HEALTHCARE FAILURE MODE AND EFFECT ANALYSIS DESIGN FOR INDONESIAN HOSPITAL LABORATORIES: A LITERATURE REVIEW Desain Healthcare Failure Mode and Effect Analysis untuk Laboratorium Rumah Sakit Indonesia: Tinjauan Pustaka

Background: Error rate in medical laboratories is very low. Only one error is identified every 330–1,000 events. The goal of laboratory services should outweigh patient safety in a well-structured manner. Healthcare Failure Mode and Effect Analysis (HFMEA) is a proactive preventive method for identifying and evaluating potential failure. Aims: This study identified factors affecting patient safety in hospital laboratories and described potential risk identification process using the HFMEA. Methods: This study was conducted between March-July 2020 and retrieved data from PubMed, Scopus, and Google Scholar. The data were generalized and extracted into Table 1 based on factors dealing with patient safety in hospital laboratories. This study performed a risk identification design based on the steps of HFMEA. Results: Out of 4,062 articles collected, only 8 articles between 2013–2020 were included for analysis. The highest error rate in laboratories occurred in the pre-analytic phase (49.2%–84.5%). The errors included clotted and inadequate specimen volume, and thus the specimens were rejected. Factors related to patient safety in laboratories were patient condition, laboratory staff performance (including training, negligence, and burnout), facilities, and accreditation. Conclusion: The HFMEA process used the result of hazard analysis with severity and probability criteria categorized into health sector. Decision tree analysis could determine the next step of the work process. The HFMEA must be adjusted to the equipment and technologies in each hospital laboratory. Leader’s commitment in monitoring and evaluation is required to maintain patient safety culture. More comprehensive data from Indonesian hospital laboratories are needed to generate more representative and applicable results.


INTRODUCTION
Patient safety is fundamental to provide essential health services (WHO, 2019). Medical practices have risks that can endanger patient safety and contribute to patient safety incidents due to unsafe health services. According to WHO, adverse events due to unsafe medical services are one of the top 10 causes of deaths and disabilities in the world, where nearly 50% of them can be prevented. Based on the previous study on patient safety in Spanish primary health centres, every 9.6 of 1,000 patient visits would produce adverse effect, although the degree of seriousness was low (Romero et al., 2017).
Hospitals, a complex organization in providing quality services, need to care for the possible risks. The laboratory has an vital role in hospital activities and diagnosis of infectious diseases caused by parasites, fungi, bacteria, and viruses (Megiwati, 2015). According to Plebani (2002), almost 70% of patient's diagnoses and clinical treatment are based on laboratory results. Additionally, Jiang et al. (2014) mention 80%-90% of the laboratory test results play an essential role in establishing a patient's diagnosis (Carraro P, 2002;Jiang et al., 2014). Another study has shown despite the low error rate in the medical laboratory compared to other units at the hospital, most of these errors rarely become an adverse event. Patient safety should be the goal of laboratory services, and the principles should be applied systematically in a well-structured manner (Aita et al., 2017).
Research at Cantonal Zenica Hospital Bosnia-Herzegovina from December 2016-March 2017 analyzed 35,343 blood samples that were rejected in the laboratory information system due to pre-analytic errors, including haemolysis at 48.50%, clotting at 39.87%, unsuitable volume samples at 7.81%, wrong tube test at 2.16%, and identification errors at 1.66% (Kadić, Avdagić-Ismić, and Hasić, 2019). Lichenstein (2016) states that 82.8% of the errors in the laboratory were caused by human factors (errors performed by staff (43.5%), nurses (22.6%), and physicians (4.8%)). The majority of the laboratory errors (51.5%) were not associated with harm. As many as 17.4% of the patients were harmed due to the errors, 98.6% of the patients were temporarily harmed, and required treatment, and 0.7% of the patients were hospitalized or had their hospitalization prolonged due to the errors (Lichenstein et al., 2016).
In 2001, the National Center for Safety Patient (NCPS) chose the traditional the Failure Mode and Effect Analysis (FMEA) approach as the basis for developing a risk analysis in the health sector (DeRosier et al., 2002;Widianti, 2015). Based on the investigation results,, the NCPS explains the traditional FMEA was unable to meet the needs of the health sector. The NCPS believes that it is necessary to adjust risk assessment indicators in the traditional FMEA according to the health needs, and thus the concept of Healthcare Failure Mode and Effect Analysis (HFMEA) was born (DeRosier et al., 2002;Widianti, 2015). Risk identification using the HFMEA is a proactive method to identify, evaluate, and record failure modes that cause problems and impacts. These will prevent potential risks that can endanger patients and health service staff (Colman et al., 2019).
Some studies have found that the application of HFMEA reduced cases of specimen rejection from 0.92% to 0% in -2013(Chadwick and Fallon, 2012Hung et al., 2015). In this article, a literature review was performed to identify risks and factors of patient safety and also design a HFMEA for hospital laboratories.

