DIFFERENT EFFECT OF SHIFT WORK ON FATIGUE AND WORK RELATED STRESS IN EMERGENCY ROOM NURSES AT THE HOSPITALS IN BADUNG AND DENPASAR REGENCY, BALI, INDONESIA

Background: Shift work is a way to maintain proper health care daily at hospital. Shift work may affect fatigue level of nurse and work related stress in Emergency Room (ER) nurses. Generally there are two types of shift work, such as two and three division time a day. The three-division time shift includes 6-6-12 shift and 7-7-10 hours shift, and twodivision time includes 12-12 hours shift. Lack of studies discusses about shift work on fatigue and work related stress in ER nurses. Objective: The aim of this study was to identify the differences of the effects between 6-6-12 shift, 7-7-10 shift, and 12-12 shift on fatigue and work related stress, and determine the dominant indicator influencing fatigue and work related stress in emergency nurses at the hospitals in Badung and Denpasar regency, Bali Indonesia. Methods: This was an observational analytic study with cross sectional approach. Purposive sampling was done to recruit 102 nurses from eight emergency departments at Badung and Denpasar regency. Occupational Fatigue Exhaustion Recovery (OFER) questionnaire was used to measure fatigue, and Expanded Nursing Stress Scale (ENSS) questionnaire was used to measure work related stress. Univariate analysis was used to analyze demographic characteristics of each ER. Kruskall-Wallis test with post hoc Mann Whitney were used to determine the different score of fatigue and work related stress between each group, and Structural Equation Modeling (SEM) was used to identify contribution of fatigue toward work related stress of ER nurses. Results: There were significant difference of the effect of shift work on fatigue, between 6-6-12 shift and 12-12 shift with p = 0.037, and between 7-7-10 shift and 12-12 shift with p = 0.003; and significant difference of the effect of shift work on work related stress, between 6-6-12 shift and 12-12 shift with p = 0.474, and between 7-7-10 shift and 12-12 shift with p = 0.128. SEM results show that fatigue contributed about 61% to increase work related stress in ER nurses. Conclusion: There was significant difference of the effects on fatigue between two and three-division time of shifts, and there was no difference of the effect on work related stress for each group. Fatigue statistically increased work related stress in ER nurses.


INTRODUCTION
The National Health Insurance Program (JKN) as the embodiment of the National Social Security System (SJSN) has been implemented since January 1, 2014. After implementation of National Health Insurance, there has always been a deficit in the year of 2014, 2015, 2016, and 2017, i.e. there was negative balance between the income of BPJS Kesehatan from patients' premium and the amount of money that BPJS Kesehatan had to pay to first level health facility and Indonesia Case Base Group (INA-CBG's) claims in advanced referral health facility (FKRTL)/hospital. One of the causes was JKN fraud inpatient JKN participant hospital service (Cahyono, 2015;Hartati, 2016;Tariden, 2017).
JKN Fraud is intentional dishonest or unfair action to obtain claims that is larger than normal for fraudulent and financial loss for others in the National Health Insurance Program (JKN) service (Ariati, 2015). JKN fraud is a white-collar crime, which one of the causes is the difference between INA-CBG's tariff rates based on the severity of diagnosis or procedure. Fraudulent healthcare is contagious if the Ministry of Health, as regulator, or BPJS Kesehatan, as executor of the guarantee, does not act. Healthcare facility that commit undetected JKN fraud and unpenalized would be an example to other healthcare facilities. Without prevention and penalizing action, BPJS Kesehatan financial losses will continue to grow (Ariati, 2015;Hendrartini, 2014;Honer, 2015;Sutoto, 2014;Trisnantoro, 2014). JKN fraud could be conducted by participants of National Health Insurance, BPJS Kesehatan, advanced referral health facilities (FKRTL), drugs and medical devices providers (Busch, 2012;MOH, 2013).
Fraud prevention in healthcare services had been done by stakeholders of National Health Insurance services as the following: a) The government as regulator has taken precautions by establishing fraud indicator, service standards, therapeutic standards, drug standards and medical devices that can be used in all healthcare services. The Government, together with BPJS Kesehatan, would monitor and evaluate the implementation of the National Health Insurance in relation to the potential fraud; b) Routine investigations by insurance companies on claims filed by healthcare facilities; c) Healthcare providers/hospital facilities filed claims in accordance to services provided to National Health Insurance participants, providing standardized services and benefits such as fulfilling the right of participants, hospitals conduct internal verification by Internal Supervisory Unit (SPI); and d) Insurance participants provide their identity so as not to be abused by unauthorized parties, requesting information pertaining to services provided by healthcare providers (Jasri, 2016).
The amount of potential loss caused by JKN fraud prompted the government to issue Permenkes No. 36 Year 2015 on fraud prevention in National Health Insurance Program (JKN) within National Social Security System (SJSN), as a legal basis for the development of JKN's anti-fraud system in healthcare services in Indonesia. Since its launch in April 2015, the regulation has been implemented ineffectively, which causes fraudulent impact on healthcare services and potentially increases fraud case, and yet there was no sufficient fraud control system. Healthcare providers are in the spotlight in healthcare fraud prosecution, as worldwide research shows that 60% of healthcare fraudulence comes from healthcare providers (Fadjriadinur, 2015).
In early 2017, the Corruption Eradication Commission (KPK) reported 1 million claims with potential JKN fraud, hence currently KPK is trying to build a JKN fraud prevention, detection and management system which involve all JKN stakeholder executives such as Ministry of Health, BPJS Kesehatan, healthcare facilities, medicine and medical device providers (Suparman, 2017). The purpose of this study was to assess the validity, reliability and effectiveness of prevention and early detection system of national healthcare insurance fraud based on computer application that contains fraud indicators based on Permenkes No. 36 Year 2015 on fraud prevention in National Health Insurance Program (JKN). We conducted this study at RSUP dr. Soeradji Tirtonegoro Klaten as the advanced referral health facility (FKRTL).

Study Design
The research method used was cross sectional (Creswell & Creswell, 2017). Potential fraud was conducted by administration officer, BPJS Kesehatan officer and internal hospital verification officer that were taken in one episode in inpatient JKN participant service, which started from registration until claim submission to BPJS Kesehatan.

Population and Sample of this Study
Population is inpatient JNK participants. Sample of this study were as follows: 1) Inpatient JKN participant, 2) BPJS Kesehatan officer, and 3) hospital internal verification officer.

Research Material
Research materials consisted of: 1) Informed consent, for patient/family approval, admission officer (TURP), BPJS Kesehatan officer and hospital internal verification officer willing to be involved in the research; 2) Computer application ( Figure 1) containing fraud indicators (Table 1); 3) The Kappa test questionnaire to test for the agreement on fraud indicators. TURP officers, BPJS Kesehatan officers and internal verification officers filled questionnaires in two different times, with one week interval; and 4) Hot-fit questionnaire to assess the reliability of prevention and early detection system for potential fraud.  Reducing the benefits that the participant is entitled to 3 Changing uninsured service into insured service 4 Conducting downcoding 5 Conducting bundling of service C Hospital 1 Conducting self-referral 2 Conducting kickback 3 Conducting readmission intentionally 4 Conducting unnecessary treatment 5 Conducting no medical value 6 Conducting no standard of care 7 Conducting over-utilization 8 Conducting unbundling / fragmentation 9 Conducting outpatients service into inpatients service 10 Manipulating length of stay into longer duration 11 Manipulating date of service 12 Conducting phantom visit 13 Conducting phantom procedure 14 Conducting cancelled service and still claim the service 15 Raising type of room charge 16 Conducting up coding 17 Conducting Diagnostic Related Group (DRG) creep 18 Separating one diagnosis into more than 1 19 Adding symptoms from a diagnosis 20 Conducting keystroke mistake 21 Conducting error in determining main diagnosis 22 Conducting error in determining main procedure 23 Conducting cloning 24 Conducting phantom billing 25 Conducting inflated bills 26 Conducting repeat billing 27 Charging fee to the patients treated according to his class' rights 28 Conducting cream skimming 29 Referring patient when INA-CBG's claim is used up 30 Manipulating ventilator usage into longer duration Independent variable was the prevention and early detection system of fraud in JKN participant's inpatient services. Dependent variables were validity, reliability, effectiveness of prevention and early detection system of fraud in JKN participant's inpatient services.
Validity is often defined as the extent to which an instrument measures what it purports to measure, the instrument measure what it is intended to measure. Validity requires that an instrument is reliable, but an instrument can be reliable without being valid (Eldridge, 2007;Kimberlin & Winterstein, 2008). Reliability estimates are used to evaluate: (1) the stability of measures administered at different times to the same individuals or using the same standard (test-retest reliability) or (2) the equivalence of sets of items from the same test (internal consistency) or of different observers scoring a behavior or event using the same instrument (interrater reliability). Reliability coefficients range from 0.00 to 1.00, with higher coefficients indicating higher levels of reliability (Kimberlin & Winterstein, 2008). The reliability of a product (or system) can be defined as the probability that a product will perform a required function underspecified conditions for a certain period of time When a system fails to perform satisfactorily, repair is normally carried out to locate and correct the fault. The system is restored to operational effectiveness by making an adjustment or by replacing a component. Maintainability is defined as the probability that a failed system will be restored to specified conditions within a given period of time when maintenance is performed according to prescribed procedures and resources (Pham, 2006).
Effectiveness of prevention and early detection system of national healthcare insurance fraud based on computer application depends on: a) Willingness of hospital director to implement anti-fraud system; b) Socialization of fraud indicator to all hospital officers; c) Acceptance of hospital officer to system and computer application that can simplify their task in prevention and early detection of fraud; d) Competence of hospital officers to operate computer application; e) Competence of hospital/BPJS Kesehatan officers to decide whether inpatient JKN participant service is a potential fraud or not.

