SEPSIS Management in Scotland

SEPSIS Management in Scotland A Report by the Scottish Trauma Audit Group November 2010 STAG

NHS National Services Scotland/Crown Copyright 2010 Brief extracts from this publication may be reproduced provided the source is fully acknowledged. Proposals for reproduction of large extracts should be addressed to: ISD Scotland Publications Information Services Division NHS National Services Scotland Gyle Square 1 South Gyle Crescent Edinburgh EH12 9EB Tel: +44 (0)131 275 6233 Email: nss.isd-publications@nhs.net Designed and typeset by: ISD Scotland Publications This publication can be made available in different languages, larger print or Braille (English only). For information on translation of this publication in your community language please phone the number below.

Contents Foreword Introduction Summary and Key Findings Methodology ii iii iv v Section 1: Demographics 1 Section 2: Patient Journey 6 Section 3: Early Warning Systems (EWS) 14 Section 4: Detection of Sepsis 19 Section 5: Severe Sepsis and Septic Shock 22 Section 6: Severe Sepsis in the Emergency Department 29 Appendix 1: Sample Representativeness 35 Appendix 2: College of Emergency Medicine Standards for Severe Sepsis and Septic Shock Management 36 Appendix 3: Proforma 37 Appendix 4: Steering Group 2009/10 40 Appendix 5: Local Audit Teams 2009/10 41 References and Further Reading 42 Acknowledgements 42 Contact details 43 i

Foreword The Scottish Trauma Audit Group (STAG) was established in 1992 and is one of the outstanding examples of how to carry out high quality clinical audit. Its purpose is to improve the quality of care for emergency patients through measurement, comparison and clinical feedback. It is a nationally funded audit through Information Services Division (ISD). The initial 10 year audit into trauma outcomes ran from 1992-2002 and was associated with multiple audit cycles, publications and improvements in the quality of care for injured patients in Scotland. This work was successful both in confirming what constituted quality care and in driving service provision in a direction that benefited patients. Following completion of the trauma audit, STAG reviewed the management of patients who present to the emergency care setting with a range of alcohol problems. This successful project was aligned with public health and government priorities and provided important information for strategic planning. This is an exciting time for patient care. The Cabinet Secretary for Health and Wellbeing launched the Quality Strategy in May 2010, concentrating on the themes of safety, effectiveness and patient-centred care set in the context of the six pillars of quality. Associated with this strategy there has been an enormous effort to find clinical indicators that measure quality care. These include Surgical Profiles, Medical Profiles and the Hospital Standardised Mortality Rates (HSMR) project. However the challenge of defining and measuring quality in an emergency care setting has not yet been resolved. Over the last two years, the work of STAG has been focused on auditing the care of emergency patients with sepsis. The sepsis project represents the first national review of systems and processes involved in the response to patients with this specific medical emergency by assessing the patient journey. The multiple treatment modalities for sepsis, and the complex way in which the healthcare system responds to such patients, made this review extremely challenging. A total of 27,046 patients required case note review in order to determine the presence of Sepsis criteria in 5285 patients. It is through the commitment and collaboration of the whole STAG team including the Local Audit Coordinators, Local Audit Clinical Leads and the STAG team at ISD, that such a large volume of clinical information has been gathered, analysed and reported. This report is the culmination of that work. Through this, clinical feedback will be given on the current emergency care provided to patients with sepsis, clarifying areas of good practice and areas where further improvement work should be targeted. In addition, the experience of solving the practical problems of this complex audit will prove invaluable in planning further audit of medical emergencies which will need to be assessed as the Quality Strategy gathers momentum. Crawford McGuffie, Chairman ii

Introduction Sepsis, severe sepsis and septic shock are complex healthcare problems. The associated morbidity and mortality remains high and poses a significant burden both to NHS resources and society in general. Awareness and understanding have been raised by the Surviving Sepsis Campaign (SSC) which ran from 2002-2008. The SSC was a performance improvement initiative that provided greater direction for clinicians around illness recognition, diagnostic certainty, treatment goals and treatment standardisation. Although the optimal strategy for treatment is constantly evolving, there is an accumulating evidence base around fluid resuscitation, rapid diagnosis, antibiotic administration, source identification and both resuscitation and management bundles covering the first six hours and 24 hours of care respectively 1,2. The ultimate aim of STAG through this work is to improve the quality of care and outcomes for emergency patients with sepsis through effective use of audit data. The STAG team identified patients who had signs of sepsis within two days of emergency attendance to participating acute hospital sites in Scotland, over a three month period in 2009. These patient journeys were then audited to review healthcare processes, interventions and outcomes. This information represents a comprehensive review of Emergency Department (ED) and admission unit attendances with potential sepsis over the index period. It clearly identifies healthcare responses of good practice and also areas where efforts for improvement could be targeted. The report contains information on 308,910 emergency attendances, patients with sepsis and patients who met the criteria for severe sepsis. The format of the report is aligned with the patient journey which it follows. Sections 1-4 deal with patients who met the criteria for the diagnosis of sepsis within two days of initial attendance and reports on the associated healthcare responses. Although the associated infection is likely to have been community acquired, the manifestation of the signs of sepsis could have presented at any time during the first two days following attendance. Section 5 deals with patients who met the criteria for severe sepsis within two days of initial attendance. Each patient journey was followed from inclusion in order to assess the specific treatments and subsequent responses to treatment. All patients were followed to discharge or death, but the majority of data contained in this report are from the first two days following initial attendance. Section 6 reports on patients who met the criteria for severe sepsis before leaving the ED and reviews the care given. iii

