Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Master's Programme in Health Informatics Spring Semester 2015 Degree thesis, 30 Credits Vital signs in Swedish emergency departments Qualitative analysis of factors influencing data quality Author: Joana Vicente Author: Joana Vicente Main supervisor: Dr . Niclas Skyttberg, Department of LIME, Karolinska Institutet and Capio St. Göran Hospital Co supervisor: Prof. Dr. Sabine Koch, Department of LIME, Karolinska Institutet Co supervisor: Dr. Rong Chen, Department of LIME and Cambio Healthcare Systems Examiner: Dr. Maria Hägglund , Department of LIME, Karolinska Institutet Master's Programme in Health Informatics Spring Semester 2015 Degree thesis, 30 Credits Affirmation I hereby affirm that this Master thesis was composed by myself, that the work contained herein is my own except where explicitly stated otherwise in the text. This work has not been submitted for any other degree or professional qualification except as specified; nor has it been published. Stockholm, May 2015 Joana Vicente 2 Master's Programme in Health Informatics Spring Semester 2015 Degree thesis, 30 Credits Vital signs in Swedish emergency departments Qualitative analysis of factors influencing data quality Abstract Background: Triage is a commonly used sorting method for emergency department patients and includes several assessments such as vital signs measurement. Patient related information collected during triage, needs to be documented in the patient´s electronic health record in order to be used during and after the emergency care visit. The potential to use and re-use that information depends on its quality. Aim: The aim of this thesis was to explore the factors that affect data quality during triage in Emergency Departments. Objectives: The first objective was to describe the patient flow and documentation processes in Swedish EDs. The second objective was to collect staff´s perspectives in the quality of information that is documented in these departments. Methods: A qualitative study was performed through interviews with fourteen care professionals analyzed by thematic content analysis. ED interview results were analyzed using a framework for definition of data quality dimensions. Results: Two main factors influencing vital sign´s data quality at the ED were revealed: human and technical factors. All coded factors that influenced data quality at the ED could be related to completeness, correctness and currency dimensions for data quality. Analysis of patient flow and the documentation processes at the ED showed that clinical measurement of vital signs leads to a simultaneous need for documentation routines. The care providers identified several potential data quality influencing factors and pointed towards a symbiosis between their documentation needs and the available technical solutions as a possible way to improve vital signs data quality at the ED. Conclusion: The main contribution of this thesis is the identification of human and technical factors that impact completeness, correctness and currency of vital sign data in the emergency department. For electronic systems to work reliable on this data the introduction of standardized documentation is essential. Keywords: Data quality; Triage; Vital signs; Emergency department. 3 Acknowledgements To my Supervisors Thank you to all my supervisors, for supporting me through this work by providing encouragement, knowledge, experience and support. Dr. Niclas Skyttberg (my main supervisor) thank you for making this study possible by providing clinical insights, engagement, feedback and support during all times. Prof. Dr. Sabine Koch, thank you for the academic guidance and Health Informatics knowledge as well as the patience and support during this work. Dr. Rong Chen, very special thanks for the technical Health Informatics insights, experience, feedback and knowledge. To all involved in this work I want to thank all the study participants for giving me the chance to create this work and taking the time to participate in the interviews. To Karolinska Institutet Thank you to all the 2015 Health Informatics teachers at Karolinska Institutet, for two amazing academic years and for showing me that ”O saber não ocupa lugar” [Knowledge doesn´t take place]. A special thanks to the 2015 Health Informatics class students for making me rich in friendships, experiences and languages! A special thanks for both Blanca and Julie for the muffin therapy sessions. To my family I want to thank my parents, with all my love, for always believing in me and giving me the strength to move forward. Obrigado Mumy e Papi. A loving thanks also to Tobias for, unconditionally, supporting me through these two years. Thank you. In the memory of my grandparents, Maria Vicência and Joaquim Quitério, because “Relembrar é viver” [To remember is to live]. Finally, I would also like to thank Be (my dog), for ensuring the stress free moments and inspiration. 4 Table of Contents List of abbreviations ……………………….………………………………...…….…....7 List of figures…………………………………………………….………………...........7 List of tables…………………………………………………….……………….…........7 1. Introduction………………………………………………………..................................8 1.1. The Swedish emergency care…………………………………………………..............8 1.1.1. Triage and flow ………………………………………………………............8 1.1.2. Documentation…………………………………………………………..........9 1.2. Data quality………………………………………………………………...……………9 1.2.1. Definition and Impact of data quality……………………………………........9 1.2.2. The Sepsis example………………………………………………….............10 1.3. Problem statement………………………………………………………....….............11 1.4. Aim and objectives……………………………………………………...….…............11 1.5. Research question.……………………………………………………..…….………...11 2. Methods…………………………………………………….……………...……….......12 2.1. Research Methodology and Approach………………………………...……...............12 2.2. Data collection tools…………………………………………………..……................13 2.2.1. Document analysis……………………………………………..……….........13 2.2.2. Study Settings…………………………………...………..............................13 2.2.3. Recruitment of participants……………………..…......................................14 2.2.3.1. Sampling…………………………………….…………...........16 2.2.4. Interviews……………………………………………………..……………...16 2.2.4.1. Pilot interview…………………………………..…………......16 2.2.4.2. Main Interviews……………………………......………...........17 2.2.4.3. Continuous development and saturation……..………..………17 2.3. Data analysis……………………………………………….………….........................18 2.3.1. Deductive Approach……………………………….…………………...........18 2.4. Ethical Considerations………………………………….……………….....................19 3. Results…………………………………………………….…………………………….19 3.1. Process Results…………………………………….…………………………………..19 3.1.1. General emergency care patient flow and documentation processes…..……19 3.1.2. An insight into three different ED patient flow and documentation processes……………………………………………………………….…..…24 3.1.2.1. ED paper based documentation process………….....................24 3.1.2.2. ED digital documentation process………………………...…....25 3.1.2.3. ED paper and digital documentation process……....................26 3.2. Conceptual analysis from the interviews………………………………………..,,,,,,,,..27 3.2.1. Coding and main concepts…………………………………………...….........27 3.2.2. Data analysis and the adopted data quality framework.………………………35 4. Discussion……………………………………………………….……………..……......37 4.1 Patient flow, documentation processes and staffs insights on data quality in the Swedish ED …………………………………………………………………………38 4.1.1 Standardization of Processes……………………………………………….....39 4.1.2 Interoperability of IT systems, Education and Competence……………….....41 4.1.3 Management…………………………………………………………………..42 4.1.4 User need for System Support………………………………………………...43 4.2 Methodology considerations…………………………………………..……..…......45 4.2.1 Limitations and strengths of the study…………………..………………….45 4.2.2 Trustworthiness, credibility and dependability…...……………………… 46 4.3 Implications of this study: Recalling the sepsis example.……………………........46 4.4 Future research…………………………………...………………………………...47 5. Conclusions………………………………………………....…...................................48 References………………………………………..……………...................................49 Appendices………………………………..…………………………………………….52 Appendix A –Main interview guide (in Swedish) ……..……..………….....52 Appendix B – Patient flow and documentation processes for settings A;C;D.……………………………..……..……………….......53 Appendix C- Mind map of data analysis code results...………….………….56 6 Abbreviations CDSs- Clinical decision support systems ED- Emergency Departments EHRs- electronic health records MD- medical doctor RN- Registered Nurse VS- Vital signs List of figures Figure 1- The research Stages. Figure 2- “Zigzag Data Collection and Analysis to Achieve Saturation of Categories”, by Creswell, 2008 (38). Figure 3- The patient flow process. Figure 4- The re-evaluation at the ED. Figure 5- Patient flow and documentation processes at an ED with paper based documentation. Figure 6- Patient flow and documentation processes at an ED with digital documentation process. Figure 7- Patient flow and documentation processes at an ED with paper and digital documentation process. Figure 8- Diagram of codification results after interviews analysis. Figure 9- Occurrences of the two major themes. Figure 10- Occurrences of the three main categories of the human factors theme. Figure 11- Occurrences of the categories and sub-categories created of the human factors theme. Figure 12- Occurrences of the two categories created and included in the technical theme. Figure 13- Occurrences of the categories and sub-categories coded for the technical theme. Figure 14- The relation between the 3 dimensions for data quality and the code generated from the data analysis of the study interviews. Figure 15- Most influencing factors of the three data quality dimensions. Figure 16- The Emergency Department cycle. List of tables Table 1- Description of the study settings. Table 2- An overview of the study participants. 7 1. Introduction 1.1 The Swedish Emergency Care Emergency care is a central piece within healthcare and includes all treatment provided in declared medical emergency situations(1). Sweden’s national government is responsible for overall healthcare processes and policies(2) and each one of the Swedish County Councils is responsible for ensuring universal access to good healthcare to the citizens(1)(3). Over the last few years emergency care in Sweden went through intense reforms and restructuring that allowed the healthcare sector to offer improved quality and results while serving an expanding, ageing patient population (4). Improvement policies in Sweden focused on consolidation, differentiation and specialization, and that is why nowadays, emergency care is no longer exclusively associated with hospital settings but can also be provided by different settings such as open-clinics (1)(4). According to an official report 2,4 million visits to Swedish hospital EDs occurred only during 2013(1). It is therefore important, to have strong organizational and structural policies in order to ensure that care is provided with the highest quality. In emergency care one of the ways of sorting or selecting the patients is by using triage, a process described below. 1.1.1 Triage and Patient Flow Triage is a dynamic decision-making process that prioritizes a persons need for medical care prior to or on arrival at an ED (5). Triage has been developed since the 1990´s and is nowadays broadly used in several different countries(6). Triage includes the analysis of emergency symptoms and signs with special attention to detect risk factors related to objective physiological vital signs (VS), vital history and risk factors (7). Triage is often carried out by registered nurses (RN) that use a guiding triage scale to help in the decision of allocating a patient to an acuity level (8). However in Sweden, the degree of medical urgency can also be assessed by a care team involving different staff categories such as physicians, nurses, nurse assistants and secretaries (9). Triage represents only a part of the patient´s emergency care and can be seen as the primary step into the ED. Triage scales can be used partially to perform an “outer triage” upon patient´s registration at the ED that is followed by a later “full triage” or physician´s assessment (10). In Sweden three triage methods are commonly used, the Medical Emergency Triage and Treatments System (METTS), Adaptive Process Triage (ADAPT) and the Manchester Triage Scale (MTS)(9). According to a 2010 report, referring to 2009 numbers, one third of the Swedish hospitals uses METTS, one third uses ADAPT, and one third uses MTS or locally developed triage scales(9). The same report shows that scientific evidence is insufficient to appraise the validity of the different triage scales when regarding their capacity to accurately predict a clinical event that requires the immediate physician´s attention (8). Both ADAPT and METTS are process-oriented triage scales and the development of these scales has preceded the development of patient flows in the Swedish ED(8). The work of processing the patients in the ED, with the aim of speeding up patient throughput is called flow processes. Different flow processes are in use in Sweden today, depending on the hospital organization and management. Flow processes are used in the ED to organise routines and contribute to reduce waiting times and overall length of stay(9). They are an efficient strategy to reduce variability in care and ensure that patients receive standardized care that has been shown to improve the outcomes (11). An example of a flow process within 8 ED is the “fast-track”, where a special coherent process is used in order to manage patients with minor disorders or injuries (9), allowing patients to “flow” within the ED in a smooth pre-defined and efficient way. 1.1.2 Documentation During ED triage information about the patient´s clinical status, such as vital signs, is collected and measured and needs to be introduced in to the patient´s EHR, in order to be used, documented and stored. At the ED, there is a need for secure and accurate transfer of information between different caregivers (12), which means that documentation in patient records is a priority. Since ineffective transfer of information and communication between healthcare professionals has been reported to be a vulnerability of the triage process(13), it is crucial to ensure that a correct and complete documentation of the triage moment, and specially vital signs assessment is performed. The patient record is the main source of information regarding patient´s healthcare history. Swedish regulations state that the procedure for documentation of patient data must ensure that the patient record can provide a ground for monitoring health care and it´s quality (7). It is known that, by working in adherence to the implemented triage system, it is more likely to achieve a higher level of correctness in documented triage level, which will also improve patient safety and even increase cost effectiveness (7). Patient safety is therefore strictly connected to triage compliance and good documentation routines (14). Documentation and the use of common triage systems and terminology is essential throughout the healthcare chain (7). Adherence to triage systems by measuring all vital signs is important to be able to document the correct triage level of a patient (15). Neglect to comply with triage documentation can result in a negative outcome for the patient´s health (16). In Swedish EDs triage documentation is performed by using EHRs and/or paper based forms. EHRs, in particular, have potential to transform healthcare systems by allowing clinical data to be transferred, stored, analysed and exchanged in an efficient way(17). But EHRs are highly dependent on the quality of the documented data to have that transformation potential. It is therefore important to gather more knowledge in the process of collecting and documenting data within the ED and the factors that might influence those processes. Both clinical and documentation processes within the ED have therefore to be understood and studied. 