Data Analyses
Validity was defined as decision whether the inpatient JKN participant service was fraudulent or not. Validity was assessed by Fischer exact test of the interpretation of fraud indicator between hospital internal verificator and BPJS Kesehatan officer.
Reliability was defined the consistency of system using by user to prevent and detect a potential fraud. Reliability was assessed using HOT-Fit research questionnaire (Human Organization Technology and Benefit)

Distribution of research subject
As shown in the table 3, research participants were as follows: a) Inpatients JKN participants by purposive sampling 1.106 of 5.548 (19.93%). Only 20% of the population agreed to be enrolled because: they had no time, they were in a hurry, especially emergency patient; not interested in the research, etc; b) 9 of 9 (100%) admission administrator officers were interested in the research; c) 2 of 2 (100%) BPJS Kesehatan officers were interested in the research; d) 20 of 20 (100%) internal hospital verification officers were interested in the research.  Table 4 shows that there were 9 potential fraud cases in RSUP dr. Soeradji Tirtonegoro from May-July 2017 classified as follows: 1) Readmission: 4 (44.44%); 2) Changes from uninsured into insured by JKN: 2 (22.22%); 3) Keystroke mistake: 1 (11.11%); 4) Fragmentation/unbundling: 1 (11.11%); and 5) Cancelled service: 1 (11.11%). Readmission was the highest case in this study. Because RSUP dr. Soeradji is a regional referral hospital in Klaten and the cases admitted were usually severe, the potential for rehospitalization in one month as limitation for readmission term was high. There was no fraud by inpatient JKN participant or BPJS Kesehatan. Validity of prevention and detection system Table 5 shows that Fischer exact test shows that only 2 out of 1.106 service claims were interpreted differently between hospital internal verification officer and BPJS Kesehatan officer. P value < 0.001, data show that system for prevention and early detection of national healthcare insurance fraud based on computer application were valid to be implemented as anti-fraud system in the hospital.

Reliability of prevention and detection system
Reliability test as shown in the table 6 for prevention and early detection system of national healthcare insurance fraud based on computer application was done using HOT-Fit questionnaire and Stata® software to obtain Cronbach's Alpha value with the following results: 1) TURP officer score 0.8088-0.8849; 2) BPJS Kesehatan officer score 1.00; 3) Hospital internal verification officer score 0.8329-0.9458. The data shows that prevention and early detection system of national healthcare insurance fraud based on computer application is reliable to be implemented as anti-fraud system in hospital.   Table 7 shows that there were 9 potential fraud findings within May-July 2017: 8 cases in May (88.88%), 1 case in June (11.12%), and 0 (0%) in July. Potential fraud in hospital was significantly reduced, therefore, prevention and early detection system of national healthcare insurance fraud based on computer application is effective to be implemented as anti-fraud system in hospital.

DISCUSSION
Tests or instruments that are valid and reliable to measure such constructs are crucial components of research quality. Key indicators of the quality of a measuring instrument are the reliability and validity of the measures (Kimberlin & Winterstein, 2008).
Validity is often defined as the extent to which an instrument measures what it purports to measure, the instrument measure what it is intended to measure. Validity requires that an instrument is reliable, but an instrument can be reliable without being valid (Eldridge, 2007;Kimberlin & Winterstein, 2008). In this study, the system measured what it was intended to measure such as potential fraud or not, although there were only 2 out of 1.106 claims with different interpretation (disagree) of fraud or not between BPJS Kesehatan and internal hospital verification officer, 1.097 claims had the same interpretation (agree) of no potential fraud, and 7 claims with the same interpretation (agree) of potential fraud.
Reliability estimates are used to evaluate: (1) the stability of measures administered at different times to the same individuals or using the same standard (test-retest reliability) or (2) the equivalence of sets of items from the same test (internal consistency) or of different observers scoring a behavior or event using the same instrument (interrater reliability). Reliability coefficients range from 0.00 to 1.00, with higher coefficients indicating higher levels of reliability (Kimberlin & Winterstein, 2008). The reliability of a product (or system) can be defined as the probability that a product will perform a required function underspecified conditions for a certain period of time When a system fails to perform satisfactorily, repair is normally carried out to locate and correct the fault. The system is restored to operational effectiveness by making an adjustment or by replacing a component. Maintainability is defined as the probability that a failed system will be restored to specified conditions within a given period of time when maintenance is performed according to prescribed procedures and resources (Pham, 2006).
Effectiveness of system for prevention and early detection of national healthcare insurance fraud based on computer application depends on: a) Willingness of hospital director to implement anti-fraud system; b) Socialization of fraud indicators to all hospital officers; c) Acceptance of hospital officers to apply the system and computer application that can simplify their task in prevention and early detection of fraud; d) Competence of hospital officers to operate computer application; e) Competence of hospital officers to decide whether a health service is potentially fraudulent or not. In this study, there was a significant decrease in potential fraud from 8 findings in May to 1 finding June and no finding in July (effective).
Development of fraud prevention system, as Permenkes No. 36, 2015 stated, must be done through three processes: a) Hospital formulates internal regulation in the form of good governance of organization and clinical services b) Hospital can develop healthcare facilities oriented in quality and cost control by utilizing effective and efficient management, evidence-based information technology and formation of fraud prevention team in the hospital; c) Hospital can develop a fraud prevention behavior as part of organization management and clinical management oriented to quality control and cost control based on TARIK (Transparency, Accountability, Responsibility, Independency, Fairness) principle (Hartati, 2016;Jasri, 2016).
Hospitals can prevent any potential fraud by forming fraud prevention team in the hospital that is responsible for: a) Creating director circular letter on fraud prohibition; b) Early detection of fraud based on service claim data; c) Socialization policy, regulations and new customs oriented on quality control and cost control; d) Improving coder, medical doctor and other officers' capability regarding claims; e) Taking precautions, detection and manage fraud; f) Monitoring and evaluation; g) Establishing commitment between hospital and BPJS Kesehatan in case of overpayment, steps on how to cooperate, and in case of fraud suspicions, clarification should be made by the hospital; h) Internal verification by SPI before submitting the claim; i) Developing clinical practice guideline and the clinical pathway for each diagnosis; j) Reporting to hospital chief director every six months (Hartati, 2016;Sutoto, 2014).
Hospitals should optimize fraud prevention team who would spearhead the development and implementation of fraud prevention and detection system. Ministry of Health Regulation (Permenkes) No 36/2015 stated that this team should at least consist of internal examination unit element, medical committee, medical recorder, coder, and other related elements. The team's task is conducting prevention and early detection for fraud based on claim data to BPJS Kesehatan, socializing regulations orienting on quality and cost control to support implementation of good organizational and clinical governance. Fraud prevention team's competence on fraud prevention and detection should also be improved (Hartati, 2016;Jasri, 2016).
In this study, most potential fraud event in National Health Insurance by hospital was readmission caused by: 1) RSUP dr. Soeradji Tirtonegoro as referral hospital in Klaten district, handled severe referral cases with a potency to relapse within less than one month after discharge; 2) Hospital staffs were not careful in detecting patients who were readmitted in less than 30 days after being discharged by the doctor, where the claim should be in the same episode as the previous admission. Hospitals should pay more attention for potential fraud in readmission so that the same mistake would not be repeated again.
Readmission reduction program could be achieved by: 1) Hospital staffs manage patients according to standard of quality; 2) Home visit service conducted soon after discharge; 3) Disease management: (a) support the physician or practitioner/patient relationship and care plan; (b) prevent exacerbations and complications utilizing evidence-based practice guidelines and patient empowerment strategies; and (c) evaluate outcomes in an ongoing basis; 4) Post-acute care: many patients are not discharged directly from the hospital to the home, but instead go to longterm acute care hospitals, inpatient rehabilitation hospitals or skilled nursing facilities where, in addition to rehabilitative services, they can receive around-the-clock medication management (Hubbard & McNeil, 2012;Wier, Barrett, Steiner, & Jiang, 2006).
Barriers to implementation and successful outcomes: 1) Incomplete and inaccurate patient medication lists: Hospital staff report many of the same difficulties faced by office-based physicians in assembling an accurate list of each patient's prescription medications on a timely and cost-effective basis; 2) Limitations of family caregiver or other sources of patient support: For patients experiencing a decline in cognitive function, a family caregiver can be the de facto medication manager; 3) Difficulty scheduling timely follow-up visits for primary care physicians or community-based specialists; 4) Funding challenges: The new discharge planning and transitional care models represent intensive, high-touch patient care approaches that can be difficult to fund in the long term (Hubbard & McNeil, 2012).

Potential fraud event in National Health
Insurance which uninsured patients were changed into insured in this study occurred in: a) healthy newborns who should be claimed together with the mothers' claim, but were claimed separately; b) administration error such as incomplete admission requirement after 3 days but was still submitted for claim (BPJS Kesehatan, 2014). Keystroke mistake could occur due to hospital staffs' carelessness when inputting patients' entry data; however potential fraud cases in this study did not increase the hospital's claim amount and hence was not detrimental to BPJS Kesehatan's finance (Mardha, 2014).
Fragmentation/unbundling were caused by submission of inpatient JKN participant claim together with outpatient claim; there should only be one inpatient JKN participant claim submitted to BPJS Kesehatan (Dodaro, 2015). Canceled service happened because hospitals have treated patients in the emergency room and had spent resources during temporary treatment before referring patients to other hospitals because the hospital was unable to provide comprehensive patient care (Thorpe, Deslich, Sr, & Coustasse, 2012). Limitation of this study include: 1) the success of prevention and early detection system implementation depend on capability of admission administrator, internal hospital verificator, BPJS Kesehatan to decide whether there is fraud or not; 2) computer application was not bridging yet with information system of hospital or INA-CBG's software to simplify implementation, the officer only need to input patient's medical record number, not the entire patient identities were put into the fraud information system; 3) Fraud indicator in this study was still incomplete, further studies are needed to add new fraud indicators to complete fraud indicator.

