categorisation and unification of errors by a pharmacist; and 6. final review and approval of the record by the principal investigator (senior pharmacist). Observers recorded data of all hospitalised patients from the MAR into the secured web database created for the purpose of data collection. The database was updated daily, whenever a medication had been changed or an inpatient had been admitted or discharged. The database served both to print a standardised data collection form and to record, categorise, review, or export the data obtained through observation.

Data Analysis

Any single medication administration, including omission, was observed as the standard event, used also as a common denominator. Each medication administration could contain any number of medication errors. However, an omission (i.e., drug omission) excluded other medication errors since it was impossible to observe. For the purposes of MAE assessment, information from SmPC, factual databases (UpToDate, Micromedex, and Drugbank) and process standards of individual hospitals were used. All steps (data collection, data entry, and evaluation) were performed according to a standardised protocol. All medications were categorised following the ATC classification (WHO Collaborating Centre for Drug Statics Methodology 2022).

Data was analysed using Wolfram Mathematica 12 (Wolfram Research Inc., Illinois, USA). The output in the form of quantity and percentage for binomial variables (e.g., the occurrence of a given major MAE/specific MAE/procedural MAE) and in the form of mean ± standard deviation or in the form of median and quartiles for numerical variables (e.g., aggregated quantities of some type of major MAE/specific MAE/procedural MAE) was performed. The sum of relevant medication administration was used as the denominator.

The dependency of either major MAE or specific MAE frequency on the nurse or inpatient characteristics, respectively, was tested using the general linear model. In this model, clustering was considered only at the level of the nurse administering the drugs, while data at higher levels (wards and hospitals) were treated as independent and identically distributed. Results were considered statistically significant if the p-value was below the Sidak-adjusted threshold (Lee 2010):$$$ 1-{\left(1-0.05\right)}^{1/\left(\mathrm{number}\kern0.34em \mathrm{of}\kern0.34em \mathrm{major}\ \mathrm{MAE}+\mathrm{specific}\ \mathrm{MAE}\right)} $$$

For each of the significant dependencies, the impact was evaluated by estimating its effect size using η² and classified as small (η² > 0.01), medium (η² > 0.06), and large (η² > 0.14) according to Cohen's convention (Cohen 1992).

For the dependence of either major MAE or specific MAE frequency on the type of medication (according to the first level of the ATC class), their real occurrence with the probability distribution corresponding to the frequency of a related administration was compared by the Goodness-of-fit test. In this model the Sidak correction was also used.

The dependency of the occurrence of major MAE on procedural MAE was analysed using two methods. First, the generalised linear model with a Bernoulli distribution and logit link function (Fahrmeir and Tutz 1994) was used for estimation and testing the impact of any procedural MAE per se, assuming the other procedural MAE remains constant. Second, the decision tree model with the CHAID algorithm (Hastie, Friedman, and Tibshirani 2009) tested the combinations of procedural MAE with the critical impact on major MAE occurrences. The output in the form of a risk ratio with a 95% confidence interval and the threshold for significance of p-value < 0.05 were used for both models.

Results

In total, 58 nurses administering medication to 331 inpatients at 12 departments of four hospitals were observed. The characteristics of nurses did not differ significantly across hospitals or department types (p > 0.05). Inpatient characteristics differed across hospitals and department types in terms of age and the number of medications used. An overview of both characteristics is shown in Table 2.

TABLE 2 Individual characteristics of nurses and inpatients.

Individual characteristics
Nurses
Total; n	58
Age, years; mean ± SD	35.78 ± 11.67
Education level; n (%)
Secondary education	16 (27.6)
Higher/Vocational education	29 (50.0)
University education (Bachelor's degree)	8 (13.8)
University education (Master's degree)	5 (8.6)
Experience in healthcare total; n (%)
< 2 years	5 (8.6)
2–6 years	21 (36.2)
6–16 years	12 (20.7)
> 16 years	20 (34.5)
Experience in the individual department; n (%)
< 2 years	6 (10.3)
2–6 years	23 (39.7)
6–16 years	19 (32.8)
> 16 years	10 (17.2)
MA per medication round; median (IQR)	20.0 (9.0–41.0)
MA per day; median (IQR)	52.0 (27.5–83.5)
Inpatients in care per day; median (IQR)	8.0 (5.5–15.0)
Inpatients
Total; n	331
Age, years; mean ± SD	72.85 ± 13.63
Gender; n (%)
Male	147 (44.4)
Female	184 (55.6)
MA per day; median (IQR)	8.0 (4.0–12.0)