1.2 Data Quality 1.2.1 Definition and Impact of data quality Quality of data can be evaluated by using both qualitative and quantitative evaluations(18), and it can be used as a quality of care measurement. In Sweden both official inspections and filed complaints within healthcare, often show poor or insufficient documentation within several healthcare departments (19). Inaccurate data may lead clinicians to make treatment errors(20), researchers to underestimate disease prevalence(21), health-system managers to underestimate compliance with care standards and alerting systems to send false alarms to physicians (22). Clinical data, or information recorded in the patient health records, needs to have quality in order to be able to be transferred, communicated and reused in a trustful way. Reuse of information recorded in the patient’s health record, is important not only while patient´s receive care at the ED, but also in a later stage during other healthcare related contacts. Also, the retrospective use of health data holds the promise to expedite scientific 9 discovery (23), but currently secondary use of clinical data is still at an early stage (24). Despite the well-known benefits of EHRs, reuse of data has been limited by several factors including concerns about the quality of the documented data (18). Quality has been defined through the “fitness for use”(25) which, in the context of data quality, means that data is of sufficient quality when it serves the needs of a given user for a specific goal(18). In the ED setting it would mean that information or data recorded in the patient records serves the purposes needed, such as making clinical decisions, treatment planning and execution (14). Few studies reflect on how quality of data can be achieved, but in the nursing field, several studies provide recommendations that facilitate nurses documentation strategies(26). However, the impact of those changes and recommendations in the quality of data is not known. The framework for data quality analysis used during this study was the one proposed by Weiskopf and Weng in 2013 (23). The authors aimed for the reuse of the EHRs data for research and proposed five dimensions of data quality: completeness, correctness, concordance, plausibility and currency. These five dimensions were defined by the authors in the following way (23): Completeness- referred to whether or not a truth about a patient was present in the EHR; this dimension aims to ensure that all information collected from the patient is present in the EHR; Correctness- data in the EHRs was considered correct when the information they contained was true; Concordance- concordance exists when there is agreement or compatibility between data elements; Plausibility- data was considered plausible if they were in agreement with general medical knowledge or information and were therefore feasible; Currency- It refers to the timeliness or recency, in the EHRs data is considered current if recorded within a reasonable period of time. In order to better understand the importance of correct documentation and EHRs data quality upon triage at the ED, an example is provided below. 1.2.2 The Sepsis example Sepsis is an infection based systemic inflammatory response syndrome that is clinically classified in three stages: sepsis (primary symptoms of the disease), severe sepsis (associated with organ failure hypotension or hypo perfusion) and sepsis shock (the most severe clinical stage of sepsis characterized by hypotension that persists despite adequate intravenous fluids) (27). Sepsis can be either a primary or hospital acquired infection, and it was pointed out as one of the major avoidable hospital related health complications in Sweden in 2013 (28). Due to the quick progression of the disease and in order to increase the survival rate for Sepsis patient´s, it is extremely important to perform an early diagnose where early inspection of vital signs is not only a quality factor for the measurement of healthcare quality in Sweden, but also a necessary step upon triage (27)(29)(30). Even if documentation of vital signs is one of the most commonly performed tasks in EDs (30), statistics show that in the year of 2011 29% of the Sepsis diagnosed patients missed any register of a specific vital sign, breathing rate, in their medical records (27)(31) a trend that continued through 2013 (32). Several other studies suggest that vital signs documentation is 10 commonly incomplete (33)(34). In the particular case of sepsis, the fact that vital signs information either is missing, incomplete, or not documented in the patient record, means that vital signs data in the patients EHRs is missing quality. This fact can have a big impact in the medical outcome and assessment of this particular group of patients. Since this disease progresses in a very quick way, it is fundamental to detect vital signs alterations on time in order to improve the Sepsis patient´s clinical outcome. This example raises questions on both the documentation routines or processes and the quality of the documented patient data, specifically vital signs, upon triage in the Swedish EDs. 1.3 Problem Statement Swedish EDs follow different triage protocols in order to sort, prioritize and select patients in to a specific patient flow process aimed to improve healthcare delivery in both quality and time. Documentation is performed throughout this process but, even if the different triage processes are well structured and defined, the way patient related information is collected, communicated and recorded in the patient´s record is pointed out as a feasible reason for many of the healthcare reported errors and complaints filled to the Swedish National Board of Health and Welfare (19). Furthermore, the factors affecting vital signs data quality during the triage process and the quality of the EHRs data per se are unknown. As stated before, good quality documentation of triage results and contents, such as vital signs, is essential for both patient safety and quality of care. There is therefore an interest in knowing which factors can affect the quality of data recorded in patient records upon triage. From a health informatics point of view, data quality affecting factors, have to be known so improvements can be made in the quality of data, before it can be reused in clinical decision support systems (CDS), research or quality management within healthcare. Data reuse within healthcare may then be improved if data is collected and documented with quality. 1.4 Aim and Objectives The aim of this study is to explore the factors that affect data quality during triage in EDs. The first objective of this study is to describe documentation processes and patient flow in the Swedish ED. This will evaluate if there is a relation between documentation processes and data quality and increase the knowledge about the factors that may affect data quality in the Swedish EDs. The second objective is to collect ED staffs perspectives regarding the quality of data documented in the ED. A description of barriers and facilitators in documentation processes that affect data quality is aimed as well as the link between barriers/opportunities and data quality. 1.5 Research Question This study aims to answer the following question: Which factors impact data quality in the Swedish Emergency departments? 11 2. Methods 2.1 Research Methodologies and Approach In order to achieve the aim and objectives of this study and answer the proposed research question, a qualitative study was performed. The choice of performing a qualitative study is due to fact that, even if data quality can be quantified, the factors that have impact on the quality of data during the documentation process at the ED, are hard to quantify. These factors are linked to organizations and individuals that will be evaluated in perceptions and experiences, not quantifiable in a quantitative way. During the first phase of this study meetings were held with Dr. Niclas Skyttberg, a clinician and expert in triage and medical management as well as Prof. Sabine Koch and Dr. Rong Chen, experts in Health Informatics. During these meetings in depth knowledge of the triage protocols, documentation and variations in Sweden was provided and gathered. Parallel to this introduction phase an extensive literature research was performed in both triage and data quality. In order to be able to study the impact of different factors in data quality within the Swedish EDs, a focus was set in the vital signs collection and documentation as part of the triage protocols. By focusing on the vital signs data collected, documented and recorded in Swedish EDs, a narrow specific section of the triage process was selected in order to be analysed and lead to the aimed answers. As mentioned before, Weiskopf and Weng in 2013, proposed five dimensions for EHRdata quality assessment: completeness, correctness, concordance plausibility and currency (18). In order to be able to analyse the quality of data in the documentation of vital signs within the triage process in Swedish EDs, and for the scope of this study, three data quality dimensions were chosen: correctness, completeness and currency. Both plausibility and concordance were considered a part of the correctness dimension and therefore not specified for this study. These three data quality dimensions were used in the second phase of this study, where interviews took place, for both the development of the interviews questions and the analysis of the interviews data. Interviews were held with ED staff members of bigger hospitals in the Svealand and Götaland region in Sweden. A pilot interview guide was developed from the three selected dimensions for data quality that were the base for a range of in depth questions. This guide was tested in one interview and led to the final interview guide that was organised in to three sections, each one focused in one of the three mentioned data quality dimensions. That is, the final interview guide questions were built to gather information on how each one of the data quality dimensions is thought to be affected at the ED. Interviews were performed in a structured way in several different settings and the collected data was analysed by constant comparative analysis as described by Strauss and Corbin 2008 (35). During the interviews data analysis the chosen data quality dimensions, were once more used in order to compare the coding results of the interviews with the framework for data quality. Data analysis was performed by using the deductive content analysis method (23) in which the content of the data analysis is used to gain a logical support to the findings and confirm eventual previous knowledge (36). Inductive data analysis methods were also used in order to claim that some of the findings were probable or likely to be found in the ED setting. Data gathered from the interviews was also used during the third and final study phase, where the description of specific patient flow and documentation processes, for each one of the studied hospitals settings and interview data analysis, occurred. Data from the interviews was 12 enriched by the analysis of specific EDs documentation such as paper based triage forms. A mapping of the patient flow and documentation processes for different settings was created in order to analyse and compare documentation strategies between them. The VUE(42) tool was used to draw the patient flows and documentation processes for each one of the clinical settings. These patient flow and documentation processes were cross checked with official reports gathered during literature research(10,37). A general patient flow and documentation processes were then extrapolated, representing the general processes that occur in Swedish EDs. This generalisation of processes allowed a better and graphical analysis of the factors that influence the quality of data in the ED. An overview of the research stages of this thesis is provided below. Figure 1. The research Stages. 2.2 Data collection tools 2.2.1 Document analysis Both official triage paper forms and reports from the emergency care unit from the different hospitals were used in order to collect information regarding the patient’s flow through the ED and the staff’s documentation process. These documents included official triage and ED patient forms that allowed the researcher to collect information on how vital signs data are measured, documented and used during before and after triage. An official report (10) on some of the EDs from the Svealand region, allowed the verification of the patient and documentation processes designed for those settings. Findings from this document analysis will be presented later on with the processes description during the results section. 2.2.2 Study Settings In order to collect information onsite, several emergency care departments of major hospitals in Sweden were selected. The clinical advisor of this study, Dr. Niclas Skyttberg, was the 13 main source for these hospital contacts. The settings were selected in first place by their geographic location around Stockholm, in the Svealand region, in order to allow face-to-face interviews. However, some hospitals settings were located in the Götaland region, requiring interviews to be performed by telephone. This different location allowed the study to not only compare results between regions but also discuss if factors that influence the quality of data in the EDs, vary or not according to the setting or geographic location. A total of eight different EDs from different Swedish hospitals were selected to participate in the study. A brief description of each one of the settings is provided below. Table 1. Description of the study settings. n=8. Setting Location Patient intake 80 000 patients per year, Svealand Hospital A excluding the gynaecological Region emergency patients Svealand 73 000 patients per year also Hospital B Region excluding gynaecological patients Svealand 75 000 per year within 8 different Hospital C Region medical specialities Description of the ED It is a general ED with specialisation in intern medicine, cardiology, surgery and orthopaedics (10) It is a general ED with 9 different medical specialities as well as a regional specialised trauma centre (10) Hospital D Svealand Region 73 000 patients per year The hospital is specialised in intern medicine, surgery and orthopaedics. (10) Hospital E Götaland Region Together with another major hospital this setting treats around 400 patients per day (40) Standard ED 42 000 patients per year Standard ED (10) 34 000 patient per year only in the emergency department One of the 20 main regional hospitals in Sweden. (41) Provides specific emergency departments for children, infectious patients, gynaecology and psychiatry (44). Hospital F Hospital G Götaland Region Götaland Region Hospital H Svealand Region 53 000 patients per year Hospital I Svealand Region During 2013 it received and treated 42 702 at the ED(1). Standard ED. 2.2.3 Recruitment of participants Because this was a qualitative study, participants were chosen by “purposeful sampling”, that is, since the intent was not to generalize to a population but to develop an in-depth exploration of a central phenomenon (38), in this case factors influencing vital signs data quality in the Swedish EDs, we purposefully selected individuals and sites. The standard used to select participants was whether they are “information rich” (42) or not. Participants were then recruited from all the presented hospital settings. Hospitals settings were approached via email sent to either the chief nurses department or physicians working on site. The email explained the purpose of the study and asked for participants that had experience in ED triage, triage education, quality control or administration. Since triage in Sweden is mainly performed by registered nurses (14), we asked for the participants to be RN but two of the interviewees were MD. It was important to select participants that had experience of working with triage at the ED, since their clinical experience and knowledge was considered a main knowledge source for triage and vital signs measurement and documentation processes. RN with experience in triage education were a valuable source in order to evaluate and obtain an in-depth sight of the 14 way triage methods are taught in the ED. Participants with leading roles in healthcare quality evaluation and administration, with combined clinical work, were also considered since their knowledge in healthcare quality factors and ED performance provided an insight into the actual status of the quality of data registered and collected at the ED. A total of sixteen professionals were interviewed, two medical doctors and fourteen registered nurses. An overview of the interviewee’s profession, clinical role at the ED and experience is presented in Table 2 below. Table 2. An overview of the study participants. Their role at the ED, clinical experience in years, experience in the ED in years and clinical workload. NA- information not available. n=16. Interview Number Setting Number Interviewees Profession Clinical Experience in years ED experien ce in years % of clinical work Medical Doctor (MD) NA NA 50 9 9 60 NA NA NA 1 Hospital E 1 2 Hospital D 1 3 Hospital F 1 4 Hospital C 1 RN, Healthcare quality responsible More than 10 10 50 5 Hospital D 1 RN More than 18 18 100 2 RN and responsible for triage education at ED More than 17 17 50 RN at children´s ED 9 9 100 RN specialist in emergency care 4,5 1 100 RN and ED education leader 24 18 0 20 11 50 27 27 50 6 7 8 Hospital C Hospital B Hospital A 1 2 Chief RN at ED and ED healthcare developer MD 9 Hospital G 1 RN responsible for quality of care development ED chief RN 10 Hospital H 1 ED chief RN 27 21 11 Hospital G 1 RN More than 15 15 100 12 Hospital I 1 RN More than 13 13 100 13 Hospital G 1 12 26 50 14 Hospital H 1 10 16 50 RN, worked before as an ED administrative RN 15 All participants were contacted by email and either a face-to-face or a telephone interview was scheduled accordingly. The pilot interview was a “test-up” interview that allowed the testing of the interview guide for the aim of obtaining the pretended information. The pilot interview was performed face-to-face with one RN participant and lasted around 40 minutes. The pilot interview guide was then restructured and used in all the remaining 13 interviews with a new improved guide, structure and set of questions based in the data quality framework dimensions considered for this study. The main interviews lasted no more than 30 minutes and were performed either face-to-face or by telephone. 