Introduction

Healthcare demand during severe influenza pandemics can exceed clinical capacity to provide normal standards of care [1], a phenomenon now known as “surge”. National pandemic preparedness plans include contingencies for surge capacity including reducing non-urgent activity and expanding critical care facilities. [2], [3] In exceptional circumstances the use of triage is recommended to rationalise use of limited hospital resources. [3].

Assessment tools could provide a framework for consistent and rational triage decisions pertaining to hospital referral. Ideally, tools used in the community by a General Practitioner (GP)* should match hospital admission triage criteria, and be valid predictors of the likelihood to benefit from use of hospital-based interventions. [*The term General Practitioner is used in the UK and some other countries to indicate a registered medical practitioner specialised in providing acute and chronic care to people of all ages and sexes in the community and in other countries may be known as a usual community doctor or primary care physician.].

During peak pandemic activity, healthcare worker absence is potentially increased due to personal sickness and caring for relatives. Imaging and laboratory services may also be limited. Together these place increased reliance and emphasis on clinical history-taking and examination skills for triage decisions, which may be devolved to less experienced staff.

Provisional guidance [4] suggested the use of the CURB-65 pneumonia score [5] and the Pandemic Medical Early Warning Score (PMEWS) [6] for hospital triage of adults in the UK. Importantly, neither score was ever intended for use in children. Most children can benefit from access to adult critical care facilities and general intensive care units either if they are unlikely to need transfer to a paediatric intensive care unit (PICU), or if in extreme circumstances there is no PICU capacity.

Recognising this gap, a “tool-kit” of national guidance was developed in 2008 in the UK which included newly developed Community Assessment Tools (CATs) for both children (Appendix S1) and adults (Appendix S2) in primary and secondary care, and matched hospital care pathways. [3], [7] CATs were designed to help front-line staff identify sick children and adults most likely to benefit from hospital care during pandemic surge periods. Although not specific to influenza, they nevertheless address the most likely modes of critical illness arising from the disease or its complications. Criteria fields are common to adult and paediatric CATs, however specific physiological thresholds and definitions (e.g. respiratory rate) are age-appropriate.

CATs use six objective criteria and one subjective criterion, any one or more of which should prompt urgent referral and admission to hospital (see Appendices 1 & 2). The evidence base for the objective criteria is that which supports the recognition of severe influenza and severe pneumonia in the community (including resource-limited settings) for adults and children, severe chronic obstructive pulmonary disease in adults, serious feverish illness in children and severe bronchiolitis in infants. [5], [8]–[17] Evidence has recently been published supporting use of subjective criteria in children [18].

The concept was developed by a group at Alder Hey Children’s Hospital NHS Foundation Trust Liverpool UK (Appendix S3), as part of an integrated package of care, which included community and hospital pathways. The pathways provide safety measures for people who are moderately unwell or who may have other diagnoses but cannot be admitted to hospital at the time of triage. The CATs and pathways were further developed by paediatric and adult expert clinical development groups and published by the Department of Health (England) for use by the National Health Service. [3] Clinicians were, and are, warned not to use the CATs and the pathways unless the local situation precludes normal admission and discharge processes.

During the somewhat mild 2009–10 influenza pandemic in the UK, some local areas came under surge pressure, especially for critical care, but overall national clinical capacity was not exceeded and CATs and surge pathways were not implemented. Notwithstanding, data collected by the UK Influenza Clinical Information Network (FLU-CIN) cohort provided an opportunity for a retrospective evaluation of the predictive ability of CATs in a pandemic scenario.