CONCLUSION
Prevention and early detection system of national healthcare insurance fraud based on computer application used in this study is valid, reliable, and effective to use as prevention and early detection system for potential fraud in inpatient JKN participant service in the hospital.
The system in this study can be adopted by other hospitals with several conditions, such as: 1) hospital should have an information technology system or server with local area network, 2) hospital has a willingness for antifraud system, 3) hospital has a hospital internal verification staff as well as workshop and techincal support and good understanding of fraud indicators that are needed for all staffs to operate properly.
It is suggested that further study is needed for automatization of computer application to minimize or simplify the activity of officers to prevent and early detect potential fraud, and need to be implemented in other hospitals.

INTRODUCTION
Emergency Room (ER) is a special unit in every hospital that gives 24-hour health service, and. to support its optimal service, shift work is applied (von Treuer, Fuller-Tyszkiewicz, & Little, 2014). Shift work, although having a positive impact, also has a negative impact for ER nurses as a profession that have shift work. Some of the negative impacts identified from the application of shift work such as increased blood pressure and pulse rate, physiological headache, changes and decreased sleep time, risk for weight gain, increased work related stress and increased fatigue for nurses who works in the ER (Kim et al., 2013). Among those impacts, fatigue and work related stress become a concern nowadays in emergency nurses.
Viewed from its type, most of the shift work applied in Indonesia is a rotational shift work and characterized by dividing working hour to two or three times a day (Dillingham, 2017). According to the preliminary study at several hospitals in Badung and Denpasar regency, there are two kinds of shift work that is applied, namely: 1) shift work that divides working hour become three times a day, such as six hours for morning shift, six hours for noon shift, and twelve hours for night shift (6-6-12); and seven hours for morning shift, seven hours for noon shift, and ten hours for night shift (7-7-10). 2) shift work that divides working hour into two times a day, 12 hours for half day shift and 12 hours for night shift (12-12). The 6-6-12 shift is applied in Surya Husadha Hospital, Prima Medika Hospital, BaliMed Hospital, and Kasih Ibu Hospital; the 7-7-10 shift is applied at Bali Royal Hospital (BROS) Denpasar, and Siloam Bali Hospital, and 12-12 shift is applied at BIMC Hospital in Kuta and BIMC Siloam Hospital in Nusa Dua.
Based on the results of a survey conducted in the United Kingdom, out of 700 nurses, as many as 60% (420 people) experienced work related stress (Ford, 2014). In addition, another study conducted in Brunei states that nurses who work in the ER are 4 times more likely to experience violence from patients and their relatives compared to nurses working in other rooms, and 2.8 times more likely to experience work-related fatigue (Rahman, Abdul-Mumin, & Naing, 2017). Other literature also mentions that ER nurses tend to experience fatigue more quickly and 66% higher work related stress than nurses working in wards (Weigl & Schneider, 2017). In this case, it can be said that the ER nurse is one of the health professions who work with shift and are at risk to get stress and fatigue (Salilih & Abajobir, 2014).
Fatigue and work-related stress are the problems not only for nurses as a shift worker, but also for quality of care in hospital. However, fatigue can also affect work related stress of ER nurses (Undap, Ratag, & Kawatu, 2016). ER is a unit with full stressors; such as process of handling patients in limited time, unexpected number of patients, and also types of complaints that vary from all patients, which requires a quick decision-making. All of these things will increase fatigue and work related stress. In fact, more problems will be coming if fatigue and work related stress occur continuously, which will result in chronic fatigue or exhaustion (Debora, 2016). If it is not immediately identified and solved, both of these factors obviously will affect patient safety in the hospital.
Additionally, working long hours will incriminate the physical and mental of nurses, especially if the number of patients out of the capacity of the ER. It can also raise the conflict between professions in the ER team. In addition, the lack of time to socialize both in the family and the community become one of the impacts of the shift work.
Based on the results of preliminary study, researcher also obtained information that the measurement and evaluation of fatigue and work related stress of nurses in ER are never been done. Therefore, this study aimed to compare the effect of shift work on fatigue and work related stress, and determine the dominant indicator for each shift work groups in emergency rooms of the hospitals in Badung and Denpasar regency.

Study design
This research was a quantitative research with observational analytic design and crosssectional approach. This research was conducted from January 20 th , to February 27 th , 2018 in eight Emergency Rooms of the hospitals at Badung and Denpasar regency.

Conceptual framework
Emergency nurses work in shift work, which leads to fatigue as one of its negative effect. There are three indicators affecting the level of fatigue, namely: i) Chronic fatigue condition, which is from accumulation of fatigue that ER nurses feel from work, ii) Acute fatigue condition, which come directly after work, and iii) Intershift recovery, which is the time needed to recover after work. Those three indicators will affect fatigue level as long as they work. Besides, the other important problem, which may affect quality of care, is work related stress. According to Expanded Nursing Stress Scale (ENSS) questionnaire, there are 9 indicators were identified as condition which can influence work related stress of ER nurses. These indicators were 1) lack of experience related to death and critical condition, 2) conflict with other professional, 3) feel lack of preparation to support patient or patient family emotionally, 4) problems with fellow co-workers, 5) conflict with the supervisor, receiving minimal support from administrator, 6) working more due to lack of organizational capability, unfamiliarity with work units, staffing and unexpected scheduling in new complex work environments, 7) uncertainty of care and less information related to the patient's health condition, 8) fear of failure to do nursing intervention because patients and their families make unreasonable requests, and 9) experiencing discrimination and isolation. From these indicators, lack of studies conducted in Indonesia to determine dominant indicators influencing fatigue and work related stress in ER nurses.

Sample
The total sample of this study was 102 ER nurses selected using purposive sampling technique from eight hospitals. The inclusion criteria were functional nurses who worked based on shift, had experiences in morning, noon, and night shift, had working experience in ER at least for three months, and agreed to become a respondent by signing informed consent. The exclusion criteria were nurses in charge as a leader or structural in ER and also functional nurses who took a break/annual leave from work.

Data analysis
Kruskall Wallis test and post hoc with Mann Whitney with α < 0.05 were used to determine the difference of score of fatigue and work related stress between three groups of ER nurses based on their shift work. Structural Equation Modeling was used to determine the dominant indicator that influences fatigue and work related stress. In multivariate analysis, each variable and its indicators were given a code to make the analysis easier. Independent variables were transformed into dummy variables, from three variables become two variables. Dummy 1 stands for 6-6-12 shift group with code 1 and others are 0; Dummy 2 stands for 7-7-10 shift group with code 1 and others are 0. As for dummy variables were created, it could be said that 12-12 shift group becomes a comparison group. Fatigue variable was given a code F, and for its indicators were given code F1 (Chronic fatigue condition), F2 (Acute fatigue condition), and F3 (Intershift recovery). Work related stress was given a code S, and for its indicators were given code S1 (Lack of experience related to death and critical condition), S2 (Conflict with other professional health such as surgeons and doctors), S3 (Feel lack of preparation to support patient or patient family emotionally), S4 (Problems with fellow co-workers), S5 (Conflict with the supervisor, receiving minimal support from administrator), S6 (Working more due to lack of organizational capability, unfamiliarity with work units, staffing and unexpected scheduling in new complex work environments), S7 (Uncertainty of care and less information related to the patient's health condition), S8 (Fear of failure to do nursing care because patients and their families make unreasonable requests), and code S9 (Experiencing discrimination and isolation).

RESULTS
As shown on Table 1, most respondents on this research are males as many as 65 nurses (63.7%), Majority of respondents aged 26-35 years as many as 73 people (71.6%), and had bachelor level as educational background as many as 53 people (52%), and were married as many as 58 respondents (56.9%). In addition, the respondents had working experiences for 25-48 months as many as 34 people (33.3%).
Mostly respondents in Surya Husadha Hospital were working on 6-6-12 shift, BROS Hospital in 7-7-10 shift, and BIMC Nusa Dua in 12-12 shift. Bivariate analysis result were shown in Figure 1 and Figure 2 with box plot.   Figure 1 shows that there was a significant difference in fatigue score between the 6-6-12 shift group and 7-7-10 shift group against shift 12-12 group (p < 0.05). There was no significant difference between 6-6-12 shift and 7-7-10 shift group (p = 0.218). From the figure 1, it shows that the 12-12 group has slightly a higher score than 6-6-12 and 7-7-10 shift groups.  Figure 2 shows that there was no significant difference of work related stress score between three shift groups (p > 0.05). It can be said that shift work affects work related stress, but there was no different score between these groups. For multivariate analysis results were shown in figure 3. SEM analysis results shows that the most dominant indicator increasing the nurse's fatigue was chronic fatigue condition (F1) with value of 0.62. For the work related stress, the most influencing indicator was the uncertainty of care and less information related to the patient's health condition (S7) with a value of 0.78. In Dummy 1 and Dummy 2 variables which lead to fatigue variable, the difference value results in negative which means 12-12 group were more tiring compared to the 6-6-12 and 7-7-10 group with value of -0.25 and -0.44 respectively. As for work related stress, the difference between three shift groups was not too significant. The different value was positive, which means that 6-6-12 shift group and 7-7-10 shift group have a slightly higher stress score when compared to 12-12 group. In other words, although the difference in fatigue score was significant for the three groups, work related stress between the three groups was not statistically different.