There were 6356 medication administrations. The majority of medication administrations consisted of solid oral drugs (5382; 84.7%), followed by liquid oral drugs (412; 6.5%), drugs administered subcutaneously (378; 5.9%), drugs inhaled (112; 1.8%), and other drug forms (72; 1.1%). The final sample included 2228 (35.1%) observed (or omitted) medication administrations in hospital A, 1539 (24.2%) in hospital B, 1489 (23.4%) in hospital C, and 1100 (17.3%) in hospital D. Most medication administrations were observed in the follow-up care department (2937; 46.2%), followed by the internal department (2472; 38.9%), and the surgical department (947; 14.9%).

Number and Prevalence of MAEs

In total, 461 major MAEs, 1497 specific MAEs, and 12,045 procedural MAEs were observed. On average, there were 0.61 ± 1.43 major MAE, 1.99 ± 2.05 specific MAE, and 16.02 ± 13.77 procedural MAE observed per patient-day. At least one major MAE occurred in 437 (6.9%) medication administrations; 24 of these administrations had two major MAEs. At least one specific MAE occurred in 1430 (22.5%) medication administrations; 67 of these administrations had two specific MAEs. At least one procedural MAE occurred in 5446 (85.7%) administrations. There were 5690 (89.5%) medication administrations with any MAE. An overview of the MAE prevalence is demonstrated in Table 3.

TABLE 3 Prevalence of individual medication administration errors.

MAE type	MA (denominator)	Number of MAEs	Prevalence (%)
Major MAE
Wrong dose	6356a	127	2.00
Wrong way of use	6356a	104	1.64
Wrong time	6356a	76	1.20
Drug omission	6356a	66	1.04
Wrong drug form	6356a	30	0.47
Wrong drug	6356a	28	0.44
Unordered drug	6356a	15	0.24
Deteriorated drug	6356a	12	0.19
Wrong inpatient	6356a	3	0.05
Specific MAE
Inappropriate food timing	5992b	1179	19.68
Inappropriate drug form modification	5382c	190	3.53
S.c. application error	378d	76	20.11
Inappropriate drink	5794e	37	0.64
Other specific MAE	6356a	15	0.24
Procedural MAE
Missing hand disinfection	6356a	3653	57.47
Missing identification of an inpatient	6356a	2561	40.29
Inappropriate hygiene handling	5382c	1785	33.17
Unchecked drug intake	6356a	1472	23.16
Administration by a different nurse	6356a	766	12.05
Trolley not under supervision	6356a	532	8.37
Generic substitution	6356a	515	8.10
Wrong strength	6356a	290	4.56
Nurse interruption	6356a	223	3.51
Unclear prescription	6356a	166	2.61
Missing education of an inpatient	399f	36	9.02
Documentation error	6356a	27	0.42
Missing opening date report	6356a	19	0.30

Association of Nurse and Inpatient Characteristics With Major and Specific MAEs

Statistically significant associations between major MAE or specific MAE and characteristics of nurses (hospital, department, age, working experience, number of medication administrations per day) or of inpatients (hospital, department, age, gender, number of doses per day) are summarised in Table 4. The most common predictor of the occurrence of major MAEs and specific MAEs was the hospital. However, no hospital was the worst in the occurrence of every major MAE or specific MAE. Another significant predictor of major MAE was the nurse's length of practice (less than 2 years).

TABLE 4 Statistically significant associations between major/specific MAE and individual characteristics of nurses or inpatients (including hospital and department type).