Methods

FLU-CIN was an emergency surveillance study established by the Department of Health England. The details of data collection and the findings have been described elsewhere. [19] In brief FLU-CIN used a purposive sampling frame based on 13 sentinel hospitals situated in five clinical ‘hubs’ in Nottingham, Leicester, London, Sheffield and Liverpool, with contributions from a further 45 non-sentinel hospitals in England and 17 in Scotland, Wales and Northern Ireland. Between April 2009 and January 2010, clinico-epidemiological and outcome data were collected on 1520 patients (800 female, 480 children <16 years) admitted to participating UK hospitals with A(H1N1)pdm09 infection, confirmed by reverse transcribed polymerase chain reaction (PCR) of respiratory samples obtained during the admission episode. Table 1 summarises socio-demographic characteristics of the FLU-CIN cohort. The clinical characteristics of the FLU-CIN cohort have been described elsewhere. [20] This was a case-control analysis using retrospective data from the full FLU-CIN cohort data. Criterion data were gathered from routine case notes using the first recorded routine clinical assessment on or shortly after admission.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic characteristics of 1520 UK patients hospitalised with pandemic H1N1 infection during the 2009/10 pandemic (The FLU-CIN Cohort) compared with source population and pre-pandemic hospital data on acute respiratory infection admissions.

https://doi.org/10.1371/journal.pone.0075384.t001

We investigated the association between each CATs criterion (separately) and the use of hospital-based clinical interventions as well as patient outcomes. Next, we used multivariable logistic regression modelling to identify which of the CATs criteria were independent predictors of the use of clinical interventions and patient outcomes. All criteria were included in the multivariable model in the first instance irrespective of statistical significance. Thereafter, we used an automatic stepwise forward selection approach (threshold for selection p<0.05) to compare the resultant multivariable model with our original approach.

The CATs criteria are:

1. severe respiratory distress,
2. increased respiratory rate,
3. oxygen saturation ≤92% on pulse oximetry breathing air, or “on oxygen”,
4. respiratory exhaustion,
5. evidence of severe clinical dehydration or clinical shock,
6. altered consciousness level and
7. causing other clinical concern to the patient’s GP or clinical team.

All criteria were validated using separately recorded clinical observations made by regular clinical staff on or shortly after admission.

Outcome measures included clinical interventions and patient outcomes: any use of supplemental oxygen; mechanical ventilation; intravenous antibiotics; length of stay (categorised as >48 hours; ≥6 days and ≥12 days); higher levels of care (high dependency – level 2 or intensive care – level 3*); death; and “severe outcome” defined by a composite of any use of higher levels of care (levels 2 or 3) or death during admission. [*Level 0: patients whose care needs can be met through normal ward care; level 1: patients at risk of deteriorating or recently relocated from higher levels of care whose needs can be met on an acute ward with additional advice and support from the critical care team; level 2: patients requiring more detailed observation or intervention, including support for a single failing organ system and those ‘stepping down’ from higher levels of care high dependency unit; level 3: patients requiring advanced respiratory support alone or basic respiratory support together with support of at least two organ systems. This includes all complex patients requiring support for multi-organ failure in intensive care units].

Analyses were conducted separately for children (age <16 years) and adults (≥16 years). Results are presented as unadjusted and adjusted odds ratio (OR) with 95% confidence intervals (95%CI) and p values. The OR describes the likelihood of a given outcome when an individual CAT criterion is positive. This was a secondary data analysis based on pragmatic recording of routine clinical assessments by regular clinical staff. The underpinning assumption is that the data recorded on presence of criteria was complete; however there is no way to verify this. By default, where a criterion was not recorded as being present, that criterion was assumed to be absent i.e. “If it isn’t in the notes it didn’t happen”. [21] Thus there are no missing data in the analysis. All analyses were carried out using Stata version 11.0 (StataCorp. 2009. Stata Statistical Software: Release 11.0; Stata Corporation, College Station, TX, USA).

Results

The unadjusted analysis investigating the association between each CATs criterion at admission (separately) and outcome measures (hospital-based intervention and patient outcomes) are presented for children (<16 years) in Table 2 and for adults (≥16 years) in Table 3. The distribution of study subjects meeting each criterion, tabulated by every outcome measure is reported in Table S1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Unadjusted analysis investigating the association between meeting CAT criteria and outcome measures in children (<16 years).

https://doi.org/10.1371/journal.pone.0075384.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Unadjusted analysis investigating the association between meeting CAT criteria and outcome measures in adults (≥16 years).

https://doi.org/10.1371/journal.pone.0075384.t003

Criterion A: In children and adults “Severe respiratory distress” was significantly associated with use of supplemental oxygen, use of mechanical ventilation, length of stay >48 hours & ≥6 days and severe outcome (level 2/3 admission and or death). In children “Severe respiratory distress” was also significantly associated with death, and in adults with use of intravenous antibiotics and length of stay ≥12 days.

