Table 1.
The Modified Early Warning Score.
This table shows the vital sign parameters comprising the Modified Early Warning Score. Adapted from Subbe et al, 2001 [12].
Table 2.
Baseline characteristics of enrolled patients.
* Baseline characteristics of 452 study patients are presented here, as collected at the time of enrollment. Attendants are family and friends of patients who stay with them in hospital to provide them with food, personal care, and transport. They are also responsible for obtaining all prescribed medications and test results.
Fig 1.
Distribution of MEWS across all patients.
Modified Early Warning Scores were calculated for all patients at the time of study enrollment, based on vital signs recorded by research personnel. Scores ranged from 0 to 9, with a median of 2 (IQR 1–3). Mortality increased with higher MEWS.
Fig 2.
Distribution of MEWS across all patients who survived.
Modified Early Warning Scores were calculated for all patients, and the majority of patients who survived had a MEWS of 1, as illustrated in this distribution of MEWS across all surviving patients.
Fig 3.
Distribution of MEWS across all patients who died.
The distribution of MEWS across patients who did not survive illustrates that MEWS ≥4 was documented in 21.5% of patients; 11.7% of patients had a MEWS ≥5.
Table 3.
Factors Associated with Mortality: Univariate Analysis.
MEWS ≥4 and ≥5 were significantly associated with 7-day mortality. Of patients’ baseline features, HIV positive status; admission with a medical diagnosis, and documented blood pressure measurement in the emergency department were also significantly associated with mortality.
Table 4.
Factors Associated with Mortality: Multivariable Analysis.
Of the 4 factors significantly associated with mortality in the univariate analyses, MEWS ≥ 5, medical admission, and systolic blood pressure measurement in the ER were included in the backward stepwise selection procedure in this multivariable analysis. HIV positive status was not included due to the high proportion of missing values. In the final model, medical admission and MEWS ≥ 5were independently associated with mortality.
Table 5.
Calculated indicators for MEWS with cutoffs of 4 and 5.
Two-by-two tables showing derivation of prognostic indicators sensitivity, specificity, positive predictive value (PPV), positive likelihood ratio (PLR), and number needed to evaluate (NNE) based on MEWS ≥4 and ≥5, in the study population. Number needed to evaluate refers to the number of patients required to evaluate to detect one outcome; it is an estimate of the effort-yield of each alert [22].