MAE type	Characteristics of nurse/inpatient	Category	Mean of current MAE frequency ± SDa	η2; effect size
Mean change in current MAE frequency (95% CI)b
Major MAE
Wrong drug	Age (nurse)	N/A	+0.0028 per 10 years (0.0001–0.0062)	0.06; small
Total experience in healthcare (nurse)	< 2 years	0.0103 ± 0.1471	0.08; medium
2–6 years	0.0015 ± 0.0066
6–16 years	0.0018 ± 0.0108
> 16 years	0.0054 ± 0.0186
Wrong dose	Hospital (nurse)	Hospital A	0.0161 ± 0.0228	0.12; medium
Hospital B	0.0096 ± 0.0206
Hospital C	0.0366 ± 0.0582
Hospital D	0.0062 ± 0.0228
Hospital (inpatient)	Hospital A	0.0254 ± 0.1165	0.02; small
Hospital B	0.0095 ± 0.0572
Hospital C	0.0387 ± 0.1268
Hospital D	0.0060 ± 0.0328
Wrong time	Hospital (inpatient)	Hospital A	0.0074 ± 0.0538	0.04; small
Hospital B	0.0063 ± 0.0412
Hospital C	0.0002 ± 0.0049
Hospital D	0.0459 ± 0.1638
Wrong inpatient	N/A
Wrong drug form	N/A
Wrong way of use	Hospital (nurse)	Hospital A	0.0426 ± 0.0679	0.20; large
Hospital B	0.0069 ± 0.0226
Hospital C	0.0007 ± 0.0037
Hospital D	0.0034 ± 0.0134
Hospital (inpatient)	Hospital A	0.0377 ± 0.1382	0.03; small
Hospital B	0.0064 ± 0.0471
Hospital C	0.0009 ± 0.0141
Hospital D	0.0054 ± 0.0427
Gender (inpatient)	Male	0.0253 ± 0.1182	0.01; small
Female	0.0069 ± 0.0447
Unordered drug	N/A
Drug omission	N/A
Deteriorated drug	N/A
Major MAE total	Total experience in healthcare (nurse)	< 2 years	0.1313 ± 0.2264	0.08; medium
2–6 years	0.0506 ± 0.0830
6–16 years	0.0387 ± 0.0575
> 16 years	0.0519 ± 0.0842
Hospital (inpatient)	Hospital A	0.0797 ± 0.1953	0.04 small
Hospital B	0.0349 ± 0.1199
Hospital C	0.0571 ± 0.1586
Hospital D	0.0607 ± 0.1737
Specific MAE
Inappropriate drug form modification	Age (inpatient)	N/A	+0.0006 per 10 years (0.0002–0.0011)	0.01; small
Gender (inpatient)	Male	0.0112 ± 0.0526	0.01; small
Female	0.0388 ± 0.1244
Inappropriate food timing	Hospital (nurse)	Hospital A	0.2459 ± 0.2209	0.13; medium
Hospital B	0.1727 ± 0.1099
Hospital C	0.0944 ± 0.0878
Hospital D	0.2398 ± 0.2062
Hospital (inpatient)	Hospital A	0.1915 ± 0.2654	0.04; small
Hospital B	0.1804 ± 0.2342
Hospital C	0.0820 ± 0.1589
Hospital D	0.2056 ± 0.2666
Department type (inpatient)	Internal	0.1949 ± 0.2597	0.02; small
Surgical	0.1582 ± 0.2742
Follow-up care	0.1366 ± 0.1948
MA per day (inpatient)	N/A	0.0151 per dose taken (0.0107–0.0196)	0.03; small
Inappropriate drink	Hospital (inpatient)	Hospital A	0.0011 ± 0.0143	0.01; small
Hospital B	0.0060 ± 0.0338
Hospital C	N/A
Hospital D	0.0146 ± 0.0944
S.c. application error	Department type (inpatient)	Internal	0.0043 ± 0.0496	0.01; small
Surgical	0.0238 ± 0.1427
Follow-up care	0.0224 ± 0.1050
Other specific MAE	N/A
Specific MAE total	Hospital (nurse)	Hospital A	0.3222 ± 0.2687	0.14; large
Hospital B	0.1940 ± 0.1189
Hospital C	0.1251 ± 0.1016
Hospital D	0.2858 ± 0.2254
Hospital (inpatient)	Hospital A	0.2408 ± 0.2834	0.04; small
Hospital B	0.2105 ± 0.2585
Hospital C	0.1185 ± 0.1996
Hospital D	0.2561 ± 0.3011
MA per day (inpatient)	N/A	0.0130 per dose taken (0.0080–0.0180)	0.02; small

