Write a short review of the literature.? Use ONLY the 3 articles listed below that must be quantitative research reports. All articles must be published in the last 5 years.? Write an intro

Work 74 (2023) 327–339 DOI:10.3233/WOR-211221 IOS Press

327

How do medication errors occur in the nursing communication process? Investigating the relationship between error types and error factors

Haizhe Jina, Zhibin Xiaoa, Mingming Lia,∗, Quanwei Fub and Vincent G. Duffyc

aDepartment of Industrial Engineering, School of Business Administration, Northeastern University, Shenyang, China bDongguan Kanghua Hospital, Dongguan, China cSchool of Industrial Engineering, Purdue University, West Lafayette, IN, USA

Received 17 September 2021 Accepted 20 December 2021

Abstract. BACKGROUND: Human error types and error factors are two important elements of error analysis. Understanding the relationship between them can contribute to new case analyses, the tendency of error occurrence statistics, error factor identification, and prevention of error recurrence. OBJECTIVE: To provide evidence and guidance for the prevention and improvement of medication communication errors by quantitatively exploring the relationship between error types and error factors. METHODS: Data were collected on self-reported errors in the medication administration process by nurses in all departments of three cooperative medical institutions, and an error sheet of specified style was adopted. Error types were determined by the systematic human error reduction and prediction approach method and human cognition processes. Error factors were extracted using the root cause analysis combined with Berlo’s communication model, and the relationship between error types and error factors was quantitatively studied using the partial least-squares regression method. RESULTS: After a one-by-one analysis of 303 error cases, the communication errors occurring in the nursing medication process could be explained by six error types and 12 error factors. In addition, 20 correlation patterns between the error types and error factors were quantitatively obtained, and their path coefficient distributions ranged from 0.088 to 0.467. CONCLUSION: The results of this study may provide reference to understand errors and establish countermeasures from the statistics of error occurrence trends, extract error factors related to error types and determine key error factors.

Keywords: Medication error, nursing communication process, error type, error factor, healthcare

1. Introduction

Healthcare is a complex, variable, and unpre- dictable professional activity system [1, 2]. Medical

∗Address for correspondence: Mingming Li, Doctoral Student, Department of Industrial Engineering, School of Business Administration, Northeastern University, No. 195 Chuangxin Road, Hunnan District, Shenyang 110167, China. E-mail: [email protected].

errors occasionally occur when medical personnel conduct complex professional operations during the medical processes and it is mainly caused by human errors [3, 4]. Medication processes refer to the entire operational process of medication, including drug prescription, dispensing, administration, and obser- vation after administration. Medication errors in nursing (MEs-N) account for a high proportion of all medical errors [5, 6]. Multiple departments with

ISSN 1051-9815/$35.00 © 2023 – IOS Press. All rights reserved.

328 H. Jin et al. / Investigating the relationship between error types and error factors

different operational characteristics coexist within a medication process; therefore, communication is very important in healthcare, and patient safety can be affected by communication errors [7, 8]. A previous report pointed out that MEs-N caused by commu- nication problems were proportionately higher [9]. Therefore, reducing communication errors in the nursing medication process is crucial for ensuring patient safety [9, 10].

The following questions should be answered to improve the understanding of communication errors in the medication administration process. First, which types of errors may occur in the medical communica- tion process? This can be determined by classifying medical errors based on their characteristics (error types). Second, why do such errors occur? Factors that result in errors during the operational process should be extracted to answer this question (error factors).

The error types are used to summarize the errors, the granularity of the error types depends on the expression level of errors [5, 6]. For example, the expression of error types can be systematic, behavioral, or psychological [11]. The nursing com- munication error in the medication administration process is the deviation of the expected commu- nication behavior from the actual communication behavior [12]. Therefore, in this study, communica- tion error types are defined as the classification of error behavior characteristics from the perspective of the deviation between expected communica- tion behavior and actual communication behavior. Observing human errors in healthcare, Mitchell et al. divided these errors into four categories (medi- cal delivery, medical equipment, complications, and care), each with several subcategories [13]. Although these classifications of error types can include com- munication errors, they lack relevance and specificity for research on communication errors. The system- atic human error reduction and prediction approach (SHERPA) is a human error identification technology that proposes error classification and combines hier- archical task analysis to identify possible errors in human activities [14, 15]. It has the characteristics of structure, comprehensiveness, and easy application, and is widely used to identify human error in various fields. Kirwan compared SHERPA, cognitive reliabil- ity error analysis, generic error modeling system, HEI in systems tool, predictive human error analysis, and other human error recognition technologies, and pro- posed that SHERPA has the best overall performance [16]. The SHERPA technology is also widely used in

the medical field. For example, Ghiyasi et al. used the SHERPA method to evaluate and analyze the human error of nurses in the emergency department [17]. The SHERPA model divides human error into five categories: action errors, checking errors, retrieval errors, communication errors, and selection errors, and each category contains several error types. The communication category is divided into three error types: information not obtained, wrong information communicated, and incomplete information commu- nicated [15]. To summarize, the SHERPA model has a wide range of applications in the classification of human errors, and it is also applicable to the medical field.

