This section provides a detailed account of the preparation for data collection for the quantitative study in the current research. There were two main issues to consider
regarding data collection for the quantitative study in phase one. The first was to establish a suitable source of the necessary data and the second was to find the best means of
investigating documentation of vital signs in the EHR. With these issues in mind, and guided by the literature review in Chapter 2, preliminary choices were made regarding the type of data to be collected, and a strategy for data collection was planned.
The rationale for decisions on data collection was based on four intentions. First, it was necessary to identify a source for a suitable cohort of patients in an acute hospital setting and to be able to access their electronic records. More specifically, it would be useful if the patients were from a group of patients whose condition was potentially at-risk of
deterioration, and therefore information on vital signs might need to be collected. Secondly, the literature had revealed many constellations of vital signs used in detecting patient deterioration and decisions had to be made on which would be the most
appropriate vital signs to examine. Thirdly, the way in which these observations were documented in an electronic record should be examined to get a clear picture of which signs were present, e.g., temperature, pulse and respiration, where each sign was documented, and how each vital sign was documented in the EHR. Finally, it was important to determine whether documented vital signs could reveal information about each patient's physiological condition and, therefore, the degree of risk of deterioration. The goals of the data collection are outlined in Table 4.1
Table 4.1 Goals of data collection
Goals
1 to identify a sample of patient records and sources of data 2 to decide which vital signs should be collected
3 to examine if, which, where and how vital signs were represented in electronic health records
4 to assess the documented vital signs to find out what they could reveal about a patient's risk of deterioration
These four goals are examined in detail in the next sections. 4.3.4.1 Identifying a sample and sources of data
To find a cohort of patients who had been at-risk of deterioration, patients who had previously suffered a cardiac arrest were considered a suitable group of patients to study. It is well-established that, prior to a cardiac arrest, the physiological status of patients
deteriorates (Bellomo et al., 2003; Bristow et al., 2000; Buist et al., 2002; Jones et al., 2005) and, in fact, deteriorates to such a degree that their hearts cease to function, which is when the cardiac arrest occurs. Such patients were therefore deemed to be a suitable cohort. Sources of data for a group of patients who had previously suffered cardiac arrest were also identified. In Sweden there is a register for patients who have suffered cardiac arrest in a hospital setting: the SRICA (Herlitz, 2009) The SRICA was initiated in 2005. It is an on- going national survey. Fifty-four of the 73 hospitals in Sweden (74%) have joined the SRICA (Herlitz, Aune, Claesson, & Svensson, 2010); the hospital in which this study was performed joined in 2007. This database registers all patients who have suffered an in- hospital cardiac arrest and on whom resuscitation attempts have been made, and was therefore an important source of information. In the SRICA, it was possible to select the registry entries from specific hospitals and therefore the study hospital could be selected to locate all of the patients from the study hospital who had suffered a cardiac arrest since 2007. The patients from the hospital were a sub-group of the entire population contained in the register. Therefore, the sample consisted of patients who had suffered a cardiac arrest in the study hospital and who could be located in the register. This would then be considered a case study sample of an entire set of patients (Polit & Beck, 2008) from the study hospital.
The SRICA was used to obtain information about the cardiac arrest and its outcome: i.e., the date and time of cardiac arrest, the cause and type of cardiac arrest, whether the patient survived to discharge, or the date and time of death. From the register, it was also possible to obtain the patient identification number, which could then be used to locate each patient's hospital records. In 2007, the study hospital implemented an EHR system, a few months after it began to register all in-hospital cardiac arrests. The patient identification number obtained from the register could be entered into the EHR system, Cambio Cosmic, and each patient's individual electronic records accessed. Access to these records made it possible to review patient status in the period prior to cardiac arrest retrospectively by examining their vital signs.
The patients in the Register included patients from both general ward care and higher levels of care such as cardiac intensive care and general intensive care. Although most studies on identifying clinical deterioration in the literature review had been carried out in general wards (Bright et al., 2004; McQuillan et al., 1998; Schein et al., 1990), in the present study a decision was made to include patients from both general wards and intensive care areas. The rationale for this was that all areas of the hospital had recently implemented electronic records, which were the focus of this study, and therefore should be examined. Patients were classified into two categories: those receiving care in general wards and those cared for in intensive care areas. Within intensive care areas there is a higher level of
observation and a higher nurse to patient ratio. Intensive care areas comprised the coronary care unit, the cardiac catheterisation laboratory and the intensive care unit. For the purpose of this study, other areas were classified as general ward care. In the event of a patient having had a cardiac arrest in other parts of the hospital, for example, in the X-Ray
department or out-patient department, this was classified as general ward care, considering the level of observation was expected to have been similar to that on a general ward. Thus, the study sample included patients from these two levels of care.