Association of ATC Class With Major and Specific MAEs

The prevalence of major MAEs and specific MAE types across ATC classes is shown in Table 5. The major MAE wrong dose was associated with ATC classes A (Alimentary tract and metabolism) and R (Respiratory system). When compared to the average prevalence of this major MAE (2%), there was more than a twofold and nearly fourfold increase detected, respectively. The major MAE wrong time was associated with ATC classes B (blood and blood-forming organs) and J (anti-infectives for systemic use). ATC class C (Cardiovascular system) was found to be the least risky for major MAE drug omission. Of the specific MAE inappropriate food timing, higher than average prevalence was recorded for drugs of ATC classification A (Alimentary tract and metabolism), C (Cardiovascular system), and H (Systemic hormonal preparations, excl. sex hormones and insulins). For specific MAE inappropriate drinks, a higher prevalence was observed for drugs of ATC classification H and J (Anti-infectives for systemic use).

TABLE 5 Prevalence of major and specific medication administration errors across ATC classes.

ATC class	Number of MA includinga/excludigb omissions	Prevalence of major MAE (%)	Prevalence of specific MAE (%)
Wrong drug	Wrong dose	Wrong time	Wrong inpatient	Wrong way of use	Wrong drug form	Unordered drug	Drug omission	Deteriorated drug	Inapp. drug form mod.	Inapp. food timing	Inapp. drink	S.c. application error	Other specific MAE
A	1553	1540	0.59	↑4.23*	0.33	—	↑2.93*	0.13	0.39	0.84	0.13	2.08	↑31.03*	0.13	↑2.67*	0.13
B	662	657	0.61	0.46	↑3.35*	0.15	0.30	0.30	—	0.76	0.30	↓0.30*	↓10.65*	↑1.67*	↑5.33*	—
C	1797	1791	0.34	0.84	0.34	—	↓—*	0.73	0.22	↓0.33*	—	↑5.53*	↑23.56*	0.06	↓—*	0.39
D	4	2	—	—	—	—	—	—	—	50.00	—	—	—	—	—	—
G	69	69	1.45	—	—	—	—	—	—	—	—	2.90	11.59	—	—	—
H	117	116	—	0.86	2.59	—	1.72	—	—	0.85	—	2.59	↑49.14*	↑15.52*	—	—
J	86	86	—	—	↑23.26*	—	—	—	—	—	—	1.16	17.44	↑5.81*	—	—
L	29	28	—	—	—	—	—	—	—	3.45	—	—	25.00	—	—	—
M	190	190	—	—	1.05	—	—	—	0.53	—	—	—	25.79	—	—	—
N	1343	1331	0.30	0.82	1.35	0.15	1.05	0.07	0.15	0.89	—	3.82	↓3.67*	—	↓—*	0.15
R	426	407	—	↑7.86*	—	—	↑9.34*	↑2.95*	—	↑4.46*	↑1.97*	—	↓5.65*	—	—	—
S	38	33	↑12.12*	—	—	—	6.06	—	↑6.06*	↑13.16*	—	—	—	—	—	↑12.12*
V	42	40	—	—	—	—	2.50	—	—	4.76	—	—	5.00	—	—	—
Total	6356	6290	0.45	2.02	1.21	0.05	1.65	0.48	0.24	1.04	0.19	3.02	18.74	0.59	1.21	0.24

