A Medication Safety Model: A Case Study in Thai Hospital

Reaching zero defects is vital in medication service. Medication error can be reduced if the causes are recognized. The purpose of this study is to search for a conceptual framework of the causes of medication error in Thailand and to examine relationship between these factors and its importance. The study was carried out upon an in-depth case study and survey of hospital personals who were involved in the drug use process. The structured survey was based on Emergency Care Research Institute (ECRI) (2008) questionnaires focusing on the important factors that affect the medication safety. Additional questionnaires included content to the context of Thailand's private hospital, validated by five-hospital qualified experts. By correlation Pearson analysis, the result revealed 14 important factors showing a linear relationship with drug administration error except the medication reconciliation. By independent sample t-test, the administration error in the hospital was significantly related to external impact. The multiple regression analysis of the detail of medication administration also indicated the patient identification before administration of medication, detection of the risk of medication adverse effects and assurance of medication administration at the right time, dosage and route were statistically significant at 0.05 level. The major implication of the study is to propose a medication safety model in a Thai private hospital.


Introduction
During 1995 to 2010 The Joint Commission of the United States (2011) showed 6,782 serious adverse events and 67 percent resulted in death. Medication error was in the top ten of serious adverse events in 2010. Medication errors harm an estimated 1.5 million people in the United States each year and result in increased medical costs to treat adverse events from using the drug estimated at 5.3 million dollars per year (The Institute of Medicine of the United States, 2006). A survey of medication errors in 1,116 hospitals in the United States, the figure was 5.07 percent (Bond et al., 2001). In Europe, it was found that adverse events occurred at a cost of billions of pounds a year and 48-49 percent was caused by drugs and could be prevented (Taxis & Barber, UK and Germany, 2004). In Australia, adverse event occurred at a cost about 350 million U.S. dollars per year and 43 percent was caused by drugs and could be prevented (Hodgkinson, 2006). In Japan, it was found that 46.6 percent of the adverse event was caused by drugs and could also be prevented (Nakajima, 2005). In Thailand, there is no national data collection and this problem has never been studied in the economic sense.
Medication errors are caused by many factors during the medication administration. Reduction of human error was emphasized to be the major source to minimize errors in many literatures. Human error was associated with failure of action causing the deviation of doing what is right (Hansen, 2006). Kleinpell (2001) pointed out that many factors which lead to errors in medication use were associated with the human experience such as new or temporary staffs who were lack of knowledge and would trend to produce incorrect documentation. Noise and other factors such as fatigue, stress and indolence cause harm patients. These are factors associated with human error (Rassin et al., 2005). Although human error is the cause of medication error; it is believed to have little effect (Henneman & Gawlinski, 2004). Since human error is relatively easy to be recognized, it raises the culture of blaming human error in health care organizations (Institute of Medicine, 2001). Blame culture is an adverse effect on workers and certainly affects patients as well. Once confidence has been destroyed it will affect Brady and Fleming (2009) concluded five contributing factors to medication errors as barriers to report, knowledge & skill, deviation from procedures, reconciling medical history, prescriptions and drug distribution systems. The study of Jordan by Mrayyan et al. (2007) found that the causes of the medication error was related to defective drug labeling or packaging, the confusion of different types of injection device and noise at work. Teinila et al. (2011) found that the medication error was associated with five origins such as physician, organization, information technology, patient and hospital. Emergency Care Research Institute (ECRI Institute) (2008) has issued a set of self-assessment questionnaires for the hospital managers to assess the safety of drugs in the hospital (medication safety). It summarizes 15 important matters collected from leading medical institutions in the United States covering national and international medical errors (Figure 1). The definition of these factors (list of resource) is shown in Table 1. Prescribing and transcribing The prescription involves an action of a legitimate prescriber to issue a medication order.
The transcription involves anything related to the act of transcribing an order (by nurse, pharmacist, or clerk) for order processing (e.g., electronically or manually in the patient's record).