4.3.4.2 Investigating which vital signs to include in the data collection
Appropriateness and accuracy of the data collected will influence the validity of research and emphasises the importance of collecting the correct data to answer the research questions (Kumar, 2011). According to the review of the literature in Chapter 2, section 2.8.7, there were no studies detected which examined the 'completeness' of vital sign recording in the EHR and, hence, there was a need for this investigation. First of all, it was important to determine exactly what was implied by 'complete' vital signs. Once that was decided, it would be possible to compare the documentation of vital signs in the study hospital to that which was considered 'complete'.
In deciding what 'complete' vital sign recordings implied, universal recommendations for observation of patients in hospital were examined. A fundamental aim of patient
observation is to recognise changes in a patient's clinical condition and, in particular, to detect signs of deterioration (Bellomo et al., 2004; Buist et al., 2007). According to the UK National Institute for Health and Clinical Excellence (NICE 2007), minimum physiological observations should include heart rate, respiratory rate, systolic blood pressure, level of consciousness, oxygen saturation and temperature, and these should be monitored at least every 12 hours (NICE 2007). The same parameters were recommended by a group of
experts who convened to discuss how to identify and monitor patients at risk of
deterioration (DeVita et al., 2010). In the literature regarding detection of deterioration in a patient’s clinical condition, many systems were available. Although, the general features of each system were similar, there were some variations regarding which vital signs should be included, as well as varying threshold levels. Details of these vital signs are given in the literature review (Section 2.4).
The duration of the retrospective time period for the data collection was also considered and guided by the current literature on pre-emptive signs of cardiac arrest. Studies have shown that signs of deterioration are often present in the 24-hour period prior to cardiac arrest (Goldhill, White, et al., 1999); therefore, the presence or absence of each vital sign being recorded in the 24 hours prior to cardiac arrest was required for the study.
4.3.4.3 Examining where and how vital signs were represented in electronic records The study set out to investigate how vital signs were represented in the EHR; therefore, the location of each vital sign and how it was represented in the EHR was important for the study. Within the electronic patient record, there were three possible locations where vital signs could be documented. These were referred to as the patient journal, the report sheet and a template, and data from each of these locations were examined.
4.3.4.4 Assessing documented vital signs to find out what they revealed about the patient’s risk of deterioration
In addition to examining the completeness of vital signs, and how these were represented in the EHR, another consideration in planning the data collection was to find out what the documented vital signs could reveal about patient status and risk of deterioration. This could be achieved by checking each patient's vital signs in relation to currently available track and trigger systems (TTS) for recognising patients whose condition was deteriorating. At the outset of this study, it was difficult to discern which of the many TTS would be most suitable for analyses in the descriptive study. In order to ensure that the necessary information was collected, four systems were initially considered for possible standards for the analyses. One system, BAS (see below and table 4.2), was currently in use in the study hospital, so this was an obvious choice of TTS for inclusion in the study. The other three systems initially selected were early warning score (EWS) systems (Moon et al., 2011; Morgan et al., 1997; Prytherch et al., 2010), which are outlined below.
The system used in the study hospital, BAS, was a type of track and trigger system. BAS is an acronym for B, 'blodtryck' (systolic blood pressure), A, 'andnings frekvens' (respiratory rate) and S, 'saturation' (oxygen satuaration). It is a single parameter system, i.e., if any one (or more) of the vital signs is abnormal, according to the threshold values shown in Table 4.2, this could indicate patient deterioration. Thus, a systolic blood pressure of less than 90, a respiratory rate of more than 30 or an oxygen saturation of less than 90% would be considered warnings of patient deterioration.
Table 4.2 BAS 90-30-90 (Blodtryck, Andning, Saturation - system in use at study hospital)
Vital sign Threshold values
B Systolic blood pressure <90 mm/Hg
A Respiratory rate >30 breaths per minute
S Oxygen saturation <90%
The selection of the remaining TTS was guided by the current literature described in the literature review (Chapter 2). Although there were many systems, it was decided to begin with three possible models of EWS systems for the reasons explained below. The first was a modified early warning scores (MEWS) system derived by Moon et al. (2011) which appeared to be widely used in the UK (see table 2.7, Chapter 2). The second was ViEWS (see table 2.9, Chapter 2) and was selected because it represented the results of a study which claimed to be based on sound clinical data in a system that was validated, and
demonstrated a performance superior to all other published TTS (Prytherch et al., 2010). A third EWS option was the MEWS system used at Växjö County Hospital, Sweden and the University Hospital, Linköping, Sweden as it represented a EWS system which had been implemented in Sweden. Thus, all of the vital signs included in BAS and three EWS systems were included in the data collection to ensure that a wide range of systems was covered. A summary of the rationale guiding the data collection is given in Table 4.3.