Association of Procedural MAEs With Major MAEs

For the association analysis between individual procedural MAEs and the occurrence of major MAEs, two models were used. Based on the generalised linear model (Table 6.), which analyzes each parameter separately, the increased risk of major MAEs was statistically significantly associated with the following procedural MAEs: unclear prescription, wrong strength, missing education of an inpatient, and trolley not under supervision. On the other hand, the risk of major MAE was statistically significantly reduced by the following procedural MAEs: missing identification of an inpatient, inappropriate hygiene handling, administration by a different nurse, and unchecked drug intake. According to the decision tree model (Figure 1.), at the first level (node 0) of the data set comprising 6290 medication administrations (371 with and 5919 without the occurrence of major MAE), the procedural MAE unclear prescription was identified as the most critical predictor of major MAE incidence (OR = 6.96, 95% CI 4.86–9.96). At the second level of this model, in both branches, with (node 2) or without (node 1) the occurrence of unclear prescription, the procedural MAE wrong strength was found to be significantly associated with increased major MAE, thus was identified as the second most critical predictor. Furthermore, at the third level of this model (node 3), in a subgroup of 5841 medication administrations, the procedural MAE inappropriate hygiene handling was found to be associated with reduced major MAE incidence. Similarly, in a small subgroup, the procedural MAE unchecked drug intake was found to be a protective factor of major MAE incidence as well (node 5). Finally, also in a small subgroup, the procedural MAE missing hand disinfection was found as a predictor of major MAE incidence (node 4).

TABLE 6 Association between major MAE occurrence and procedural MAE occurrence (generalised linear model).

Procedural MAE	Risk ratio	95% Confidence interval	p
Unclear prescription	7.34	5.04; 10.68	< 0.001
Wrong strength	4.38	3.08; 6.24	< 0.001
Missing education of an inpatient	3.75	1.60; 8.76	0.002
Documentation error	3.07	0.90; 10.50	0.073
Trolley not under supervision	1.66	1.16; 2.39	0.006
Missing opening date report	1.63	0.37; 7.19	0.516
Missing hand disinfection	1.11	0.88; 1.39	0.380
Missing identification of an inpatient	0.78	0.61; 0.99	0.042
Generic substitution	0.67	0.44; 1.01	0.058
Nurse interruption	0.57	0.28; 1.17	0.126
Unchecked drug intake	0.56	0.40; 0.77	< 0.001
Administration by a different nurse	0.56	0.37; 0.83	0.004
Inappropriate hygiene handling	0.52	0.39; 0.70	< 0.001

[IMAGE OMITTED. SEE PDF]

Discussion

MAE Prevalence

The entire medication administration process carried out by nurses in four Czech hospitals was explored and analysed in this study to emphasise the role of nurses as being important medical professionals with a great impact on the therapeutic value of the administered drugs, and therefore on the health safety of inpatients.

The overall prevalence of major MAE (6.9%) in this study was comparable to previous systematic reviews (Berdot et al. 2013; Keers et al. 2013b), although this should be read with caution due to differences in definitions of MAE and in direct observation methodology. Firstly, observers had no access to inpatient pharmacotherapy in advance. Secondly, the length of observation varied. Lastly, a different approach to nurse or inpatient randomisation was applied. Unlike in other studies (Berdot et al. 2021; Jessurun et al. 2023), we did not exclude the major MAE wrong time. Although its prevalence was very low (1.2%), unlike other studies (Berdot et al. 2013; Blignaut et al. 2017; Keers et al. 2013b), it was comparable with other major MAEs. The reason for low prevalence in this regard was prescription habits, since for most of the drugs usually only the time of day of administration (morning, noon, evening) was determined. Specification of the exact time of administration was used rarely, also due to the strict definition of major MAE wrong time (±15 min); this type of major MAE occurred most often with antibiotics and low-molecular-weight heparins. Except from major MAE wrong time, the most prevalent were wrong dose, wrong way of use and drug omission, which are often described as most common errors at all (Cottney and Innes 2015; Jessurun et al. 2023; Labib, Labib-Youssef, and Fatah 2018; Al Tehewy et al. 2016). Despite the absence of a closed-loop electronic system, the overall prevalence of major MAEs was lower than expected. In addition to the existing quality-of-care management in the hospitals, all study inpatients were adults and 85% of medications administered comprised oral solid drug forms, both generally related to the lower risk (Jessurun et al. 2023; Westbrook et al. 2024).

