This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIRx Med, is properly cited. The complete bibliographic information, a link to the original publication on https://med.jmirx.org/, as well as this copyright and license information must be included.
The modified early warning score (MEWS) is an objective measure of illness severity that promotes early recognition of clinical deterioration in critically ill patients. Its primary use is to facilitate faster intervention or increase the level of care. Despite its adoption in some African countries, MEWS is not standard of care in Ghana. In order to facilitate the use of such a tool, we assessed whether MEWS, or a combination of the more limited data that are routinely collected in current clinical practice, can be used predict to mortality among critically ill inpatients at the Korle-Bu Teaching Hospital in Accra, Ghana.
The aim of this study was to identify the predictive ability of MEWS for medical inpatients at risk of mortality and its comparability to a measure combining routinely measured physiologic parameters (limited MEWS [LMEWS]).
We conducted a retrospective study of medical inpatients, aged ≥13 years and admitted to the Korle-Bu Teaching Hospital from January 2017 to March 2019. Routine vital signs at 48 hours post admission were coded to obtain LMEWS values. The level of consciousness was imputed from medical records and combined with LMEWS to obtain the full MEWS value. A predictive model comparing mortality among patients with a significant MEWS value or LMEWS ≥4 versus a nonsignificant MEWS value or LMEWS <4 was designed using multiple logistic regression and internally validated for predictive accuracy, using the receiver operating characteristic (ROC) curve.
A total of 112 patients were included in the study. The adjusted odds of death comparing patients with a significant MEWS to patients with a nonsignificant MEWS was 6.33 (95% CI 1.96-20.48). Similarly, the adjusted odds of death comparing patients with a significant versus nonsignificant LMEWS value was 8.22 (95% CI 2.45-27.56). The ROC curve for each analysis had a C-statistic of 0.83 and 0.84, respectively.
LMEWS is a good predictor of mortality and comparable to MEWS. Adoption of LMEWS can be implemented now using currently available data to identify medical inpatients at risk of death in order to improve care.
Critical illness is a leading cause of morbidity and mortality in sub-Saharan Africa, including Ghana [
One of the main reasons why patients deteriorate and die in hospitals is delayed recognition of illness severity in understaffed inpatient wards. Early warning tools to help identify patients at the highest risk of death could help countries like Ghana with resource allocation and clinical decision making (
Conceptual framework showing predictors of in-hospital mortality and the role of the modified early warning score (MEWS) among ill patients.
Multiple studies have shown that critical illness and serious adverse events in hospitalized patients are preceded by signs of clinical deterioration in up to 80% of those affected [
MEWS is a commonly used illness severity score that is calculated by combining five physiologic bedside parameters: systolic blood pressure, heart rate, respiratory rate, temperature, and level of consciousness assessed by the AVPU (alert, [responds to] voice, [responds to] pain, unresponsive) scale or RASS (Richmond Agitation Sedation Scale) score. These four vital signs and the observation of consciousness are individually scored and summed to yield a combined score between 0 and 14, with higher scores representing increased illness severity.
In a systematic review conducted by Smith et al [
This study sought to validate the use of MEWS as a clinical decision-making tool to improve early identification of hospitalized medical patients at increased risk for death in Ghana. In addition, since level of consciousness is not routinely recorded in current clinical practice, we aimed to investigate the predictive utility of a limited MEWS (LMEWS) calculation based on vital signs alone. Most studies in similar settings have found that the level of consciousness is generally high (ie, the patient is well oriented) even when other aspects of the MEWS value are abnormal [
This was a retrospective chart review study of hospitalized medical patients, aged ≥13 years, admitted to the Korle-Bu Teaching Hospital in Accra, Ghana. The Korle-Bu Teaching Hospital is the national hospital of Ghana and the leading tertiary care referral center in the country [
Flow chart demonstrating the creation of the modified early warning score (MEWS) cohort. LMEWS: limited MEWS.
Demographic data were collected to analyze covariates. Patients’ vital signs recorded at 48 hours after admission were recoded and scored to generate the LMEWS value, using thresholds as previously described (
Our study was based on the conceptual framework depicted in
Due to the confidential nature of patient information, and the need to protect anonymity and obtain consent during health record reviews, ethical approval and waiver of documented permission was obtained from the Institutional Review Board (IRB) of Johns Hopkins University, and from the Scientific and Technical Committee (KBTH-STC 00017/2019) and the IRB of the Korle-Bu Teaching Hospital. Although reporting was anonymous, patients’ records were not, so researchers involved in data collection and handling also signed a confidentiality clause.