Summary and Key Findings Over a three month period, 5285 patients (1.7% of new patient attendances) developed the criteria for sepsis within two days of initial attendance. At the participating hospitals, this is likely to equate to an annual incidence of approximately 21,000 unscheduled care patients who have signs of sepsis within two days of attendance. The median duration of stay in hospital was seven days for patients with sepsis. Most of these patients were looked after within a ward setting. 14% of patients with sepsis were admitted to a critical care facility (High Dependency Unit/Coronary Care Unit or Intensive Care Unit) within two days of admission. The median length of stay for patients within a critical care facility was three days. Using estimated bed-day costs, the total annual cost of care for emergency patients with sepsis within NHS Scotland is likely to be in excess of 79 million. Attendances with sepsis were more common among females (53%) and older patients. Documentation suggests that the vast majority of patients were managed initially by doctors in training. Early Warning System (EWS) charts were commenced on 71% of patients within two hours of initial attendance. When the EWS chart indicated the need for review, documented confirmation that this had occurred was present in 91. The lack of standardisation of EWS charts should be addressed to enable meaningful comparison between units. The most commonly documented sources of infection were respiratory or urinary tract which together occurred in 73% of patients. At the point of inclusion in the audit, sepsis patients typically had two or three physiological derangements. Of the immediately available physiological parameters, heart rate, respiratory rate and temperature were the most frequently deranged in the septic patient. If a standardised EWS were to be introduced, it should be a sufficiently sensitive tool to predict the development of serious illness or deterioration in patients. These three physiological parameters should be included and weighted in a standardised manner. Systolic blood pressure was noted to be deranged in only 10% of patients at the point of inclusion in the audit. Clinical reliance on the development of hypotension prior to the consideration of potential sepsis may delay diagnosis and treatment, and, potentially worsen the outcome for patients. Overall, 34% (1325) of patients with sepsis met the criteria for severe sepsis within two days of initial attendance. Of these patients 48% (637) met the criteria before leaving the ED. Reinforcement of good practice in aspects of illness recognition, fluid resuscitation, antibiotic administration, prior blood cultures, critical care assessment and resuscitation bundle compliance in the ED is likely to improve the outcome for patients with severe sepsis 2. The overall mortality for patients who met the criteria for sepsis within two days of initial attendance was 14%. For patients who met the criteria for severe sepsis within two days of initial attendance mortality was 24%. iv

Methodology Data were collected by dedicated Local Audit Coordinators in each of the 20 participating sites in Scotland (Appendix 5). The hospitals consisted of both district general and teaching hospitals. All hospitals in mainland Scotland with an Emergency Department (ED) were eligible to contribute, but five hospitals did not have the required resources to participate. The data collection period had both a prospective and a retrospective phase. nd Patients who attended hospital as an emergency, from the community between 2 March and the 31 st May 2009 were identified prospectively from ED or admissions unit records. This process identified 308,910 initial patient attendances. Hospital information systems were then interrogated to ascertain whether the patient had an inpatient stay of at least two days. Patients who died within the first two days, and therefore may have been omitted from data collection, were identified retrospectively by the Local Audit Coordinators using General Register Office Scotland records and eligible patients were included for analysis. Local Audit Coordinators were able to exclude, at this stage, those patients who had an obviously non-infective cause for attendance such as acute cardiac ischemia, trauma or stroke. All patients aged under 17 years were excluded. This process identified 27,046 patients who required case note review in order to determine the presence of Sepsis criteria. The criteria used to determine the presence of sepsis were: within the first two days from attendance there was a suspicion or confirmation of infection. and within the first two days from attendance the patient had two physiological derangements at the same time. The physiological derangements (taken from the SSC guidelines) were defined as: temperature > 38.3 C or < 36 C heart rate > 90bpm, respiratory rate > 20/min, white cell count of > 12000/µL or < 4000/µL or >10% immature forms. acutely altered mental status systolic blood pressure < 90mmHg blood glucose > 7.7mmol/L in the absence of diabetes Data were collected retrospectively on the entire patient journey, from initial attendance to discharge from acute care, using a standardised audit proforma. All patients were followed to discharge or death, but the majority of data contained in this report are from the first two days following initial attendance. A total of 5285 patients fulfilled the entry criteria. The difficulties encountered in obtaining and extracting data from casenotes (27,046 sets of casenotes were reviewed) were such that a pragmatic decision was taken to stop data collection in July 2010. This meant 1395 (26%) patients did not have full data collection and so were not available for analysis. Full data collection was completed for 3890 (74%) cases. v

In order to assess whether attendances for which proforma were completed (n = 3890) were representative of all cases that met the sepsis criteria (n = 5285), a number of key demographic and patient journey variables were collected for all attendances where it was not possible to complete a full proforma (Appendix 1). It should be noted that some patients attended hospital and met the sepsis inclusion criteria more than once during the audit period. The 3890 cases for which proforma were available represent 3733 individual patients. Throughout this report data were analysed by patient attendances (n = 3890). However, for the purpose of gender and age profile analysis (Section 1), each patient was only included once (n = 3733). The term sepsis is used to define the entire spectrum of sepsis, severe sepsis and septic shock in Sections 1-4. In Sections 5 and 6, details of the management of patients with severe sepsis and or septic shock are described. Means and medians are used at different points in the report as measures of central tendency within a data range. A mean is the mathematical average (i.e. sum of the values divided by the number of items in the data range). This is an exact measure of central tendency, but can be unsuitable in data ranges where there are outliers which may skew the results. The median on the other hand, is simply the point at which, if values in the data range were to be sorted from high to low (or vice versa), the middle point would lie. While this avoids the distortion problems encountered using the mean, it is not an exact measure and may not reflect clustered values either side of the middle point in the data range. To overcome these issues, both measures are sometimes quoted within this report. Where median values are reported, the inter-quartile range (IQR) is also given. The IQR represents the data range within which the middle 50% of values lie. Where a mean value has been reported the standard deviation (St. Dev.) of the mean is also given. The standard deviation measures the spread of the data about the mean value. vi