Illegible handwriting, verbal or telephone orders
It is important to be alert for illegibility and to the prescription orally or by phone. Any doubt or confusion must be resolved before dispensing or administering the medicine.

Predispensing and dispensing medication
Predispensing activities include printed data, patient's name, drug name, drug use in the drug label. Dispensing activities include order review, entry/processing, preparation, and dispensation (including stocking automated dispensing devices).

Medication administration
Administering activities begin in the patient care unit, care delivery area, or patient bedside and continue through actual drug administration to the patient. It includes giving the right medication to the right patient at the right time and informing the patient about the medication.

Surveillance of drug monitoring
Monitoring activities involves evaluation of patient's physical, emotional, or psychological responses to the medication with record of such findings.
High-risk medications High-alert medications are drugs that have a higher risk of causing significant patient harm when they are used erroneously. Based on the stated purposes, the following research questions were formulated: 1). What are the levels of importance of these factors that affect the medication errors in a Thai private hospital?
2). How often do these factors implement were practiced to reduce medication errors?
3). How do these factors affect the most common type of medication errors? 4). What are the detail description of these factors that affect the most common type of medication errors?

Questionnaire
Questionnaire covering the causes and preliminary findings of medical error were presented to forty six hospital staffs during October to December 2012 (crossectional approach) for answering. The medication error is the dependent variable, and the set of the important factor variable is the independent variable. The direction of expected effects of these variables is indicated by arrows in Figure2. It is proposed that these important factor variables would positively contribute to the medication errors. The questionnaire was divided into three sections: A) General information of the organization B) The prevention of medication errors and C) suggestions and comments. The respondents were required to complete all three sections. There were 8 questions in section A, 216 questions in section B and 2 questions in section C. Data were scored through questionnaire on 5 point Likert scale, 1=strongly disagree to 5=strongly agree. All the answers were received from 46 respondents.
There were two main objectives in designing the questionnaire. First was to maximise the proportion of population answering the questionnaire-that is, the response rate and to obtain accurate relevant information for the survey. Second was to maximise response rate-that is the author had to carefully consider the way questionnaires are administered, explaining the purpose of the survey, and keep reminding those who had not responded. In order to obtain accurate and relevant answer, the author was present at the interview session to explain and clarify any question which might arise.

Case study Sampling
Case study sampling of the population was by randomization, regardless of the probability (non-probability) and sampling method specific (purposive sampling) due to sampling convenience and limitation. Critical case sampling is a type of purposive sampling technique that is particularly useful in exploratory qualitative research, which is of limited resources, as well as research where a single case (or small number of cases) can be concluded in explaining the phenomenon of interest. The case study selected a large hospital occupying more than 100 beds with a standard quality of hospital accreditation in Thailand (HA). Sources of primary data were obtained from 46 staffs who work with drug use process in the hospital, consisted of 26 general staffs in nursing department, 11 head-nurses in different sections, 2 managers of nursing department, 1 physician, 4 pharmacists, 1 hospital quality staff and 1 medical director. All the proposed factors affecting medication error variables were based on ECRI (2008) questionnaire and two additional factors from Thai experts' opinion. The data for these variables were computed and used in data analysis.

Expert Interview and Validity
Five Thai healthcare quality experts were interviewed. Semi-structured interview was adapted for each expert.
There were two quality directors from two private hospitals, one quality manager from a private hospital and another one from a public hospital, the last one was an expert surveyor from the Institute and Hospital Accreditation (HA) They were interviewed in order to get an actual and in-depth view of medication errors in Thai hospitals; In addition, 226 questions were constructed and presented to these five experts to verify the legitimacy by applying. [Item-Objective Congruency index (IOC)] Each question was rated in three scales: '-1' representing disagreement, '0' representing uncertainty, and '+1' representing agreement. IOC index from all questions was 0.852 (n = 5) which illustrated the acceptable level of content validity (IOC index must more than 0.5)

The Reliability
Calculating Cronbach's alpha is the most commonly used method to estimate reliability. The questionnaires were constructed for this study to pilot test with target population of 46 persons who were involved in the hospital drug use process. The Cronbach's coefficient alpha of the significance of important factors affecting the cause of medication errors was 0.9942 and the actual implementation was 0.9964. Both demonstrated acceptable level of internal consistency.