2.2.3.1 Sampling Sampling can be defined as a process or method to select a part of the population, or sample, in order to determine characteristics of the entire population (38). There are different types of sampling and two were used in this particular study: theory or concept sampling and homogenous sampling. Theory or concept sampling was used in order to help the researcher sample individuals because they can help the researcher generate or discover a theory or concepts within a theory (42), in this case factors affecting the quality of data at the Swedish ED. Homogenous sampling was also used since the researcher purposefully sampled individuals based in a subgroup with specific characteristics (42), that is all the participants needed to work within ED, have experience in triage methodology and even some other responsibility connected either to ED education or healthcare quality administration. 2.2.4 Interviews Interviews were the main data collection tool selected for this study. Semi-structured interviews are pointed as one of the best methods to collect and try to understand information from other human beings (38). This type of interviews allows researchers to address issues of trust and organization, while maintaining a feeling of openness (39). Interview open-ended questions were developed in a pilot interview guide that was tested during one pilot interview. This guide was then developed to a new range of questions that were used for both face-to-face and telephone interviews. Each interview was performed in Swedish, audiotaped, with the consent of the interviewee, and transcribed verbatim in order to facilitate the data analysis. Some of the interviews were performed with the great support of Dr. Niclas Skyttberg, and only I conducted others. 2.2.4.1 Pilot interview A preliminary interview guide was developed, using the three dimensions for data quality: correctness, completeness and currency. This guide aimed to collect information about the triage process, vital signs measurement and documentation and the interviewee own perception on triage, the documentation strategy in place and the quality of vital signs data collected and registered on ED patient records. This preliminary interview guide was tested in a pilot-interview that allowed the testing, restructuration and narrowing of the interview questions. This step was fundamental in order to both reduce the amount of questions and narrow the collected information by using more specific questions. The pilot interview was performed with a RN in a face-to-face interview. This interview was performed in 40 minutes and gave an overview and broad insight on both triage and vital 16 signs measurement and documentation processes. This interview was performed in Swedish, audiotaped and later transcribed in verbatim. 2.2.4.2 Main Interviews A new interview guide with a new set of questions was developed from the experience gathered with the pilot interview. This new interview guide (available in Appendix 1), was also divided in main sections: ED vital signs measurement process, documentation process and paper based documentation. Questions related to the ED VS measurement process aimed to have the interviewee describe the triage process in use at that specific hospital, what can determine if a triage process is complete or incomplete and the stakeholders opinion on if patients that are not submitted to triage are healthier than those that are. Questions related to ED documentation processes pretended to collect insights and information related to the way both triage and vital signs are documented in that particular setting. Specific questions were introduced in this section of the interview guide regarding the three foci of this study: correctness, completeness, currency, as well as the factors that the interviewee though could influence these three data quality dimensions. Finally, a question was introduced in the end of the interview guide to collect information and insights on the eventual use of paper-based documentation. A total of 13 interviews were performed both in face-to-face and telephone and lasted no more than 30 minutes. All study settings were represented with at least one interview. All interviews were performed in Swedish, audiotaped and later transcribed in verbatim in order to facilitate the on-going data analysis. 2.2.4.3 Continuous development and Saturation While the interviews were being held, data analysis of the interviews was being performed in parallel using an emerging design. An emerging design is a process in which data is collected and analysed immediately, rather than waiting until all data collection is achieved (38). Continuous analysis, development and improvement of the collected data were performed from interview to interview, which allowed the researcher to progress to data saturation. Saturation was achieved when no major themes and no new information could be added to the data analysis results (38) from the already performed and analysed interviews. By achieving saturation with the interviews data, the researcher made the subjective determination that new data, collected with more interviews, would not provide any new information or insight to the developing categories (38). Figure 2 illustrates the zigzag data collection and analysis process proposed by Creswell, 2008 to achieve saturation of categories that exemplified the parallel data collection and analysis performed during this study. 17 Figure 2- “Zigzag Data Collection and Analysis to Achieve Saturation of Categories”, by Creswell, 2008 (38) 2.3 Data analysis 2.3.1 Deductive approach A common, widely used method for qualitative data analysis (45) results from a systematic, qualitative study procedure that is used to obtain a theory. That theory explains, in a broad concept level, a process, action or interaction about a specific topic (38). In this particular study, by using this approach it is possible to start the study without predefined theories and to build theories and deep explanations and descriptions of the factors affecting the quality of data today in Swedish EDs. Since a broader explanation of the impact of the documentation processes in the quality of data is aimed, and few studies evaluating the factors that affect quality of data in ED exist, the use of this method for qualitative data analysis is the optimal tool to analyse the data collected during the study. As mentioned before, rather than beginning by researching and developing hypothesis the researcher´s starting point was the data collection through the use of interviews. A comparative data analysis was performed in order to move back and forth among the data while it was being collected, as suggested by Strauss and Corbin 2008 (35). The aim of the study was to identify factors affecting data quality, so a deductive content analysis was also used, in which the structure of data analysis is operationalized based on previous knowledge and the purpose of the study is theory testing (36). A deductive approach is based in an earlier theory or model and therefore moves from general to specific (27). In this particular study, as mentioned before, the framework for data analysis is the one proposed by Weiskopf and Weng in 2013 (23). The researcher departed from this framework and after making sense of the collected data conducted a data analysis using a deductive approach. The theoretical framework for data quality was used in this study (23) throughout the methods and the results section and worked as a constant resource for development, collection and interpretation of the study results. The interviews were analyzed using thematic content analysis to collect staffs perspectives on data quality at the ED of the different study settings. The interviews, which were conducted with registered nurses with different functions within the ED, and two medical doctors, allowed the researcher to collect information on how ED staff thinks quality of vital signs data recorded in the patient´s record is today and why. Perspectives and insights were collected on how completeness correctness and currency of vital signs data might be affected at the ED today and why. After transcribing the interviews in detail a codification process was performed. Codification 18 was made at the same time that the interviews took place and both themes and concepts were constantly reviewed, generated and checked. Since the interviews were performed in Swedish, codification and creation of the different categories was performed in English language in order to allow a better presentation of the thesis findings. The process used to perform the codification of the interviews text was based in a deductive method. Information contained in main or key interview text sentences was classified and deducted in to themes, categories, sub-categories and concepts. The study participant´s insights were classified from themes to concepts based on the research question (which factors impact data quality in the Swedish EDs) and the adopted framework for classification of data quality. 2.4 Ethical Considerations Ethical approval was applied for at the Stockholm representation of the Swedish Central Ethical Review Board (Regionala Etikprövningsnämnden i Stockholm) which decided that an ethical approval was not necessary to perform the study. All participants were fully informed about the goals, aims and objectives of this study. They were also informed about the anonymity of the collected data in order to ensure protection of their integrity. Interviewees voluntarily participated in the study and gave verbal consent prior to the audiotape of the interviews. 3. Results This section will present the results and finding of this study. Results are presented in two sections: the patient flow and documentation processes results, named process results, and the conceptual analysis results from the interviews. 3.1. Process Results The aim of this study was to explore the factors affecting data quality by achieving the following objectives: describe documentation processes and patient flow in the Swedish ED as well as to collect ED staffs perspectives on the quality of data documented in the ED. In order to achieve the first objective information was gathered from both interviews and local documentation on triage forms and ED routines documents. Information collected from these sources was then compared to official reports on both triage and ED organization (10)(8)(1). This information led to the creation of two different types of processes: the ED patient flow and the ED documentation. The general patient flow and documentation processes for the ED are presented below. Study results provided a deeper knowledge on three different ways of supporting documentation at the Swedish EDs today: digital, paper or both. 3.1.1 General emergency care patient flow and documentation processes Both interviews and local documentation, such as triage and ED registration forms, provided a good understanding about the way patients visit to the ED is organized and how documentation is performed on the way. All ED settings considered for this study had a similar organization and structure. However, differences were visible not only on how triage was performed, but also how vital signs information was measured and documented. 19 Patient flow The way patients move through the ED ward differed from setting to setting according to local routines and organizational strategies. The way a patients visit is organized and structured within the ED can be referred to as a patient flow process. This flow is deeply connected to the way documentation of the patients visit at the ED is performed by staff. That is, while the patient moves through the ED, information is collected and needs to be documented. One can think of two parallel processes that simultaneously take place at the ED, the patient flow and the documentation processes. Interviews showed that the way of performing ED documentation is extremely variable from site to site, but in the end the aim is to document and make information collected and generated during an ED episode, accessible. In order to describe the patient flow and documentation process at the ED, a general description of both processes is provided below The patient flow was generalized to the one presented in figure 3. The main steps in the patient flow process are arrival, registration, triage, monitoring and treatment, decisions and discharge. The figure provides an overview of this process together with the main clinical steps and documentation moments that happen at the same time while the patient flow process took place. Figure 3. The patient flow process. Figure 3 represents the main clinical and documentation steps at the ED. The vertical oriented process on the left illustrates the patient flow process through ED from start to end. The two horizontally oriented processes represent the clinical steps that occur during the patient´s process and the main documentation support used at the ED. 20 Arrival and registration Patient´s arrival to the ED was marked by one of three methods: registration in a specific desk (together with a primary sorting); together in combination with triage or by the use of a queuing system without sorting. All hospitals performed patient registration and this was the first step for all patients arriving at the ED on their own, without ambulance. Registration performed through desk was described in some hospital settings as a way of sorting the patient inflow to the correct ward. This was particular important since those hospitals had a minor ED, or “lättakuten”, responsible for the care of considered minor emergency situations. This primary sorting was based on a pre-triage. The desks were used to classify patients and forward them to the correct emergency ward in avoiding queuing and long waiting times at the “main” ED. Desks were available during the day but their use was limited during night time. Other settings performed patient registration upon triage that is the patient was admitted to the triage team and registration performed at the same time. Queuing systems, where patients collect a number and wait for their turn, were used in one hospital. There patients were admitted at once to the ED team, without needing to register or perform a pre-triage step. If patients arrived by ambulance registration, and sometimes triage, was not required since it would already have been performed by ambulance staff. For the scope of this study patients arriving via ambulance will not be considered in depth. Triage Once the patient was admitted / registered, triage could start. Triage was described in a similar way in all study settings. It included several clinical steps such as patient history taking, clinical measurements, communication and documentation that resulted in the end in a priority scale result that classified the patients according to their need for emergency care treatment. International triage scales such as METTS and ADAPT, were used in all sites in order to prioritize patients upon triage. Triage scales were not used during pre-triage, where the patient only described the actual symptoms and a decision tool was used in order to sort patients to the correct ED ward. Triage included, patient history taking with collection of new information used to update or create a patient EHR. Information considered relevant such as allergies, previous medical care and medication was collected together with patient´s description of symptoms that led to the actual visit to the ED. Clinical measurements follow this first triage step and are performed accordingly to the information collected previously. Vital signs measurements are here performed, communicated to the patient and other staff members and documented in the patients record. Vital signs include, blood pressure, heart rate, breathing rate, temperature and conscience level. According to the Swedish triage recommendations (9), all vital signs have to be measured and documented during triage. Even if this is the rule, in particular cases such as orthopedics patients and children, patients admitted to the ED were reported to go through triage without measurement and documentation of all vital signs. This decision to not measure all vital signs upon triage was associated with factors described further on the discussion section of this study. Triage was performed, in all studied settings, by registered nurses that either worked alone or in two teams. Triage was concluded by a scale result that classified the patient according to their emergency. Triage was also the first main documentation step within ED. The way triage was documented by staff at ED varied a lot from one