Section 1: Demographics Summary Overall hospital emergency attendances were recorded over the index three month period. All 20 sites which contributed to the audit had an Emergency Department (ED). At 11 of the contributing sites, the route of admission to hospital was either via an ED or directly to an admission unit. A total of 308,910 patients attended the participating hospitals over this period with a fairly even distribution over the three months (Figure 1.1). Of the total attendances, 27,046 patients were identified for case note review. Preliminary analysis identified 5285 patients who met the inclusion criteria. The proportion of patients who developed sepsis within two days of attendance is illustrated in Table 1.1. At the end of the audit, complete data were available for 74% of total cases (Figure 1.2). Difficulties in data collection were related to the complexity of the information, records storage systems and availability of case notes. Analysis of baseline characteristics of the patients with incomplete data is shown in Appendix 1. Attendances with sepsis were more common among females (53%) and in patients in their seventies and eighties (Figure 1.3). 1

Figure 1.1: What were the total initial attendances at each hospital* during the audit period (March 2 nd - May 31 st 2009)? Hospital (number of attendances) Perth (7,819) Dumfries & Galloway (9,495) Inverclyde (10,015) Raigmore (10,449) Ayr (10,989) Queen Margaret (12,027) Victoria, Kirkcaldy (12,397) Southern General (12,756) Western Infirmary (15,615) Hairmyres (16,049) Ninewells (16,053) Stirling (16,461) Crosshouse (16,944) Victoria, Glasgow (17,600) Royal Alexandra (18,501) Wishaw (19,067) Glasgow Royal (19,280) Monklands (19,334) Aberdeen (19,754) March (101,572) April (101,800) May (105,538) Royal Infirmary, Edinburgh (28,305) 0 5,000 10,000 15,000 20,000 25,000 30,000 Initial attendance * For full hospital names see Appendix 5. A total of 308,910 patients attended hospital as an emergency between 2 nd March and 31 st May 2009. Of these, 33% attended in March, 33% in April and 34% in May. 2

Table 1.1: What proportion of initial attendances were patients who developed sepsis within two days of presenting to hospital? Hospital Initial attendance Sepsis case Non-sepsis case Unknown * N o. cases % cases N o. cases % cases N o. cases % cases Aberdeen 19,754 331 1.7 19,310 97.8 113 0.6 Ayr 10,989 289 2.6 10,671 97.1 29 0.3 Crosshouse 16,944 369 2.2 16,569 97.8 6 0.0 Dumfries & Galloway 9,495 69 0.7 9,426 99.3 0 0.0 Glasgow Royal 19,280 512 2.7 18,673 96.9 95 0.5 Hairmyres 16,049 121 0.8 15,904 99.1 24 0.1 Inverclyde 10,015 131 1.3 9,869 98.5 15 0.1 Monklands 19,334 182 0.9 19,085 98.7 67 0.3 Ninewells 16,053 409 2.5 15,568 97.0 76 0.5 Perth 7,819 168 2.1 7,634 97.6 17 0.2 Queen Margaret 12,027 99 0.8 11,907 99.0 21 0.2 Raigmore 10,449 274 2.6 10,129 96.9 46 0.4 Royal Alexandra 18,501 184 1.0 18,285 98.8 32 0.2 Royal Infirmary, Edinburgh 28,305 487 1.7 27,800 98.2 18 0.1 Southern General 12,756 341 2.7 12,370 97.0 45 0.4 Stirling 16,461 139 0.8 16,322 99.2 0 0.0 Victoria, Glasgow 17,600 401 2.3 17,107 97.2 92 0.5 Victoria, Kirkcaldy 12,397 164 1.3 12,216 98.5 17 0.1 Western Infirmary 15,615 484 3.1 15,091 96.6 40 0.3 Wishaw 19,067 131 0.7 18,936 99.3 0 0.0 Total 308,910 5,285 1.7 302,872 98.0 753 0.2 * It was not possible to determine if these patients developed sepsis within two days of attendance as the necessary information was not available (e.g. case notes unavailable, poor documentation). The proportion of patients who developed sepsis within two days of attendance at each hospital ranged from 0.7% to 3.1%, with a national average of 1.7%. Some of the variation amongst hospitals may have been due to differences in local case mix and/or due to the complexity of data required. Some sites were not able to submit a proforma for all of their sepsis cases before the end of the audit (Appendix 1). 3

Figure 1.2: What proportion of sepsis cases were complete data collected for? 80% 70% 60% 50% 40% 30% 20% 10% 0% Data collection not completed Data collection completed (1395) (3890) Data collection completed (number of cases) Full data collection was completed for 3890 (74%) of sepsis cases. Variations in local practices for storing and accessing medical records and the resource available for data collection were seminal issues. 4

