ARTICLE

Vol. 137 No. 1590 |

DOI: 10.26635/6965.6298

Raise the Flag II: sepsis mortality before and after the introduction of a whole-of-system quality improvement programme at a tertiary hospital in New Zealand

Ensuring reliable recognition and early resuscitation of sepsis in frontline health services should be a priority globally and for New Zealand, a country with high rates of invasive bacterial infection.

Full article available to subscribers

Ensuring reliable recognition and early resuscitation of sepsis in frontline health services should be a priority globally and for New Zealand, a country with high rates of invasive bacterial infection.1–7 For example, the incidence of invasive skin and soft tissue infection caused by Staphylococcus aureus rose from 81 to 140 cases per 100,000 population between 2000 and 2011, while rates of invasive Group A beta-haemolytic streptococcal disease rose from 4 to 9 per 100,000 population between 2002 and 2016.3,4 Using discharge coding data collected between 1989 and 2008, Baker et al. demonstrated a 5% increase in the proportion of hospital admissions caused by infectious diseases. The risk of infection-related hospital admission at least doubled in association with ethnic group (Māori or Pacific Island), socio-economic deprivation and age (below age 5 or above age 70).5 In this context, it would be logical to expect an increase in the number of acute presentations with sepsis, which is currently defined as a “life-threatening illness due to a dysregulated host response to infection”.6 In a study conducted in the Waikato Region, Huggan et al. reported that sepsis admissions were more frequent in 2012 compared with 2008 (age-standardised rate ratio [ASRR] 1.62, 95% confidence interval [CI] 1.18–2.24), and that sepsis was over three times more likely among people of Māori ethnicity (ASRR 3.22, 95% CI 2.85–3.65).7

At our hospital in 2016, a group of clinicians with expertise in sepsis management was tasked with improving sepsis management in publicly funded facilities in the Waikato Region. This led to the design and implementation of a whole-of-system quality improvement intervention, known as “Raise the Flag”. The Raise the Flag programme led to an early but non-sustained improvement in the delivery of an acute sepsis resuscitation bundle.8 Beyond immediate resuscitation efforts, however, the programme encouraged a culture of prioritising sepsis management and removing barriers to this objective. We therefore hypothesised that the Raise the Flag programme could impact sepsis mortality independently of immediate resuscitation practice, and conducted a retrospective study of this outcome using discharging coding data.

Methods

Of the Waikato resident population at the 2018 New Zealand Census (n=458,202), 24% identified as Māori (n=109,488).9 Waikato Hospital is the tertiary referral centre for four public hospitals in the Waikato Region. In 2015, 10% of New Zealand’s emergency department presentations were managed across this hospital network (105,347/1,062,047). Sixty-eight percent of these were direct to the Waikato Hospital emergency department (n=72,070).10

In 2016, a multi-disciplinary Sepsis Action Group (SAG) was established by the Waikato District Health Board Quality and Patient Safety executive. By 2017, the group included infectious disease, intensive care, paediatric medicine, emergency medicine, rural medicine, intensive care nursing, quality improvement, public relations, communications and graphic design experts. Consumer and resident medical officer representatives were also appointed. The group received advice from experts in Māori health and reported to the Waikato Hospital Iwi Māori Council.

The position of sepsis nurse coordinator was established in 2018, and a sepsis recognition and action tool was introduced to all acute care facilities in August of that year. The tool listed high-risk findings necessitating urgent treatment and, on the reverse, specified six actions to be completed urgently, ideally within 60 minutes. These actions were i) administer oxygen if required, ii) give fluid if required, iii) measure serum lactate, iv) send blood cultures, vi) give appropriate antibiotics and vi) measure urine output.8

Alongside introduction of the tool, concerted efforts were made to increase health workforce awareness of sepsis as a product of health inequity, and a major cause of mortality. Dedicated educational resources were offered to all staff, including an e-learning package, while mandatory orientation to the programme was included in new staff orientation. Summary resuscitation outcomes were shared with staff in newsletters and grand rounds. The sepsis nurse coordinator provided in situ education and training for frontline staff. Senior clinical leaders within individual departments, termed “sepsis champions”, were recruited to promote programme adoption and to explore barriers to timely recognition and treatment of sepsis. Continuous improvement based on this feedback led to the post-launch development of new resources. These included an acute abdomen pathway (based on the need to prioritise abdominal imaging and transfer to theatre in abdominal sepsis), a hypoperfusion pathway (to define intensive care unit [ICU] referral criteria and indications for use of vasoactive medications) and a multi-media design package to increase programme visibility in clinical and non-clinical areas. Māori and Pacific ethnicities were added as an “amber flag” to the sepsis recognition and action tool early in 2020 based on concern for higher prior probability of infection and sepsis as a cause of acute illness. Changes in programme and policy were communicated to staff by the SAG within the wider communication and education efforts given above, and through the network of sepsis champions.