Here, the phrase “error factors” refers to the causes of errors. Regarding the error factors that influ- ence healthcare, Handayani et al. proposed that the two main factors inducing medication administration errors were the safety culture of medical institutions and the understanding of patients’ families on medi- cal treatment [18]. Thomas et al. proposed that the complexity of nurses’ work, interruption of work, and high workload were the main factors induc- ing medical errors [19]. Brigitta et al. proposed that ineffective communication, work environment, and human factors were the main factors causing the occurrence of medication administration errors in nurses [9]. In conclusion, the current studies on error factors mainly focused on safety culture, organiza- tional management, and work environment, and did not pay enough attention to the operation process. Good results were produced by good operational pro- cesses [20]. Medical errors were the embodiment of operational process defects, and the extraction of error factors from the perspective of operational processes was an effective way to prevent the recur- rence of medical errors [20, 21]. The communication model helped to extract communication error factors from the perspective of the operation process. The Shannon–Weaver model is called the initial model in the field of communication [22]. Berlo extended the Shannon–Weaver model and proposed an SMCR communication model that contains five elements: source, message, channel, receiver, and noise. SMCR model is widely applicable to communication in mul- tiple industries, including medical institutions [23].

To further prevent medication communication errors made during clinical nursing, it is also neces- sary to understand which error factors are associated with specific error types and the strength of these relationships. However, many previous studies have focused only on error types or error factors, rather

H. Jin et al. / Investigating the relationship between error types and error factors 329

Fig. 1. Research framework.

than considering them together [9, 13, 14, 18, 19]. In recent years, quantitative research on medical errors has become a research hotspot [24, 25]. By quantifying the relationship between error types and error factors, we can understand the deep interaction between them and provide quantitative guidance to diagnose and improve errors.

2. Methods

2.1. Aim and research framework

The objective of this study is to provide evidence and guidance for the prevention and improvement of medication communication errors by quantitatively exploring the relationship between error types and error factors. The overall research framework of this study is shown in Fig. 1.

2.2. Study setting

The error cases collected from three participating acute care medical centers (number of beds: 1313, 1228 and 892; location: China) over a one year period (Jan. 2019 to Dec. 2019) were the data sources used in this study. This study collected medical errors related to the nursing administration process from all depart- ments of three cooperative medical institutions.

2.3. Data collection and analysis procedure

The case data for this study were collected using the following procedure. When nurses encountered

an error in the medication process, they recorded the error in the format-specified error sheet, which they submitted to the medical safety administra- tion office (the format-specified error sheet is shown in the attachment). The error data were based on self-reported cases by nurses in all department of cooperative medical institutions. The error data used in the present study were provided by the safety administration office of each participating medical institution. This study was not able to directly contact medical settings or patients. The error-related content is included in our research data and does not contain any personal information of patients or nurses.

Each collected error case was jointly analyzed by two medical error research experts and one medical safety manager. Both medical error research experts and medical safety managers have more than 10 years of work and research experience related to medi- cal safety. To extract communication error cases, we based the definition of medication communication error in nursing and identified cases of communica- tion errors in the nursing medication process based on the content of the error sheet.

2.4. Extraction of communication error types

The error type is the classification of error behavior characteristics from the perspective of the devia- tion between the expected operation behavior and the actual operation behavior [11, 12]. Consider- ing that the three SHERPA communication error types (i.e., information not obtained, wrong infor- mation communicated, and incomplete information communicated) are relatively abstract [15], we com- bined them with the human cognition process [26] to establish six specific error types in this study. The following uses a specific error case to introduce the extraction process of the error type.

Step 1: Summarize error case content. The cases of the analyzed objects are

summarized as follows: A drug that had to be kept refrigerated was removed imme- diately before medication administration. The nurse forgot to prepare the drug before the medication was administered.

Step 2: Determine in which stage of the cognitive process the case occurred.

According to the content of the error case, it is evident that the failure to pre- pare the drug occurred because of a lapse in memory (i.e., forgetting). Therefore, from

330 H. Jin et al. / Investigating the relationship between error types and error factors

the perspective of the human cognitive process, this case belongs to the stage of “Memory of the information.”

Step 3: Establish to which of SHERPA’s three communication error types this case belongs.