Criterion B: In children and adults “Increased respiratory rate” was significantly associated with use of supplemental oxygen and length of stay (≥6 & ≥12 days). In adults “increased respiratory rate” was also significantly associated with use of mechanical ventilation. Increased respiratory rate was significantly associated with a decreased likelihood of intravenous antibiotic use in both children.

Criterion C: In children and adults “Peripheral oxygen saturation ≤92% breathing air, or on oxygen” was significantly associated with mechanical ventilation, intravenous antibiotic use, length of stay (>48 hours and ≥6 days) and severe outcome. In adults “Peripheral oxygen saturation ≤92% breathing air or on oxygen” was also significantly associated with length of stay (≥12 days) and death. Use of supplemental oxygen as an outcome measure was excluded ad oculos from this analysis, as peripheral oxygen saturation ≤92%, would with few exceptions trigger oxygen supplementation in adults and children. [5].

Criterion D: It was not possible to calculate odds ratios for “Respiratory exhaustion or apnoea” in children due to insufficient cases meeting this criterion. In adults “Respiratory exhaustion or apnoea” was significantly associated with use of supplemental oxygen, mechanical ventilation, increased length of stay (≥6 & ≥12 days), severe outcome and death.

Criterion E: In children and adults “Severe clinical dehydration or shock” was significantly associated with use of mechanical ventilation, severe outcome and death. In adults “Severe clinical dehydration or shock” was also significantly associated with use of supplemental oxygen, intravenous antibiotics and length of stay (≥6 & ≥12 days).

Criterion F: In children and adults “Altered conscious level” was significantly associated with use of supplemental oxygen, mechanical ventilation, severe outcome and death. In adults “altered conscious level” was also associated with intravenous antibiotic use and length of stay (≥6 & ≥12 days).

Criterion G: In children and adults the subjective criteria “Causing other clinical concern” was associated with intravenous antibiotic use and severe outcome. In children “causing other clinical concern” was also significantly associated with death, and in adults with length of stay (≥6 days). The most frequent examples of meeting this criterion included in children: fever alone or in combination with rash or mottled skin or pain or cough; and in adults: chest pain, fever alone or in combination with cough or pain.

Summary results of multivariable analyses using the automatic stepwise forward selection procedure of CATs criteria at admission as independent predictors of outcomes (hospital-based intervention and patient outcomes) in children and adults are presented in Tables 4 & 5, and in Figures 1 & 2. Results were very similar when all criteria were retained in the model (Tables S2 & S3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. CAT criteria as independent predictors of use of oxygen, mechanical ventilation and severe outcomes in children.

Multivariable analyses by forward stepwise regression of CAT criteria as independent predictors of use of oxygen [upper panel], mechanical ventilation [middle panel] and combined severe outcomes (level 2/3 admission or death)[lower panel] in children (<16 years).

https://doi.org/10.1371/journal.pone.0075384.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. CAT criteria as independent predictors of use of oxygen, mechanical ventilation and severe outcomes in adults.

Multivariable analyses by forward stepwise regression of CAT criteria as independent predictors of use of oxygen [upper panel], mechanical ventilation [middle panel] and combined severe outcomes (level 2/3 admission or death)[lower panel] in adults (≥16 years).

https://doi.org/10.1371/journal.pone.0075384.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Multivariable analyses of CAT criteria as independent predictors of outcomes in children (<16 years), forward stepwise regression.

https://doi.org/10.1371/journal.pone.0075384.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Multivariable analyses of CAT criteria as independent predictors of outcomes in adults (≥16 years), forward stepwise regression.