Figure 1.3: What was the age and gender distribution of patients* who presented with sepsis? 30% 25% Female (1980) Male (1753) % of total patients 20% 15% 10% 5% 0% < 20 20-29 30-39 40-49 50-59 60-69 70-79 80-89 > 89 (42) (135) (174) (286) (389) (672) (969) (868) (198) Age group (number of patients) * There were 3733 individual patients included within the audit. In all other sections of this report, data were analysed by patient attendances (n=3890). For the purpose of gender and age profile analysis, those patients with multiple attendances were included only once. A total of 3733 individual patients were included in this audit. The majority (3595) only presented with sepsis once during the audit period. 138 (3.7%) patients presented more than once. The maximum number of times that a patient presented with sepsis was four. Of the total patients admitted with sepsis, 47% were male and 53% female. The median age of all patients was 71 years (IQR = 58-81) and there was no notable difference in age between males (median = 71, IQR = 58-79) and females (median = 72, IQR = 58-82). 5

Section 2: Patient Journey Summary A total of 73% of patients in the group arrived at hospital by ambulance (Figure 2.1). Patients who had not been pre-assessed by their GP presented to the ED by emergency ambulance, car or other means of self-presentation. Patients who had been pre assessed by their GP presented to the ED or went to an Admissions Unit utilising the same modes of arrival. Some hospitals had a policy that all patients went to the ED prior to admission even if they had been GP pre-assessed. The initial area of attendance is illustrated in Figure 2.2. In the initial area of attendance, a consultant was documented as being present in 9 (Figure 2.3). This may have been underestimated as a result of poor documentation. It would appear that the majority of patients in the audit were initially assessed and managed by doctors in training in both the admissions unit and in the ED. It is worth noting that 8% of patients were assessed by a nurse only. This is likely to be related to patients who had been GP pre-assessed and admitted via the ED and deemed as not requiring a medical intervention prior to admission to the admissions unit. The added value associated with this step in the patient journey (i.e. via the ED) could be questioned. The proportion of cases initially managed by doctors in training was greater in the out of hours period from 20.00 07.59 (Figure 2.4). It is important to note that patients in the audit developed sepsis within the first two days from attendance and therefore may not have been exhibiting signs at initial attendance. The progression of the patient journey both in and out of hours is detailed in Figure 2.5, indicating the destination from the initial area of attendance. In 8, patients were admitted directly to a critical care area (HDU/CCU or ICU). Figure 2.6 shows that critical care was the highest level of care achieved for 14% of patients in the first two days. Patients were more likely to be admitted to a critical care facility within two days of attendance if the initial area of attendance was the ED. The length of time taken to reach the critical care facility was less if the initial area of attendance was an ED (Figure 2.7). The mean and median length of stay within critical care was five (St. dev = 8) and three (IQR = 2-6) days respectively. Maximum length of stay was 102 days. Figure 2.8 illustrates the profile of length of stay within critical care and greater detail around the duration of patient stay within HDU/CCU and ICU is given in Table 2.1. In-hospital stay had a range of one to 231 days within a mean and median of 12 (St. dev = 16) and seven (IQR = 4-14) days respectively. The length of in-patient stay profile is illustrated in Figure 2.9. Patients with sepsis spent a notable amount of time in hospital both within critical care and in the ward setting. Sepsis placed a considerable demand on resources per individual patient both in critical care and for the hospital-wide system. The overall mortality for patients who developed sepsis within two days of attendance was 14%. 6

Figure 2.1: What was the most common mode of arrival for patients who developed sepsis? 80% 70% 60% 50% 40% 30% 20% 10% 0% Self Ambulance Air Unknown (735) (2833) (6) (316) Mode of arrival (number of cases) Figure 2.2: What proportion of patients presented to an Emergency Department and what proportion to an Admissions Unit? 100% 80% 60% 40% 20% 0% Emergency Dept. Admissions Unit (3271) (619) Initial area of attendance (number of cases) At 11 of the contributing sites, the route of admission to hospital was either via an ED or directly to an admissions unit; in the remaining nine all emergency patients were admitted via the ED. As a result, 619 (16%) of sepsis cases were admitted directly to an admissions unit with the remaining 3271 (84%) admitted through the ED, regardless of the source of referral. 7

Figure 2.3: What was the grade of the most senior doctor documented as present on attendance? Most senior doctor (number of cases) Consultant (357) ST / FTSA* (2041) Foundation Dr (955) Other career grade (120) Grade unknown (119) None: Nurse only (298) 0% 10% 20% 30% 40% 50% 60% * Specialist Trainee / Fixed Term Specialty Training Appointment For patients whose initial area of attendance was an admissions unit, the most senior doctor documented as present was recorded within the first six hours following attendance. Figure 2.4: What was the grade of the most senior doctor documented as present on attendance and did this vary according to time of attendance*? Most senior doctor (number of cases) Consultant (320) (37) ST / FTSA (1342) (698) Foundation Dr (613) (340) Other career grade (92) (28) Grade unknown (86) (32) None: Nurse only (243) (55) 08:00-19:59 (2696) 20:00-07:59 (1190) 0% 10% 20% 30% 40% 50% 60% attending in each time period * In four cases the time of attendance was unknown. 8

Figure 2.5 What was the most common destination after the initial area of attendance and did this vary according to time of attendance*? Destination (number of cases) Mortuary (27) (5) ICU (44) (17) HDU / CCU (162) (96) Short stay ward (304) (193) Ward (2095) (851) Other destination (29) (11) 08:00-19:59 (2696) 20:00-07:59 (1190) Home / Irregular discharge (35) (17) 0% 10% 20% 30% 40% 50% 60% 70% 80% attending in each time period * In four cases the time of attendance was unknown. Intensive care Unit High Dependency Unit/Coronary Care Unit The vast majority of the short stay ward cases were from two hospitals (70% from Royal Infirmary, Edinburgh and 30% from the Western Infirmary, Glasgow) Other destinations include radiology, percutaneous coronary intervention, theatre and other hospital. 9