This was a low-risk observational study, registered prospectively with the local audit committee, but was considered out-of-scope for a Health and Disability Ethics Committee review. Programme evaluation was informed by Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.11

Sepsis cases were identified in discharge coding data from May 2015 to July 2021 using the New Zealand Sepsis Indicator (NZSI), developed to mirror traditional approaches to the study of sepsis epidemiology.12 The NZSI makes use of the International Statistical Classification of Diseases and Related Health Problems 10th Revision, Australasian Modification (ICD-10-AM). It identifies primary discharge codes specifying sepsis (i.e., A40.0 Sepsis due to streptococcus, group A), which are associated with secondary codes defining organ failure (i.e., N17 Acute renal failure).12 Eighty-six percent of patients identified by the NZSI meet current clinical criteria for sepsis.7,12

Although quality improvement efforts covered paediatric admissions, earlier work suggests that the NZSI performs poorly in identifying paediatric sepsis cases.6,12 Patients aged 15 and over are admitted to adult services at Waikato Hospital so were exposed to the intervention. For these reasons, we included in this study all acute, overnight admissions to Waikato Hospital of patients aged 15 and over. To avoid potential bias introduced by the need for transport for definitive care, we excluded cases presenting to other hospitals in our network in this analysis. The primary outcome was in-hospital mortality measured at acute hospital discharge. This excludes time in convalescent care and rehabilitation facilities. Secondary outcomes of interest were: 30- and 90-day mortality; need for ICU admission; ICU length of stay (LOS); and acute in-patient LOS. Exposures measured were: age; prioritised ethnicity; an address-based measure of socio-economic deprivation; and the Charlson Comorbidity Index (calculated using a method validated for use in discharge coding data).13,14 Based on the study of Burrell et al., we divided time exposure into two periods, an extended baseline period (May 2015 to July 2019) and a post-implementation period (August 2019 to June 2021).1

A final record extract was undertaken on 16 October 2021, which included mortality data through to 30 September 2021. Regional population estimates derived from the 2018 New Zealand Census were used to estimate age-standardised sepsis incidence. The Chi-squared test was used to compare categorical variables with results considered significant at a p-value of less than 0.05. Binary logistic regression was used to calculate adjusted odds ratios [aOR] and 95% confidence intervals [CI] for individual variables (gender, age, ethnicity, socio-economic status, Charlson Comorbidity Index score and year of admission) against outcomes of interest (in-patient, 30- and 90-day mortality). For independent variables included in the binary logistic regression, missing values were coded as “unknown”. This was to avoid excluding patients with missing values in some variables and to include all patients in the analysis. All data analyses were performed in IBM SPSS statistics version 27 (New York, United States).

Results

Four thousand, two hundred and sixty-eight sepsis cases were identified and are described in Table 1. There were 2,432 cases in the extended baseline period and 1,836 in the post-implementation period. Sepsis was more common among men (58%), those aged over 55 (81%) and in the presence of significant comorbidity based on a Charlson score ≥1 (61%). Over two thirds (68%) of all cases were recorded as living in the two highest quintiles of socio-economic deprivation. One third (34%) of all patients were of Māori or Pacific ethnicity. The majority of patients (81.5%) were managed on the general wards without ICU admission. The proportion of patients with no comorbidity was higher in the post-implementation period (proportion of patients with a Charlson score of 0, 41.9% vs 36.1%, p=<0.01). In the total study population, in-patient mortality was 20% (838/4,268). Table 2 demonstrates that crude mortality was lower in the post-implementation period (in-hospital mortality 17.4% vs 21.3%, p<0.01). There was no crude difference in ICU or in-patient LOS.

View Table 1–3.