METHOD
This study was literature review collecting articles from electronic databases such as PubMed, Scopus, and Google Scholar published in English and Indonesian between January 1, 2010-May 31, 2020. The literature search was conducted from March to July 2020 to search for electronic databases and selected journals, as well as crosschecking bibliographies from other published review articles through Mendeley to prevent duplicacy. The last step was all articles were reviewed by the clinical phatologist in an online final project presentation. This current literature review was a qualitative study that focused on identifying patient safety risks in hospital laboratories worldwide. Articles under review were journal articles, clinical trials, systematic reviews, observational studies, and descriptive studies that met the criteria based on Method Adopted for Literature Review by Turrini et al., 2010. The primary aim of this literature review was to identify risks and factors dealing with patient safety in hospital laboratories using the HFMEA or other methods. After the review process, the Healthcare Failure Mode and Effect Analysis for Indonesia hospital laboratories were designed based on the steps initiated by DeRosier et al. (2002). Articles using the HFMEA or FMEA were prioritized for more applicable analysis design in hospital laboratories. The fulltext versions of all potentially relevant articles were read independently by the researchers. Eligibility for article inclusion criteria was determined by a structured flow chart and detailed guidelines using PRISMA Flow Diagram (Figure 1). A summary of literature review findings is presented in Table 1. This table outlines the basic characteristics of each article reviewed and briefly summarizes the key findings.

ANALYSIS AND DISCUSSION
A Figure 1 shows 4,062 articles collected were identified, but only 20 articles of them were full text. Finally, 8 relevant articles that met the criteria were included for review. All these included articles identified patient safety risks and factors in laboratories by using the HFMEA and other methods. The eight articles demonstrated the total testing process (TTP) in laboratories was divided into 3 phases, namely pre-analytic, intraanalytic, and post-analytic. The preanalytical phase is the phase where the laboratory has no direct control on the process (Tournis and Makris, 2018) and occurs first in the laboratory process (Automation and Technology in the Histology Laboratory, 2018). The second phase is the intra-analytic phase, where the "actual" laboratory testing or the diagnostic procedures, processes, and products are conducted to ultimately produce results (Automation and Technology in the Histology Laboratory, 2018). The post-analytic phase is the final examination process which generates laboratory results. The highest error rate in laboratories occurred in the pre-analytic phase from 49.2%-84.5% (Hung et al., 2015;Patel et al., 2018). Errors in this phase were clotted specimens and inadequate specimen volume that may cause specimen rejection.
Factors that influenced patient safety in laboratories were patient condition, laboratory staff attributes including training, negligence, and burnout, facility, and accreditation. The review found only one article used the FMEA to identify risk and factors of patient safety in clinical laboratories, and one article integrated the Value Stream Map and HFMEA into Six Sigma method (see Table 1).