Scoring scale for the modified early warning score (MEWS) adopted form Subbe et al [
Physiological parameter | MEWS value | ||||||
|
3 | 2 | 1 | 0 | 1 | 2 | 3 |
Systolic blood pressure (mmHg) | <70 | 71-80 | 81-100 | 101-199 | —a | ≥200 | — |
Heart rate (bpm) | — | 41-50 | 41-50 | 51-100 | 101-110 | 111-129 | ≥130 |
Respiratory rate (cpm) | — | — | — | 9-14 | 15-20 | 21-29 | ≥30 |
Temperature (°C) | — | — | — | 35-38.4 | — | ≥38.5 | — |
AVPUb score | — | — | — | Alert | Reacting to voice | Reacting to pain | Unresponsive |
aNot applicable.
bAVPU: alert, voice, pain, unresponsive.
Data were analyzed using STATA (version 15.1, StataCorp LLC). The estimated sample size was determined a priori based on work by Kyriacos et al [
The sample comprised 112 patients admitted for medical reasons during the study period. Of these, 62% (69/112) were male with a mean age of 47 years (SD 17.5), and 38% (43/112) were female with a mean age of 52 years (SD 20) (
At 48 hours post admission, patients’ mean systolic blood pressure was 125 mmHg (SD 2.9), average pulse rate was 91 mmHg (SD 2), mean axillary temperature was 36.9°C (SD 0.1), and average respiratory rate was 24 cpm (SD 4.7). Only temperature and respiratory rate were individually associated with mortality (
A significant MEWS was associated with a relative risk of 2.01 (95% CI 1.33-3.04) for death in the univariable analysis, while a significant LMEWS had a relative risk of 2.19 (95% CI 1.46-3.30) in the univariable analysis (
Showing baseline characteristics.
Characteristic | Survival to discharge (n=66) | Death in hospital (n=46) | |||
Sex (male), n (%) | 45 (68.2) | 24 (52.2) | .09 | ||
|
<.001 | ||||
|
25-64 | 46 (69.7) | 27 (58.7) |
|
|
|
≥65 | 7 (10.6) | 18 (39.1) |
|
|
|
.01 | ||||
|
Cardiopulmonary | 15 (23.1) | 13 (28.3) |
|
|
|
Neuroendocrine | 11 (16.9) | 18 (39.1) |
|
|
|
Hemaoncological | 11 (16.9) | 1 (2.2) |
|
|
|
|
||||
|
Systolic blood pressure (mmHg) | 127.8 (29.4) | 120.7 (32.1) | .23 | |
|
Pulse rate (bpm) | 89 (17.6) | 94 (18.1) | .17 | |
|
Axillary temperature (°C) | 36.7 (0.7) | 37.3 (1.2) | .002 | |
|
Respiratory rate (cpm) | 23 (4.7) | 25 (6.9) | .03 | |
|
Average length of admission | 7 (6.3) | 8 (7) | .60 |
a
Multivariable logistic regression of death using full modified early warning score (MEWS) and the limited MEWS (LMEWS).
Covariate | MEWS, odds ratio (95% CI) | LMEWS, odds ratio (95% CI) |
Age | 1.08 (1.04-1.12) | 1.08 (1.04-1.12) |
Sex (male) | 0.44 (0.16-1.23) | 0.40 (0.14-1.13) |
MEWS (significant) | 6.33 (1.96-20.49) | 8.22 (2.45-27.56) |
Duration of admission | 0.99 (0.93-1.07) | 1.01 (0.94-1.08) |
Diseased organ system | 0.59 (0.31-1.13) | 0.59 (0.31-1.12) |
The death rate calculated by the Poisson regression after adjusting for only age was 2.02 (95% CI 1.40-2.91) times higher in patients with a significant MEWS compared to those with a nonsignificant MEWS. The death rate for a significant MEWS value using LMEWS was 2.13 (95% CI 1.48-3.07) times that of nonsignificant MEWS after adjusting for age.
In the multivariable predictive model adjusting for age, sex, duration of admission, admission to the ICU, organ system involved, and comorbidities, the odds of death among patients with a significant MEWS was 6.33 (95% CI 1.96-20.50) times that of patients with a nonsignificant MEWS. The death rate among patients with a significant LMEWS was 8.2 (95% CI 2.5-27.6) times that of patients with a nonsignificant LMEWS in the multivariable analysis. The best multivariable regression model was selected based on the Akaike Information Criteria, with a value of 116.4. The odds of death for every year increase in age was 8% (OR 1.08, 95% CI 1.04-1.12). Other covariates were not statistically significant.
Both MEWS and LMEWS were found to have good discrimination based on the ROC curves, with a C-statistic of 0.833 and 0.838, respectively (
Sensitivity analyses using a significant MEWS or LMEWS cut-off score of ≥5 yielded a multivariable OR of 12.4 (95% CI 2.5-61.2) and 15.1 (95% CI 2.5-91.8), respectively. The ROC curves for MEWS and LMEWS was found to be 0.838 and 0.840, respectively, when a cut-off of ≥5 was adopted, as captured in
Receiver operator characteristic (ROC) curve for the modified early warning score (MEWS) using a cut-off of 4.