study setting to another. Some settings based the ED documentation in a paper based form. This form provided a structured way of collecting all 21 information needed to perform patient´s triage, monitoring and follow their treatment at the ED. It was based on tables and charts that were filled in directly after the collection of the information. The paper form followed the patient through all time at the ED and was constantly updated when new information on VS or patients clinical status was collected. Settings using the paper based forms mentioned an instant documentation of information directly after collection. For example, when vital signs were measured results were documented instantly on paper. When this paper based documentation was used during all the process, the paper would either follow the patient to the next hospital ward, or be saved and later on scanned into the EHR, if the patient was discharged home. Other settings adopted a digitalized documentation policy, where information was collected and inserted directly into the EHR. In order to achieve this digitalized way of documentation different strategies were mentioned during the interviews. Some of the settings performed triage in special triage rooms with own computers, where measurements and information could be collected by the registered nurses and immediately inserted and documented into the EHR, with no need for delays in documentation. However, in case of high patient inflow, a mixed documentation strategy could also be used. That is upon triage and when no computers were available to use at the time, or when the amount of patients didn’t allow access to the computers, paper-based documentation would be performed in order to register patient´s information in a quick and easy way. That paper would be used by staff until there was a chance to document the information in the EHR, later on. Specifically with vital signs measurements these settings described that VS would be recorded on paper, and later on introduced into the EHR. The time slot between measurement and documentation of vital signs into the patient´s journal was reported to vary a lot between 5 minutes to several hours, depending on multiple factors presented later on this study. Monitoring and treatment When triage was concluded the patients entered the next phase in the ED patient flow process: monitoring and treatment. This was the actual treatment phase where medical treatment was provided and monitoring of the patient allowed staff to supervise their clinical status. During this phase, vital signs measurement, a step called re-measurement of VS, could happen in order to monitor patient’s condition. However, not all vital signs needed to be re-measured and the selection of the vital signs that needed to be followed during the monitoring and treatment phase was associated to the patient´s individual clinical status. Re-measurements were also mainly performed by registered nurses. Documentation of information collected in this stage would follow the documentation pattern adopted at previous step, triage. However, even if documentation was performed by either paper or digital documentation, when considering vital signs re-measurements, they might even not be recorded or documented in some settings. Patients arriving at the ED via ambulance, were admitted directly in this step of the patient flow process, since triage and registration, as mentioned before, had already been performed by ambulance staff prior to arrival at the ED. Decision point The next step in the patient flow process was a decision point where patient´s clinical condition was analyzed and a decision made, by medical doctors, on whether the patient could be discharged from the ED or not. This decision was based on information gathered all the way through the actual emergency episode, documented in either paper, digital or both and resulted in a decision to maintain the patient under motorization and treatment or to proceed to discharge from the ED. If patient´s clinical condition required further monitoring, the 22 decision to discharge them to another hospital ward was taken. If the patient was considered well enough a decision would be made to discharge and send the patient home. The patient flow process ends in this way with the final discharge decision. This decision was also communicated and documented, representing both the final clinical and documentation steps in this process. Re-Evaluation As mentioned before both patient and documentation processes occur at the same time at the ED meaning that during the ED processes, events occur based on special actions and main decisions. Information collected from this study interviews revealed that in some particular cases, such as high patient inflow at the ED, patients would go through triage, face a waiting time and needed to be re-evaluated in a second triage step in order to access if their condition had deteriorate or not during the waiting time. During this re-evaluation process all or some of the patient´s vital signs would need to be re-measured and documented in order to detect if the previous triage priority result had changed. A deeper overview of the ED patient flow and documentation processes when vital signs re-evaluation is needed is presented in Figure 4. Figure 4. The re-evaluation at the ED. Re-evaluation after triage might occur upon long ED waiting times. The first horizontal row in Figure 4 shows the ED patient flow process with patient arrival, triage and re-evaluation. Triage is, as mentioned before a combination of history taking, measurement, communication, documentation and final priority scale. The second row shows that triage documentation could be performed with paper or digital support. However, as the last row shows, even if paper was used to document triage and triage results, and later on introduced into the EHR, that does not mean that previous steps, such as registration of the patient arrival, were not performed with a digital information system. Depending on the 23 hospital, documentation was performed in paper, digital or both. The main events that accompany the ED re-evaluation process were the arrival of the patient at the ED, the handover to the triage team, the triage moment and a new handover prior to re-evaluation and new triage. From a clinical point of view, actions occurred mainly during and after triage. Measurements, vital signs documentation, calculation of priority and priority documentation represented the main clinical related actions during the process. Triage priority final scale was the main decision point during triage and was supported by the use of specific triage tools and scales, mentioned before. 3.1.2 An insight into three different ED patient flow and documentation processes Since both interviews and document analysis from study settings, showed that differences occur in both patient flow and documentation processes at the ED, an insight of the documentation process supported by three different methods is provided in this section. A description of three different hospital settings is provided representing a paper based documentation ED, a digitalised ED and one that uses both paper and digital documentation. A description of patient flow and documentation processes, for other study settings is provided in Appendix B. 3.1.2.1 ED paper based documentation example From interviews performed with participants of one hospital setting, an overview of both the patient flow through the ED and how the documentation was performed was created. The following figure represents the patient flow and documentation processes at this particular hospital. Figure 5. Patient flow and documentation processes at an ED with paper based documentation. Grey= self arrival patient flow; red= ambulance patient flow; blue= documentation process. Patients arriving at the ED on their own would take a queiuing number and wait for their turn in the ED waiting room. If the clinical status of one of this patients deteriorated while waiting, they would enter a “priority lane”, the same used by patients arriving at the emergency deppartment by ambulance, where they would skip triage and be immediately admited to the emergency room. Patients were afterwards transferred to one of the ED section : internal medicine, surgery, orthopaedics or other for observation and treatment. Patients that would 24 have been waiting at the ED, were admited to a special triage room where triage was performed and for example, vital signs measured and documented. Patients were then attributed a triage priority scale result and sorted in to the correct ED section. When considering documentation, this hospital setting approached a paper based methodology. That means that all information collected through the patient´s visit at the ED was documented on a pre-defined paper based triage and ED form that was later on scanned into the EHR, upon patients discharge. Time for scanning this paper based record into the EHR could vary between some hours to some days. For patients arriving via ambulance information was transferred from the ambulance personal to the ED staff verbally, and immediately recorded on paper, or by using a print-out of the ambulance record. This paper based ED form was described as an information source that followed the patient throught the ED. 3.1.2.2 ED digital documentation process example Interview results from another hospital setting provided information on how this setting used a hole digital documentation strategy. The way this setting organised the patient flow process through the ED and the way information was documented at the same time is presented in Figure 6 below. Figure 6. Patient flow and documentation processes at an ED with digital documentation process. Grey= self arrival patient flow; red= ambulance patient flow; blue= documentation process. Patents arriving on their own at this ED could either be received by a triage team of two registered nurses or directly by the medical team, depending on how busy the ED was. That is when there were less patiens and no waiting time, patients were directly forward to the medical team skipping the triage team. Triage was in these cases performed by the medical team at the same time that treatment was delivered. Both triage and medical team had access to own computer stations and performed documentation of information, such as vital signs measurements, directly in to the patients EHR. For example if blood pressure was measured, the value was recorded immediately in to the patients electronic heath record. If the ED was facing high patient pressure, patients would wait for their turn to meet the triage team. However if their clinical status deteriorated, they would enter the priority lane and, likely 25 ambulance arriving patients, directly taken to one of the emergency rooms for immediate care. Both ambulance and emergency rooms also performed the digitalised documentation strategy. Ambulance personal would provide patient information verbally that would immediately be documented in to the patient record. This ED also had the possibility to directly receive digital patient ambulance records in to the EHR. 3.1.2.3 ED paper and digital documentation process example Finally, from interviews performed with participants from a different hospital, an overview of how the patient flow and the documentation process using sometimes both paper and digital documentation, was designed. Figure 7 presents an overview for this particular study setting. Figure 7. Patient flow and documentation processes at an ED with paper and digital documentation process. Grey= self arrival patient flow; red= ambulance patient flow; blue= documentation process. Patients arrived at the ED either by their own or by ambulance. Ambulance arriving patients entered a “priority lane”, where they arrived at the ED and were immediately transferred to the adequate ED section (ortophaedics, infection medicin). Ambulance arriving patients had, upon arrival, triage and registration already performed and that information was transferred to the ED staff either verbally, directly electronically or as a print out paper based form that was later on scanned into the patient EHR. Electronic transfer of ambulance records was not immediate, it was described to could take several days to happen. Patients arriving on their own could also enter this “ priority lane” if , while waiting at the ED waiting room, their clinical status deteriorated so much that they were considered high priority patients. Patients arriving on their own and that can wait, are submited to an outer triage performed at a reception desk where both registration and and symptoms description occurs. Vital signs were not normally collected in this primary stage in the outer triage. Patients would then be forward to triage and from there to the adequate ED section. Documentation during outer triage was performed directly into the patients EHR. Triage results could be documented either directly into the EHR or in a paper, depending on multiple factors such has high staff workload and 26 computer availability. Information documented temporary in paper was later on introduced in to the EHR, within a variable time window. Paper based documentation was reported to be used in rare cases, and the hospital policy and priority was to only use digitalised documentation strategies. 3.2 Conceptual analysis from the interviews To collect ED staffs perspectives in the quality of data documented in the ED was the second objective of this study. Perspectives on data quality were collected through the interviews and compared to the previously described framework proposed by Weiskopf and Weng in 2013 (23). 3.2.1 Coding and main concepts From an early stage two main themes emerged from the interviews, while considering the factors affecting data quality at the ED today: human and technical factors. A mind map of all the themes, categories, sub-categories and concepts obtained through analysis of this study interviews is available in Appendix C. Figure 9 provides an overview of these two themes their categories and sub-categories. Figure 8. Diagram of codification results after interviews analysis. Two main themes were identified, human and technological factors. Several categories and sub-categories were described within each theme. 27 During this section quantified results for the occurrences of each one of the coded categories is used in order to give an overview of the most common or most mentioned codes gathered from this study interviews. This quantification allowed the author to reflect on the importance of each one of the code categories regarding their occurrence during the interviews. Interviews data analysis revealed that human factors affecting data quality at ED occurred in 58% of the interviews data and technical factors in 42%. Figure 9 illustrated the occurrences of these two main themes. Factors Associated with data quality at ED. Technical Factors 42% Human Factors 58% Figure 9. Occurrences of the two major themes. This graph resulted from the interview data analysis of the study. Technical factors associated with data quality represented 42% of the data occurrences and human factors 58%. Theme 1: Human factors Early in the process of data analysis of the study interviews, the influence of human factors in the quality of data at the ED was considered important. The human factors theme relates to the interviewee’s reflections on the influence that ED staff has in the quality of data that is collected, documented and used at the ED. The three main categories included in the human factors theme were the education and competence, the management and the user need for system support. Those three occurred in different percentages in all 14 interviews according to Figure 10. 