As specific MAEs were considered separately, they occurred in almost a quarter of medication administrations. The specific MAE inappropriate food timing was the most common, i.e., almost every fifth medication was administered incorrectly, probably due to the lack of systemic measures in hospitals and related departments. At the prescription level, the drug and food timing were defined rather exceptionally. There has been no comparison in the literature for this type of specific MAE so far, probably because there is only limited evidence for most drugs. Nevertheless, specific MAE inappropriate food timing was observed for some drugs, which could fundamentally affect the goal achievement of treatment and increase risks of medications such as levothyroxine (Skelin et al. 2017), rivaroxaban (Stampfuss et al. 2013), levodopa (Agnieszka, Paweł, and Małgorzata 2022), or metformin (Blonde et al. 2004). Furthermore, this error often occurred with the administration of insulins replacing postprandial secretion when the difference in food timing of different insulin types was not reflected (NHS 2021). In addition, this specific MAE could also affect the inpatients´ behaviour in their domestic environment. To eliminate the risk, incorporating the default proper food timing of drugs into an electronic prescription system can be recommended. Drugs that require administration before food should be administered first, followed by a meal at least 30 min after taking the drug, whereas other drugs should be administered soon after a meal. There are some exceptions that require different food timing to achieve an optimal effect, such as levothyroxine (Skelin et al. 2017) and bisphosphonates (Barrett et al. 2004). Therefore, pharmacists should be involved in the process, taking responsibility for setting up the electronic prescription system, including the proper food timing of individual drugs. Still, it is also necessary to adapt nursing processes accordingly.

For the specific MAE inappropriate drink, similarly, a lack of systemic management was noticed: instant tea (without tannins) was used most frequently, while plain water was used in less than a quarter of all medication administrations. Surprisingly, coffee with a high milk content being considered as the most hazardous drink with some drugs, was used in 5.16% of all medication administrations. This drink is inconvenient for the administration of drugs such as fluoroquinolones and levothyroxine because of forming an unabsorbable chelate, consequently reducing the effectiveness of these drugs (Kays et al. 2003; Skelin et al. 2017). Avoiding drinks containing milk and a high amount of free divalent ions (some types of mineral water) should be recommended, while pure water must be preferred. This should be implemented in the nursing standards of healthcare facilities.

The prevalence of the specific MAE inappropriate drug form modification was 3.0% of all medication administrations but accounted for 27.5% of the 691 medication administrations in cases of modified drug forms. The form disruption can significantly affect the extent and speed of drug absorption or cancel its protective function, e.g., stomach degradation, and mucosa irritation protection (Fodil et al. 2017). The crushing of proton pump inhibitors or potassium chloride in form of enteric tablets or micro-pellets and the splitting or crushing of metoprolol or dihydrocodeine in modified-release drug forms were also observed.

In addition, a high prevalence (20.11%) of the specific MAE s.c. application error within the subgroup of drugs administered subcutaneously was identified which further necessitates continuous education of nurses in this area.

In the procedural MAE category, unlike other errors, missing hand disinfection and no inpatient identification are reported frequently and are highly prevalent (Blignaut et al. 2017; Schutijser et al. 2018; Westbrook et al. 2011, 2020; Wondmieneh et al. 2020). In our study, missing hand disinfection was the most common error, observed in almost 6 out of 10 medication administrations. Since the quality of the disinfection was not evaluated, the reality is probably even worse. Surprisingly, the data collection took place during the waning wave of the covid-19 pandemic, when hygiene was essential. The high frequency of the procedural MAE missing hand disinfection points to the suboptimal setting of nursing processes. Nurses usually disinfected their hands while leaving the room only, so they were likely to have touched other objects (door handles, patient tables, etc.) before reaching the next room or the next inpatient. Moreover, nurses frequently used protective gloves inappropriately—only one pair during the entire process of administering drugs to all patients.

Even though the identification of an inpatient is a basic precaution to avoid inpatient confusion (especially during stressful situations), the procedural MAE no inpatient identification was very frequent (in about 40% of all medication administrations). According to the internal standards, nurses should check the patients´ wristband and ask for the inpatients´ name. Still, it was often noticed during the observation that the nurses performed the identification only because of the sake of observation, not as a routine procedure. Hence, the prevalence of this procedural MAE was found to be underestimated. There could be several reasons or barriers to identifying an inpatient, such as familiarity with the patient (especially during long-term hospitalizations) or feeling it unnecessary to repeat this process several times a day. Nevertheless, one event of inpatient confusion in a stressful situation was identified, which could have been prevented by the proper identification of the inpatient. To decrease the risk of inpatient confusion, the introduction of a so-called barcode-assisted medication administration system (Hutton, Ding, and Wellman 2021) might be beneficial, where the verification of an inpatient is integrated into the medication administration process.