Receiver operator characteristic (ROC) curve for the limited modified early warning score (LMEWS) using a cut-off of 4.
Receiver operator characteristic (ROC) curve for the modified early warning score (MEWS) using a cut-off of 5.
Receiver operator characteristic (ROC) curve for the limited modified early warning score (LMEWS) using a cut-off of 5.
MEWS has been validated in several settings as a robust predictor of both clinical deterioration and death in hospital [
Serious adverse events and some portion of in-hospital mortality can be prevented by limiting human error, such as failure to recognize the early warning signs of a deteriorating patient or failure to act on this information in a timely manner [
We found that, in this setting, having a LMEWS value of 4 or greater was highly associated with in-hospital mortality. The area under the curve (AUC) of 0.84 for the LMEWS is consistent with good model accuracy in the discrimination of patients who are critically ill. The combination of LMEWS with clinical judgment is therefore likely to be as effective in Ghana as it has been in other similarly resourced settings [
The standard inpatient vital signs monitoring charts used in many Ghanaian hospitals includes a 4-hourly graphic to plot temperature, pulse rate, respiratory rate, and blood pressure. Additional parameters may also be serially recorded in some instances or centers; however, the typical bedside observation chart does not record the level of consciousness for patients, as captured in the MEWS by including either the AVPU or RASS score.
Although the original description defined a significant MEWS as any single score ≥5, or any increase of 2+ points in patients with initial scores above 5, a cut-off of 4 was adopted for this study [
Interestingly, using MEWS or LMEWS with a cut-off of ≥5 did not only yield higher discrimination, based on the C-statistics, but also had better calibration in terms of correctly assessing the risk of disease severity. Based on the receiver operating characteristics and the Hosmer-Lemeshow goodness-of-fit test, LMEWS with a cut-off of ≥5 was superior to both MEWS and LMEWS with a cut-off of ≥4.
Encouraging complete, accurate documentation and a standardized interpretation of vital signs with appropriate actions by nurses, doctors, and other allied staff can potentially improve the outcomes of patients admitted to hospitals, even in a setting that lacks rapid response teams. Many interventions such as fluids or antibiotics do not require advanced equipment or costly supplies, making the implementation of the afferent arm of a rapid response system important even in settings where the efferent arm is more limited [
This study is subject to all the limitations of a single-center, retrospective chart review. Sources of bias include the potential for differential clinical care based on perceived patient status in the absence of a standardized rapid response team or protocol. In addition, the study only examined vital signs collected at a single time point for each patient. Changes in serially measured physiological parameters were not evaluated. A study published by Ludikhuize et al [
More prospective research is needed to help define the utility of LMEWS for physicians looking to allocate resources and develop rapid response teams that can act on predictive information to improve patient outcomes and patient care.
This study was the first to examine the ability of an early warning system to predict inpatient mortality based on routinely collected clinical data in a low-resource setting. Early recognition of clinical status decline is critical even in low-resource settings, where bedside interventions may prevent ICU admissions and disease complications including death. Though the MEWS system provides good discrimination, the LMEWS provides better discrimination and calibration in the prediction of mortality and can identify critical illness among inpatients with primarily medical diagnoses. Additional prospective studies will be useful to validate LMEWS among other categories of inpatients and to investigate its impact on health resource allocation and clinical outcomes in low-resource settings.
area under the curve
alert, voice, pain, unresponsive
intensive care unit
Institutional Review Board
incidence rate ratio
limited modified early warning score
modified early warning score
odds ratio
Richmond Agitation Sedation Scale
receiver operating characteristic
This research was funded by the Leroy Burney Family Fund, which had no role in the design, collection, interpretation of data, or writing and submission of the manuscript.
We are grateful to the Korle-Bu Teaching Hospital, medical department staff, and clientele for making patient charts available for data collection, as well as Drs Benjamin Sena Fenu and Oforiwaa Amoah, who contributed to data collection and on-site follow-up for ethical and technical committee approval; Victor Nortey, who was instrumental in guiding us through the ethical review processes at the Korle-Bu Teaching Hospital; and George Mwinnyaa and Seth Bennett, who reviewed the statistical analysis.
EJA was responsible for the concept, study design, partial data collection, analysis and interpretation, and writing of the manuscript. PA, JSRM, MC, and SES participated in concept development, critical revision, and review of the manuscript. PA, JSRM, and MC also served as academic mentors, while SES was the on-site preceptor as well.
None declared.