https://doi.org/10.1371/journal.pone.0075384.t005

In children, use of supplemental oxygen was independently positively predicted by criteria A “Severe respiratory distress” and F “Altered conscious level, while association with criterion B “Increased respiratory rate” was weaker (Figure 1 upper panel). Use of mechanical ventilation was independently predicted by criteria C “Peripheral oxygen saturation ≤92% in air or being on oxygen”, E “Severe clinical dehydration or shock”, and F “Altered conscious level”, while association with criterion A “Severe respiratory distress” was weaker (Figure 1 middle panel). Decision to use intravenous antibiotics was predicted by criterion C “Peripheral oxygen saturation ≤92% on air or being on oxygen” but negatively predicted by criterion B “Increased respiratory rate” (Table 4). Length of stay >48 hours was independently predicted by criteria C. Length of stay ≥6 days was independently predicted by criteria B and C. Length of stay ≥12 days was independently predicted by criterion B (Table 4). Severe outcome (level 2/3 care or death) was predicted by criteria A “Severe respiratory distress”, C “Oxygen saturation ≤92%”, E “Severe clinical dehydration or shock” and F “Altered conscious level” (Figure 1 lower panel). Of note, mortality was only independently predicted in children by criteria G “Other clinical concern” [OR (95%CI); 4.24(1.33–13.47), p = 0.014] (Table 4).

In adults use of supplemental oxygen was independently positively predicted by criteria A “Severe respiratory distress”, B “Increased respiratory rate”, D “Respiratory exhaustion”, F “Altered conscious level” and negatively by G “Other clinical concern” (Figure 2 upper panel). Use of mechanical ventilation was independently positively predicted by criteria B “Increased respiratory rate”, C “Oxygen saturation ≤92%”, D “Respiratory exhaustion”, E “Severe dehydration or shock”, F “Altered conscious level” and G “Other clinical concern” (Figure 2 middle panel). Intravenous antibiotic use was independently predicted by criteria A, C, and F, but negatively predicted by criteria B and G (Table 5). Length of stay >48 hours was independently predicted by criteria A and C (Table 5). Length of stay ≥6 days was independently predicted by criteria B, C, F and G. Length of stay ≥12 days was independently predicted by criteria B, C, E and F. Severe outcome was independently predicted by criteria B “Increased respiratory rate”, “Oxygen saturation ≤92%”, D “Respiratory exhaustion”, E “Severe clinical dehydration or shock”, F “Altered consciousness” and G “Other clinical concern” (Figure 2 lower panel). Mortality was independently predicted by criteria B “Increased respiratory rate”, C “Oxygen saturation ≤92%”, and E “Severe clinical dehydration or shock” (Table 5).

Discussion

Hospital capacity and resources are strictly limited during a pandemic. Each of the CATs criteria had an independent role in identifying both use of clinical interventions that would in normal circumstances only be provided in hospital and adverse patient outcome measures, suggesting that had a “surge” threshold been reached in the UK during the 2009/10 pandemic, these triage tools would have been fit for purpose. Based on multivariable analyses none of the criteria were redundant. This is reassuring given that criteria A to F were selected for proven performance in identifying severe illness in previous well regarded studies and all criteria are recognised in current national and international guidance. [5], [8]–[17] This also suggests that the potential exists for creating a simple unweighted scoring system similar to CURB-65 or PMEWS using the CATs criteria. A direct comparison of the overall performance (whole tool rather than individual criteria) of CATs, CURB-65 and PMEWS as triage tools using the same cohort data set is the subject of a related paper. [22] CATs were, if anything, slightly better performing than CURB-65 and PMEWS during the first two waves of the 2009/10 pandemic in the UK.

Criteria C – “Peripheral oxygen saturation ≤92% breathing air, or on oxygen” performed well (in terms of magnitude of odds ratio and statistical power) in predicting use of resources and outcomes across all analyses in both adults and children. Hypoxaemia is particularly difficult to assess in children by clinical observation. Our findings support the routine use of age-appropriate peripheral oxygen saturation measurement in assessing severity of disease in patients of all ages. Given the relatively low cost of such monitoring equipment, this observation may be particularly pertinent in terms of resource poor settings in other countries.