From the content of the case, it is clear that the message “Remove medication from the refrigerator before administra- tion” did not get through. Therefore, from the perspective of SHERPA’s three communication error types, this case is classified as “Information not communi- cated.”

Step 4: Determine the expression form of the error type based on the characteristics of error behavior in this case.

This case occurred in the human cogni- tive stage of “Memory of the information,” which belongs to SHERPA’s error type “Information not communicated.” Com- bined with the characteristics of the case, we integrate the above two into the expres- sion “Total lapse of information memory” and regard it as one of the error types in this study.

2.5. Extraction of communication error factors

Communication errors may be caused by various factors [27]. Therefore, even if a particular commu- nication error is identified, effective countermeasures cannot be addressed. Thus, identifying communica- tion error factors and taking countermeasures against them is an effective methodology [15, 25]. Root cause analysis (RCA) is widely used in the field of health- care and can deeply explore the causes of medical errors [28]. In this study, error factors were extracted from the perspective of the operation process based on the RCA method. Error factors were sorted based on five elements of SMCR communication model [23]. A schematic of the SMCR communication model is shown in Fig. 2.

The following steps illustrate the extraction method for error factors using RCA:

Step 1: Select team facilitator and team members. The purpose of this step is to identify

the personnel who will analyze error cases based on RCA. In this study, error cases were jointly analyzed by two medical error research experts and one medical safety manager.

Fig. 2. SMCR communication mode.

Step 2: Identify the error case to be investigated. This step is used to identify specific

error cases to be analyzed. An actual error case is summarized as follows: a drug that had to be kept refrigerated was removed immediately before medication adminis- tration. The nurse forgot to prepare the drug before the medication was adminis- tered.

Step 3: Describe what happened. This step involves combining the time

sequence of the selected error cases to fur- ther clarify the error-inducing process. We conducted time sequence sorting of the error induction process of the above cases, and the results are shown in Step 3 of Fig. 3.

Step 4: Identify the contributing items. In this step, a more in-depth discussion

will be carried out for each stage in Step 3, and the items closely related to the error are proposed. Contributing items for the above case are shown in Step 4 of Fig. 3.

Step 5: Identify the error factors. In this step, we first confirmed the spe-

cific stage of the communication error based on the SMCR model. Second, according to the communication error stage and error contributing items, the error factors are identified by repeatedly asking “Why?”.

In the above case, an error occurred in the nurse’s information reception, which belongs to the “receiver” stage of the SMCR model. Then, according to the information reception stage, we asked “Why?” four times and extracted the error factors as “Depend on memory.” The details are shown in Step 5 of Fig. 3.

2.6. Statistical analysis

Partial least-squares (PLS) regression is a multi- variate statistical analysis method proposed by Wold

H. Jin et al. / Investigating the relationship between error types and error factors 331

Fig. 3. The diagram of error factor extraction.

[29]. PLS combines the advantages of principal component analysis, canonical correlation analysis, and multiple linear regression analysis. Addition- ally, when there is a high degree of correlation between variables or a small sample size, PLS can provide more reliable results than other methods [30]. This study chose PLS to clarify the relationship between error types and error factors for several rea- sons. The dependent and independent variables were extracted by analyzing error cases. Therefore, there may be multicollinearity and nonlinear data distri- butions. Additionally, the sample size was small. In data with these statistical problems, PLS can provide more reliable statistical analysis results than other methods.

3. Results

3.1. Case collection

Using the error sheet designed for this study, we collected 597 MEs-N in three hospitals. Of these cases, 303 (50.8%) were caused by communication errors. The details of these cases are presented in Table 1. The proportion of CE-NMP was highest in hospital C, followed by hospitals B, A.

3.2. Communication error types

To extract error types in the communication pro- cess, based on the method of extracting error types

described above, we analyzed the 303 communi- cation error cases collected in this study one by one and then extracted error types for each case. We then unified these classifications and described the outcome as the study’s error type, as shown in Table 2.

The structure of Table 2 is as follows. The left list shows the three steps of the human cognition pro- cess, the left two columns represent SHERPA’s three error types, the left three columns represent the six new communication error types in this study, and the remaining four columns represent the number of cases. As can be seen from Table 2, the six error types extracted in this study are the materialization of SHERPA’s three error types from a medical per- spective. Theoretically, the three steps in the human cognition process and the three SHERPA error types should produce nine error types. However, some com- binations are not theoretically feasible. For example, for perception of information source, no error type corresponds to the SHERPA error type of incomplete information communicated because the process of perceiving an information source is momentary, and errors occurring at this stage mainly result from a lack of detection or incorrect detection. Therefore, the six kinds of error types described above are con- sistent with the actual error data and with logical reasoning. Table 2 shows that information cogni- tion failure, at 35%, accounted for most cases of all error types, followed by selection of the wrong information source (22.4%) and missing information cognition (17.4%).