Figure 2.6: What proportion of patients were admitted to Critical Care within two days of attendance and did this vary depending on the initial area of attendance? Crit. Care within two days (561) No Crit. Care within two days (3328) 100% 80% 60% 40% 20% 0% Emergency Dept. Admissions Unit* Overall* (3271) (618) (3889) Initial area of attendance (number of cases) * In one Admissions Unit case it was unknown whether the patient went to Critical Care within two days of attendance. 561 (14%) patients included in the audit were admitted to a Critical Care bed within two days of attendance. 10

Figure 2.7: How long did it take patients who were admitted to Critical Care within two days to get there and did this vary depending on the initial area of attendance? (a) Emergency Dept. patients admitted to Critical Care within two days (501 cases) 60% 50% 40% 30% 20% 10% 0% < 6 hrs 6 - < 12 hrs 12 - < 24 hrs 24-48 hrs > 48 hrs Within two days, but timing unknown (285) (79) (54) (52) (21) (10) Hours from initial attendance (number of cases) (b) Admissions Unit patients admitted to Critical Care within two days (60 cases) 60% 50% 40% 30% 20% 10% 0% < 6 hrs 6 - < 12 hrs 12 - < 24 hrs 24-48 hrs > 48 hrs Within two days, but timing unknown (18) (13) (16) (8) (2) (3) Hours from initial attendance (number of cases) 11

Figure 2.8: How many days did Critical Care patients (561 cases) spend in Critical Care? 35% 30% 25% 20% 15% 10% 5% 0% 1 day (or part of) 2-3 days 4-7 days > 7 days (116) (187) (165) (93) Days in Critical Care (number of cases) Table 2.1: Where a patient was admitted to Critical Care within two days of attendance (561 cases), what was the breakdown of the length of stay (in days) spent in HDU/ CCU, ICU and overall? HDU/CCU* ICU * Critical Care* Cases 465 183 561 Total days 1565 1457 2992 Median days 3 4 3 Inter-quartile range 1-4 2-9 2-6 Minimum days 1 1 1 Maximum days 28 102 102 Mean days 3 8 5 Standard Deviation 3 12 8 * Patients stays of one day (or part of) were counted as one day. 87 patients went to a High Dependency Unit/Coronary Care Unit and an Intensive Care Unit. Where a patient first attended both levels of care on the same day (n=30), this has been counted as one Critical Care day. 12

Figure 2.9: How many days did sepsis patients spend in hospital? 60% 50% 40% 30% 20% 10% 0% 1 day (or part of) 2-3 days 4-7 days > 7 days (120) (473) (1375) (1922) Days in hospital (number of cases) 49% of patients included in the audit had a hospital stay of more than seven days. 13

Section 3: Early Warning Systems (EWS) Summary The Scottish Patient Safety Programme (SPSP) recommends the use of EWS. These scoring systems allocate a numerical value based upon the level of physiological derangement present in a patient. Dependent on that numerical value, advice should be sought from senior nursing and / or medical staff. One of the aims of the sepsis audit was to document and report on compliance with EWS, in particular escalation of care to senior staff. This aim was in part confounded by the absence of a national standardised EWS. EWS scoring may play an important role in reducing inter-operator variation and alerting healthcare professionals that action is required to prevent further patient deterioration. EWS scoring charts were used at some point in the patient journey on 97% of occasions (Figure 3.1). In cases where charts were used, they were initiated within two hours of attendance in 73% of patients (Figure 3.2). Compliance with local guidance on chart usage was achieved in 46 (Figure 3.3). There were different chart systems used by Health Board areas, hospitals in the same Health Board area and even within the same hospital which may have contributed adversely to compliance. Following the use of EWS charts, there was an identified need for review in 51 (Figure 3.4). Documented confirmation within the notes that this action had been taken was present in 91 (Figure 3.5). One of the recommendations arising from the audit is that there should be a national standardised Early Warning System in place across Scotland. 14

Figure 3.1: How often was an EWS chart used? 100% 80% 60% 40% 20% 0% Chart not used / not documented as used Chart used (127) (3763) EWS chart used (number of cases) In 97 an EWS chart was commenced at some stage in the patient journey. Figure 3.2: Where an EWS chart was used (3763 of 3890 cases), how long after attendance was a first total score available? 70% 60% 50% 40% 30% 20% 10% 0% < 1 hrs 1 - < 2 hrs 2 - < 4 hrs 4 - < 6 hrs 6-48 hrs > 48 hrs Date/times missing (2589) (170) (434) (180) (112) (3) (275) Hours after attendance (number of cases) In 69 where a chart was used, the first total score was available within the first hour of attendance and 73% within two hours. The recommended target for the time of first completion may differ between Health Board areas. 15

Figure 3.3: Where an EWS chart was used (3763 of 3890 cases), how often was it completed in compliance with local policy? 60% 50% 40% 30% 20% 10% 0% Non-compliant Compliant Unknown (2016) (1719) (28) Compliance with local policy guidelines (number of cases) In 46 charts were completed in line with local policy. The main factors contributing to non-compliance were the failure to complete a total score and failure to complete all categories. Standardisation of EWS charts across acute hospital sites had not been achieved. 16

