- Top of page
- Sources of support
- Conflict of interest
Ethical approval for the study was granted by the local research ethics committee. The study setting was a single-site 700-bed general hospital situated in the South of England.
The study was an audit of nursing practice against an EWS protocol in the detection and initial management of deteriorating ward patients. Patient deterioration was identified using the surrogate marker of cardio-respiratory arrest (CRA), a marker that has been well used in other deterioration studies (Schein et al. 1990, Bedell et al. 1991, Hodgetts et al. 2002, Jones et al. 2013).
The inclusion criteria were all adult (18 years and above), nonobstetric inpatients who had suffered a CRA on a general ward during one year (1 January–31 December 2007). CRAs were identified through resuscitation department records, switchboard logs, CCO records and patient hospital records. CRA cases were coded to maintain patient anonymity.
Data were collected using a predesigned data collection pro forma. The patients' hospital records and charts were scrutinised for evidence of documentation and recording of the elements of the predefined protocol standards during the 12 hours prior to the CRA, deemed the deterioration phase. The 12-hour study phase was based on the maximum length of a ward nurse's' shift and the recommended maximum time that should be between patients' vital signs recordings (NICE 2007, DeVita et al. 2010). The demographic and event details of all included CRA cases were entered and analysed using a Microsoft® Excel database (Microsoft Excel. Redmond, Washington: Microsoft, 2003.).
The protocol included eleven standard elements that were derived from recommendations from the literature and national guidance and were identified as follows:
- Vital signs recorded within 4 hours.
- Vital signs recorded within 6 hours.
- *Vital signs recorded within 12 hours.
- Temperature recorded.
- *Heart rate recorded.
- *Blood pressure recorded.
- *Respiratory rate recorded.
- Oxygen saturation (SpO2) recorded.
- *EWS recorded.
- *EWS accurate.
- *Referral made according to EWS trigger threshold.
*Denotes essential practice standards.
Of the 11 elements, seven were considered essential and are identified by an asterisk. Data relating to these 11 elements were sought in the patients' records. Nursing practice was audited by documented evidence of adherence to the protocol elements as outlined above and was categorised as such:
- Good - All 11 elements of the protocol were met.
- Basic - <11 elements of the protocol were met, but including the seven essential elements*.
- Poor - Failure to meet the seven essential protocol elements*.
The three categories of protocol adherence by ward nurses were then statistically analysed against the predefined context elements relating to the CRA event using Pearson's chi-square test. Statistical significance was set at p ≤ 0·05. The aim was to explore which context elements may have an influence on the degree to which nurses adhered to the protocol.
While adherence to the protocol was under the control of the nurse, the context elements were not. The predetermined context elements emerged from the literature, hospital and departmental statistics and personal clinical experience.
Eight context elements were evaluated as follows:
- Admission route type: emergency or elective.
- patients' age: below 65 years or 65 years and above.
- Patient gender.
- Time of day of CRA: social: 07:00–20:59 hours, and unsocial hours.
- Day of week of CRA: week days, and weekends and bank holidays.
- Ward where CRA occurred.
- Speciality of care: medical, surgical or elderly care.
- Nursing staff levels: locally agreed recommended level, under the recommended level and over the recommended level.
- Top of page
- Sources of support
- Conflict of interest
During the study year, 214 adult CRAs on the general wards met the inclusion criteria, and the hospital records for these CRA cases were examined. The final analysis involved 123 (57·5%) CRA events involving 120 patients. Ninety-one (42·5%) cases were excluded due to no information about the CRA event, not a CRA when records examined, incomplete records, duplicate event record or records not found in storage.
The 123 CRA cases occurred on all 26 general wards in the hospital. The wards comprised 13 medical, eight surgical and five elderly care wards. The number of CRAs on the ward ranged from one CRA on five wards to 13 CRAs on two wards. The demographic and event details of the cardiac arrest cases are outlined in Table 1.
Table 1. Demographic and event summary of 123 cardiac arrest cases
| ||Number (%)|
|Emergency admission||117 (95·1)|
|Elderly care||9 (7·3)|
|Age 65 and above||97 (78·8)|
|Left hospital alive||8 (6·5)|
|Occurred during unsocial hours (21:00–06:59 hours)||56 (45·5)|
|Occurred on weekdays (excluding bank holidays)||88 (71·5)|
|Length of stay||1–73 days|
Nurses' adherence to the EWS protocol was graded according to the predetermined criteria and categorised as previously described.
The number of cases in each category is reported in Table 2.