Table 4.3 Rationale guiding data collection
Goal Rationale
1 To identify a sample and sources of data
The Swedish Register for In-hospital Cardiac Arrest identified patients who had been at risk
The register subsequently led to these patients' EHR
2 To decide which vital signs should be used in the data collection
The literature review guided the choice of vital signs
3 To examine whether, which, where and how vital signs were represented in electronic records
Within the EHR, the relevant information on the representation of vital signs could be found
4 To assess the documented vital signs to find out what (if anything) these could reveal about the patient's risk of deterioration
Four track and trigger systems were preliminarily selected to establish if patients showed signs of deterioration
4.3.4.5 Additional data
All potentially useful data should be collected at the time of data collection as this is preferable to collecting too little data and having to return to the data at a later date (Polit & Beck, 2008). To supplement the data described above, additional data that would be or might be needed for the analyses were collected. One source of additional data was from the SRICA itself, which provided important demographic information: the patient's age and gender, and the name of the department in which the patient was housed at the time of arrest. In addition, it provided information about the actual cardiac arrest: the date and time of the cardiac arrest, the initial cardiac rhythm, the cause of the arrest, whether the arrest was witnessed, the time to treatment, whether the patient was intubated during the
procedure, any medication given during the resuscitation procedure, if the patient survived the resuscitation procedure and if the patient survived to discharge. The admission and discharge cerebral performance categories were also included. Incorporating all of this information within the data collection ensured that a wide range of analysis options was possible. The other source of additional data was the EHR from which data on any interventions were included, e.g., if intravenous fluids were commenced because of low blood pressure. This was because it might be possible to analyse associations between
interventions and patient status. Also, data on each patient's latest ECG and prescribed cardiac medications were included with a view to future analyses.
4.3.4.6 Additional time period for data collection
As indicated in section 4.3.4.2, the appropriate time period for data collection of vital signs was the 24-hour period prior to cardiac arrest. In the event that any of these vital signs had not been documented in the 24-hour period prior to cardiac arrest, a further piece of data was collected: the last documented recording of the vital sign, for the period prior to the 24-hour period preceding cardiac arrest, stating when it was documented.
4.3.4.7 Summary of data collection planning
The records of all patients who had suffered a cardiac arrest in the study hospital between 2007 and 2011 were included. The study sample was obtained by identifying patients who had suffered an in-hospital cardiac arrest in the study hospital who were registered in the SRICA. Data were collected from the SRICA and each patient's EHR.
Documentation of vital signs was reviewed by collecting and studying all patients' vital signs documented in the 24-hour period prior to cardiac arrest. A wide range of vital signs, based on current systems for detecting patient deterioration was collected to enable analysis on the completeness of vital sign recording. These included the vital sign parameters recommended by NICE (2007), as well as the additional vital sign parameters used in the models described in the literature review. Thus, the data collection included the following vital signs: respiratory rate; systolic blood pressure; heart rate; temperature; oxygen saturation; conscious level and hourly urinary output. 'Complete' documentation of vital signs would imply that all of these vital signs were measured at least twice per day. These data provided a detailed account of the completeness of documentation of vital signs in the study hospital. To find out how this information was represented in the EHR, data were collected relating to the location and representation of each piece of data in the record. 4.3.4.8 Designing the data collection tool for the phase one study
The previous sections provided details of which data needed to be gathered to answer the research questions. Using this information, a data collection tool was developed. The data collection tool had four sections, as described below in Table 4.4.
Table 4.4 Sections in data collection tool
Sections in the data collection tool
1 Demographic information and details about the cardiac arrest 2 Overview of latest ECG and prescribed cardiac medications
3 Which vital signs, vital sign values and the location of each vital sign within the EHR
4 Possible clinical interventions, such as oxygen therapy, intravenous therapy, blood tests or medication
The data collection tool was created in MS Word and paper copies produced for hand- written data collection. An example of the final data collection tool can be viewed in Appendix II. In order to test the data collection tool, a pilot study was carried out. The pilot study is described in section 4.5.