DISCUSSION
In accordance with the results of Kruskal-Wallis bivariate analysis with post hoc Mann-Whitney, there was statistically significant difference in fatigue score between three groups with value of p < 0.05. Besides, it can be seen between the group of 6-6-12 and group of 7-7-10 that the fatigue score was not significantly difference (p = 0.218). It can be explained that the longer duration of daily shift work will result in a significant increase in fatigue, especially for nurses who work in the ER (Pryce, 2016). The same result also found in previous research conducted in Surakarta, which states that there was a significant influence (p = 0.001) between the shift work toward the level of fatigue of nurses (Pramitasari & Sri Darnoto, 2016). In this case, it can be said that the longer duration of work performed by a nurse in an environment with high stressors such as the ER, fatigue will increase significantly.
Generally, nurses are required to provide optimal services for patients who come to the ER. However, as one effect of shift work, fatigue could decrease nurse capacity to maintain a comprehensive service (Antill, 2016). It is said that the decrease in nurse ability to perform optimal services is influenced by the duration of work and the duration of rest. Shift work will affect the duration of work as well as the optimal Canadian Nurse Association also mentions that the existence of a shift work will affect sleep patterns, individual stress, work patterns, and demands from the work (CNA, 2008).
Viewed in terms of duration of work time, 12-12 shift has longer time of night work or curfew compared to 6-6-12 shift and 7-7-10 shift. Curfew is a biologically clock of human to sleep and recover their physic and mental after daily activities. The existence of longer night shift will certainly result in changes in the circadian rhythm. Study said that the longer night shift work they get, shift workers will experience chronic fatigue higher than nurses who rarely get night shift (Øyane, Pallesen, Moen, Åkerstedt, & Bjorvatn, 2013). This is the reason why fatigue score was higher on 12-12 shift than 6-6-12 shift and 7-7-10 shift group.
Every single nurse has their own resilience in managing fatigue, it might be the reason why the data distribution for fatigue scores uneven for each group. Seen from the duration of work, 12-12 shift group was higher than 6-6-12 shift and 7-7-10 shift group. In a week, nurses with 12-12 shift work three to four times, and the rest are off or free day. Although the duration of the free day can be said to be longer than the shift 6-6-12 and shift 7-7-10 groups, nurses with shift 12-12 are expressed their feeling that they are more tired due to previous work accumulations. Besides, there is an irregular schedule that leads unbalanced between work hours and recovery time, which is caused by the hindrance that comes from nurses. For example, the nurses who should have a morning shift, because of something like religious activities, cultural activities, family events, or illness so that he/she switch their schedule to another day or exchanged with other nurses or applying for annual leave. This event could affect the duration of work, as well as the duration of rest in the future. In accordance with Yumang and Burns statements, this irregular shift work hours, and longer duration of work than rest periods are the main factors leading to chronic fatigue in nurses (Yumang-Ross & Burns, 2014).
Nurse work related stress is a problem as well as a challenge for the management of an organization particularly in the hospital. Based on results of the analysis, it is statistically showed that there was no significant (p > 0.05) difference in score between each group. This may be due to difference of workload, work duration each shift, and self-motivation. Supported by previous research results indicated that workload, work motivation, and different work environment will affect the work related stress of a nurse (Murharyati & Kismanto, 2015). First, from the workload, according to the data obtained during the study, 6-6-12 shift group has the number of visits ranged from 400 to 1574 patients per month or equivalent to 13 to 53 patient visits per day. The 7-7-10 shift group has the number of visits ranged from 944 to 1583 patients each month or about 32 to 53 patient visits per day., and 12-12 shift group has the number of visits ranged from 150 to 481 patients per month or equivalent to 5 to 16 patient visits per day. From those data, it is concluded that the number of patient visits for 12-12 shift group significantly less compared with 6-6-12 and 7-7-10 shift group visits. According to the previous study, the number of patients who come in a day will affect the workload of the nurse, so it can be said that the nurse group with 6-6-12 and 7-7-10 shift have higher workload than 12-12 shift group (Pramitasari & Sri Darnoto, 2016).
Second factors that increase work related stress is duration of the work. It is said that work duration will change nurses' circadian rhythm and cortisol's level. There was a significant increase in cortisol level from the day off to the first and subsequent work first and significantly decreased after shift work towards holidays and off after night shift (Marchand, Durand, & Lupien, 2013). These results are also supported by another study which reveals that the longer the duration of recovery time for shift worker, the more optimal the person's performance will be (Lombardi et al., 2014). Based on recent study, it is said that the working hours and workload arrangements for each time of shift are not the same. This will lead to differences in coping mechanisms, differences in fatigue levels and of course different levels of work related stress. If connected to the situation in ER, which is characterized by unpredictable amount of patients, unpredictable emergency cases, different patient triage systems, priority mechanism of patient handling, fast decision making, and heavy workload, nurse will get fatigue easily than nurses' who work in other units. In other words, work related stress would continue to occur as long as a person works with a shift, but will decrease with sufficient recovery time (Rodwell & Fernando, 2016).
The last one is self-motivation as a reason why there is no difference in work related stress scores between each group. Self-motivation is an internal factor of nurses (Gunawan, 2016). Because of this factor nurse will give optimal effort to finish all their jobs (Drake, 2017). In addition, to achieve a goal, a person with high self motivation will have a strong desire to achieve something, then begin to assess the risks from efforts by taking the most minimal portion of risk, taking the initiative movement and always be prepared when the opportunity has arrived (Cooper, 2015). The existence of self-motivation will also help to realize the purpose and reason why a person survives despite unfavorable conditions. Similarly, in this study, nurses group with 12-12 shift even though their fatigue scores are higher than group of 6-6-12 and 7-7-10 shifts, because of their self motivation was high enough, work related stress score was not significantly different in each group.
Based on the results of multivariate analysis, each type of shift work can affect nurse's fatigue and work related stress. Shift work plays an important role in increasing and decreasing fatigue and work related stress from nurses. In SEM analysis, groups with 12-12 shift have fatigue score higher than group with 6-6-12 and 7-7-10 shift, shown by the difference value on dummy variables, which is -0.25 and -0.44. In other words, the work duration becomes a factor, which greatly affects the fatigue of ER nurses (Øyane et al., 2013). The longer work duration on one day will affect the nurse's fatigue, which further affects their performance, decreases quality of services, and increases the negative risk to the patient and the ease of conflicts within teammate (Caruso, 2014;Pryce, 2016). A nurse who works with a shift will experience a change of time to rest and sleep. The nurse will undergo a process of adaptation ranging from rest period changes, sleep quality and quantity, changes of heart rate frequency, chemical changes in the body such as cortisol and adrenaline. From the changes in the circadian rhythm, the nurse will experience tired, lackluster, muscular weakness and cognitive function, all of which are a definition for fatigue (Antill, 2016).
In accordance with SEM analysis, the indicator that plays a significant role in improving work related stress is an uncertainty of care and less information related to the patient's health condition. For this indicator, some points that include in this are the lack of exposure to information from doctors regarding the health status of patients, doctors who are absent in emergency situations, thus making the nurses fearful of making mistakes in treating patients. Similar statements are also mentioned by Stiell said that about the problem in the delivery of information by doctors to nurses will affect the results of patient care in the ER (Stiell, Forster, Stiell, & van Walraven, 2003). It can be explained that uncertainty of care caused by the lack of communication between team can affect the response given by nurse to the patients. As we know, nurse is the closest health profession and interacts directly with the patient for 24 hours. If there is a miscommunication between team, nurse will be the first one to be blamed by patients. Another point that became the vital point was the presence of doctors in the ER, most respondents answered the absence of doctors when the emergency situation as a cause of their stress while working in the ER. The nurse will feel in a burden and dangerous position when the responsibility to make a decision is submitted to them. All of these points are indicators of uncertainty of care process, which can lead to feeling guilty if bad happen after the decision was made, and lastly become the most indicators of work related stress for nurse in this study. As we know the most indicators, which contribute, to increase work related stress of ER nurses, it must become a concern by ER team as well as hospital management team as well. Based on this findings, management of each hospitals must conduct an evaluation about nurse's fatigue and work related stress especially in ER periodically to prevent negative effects on quality if care.

CONCLUSION
There is a significant differences score of fatigue between shifts that divided their work hour into three times a day toward group of shift that divided their work hour two times a day. But there is no significant differences score of work related stress between each group were studied.

INTRODUCTION
Phlebotomy is a routine process of blood sampling in laboratory and it should be done professionally. Therefore, a phlebotomist is expected to handle the situation according to guidelines based on their job (Pendergraph & Pendergraph, 1998). There are several types of phlebotomy, one of them is venous blood sampling. It used larger syringe sizes and sharp needle. Blood sampling (venipuncture) requires particular technique and it must be done by profesional health workers because the venous blood sampling is important (Booth, Wallace, & Fitzgerald, 2009). Venous blood sampling has several risks of complication. There are some complications of blood sampling. One of them caused by physiological or medical factor such as syncope (fainted). The most factor of syncope is emotional factors such as nervous, pallor and anxiety (Booth et al., 2009).
The feeling of nervous, pallor and anxious can be referred as anxiety. This anxiety arises from lack of unexplained knowledge and fear, a worry, no specific object, and it is subjective (Stuart & Sundeen, 2007). It takes a way to lower anxiety and pain in venous blood sampling. This reduction must be done by providing a comfort situation to patients so the syncope would not happen during venous blood sampling.
Anxiety and pain reduction can be done by pharmacological and non-pharmacological therapy. Non-pharmacological interventions need to be developed, related to the psychological problems that affected by anxiety and pain problems. The development of non pharmacologic therapy has also evolved to complementary therapies. People choose this therapy based on the low side effects (van der Watt, Laugharne, & Janca, 2008).
Hypnosis is one of the conscious conditions where people are able to accept the suggestions (information). Hypnosis also defined as the optimal condition of the subconscious mind or critical area (Nurindra, 2007). Based on the principles of human behavior 88% by sub-conscious and new values can be implanted in the sub-conscious through the Hypnosis process. This hypnosis process can be used for the initial process of venous blood sampling in patients to reduce the pain. The early phase of hypnosis is preinduction phase (early process before the actual hypnosis session) (Nurindra, 2007).

Emotional freedom technique (EFT) is a complementary therapy developed by Gary
Craig. EFT is a non invasive action and part of emotional acupressure based on energy power of acupunture. It was developed to manage stress and anxiety (Prameswari & Ariyani, 2015). EFT has proved and able to treat a lot of problems. Through the research of holistic hypnotherapy modification, it has been found that negative thoughts and emotions come from the subconscious mind. By this method, people are directed to be positive thinking. Other researh showed that the hypno EET method can reduce stress and anxiety (Iskandar, 2010;Temple & Mollon, 2011;van der Watt et al., 2008).

An early study in Medical Laboratory
Technology of Health Polytechnic of Palembang showed that 42% of students had anxiety and 34% of students got pain during venous blood sampling. In every year at phlebotomy class or practice of venous blood sampling found that 5% students were fainted during the class. The aim of this study was to determine the effect of hypno -EFT to reduce the anxiety and pain during venous blood sampling. This research was conducted by observing anxiety and pain during venous blood sampling.