28 Factors Associated with data quality at ED. Human Factors Education and Competence (57 ocurrences) 21% User need for system Support (168 occurences) 61% Management (49 ocurrences) 18% Figure 10. Occurrences of the three main categories of the human factors theme. The graphic shows the percentage of occurrences of each one of the categories of the human theme during all the 14 interviews. Education and competence represented 21%, management 18% and user need for systems support category 61% of the Humans factors theme. Figure 11 provides an overview of the occurrences of all the categories and sub-categories included in the Human factors theme. Syste Mana m Education and geme Supp nt ort competence Human factors associated with data quality at ED. Documentation Workflow Change Management Quality control Experience Control of staff competence Slopiness and carelessness Methodology competence and knowledge Clinical competence and knowledge 87 81 10 39 15 18 14 10 16 0 20 40 60 Ocurrences 80 100 Figure 11. Occurrences of the categories and sub-categories created for the human factors theme. Occurrences through all the 14 study interviews. Education and competence The education and competence category is associated to the influence of staff´s education and experience in the quality of data at the ED. This category had 57 occurrences through all 14 interviews and was divided in 5 specific sub-categories that will now be described. Clinical Competence and Knowledge During interview number #4 it was stated that “ we record the vital signs values that appear on the screen, I think that is correct, but I don’t think we always reflect if those values are 29 plausible or not”. This statement reflects on the fact that even if the correct measurements information is collected and documented, reflection on how those measurements are adequate to that actual patient or not, may not always happen. Clinical competence and knowledge is therefore the first sub-category of this theme that defines this reflection and occurred 16 times through all the interviews. Methodology Competence and Knowledge Methodology, competence and knowledge, the second sub-category within education and competence, specifically relates to the lack of knowledge of error sources described in some interviews. Interviewee #5 reflected about error sources that could influence data quality at ED and stated that “During temperature measurement, wax in ear canal may be an error source”. Interviewee #12 mentioned that “Incorrect vital signs may always occur. Saturation measurement may be sensible to finger temperature, dirty saturation probe or wrong position of the sensor”. These two statements represent the influence of the methodology competence and knowledge of error sources during the practical measurement of the vital signs at the ED. This sub-category was the one that occurred less in all interviews. Sloppiness and carelessness The third sub-category is the presence of sloppiness and carelessness. Interview #2 showed that “One maybe does not measure pulse rate during one minute and can mix pulse rate with hearth rate.” Reflections on the presence of sloppiness and carelessness during vital signs measurements at the ED were present 14 times during all the interviews. Control of Staff Competence Interview #6 showed that “We also have a certification procedure for all staff that uses technical tools, so we can ensure that those tools are not error sources”. This reflection led to the creation of the fourth sub-category: the control of staff competence that represents eventual local measures or policies that control staffs knowledge when using technical tools. Control of staff competence occurred 18 times being the most recurrent concept within education and competence. Experience Finally within this category, experience was described in interview #6 the following way “It can depend on if you have worked a few years. Not that one is extreme experienced, but if you have some clinical experience you learn to see if this is plausible or not”. This final subcategory was described 15 times during the interviews. Management The management category within the human factors associated with data quality in the ED theme, had 49 occurrences through all the interviews. This category included quality control and change management representing organizational policies and strategies present at the study settings in order to control, monitory and improve quality of care. Quality Control Quality control sub-category was represented by 39 occurrences during the 14 interviews. It 30 included reflections on follow up of data quality, adherence to guidelines and discussions on quality. Follow up of data quality related to interview #6 statements “The only way to prove this [completeness of patient records] is to make a control check, measure and see, for example now only 40% of the registers are complete”. Follow up of data quality was mentioned as a way of controlling data quality at the ED. Discussions on quality and the follow up the adherence to actual guidelines and triage methods were other two concepts included in the quality control sub-category. Interviewee #1 mentioned that “From all the hospital perspective there are different ways of doing it [triage]”, that is adherence to established triage guidelines seemed to vary. The same interviewee reflected on the fact that “When I, as a doctor, get the triage record it can have been half an hour since the measurements in the record were taken”, showing that discussions on quality are present at the ED. Change Management The change management cub-category was mentioned 10 times through the data analysis of the interviews. Change management sub-category reflected the reflections on how hard it is to change the habit of paper based documentation at the ED. Interview #3 showed that “It is both routine and tradition to perform [documentation] on paper”. User need for System Support The final category within the human factors associated with data quality at ED theme is the user need for system support and represents 61% of the human factors theme. This category was created by reflections on how the computer systems in use influence and should support the humans working at the ED. While working at the ED and as described before, two processes occur at the same time. The patient process flow (the workflow for the staff) and the documentation process. The two main sub-categories in the system support code are therefore the workflow and the documentation. Workflow The workflow sub-category had 81 occurrences in all 14 interviews. It was described with four main concepts related to needs or requirements for the computer system to adapt to the ED staffs workflow: the mobility, the calculation of triage score, the overview and the checklist and reminders. Mobility is exemplified with an extract of interview #8 “The technological development at the wards is delayed, so we don’t have tablets or mobile devices”. Interview #1 mentions “Automatic calculation of the RETTS score? No we don’t have it. We have to do it ourselves”, and led to the concept of calculation of triage score. Overview is another concept in the workflow category. Overview of the patient status is mentioned in interview #11 “We need to see how sick the patient is and if the clinical status is progressing”. Interview #5 mentions that “One has an overview of all the patients in the ward, that is not enough with paper for that you need a computer”. This supports the overview of the ward concept. An overview of the ED process was mentioned by interviewee #3 that stated “A quick overview of the different parts of the process” was desired. Interview #9 mentioned that “A system with an overview of the ED allows us to see how much time is left until I need to measure a blood pressure for that patient”. The checklist and reminders concept was therefore created in order to represent the need to have reminders or organized checklists to what needs to be done for that specific patient during the time at the ED. 31 Documentation Regarding the documentation process at the ED, the documentation sub-category was created. Interview #2 stated that “There should exist a direct connection between maybe the patient´s id band and the [vital signs] monitor”, and reflected that automatic registration of measurements are a concept to consider. Logical control of the entered data is another concept within the user need for system support in documentation. In a statement from interview #8 “ The system does not alert if I enter a 300 value in blood pressure, it should alert in red for me to be able to see the error”, the need for the logical controls of entered data concept was supported. Reflections were made on if documented data or patient information was used retrospectively. A study participant from interview #8 mentioned that “I wonder if we look that much to older documented values”. Point of care documentation was a concept based on several statements like this one from interview #11 “Everything is documented in the triage room computer, just after measurement”. The final concept in the documentation subcategory is the completeness checks in the EHR. A statement from interview #7 shows why this concept is important “In order to perform a complete triage we have to introduce all the vital signs included, that is a system requirement to calculate the triage score”. Theme 2: Technical factors Technical factors were the second theme that resulted from this study interviews data analysis code. This theme was related to the influence of technological factors in the quality of data at the ED. The two main categories in this theme were the interoperability of IT systems and the standardization of processes and occurred in different percentages throughout the interviews according to Figure 12. Factors Associated with data quality at ED. Technical Factors Interoperabili ty of IT systems (35 ocurrences) 18% Standardizati on of Process (160 ocurrences) 82% Figure 12. Occurrences of the two categories created and included in the Technical factors theme. Results to all the 14 study interviews. Interoperability of IT systems represented 19% and standardization of processes 82% of the Technical factors theme. The figure below provides an overview of the categories and sub-categories included in the technical factors theme. 32 Interop erabilit Standardization of y process Technical factors associated with data quality at ED. Patient safety standards 14 Standardized documentation 40 Lack of standard 15 Failure to comply 55 Standardization of triage process 72 No interoperability within system 17 Non communicating systems used for… 18 0 10 20 30 40 50 Ocurrences 60 70 80 Figure 13. Occurrences of the categories and sub-categories coded for the technical factors theme. Results for all interviews. The most common sub-category was the standardizations of triage process, followed by failures to comply and standardized documentation. Patient safety standards occurred only 14 times through the interviews data. Interoperability of IT systems The interoperability of IT system category in this theme included two main sub-categories: non communicating systems used for documentation and no interoperability within ED systems. Non-Communicating Systems Lacking of communicating systems used within the ED was a recurrent statement that occurred 18 times in all interview data. Interviewee #11 mentioned that “We don’t need to use paper [in documentation], but when we take an electrocardiogram we get a printed copy and it isn’t available in the patients EHR at once”. This statement shows that printouts from digital monitors or exam machines are reported to be available on paper during treatment, and later on scanned in to the EHR. That is the exam and monitoring tools don’t communicate in an automatic way with the local EHR. The same happens to the settings where a paper based triage form is used. “We fill in the paper record where we have all the vital signs and all the medication information. The record is later on scanned” a statement from interview #12. Non-Interoperability within Systems The non-interoperability within systems is the second sub-category of interoperability and includes the presence of “free text” documentation and multiple ways of documenting the vital signs. According to interview #1 “Yes we dictate and document in free text”. The multiple ways of documenting vital signs concept, is related to the fact that VS are reported to be documented in paper, later in the EHR, as free text, in tables or other forms. Interview #5 showed that “We document directly in our paper report form and later on we introduce the values in the [EHR] tables”. 33 Standardization of Processes The standardization of processes category represents the biggest category within the technical factors theme with an 82% occurrence in the interviews data. Sub-categories in this section include: the standardization of the triage process, the failure to comply to and lack of standards, standardization of the documentation and patient safety standards. Standardization of Triage Process Standardization of the triage process is the first sub-category and is the most common in this category represented by 72 occurrences in the interviews data. A standardized triage process was reported to be organized and structured according to several concepts. The flow based triage concept was created from the report that extra actions may be taken depending on the triage result. That is if after triage “The patient shows changed consciousness level we always perform a standardized blood sugar test” interview #2. The pre-triage concept represents a clinical step before triage. The presence of this step would depend on the patient´s emergency level and whether or not it has already been performed. “Patients arriving at the ED on their own, may wait in the waiting room until triage”, interview 3, which exemplifies the low level of acuteness patients where pre-triage may not be present. Ambulance patients represented the high emergency cases where “Patients arrive via ambulance at the ED and are immediately transferred to the adequate ED sector” interview #4. Those patients had already been submitted to pre-triage or triage prior to arrival at the ED. Even if the triage processes were standardized, some settings reported a variability of this standard according to the time of day. Study setting #4 participants reported that “We have decided that Monday to Friday between 08:00 am and 21:30 pm, we have a common triage team for surgery and orthopedics”. This showed that standards used for triage methods at the ED could be adapted to the time of day. Failure to Comply Moving on to the next subcategory within the standardization of processes category in the technological theme, we enter the failure to comply with standards. Failure to comply with standards was reported to be due carelessness and sloppiness coded by these two extracts from the interviews: “Not all staff documents re-evaluation measurements of vital signs”, interview #8 and “[Not documenting re-measurements] may be due to sloppiness or maybe because we just decide not to do it” #5 interview. High workload coded from the statement “High patient inflow is often why fail for” in interview #6, is another concept created within the failure to comply with standards sub-category. However, individual variation mentioned during interview #5, “It can fell like, if it is a patient that comes for uncomplicated symptoms, unnecessary to take all vital signs” and the fact that standards are not felt to be important as described in interview #6 “We only measure breathing rate if that is why they came to the ED”, are two other concepts included in the reasons or failure to comply to standards subcategory. Even if the main triage processes were standardized in all study settings there was a reported lack of standard for the measurement and documentation of vital signs upon reevaluation of the patient at the ED. Interviewee #2 stated that “It depends on why we have to measure the vital signs and that is one of the biggest problems in the ED, to have continuity and measure continuously”. Standardized Documentation The standardized documentation sub-category had an occurrence rate of 40 in all the interview data and was represented by two main concepts the paper based triage record and 34 the template in the EHR. Interview #12 reported that “We write directly on paper and we follow the vital signs also on paper” showing the use of a paper based standardized documentation form. Interview 6 showed that templates in the EHR are another way of standardizing the documentation process “We have the triage template in the EHR where we document the vital signs”. Patient Safety Standards The last sub-category within the standardization of processes was the patient safety standards. Even if this was the less occurring sub-category, it was important to define it in both check in check out and identity control concepts. It was reported during interview #2 that “When I leave or receive the patient at the ED it feels safer if all the vital signs are documented”. This statement supports the presence of patient safety standards to transfer patients from and to the ED. Finally, standards were also reported in the patient identification and id control. Also during interview #2 it was said that “All patients at the ED are registered and provided an id band on their arm”. 