Asthma is one of the most common chronic respiratory diseases worldwide [1]. Despite local and international agreement on the classification of asthma severity based on symptom frequency and control [1–3], there is still no broad consensus on standardized definitions of the severity of asthma exacerbations.

The SEPAR-ALAT classification addresses this gap by incorporating markers of acute severity, such as acidosis, the need for intubation, or previous respiratory arrest, which can be used to identify patients with different risk profiles [4,5]. However, its potential to identify patients likely to have different outcomes remains unclear. Therefore, our objective is to characterize the factors associated with the SEPAR-ALAT categories and to explore the prognostic utility of this classification in identifying patients likely to have different clinical outcomes, specifically in-hospital mortality and prolonged hospitalization. Adopting a risk-based approach, this study provides information for triage, monitoring, and escalation strategies based on the stratification of risk for asthma exacerbations.

Patient data from the multicenter, international, observational, retrospective, 10-year EAGLE study that characterized the clinical profile and healthcare management of a large cohort of patients from Spain and Latin America with severe asthma exacerbations were used for the purpose of this study [6]. The SEPAR-ALAT criteria were used to classify asthma exacerbations: patients categorized as very severe had to meet at least one of three criteria, including arterial pH <7.35, intubation, or respiratory arrest; all other patients were classified as severe [5]. The variables were analyzed descriptively, employing a complete case approach.

To identify factors independently associated with the “very severe” category, we constructed a multivariate logistic regression model. To mitigate the risk of model overfitting, a limited set of clinically justified variables was selected, including age, prior hospitalization history, and regular medical follow-up. A manual backward stepwise selection process was utilized to refine the model, with a P<.05 considered statistically significant. Potential multicollinearity between independent variables was assessed using the variance inflation factor, with values <5 considered acceptable to ensure the stability of the effect estimates. For further details regarding the multivariate models, the supplementary material should be consulted. Descriptive and multivariate analyses were performed using all available data for the relevant parameters, and no imputation methods were applied.

Data from 1437 patients from the EAGLE study who had been hospitalized for severe asthma exacerbation were analyzed using the SEPAR-ALAT asthma exacerbation criteria. The mean age was 47 years (IQR, 31–61), with the majority being women. A total of 260 (18%) patients were identified as having very severe asthma exacerbations. In this group, the proportion of patients previously classified with severe asthma was higher, while the proportion with intermittent and mild asthma was significantly lower. Regarding prior treatment, the very severe group showed a significantly higher use of SABAs and a different pattern of LABA use compared with the severe group. In addition, previous PEF and FEV1 values were lower in the very severe group. In terms of clinical events, previous hospitalizations were more frequent in the very severe group. See Table 1 for more details.

Multivariate regression analysis of the SEPAR-ALAT asthma exacerbation classification showed that patients who were not followed up were almost twice as likely to have their asthma exacerbation classified as very severe (OR, 1.78; P=.049). In addition, patients with severe asthma were around seven times more likely to be classified as very severe compared with those with intermittent asthma (OR, 7.36; P=.009). Concerning hospital admissions, the risk of being classified as very severe rather than severe was 88% lower in patients with no prior hospitalization compared with those with prior hospitalization (OR, 0.12; P<.001).

In terms of differences in outcome variables between asthma exacerbation groups, length of stay and death from asthma were significantly associated with the SEPAR-ALAT classification, with worse outcomes in the very severe group. Very severe patients were hospitalized for an average of 8.9 days vs 7.3 days for severe patients; the association was significant, although it weakened after adjusting for covariates (Fig. 1). In addition, the very severe group presented a 70-fold increased risk of death, with 93.8% of deaths occurring in this group (15/16; 5.8% in the very severe vs <0.1% in the severe group; adjusted OR, ≈25).

Fig. 1.

Days of hospital stay according to the SEPAR-ALAT asthma exacerbations severity classification. The values estimated (and their 95% confidence intervals) for each classification group using the multivariate stepwise regression model are shown in black. The mean values observed in the original data are displayed in red.

The differences observed between the groups in the multivariate analyses show that the SEPAR-ALAT classification (based on markers such as acidosis, intubation, or respiratory arrest) allows identification of in-hospital risk in severe asthma exacerbations. In the EAGLE cohort analyzed, the very severe category was associated with longer hospital stays (1.2 days; 95%CI, 1.1–1.3; P<.001) and, most notably, higher mortality (OR, 25.0; 95%CI, 3.9–482.0; P=.003). Furthermore, a prior history of hospitalizations and lack of follow-up were independently associated with the very severe classification, reinforcing the clinical validity of a protocol that integrates signs of acute physiological decompensation and absence of regular medical follow-up (OR, 8.3; P<.001; and OR, 0.6; P=.049, respectively). These findings are consistent with the results of the EAGLE study, which reported average hospital stays of around one week and regional differences in care [6–8]. The concentration of mortality in the very severe subgroup provides a prognostic detail that historical series could not fully capture because they lacked the specific very severe (life-threatening) classification included in SEPAR-ALAT.

Severe asthma exacerbations can progress to respiratory failure, profound respiratory acidosis, and respiratory arrest requiring intubation [9]. Therefore, an algorithm that includes these events may be more effective at predicting the risk of in-hospital severe events than algorithms based on symptoms or maintenance treatment.

Our results show that the SEPAR-ALAT classification further stratifies in-hospital risk even in patients classified as severe by chronic criteria, especially those with a history of hospitalization or no structured follow-up. While the variables defining the “very severe” category (arterial pH <7.35, intubation, and respiratory arrest) are intrinsically associated with mortality – introducing a degree of conceptual circularity – this classification functions as a standardized clinical staging tool for rapid triage and resource allocation. Importantly, its prognostic relevance extends beyond these defining events, as the “very severe” category was independently associated with longer hospital stays (8.9 vs 7.3 days; P<.001) and identifiable prehospitalization risk factors, supporting its ability to identify a distinct high-risk phenotype requiring intensified surveillance.

The limitations of the study include its retrospective design and the use of outdated data. However, the use of historical data does not diminish the clinical relevance of the findings, as the physiological markers defining the “very severe” category (acidosis and respiratory failure) are universal indicators of life-threatening asthma. The substantial association with mortality (OR, 25) suggests that the SEPAR-ALAT classification captures a fundamental risk profile that remains applicable today for identifying patients requiring maximal surveillance and escalation of care.

Furthermore, the low number of mortality cases (n=16) resulted in wide confidence intervals (OR, 25.0; 95%CI, 3.9–482.0), meaning the exact magnitude of the mortality risk should be interpreted with caution. Nevertheless, the fact that nearly 94% of deaths were concentrated in the “very severe” group reinforces the clinical signal of the classification.

The SEPAR-ALAT classification complements existing systems by identifying a high-risk clinical profile, thereby improving risk stratification and personalized management. Although these exploratory results must be confirmed through formal validation in prospective cohorts, incorporating this risk-based approach may improve management strategies in emergency settings.