Potential risks in laboratories
According to the Regulation of Indonesian Ministry of Health No. 411/MENKES/PER/III/2010, clinical health laboratories provide clinical specimen examination services to obtain individual health information, support disease diagnosis, cure the diseases and restore the health (Ministry of Health Republic of Indonesia, 2010). Aproximately 70%-80% of the laboratory results are related to diagnosis and treatment, and thus laboratories are important for health care and patient safety. They may have potential errors such as misdiagnosis, late diagnosis, inappropriate tests, patient safety incidents, cost and time loss (Sinici Lay, Pinar and Akbiyik, 2014;Aita et al., 2017). The error rate in medical laboratories is very low (one error identified every 330-1,000 events) (Kalra, 2004); and the majority of these error rarely become adverse events (Plebani M, 1997), patient safety should be considered the goal of laboratory services and its principles must be systematically applied in a well-structured manner (Aita et al., 2017).
The pre-analytical phase includes test request, patient or sample identification, sample collection, handling and transport, whereas pre-analytical phase involves the steps of samples preparation for analysis such as centrifugation, aliquoting and sorting. It has been demonstrated that most occur in the pre-analytical phase by healthcare personnnel who are not under control of the laboratory, but the majority of these errors are preventable (Gunnur Dikmen, Pinar and Akbiyik, 2015). Pre-analytical phase start following specimen acceptance by the laboratory staff (Lippi and Guidi, 2006).
According to Hung et al, most testing errors occur in the pre-analytical phase accounting for 84.5% of the total errors detected (Hung et al., 2015). Patel et al,collected 172,800 data test performed on 43,200 samples, total number of errors were identified in pre-analytical phase (49.2%), intra-analytical (17.4%), postanalytical (33.4%) (Patel et al., 2018). Jiang et al. who used FMEA in their research explain the highest RPN (Risk Potential Number ≥ 200) in clinical laboratories was in the pre-analytic phase. It occurred due to hemolysis, delay on delivery samples to the laboratory, and inadequate specimen volume (Jiang et al., 2014).
Specimen rejection was caused by inappropriate test requests. As many as 16.97% of the physicians incorrectly determined the test according to patients' needs. Kachalia A, et al (2007) and Wahls and Cram (2007) support this result by stating that 55%-58% of the errors or delays in diagnosis were caused by the failure of the service unit or emergency department to determine appropriate tests according to patients' need (Kachalia et al., 2007;Wahls and Cram, 2007;Patel et al., 2018).
The errors frequency in the intraanalytic phase is around 7%-17.4%. The common errors include delay in reporting laboratory results due to malfunction of the equipment, sample mixing with other samples/ materials, interference (endogenous or exogenous substances), and undetected errors (Sakyi et al., 2015;Patel et al., 2018). This failure occurred in the re-sampling phase (Jiang et al., 2014;Sciacovelli et al., 2017). The final examination process, the post-analytic phase, Patel et al. (2018) stated that the percentage of the errors in the postanalytic phase was 33.4%, of which 12% were failure to report laboratory test results according to the predetermined time (Patel et al., 2018). Jiang et al. (2014) obtain the highest RPN (RPN ≥ 200) due to failure to report test results according to the standard time (RPN = 210) and critical results (RPN = 200) resulting in delayed emergency treatment. The failure occurred due to the carelessness of laboratory staff, such as ignorance about complex work procedures in the laboratory and the inability to operate Laboratory Information System (LIS) or a computer-based technology (Jiang et al., 2014). Specimen labeling errors and incomplete forms of specimen requisition were the major causes of specimen rejection. Some factors contributed to the errors in the specimen handling process. These include lack of discipline in completing surgical records, late data accuracy checking, lack of inspection mechanism, as well as the insufficient labor force. Errors in the post-analytic occurred due to carelessness of laboratory staff; for instance, the clinical laboratory results were often overlooked or not reported due to its complexity.

Laboratory Staff
Physician burnout Kroft, S. H (2020) American Society for Clinical Pathology Well-Being, Burnout, and the Clinical Laboratory 1. As many as 71% of the pathologists experienced burnout at some point. with a full one-third indicating that it was something they were currently experiencing. 2. Pathologists in the ASCP survey reported job-related stress, and 43% of them reported being moderately or very overwhelmed by their workload. 3. Burnout in lab professionals appears to be more prevalent than in pathologists and pathology trainees. About 85% of lab professionals sometimes experienced burnout. A half of them perceived it as a current issue. Over half experienced a lot of stress, and nearly half were moderately or very overwhelmed. Almost 40% were moderately or very anxious. This implies that the results were correlated with burnout.

Facility
Obsolete equipment

Gunnur Z D et al. (2013) Turki
Specimen rejection in laboratory medicine: Necessary for patient safety?
Barcode scanners may misread patient identifiication barcodes due to incompatible size of symbols on patient ID bands or specimen labels and the scanner settings. Equipment malfunction and undetected failure in the internal QC were identified mainly as intra-analytical errors. Automation, training of laboratory staff and espousal of internal and external QC programs contributed immensely to the remarkable decline in intra-analytical errors and also the good condition of the Art Analyzer.

Patient safety factors in laboratory
Based on Table 1, there are 4 factors affecting patient safety in the laboratory.