3.2.2 Data analysis and the adopted data quality framework Several concepts sub-categories, categories and themes resulted from the analysis of this study interviews data. The presented data will be discussed in detail in the next section of this thesis, the discussion. From the results of the data analysis of the interviews, a relation was established between the themes, categories, sub-categories and concepts of the data analysis code, with those three dimensions for data quality based on the framework proposed by Weiskopf and Weng in 2013 (23).With this step we related the created code with the chosen data quality framework. Figure 14 summarises the relation between the data quality framework and the generated code. 35 Figure 14. The relation between the 3 dimensions for data quality framework proposed by Weiskopf and Weng in 2013 (23) and the code generated from data analysis of the study interviews. In figure 14, in each code sub-category the three data quality dimensions appear in importance order. That is the first symbol, after each one of the sub-categories represents the data quality dimension that is more influenced by that factor at the ED. The two major themes, human and technical factors, also have a connection to the three main data quality dimensions. Again the first symbol in the red chain represents the dimension that is more affected by that specific factor. All three data quality dimensions were represented in the interviews data analysis results code and within both of the major themes. All subcategories could be related to all three data quality dimensions described in the data quality framework used during this study. However, it was possible to identify the dimension that was more affected by a specific data quality influencing factor. For example, if clinical competence was deficient the main dimension for data quality to be affected would be the correctness of the collected data. If staff didn´t know how to collect vital signs or handle error sources, VS could be wrongly measured and documented. Furthermore, if clinical competence failed the completeness of the vital signs data might also be affected, since problems with measurements would result in incomplete records. The less affected data quality dimension upon deficiency in the staff´s clinical competence was considered to be currency. This interpretation can be extended to all subcategories of the data analysis code represented in figure 14. From a general point of view correctness was the more sensible dimension for data quality that would be easier to affect with both human and technical factors and was followed by completeness and currency. The way the different categories and sub-categories relate to the proposed dimensions will be explore further during the next section of this thesis, the discussion. Figure 15 represents the main founded influencing factors for each one of the data 36 quality dimension. Figure 15. Most influencing factors of the three data quality dimensions. This figure illustrates which factor has an higher impact in each one of the three considered data quality dimension at the Swedish ED. The code obtained from the data analysis of the interviews was also used to complete and strengthen the previous process description of both patient flow and documentation processes at the ED. With the visualizations of those two ED processes, it was easier to see exactly where and when each one of the coded factors influenced data quality at the ED. Themes, categories and sub-categories were used together with the mapping of the patient and documentation processes at the ED to relate the staff´s perspectives on which factors affect vital signs data quality at the ED to the different steps of the ED processes. This connection related the coded results with correctness completeness and currency, and mapped those results through the patient flow and documentation processes at the ED. 4. Discussion This thesis aimed to explore the factors that influence the quality of data collected and documented at the ED. Different ED settings were studied in order to gather information to map patient flow and documentation processes within the ED. The code that resulted from the analysis of the interviews was both compared to a proposed framework from data quality and used to better understand and map the different factors affecting that affect vital signs data quality in the ED. A general description of the main findings of this thesis will be presented, followed by a reflection on the study methodology. The methodology reflection of this study will consider limitations and strengths, trustworthiness, credibility and dependability and will conclude this section. 37 4.1 Patient flow, documentation processes and staffs insights on data quality in the Swedish ED This thesis showed that patient flow and documentation processes take place at the same time at the ED. Quality of vital signs data documented in the ED can be affected by both human and technical factors that have an impact in the data completeness, correctness and currency in the patients EHR. This study showed that human factors represented 58% of the factors that could have an impact in data quality at the ED suggesting that, even if technical factors exist, human factors seem to be the most influencing ones. At the ED both patient flow and documentation processes need to happen without friction and are therefore dependent on a pre-defined structure in order to work. Constant evaluation and improvements of those ED processes allow staff to work with those processes in a constructive efficient way. This generates a cycle of constant structured based work that is often evaluated and improved in order to support ED staff in their work. This cycle was named the ED cycle and is represented in figure 16 together with the major themes, categories and data quality dimensions studied. Figure 16. The ED cycle. This cycle results from both patient flow and documentation processes at the ED. In figure 16, the human factors that affect data quality at the ED are represented on the right side, in orange together with their major categories. The technical factors theme and categories are represented on the left side of the figure, in blue. The three data quality dimensions, proposed by the framework used for this study, are represented at the center of 38 the cycle. 4.1.1 Standardization of Processes Both patient flow and documentation processes in the ED need to be structured. The patient flow process is connected to several clinical and standardized steps that allow the patient visit at the ED, to follow a similar pattern independently of the considered hospital setting. Standardized procedures Standardized triage procedures were present at the ED of all studied clinical settings. Triage structure and standards were pointed out during the interviews, as one of the factors that may affect the quality of data in the patient records within the ED. This factor was coded as the standardization of triage process sub-category within the technical factors theme. The documentation process follows side by side with the patient flow process, and represents the moments within the patients visit to the ED where documentation of patient related information, such as vital signs, is performed in to the patient record. Triage documentation was described in the settings to happen in one of three ways: on paper, digital or both. Regardless of the way it was performed documentation was structured, which led to the standardized documentation sub-category. Interview data analysis results indicate that the presence of standards for triage procedures and documentation are needed, in order to ensure that ED visits are structured and organized. A structured triage procedure seems to ensure that all patients are received and treated in a similar way. Furthermore, triage process standardization seems to contribute for the completeness, correctness and currency of the information documented in the patient record. When using a standard to perform triage and when structure for triage variations (such as the need for extra exams in special cases, the time of day or the way patient s arrive at the ED) is present, staff is given the opportunity to collect enough information that ensures that the patient record is complete. Compliance to Standards The patient flow process suggested, however, that triage is performed in full scope for all patients arriving at the ED, but orthopedics patients were often reported to not be submitted to a full triage indicating that, even if standards and structure are available, there are failures in standards compliance. Data analysis findings prove that there is lack of compliance with the standardized process of triage due to, either patient individual variation or the fact that standards are not felt to be relevant. Even high workload and carelessness were reported as reasons not to comply with triage standards. This implies that the completeness of the triage information recorded or documented in the patient record may be affected by the same reasons that lead staff to fail to comply with standards. How to improve standards Documentation of triage information in a structured paper form or in an EHR template can help to improve completeness of the record, by allowing staff to have a form or guide for documentation. This specific point is related to the impact and the needs of a supporting EHR system that provides staff with checklists and process overviews, and that will be discussed later on. By providing an overview of the triage process and the patient information, failure to comply can be minimized as staff is reminded of all the information that needs to be 39 collected. Furthermore, if standards are extended to all clinical steps within the patient flow process even completeness of the information of the patient´s re-measurements, for example vital signs, after triage can be improved since results show that standards for both measurement procedures and documentation of these repeated measurements is lacking. But even if standards may help to achieve completeness in the EHR, the way standards are supported and delivered seems to have a big impact on both correctness and currency. If triage documentation is performed on a paper-based standardized form, it is reported that, for example, vital signs information is documented immediately after measurement with no need for intermediate information support, and therefore minimizing the risk for errors. Because information is written exactly after measurement timeliness is extremely accurate. Correctness and currency seems therefore to be optimal with paper based documentation. Digital based settings reported that computer stations were available physically “at point of care”. This means that when information was collected it was immediately timed and documented in to the EHR. This point of care documentation had been achieved by several organizational strategies such as, having a triage team consisting of two RN where one performs the patient contact and measurements and the other documents collected information directly in to the patient record. Digital Settings Digital based settings documented in a standard EHR template that allowed the user to move further without introducing all the information in the template. For example, if vital signs were measured, the user could introduce only blood pressure and temperature and be able to move forward in the system. This possibility represents a risk for the completeness of the patient records. Since there was no obligation to introduce information on all the vital signs, there is a risk for the patient record to be incomplete. This was common in the different EHRs available at the settings. Furthermore, systems did not provided logical controls of introduced data, another concept that will be described further later on, affecting therefore the correctness of the documented information. Currency seemed to be present in digital based settings since, as mentioned before, information was introduced in to EHR directly after collection. When triage documentation was performed with both paper and digital support, data quality dimensions seemed to be highly affected. These settings described that triage data could either be only collected or collected and documented in a paper and later on introduced in to the EHR. This means that using the vital signs example once more, if measurements were performed either staff would document it in a piece of paper or just don’t document at all until available time and computer allows the introduction of the information in to the patient record. This primary paper documentation implies that correctness, completeness and currency are affected. Lack of standards during the collection and primary documentation of the patient data leads to higher risk for incomplete gathering of information affecting completeness. The fact that information from several patients might be primary documented in this paper increases the risk for information to be incorrectly introduced in to the patient record. Finally, because data is only introduced in to the electronic patient record later on and after the measurement, the user needs to set back the clock to the time measurements were performed, which normally is forgotten resulting in a lack of accuracy in the data currency. Standards and Data Quality Standards are the base and building blocks for achieving better structures and routines that allow patient flow and documentation process in the ED to work. Standardization of processes 40 was the strongest category included in the technical factors theme confirming that this is a major factor associated with data quality at the ED. This finding is consistent with “the fact that structured entry, compared to handwriting, may encourage recording of specific or otherwise unincorporated data elements resulting in a more detailed record”(44). Data quality can be affected in its completeness, correctness and currency if standards fail to exist, work properly or even if staff fails to comply. Patient safety, specifically during ED treatment and upon discharge to home or other ward, is highly dependent on the existence of standardized process that help documented patient data to be complete, correct and current so it can trustfully be used. 4.1.2 Interoperability of IT systems, Education and Competence When standards exist they need to be tested and used in the next ED cycle phase, the working phase. In order to work, standards need both technical interoperability of IT systems and human education and competence. The patient flow and documentation processes show that patient record information is collected and used from the patient´s arrival at the ED, all the way until after discharge. Human requirements Both processes are dependent on the education and competence of staff members that ensure that the processes happen in a structured, organized and quality way. To be able to use, collect and know how to find patient related information through the ED, education and competence is required from the staff. Competence and knowledge on both clinical and methodological procedures is extremely important in order to ensure correctness of the collected data. During triage, vital signs measurements are dependent on several tools that need to be known by the staff in order to provide accurate measurements. But sometimes plausibility of the obtained values needs to be accessed in order to see, whether that measurement is plausible for that particular patient or not. Plausibility appeared to be related to clinical experience. It was implied that the correctness of the documented patient data in the EHR could be influenced by clinical, methodology competence and experience. Senior staff members were reported to easily react to an error suspicious measurement. Several settings were aware of this factor and performed, therefore, several controls on staff´s competence. Measures such as educational sessions on new measurement tools and spontaneous control of ED patient records seemed to be used in order to detect failures and improvements areas. But even if education was provided and competence tested, sloppiness and carelessness was reported, showing that even if standards and structures are present, several factors may influence humans working at the ED and therefore affect the quality of data in the patient records. All three data quality dimensions correctness, completeness and currency can be affected by these attitudes since effects of sloppiness and carelessness can go from wrong values documented in wrong patient record, incomplete introduction of data and even no attention to the time when measurements were taken. As mentioned before patient flow and documentation processes show that information moves from patient arrival to after ED discharge. Technical requirements In order to allow this information to be available and to work in the ED, technical interoperability of IT systems is required. When paper based documentation procedures are 41 present in triage it is important to consider that patient information recorded in those forms is only accessible physically at the ED. The use of information in a distance form (for example if other wards need to be consulted) is not possible since there is no interoperability between the paper form and the patients EHR. The same problem occurs when ambulance services provide a print out of their journals or when specific exams, such as electrocardiograms, are required. Information contained in those papers is reported to be scanned after the patients visit to the ED. Digital settings may also have a delay in introducing patient related information, as mentioned before, when they use a paper based primary documentation. The fact that ED data is not available at once in the patients record, affects the completeness of the record, since information of actual or recent treatments may be missing. Even if paper form for documentation were considered to be optimal in their completeness, correctness and currency related to their standardized form and use, they fail grandly from a communication or interoperability of systems used for documentation point of view. But even digital based documentation settings may see completeness and correctness of ED EHR data affected when free-text based documentation is allowed or when there are multiple ways of introducing the same information. Using again the vital signs example, EHR systems may be organized with different sections where free text documentation is possible when tables or templates are not provided. Information on vital signs measurements may be therefore documented in different sections depending on the staff. This fact may affect completeness since not all vital signs might be documented and correctness since they even might be recorded in the wrong sections of the EHR. ED Work Support is Essential When standards exist and, are well known and used with the correct support they contribute to working patient flow and documentation processes. Education and competence together with interoperability are the perfect example of how human and technical factors interact and may affect the quality of data in the ED. Human factors such as the lack of physicians with advanced experience “stresses the need for solutions that simplify documentation and allow more time for direct patient care” (45). It is therefore essential to develop solutions that can create a balance between the human needs and the technical solutions. A way to develop these solutions is to gather information from the clinical sites. That is why constant evaluation and monitoring of ED policies and work is needed. 