Study design
This was a quasi-experimental study with pretest posttest with control group design.

Sample
Population of this research were all students in SMAN 22 Palembang. The technique sampling was simple random sampling. The total sample of the research was 35 students for a treatment group and 17 students in a control group.

Intervention
The researchers did intervention to both of groups. The control group was intervented by venous blood sampling for 2 -3 minutes, while the treatment group was intervented by giving the hypno-EFT method and standard phlebotomy procedure for 4-5 minutes. Measurement of anxiety and pain was done before and after intervention in both groups.

Instrument
Anxiety and pain were measured in this study. The anxiety of respondents was measured by using the HARS Modified Questionnaire. It is a standard questionnaire which consist of 25 questions to measure anxiety of respondent. Meanwhile, the pain was measured by Visual Analogue Scale (VAS) Tool, with a line span of 10 centimeters (Nursalam, 2008;Yudiyanta & Novitasari, 2015).

Ethical consideration
Ethical approval was obtained from the Health Research Ethics Commission at Health Polytechnic of Makassar. The researcher explained the objectives and procedures of the study, and asked for the subject's willingness to be the respondent in the study and asked participants to sign the informed consent.

Data analysis
Wilcoxon signed rank test was used for data analysis because the result of normality test using Kolmogorov smirnov was <0.05, which indicated that the data were not in normal distribution. In multivariate test after analysis between treatment group and control group, normal and homogeneous distribution data were obtained, then the analysis was continued to multivariate test of Manova (Kuzma & Bohnenblust, 1992). Table 1 about the distribution of anxiety before intervantion of treatment group showed that 1 person (2.9%) had no anxiety, mild anxiety was found in 9 people (25.7%), moderate anxiety in 20 people (57.1%), and severe anxiety in 5 people (14.3%) . Meanwhile after treatment, 23 people (65.7%) with no anxiety, mild anxiety in 11 people (31.4%) and severe anxiety in 1 person (2.9%).  Table 2 shows the mean of pain in the therapy group before the hypno-EFT was 3.20, the median was 3.00 and standard deviation was 1.91. The minimum value of pain was 0 and maximum value was 9. At the 95% level of confidence, the respondent got the pain at 2.55 -.3.85. While after intervention, the averages was 1.54, median was 1.00 and standard deviation was 1.597. The minimum value of pain was 0 and maximum value was 6. At the 95% level of confidence, the respondent got the pain 0.99 -2.09. While Wilcoxon signed rank test as shown in the Table 3 showed that p-value 0.000 (<0.05), which indicated that there was a statistically significant effect of hypno-EFT to decrease of anxiety and pain on treatment group compared to control group.   4 showed that there was significant effect of hypno-EFT method to lower the anxiety with p-value = 0.001 (<0.05) And also was affected to decrease of pain with p-value = 0.003 (<0.05).

Anxiety
Findings showed that there was a significant reduction of anxiety after given hypno-EFT, which shows that 23 people (65.7%) with no anxiety, 11 people (31.4%) with moderate anxiety, and 1 person (2.9%) with severe anxiety. According to Ali FM, et all showed that 13.95% adult got fear or phobia to syringe needle. Moreover, study said that there was another fear to get infectious diseases after using former syringe, scared of blood, scared of pain, and scared to get accidently sample exchanged (Prodia, 2016). In addition, there are several complications of venous blood sampling (Booth et al., 2009). One of them is fainting. It is occurs due to the lack of blood flow to brain. It caused a low blood volume, heartbeat fast and anxiety due to emotional factors (Boonstra, Schiphorst Preuper, Balk, & Stewart, 2014).
Hypnosis is a way of changing the condition of normal state to hypnosis state. This process was done by changing the external focus to internal focus. Humans were more suggestive at hypnosis conditions, so they can accept suggestions that can be turn into new values (Nurindra, 2007). EFT is a meridian energy therapy such as acupuncture. EFT works directly on the body's meridian system. As well as using a needle, EFT could stimulate the main meridian point by tapping it. Imagine that meridian is like river flow. Emotional or physical problems are the same as river flow inhibitors. The point at the meridian sends kinetic energy to the system energy and release the obstacles that block the flow of energy (Iskandar, 2010).
The research used hypno-EFT technique. It was performed only at the pre-induction stage and then continued with verbal suggestive EFT technique. There are several requirements in this process. As an example, respondents should have a willingness as volunteer, able to communicate and have the ability to focus. The result was respondent's anxiety decreased, but one respondent was in severe anxiety, which was caused by the respondent who was not convinced by hypno-EFT therapy and had small veins.

Pain
Results of this study also showed that there was a significant decrease of pain after given hypno-EFT, which found that the average of pain was 1.54, the median was 1.00 and standard of deviation was 1.597, with the minimum value of pain was 0 and 6 for maximum. According to Boonstra, et all (2014) the range value on VAS for measuring the pain is ≤ 3.4 as mild pain, 3.5-7.4 as moderate pain and ≥ 7.5 as severe pain. Based on this category it can be concluded that the average pain of respondent was mild pain. The range of pain was mild pain to severe pain (Boonstra et al., 2014).
This result was appropriate to early study in medical labolatory technology of palembang indicated that 34% of students got pain during venous blood sampling. The average of pain of respondent was decreased, but, there were 4 people (11.42%) had increased pain and anxiety, which is caused by small veins, a deep veinpuncture and nervousness during blood sampling.

Effect of Hypno-EFT in reducing anxiety and pain
The statistical result obtained p-value of 0.001 (<0.05) for anxiety, and 0.003 (<0.05) for pain level, which indicated that there was a significant effect of hypno-EFT in decreasing the anxiety and pain during venous blood sampling. These results were in line with previous study who found the therapeutic benefits of EFT (Emotional freedom techniques) therapy (Prameswari & Ariyani, 2015). Other studies also showed that 50.91% respondents got severe anxiety during tooth removing. It because people are afraid of syringe and related tools. EFT had an effect on decreasing anxiety in pre-surgery patients (Yahya, 2015). Another study also showed that there was an effect of hypnoteraphy to derease stress levels in psychosomatic gastritis patients (Marthaningtyas, 2012). This result was also in line with previous study that there was effect of hypnoteraphy in decreasing anxiety and pain in post-surgery patients (SUMARWANTO, 2015), and also an effect of hypnoteraphy with a combination of analgesic in decreasing pain in post-sectio caesarea patient (Niraski, 2015). Moreover, EFT could reduce the pain in patient of post fremur fracture surgery (Mudatsyir, 2012). In principle, 88% of human behavior is influenced by the subconscious. New values can be implanted with hypnotherapy and energy balancing by the EFT method. In this process, the suggestions were given to the respondents to relax and feel comfort, which is purposely to decrease the pain during venous blood sampling.

CONCLUSION
It can be concluded that there were significant effects of Hypno-EFT therapy to decrease the anxiety and pain during venous blood sampling. This method is recommended as an alternative procedure in venous blood sampling with complicating factors.

INTRODUCTION
Globally, it is estimated that 10.7 million women died from complications related to pregnancy and childbirth between 1990 and 2015 (Alkema et al., 2016). Most maternal deaths and birth complications in low-and middle-income countries are caused by health problems, which largely can be prevented (Unicef, 2014). If women have access to emergency obstetric care under the supervision of skilled birth attendants then maternal mortality can usually be avoided. Globally, interventions to reduce maternal mortality rate (MMR) have decreased from 385 deaths per 100,000 births in 1990 to 216 in 2015 (Alkema et al., 2016). However, there is still a gap between skilled birth attendants (SBA) coverage in developed and developing countries, which may also occur between regions within a country (Graham, Bell, & Bullough, 2001).
Despite many efforts in developing countries, a high proportion of deaths continue to occur at home in unhygienic conditions without presence of SBA and the necessary infrastructure for referral if complications arise (Dogba & Fournier, 2009). Various factors are related to preference for homebirths, such as lower cost, geographical barriers to access health facilities, and low status of women in the family in India and Laos (King, Jackson, Dietsch, & Hailemariam, 2015). Likewise, women in Indonesia also prefer home delivery assisted by TBAs for normal delivery (La Ode & Asfian, 2016;Titaley, Hunter, Dibley, & Heywood, 2010).
Cultural and social factors such as tradition, gender and social structure of decision making are important and may hinder delivery at health facilities, both for normal delivery and when complications occur (I. Jansen, 2006). In Indonesia, despite availability of village midwife and insurance scheme for birth, a study in 12 districts indicated that many mothers are still unaware of the need for safe delivery in health facilities (Handayani, Suharmiati, Kurniawan, Nuraini, & Wasito, 2014). Therefore, a qualitative research is still needed to deeply understand these socioeconomic and cultural factors surrounding childbirth. This study aimed to explore whether there has been a shift from homebirth to delivery in health facilities and reasons associated with the phenomenon.

Study design
This qualitative ethnography study was carried out in the village of Muara Kaman Ulu and Muara Kaman Ilir, Kutai Kertanegara, i.e. the working area of Muara Kaman Health Center. In these two villages the majority of the population is Kutai tribe who mostly understand and speak Indonesian. This study is part of a qualitative longitudinal study that aimed at exploring patterns and dynamics in decision-making about places and birth attendants. This paper focuses on the shift from homebirth to health facility-based delivery and its enabling and inhibiting factors from women's own perspective. The main author works in a university setting, focusing on maternal and child health, is familiar with Kutai culture and language although not fluent. Data collection was assisted by one female research assistant with a public health background, fluent in using the local language. The research assistant is familiar with the study purpose and methods. Indepth interviews were conducted by the main author, with assistance in writing the transcripts and documentation.

Population and sampling
Pregnant women of Kutai tribe in the first or second trimester were included in the study. A purpose, intensity sampling was applied to select those who have homebirth experience assisted by TBAs, SBAs or had a history of delivery at health facilities.