The results of binary logistic regression are shown in Table 3. Admission during the post-implementation period was associated with reduced odds of in-hospital mortality (odds ratio [OR] 0.83, 95% CI 0.7–0.98, p <0.05). There was weak evidence of lower 30-day mortality (OR 0.86, 95% CI 0.74–1.00, p=0.051). There was no evidence of a difference in 90-day mortality (OR 0.90, 95% CI 0.78–1.04, p=0.152). There was a significant association between mortality and socio-economic deprivation. Among cases in the fifth (compared with the first) quintile of deprivation, odds of in-patient mortality were 1.47 (95% CI 1.06–2.04, p=0.02). Increasing age was strongly associated with mortality. The adjusted odds of patient death increased by 1.03 for every additional year of age (95% CI 1.03–1.05, p<0.01). Comorbidity was also a significant determinant of mortality. For example, compared to a comorbidity score of 0, the odds of in-patient mortality were increased by 2.72 for a comorbidity score of 1 (95% CI 2.10–3.52, p<0.01), and by 4.76 for a score of 3 or more (95% CI 3.81–5.95, p<0.01). There was no statistically significant evidence of an association between non-Māori, non-Pacific ethnicities and either an increased or a decreased risk of mortality (i.e., aOR for in-patient mortality 1.16, 95% CI 0.95–1.41, p=0.15).

Discussion

Following implementation of a sepsis quality improvement programme, after adjusting for important confounding variables such as age and comorbidity, we observed a 17% reduction in the odds of in-patient mortality among patients with sepsis (aOR 0.83, 95% CI 0.70–0.98, p=0.03). A weak association persisted at 30 days (OR 0.86, 95% CI 0.74–1.00, p=0.05), but not at 90 days (OR 0.90, 95% CI 0.78–1.04, p=0.15). We observed no change in ICU or in-patient hospital LOS following the intervention. Socio-economic deprivation was independently associated with increased mortality at all time points (i.e., aOR for 90-day mortality comparing the highest with the lowest quintile of deprivation 1.39, 95% CI 1.05–1.85, p<0.05).

We report what we believe to be the largest single-centre study of sepsis outcomes conducted in New Zealand. Strengths of our approach include the use of administrative data to allow continuous reporting of incidence and clinical outcomes over a 6-year period. In earlier work using this approach we demonstrated that 86% of patients satisfy contemporary sepsis definitions.12 We are therefore confident that the significant majority of patients identified here presented with a critical illness. In this context, a crude reduction of 4% in acute mortality is clinically relevant and is consistent with the reported outcomes of state-wide sepsis quality improvements in Australia.1,2 This study adds to these findings by adjusting for important confounding variables, including differences in age and comorbidity between the pre- and post-launch periods. This strengthens the argument that the association of quality improvement programmes with reduced odds of in-patient mortality are due to the programmes themselves, rather than to changes in underlying patient vulnerability.

The limitations of this study include its retrospective design, and use of a dataset lacking clinical granularity. This means that we have not been able to measure all confounding variables known to affect mortality, and are therefore unable to determine if observed improvements in mortality relate to residual confounding. We acknowledge that the observation period includes the early stages of the COVID-19 pandemic, which in New Zealand was managed by significant restrictions on population movement and association, with as yet unknown impacts on the underlying causes of sepsis. Our observation period may have been too short to document significant improvements in LOS. Finally, we acknowledge that no observational study reporting an association between an intervention and an outcome can prove causation.

Beyond immediate resuscitation efforts, clinicians managing sepsis provide a range of interventions and complex therapies, each representing an opportunity to expedite and improve resuscitation and treatment. Successful sepsis care relies not just on a single intervention, such as introduction of a resuscitation bundle, but on a raft of education and process changes that raise awareness and lead to changes in process and prioritisation. For example, although the proportion of patients admitted to the ICU during their hospital stay did not change in this analysis, Walland et al. reported a sustained increase in the odds of direct transfer to ICU following a sepsis diagnosis after the programme launch in 2018 (aOR 2.81, 95% CI 1.13–6.97, p=0.03).8 Early ICU access for patients with sepsis may be an important determinant of outcome in resuscitation-eligible groups. Sepsis mortality is lower in the USA where a higher proportion of cases are admitted to ICU, compared to European settings where ICU admission is less common.15 Measures of ICU access are associated with outcomes in critical illness. In one study, a delay in ICU transfer of 6 hours or more was associated with a 30% increase in the adjusted odds of 30-day mortality.16 In another, the opening of an ICU within a large metropolitan emergency department reduced the odds of mortality at 30 days by 15%.17

We observed that the majority of patients with sepsis are managed outside intensive care, across a range of clinical specialties. This implies that improvements in sepsis care rely significantly on care in environments remote to critical care services. Real-world evaluations of practice have shown that staff working outside an ICU are poorly adherent to sepsis resuscitation bundles. For example, in a study using annual point-prevalence studies in 14 Welsh hospitals, the full sepsis resuscitation bundle was completed in 14% of eligible patients (223/1,651), but in 11% (190/1,651) none of the bundle elements were completed.18 Those patients seen by a critical care outreach team received the complete bundle in 32% of cases (54/170). In Australia and New Zealand, where rapid response teams operate, sepsis is the most frequent reason for their activation.19,20 Critical care outreach may be key to the delivery of the immediate resuscitation steps on which most quality improvement efforts are typically focussed.