4.1.3 Management Management strategies are of importance throughout the patient flow and documentation processes. This is a major part of the ED cycle where standards working on a particular setting are evaluated, controlled and eventually changes in order to improve the quality of care and consequently the quality of EHR data in the ED. Management and Quality Control Interview results revealed that the described patient flow and documentation processes are a result of management strategies, associating this category to 18% of the human factors influencing quality of data in the ED. Management includes both quality control and change management and is a category that can have impact in all three data quality dimensions. Quality control by follow up on data quality, adherence to standards or guidelines and even discussions on quality has the power to raise the data quality subject within the ED working setting. Both follow up on the quality of data recorded in to the ED EHR, by retrospective 42 controls or the follow up on the adherence of the established standards for triage and ED, such an evaluation of the completeness of the vital signs in ED patient records, provide valuable information about the actual status of data quality in that specific setting. To discuss data quality with staff was often referred during the interviews as one of the measurements that would help improve the quality of data recorded in to the EHR. Interviewees reported that they feel more engaged to perform changes if they know why changes are needed. Taking the vital signs example again if discussion regarding the importance of vital signs documentation in the patient records is raised, staff may feel more engaged to perform change and reduce factors previously mentioned such as sloppiness and carelessness. It is assumed therefore that if ED staff was provided with knowledge on the factors that define quality of data in the patient records, they might help to improve correctness, completeness and currency. Settings that used a paper based documentation process named that habits may be hard to change, paper forms seems to be a part of their cultural ED setting and therefore considered almost as a tradition. Both evaluation and change are needed to improve data quality in the ED, and both are dependent on management and quality control strategies. Finally this section support previous studies on how important motorization and management of change is for the improvement and development of EDs (45)(26). Rao et al wrote that: “Implementers should carefully consider how implementations will affect information flow and then expect the unexpected”(46). 4.1.4 User need for System Support The final step within the ED cycle is the one that takes place when standards are present, used and evaluated and new support is developed to help staff with the execution of both process flow and documentation processes. These processes show that documentation in the ED is tightly related to the different steps in the patient flow process. Staff Workflow Support Factors related to the way EHR systems support staffs work in the ED, were coded system support and represented the biggest part (61%) of the considered human factors that affect data quality in the ED. As mentioned before, the patient flow process can be related to the ED staff workflow, and an EHR system needs to adapt to that workflow in order to allow staff to perform better. During the study interviews factors such as mobility and calculation of triage scale were mentioned as possible sources for improvement of data quality in the ED. Paper based documentation settings showed us that the bedside documentation provided a high level of data correctness, completeness and currency. It is therefore proven that in order to allow improvement of the quality of data in the ED, a system needs to be portable and mobile to follow both staff and patient through the ED process. This finding is supported by previous study that concluded mobile devices “improved patient physician interaction, improved workflow and structural iPad benefits promoted their demand. Physicians perceived the structural benefits of iPads would improve patient physician interaction and improve workflow in the ED”(46). Support and Data Quality Data quality correctness may even be further improved if the system provides a tool that automatically calculates total triage scores. If the system allows an automatic calculation of the total triage score for a patient after the introduction of all the contents and information collected during triage, correctness may improve since calculations errors are less likely to 43 occur. However, if the wrong values are introduced in the wrong EHR, or if incomplete or wrongly measured values are used, correctness will instead decrease, as mentioned before during the structure and standards section of the ED cycle. As discussed also in that section the use of checklists and reminders may also have a positive impact in the data correctness, completeness and currency. If checklists are used the risk of forgetting a task, measurements or introduction of needed information seems to be less. These solutions may also have time planed tasks that allow staff to plan their actions and improve currency by allowing timeliness in the EHR data to be current. Patient flow process showed us an overview of the patient´s process through the ED. In a similar way staff mentioned that they need to have a constant overview of everything related to that patient during that ED visit. Again this is easily achieved by the paper based form where it was mentioned that one can easily go through the patient´s information in a quick and direct way. Systems supporting staffs workflow in the ED need therefore to be able to provide overviews on the patient and their process in the ED. Some settings also described that their current system allows also an overview of all ED providing also the support to communicate via instant message via computer with other staff members. The fact that an overview of the ED was provided seemed to be a great feature because participants appreciated the fact that they could see staffs availability and from there regulate and plan their actions in a more structured way. Structured support adapted to the staff´s workflow was suggested to help improve the quality of data but support on for the documentation process represented the majority of the occurrences within the system support category suggesting that this may be the support that could have greater impact in the quality of data in the ED. All three considered data quality dimensions were suggested to be able to be improved when documentation support was present in the ED system. Monitors for vital function are used in some of the studied settings and were referred in some interviews as a way to improve correctness of, for example the vital signs documentation. It was stated that if monitors were automatically connected the ED EHR, they would allow values to be introduced in the patient record in automatic way without the staffs intervention helping therefore to improve correctness, currency and even completeness. However, discussions from automatic registration of measurements form monitors mentioned a previous described factor, patient security. Some interviewed RN mentioned the fact that patient identity would need to be controlled in order to be able to use this direct documentation from monitors in to the EHR. There is an obvious knowledge of the fact that the use of direct documentation from monitor to the patient record may either improve the studied data quality dimensions or damage them, depending on both human and technical factors. Improvement of Recorded Data by System Support Mentioned before during the description of the working part of the ED cycle, was the need for systems to have a logical control of entered data in order to avoid errors and improve correctness. An example is the fact that actual EHRs, allow the introduction of systolic blood pressure measurements of 300 millimeters of mercury, a value impossible to obtain. Also completeness checks, mentioned before improve both the staff´s compliance to standards and the completeness of the data documented in to the patient record. As mentioned before when documentation occurs directly after the execution of the measurements or clinical treatments, all three data quality dimensions seem to be improved. That is why a point of care documentation should be provided in order to adapt the system to the ED documentation process needs. Finally it was previously described that settings that use a paper based documentation process, lacked the opportunity to both allow ED patient information to be available to other professionals, outside the ED and to consult previous patient related information stored in the patient record. The retrospective use of the data should therefore be 44 available in order to allow documentation performed during the patients visit at the ED to be available in a complete, correct and current way. This final step in the designed ED cycle continues further on with a new structure and standards phase, that will be again tested and evaluated leading to new improvements and changes that will support the ED processes. Even if the presented ED cycle structured this process in separate phases and with distinct influencing factors for data quality, it is important to have in mind that both human and technical factors can be related to all cycle phases. The ED seems to need to have a sensible balance between both human and technical factors that support and ensure both patient flow and documentation processes, in order to achieve EHR data quality. Completeness correctness and currency don’t seem to be able to achieve only by making changes or improving either human or technical influencing factors. An adequate balance between them seems to be the key to success. 4.2 Methodology considerations Qualitative methods were used in this thesis to gather information and knowledge on the study subject. The sampling strategy described was used in order to try to select participants with a deep or good understanding in both ED patient flow and documentation processes. Multiple settings were considered in this study which allowed different perspectives and strategies to be known and studied. The use of a known and previously described framework, allowed the researcher to search for factors influencing the quality of data in Swedish ED. Interviews were a powerful tool to collect insights and perceptions from the ED staff. The structure used during the first interview, the pilot interview, was optimized and strengthen in order to focus on completeness, correctness and currency of the ED data. At the end saturation was present since no new information was gathered and the few last interviews helped reinforce collected data from the previous ones. In the following sections issues such as study limitations and strengths, trustworthiness, credibility and dependability will be discussed. 4.2.1 Limitations and strengths of the study Even if recognised qualitative research methods were used in this study the researcher knows that the validity of the findings may be affected by limitations. The first limitation of this study is the language. The fact that all interviews and part of the literature research was performed in Swedish, means that collected data needed to be translated to English prior to data analysis. This step may represent a risk to misinterpretation or wrong translation. The initial use of open-ended questions during the pilot interview allowed the interviewee to reflect and describe both triage and documentations methodologies in depth. But the information collected during this interview might have been insufficient regarding the three main factors for data analysis used during this study: correctness, completeness and currency. Another limitation is the fact that an extended relationship was not established with the interviewee since they were only interviewed once. Multiple interviews or group interviews may have been ideal. However, by performing small group and mainly individual interviews 45 data was collected and analysed in a more in depth way which strengths the conclusions gathered to each one of the settings. Since the inclusion criteria used in this study required or asked for participants to work as RN, because they are the ones responsible for triage in Sweden, a limitation may be the fact that the pilot interview was performed with a medical doctor and not a RN. A new and improved questionnaire was developed after this pilot interview and again tested once more with the first RN interview, which ensures that the achieved and selected questions were adapted to the information that needed to be collected even with RN participants. Limitations might also be considered in the process results. A general patient flow and documentation process for the ED was created by the researcher from information collected during the interviews and the analysis of the settings ED documentation. These generalised processes might not be able to be true in all ED settings, such as smaller hospitals or ED outside Sweden, since they were developed based on information from each one of the study settings. But the fact that eight different main hospitals were studied might also be seen as one of the study strengths for the use of the generalised processes in other main Swedish ED. Finally, I believe it is important to consider that data was collected from healthcare professionals and that may represent a bias in this study when considering the fact that human factors influencing data quality at the ED occurs more often than the technical ones. This might be explained by the fact that for healthcare professionals human factors are more natural than technical ones which may have influenced the fact that human factors were more recurrent in the data collected in the study. 4.2.2 Trustworthiness, credibility and dependability Trustworthiness in this study was ensured by peer debriefing of the study process from the researcher team. As recommended, “the team sough feedback on the evolving findings and interpretation of the data”(38) and the detail in coding and analysis was consolidated by saturation. A strategy that was used to validate this study is a primary qualitative researcher´s strategy called triangulation. According to Creswell, 2008 triangulation is the process of gathering evidence from different individuals, types of data or methods of data collection in descriptions and themes in qualitative research. This thesis used triangulation of analysed data that is, evidence from the main supervisor and clinical advisor, data collected from interviews and even documents related to the triage and ED from the different clinical settings contributed to the data analyses process. The researcher examined each one of the information sources and founded evidence to support the created themes providing this report with both accuracy and credibility. Finally, dependability is an evaluation criteria used when a phenomena may change depending on the environment, research method or time resulting in instability to the research (39). Even if findings differed from setting to setting the influencing factors for data quality at the ED described, should be able to be found in similar settings by similar methods. 