Table 2. Ward nurses' adherence to early warning scoring protocol
|Standard of practice||Number (%)|
Twenty-five (20·3%) cases scored maximum for adherence to the protocol where all 11 elements were evident in the records. Just over half (50·4%) failed to reach the minimum standard of practice.
When the 11 elements were individually measured as shown in Fig. 2, the frequency and completeness of physiological observations were more consistently recorded than EWS recording, EWS accuracy and referring for more expert help.
Early warning scoring was completed in 103 (83·7%) cases, but nearly a quarter (24·3%) of these were inaccurate. Sixteen of the inaccuracies were scored below the trigger of 3 so would not have generated a referral, but 15 of these had actual EWS score of 3 or more, which should have triggered a referral. The rate and details regarding inaccuracies are outlined in Fig. 3. Of all 123 CRA cases, lack of observations (n = 3), lack of EWS recorded (n = 17) and inaccurate EWS calculations (n = 25) meant that 45 CRA cases (36·5%) had an ineffective recording of an EWS, which probably contributed to suboptimal referral decisions. The protocol for referring the patient for more expert help when the EWS trigger threshold was reached was not adhered to in 39% of the total number of CRA cases, due to the absence of EWS recording (n = 20) and not referring according to protocol (n = 28).
The three categories of nurses' adherence to the EWS protocol were evaluated against the predetermined context elements, and their statistical probabilities are shown in Table 3. Due to the very large majority (95·1%) of patients being emergency admissions, it was inappropriate to apply probability statistical analysis for this context element. Ward location was analysed against the degree to which nurses adhered to protocol to see whether there were any consistencies that might point to cultural influences, but due to the small numbers of CRAs on some wards, statistical analysis was not possible. Overall, there appeared to be no discernible patterns of EWS protocol adherence related to individual wards.
Table 3. Comparison of ward nurses' adherence to early warning scoring (EWS) protocol with context elements
|Context element||No.||Adherence to EWS protocol (total no. of cases = 123)||p-Value|
|Below 65 years||26||5||10||11||0·496|
|65 years and above||97||20||26||51|
|Time of day|
|Social (07:00–20:59 hours)||67||16||18||33||0·537|
|Unsocial (21:00–06:59 hours)||56||9||18||29|
|Day of week|
|Weekends and bank holidays||35||10||3||22|
|Staffing levels (staffing levels were determined using computerised rotas which were available in 83 cases, for the shift during and immediately preceding the cardio-respiratory arrest event)|
Of all the context elements analysed for statistical significance, only the day of the week showed that patients who suffer a CRA event at the weekend or bank holidays were more likely to have had a poorer adherence to the EWS protocol in the hours preceding their CRA.
- Top of page
- Sources of support
- Conflict of interest
The audit of the EWS protocol has shown that ward nurses' adherence to the minimum accepted standards of practice was achieved in less than half of the patients included in this study. Only one-fifth of patients received the optimum standard of practice outlined in the EWS protocol. As these patients represented some of the most acutely unwell in the hospital, the results are of some concern.
The frequency and completeness of observation recording were better when compared with previous findings (McBride et al. 2005, NPSA 2007a,b, Odell et al. 2007). This may be due to the increased interest in the field, the influence of national policy and patient safety initiatives over the last 10 years. The main issue of concern was the inaccuracies of early warning scores and, coupled with the small number of missing EWS, meant that over a third of cases had an ineffective EWS. These findings resonate with those from other studies (Hillman et al. 2005, NCEPOD 2005, NPSA 2007a), and it seems further research needs to be carried out with RRSs if we are to realise their full potential. The problems this study has shown with elements of the afferent phase, such as inaccuracies and absence of EWS, might be a significant reason that studies to date have failed to adequately show effectiveness of rapid response teams (Hillman et al. 2005).
Seventy-six cases had an appropriate referral decision, but of these, 15 were based on inaccurate calculations of EWS. Therefore, in 62 cases, the referral decision was not optimal. Referral decisions are not only influenced by the protocol elements identified for this study, but also affected by complex factors such as the nurses' confidence, knowledge, skill, communication and relationships with members of the wider healthcare team. In an Australian study using structured interviews with 91 staff, Shearer et al. (2012) identified a number of sociocultural factors that were barriers in the activation of a rapid response team. One of the main factors was that ward staff felt that they had the experience and skill to manage the patients themselves even though they recognised that the patients met the rapid response trigger criteria. It was speculated that ward staff may feel that rapid response trigger criteria lack the necessary sensitivity and specificity, as few patients who triggered had a subsequent adverse event (Shearer et al. 2012).