Data collection method
In-depth interviews were held at least three times, in each trimester of pregnancy until postpartum during the period of April 2015 to April 2016. Twenty pregnant women were initially interviewed, but three women were then excluded due to miscarriage, premature birth and moving out of the village. Both Indonesian and local languages were used. Each interview lasted between 30-90 minutes and conducted at home. Interview questions include history of previous labor, place and birth attendants of previous and current pregnancy during the study, decisionmaking process and various reasons behind women's choice of place and birth attendants. To enrich the information from the women, four midwives were also interviewed to obtain a better understanding and for triangulation purpose to increase data validity. Saturation was achieved after 17 informants, through continuous discussion between the authors.

Data analysis
Data analysis was carried out simultaneously from the beginning of data collection, and transcripts were made available for all interviews. Coding was carried out by the main author, and selected transcripts were coded by the last author as an attempt to improve reliability during coding. A thematic analysis method was applied to identify women's experiences, meanings and realities as well as to come up with the main topics of discourse to produce the themes.

Ethical approval
This study obtained ethical clearance from the Medical and Health Research Ethics Committee, Faculty of Medicine, Public Health and Nursing, Universitas Gadjah Mada, Yogyakarta, Indonesia. At the beginning of the interview, all informants received information regarding purpose of the study. They have agreed to be interviewed more than once in the study and gave permission to record the interview.

RESULTS
A total of 17 pregnant women aged 19-37 years were interviewed, and all of them were Muslim. Three informants worked as teachers at various levels of education located in Muara Kaman sub-district and other informants were housewives. The last education successfully completed by most informants was junior high school. Of the 17 pregnant women, there were 15 multiparous pregnant women and two primigravida pregnant women. One informant had high risk of pregnancy with age over 35 years and two were already pregnant at the age of 19 years.
Thematic analysis highlighted a shifting phenomenon from homebirth to facilitybased delivery, enablers and barriers toward delivery at health facilities. Awareness of homebirth risk, positive attitudes toward facility-based delivery, and women autonomy in decision making were the categories found as enablers for delivery at the health facilities, while perception surrounding birth, preference for homebirth and lack of partnership between the midwife and TBA were considered as barriers.

A shift in choice of place and birth attendant
We observed a slight shift from homebirth assisted by TBAs to delivery at health facilities (Table 1). An increase in the number of women with deliveries at health facilities was observed from five women in the last pregnancy to nine women in the current pregnancy during the study. This was nearly equal to the number of women who had homebirth (eight women). Two women consistently gave birth at health facility outside Muara Kaman village. Three deliveries took place at home assisted by TBAs only. Two of these three women initially planned the delivery at the health facility, but because the delivery took place at night and they were delayed in taking action despite the sign of delivery, they remain assisted by TBAs with the midwife visiting them in the next day. The shift were more prominent in the choice of birth attendant, of which 14 out of 17 deliveries were currently assisted by SBAs, i.e. midwife. and were supported by their spouses and families to make own decisions. . Younger age and primiparous women often preferred to give birth assisted by midwife. They tend to adhere to advices given by the midwife.
"Right now I realize that primipara women are more often to listen to midwife, unlike the older multipara women who still prefer to have homebirth assisted by TBAs" (midwife 2, 27 years).
This shift is certainly supported by easy access to Muara Kaman Health Center, which is located between the two villages, i.e. Muara Kaman Ulu and Ilir. The Health Center was situated in close proximity to the residential area, i.e. only 3-5 km away. Antenatal care was available at the community health post (Posyandu) and there are three community health posts, one in Muara Kaman Ulu village and two in Muara Kaman Ilir village. In addition to these public health facilities, women can also seek care at the midwife private practice. Four health centre midwives had their private practice at home which opened in the afternoon. While all midwife private practice provide antenatal care, family planning and willing to assist homebirth, only one midwife provides birth facility in the private practice.
Another reason to support delivery at health facility was safety. Women who delivered at health facility believed that hospital has complete facilities, sterile equipment, effective labor management and skilled health workers. These made the women feel safe. "Yes I think the hospital has complete [facilities], the facilities are better there" (women 5, 33 years).
Health facilities were also selected because the women were aware that childbirth is also an event that may pose a risk. This reality encourages the mother to fear giving birth at home assisted only by the TBAs.
"I fear if I giving birth with TBAs and if there is an emergency, she can't do anything. But if I deliver at the hospital, they can handle it." (women12, 29 years old) The quotation also shows that women are more confident with the health care services compared to TBAs. In choosing place of delivery, women also consider their health condition during pregnancy and advice from the midwife.
"When I checked my hemoglobin level, the midwife told me if I had anemia I have to deliver at the hospital because I may need blood transfusion" (women 1, 32 years old) In addition to knowledge and awareness, women also seem to have the autonomy in making decision regarding place of delivery and birth attendants. In a patriarchal society, the spouse who is the head of the family play a significant role in making decisions of all aspects in life. Slightly different from that norm, is the decision related to health. The choice of place of delivery and the birth attendants were shown to be given to the women in order to make them feel comfortable, and because it is believed that women know their health condition better. "My husband said it's up to me if I want to deliver at the hospital; the important thing is safety first " (women11, 27 years).
Women also have the freedom to travel or do activities outside the house, and manage financial arrangement for daily life, child education, and health including preparation for costs to give birth.
At the time of data collection, insurance scheme for antenatal care in Posyandu and delivery at government health facilities has been covered using the district government health insurance (Jamkesda) from Kutai Kertanegara.
Jamkesda has been implemented since 2009 and is intended for all residents in the Kutai Kartanegara regency who does not have other health insurances. Nevertheless, only three out of nine women who gave birth in health facilities used Jamkesda, and the remaining six women who delivered at midwife private practice paid with out-ofpocket payment. Although women are aware that delivery at the health center was free, but they preferred to deliver at midwife private practice because they did antenatal care with the midwife.
Even though women have autonomy to make their own decisions, not all women choose to deliver at a health facility. As an illustration, one woman insisted on giving birth at home assisted by TBA, but this decision was not supported by the family who believes that to ensure safety for both the mother and baby during home delivery, it should be assisted by a midwife. This situation suggests that decisions determined by others are not necessarily detrimental to the health of the mother, if the family is aware that every delivery should be assisted by SBA to ensure health of the mother and newborn baby.

Barriers to give birth at health facility
Three common reasons appeared to hinder delivery at health facilities. These are: cultural barriers related to beliefs that delivery at the health facilities is only appropriate for abnormal delivery; perception that planning the delivery will only create complications for the women and the family; and partnership between midwife and TBAs was not optimum yet. Specifically for hospital, long distances and poor condition of road heading to hospitals outside the village created unsafe and uncomfortable situation for the women. The main inhibiting factor is still a strong preference for home birth. The majority of women in Muara Kaman perceived that birth is a natural and normal phenomenon; therefore if the women are in good health, homebirth is the first choice. Women would go to the health facilities only when there is a problem. Even some women still insist on having homebirth when different views on place of birth occurred between women and the spouses. Although the spouse wanted to have the delivery at the health facility, but the final decision was returned to the women. "My husband wishes to take me to deliver at the health facility. He fears something bad could happen if I don't want to but I prefer homebirth" (women 13, 19 years).
Another strong reason for home birth is sense of comfort with the presence and support from the family. In Muara Kaman, there is a habit for the family and close neighborhood to accompany women who will give birth. This practice keeps the women comfortable during labor, by receiving care and having support from the family. In a health centre with limited space in the delivery room, only the spouse or parent could accompany during the delivery process. This makes women reluctant to choose health center as a place of birth, as stated below. "In the health center here, family couldn't enter the delivery room, it is usually only possible to visit during the visitor time" (woman 4, 34 years) Perception on planning a delivery at a health facility is considered to be similar to asking for problems during labor. If it has been planned to give birth in a certain place assisted by a person, then it is believed that the baby will only be born in that place and attended by the person as planned. Therefore, delivery planning was considered to complicate the women themselves, leading to fear to plan the place of delivery and birth attendant. This perception is deeply believed among pregnant women because the information is passed from generation onto the next generation.
"My father told me not to intend to plan the delivery at the hospital; hopefully I have a normal delivery at home. I don't want to have a surgery " (woman13, 19 years old) ".
Although women accept that midwife is the trusted source of information and safe delivery takes place at health facilities, nevertheless they are not able to get out from their comfort zone of having the labor at home. Therefore, homebirth assisted by SBA is more acceptable as it is only considered as modifying (not changing) their local practices.
The presence of informants who made the decision for home birth puts the midwife in an awkward situation. The midwife knew that the delivery should take place at health facilities. However, if the women insisted on giving birth at home, this made them feel obliged to assist the delivery in order to prevent the childbirth from being assisted only by TBAs. Midwives are willing to assist the delivery at home given that women should be willing to be referred to the health center if complications arise.
"When a patient called you to assist homebirth, we first asked them to go to the health centre. If they refused, still we can't ignore them or else they would ask for TBAs to help them. That's why I assisted them under the condition that if problems arise, they should obey me when I referred them to a health center"(midwife 1, 35 years) Woman giving birth at home usually called the TBAs first, and just before labor started, they then call the midwife. So the role of the TBAs was to accompany the mother when the contractions started and the TBA determine when the midwife should be contacted. The role of the TBAs continues up to postpartum care for the women and infants. Overall, service provided by the TBAs is favored by the women.
"Bidan kampong [village midwife, a local term for TBA] is helpful because the midwife would soon leave after childbirth. Bidan kampung still takes care of me and my baby up to three days for baby bathing" (women 10, 22 years) In pregnancy classes, midwives provide information on danger signs that may occur during pregnancy and childbirth. Midwives also encourage women to give birth at Puskesmas. Although women seemed to understand and agree, nevertheless, in reality this is not sufficient to completely change the preference of having homebirth assisted by the TBAs only.
"There is often a pregnancy class for example in the Posyandu, where they had been told about the danger sign, but homebirth with TBAs is part of the tradition, so they didn't consider about it." (midwife 3, 24 years old) There are six TBAs in Muara Kaman Ulu and Ilir villages, with four TBAs who are still actively assisting delivery while the other two are already very senior and thus only occasionally assist delivery among their families. Out of four active TBAs, three of them have been working in partnership with a midwife because they did not want to assist home birth alone without the presence of a midwife. They even encourage the women to give birth at a health facility and are willing to accompany the women until the baby is born. One TBA who refused to work with a midwife only calls the midwife when there is a problem. If the TBAs have agreed to assist the deliver without the presence of a midwife, most people who did not understand would immediately take it for granted.
"Her husband said that he indeed wants to call a midwife, but this was prohibited by the TBA who convince him that she could assist the delivery by herself" (midwife 1, 25 years) Decision for home birth assisted by TBA was sometimes also made by others (spouse or parent), despite women's intention to give birth at health facility. "In the counseling, we [the midwife] asked the women who would be their birth attendant and if they still want to give birth with the TBAs, given that they have already been informed about the complications and danger sign. After counseling, some mother would want to have the delivery with SBA but sometimes it is the family who didn't want it, either the mother-in-law, her own mother or husband, they influenced each other." (midwife 2, 27 years old)