Figure 3.4: Where an EWS chart was used (3763 of 3890 cases), how many patients had physiology which suggested the need for review? 60% 50% 40% 30% 20% 10% 0% Review not required Review required Unknown (1820) (1915) (28) Physiology suggested need for review (number of cases) Where an EWS chart had been used, 51% of patients demonstrated physiological changes which required review. The need for review was based on local policy. There was variation in the weight given to physiological values recorded in different Health Boards, which resulted in different tracking and trigger outcomes. Direct comparisons between outcome scores were therefore not possible. 17

Figure 3.5: Where patient physiology suggested the need for review (1915 of 3890 cases), how often was action taken? 100% 80% 60% 40% 20% 0% Action not taken / not documented as taken Action taken (178) (1737) Action taken (number of cases) Action was documented as having been taken in 91 where a patient s physiology suggested the need for a review. 18

Section 4: Detection of Sepsis Summary The Surviving Sepsis Campaign (SSC) guidelines 2 defined sepsis as infection, documented or suspected and the presence of some of a number of general variables indicating physiological derangement. For the purpose of the STAG Sepsis audit, patients were included in the audit if, within the first two days of attendance, there were two or more of the following variables present at the same time and confirmed or suspected infection: Signs of Systemic Inflammatory Response Syndrome (SIRS): temperature >38.3 C or < 36 C heart rate > 90bpm respiratory rate > 20/min white cell count of >12000/µL or < 4000/µL or >10% immature forms Additional criteria: acutely altered mental status systolic blood pressure <90mmHg blood glucose >7.7mmol/L in the absence of diabetes The most commonly documented sources of infection in patients with sepsis are shown in Figure 4.1 and were respiratory or urinary tract, which together accounted for 73. At the point of inclusion in the audit, patients had either two or three physiological derangements on 77% of occasions (Table 4.1). The proportion of patients who had each physiological derangement, noted at the point of inclusion in the audit, is shown in Figure 4.2. Of the immediately available physiological parameters, heart rate (81%), respiratory rate (58%) and temperature (44%) were noted to be deranged. If a standardised EWS were to be introduced, which is a sufficiently sensitive tool to predict the development of serious illness or deterioration in patients, then these three physiological parameters should be included and appropriately weighted in a standardised manner. Systolic blood pressure was the least common derangement noted (10%) at the point of inclusion in the audit. Clinical reliance on the development of hypotension prior to the consideration of potential sepsis may delay the diagnosis and treatment and potentially worsen the outcome for patients. 19

Figure 4.1: What was the most common source of infection documented (based on the first infection recorded during patient s stay)? Type of infection (number of cases) Skin / soft tissue (223) Acute abdominal infection (385) Urinary tract infection (598) Respiratory infection (2246) Other infection* (129) Unknown (309) 0% 10% 20% 30% 40% 50% 60% * Other infection types include central nervous system infection, bone / joint, wound, endocarditis, implantable device, blood stream catheter and tonsillitis. Respiratory infection was the most common source of infection recorded, occurring in 58% of cases, followed by urinary tract infection in 15 (Figure 4.1). The high proportion of patients with respiratory infection may have been related to multiple attendances in the chronic obstructive airways disease patient group. Patients who presented with sepsis three or four times during the audit period most often did so with respiratory infection. The source was recorded as Unknown where no specific reference was made to the source of infection. 20

Table 4.1: How many physiological derangements were noted per sepsis case (at the point of inclusion in the audit)? Number of physiological derangements N o. cases % cases Two 1538 39.5 Three 1438 37.0 Four 678 17.4 Five 200 5.1 Six or seven 36 0.9 Total 3890 100.0 The majority (77%) of patients had either two or three physiological derangements at the point of inclusion in the audit. Figure 4.2: How often was each type of physiological derangement noted (at the point of inclusion in the audit)? Deranged Not deranged Unknown Systolic blood pressure Mental state Observation Blood glucose Temperature Respiratory rate White blood count Heart rate 0% 20% 40% 60% 80% 100% A heart rate greater than 90 beats per minute was found to be the most common immediately available physiological derangement (81%). Only 44% of patients were noted to have a derangement in temperature. Systolic blood pressure was the least common derangement noted (10%). Deranged white cell count was noted to be present in 70% of patients, but this measurement was not necessarily immediately available. There were no cases within the audit where the patient was recorded as having a normal white cell count with more than 10% immature forms. 21

Section 5: Severe Sepsis and Septic Shock Summary In prior sections, the term sepsis has been used to represent the entire spectrum of sepsis, severe sepsis and septic shock. In Section 5, the patients who fulfilled the entry criteria were followed to assess whether they developed severe sepsis, within two days of initial attendance. The criteria for the development of severe sepsis were incorporated from the SSC guideline definition. These were the presence of the signs of sepsis and any of the following: Systolic Blood Pressure (SBP) < 90mmHg or Mean Arterial Blood Pressure (MABP) < 65mmHg or on inotropes SBP decrease > 40mmHg from baseline Lactate > 2mmol/l SpO2 < 90% on room air or supplemental oxygen Bilateral pulmonary infiltrates with PaO2/FiO2 < 40 Creatinine >176.8μmol/l or urine output < 0.5mls/kg/hr for 2hrs Bilirubin > 34.2mmols Platelets < 100x109l INR >1.5 or APTT > 60s Following identification of severe sepsis, specific interventions, subsequent patient responses and timings were recorded. With the exception of Figures 5.1 and 5.2, all of the patients described in this section of the report met the criteria for severe sepsis. Figure 5.1 illustrates the proportion of patients who developed the criteria for severe sepsis within two days of attendance and Figure 5.2 demonstrates the proportion of patients who were admitted to critical care within this timeframe. The SSC Guidelines recommend that resuscitation begins immediately in septic patients with hypotension or elevated serum lactate > 4mmol/L. Figure 5.3 identifies the patients who were candidates for fluid resuscitation due to one or both of these derangements. Figure 5.4 shows what proportion of these patients were given immediate fluids. When both hypotension and a raised lactate were present, immediate fluids were given in 85. Of the patients in the hypotensive group, 63% responded to immediate fluid resuscitation. 26% of the patients in the raised lactate group responded to immediate fluid resuscitation (Figure 5.5). It is worth noting that the mean volume of fluids given to responders was less than that of non-responders in both groups. Between 35% and 47% of patients who did not respond to initial fluid resuscitation went on to receive central line monitoring or inotropes and central venous O 2 saturation was rarely measured (Figure 5.6). 22