332 H. Jin et al. / Investigating the relationship between error types and error factors

Table 1 Characteristics of MEs-N

Hospital Beds Care State Number of Collection Number of CE-NMP/ type MEs-N period CE-NMP MEs-N

A 1313 Acute China 224 Jan. 2019 103 46.0% B 1228 Acute China 197 – 99 50.3% C 892 Acute China 176 Dec. 2019 101 57.4% Total — — — 597 303 50.8%

CE-NMP: Communication Errors in Nursing Medication Process.

Table 2 Communication error types

Human cognition processes

SHERPA’s error types This study’s error types Number of error cases (%)

Perception of information source

Information not communicated Failure to find the information source

45 (14.9%)

Wrong information communicated

Selection of the wrong information source

68 (22.4%)

Cognition of information contents

Wrong information communicated

Information cognition failure 106 (35.0%)

Information communication incomplete

Missing information cognition 53 (17.4%)

Memory of the information

Information communication incomplete

Partial lapse of information memory

15 (5.0%)

Information not communicated Total lapse of information memory

16 (5.3%)

Total — — 303 (100%)

3.3. Communication error factors

The error factors considered in this study are fac- tors in the operational process that often result in errors. To extract error factors in the communica- tion process, based on the method of extracting error factors described above, we analyzed the 303 com- munication error cases collected in this study one by one and then extracted error factors for each case. Then, we combined the error factors with the SMCR communication model elements. The extracted error factors are listed in Table 3.

As shown in Table 3, 12 communication error fac- tors in the nursing medication process were identified in the 303 cases. Error factors in the communication process were found from the five elements: source, channel, message, receiver and noise. From the per- spective of the number of cases, the error factor “lack of clarity on changes in information content” accounted for the largest percentage of all errors (18%), which shows that timely information modifi- cation according to the patient’s situation is of major importance in the operational process of healthcare. Because the personnel structure related to infor- mation changes is complex and involves numerous individuals, it is important to consider how to ensure that the relevant operating staff members acquire

new information correctly to reduce communication errors. Additionally, “information in multiple loca- tions,” “instability of the position of information,” “usage of confusing units and figures,” and “occur- rence of information with low appearance frequency” accounted for high percentages of the errors, with each making up over 10%. Addressing these factors should be a high priority in the operating process.

3.4. Relationships between error types and error factors

This study examined the relationships between error types and error factors using PLS analysis. These relationships are presented in Table 4. The table shows that all R2 values of the endogenous variables were satisfactory, demonstrating a high and effective predictive validity. Notably, the R2 value for “selec- tion of the wrong information source” showed that all the exogenous variables together explained 57.9% of the variance in B2. Table 4 can explain the influence of error factors on the error modes. For example, the path coefficients and bootstrapping test results indi- cated that “instability of the position of information”, “information deficiency”, and “irrelevance of related information” had significant direct effects on “failure to find the information source”.

H .Jin

etal./Investigating the

relationship betw

een error

types and

error factors

333 Table 3

Communication error factors

SMCR model elements

Error factors Description Case Proportion (%)

Source Information in multiple locations

Related information isn’t written on one information source, but scattered across multiple information sources

The administration instruction of Sulbacillin was on another instruction list, and the medical staff failed to receive the information.

43 (11%)

Source Information deficiency Information extraction isn’t based on information source or fail to find information source

Due to a large number of parallel jobs, medical staff operated with memories instead of the reference to the information source, resulting in failing to receive information changes and thus operation errors.

24 (6%)

Source Instability of the position of information

Information source has no fixed storage position or isn’t stored on fixed position

Although there are injection instructions of Vancomycin on the instruction list, but medical staff did not find the injection prescription, leading to failure to perform the injection in specified time.

51 (13%)

Channel Oral dissemination of information

Information spread orally Oral transmission of information led to hearing the wrong name of patient.

13 (3%)

Message Irrelevance of related information

Highly related information is scattered on different parts of an information source

There were instructions for both catabon and dobutrex at 17 : 00, but they were not written in one place. The medical staff only saw the instruction from catabon.

15 (4%)

Message Lack of promptness in information sharing when content changes

Information is not shared timely when content changes The changed information was not transmitted to the dispensary. Then the medicine prepared following the instructions before the changes were sent to the nursing station.

26 (7%)

Message Lack of clarity on changes in information content

Information changes isn’t obvious to recognize Decadron was reduced from 2 mg to 1 mg. But the medical staff failed to notice the change and gave the previous administration amount.

71 (18%)

Receiver Similar names Similar font or pronunciation of