DISCUSSION
This study provides a deeper insight in a rural setting with strong cultural beliefs. We found that there has been a slight shift from homebirth to facility-based delivery in. Although homebirth is still common, many deliveries are already assisted by SBAs. This evidence supports the provincial data of East Kalimantan in 2016 that shows higher percentage of deliveries at health facilities or assisted by SBAs compared to the national figures, 81.1% vs 77.4%, respectively (Widodo, Amanah, Pandjaitan, & Susanto, 2017). Similarly, the finding is also consistent with the global pattern from the Demographic Health Surveys and the Multiple Indicator Cluster Surveys (in 80 low and middle income countries), which shows that in some developing countries there has been a shift in support of deliveries at health facilities rather than homebirth (Johnson, Padmadas, & Matthews, 2013).
One enabling factor for the delivery at health facility is women's autonomy in decision-making. This is in line with a study in Pakistan (Hou & Ma, 2012) that demonstrates Pakistanis women who have the power to make their own decisions is positively correlated with the utilization of maternal health services. (C. Jansen, Codjia, Cometto, Yansané, & Dieleman, 2014) reported a similar finding.
Sufficient knowledge and women's awareness about the risks and complications that may occur during childbirth as well as a growing beliefs that skilled birth attendant such as midwife and a complete health facilities to ensure safety of both the mother and newborn are found to empower women to make the decision of place of birth and birth attendant. A study in Uganda also revealed the influence of risk perceptions of pregnancy and childbirth in the community toward health care utilization (Atekyereza & Mubiru, 2014). Increased women's attention to pregnancy complications and skills of midwife in dealing with such complications will encourage delivery at health facilities (Titaley et al., 2010).
Reluctance to plan the place of delivery and birth attendant was expressed due to the belief that this would be considered as expecting problems in childbirth and only complicates mothers and their families. This fear becomes an important context for this community to emphasize the acceptance and strengthen the practice of midwife-assisted homebirth during childbirth. However, the role of TBAs is still culturally respected by the Kutai community whose existence could not be ignored. Therefore finding the best strategy for the culture and modern medicine to meet in a partnership between midwives and TBAs is the key to improve the collaboration. An alternative new role for TBAs is their involvement in the implementation of Birth Planning Program and Complication Prevention. This program is an activity facilitated by midwife in order to empower the husband, family and community in planning safe delivery and prevention of complications in pregnant women. As TBA is well respected in the community, they could motivate the women and to support facility-based delivery through increased knowledge and practice of delivery planning. Another strategy to engage TBAs is to place TBAs as a communitylevel network for the delivery referral at the health facility (Crissman et al., 2013).
Finally this study showed that there were still families that support homebirth with TBAs only, while studies have shown that in areas of low maternal mortality, 99.2% of mothers had performed safe deliveries at health facilities (Widodo et al., 2017). This condition again highlights the importance of encouraging support from the family and spouse for women to deliver at health facilities. As (Shimazaki, Honda, Dulnuan, Chunanon, & Matsuyama, 2013) stated, safe delivery behavior is influenced by social support for healthy behavior, and high maternal social values for the family. Furthermore, since women who participate in birth planning have less complications during pregnancy and childbirth (Werdiyanthi, Mulyadi, & Karundeng, 2017), health education about the importance of birth planning and complication prevention program targeted to the pregnant women and their families should continuously be strengthened using culturally acceptable strategies.

CONCLUSION
In rural areas with a strong cultural belief, there has been a shift from homebirth to delivery at health facilities. However, practice of homebirth as assisted by TBAs only is still present, indicating improvements are continuously required. The role of TBAs remains important for the women and community, and therefore it is necessary to increase opportunities to engage TBAs in a more culturally acceptable partnership with the SBAs. Better partnerships, would hence accelerate the transition from homebirth to facilitybased delivery. Wicaksana, I.G.A.T., et al. Public Health of Indonesia. 2018 June;4(2):83-90 http://stikbar.org/ycabpublisher/index.php/PHI/index ISSN: 2477-1570 Original Research

INTRODUCTION
Increasing life expectancy in various places in the world leads to increase of number of elderly. The number of elderly around the world nowadays reaches about 500 million people, with an average age of 60 years and predicted in 2025 will increase to 1.2 billion (Swain, Hancock, Hainsworth, & Bowman, 2013). The number of people aged 60 years and late in developing countries is estimated to increase to 20% from 2010 to 2050 (Ponto, et al., 2015). As one of developing country, life expectancy in Indonesia reaches 67.4 years on 2010 and will increase to 71.1 on 2012. This means that there will be a significant increase in the elderly population (Statistik, 2015).
Indonesia ranks fourth in the world's largest population after China, India and Japan (Wulandari & Rahayu, 2011) . The elderly population in Indonesia in 2010 is 9.77% of the total population, and is predicted to increase about 11.34% in 2020 or equal to 28.8 million people (Yuliati & Ririanty, 2014) . In 2010, Bali Province has a fairly population of elderly that is about 300.000 people. Bali Province is ranked 4th out of 5 provinces with the highest number of elderly in Indonesia, which is about 8.77%. It is estimated that by 2020 it will double to more than 432 thousand people or 11.4% of the total population (Statistik, 2015).
In Indonesia, the presence of Panti Werdha is functioned to accommodate neglected elderly people. However, there are still many problems occur for elderly people, especially psychosocial problems. The most prevalent psychosocial problems of elderly people living in Panti Werdha such as loneliness, sadness, anxiety, or anxiety disorders. Anxiety if left untreated may fall into a state of depression and a high risk of suicide (Jayanti, Winarso, & Madyaningrum, 2008). According to the European Study of the Epidemiology of Mental Disorders (ESEMeD), anxiety recognized as one of the most commonly diagnosed psychiatric disorders among the elderly. The prevalence rate of anxiety symptoms ranges from 15% to 52% and 3% to 15% for clinical anxiety disorders in accordance with diagnostic guidelines. As for that problem, we should pay special attention to the presence of an elderly with anxiety disorders, as these symptoms can have a negative impact on their psychosocial condition. An elderly with anxiety symptoms tend to show lower levels of autonomy, loss of visual and hearing ability, mental imbalance, cognitive impairment, physical health problems, poor quality of life and increased risk of death (Riani, 2013) . It is interesting that the incident is even more prevalent in the elderly in institutions (nursing home). Anxiety among elderly people living in orphanages tends to lead to chronic conditions that can significantly lead to mental disorders in the elderly (Fernández-Blázquez, Ávila-Villanueva, López-Pina, Zea-Sevilla, & Frades-Payo, 2015).
As the preliminary study conducted at the Panti Sosial Tresna Werdha (PSTW) Bali on September 28 th until 30 th , 2017 obtained the total number of elderly in the Panti I and Panti II was 127 elderly people. When assessed by interviewing 20 elderly who were randomly selected, all of them experienced anxiety. Most of elderly said that they feel sad and wasted, feel have no choice, and forced to live in PSTW by their relatives. Based on the results of interviews with nurses on duty, about 98% of elderly living in PSTW Bali experiencing anxiety. This is because psychosocial problems in the elderly have not received serious attention. The action given is only by holding a spiritual activity once a week. This activity is considered ineffective because there are still many elderly who have anxiety, even depression and sent to the mental hospital to get further treatment.
In the Management of Care (MOC) standard, nursing care, which performed on anxiety of clients both at general hospital and mental health hospital, is given the generalist psychotherapy. Acceptance and Commitment Therapy (ACT) is advanced psychotherapy that is being developed and can be used to solve psychosocial problems especially anxiety in elderly. Anxiety in elderly people living in PSTW tends to lead to chronic conditions, so ACT is a suitable therapy to be applied. This study aimed to analyze the extent of ACT's influence on the level of anxiety of the elderly living in the PSTW Bali.

Study design
This research was a quasi-experimental study with pretest posttest with control group design.

Sample
The number of samples in this study was 60 elderly people, divided into control and treatment group, with 30 samples in each group. The inclusion criteria of the sample were: elderly with psychosocial anxiety problem, elderly with good cognitive function (people orientation, place, and time), and elderly who were willing to be the research respondent by signing informed consent. Exclusion criteria included elderly who were following other psychotherapy, elderly who must bed rest because of their health status or given total care, and elderly with severe chronic disease. This research was conducted in Panti Sosial Tresna Werdha Bali at Panti I, which located at Wana Seraya Denpasar and Panti II which located at Jara Mara Pati Singaraja. This study was conducted from January until March 2018.

Measurement
To measure the anxiety score in elderly, researchers used Hamilton Anxiety Rating Scale (HARS), which made by Hamilton in 1959 consisting of 14 close-ended questions. Assessment and score of anxiety level as follows: 0= No symptoms, 1= Mild (one symptom of choice), 2= Moderate (half of the symptoms), 3= Weight (more than half of the symptoms present), 4= Very severe (all symptoms present). Range core of anxiety is from 0 to 56.