Criterion G – “Causing other clinical concern to their own GP or clinical team” was developed by a workgroup of experts in primary care and emergency care. In consensus they insisted on including a subjective criterion that allowed “a clinician in primary care to uses their own judgement to refer a patient to hospital” and “a clinician in hospital to use their judgement to admit a patient” in cases where other criteria may not have been met. It might be expected that very many patients would meet this subjective criterion and that the criterion would have poor discriminatory power and no validity but quite the opposite was observed. In multivariate analysis criterion G was the only criteria that independently predicted death in children. In adults criterion G independently predicted length of stay ≥6 days, mechanical ventilation and combined severe outcome (level 2/3 admission or death). Meeting this criterion in children was often based on recognition of fever plus another sign, and in adults recognition of chest pain or fever plus another sign. This finding supports the role of clinical judgment, be it founded on acumen, experience or instinctive “gut feeling”, as a important independent predictor of serious illness in people of all ages. [18].

Conclusions

This study shows that CATs are potentially useful predictors of both use of hospital-based interventions and severe patient outcomes during an influenza pandemic. Each of the CATs criteria had a role in predicting a given outcome and none are redundant. It is also notable and novel that both paediatric and adults CATs were developed upon a common framework of assessment. Importantly this confers an element of equity in a situation where children and adults may have to compete for access to a limited common resource such as mechanical ventilators.

The CATs presented in this paper represent the first attempt to provide frontline healthcare professionals working across a national health service in communities and hospitals with high threshold triage criteria that could help make consistent, albeit difficult, decisions during extreme situations when health care demand exceeds capacity.

Supporting Information

Acknowledgments

We thank Dr Elaine Gadd and Mr Colin Armstrong, Department of Health, England for their invaluable contribution in relation to the FLU-CIN programme.

We gratefully acknowledge individuals who helped identify cases and collated clinical data: Alison Booth, Margaret Charlesworth, Sarah Rodenhurst, Angela Ballard and Alison Holmes at Imperial College Healthcare NHS Trust, London, UK; Sally Batham, Phayre Parkinson, Tracy Kumar, and Aiden Dunphy at the University Hospitals of Leicester NHS Trust, Leicester, UK; Anne Tunbridge, Patty Hempsall, Joyce Linskill, Aimee Turner, Sharon Grindle, Dawn Shevlin and Eric Moulds at Sheffield University Hospitals NHS Trust, Sheffield, UK; Elvina White, Elaine Scott, Jennifer Cater, Erica Sergi and Helen Hill at Alder Hey Children’s Hospital NHS Foundation Trust, Liverpool, UK; Deborah Fleetwood, Lorna Roche, Sarah Dyas, and Maria Boswell at the Royal Liverpool and Broadgreen University Hospital’s Trust, Liverpool, UK; Gillian Vernon, Gillian Houghton, Heather Longworth and Angela Kerrigan at Liverpool Women’s Hospital, Liverpool, UK; Sonia Greenwood, Gemma Thompson, Emily Jarvis, Charlotte Minter at the Nottingham University Hospitals NHS Trust, Nottingham, UK; Kristina Lum Kin, Jacqueline Daglish, Sam Hayton, and Gemma Slinn at Birmingham Children’s Hospital, Birmingham, UK; Michelle Lacey, Kevin Rooney, Karen Duffy, Anne Gordon, Eleanor Anderson, Hilary Davison, William Carman, Mark Cotton, Arlene Reynolds, Heather Murdoch, Karen Voy, Rosie Hague and Ali McAllister for their contribution to FLU-CIN in Scotland. Brian Smyth and Cathriona Kearns from the National Public Health Agency, Northern Ireland for identifying cases and facilitating data collection; Teresa Cunningham at the Southern Trust and Leslie Boydell at the Belfast Trust for facilitating data collection. Alemayehu Amberbir, Safaa Al-Badri, Baraa Mahgoob and Nachi Arunachalam at the University of Nottingham for data entry; also Graham Watson for database development and support. We also thank Professor Sir Gordon Duff and Professor Janet Darbyshire, for their support. Disclaimer: The views expressed are those of the authors and not necessarily those of their employers, the NHS, the NIHR or the UK Departments of Health. Appendices 1 and 2 are reproduced with permission of the Department of Health England. This report was approved for publication by the Department of Health, England.