The procedural MAEs, wrong strength, and generic substitution pose significant systemic problems in our hospitals. Due to the probability that nurses are not permitted to make these deviations from the prescriber's orders, none were recorded by the nurses in the patient's MAR, and it was therefore impossible to find out which drug had in fact been administered. The lack of prescribed medication in the department can be a possible reason, because the prescriber would presumably not follow hospital drug policy. However, according to legislation in the Czech Republic, nurses are not allowed to perform any sort of generic substitution. Admittedly, the use of a different strength could be legally acceptable, but unfortunately, the procedural MAE wrong strength has occasionally led to some major MAEs, such as the wrong dose of levetiracetam, mirtazapine, levothyroxine, or metformin.

Factors Associated With MAE

It is apparent that certain major MAEs and specific MAEs are typical of certain ATC classes. The analysis showed that ATC class A (alimentary tract and metabolism) was related to problems with the lactulose dosing, especially a great variance in dose administration. It was caused by the procedural MAE's unclear prescription (no or unspecified volume units) in some cases. This could be significant in certain situations, especially in the treatment of hepatic encephalopathy, when higher doses are taken several times per day (Sharma and Sharma 2013). At the same time, in ATC class A, the major MAE wrong way of use was found more frequently related to proton pump inhibitors administration into the percutaneous endoscopic gastrostomy or to the swallowing of chewable tablets of calcium with vitamin D.

ATC classes R (respiratory system) and S (sensory organs) were mainly represented by inhalers and eye drops; associations indicated an individual approach of each nurse to administer those drugs. In our opinion, the nurse's role in administering these drug forms is not clearly defined in the system. On the other hand, the literature discusses the possibility that “these medications are usually not considered the most necessary during hospitalization” (Jessurun et al. 2023).

Interestingly, the higher prevalence of the specific MAE inappropriate drug form modification was found in ATC class C (cardiovascular system), since, in this class, tablets (coated and controlled-release drug forms) usually are not intended to be divided into equal parts or to be split in half at all. Furthermore, this specific MAE occurred mostly alongside the procedural MAEs of unclear prescription and/or wrong strength (in cases where the drug of the correct strength was unavailable at the department).

The major MAE wrong time was associated with ATC class J (anti-infectives for systemic use), and in some cases, time deviation was even greater than 2 hours. The medication administration time was observed to be suboptimal and often set in the period of shift changes. Although the clinical significance could not be assessed (as no such detailed health data were collected), in extreme cases it still might have potentially led to antibiotic therapy failure or to an increase in antibiotic resistance (Ambrose et al. 2007). Better coordination of nursing processes and optimization of individual department settings can be suggested in this regard.

Concerning the association between individual characteristics of nurses/inpatients and the occurrences of major MAE/specific MAE, the individual variations amongst hospitals were found to be the most common predictor, probably due to the individual drug administration processes set in each hospital. The same predictor was also described in published studies (e.g., Blignaut et al. 2017; Jessurun et al. 2023), thus the proper setting of individual hospitals seems to be essential to reduce the incidence of major MAE/specific MAE (Keers et al. 2013a). Furthermore, the total work experience in healthcare was found to be another predictor, i.e., major MAEs were associated with nurses with minimal work experience and with nurses with experience of more than sixteen years. A similar non-linear u-shape dependence has been described by Hwang et al. (2016) However, the increase was significant in cases where the employment length was shorter than 1 year or longer than 5 years. To our knowledge, it is not common for polypharmacy to be proven as another predictor as identified in one study only (Cottney and Innes 2015). The same applied to specific MAEs, where the number of medication administrations per inpatient per day was found to significantly increase its frequency by 1.3% for each dose of medication administered to a particular patient.