Measurement of Variables
The organizational variables were measured as follows: 1). Dependent variable represents the medication errors.
2). Independent variables represent the proposed causes of medication error, adapted from ECRI (2008) framework and Thai expert opinion.

Statistical Analysis
The SPSS for Windows was used to analyze descriptive statistics, frequency distribution, percentage and standard deviation. The correlation coefficient of Pearson Product Moment coefficient correlation, independent samples t-test and stepwise multiple regression analysis were used to examine the research questions. Enter Regression analysis was used for variable selection technique and also the assumption of regression equations to determine all the possible relationships between independent and dependent variables.

Results
The results of the data analysis showed ten respondents from the ward (21.7%), and eight from the ICU (17.4%) consisting the two major department groups involved in the drug use process, though twelve respondents were from OPD department (26%).When the respondents were classified individually, they were twenty six nursing staffs (56.5%), eleven head nurses (23.9%) four pharmacists (8.7%), two managers of nursing department (4.3%), one physician (2.2%) as well as one medical director (2.2%).
Forty four respondents (95.7%) recognized the causes of medication error except two who did not (4.3%) Only forty two persons (91.3%) of these population recognized that they were preventable errors. Despite the fact that twenty eight personals (60.9%) expressed the lack of solution for prevention, the remaining eighteen respondents (39.1%) who admitted the existence of preventive measures but expressed that these measures were solely by avoid repeating the mistakes without a definitive solution. According to the author's view point, this is not an appropriate approach for prevention of errors. Table 3 Table 3 The highest mean was the administration error, reflecting the most severe problem in the drug use system. There were a few other problems including errors in copying drug orders (transcribing error) and in dispensation (dispensing error). The rarer errors were prescribing and pre-dispensing errors. Types of medication errors are shown in Pareto charts (Figure 3).

The Ex
In this stud The study accreditati but five (1

Ques private hos
The study ranked in abbreviatio storage, m reporting, reconciliat The signif "Strongly for hospita

Ques
The study frequency medication monitoring medication the pharm and the ho applying P  Null hypothesis: There is no significant difference between the means of the two variables.
Alternate hypothesis: There is a significant difference between the means of the two variables The last column showed the significant value of less than 0.05 which rejected the Null hypothesis. There is statistically significant difference at 95% confidence level (P <0.05), demonstrating a significant difference between the means of important and actual performances. This reflected that these factors were essential and should effectively implemented. In this hospital, these factors affecting medical errors were not cautiously practiced. Therefore, it is important to improve the performance of these hospital personals effectively to reduce the medication errors.

Questions 3: How do these factors affect the most common type of medication errors?
The study revealed most respondents recognized that administration error carried multiple problems in the drug use system. Thus the most common type of medication errors in this hospital was the administration error.
It is believed that the independent variables (the 15 important factors by the theory of ECRI Institute (2008) affect medication errors. The concept of ECRI Institute (2008) was analysed concerning the correlation between independent variables and the dependent variable (administration error), to find out whether the independent variables was a linear relationship with the dependent variable, by applying Pearson correlation (Pearson Product Moment Correlation Coefficient) and the result was shown in Table 5.  In addition to the afore-mentioned results, extra-study to find out the external impacts affecting the most common type of medication errors in a Thai hospital implying the certification standards of the hospital and the professional standard of hospital staffs was carried out. The two external impacts as independent variables and administration error as dependent variables were analysed by applying the Independent samples t-test. The certification standards of the hospital was shown in Table 6 and the professional standard of hospital staffs was shown in Table 7.   Vol. 5, No. 5;2013