4.3 Implications of this study : Recalling the sepsis example Going back to the clinical example of sepsis patients, presented during the introduction part of this thesis, is a way of showing how the finding soft hi study can improve sepsis diagnosis and treatment. As mentioned before sepsis diagnose needs to happen as soon as possible upon 46 vital signs changes in order to improve patient chances to survive the disease. It was mentioned that sepsis patients missed vital signs documentation in their journals, which can mean that either vital signs were not fully measured or they were but documentation was not performed. Using this Sepsis example, the way to improve sepsis patient´s diagnosis and outcomes would be to create a clinical decision support that assists ED staff. Benefits of the creation of such a system would be the use of a standardized and organised way for documenting patient related data with the contribution for the improvement of the data quality, and the instant use of that data to assist clinical decisions and treatment at the point of care. If all vital signs data needed to be introduced in the CDS, in order to move forward within the system, completeness would be ensured. Furthermore, eventual direct link between vital signs monitoring stations and the patient EHRs would allow an instant feed of vital signs data ensuring that they are currently documented even improving correctness, since VS would be automatically measured. Alternatively, the use of bedside stations or mobile devices for ED documentation would also contribute to improve vital signs currency at the EHRs, since the measurements would be performed and documentation possible directly afterwards. Staffs education and sensibility for data quality issues would also contribute to reduce the incompleteness and incorrectness level of today´s sepsis patient´s EHRs. So a way to easy and improve documentation of vital signs upon triage and early diagnose of sepsis patients would be a clinical decision support system. The system would provide healthcare professionals warnings and treatment recommendations at point of care that eventually would have an impact in sepsis patient´s clinical outcome. A practical example is that the CDS would warn if not all vital signs were introduced in the patient record, and if the breathing rate was altered, warnings could be activated in order to alert for eventual sepsis diagnose. But the design, development and functionality of an accurate sepsis CDSS needs introduction of quality data. This is when understanding and knowing the factors that can affect data quality at the ED is essential. The findings of this study therefore a first step to identify and recognize potential influential factors of data quality in the ED. 4.4 Future research Future research may evaluate how currency can be improved on vital signs data. In this effort researchers may use again the Sepsis example to quantify data and see how currency is affected in a more detailed way by using quantitative study methods. This research may lead to the design and development of a CDS for early detection of, for example, Sepsis patients. Finally, the impact on the healthcare professional’s performance and workflow as well as the patient outcomes and quality of vital signs data recorded at the ED EHRs, call for even additional studies and evaluation aimed for the development and continuous improvement of this subject. 47 5. Conclusion This thesis concludes that Swedish EDs contain barriers for achievement of EHRs data quality. The mapping of both patient flow and documentation processes performed in this thesis, together with the described factors that can influence those processes and impact data quality in ED EHRs, provided an overview of both the barriers and the opportunities to achieve quality of data in the ED patient records. Human factors were the ones pointed to have a higher impact in the quality of data at the ED. However, the major impact in completeness, correctness and currency of data at the ED is achieved when both human and technical factors are considered. There seems to be an essential need for supporting the way documentation is performed at the ED in order to improve data quality. This study concludes that a way of ensuring that electronic systes work reliably on the data from patients EHRs, is to introduce standardized documentation. The use of standards is already in place for the clinical and patient related procedures, but seems to still be under development for the documentation routines. This study identified the major factors that can affect data quality at the ED and the findings should be extended in the future to help understand better those factors and the way to improve them. 48 References: 1. Socialstyrelsen. Väntetider vid sjukhusbundna akutmottagningar [Internet]. Stockholm. 2011. Available from: http://www.socialstyrelsen.se/publikationer 2011/2011-3-36 [Cited 2015 Jan 4] 2. Esmail N. Health Care Lessons from Sweden. Lessons from Abroad: A Series on Health Care Reform. Fraser Institute.2013. 3. Anell, Anders, Anna H Glenngård SM (2012). Sweden: Health system system review. Health Systems in Transition, 14 (5). [Internet]. Fraser Institute. 2012. 4. Association for Local Authoroties and Regions. Swedish Health Care in Transition. Structure and Methods for better Results [Internet]. Stockholm; 2009. Available from: www.skl.se [Cited 2015 Feb 4] 10. Hälso och Sjukvårdsförvaltningen Stockholm Läns Landsting. Genomlysning av Stockholms fem stora akutmottagningar [Internet]. Stockholm; 2013. Available from: http://www.sll.se/Global/Politik/Politiskaorgan/Landstingsstyrelsen/Produktionsutsk ottet/2013/2013-11-26/punkt11.pdf 11. Eitel DR, Rudkin SE, Malvehy MA, Killeen JP, Pines JM. Improving Service Quality by Understanding Emergency Department Flow: A White Paper and Position Statement Prepared For the American Academy of Emergency Medicine. J Emerg Med. Elsevier Inc.; 2010;38(1):70–9. 12. Socialstyrelsen. Omfattningen av administration i vården Innehåll. Stockholm; 2000. 5. Gerdtz MF, Bucknall TK. Triage nurses’ clinical decision making. An observational study of urgency assessment. J Adv Nurs. 2001;35:550–61. 13. Hitchcock M, Gillespie B, Crilly J, Chaboyer W. Triage: An investigation of the process and potential vulnerabilities. J Adv Nurs. 2014;70(October):1532–41. 6. Fernandes C, Tanabe P, Gilboy N, Johnson L, McNair R, Rosenau A, Sawchuk P, Thompson D, Travers D, Bonalumi N, Suter R: Five-level triage: a report from the ACEP/ENA five-level triage task force. J Emerg Nurs 2005;30:67-69. 14. Socialstyrelsen. Kompetensbeskrivning för legitimerad sjuksköterska. Stockholm; 2005. 15. Widgren, B., Jourak, M., Martinius, A., 2008. METTS-A ger underlag för rätt prioritering till rätt vårdnivå. METTS-A is a base to correct priority to correct level of care (in Swedish). Swed. Med. J. 105 (4), 201–204. 16. Nakahara, S., Matsuoka, T., Ueno, M., Mizushima, Y., Ichikawa, M., Yokota, J., et al., 2010. Predictive factors for undertriage among severe blunt trauma patients: what enables them to slip through an established trauma triage protocol? J. Trauma 68 (5),73-86 . 17. Menachemi N, Collum TH. Benefits and drawbacks of electronic health record systems. Risk Manag Healthc Policy. 2011 Jan [cited 2014 Jan 6];4:47–55. 18. Weiskopf NG, Weng C. Methods and dimensions of electronic health record data quality assessment: enabling reuse for 7. 8. 9. Jönsson K, Fridlund B. A comparison of adherence to correctly documented triage level of critically ill patients between emergency department and the ambulance service nurses. Int Emerg Nurs. 2013;21(3):204–9. Farrokhnia N, Göransson KE. Swedish emergency department triage and interventions for improved patient flows: a national update. Scand J Trauma Resusc Emerg Med [Internet]. BioMed Central Ltd; 2011;19(1):72. Swedish Council on HTA. Triage and Flow Processes in Emergency Departments Swedish Council on Health Technology Assessment SBU Board of Directors and Scientific Advisory Committee Secretariat Board of Directors. 2010. 49 clinical research. J Am Med Inform Assoc [Internet]. 2013 Jan 1 [cited 2014 Mar 28];20(1):144–51. Available from: http://www.pubmedcentral.nih.gov/articlere nder.fcgi?artid=3555312&tool=pmcentrez &rendertype=abstract 19. Socialstyrelsen. Tillsynsrapport 2013 Hälso- och sjukvård och socialtjänst [Internet]. Stockholm; 2013. Available from: www.socialstyrelsen.se 20. Leape LL, Bates DW, Cullen DJ, et al. Systems analysis of adverse drug events. ADE Prevention Study Group. JAMA. 1995;274:35–43. 21. Johnson N, Mant D, Jones L, Randall T. Use of computerised general practice data for population surveillance: comparative study of influenza data. BMJ. 1991;302:763–5. 3. 28. Nyl U, Berglund C. Markörbaserad journalgranskning.Sveriges Kommuner och Landsting. 2013. 29. Linnér A, Sundén-Cullberg J, Johansson L, Hjelmqvist H, Norrby-Teglund A, Treutiger CJ. Short- and long-term mortality in severe sepsis/septic shock in a setting with low antibiotic resistance: a prospective observational study in a Swedish university hospital. Frontiers public Health. [Internet]. 2013;1(November):51. 30. Miltner RS, Johnson KD, Deierhoi R. Exploring the Frequency of Blood Pressure Documentation in Emergency Departments. J Nurs Scholarsh. 2014;98–105. 31. Svenfors J, Vikefors T et al. Special sepsis: nationellr kvalitetsregistrer kan ge bättre vård vid livshotande sepsis. Läkartidningen. 2011;V108;6S279–81. 22. Hogan WR, Wagner MM. Accuracy of data in computer-based patient records. J Am Med Inform Assoc [Internet]. [cited 2015 March 16];4(5):342–55. 32. Nationell Kommite för Infektionssjukdomar. Kvalitetsregistret för svår sepsis / septisk chock Årsrapport för 2013. 23. Botsis T, Hartvigsen G, Chen F, Weng C. Secondary Use of EHR: Data Quality Issues and Informatics Opportunities. AMIA Jt Summits Transl Sci Proc AMIA Summit Transl Sci. 2010 Jan [cited 2014 Sep 4];2010:1–5. 33. Chellel A, Fraser J, Fender V, et al. Nursing observations on ward patients at risk of critical illness. Nurs Times 2002; 98 (46): 36-39. 16. 34. Mcgain F, Cretikos MA, Jones D, Dyk S Van, Buist MD, Opdam H, et al. Documentation of clinical review and vital signs after major surgery. MJA. 2008;189(7). 35. Corbin, J. M., & Strauss AL. Basics of qualitative research: Techniques and procedures for developing grounded theory. Sage publications, Inc.; 2008. 36. Elo S, Kyngäs H. The qualitative content analysis process. J Adv Nurs. 2008;62(1):107–15. 37. Asplund K, Castrén M, Ehrenber A, Farrokhnia N, Göransson K, Jonsson H, et al. Triage och flödesprocesser på akutmottagningen En systematisk litteraturöversikt. Scand J Trauma Resusc Emerg Med. 2010;(April):281. 38. Creswell, J. W. (2008). Educational research: Planning, conducting, and evaluating quantitative and qualitative 24. Prokosch HU, Ganslandt T. Perspectives for medical informatics. Reusing the electronic medical record for clinical research. Methods Inf Med 2009;48: 38-44. 25. Juran JM, Gryna FM. Juran’s Quality Control Handbook. 4th edn. New York: McGraw-Hill, 1988. 26. Yeung MS, Lapinsky SE, Granton JT, Doran DM, Cafazzo JA. Examining nursing vital signs documentation workflow: barriers and opportunities in general internal medicine units. J Clin Nurs [Internet]. 2012 Apr [cited 2014 Sep 7];21(7-8):975–82. 27. Brink M, Cronqvist J, Furebring M, Gårdlund B, Lanbeck P, Ljungström L, et al. Vårdprogram Svår sepsis och septisk chock initial handläggning. Svenska Infektionsläkareförening.2013. 50 research. Upper Saddle River, N.J: Pearson/Merrill Prentice Hall. 39. Kvale,Steinar(1996) Interviews:AnIntroductiontoQualitativeRes earchInterviewing. London..SAGE, pp.124135. 40. Sveriges Kommuner och Landsting. Akut förbättring ett nationellt projekt för bättre patientflöden på akutmottagningar [Internet]. 2013. Available from: www.skl.se 41. XMind [Internet]. 2015 [cited 2015 Jan 1]. Available from: www.xmind.com 42. VUE-Visual Understanding Environment [Internet]. 2015 [cited 2015 Jan 1]. Available from: www.vue.com 43. Gliffy [Internet]. 2015 [cited 2015 Mar 1]. Available from: www.gliffy.com 44. Apkon M SP. Impact of an electronic information system on physician workflow and data collection in the intensive care unit. Intensive Care Med. 2001;Jan;27(1):122–30. 45. Füchtbauer LM, Nørgaard B, Mogensen CB. Emergency department physicians spend only 25% of their working time on direct patient care. Dan Med J. 2013;60(1):1–5. 46. Rao AS, Adam TJ, Gensinger R, Westra BL. Study of the factors that promoted the implementation of electronic medical record on iPads at two emergency departments. AMIA Annu Symp Proc [Internet]. 2012;2012:744–52. 51 Appendix A- Main interview Guide (In Swedish). Intervjuunderlag datafångst sepsisstudie, dokumentation akutprocess. Intervju med: Arbetar på, sjukhus, enhet: Erfarenhet från akutsjukvård: Insyn i rutiner kring akutflöde medicin / kirurgi / ortopedi / annat? Klinisk tjänstegrad: Akutprocess mätrutiner: Beskriv hur er triageringsprocess ser ut. Beskriv vad som påverkar om om patienter triageras fullt ut eller inte? Är icke triagerade patienter friskare än triagerade? Akutprocess dokumentationrutiner Beskriv hur ni dokumenterar triagering och vitalparameterar Beskriv Hur kompletta du uppfattar att registeringarna av vitalparametrar är Vilka faktorer ser du dom stärker kompletthet? Vilka faktorer uppfattar du som hinder för kompletthet? Beskriv Hur rätt du uppfattar att registreringarna av vitalparametrar är Vilka faktorer ser du som stärker korrekthet? Vilka faktorer hindrar korrekthet? Beskriv Hur rätt tidsstämplarna kring vitalparametrar är? Vilka faktorer stärker tidsfaktorn i mätning och registrering? Vilka faktorer hindrar tidsfaktorn i mätning och registrering? Om pappersdokumentation: Vad hindrar er från att gå över till digitala arbetssätt? Appendix B- Patient flow and documentation process for Hospital settings A; C and D. Hospital Setting A In this setting patients arriving alone at the ED are registered by a primary team that forwards patients to the correct ED. Vital signs may not be measured on all patients in this step. If patients show minor emergency care need they will be forward to the minor emergency department “lättakuten”, if not patients will wait in the ED waiting room for their turn. While waiting if their clinical status deteriorates they will enter a fast line called “larm”, the same used by ambulance arriving patients and where patients meet the emergency team at once. Patients arriving by ambulance have their clinical information, including vital signs, transferred verbally to the ED team or digitally send to their EHR. This digital transfer is however not instant and might take several days. From this point and when stable patients are forward to one of the ED wards where vital signs remeasurements may happen. These re-measurements are expected to be documented at once in to the EHRs, but are reported to not always be documented. Patients that have been waiting in the ED waiting room meet a ED team that executed triage and forward the patient to one of the ED wards. This team is expected to perform documentation directly in to the EHRs, but can reportedly perform documentation first in a paper and later in the EHRs, when high patient intake happens. 53 Hospital Setting C In this setting patients arriving alone at the ED take a queuing number and wait for their turn. When their turn comes a primary team performs an outer triage that forwards patients to the correct ED. Vital signs may not be measured on all patients in this step. If patients show minor emergency care need they will be forward to the minor emergency department “lättakuten”, if not patients will be forward to another desk and later on to the respective ED ward. If their clinical status deteriorates they will enter a fast line called “larm”, the same used by ambulance arriving patients and where patients meet the emergency team at once. Patients arriving by ambulance have their clinical information, including vital signs, transferred verbally to the ED team or as a monitor print out on paper that is delivered to the ED team upon arrival. Information received verbally is written on paper and later on documented in to the patients EHR. All vital signs measurements and documentation at the different ED wards is performed directly in to the patients EHRs without paper support. 54 Hospital Setting D In this setting patients arriving alone at the ED take a queuing number and wait for their turn. When their turn comes a primary team performs an outer triage that forwards patients to the correct ED ward. During night time this outer triage is not available and patients are forward directly to the ED wards. Vital signs may not be measured on all patients during outer triage. If patient´s clinical status deteriorates they will enter a fast line called “larm”, the same used by ambulance arriving patients and where patients meet the emergency team at once. Patients arriving by ambulance have their clinical information, including vital signs, transferred verbally to the ED team or as a monitor print out on paper that is delivered to the ED team upon arrival. Information received verbally is written on paper and later on documented in to the patients EHR. All vital signs measurements and documentation at the different ED wards is performed in to a specific ED paper form that is later on scanned in to the digital patient record. This scanning process is reported to take several days. 55 Appendix C- Mind map of data analysis code results. Themes, categories, subcategories and concepts obtained through analysis of this study interviews 56