To the Editors:
We read with interest the endorsement of “barefoot medicine” by Ewig et al. 1 in a recent issue of the European Respiratory Journal. As with Niederman et al. 2, we recognise the limitations of CURB-65 (confusion, urea >7 mmol·L−1, respiratory rate ≥30·min−1, low blood pressure, and aged ≥65 yrs) scoring but the importance of its simplicity and ease of use. As part of the planning for a potential H5N1 influenza pandemic, using Dept of Health and Health Protection Agency projections 3, we have been forced to acknowledge that our urban emergency department, which normally sees ∼250 patients·day−1, will see 450 excess attenders·day−1 with influenza symptoms at a pandemic peak. We aimed to develop a rapidly applicable, purely clinical scoring system for use in primary and secondary care, to identify those in need of hospital admission and to reassure those fit for discharge. We suggest that the ideal score should reflect acute physiological derangement, as well as accommodating age, comorbidities and social factors, and could be used to triage and track for admission, intensive care unit (ICU) treatment and mortality. We believe that our proposed system has gone some way towards addressing this.
We modified our hospital pandemic medical early warning score (PMEWS) 4 to include transcutaneous oxygen saturation. We also concur with Ewig et al. 1 and Niederman et al. 2 that comorbidities and social factors have to be taken into account when making admission and discharge decisions, and our score incorporates an extra point for being aged ≥65 yrs and another single point for any of the following: 1) social isolation (defined as living alone or having no fixed abode); 2) chronic disease (respiratory, cardiac, renal, diabetes mellitus or immunosuppression of any cause); or 3) performance status of limited activity or worse (modified Karnofsky >2 5).
The validation of 195 adult patients (101 aged <65 yrs) with a diagnosis of lower respiratory tract infection presenting to our emergency department (South Manchester University Hospitals Trust, Manchester, UK) between February and December 2005 showed good discrimination for the physiological section of the score, which was further improved by the addition of age and social factors. We retrospectively calculated PMEWS, CURB-65 and CRB-65 scores from emergency department medical and nursing notes, and constructed receiver-operating characteristics (ROC) curves for the prediction of admission (fig. 1⇓). PMEWS without the arterial oxygen saturation component is shown as we recognise that not all primary care providers will have access to a pulse oximeter.
Receiver-operating characteristics curve for admission to hospital. Diagonal segments are produced by ties. ––––: pandemic medical early warning score (PMEWS) 0.953 (0.927–0.978); -----: PMEWS without arterial oxygen saturation 0.937 (0.906–0.967); – – –: CURB-65 (confusion, urea >7 mmol·L−1, respiratory rate ≥30·min−1, low blood pressure, and aged ·65 yrs) 0.858 (0.806–0.909); – - –: CRB-65 (confusion, respiratory rate ≥30·min−1, low blood pressure, and aged ≥65 yrs) 0.818 (0.760–0.877).
We extended this to assess the value of the PMEWS score in predicting requirements for higher levels of care. Figure 2⇓ shows the ROC curves for discrimination of need for high dependency or ICU care amongst 91 patients admitted with community-acquired pneumonia to our 855-bed secondary and tertiary care hospital from February to December 2005. A further number of patients (n = 19) were omitted from this analysis as invasive care and cardiopulmonary resuscitation were deemed inappropriate.
Receiver-operating characteristics curve for admission to level 2/3 care. Diagonal segments are produced by ties. ––––: pandemic medical early warning score 0.856 (0.768–0.943); – – –: CURB-65 (confusion, urea >7 mmol·L−1, respiratory rate ≥30·min−1, low blood pressure, and aged ≥65 yrs) 0.667 (0.549–0.785); -----: CRB-65 (confusion, respiratory rate ≥30·min−1, low blood pressure, and aged ≥65 yrs) 0.631 (0.497–0.765).
We suggest that as the pandemic medical early warning score is not disease specific, it can provide a more accurate assessment of need for hospital admission, clinical deterioration and improvement, and will be a valuable track and triage tool in the event of an influenza pandemic. Further validation is ongoing.
- © ERS Journals Ltd