We are not surprised to report that improvements in in-patient mortality were no longer sustained by day 90. After 30 days, sepsis deaths are rarely related to sepsis-associated organ failure, suggesting that late mortality is mediated by the interaction of a proximate sepsis event with its ultimate underlying causes (i.e., medical comorbidity, major trauma or advanced age).21 This does not clearly explain why the adjusted odds of mortality were higher at all time points for those living in the highest quintile of socio-economic deprivation. Although the association between increasing neighbourhood socio-economic deprivation and sepsis mortality is well recognised, this is the first time, to our knowledge, that this has been reported in New Zealand.22 Increased mortality among those experiencing socio-economic deprivation is likely to be mediated in part by unmeasured confounding factors, such as smoking. However, the possibility of deficits in the quality of care and follow-up following critical illness is an important topic for further research.

In summary, this is the first report from New Zealand demonstrating an association with reduced in-hospital mortality following the launch of a whole-of-system sepsis quality improvement intervention. For all patients, sepsis outcomes are worsened by increasing age, comorbidity and exposure to socio-economic deprivation. This study therefore directs attention to chronic illness and the social determinants of health as drivers of preventable sepsis morbidity and mortality. Investment in sepsis quality improvement has the potential to improve short-term sepsis outcomes.

Aim

To study in-patient mortality before and after the introduction of a whole-of-system sepsis quality improvement programme at a tertiary hospital in New Zealand.

Methods

The “Raise the Flag” sepsis quality improvement programme was launched in 2018. Discharge coding data were used to identify sepsis cases between May 2015 and July 2021.

Results

Of 4,268 cases of sepsis identified, 81% were over 55 years old, 34% were of Māori or Pacific Island ethnicity, 61% had significant co-morbid illness and over two thirds (68%) lived in the two highest quintiles of socio-economic deprivation. The adjusted odds of in-patient mortality were lower in the post-launch period (adjusted odds ratio [aOR] 0.83, 95% confidence interval [CI] 0.7–0.98, p<0.05), and were higher in association with age (aOR 1.04 for every additional year of age, 95% CI 1.03–1.05, p<0.01), socio-economic status (aOR 1.47 comparing the highest quintile of socio-economic deprivation with the lowest, 95% CI 1.06–2.04, p=0.02) and comorbidity (aOR 2.42 comparing a comorbidity score of 1 with a score of 0, 95% CI 2.1–3.52, p<0.01).

Conclusion

In patients with a sepsis diagnosis, the odds of in-patient death were lower following the launch of the Raise the Flag sepsis quality improvement programme.

Authors

Paul J Huggan: Senior Medical Officer, Departments of Infectious Disease and General Medicine, Te Whatu Ora Waikato, New Zealand.

Katherine M Walland: Senior Medical Officer, Department of Infectious Disease and General Medicine, Te Whatu Ora Waikato, New Zealand.

Chunhuan Lao: Senior Research Fellow, Te Huataki Waiora School of Health, University of Waikato, New Zealand.

Anna Gwynne: Data Analyst, Te Whatu Ora Waikato, New Zealand.

Daniel Dobbins: Senior Medical Officer, Department of Emergency Medicine, Te Whatu Ora Waikato, New Zealand.

Robert Martynoga: Intensive Care Specialist Advisor and Trustee, The New Zealand Sepsis Trust, Hamilton, New Zealand.

Correspondence

Paul J Huggan: Senior Medical Officer, Departments of Infectious Disease and General Medicine, Te Whatu Ora Waikato, New Zealand.

Correspondence email

paul.huggan@waikatodhb.health.nz

Competing interests

Nil.

1)       Burrell AR, McLaws ML, Fullick M, et al. SEPSIS KILLS: early intervention saves lives. Med J Aust. 2016;204(2):73 doi: 10.5694/mja15.00657.

2)       Brusco NK, Sykes K, Cheng AC, et al. A state-wide implementation of a whole of hospital sepsis pathway with a mortality based cost-effectiveness analysis from a healthcare sector perspective. PLOS Glob Public Health. 2023;3(5):e00006873. https://doi.org/10.1371/journal.pgph.0000687.

3)       Williamson DA, Zhang J, Ritchie SR, et al. Staphylococcus aureus infections in New Zealand, 2000-2011. Emerg Infect Dis. 2014;20(7):1156-61. doi: 10.3201/eid2007.131923.