Analysis of major MAE occurrence alongside procedural MAEs revealed some significant associations (Table 6, Figure 1). The procedural MAE unclear prescription was responsible for an approximately sevenfold increase in major MAE frequency in both statistical models, and procedural MAE wrong strength was responsible for an approximately fourfold increase. A detailed data review showed the vaguely specified dosing (as shown in ATC class A), combined drugs (e.g., calcium with vitamin D), and the drug strength labelled with some word supplement (e.g., “forte”, “S”, etc.) as the most problematic. In most cases, nurses were not aware of such potential problems, while they neglected to verify the prescription by contacting the prescriber. This lack of awareness often led to the major MAE wrong dose. To minimise those risks, the nurses´ education ought to be improved, it would be beneficial to accelerate the implementation of a more advanced electronic prescription system (selection from a list of available drugs in stock, which includes exact drug names and strengths). Nonetheless, some factors decreased major MAE frequency in our study (procedural MAEs: missing identification of an inpatient, inappropriate hygiene handling, administration by a different nurse, and unchecked drug intake). This could be related to the effect of observation, since the nurses did not follow their usual routine and tried to meticulously follow the nursing process, therefore increasing the risk of major MAE. Consequently, adhering to the routine (e.g., not ignoring the use of tweezers, tool cleaning, drug intake check, etc.) was regarded as a major MAE protective factor.

Strengths and Limitations

To our knowledge, this study is unique in its complexity of studying the whole process of medication administration, including adherence to good practice standards. Furthermore, the study is valuable for its multidisciplinary (nurses' and pharmacists' views), complex approach and significant sample size (6356 medication administrations). Also, a very detailed classification of medication errors was used. We observed and evaluated the broad range of MAEs (including hand disinfection, patient identification, drug and food/drink timing, etc.) using the direct observation method, which is considered the most accurate and relevant in data collection (Berdot et al. 2021). In addition, possible associations between identified procedural MAEs and the occurrence of major MAEs were analysed.

Admittedly, this study has some limitations. There was a potential selection bias amongst nurses due to voluntary participation. Despite using the direct observation method, an observational bias could have occurred due to the relatively large number of observers and the Hawthorne effect of short-time observation. The Hawthorne effect should not have any major effect on the overall major MAE prevalence (Dean and Barber 2001), but the effect on individual MAEs and procedural MAEs could not be excluded.

Conclusions

Our multidisciplinary and multicentric study offers some unique insight into hospital medication administration, and despite finding the overall major MAE prevalence (6.9%) similar to recent studies, we also found meaningful non-adherence to good practice standards (especially in hand disinfection and inpatient identification) and revealed the occurrence of problematic food and drug timing (19.7%).

Robust statistical methods were used to identify potential sources of MAE occurrence. Firstly, MAE occurrence across ATC classes were used to identify the most risky medications (e.g., proton pump inhibitors, inhalation systems). Secondly, associations between MAE occurrence and nurses'/inpatients' characteristics were measured revealing that the indicator respective hospital to have the largest impact on MAE occurrence. Lastly, by analysing associations between major MAE occurrence and procedural MAEs, we identified two meaningful predictors of major MAE occurrence: unclear prescription and wrong strength.

This robust and unique study provides data indicating that, to minimise the occurrence of MAE, we should optimise all processes associated with medication administration including involvement of facility management and other healthcare professionals such as pharmacists.

Relevance to Clinical Practice

To the best of our knowledge, this is the first work analysing the entire context of medication administration, focusing on medication errors. The methods used here could be considered to identify trouble spots within the medication administration process. We recommend that the determinants of MAE identified in our study should be considered by hospital stakeholders in their support programs to reduce the level of burden for nurses during medication administration.

Author Contributions

All the authors except two contributed to the conception, data collection, data analysis, and drafting the manuscript. Ales Antonin Kubena and Jiri Vlcek contributed to the conception, data analysis, and drafting the manuscript only. All the authors read and approved the final manuscript as well as agreed to be accountable for all aspects of the work.

Acknowledgements

The study was supported by Ministry of Health of the Czech Republic, grant nr. NU20-09-00257. All rights reserved.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Statistics

There is a statistician on the author team: Ales Antonin Kubena.

References

Agnieszka, W., P. Paweł, and K. Małgorzata. 2022. “How to Optimize the Effectiveness and Safety of Parkinson's Disease Therapy? – A Systematic Review of Drugs Interactions With Food and Dietary Supplements.” Current Neuropharmacology 20, no. 7: 1427–1447. [DOI: https://dx.doi.org/10.2174/1570159x19666211116142806].