Figure 5.1: How many patients showed clinical signs of severe sepsis within two days of attendance? 60% 50% 40% 30% 20% 10% 0% Severe sepsis No severe sepsis Clinical signs present, but no date available* (1325) (2033) (532) Severe sepsis within two days (number of cases) * The date of the first clinical sign of severe sepsis could not be obtained, therefore it was not possible to determine if the patient developed severe sepsis within two days of attendance. Overall 34% of patients who were eligible for inclusion in the audit had signs of severe sepsis observed within two days of attendance. 23

Figure 5.2: Were patients who developed severe sepsis more likely to go to Critical Care within two days of attendance? Crit. Care within two days (561) No Crit. Care within two days (3328) 100% 80% 60% 40% 20% 0% Severe sepsis No severe sepsis* Clinical signs present, but no date available (1325) (2032) (532) Severe sepsis within two days (number of cases) * In one No severe sepsis case it is unknown whether the patient went to critical care within two days of attendance. Of the patients who met the criteria for severe sepsis, 27% were admitted to critical care within two days of attendance. 24

Figure 5.3: What proportion of patients with severe sepsis (1325 of 3890 cases) met the criteria which recommend giving an immediate fluid challenge? Yes No Unknown 100% 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l Both (1325) (1325) (1325) Observation (number of cases) Almost half of those patients identified as candidates for fluid resuscitation had hypotension (SBP < 90mmHg) and 12% had an elevated lactate level (>4mmol.l). Not all sites had access to immediate lactate testing which is likely to have affected the frequency of testing. 25

Figure 5.4: Of immediate fluid candidate patients with SBP < 90mmHg (659 cases), Lactate > 4mmol/l (162 cases) or both (89 cases), how many were given immediate fluids? Immediate fluids given No immediate fluids given Unknown 100% 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l Both (659) (162) (89) Observation (number of cases) Of the 659 patients with signs of sepsis induced hypotension, 58 received immediate fluid resuscitation. This was defined as within one hour. When both hypotension and an elevated lactate were present, patients receiving immediate fluid resuscitation increased to 85. Information on patient co-morbidities was not recorded therefore there may have been cases where there were clinical reasons for not giving a fluid challenge. 26

Figure 5.5: Of immediate fluid candidate patients with SBP < 90mmHg or Lactate > 4 mmol/l who were given immediate fluids (385 and 117 cases respectively), how many responded? Did not respond Responded Unknown 100% 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l (385) (117) Observation (number of cases) Septic shock is defined by persistent hypotension, and / or persistently elevated lactate following immediate fluid resuscitation. Therefore in 34% of the hypotensive group and 33% of those with lactate > 4mmol//l, despite being given immediate fluids, signs of septic shock were present. The mean volumes of fluids given to patients in the hypotensive group who did respond and did not respond were 0.8 litres and 1.0 litre respectively. In the elevated lactate group the mean volumes for those who did and did not respond were 1.1 litres and 1.3 litres respectively. 27

Figure 5.6: How were immediate fluid candidate patients with SBP < 90mmHg or Lactate > 4mmol/l, who failed to respond to immediate fluids (132 cases and 38 cases respectively), treated? Yes No Unknown 100% 80% 60% 40% 20% 0% Vasopressors administered CVP* measured SCVO2 measured Vasopressors administered CVP* measured SCVO2 measured SBP < 90mmHg (132) Lactate > 4mmol.l (38) Observation (number of cases) / Action * Central Venous Pressure Central Venous Oxygen Saturation In the hypotensive group of patients, of the 132 cases where signs of sepsis induced shock were present, 35% had vasopressors administered and 36% had CVP measured. In the elevated lactate group, 42% had vasopressors administered and 47% CVP measured. The results also demonstrate that central venous O 2 saturation is very rarely measured in participating Scottish hospitals to aid with the resuscitation of patients with septic shock. The overall mortality for patients who met the criteria for severe sepsis was 24% and for the patients who met the criteria for septic shock was 43%. 28

