Published: 20/03/2003, Volume II3, No. 5847 Page 32 33
An evaluation system using easily available data can determine the severity of patients'condition - and thereby predict their likely length of stay in hospital.
Christian Subbe and colleagues report
Performance management comparisons between hospitals in England and Wales are made by analysing clinical episodes based on diagnostic ICD-codes (international classification of diseases). The current system does not take into account the fact that differences of severity of illness between patients with the same disease will contribute significantly to the length of their in-hospital stay and their risk of admission to high dependency or intensive care units.
The modified early warning score (MEWS) is a standardised evaluation of parameters that are routinely recorded by nursing and medical staff at the bedside. In patients admitted to acute emergency medical care, high scores translate into poor prognosis in terms of increased mortality and need for admission to critical care. Our study shows that MEWS can also be used to predict the length of in-hospital stay. Using it systematically would allow inter-hospital comparisons of length of stay as well as highlighting likely shortages in availability of hospital beds.
Critical care outreach has been recommended by the Department of Health as a tool to improve early recognition of patients at risk of catastrophic deterioration.
1The use of simple triage systems and knowledge-sharing from critical care professionals is recommended to improve patient care and clinical governance, avert unnecessary intensive care admissions and use critical care resources in the best way.
2MEWS is a triage tool that uses routinely available data, including systolic blood pressure, pulse rate, respiratory rate, temperature and level of consciousness using the AVPU score with A for 'alert', V for 'responding to voice', P for 'responding to pain' and U for 'unresponsive' (see table). In order to calculate the MEWS-weighted score the single parameters are added up.MEWS is the sum of the scores for single parameters and operates along the assumption that the higher the score, the sicker the patient.
Based on the value of MEWS, patients can be classified into groups at lower or higher risk of catastrophic deterioration. It can therefore be used to aid medical, nursing and critical care staff to prioritise their workload towards the care of the most vulnerable patients.
MEWS is a validated tool to identify medical patients at risk of catastrophic deterioration (death, intensive care admission or highdependency unit admissions).
3However, it is not known how the admission of patients with higher scores affects their length of hospital stay. This is particularly relevant during winter periods where the admission of large numbers of acute medical admissions has knock-on effects for bed capacity in other areas.
Data was collected for acute medical admissions during a four-month period in 2000 and 2001.
Patients were admitted to the 56-bed acute medical admissions unit of a district general hospital covering a population of 300,000 in north east Wales. Patients admitted directly to critical care areas, elective admissions to other wards, patients admitted for terminal care and re-admissions within three months were excluded from analysis.
Nursing staff and data collection clerks obtained data for all patients admitted to the medical admissions unit. This included epidemiological data such as date of birth and date of admission to hospital and basic physiological data as outlined above. From the physiological data the MEWS score was calculated.
Results For 1,974 patient episodes admissions, MEWS and discharge date were analysed.
The overall median length of hospital stay was six days.
Patients were classified into high and low risk of catastrophic deterioration by means of MEWS. Patients with a MEWS of 0-2 were classified as low risk, with a MEWS of 3-4 as intermediate risk and patients with a MEWS of 5 or above were classified as high risk.Across all emergency admissions, higher MEWS on admission was significantly associated with increasing length of stay.
More specifically, patients with MEWS of 0 had a median length of stay in hospital of three days, patients with a MEWS of 5 had a median length of stay of eight days.
However, there were important differences in this relationship between MEWS and length of stay in patients of different age groups and gender.
Length of stay was compared in patients of different age groups: there were 1,459 patients below the age of 75, of which 64 died, and 841 patients of 75 and over, of which 140 died.
Patients aged under 75 stayed a median of three days as compared with patients 75 years and over who stayed a median of 10 days.
Women stayed a median of six days and men a median of five days.
There was a correlation between length of stay and MEWS in the younger patients group. In comparison, the correlation was not statistically significant in the older patient group. In a multiple regression analysis the impact of age and MEWS but not sex on length of stay was significant.
Our study found that an increase in a simple and readily available clinical marker on emergency admission was associated with an increase in length of in-hospital stay: patients with higher MEWS had a significantly longer stay in hospital.
The relationship between severity of disease and length of stay was only weak in the elderly patients in whom other factors, including social, mental and functional parameters will better predict length of stay.
4Indeed, the study directly supports the perception that the frail, dependent elderly patient may be admitted on the acute medical take for reasons other than acute illness.
5The implications from the results of our study are perceivable in three areas. First, although more complex models may predict length of hospital stay more accurately, the advantage of MEWS is its simplicity in using routine clinical data for calculation.
6Our data suggests that MEWS can be used to aid comparison between hospitals and between time periods of data relating to acute medical admissions and length of stay.Over larger populations, MEWS could be used to benchmark hospitals for their length of stay of patients with specific diagnoses, but with an additional adjustment for severity of illness.
Second, this simple estimate of severity of illness and its relationship with length of stay can be used to inform hospital bed managers.
The admission of a larger number of seriously ill medical patients over a single day will have predictable knock-on effects on the availability of beds for routine and planned admissions.On the other hand, the admission of less sick patients over a day would allow for more intelligent organisation of planned elective admissions.
Third, our data strongly suggests that patients with high MEWS scores on admission stay longer and use more healthcare resources.They will also require more time and care from nursing staff.
Analysing admission MEWS profiles would allow greater understanding by an acute trust of, for example, resource allocations, numbers of beds needed and impact of different pre-hospital admission prevention schemes.
MEWS is particularly attractive because the data used for its calculation is already available and can be linked easily with other areas of performance-management and clinical risk-management.
The modified early warning score can be used to help predict how long a patient will stay in hospital.
MEWS is a validated tool and can be calculated relatively easily from routine data.
A MEWS score of 5 (high risk) had a median length stay of eight days, compared with a score of 0-2 (low risk) of three days; however, there were variations according to age groups and gender.
1Department of Health. Comprehensive critical care: a review of adult critical care services. London:
Department of Health, 2000.
2Morgan RJM, Williams F, Wright MM. Clin Intensive Care. 1997; 8: 100
3Subbe CP, Kruger M, Rutherford P, Gemmel L. Patients at risk: Validation of a modified Early Warning Score in Medical Admissions. QJM 2001;94:521-6.
4Di Iorio A, Longo A, Mitidieri Costanza A, Palmerio T et al.Factors relating to the length of inhospital stay of geriatric patients. Aging (Milano).1999;11:150-4.
5Flintoft VF, Williams JI, Williams RC, Basinski AS et al. The need for acute, subacute and nonacute care at 105 general hospital sites in Ontario. CMAJ.1998;158:1289-96.
6Roe CJ, Kulinskaya E, Dodich N, Adam WR.
Comorbidities and prediction of length of hospital stay. Aust NZJM.
Dr Christian Subbe is specialist registrar in thoracic medicine, Wrexham Maelor Hospital, Jonathan Falcus is directorate manager, department of medicine, Wrexham Maelor Hospital, Dr Peter Rutherford is senior lecturer and honorary consultant, department of nephrology, University of Wales College of Medicine and Dr Les Gemmell is director of critical care services, Wrexham Maelor Hospital.