Patient condition factor
Phlebotomists have difficulties to find peripheral veins of patients, especially in pediatric, neonatal, oncology, and ICU patients, during the blood specimen collection. It may result in the potential risk of inadequate specimen volume for laboratory analysis. Poor condition of patients and uncooperative attitude towards doctor's recommendations become the factors that may cause poor quality specimen (Jiang et al., 2014;Sinici Lay, Pinar and Akbiyik, 2014). Patel et al. (2018) assert that high error rates in the laboratory happen due to various reasons such as lack of training for laboratory staff, the complexity of the test procedures, involvement of many professional, lack of human resources, lack of automation in the delivery of samples to the laboratory, weak communication between laboratory staff and department staff (Patel et al., 2018). According to Jiang et al. (2014), the failure in the pre-analytic phase during the sample collection was related to poor knowledge of nurses and couriers about working procedures in the laboratory.

Laboratory staff factor
Specimens clotting and hemolysis occur because phlebotomy training is rarely available to nurse or laboratory analysts (Sinici Lay, Pinar and Akbiyik, 2014; Sakyi et al., 2015). Second, the high of turnover staff, especially doctors, also contributed to the errors in the laboratory. As new doctors need to adopt laboratory work procedures, they require training to improve their competence. The tight schedule of doctors and laboratory nurses also hampered their participation in prevention error training (Sinici Lay, Pinar and Akbiyik, 2014).
Research by Hung et al. (2015) has found that negligence of laboratory staff in labeling patient identification forms is another cause of specimen rejection. Negligence occurs due to low discipline in completing the surgical records, low data accuracy, and low inspection mechanism due to insufficient labour (Hung et al., 2015). Jiang et al. (2014) have revealed that due to complex laboratory procedures, laboratory staff were careless in performing their work and thus cause failure in the post-analytic phase (Jiang et al., 2014).
Burnout of laboratory staff also influences the potential risk in the laboratory. As many as 71% of the pathologists in the laboratory experienced burnout. Based on the American Society for Clinical Pathology, work stress is the main cause of burnout, and too much workload is another cause (43%). Most of the laboratory staff (85%) experienced burnout generally because of fatigue in handling the complexity of work in the laboratories. A study has found 40% of the staff experienced moderate to severe anxiety levels (Kroft, 2020).

Facility factor
Another factor contributing to the error is incompatible scanner that fails to read patient barcodes in the pre-analytic phase. Mismatch might occur between the barcode size, the symbol on the patient ID, andspecimen labels (Gunnur Dikmen, Pinar and Akbiyik, 2015). Equipment damage is the main cause of errors in the intra-analytic phase, causing laboratory test results not to be collected in the postanalytic phase. Equipment damage which is not detected in the internal quality control are identified as errors in the intraanalytic phase (Sakyi et al., 2015).

Accreditation factor
Accredited laboratories (47%) had a lower error rate than non-accredited laboratories (IRR 0.53; p 0.03). Laboratories of university hospitals or research laboratories had a lower error rate than private laboratories (p=0.013). Tack et al. (2018) have hypothesized a positive correlation between laboratory accreditation status and the accuracy of laboratory test results. Accredited laboratories are considered to have a higher level of caution in reporting analysis results to patients. Laboratories of university hospitals had a lower analysis error rate in conducting the analysis than non-university laboratories (Tack et al., 2018).

