Clinical features that predict the mortality risk in older patients with Omicron pneumonia: the MLWAP score

Yongjian Pei, Ting Li, Chen Chen, Yongkang Huang, Yun Yang, Tong Zhou, Minhua ShiInternal 2023 · DOI

TL;DR

Researchers have developed a new scoring system, the MLWAP score, to predict the 90-day mortality risk in older patients hospitalized with Omicron pneumonia. This score, derived from readily available clinical and laboratory data, demonstrates superior predictive ability compared to existing models and can help clinicians better stratify patients for resource allocation and improved outcomes.

What Problem Does This Paper Address?

The COVID-19 pandemic, particularly with the emergence of the highly transmissible Omicron variant, has placed immense strain on healthcare systems. Older adults are particularly vulnerable to severe outcomes, including pneumonia and mortality. While general pneumonia scoring systems exist (like CURB-65 and PSI), their applicability to COVID-19, especially the Omicron variant in older populations, is not well-established. There's a clear need for a specific, accurate, and easily applicable tool to identify older patients with Omicron pneumonia who are at high risk of death, enabling timely and targeted medical interventions.

Methods

The study employed a retrospective design, analyzing data from 227 older adult patients (≥ 60 years) hospitalized with confirmed Omicron pneumonia. The patient cohort was randomly split into a 70% training set (158 patients) for model development and a 30% test set (69 patients) for validation.

A multivariate logistic regression model was used on the training set to identify independent predictors of 90-day mortality. The regression coefficients (β) were then used to assign weights to these predictors, forming the MLWAP (Immunosuppression, Lactate, White blood cell count, Age, PaO2/FiO2) score. The model's performance was evaluated using:

  • Area Under the Receiver Operating Characteristic Curve (AUC): To assess discrimination.
  • Hosmer-Lemeshow Goodness-of-Fit Test: To evaluate calibration.
  • Calibration Curves: To visually assess how well predicted probabilities matched observed outcomes.
  • Bootstrap Resampling: To validate the calibration of the model.

The MLWAP score was then compared against established scores like CURB-65 and PSI.

Datasets

The study utilized a dataset of 227 hospitalized patients diagnosed with Omicron pneumonia between December 15, 2022, and January 16, 2023, at the Second Affiliated Hospital of Soochow University, China. This dataset was divided into a training set (n=158) and a test set (n=69). No new benchmark datasets were introduced; rather, existing scores (CURB-65 and PSI) were used for comparative analysis.

Key Insights and Ideas

  • Omicron's Impact on Older Adults: Despite initial perceptions of lower severity, the Omicron variant still posed a significant mortality risk to older adults, highlighting the continued need for vigilance and predictive tools.
  • Multifaceted Risk Factors: Mortality in this population isn't driven by a single factor. The MLWAP score effectively integrates clinical (age, immunosuppression) and laboratory (lactate, WBC, PaO2/FiO2) parameters, suggesting a complex interplay of physiological responses.
  • Clinical Utility of Simple Metrics: The MLWAP score relies on readily available data points, making it practical for immediate clinical use in busy hospital settings, unlike more complex scoring systems.
  • Superiority over Existing Scores: The MLWAP score demonstrated better predictive performance than CURB-65 and PSI, indicating its enhanced relevance for Omicron pneumonia in older patients.

Key Results

The MLWAP score was developed based on five key predictors:

  • Immunosuppression (weighted score: 5)
  • Lactate ≥ 2.4 (weighted score: 3)
  • White blood cell count ≥ 6.7 × 10<sup>9</sup>/L (weighted score: 2)
  • Age ≥ 77 years (weighted score: 2)
  • PaO<sub>2</sub>/FiO<sub>2 </sub>≤ 211 (weighted score: 1)

The total score ranges from 0 to 13.

  • Predictive Performance: The MLWAP score achieved an AUC of 0.852 in the training set and 0.875 in the test set, indicating strong discriminative ability.
  • Calibration: Calibration curves showed a good fit, meaning the predicted probabilities closely aligned with observed mortality rates.
  • Sensitivity and Specificity: The model achieved a sensitivity of 0.849 and a specificity of 0.714.
  • Comparative Performance: The MLWAP score's AUROC (0.859) was superior to CURB-65 (0.788) and PSI (0.801).
  • Risk Stratification: Patients were categorized into low (0-7 points), medium (8-10 points), and high (11-13 points) risk groups, with observed mortality rates of 9.10%, 38.30%, and 84.80% respectively in the test set.

Limitations and Future Work

The authors acknowledge several limitations:

  • Retrospective Design: This inherent limitation means potential biases and confounding factors (e.g., patients refusing maximum intervention) could not be fully controlled.
  • Single-Center Study: The findings are based on data from one hospital, which may limit generalizability.
  • Small Sample Size: While 227 patients is a reasonable number, a larger, multi-center study would improve the robustness and universality of the MLWAP score.

Future work should focus on:

  • External Validation: Verifying the MLWAP score's performance in larger, diverse international datasets.
  • Prospective Studies: Conducting prospective trials to confirm the score's utility in real-time clinical decision-making.
  • Integration into Clinical Practice: Exploring how the MLWAP score can be seamlessly integrated into electronic health records and clinical workflows to aid in patient management.

Summary updated

Abstract

In December 2022, the Chinese suffered widespread Omicron of SARS-CoV-2 with variable symptom severity and outcome. We wanted to develop a scoring model to predict the mortality risk of older Omicron pneumonia patients by analyzing admission data. We enrolled 227 Omicron pneumonia patients aged 60 years and older, admitted to our hospital from December 15, 2022, to January 16, 2023, and divided them randomly into a 70% training set and a 30% test set. The former were used to identify predictors and develop a model, the latter to verify the model, using the area under the receiver operating characteristic curve (AUC), the Hosmer-Lemeshow goodness-of-fit test, a calibration curve to test its performance and comparing it to the existing scores. The MLWAP score was calculated based on a multivariate logistic regression model to predict mortality with a weighted score that included immunosuppression, lactate ≥ 2.4, white blood cell count ≥ 6.70 × 10<sup>9</sup>/L, age ≥ 77 years, and PaO<sub>2</sub>/FiO<sub>2 </sub>≤ 211. The AUC for the model in the training and test sets was 0.852 (95% CI, 0.792-0.912) and 0.875 (95% CI, 0.789-0.961), respectively. The calibration curves showed a good fit. We grouped the risk scores into low (score 0-7 points), medium (8-10 points), and high (11-13 points). This model had a sensitivity of 0.849, specificity of 0.714, and better predictive ability than the CURB-65 and PSI scores (AUROC = 0.859 vs. 0.788 vs. 0.801, respectively). The MLWAP-mortality score may help clinicians to stratify hospitalized older Omicron pneumonia patients into relevant risk categories, rationally allocate medical resources, and reduce the mortality.
BibTeX
@inproceedings{pei2023clinical,
  author = {Yongjian Pei and Ting Li and Chen Chen and Yongkang Huang and Yun Yang and Tong Zhou and Minhua Shi},
  title = {{Clinical features that predict the mortality risk in older patients with Omicron pneumonia: the MLWAP score}},
  booktitle = {Internal},
  year = {2023},
  doi = {10.1007/s11739-023-03506-2},
}
Figure