4)       Williamson DA, Morgan J, Hope V, et al. Increasing incidence of invasive group A streptococcus disease in New Zealand, 2002–2012: a national population-based study. J Infect. 2015;70(2):127–34. doi: 10.1016/j.jinf.2014.09.001.

5)       Baker MG, Barnard LT, Kvalsvig A, et al. Increasing incidence of serious infectious diseases and inequalities in New Zealand: a national epidemiological study. Lancet. 2012;379 (9821):1112-1119. doi: 10.1016/S0140-6736(11)61780-7.

6)       Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801-810. doi: 10.1001/jama.2016.0287.

7)       Huggan PJ, Bell A, Waetford J, et al. Evidence of High Mortality and Increasing Burden of Sepsis in a Regional Sample of the New Zealand Population. Open Forum Infect Dis. 2017;4(3):ofx106. doi:10.1093/ofid/ofx106.

8)       Walland KM, Howard C, Paul O, Huggan PJ. Raise the Flag I: the impact of a sepsis quality improvement programme on delivery of a sepsis resuscitation bundle at a tertiary hospital in New Zealand. NZ Med J. 2023 Dec 15;136(1587):75-84.

9)       Stats New Zealand. 2018 Census [Internet]. 2020 [cited 2021 Nov 13]. Available from: https://www.stats.govt.nz/2018-census/.

10)    Manatū Hauora – Ministry of Health. Emergency department use 2014/15 [Internet]. 2016 [cited 2021 Nov 13]. Available from: https://www.health.govt.nz/publication/emergency-department-use-2014-15.

11)    von Elm E, Altman DG, Egger M, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335(7624):806-808. doi: 10.1136/bmj.39335.541782.AD.

12)    Huggan PJ, Helms TA, Gibbons V, et al. Counting the cost of major infection and sepsis in New Zealand: an exploratory study using the National Minimum Data Set. N Z Med J. 2021;134(1528):10-25.

13)    Salmond C, Crampton P, Atkinson J. NZDep2006 Index of Deprivation [Internet]. Wellington, New Zealand: University of Otago; 2007 [cited 2024 Jan 30]. Available from: https://www.otago.ac.nz/wellington/research/groups/research-groups-in-the-department-of-public-health/hirp/socioeconomic-deprivation-indexes-nzdep-and-nzidep-department-of-public-health.

14)    Quan H, Li B, Couris C, et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011;173(6):676-682. doi: 10.1093/aje/kwq433.

15)    Levy MM, Artigas A, Phillips GS, et al. Outcomes of the Surviving Sepsis Campaign in intensive care units in the USA and Europe: a prospective cohort study. Lancet Infect Dis. 2012;12(12):919-924. doi: 10.1016/S1473-3099(12)70239-6.

16)    Chalfin DB, Trzeciak S, Likourezos A, et al. Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med. 2007;35(6):1477-83. doi: 10.1097/01.CCM.0000266585.74905.5A.

17)    Gunnerson KJ, Bassin BS, Havey RA, et al. Association of an Emergency Department-Based Intensive Care Unit With Survival and Inpatient Intensive Care Unit Admissions. JAMA Netw Open. 2019;2(7):e197584. doi: 10.1001/jamanetworkopen.2019.7584.

18)    Kopczynska M, Unwin H, Pugh RJ, et al. Four consecutive yearly point-prevalence studies in Wales indicate lack of improvement in sepsis care on the wards. Sci Rep. 2021;11(1):16222. doi: 10.1038/s41598-021-95648-6.

19)    White K, Scott IA, Bernard A, et al. Patient characteristics, interventions and outcomes of 1151 rapid response team activations in a tertiary hospital: a prospective study. Intern Med J. 2016;46(12):1398-1406. doi: 10.1111/imj.13248.

20)    Young PJ, Gladwin B, Psirides A, Reid A. Unplanned admissions to the Wellington Hospital intensive care unit before, during and after New Zealand's COVID-19 lockdown. N Z Med J. 2020;133(1527):95-103.

21)    Daviaud F, Grimaldi D, Dechartres A, et al. Timing and causes of death in septic shock. Ann Intensive Care. 2015;5(1):16. doi: 10.1186/s13613-015-0058-8.

22)    Lusk JB, Blass B, Mahoney H, et al. Neighborhood socioeconomic deprivation, healthcare access, and 30-day mortality and readmission after sepsis or critical illness: findings from a nationwide study. Crit Care. 2023;27(1):287. doi: 10.1186/s13054-023-04565-9.