Discussion
After identify risks and factors in clinical laboratories based on the articles that we have, the author design an Healthcare Failure Mode and Effect Analysis for Indonesia hospital laboratories based on the steps from DeRosier et al., (2002). HFMEA is a traditional FMEA developed specifically for the health sector by NCPS in 2002(DeRosier et al., 2002. The HFMEA approach was developed to address criticism of using FMEA in healthcare, particularly in respect to the use of a single risk priority number (RPN) to rank vulnerabilities (Arbor, 2014). The process is the same as the traditional FMEA, but there are striking differences in the process, exactly on step 4 of calculating the risk value. In this step, the traditional FMEA assessment is the risk priority obtained from the multiplication of severity, occurrence, and detectability in non-health services. While in the application of HFMEA, the risk value of hazard analysis (see on Formula 1) is a multiplication of severity and probability with four criteria for health services ( Table 2).
The use of the conventional FMEA is commonly applied to identify risk in hospital laboratories although the method needs improvements. A report mentioned that the laboratory of PKU Muhammadiyah Yogyakarta Hospital in 2015 employed the Root Cause Analysis (RCA) and Fishbone Analysis rather than the standard FMEA (Hospital Report: RS PKU Muhammadiyah Yogyakarta, 2015). Another study has shown that the laboratory of Wirobrajan Primary Healthcare Center carried out the standard FMEA by determining the cut-off point of the Risk Priority Number (RPN) to ease risk identification according to the error level (Khairani et al., 2015). Research conducted by Sithi and Ani (2018) at Pacitan District General Hospital, East Java province, proved that the FMEA could reduce the RPN value from 250 to 125 for patient identification in inpatient supporting services (laboratory and radiology), pharmacy, and nutrition installations from June to September 2016 (Sithi and Widiastuti, 2018).
The advantage of HFMEA is the hazard scoring matrix process (Table 3). It is a grading method based on the hazard analysis value that determines the severity of the potential risk. After the determination of grading method, corrective action must be immediately planned through step 5 in the action and outcome measures, where the HFMEA team determine preventive steps to avoid potential risks or mitigation steps to minimize the impacts. Then, the team determine whether the failure mode assessed needs to be continued to the next process through the HFMEA Decision Tree Analysis. The decision tree analysis considers not only severity and probability scores, but also assesses the criticality of the failures (i.e, single point weaknesses) and whether there are controls in place to prevent or detect these failures. The use of "yes" or "no" responses in the HFMEA decision tree to asses the criticality, presence of control measures, and detectability of the failure mode is less subjective (B, NA and N, 2012) and more easily agreed upon than assigning scores (Chadwick and Fallon, 2012).

Healthcare Failure Mode and Effect Analysis Design
Based on the literature review, the author designed an HFMEA draft to identify potential risks in the laboratories. The laboratory procedures and potential risks in the HFMEA draft are retrieved from various studies that have been reviewed.
Step 1: Define the topic The HFMEA design specially is applied in the pre-analytical phase because the highest error rate there ranges from 49.2% to 84.5% (Sinici Lay, Pinar and Akbiyik, 2014;Patel et al., 2018).
Step 2: Assemble the team The HFMEA team consists of multidisciplinary hospital staff who are directly involved in the laboratory risk analysis process. Generally, laboratory personnel consist of clinical pathology specialists, health analysts, nurses, and administrative staff (Ministry of Health Republic of Indonesia, 2010). The head of the clinical laboratory leads the team members during the HFMEA process and ensures them to follow the steps and record HFMEA results. The team members include staff with 5 years experience who fully participate in the implementation of HFMEA to advice based on their respective work experiences. In designing the HFMEA, the author used blood specimen collection process initiated by Jiang et al. (2014). Phlebotomists (nurses or laboratory analysts) would perform the blood specimen collection.
Step 3: Graphically describe the process To identify potential risks, it is required to analyze all possible failure in each step. Figure 2 shows the flow of blood specimen sampling in a hospital laboratory. The sampling process consists of pre-analytic (5 stages), intra-analytic (2 stages), and post-analytic (2 stages) (Jiang et al., 2014). The researchers took the scope in point 3 for the specimen collection because the error rate there had potential risk of specimen rejection. Then the subprocesses are reconfigured into the flowchart (Figure 3) to allow a list of failure modes in each subprocess. Each failure mode will be transferred to the HFMEA worksheet (Table 4) to analyze potential causes and assess the Hazard Score.

Formula 1. Hazard Analysis
Step 4: Conduct a hazard analysis The HFMEA team determined the hazard analysis value by multiplying two components of risk severity and probability ( Table 2 with the Formula 1). After the hazard analysis values were obtained, the team determined the hazard scoring matrix (Table 3) (DeRosier et al., 2002). The scoring is a simulation number taken by laboratory analysts to provide an overview of the HFMEA process. Table 4, in the decision tree analysis coloumn must be filled by the HFMEA team to make adjustments to the hospital laboratory conditions.

Step 5: Develop actions and outcome measures
At this stage, the team identified actions in accordance with the standard outcome measures, determined people in charge and required leadership commitment in the forms of monitoring and evaluation to develop patient safety culture in a laboratory (DeRosier et al., 2002;Chadwick and Fallon, 2012).