Data analysis
Univariate analysis was done to know the respondent's characteristic and studied variables. Bivariate analysis was performed using paired and independent T-test to determine the effect of ACT on anxiety. Before testing, firstly data were tested its normality using Kolmogorov Smirnov test because the total number of samples were more than 50. Table 1 shows that, based on age, respondents in the control group were dominant in the age range of 60-74 years old, which was about 18 people (60%), while on treatment group, respondents were in the range of 60-74 years old and > 74 years old as many as 15 people (50%) on each category. Based on gender characteristics, majority of respondents in control and treatment group were female with the number 22 (73.3%) and 23 people (76.7%) respectively. Based on level of education, most of respondents in both group had elementary background level, with 29 people (96.7%) for control group and 26 people (86.7%) for treatment group.    Table 3 shows that the anxiety score during pretest in control group had median score of 40.57, and during posttest was 38.00. In the treatment group the median of the anxiety score during pretest was 41.00 and during posttest was 30.50, with p-value <0.05, which indicated that there was no significant difference in anxiety levels between control group and treatment group. While Table 4 shows that the normality test obtained p 0.072 (>0.05), which indicated that data were normally distributed. Thus, paired and independent T-Test were used for further analysis.  The independent T-test shows that the mean of anxiety in posttest in the control group was 36.30 and 30.83 in treatment group with p = 0.001, which means that the score of anxiety in the treatment group were lower that anxiety score in control group.

DISCUSSION
Based on the age of respondents, it can be concluded that the average age of respondents in this study mostly in the range 60-74 years.
As for equality test result, obtained p = 0.604 (> α = 0.05), which mean there is no significant difference between the age of respondents in the control and treatment group. Based on the sex respondents, it can be concluded that in this study respondents were dominated by female respondents in both groups. As for equality test, resulted with p = 1.000 which means there is no statistically difference between both groups. In general, male and female have the same prevalence to experience psychological disorders. Mental disorders in male mostly lead to violent behavior and personality disorder, whereas in female it tends to be more frequent in affective disorders and anxiety (Yulinda, 2015). Based on the level of education, most of the respondents in this study have the level of elementary school in both groups. The test of equality obtained p = 0.580, which means there is no statistically difference between two groups. Respondents with lower level education will be easier experiencing stress than individuals who have higher education level. It is caused by cognitive ability to perceive the appearance of stressors, lower ability to perceive stressors more susceptible. Individuals with higher education will more easily receive information, easily understand and solve problems (Notoatmodjo, 2012).
Based on the respondents' income, the average income of respondents in this study as much as 128.83 IDR in the control group and 86.33 IDR in the treatment group. Equality test obtained p = 0.54, which means statistically there was no significant difference of income between each group. Elderly income is calculated from the perceived elderly per month, whether obtained from the sale of various kinds of handicrafts they made or from donations from guests who come to visit. Based on the length of stay in PSTW, it can be concluded that the average was 79.77 months in the control group and 43.57 months in the treatment group, with p = 0.063, which means there was no statistically difference between groups. Length of stay is counted by month, starting from the beginning of the elderly into the PSTW until this study was conducted. And lastly, based on the anxiety score of respondents, it can be concluded that the anxiety of respondents in the control group before the intervention (pre-test) was 40.57, and in treatment group was 41.00., with no statistical difference between pre-test anxiety score between groups (p = 0.628).
The results of this study showed that there was a decrease in anxiety score in the control group after given generalist anxiety therapy. Measurement of anxiety level of respondents in the control group was done twice, i.e. pretest and post-test. Based on the result of bivariate analysis using paired T-test, resulted in p = 0.001, it shows that the Ho was rejected and Ha was received, which means there was a significant difference between the level of anxiety of pre-test and post-test respondents. This proves that anxiety in generalist therapy is effective enough to decrease anxiety of elderly who are living in PSTW. Based on researcher's opinion, the change of anxiety score between pre-test and post-test in control group is because of the content of generalist treatment as a pharmacological anxiety management. This therapy consist of four sessions: the first session leads to the improvement of respondent's knowledge about anxiety by giving health education, the second session is the provision of deep breath relaxation techniques, the third session is distraction or diversion technique, and the fourth session is a five-finger hypnosis. Five-finger hypnosis is a therapy that relies on the art of verbal communication that aims to bring the mind wave of respondents to trance (alpha/beta waves), so we can provide a calming effect in the hearts of respondents. The results of other studies also mention the five-finger hypnosis is quite effective for self-programming purposes, eliminating anxiety by involving parasympathetic nervous system so that it can decrease the heartbeat, blood pressure, and breathing frequencies (Evangelista, Widodo, & Widiani, 2016). Beside that, results of this study in accordance with research conducted by (Livana, Keliat, & Putri, 2016) that stated that generalist anxiety therapy effectively reduce anxiety on patients with complaints of physical illness in general hospital in Bogor. The other study has also mentioned that giving generalist anxiety therapy can increase the elderly spirit, causing peace, and tranquility in the liver and reduce tension (Livana et al., 2016).
The results of this study also indicated that there is a decrease in anxiety score in treatment group after given generalist anxiety therapy and ACT. Measurement of anxiety level of respondent in this group also done twice that is pre-test and post-test. Based on the bivariate analysis results using paired T-test, obtained p = 0.001, which means there is a statistically significant difference between the respondents' anxiety score from pre-test and post-test. This can be proven that the treatment of generalist anxiety combined with ACT is effective in reducing the anxiety of the elderly living in PSTW. Based on researcher's opinion, the change of anxiety score between pre-test and post-test is because in the combination effect advanced psychotherapy ACT. In generalist therapy combined with ACT group, respondents were taught to cultivate selfacceptance on disruptive and unpleasant thoughts, and then to place themselves in accordance with the values they held, so that respondents can accept with sincere feeling toward their conditions now. With this acceptance, the respondents is expected to determine what is best for them, and committed to do what they have chosen to live in the future. Furthermore, ACT is a therapy that help individual by using the acceptance as a coping strategy in stress situations both internal and external that are not easy to overcome. Respondents are helped to accept the unwanted events without eliminating them but rather identifying for choose the best action in accordance with the desired goals. ACT does not alter cognitive content to promote behavioral change but focuses on individual behavior and context of occurrence to reduce anxiety. That's mean ACT therapy is very effective to overcome anxiety.
Beside that, ACT has several principles such as Acceptance¸ Cognitive defusion, Self as context, Values, and Committed Action. First one is acceptance, which is the self-acceptance of bad experiences that are not pleasant without trying to change them. The purpose of this acceptance is to increase the individual's willingness to face the thoughts, feelings, and experiences they have been avoided before (Hayes, Strosahl, & Wilson, 2011). Second is cognitive defusion, based on research conducted by (Bach & Moran, 2008) suggests that avoidance behavior from interference and disruptive feelings affects individual coping mechanisms. The attempt to avoid having a bad experience is contradictory. The goal of cognitive defusion is to modify unwanted cognitive function through repetitive exercise and keep on going until the stressor has no meaning for the individual (Yovel, 2009). Third is self as context, which mean for helping individuals see themselves without having to judge or connect with true or false values (Heimberg & Ritter, 2008). ACT helps individuals to become more focus on themselves with the aim is when bad or unpleasant thoughts come to the individual, it will not affect it, so anything that is thought or perceived will not affect the individual. Fourth is a value, which mean to clarify the value that exists in their life and help individuals to take decisions or actions that support the value of life that already exists. Individuals are helped to use existing values to address current problems (Harris, 2006) stated that ACT helps individual to develop larger and more effective measurement related to the selected values through 9 areas of family, marriage, social relationships, career, education, recreation, spiritual, civic, and health. Clarifying these values will help respondents increase their desire to engage in new, and adaptive behaviors. Last is committed action, which is helping individuals to promise an act in accordance with their values. In addition, besides focusing on verbal control, respondents must also commit to do an action or behavior appropriate to verbal control (Hayes, Luoma, Bond, Masuda, & Lillis, 2006) The last results of this study showed that there was a difference of anxiety score after given intervention (post-test score) between control and treatment group. As result shows, after given generalist anxiety therapy the scores in the control group was 36.30, while in the treatment group after given generalist anxiety therapy combined with ACT the score was 30.83. Based on the results of bivariate analysis using independent T-test, with p-value = 0.001 with difference score about 5.47. This value shows that Ho is rejected and Ha is accepted, which means there was a statistically significant difference between the score of post-test anxiety between control group and treatment group. It can be concluded that, the post-test anxiety score in the treatment group was lower than the post-test anxiety score in the control group.
The generalist anxiety therapy combined with ACT therapy is more effective to reduce anxiety than given the generalist anxiety therapy alone in control group especially for elderly people living in PSTW. The researcher suggest that the difference in post-test anxiety scores in the control and treatment groups due to anxiety experienced by the elderly living in PSTW tends to lead to chronic conditions, so generalist anxiety therapy is not sufficient enough to reduce their anxiety score. In other word, generalist anxiety therapy might be more effective if given to elderly with acute anxiety problems. In addition, ACT is more effective because it is an advanced psychotherapy that is effective in building long-term attention, selfacceptance, self-commitment, and more openness in developing the capabilities possessed by the elderly themselves. The results of this study are consistent with several other research findings, such as research conducted by (Fernández-Blázquez et al., 2015) who stated that ACT has been proven effective in overcoming anxiety of clients who suffered a stroke at the Bukittinggi National Stroke Hospital. (Hayes et al., 2011) also suggest that ACT is very effective in receiving attention and more open in developing the capabilities of narcotics abuse clients, chronic pain, PTSD and is very effective as a selftraining model for chronic cases.

CONCLUSION
The results of this study indicate a significant difference between anxiety levels of elderly in control group before and after given an anxiety generalist therapy. There was also significant difference between the elderly's anxiety level in the treatment group before and after given generalist therapy and ACT. This result suggests ACT's is effective for reducing the elderly anxiety that resides in the PSTW of Bali Province. It is expected that nurses may collaborate with a psychiatric nurse specialist to provide advanced ACT psychotherapy on an ongoing basis to an anxious elderly who do not fall into depression and suicidal risk.