Section 6: Severe Sepsis in the Emergency Department The following section reviews the management of the 637 patients who met the diagnostic criteria for severe sepsis before leaving the ED (Table 6.1). The Surviving Sepsis Campaign (SSC) made a series of evidence based recommendations for optimal management of patients with severe sepsis. The College of Emergency Medicine Clinical Effectiveness Committee published standards for severe sepsis and septic shock management in adults in May 2009, around the end point of the index period in this audit. These are reproduced in Appendix 2. In this section compliance with some of these recommendations is reported (Table 6.2): Supplementary oxygen was commenced either before or within one hour of the first signs of severe sepsis on 76% of occasions. Blood cultures were taken in 21% of patients either before or within an hour. Intravenous antibiotics were administered on 32% of occasions either before or within one hour, and 66% of occasions within four hours. Lactate was measured within four hours on 55% of occasions. A greater proportion of patients who developed signs of severe sepsis before leaving the ED met the criterion for immediate fluids through having a lactate of >4mmol/l (Figure 6.1). It was noted in Figure 6.2 that patients with hypotension who had severe sepsis before leaving ED were administered fluids within one hour more frequently than those in the all severe sepsis cases group (70% compared with 58%). Figure 6.3 illustrates that there was no notable difference in the proportion of patients with severe sepsis who responded to immediate fluids when comparing ED severe sepsis and the all severe sepsis group. The ED severe sepsis group, who did not respond to immediate fluids, had comparable rates of vasopressors and central venous pressure (CVP) measurement when compared to the all severe sepsis group (Figure 6.4). The results also demonstrate that central venous O 2 saturation (ScVO 2 ) is very rarely measured in patients who develop the criteria for severe sepsis within the ED to aid with the resuscitation of patients with septic shock. Around half of the patients who met the criteria for inclusion with severe sepsis, did so before leaving the ED. The provision of a fluid challenge, re-assessment and repeating if indicated, is a simple yet effective intervention that should be achievable along with the administration of effective antibiotics. Raising awareness of this, allied to the importance of taking blood cultures prior to administration of antibiotics, may improve outcome and allow better targeting of the causative organism further on in the patient journey. 29

Table 6.1: How many of the patients who developed severe sepsis (1325 of 3890 cases), did so before leaving ED? N o. cases % cases Severe sepsis developed before leaving ED 637 48.1 Severe sepsis developed after leaving ED 496 37.4 Severe sepsis occurred, but unknown if it developed before or after leaving ED 36 2.7 Severe sepsis occurred, but initial area was Admissions Unit 156 11.8 Total 1325 100.0 Table 6.2: Of patients who developed severe sepsis before leaving ED (637 of 3890 cases), how long after the first clinical signs of severe sepsis did each intervention/investigation take place? Commenced on Oxygen N o. cases % cases Blood cultures taken N o. cases % cases IV antibiotics given N o. cases % cases No IV - Oral antibiotics instead N o. cases % cases Critical Care assessment N o. cases % cases Lactate measured N o. cases % cases Before severe sepsis 204 32.0 42 6.6 48 7.5 * * * * 56 8.8 < 1 hr after severe sepsis 280 44.0 94 14.8 158 24.8 8 1.3 11 1.7 231 36.3 1 - < 2 hrs after severe sepsis 34 5.3 15 2.4 124 19.5 5 0.8 15 2.4 37 5.8 2 - < 4 hrs after severe sepsis 22 3.5 22 3.5 88 13.8 22 3.5 19 3.0 23 3.6 4 < 6 hrs after severe sepsis 8 1.3 10 1.6 32 5.0 5 0.8 14 2.2 5 0.8 6-48 hrs after severe sepsis 17 2.7 46 7.2 71 11.1 32 5.0 41 6.4 30 4.7 > 48 hrs after severe sepsis 0 0.0 0 0.0 0 0.0 * * * * 0 0.0 Within two days, but timing data missing 8 1.3 157 24.6 17 2.7 * * 40 6.3 46 7.2 Sub-Total 573 90.0 386 60.6 538 84.5 77 12.1 146 22.9 428 67.2 Not carried out within two 61 9.6 251 39.4 96 15.1 19 3.0 483 75.8 159 25.0 days Not known if carried out within 3 0.5 0 0.0 3 0.5 3 0.5 8 1.3 50 7.8 two days Not applicable 0 0.0 0 0.0 0 0.0 538 84.5 0 0.0 0 0.0 Total 637 100.0 637 100.0 637 100.0 637 100.0 637 100.0 637 100.0 * Indicates small values that have been suppressed due to the potential risk of disclosure - carried out in accordance with ISD s Disclosure Control Protocol 3. In the 637 cases of patients who had signs of severe sepsis prior to leaving the ED, positive microbiological culture were identified within seven days in 35. 39% of patients had no blood cultures taken within the first two days following attendance. 52% of patients received IV antibiotics within two hours, and 32% within the one hour target. 30

Figure 6.1: What proportion of patients with severe sepsis met the criteria which recommend giving an immediate fluid challenge? (All severe sepsis cases vs. severe sepsis in ED cases) Yes No Unknown 100% 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l Both SBP < 90mmHg Lactate > 4mmol.l (1325) (637) All severe sepsis cases Both Severe sepsis in ED cases Observation (number of cases) 31

Figure 6.2: Of immediate fluid candidate patients with SBP < 90mmHg, Lactate > 4mmol/l or both, how many were given immediate fluids? (All severe sepsis cases vs. severe sepsis in ED cases) 100% Immediate fluids given No immediate fluids given Unknown 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l Both SBP < 90mmHg Lactate > 4mmol.l Both (659) (162) (89) (315) (122) (67) All severe sepsis cases Severe sepsis in ED cases Observation (number of cases) Information on patient co-morbidities were not recorded therefore there may have been cases where there were clinical reasons for not giving a fluid challenge. Figure 6.3: Of immediate fluid candidate patients with SBP < 90mmHg or Lactate > 4mmol/l who were given immediate fluids, how many responded? (All severe sepsis cases vs. severe sepsis in ED cases) 100% Did not respond Responded Unknown 80% 60% 40% 20% 0% SBP < 90mmHg Lactate > 4mmol.l SBP < 90mmHg Lactate > 4mmol.l (385) (117) (221) (88) All severe sepsis cases Severe sepsis in ED cases Observation (number of cases) 32