In hindsight, the implementation of EWS systems and rapid response teams might have been an oversimplified solution to a very complex and multifaceted problem, and as pointed out by Hands et al. (2013), future work is needed to help us understand the local social, cultural and professional issues that influence individual practice.
Of all the context elements that were tested for their influence on adherence to the EWS protocol, the only one with a significant statistical association was those CRA cases that occurred during weekends and bank holidays. All the context elements tested in this study have been suggested as contributory factors to suboptimal care in other studies, and by investigating them in more depth, we can understand more about what might influence nurses' adherence to defined standards of care. While staffing levels may not appear to have an effect on the standard of nursing practice in this study, the weekend influence may be due to the presence of less senior nursing staff and the use of more bank and agency staff. While the numbers of nursing staff are unaffected, the experience and expertise of the staff may be diluted at those times. Other weekend effects may be lower numbers, or less experienced medical staff, as well as reduced availability of other support services. The weekend effect has been reported in other studies: an increased risk of death associated with emergency admission at the weekend (Aylin et al. 2010, Freemantle et al. 2012) and increased mortality for patients having elective surgery towards the end of the week and at weekends (Aylin et al. 2013). More research into the weekend effect would be of benefit.
There were particular challenges in the audit process. The considerable number of missing records and data within those records was a drawback to the completeness of data analysis. The problems with storage, care and retrieval of patient records are common and are unlikely to be restricted just to the study hospital. The advent of the electronic patient record is likely to make a positive impact on this problem, but the quality of data entry and record keeping will continue to depend on the professionalism, skill and the meticulousness of healthcare staff. While it might be argued that documentary evidence may not reflect practice, the documentation of the identified elements was in itself a measure of adherence to the defined standard of care. The use of records can minimise the inherent bias that might be evident in questionnaires and interview methods.
The use of CRAs as a surrogate marker of deterioration was necessary in the absence of a satisfactory objective definition of deterioration. Some of the CRAs could have been sudden events with no previous physiological deterioration signposted in the patients' observations. However, the focus of the study and the data that were analysed were not dependent on whether the patient had signs of deterioration, as the standard outlined in the EWS protocol should be applied to all acute inpatients. It could be argued that some of the patients should have had treatment limitation decisions, and this may have affected nursing practice decisions for that patient (either positively or negatively). This was not included in study remit but might be a useful focus for further research.
The choice of standards for the EWS protocol in the detection and immediate management of the deteriorating patient is open to challenge, but were developed from the literature and national recommendations. The standards and audit tool can be used by others to measure adherence to early warning systems protocols and can be added to and updated as new practice evidence becomes evident.
Lastly, the data for this study were collected in 2007, and the data may be considered out of date. However, to date, there have been very few published studies detected in this area. Two recent literature reviews on RRSs and vital signs monitoring (Massey et al. 2010, Kyriacos et al. 2011) have highlighted the paucity of research in this area and the need for more research on vital signs recording on general wards and the effective use of RRSs. Hands et al. (2013) reviewed patterns of vital signs recording on a large electronic system and assessed compliance with the escalation protocols using data collected during 2010/2011 and reported similar findings to the study reported here. This demonstrates the ongoing and widespread problem of inconsistent observation recording and lack of risk score documentation that were highlighted in the recently published enquiry into the Mid-Staffordshire hospital (Department of Health 2013). The report related to patient care issues between 2005–2009.
In reviewing the literature from the last 30 years, it has become apparent that changes in practice take many years to embed into culture, and the findings in the study by Hands et al. (2013) demonstrate that it is unlikely that there have been major changes in the intervening time that will make the findings from this study out of date.
- Top of page
- Sources of support
- Conflict of interest
The aims of the study were to audit nursing practice against a standard of care in the monitoring and use of an EWS system protocol and examine possible influences of predetermined contextual factors on that practice.
The findings suggest that ward nurses' monitoring of patients' vital signs has improved compared with earlier research, but errors in EWS and nonadherence to EWS protocols were still a problem.
In seeking factors that might influence nursing practice decisions in adhering to EWS protocols, a number of contextual elements were tested for their effect, and the only factor found to have a significant association with the quality of nursing practice was in those CRA events that occurred during weekends or bank holidays. Auditing nursing practice from documentary evidence adds to the knowledge gained from nursing interviews and questionnaires in areas such as ward culture, poor staffing numbers and poor night-time care. Issues to do with nurses' skills and knowledge, communication and relationships with other healthcare personnel have all been highlighted in the literature. Further research into the afferent phase of the rapid response system would provide valuable insight into how we might shape future system development, improve clinical practice and positively affect patient experience and outcome.