In this large multicenter cohort, we identified six easily available factors that are associated with in-hospital mortality and derived the new ABCDMP score in patients with AECOPD and CVDs. To our knowledge, this was the first clinical score to assess individuals’ risk of poor prognosis in patients with AECOPD and CVDs. The clinical score can be calculated early after patient presentation and performs well in predicting mortality in patients with AECOPD and CVDs. It is superior to the existing optimal risk scores for predicting adverse outcomes in AECOPD. Additionally, the new score had moderate predictive performance for long-term mortality and could discriminate between patients at low, medium and high risk of mortality in patients with AECOPD and CVDs.
CVDs risk in COPD is very high compared with the general population due to lung hyperinflation, pulmonary hypertension and systemic inflammation [13, 23]. In a large cohort of patients with COPD admitted to a Veterans Administration hospital, the prevalence of coronary artery disease was 33.6%, significantly higher than the 27.1% prevalence seen in a matched cohort without COPD . In a large and possibly most conclusive systematic review, Chen et al. found a nearly 2.5-fold increased risk of cardiovascular disease overall and a two- to five-fold higher risk of major cardiovascular disease types (ischemic heart disease, cardiac dysrhythmia, heart failure, diseases of the pulmonary circulation, and arterial diseases) in patients with COPD . Hospitalized acute exacerbations are associated with mortality of cardiovascular events in COPD. Using a health insurance research database in Taiwan, Wang et al. found that the 90-day mortality rates of acute myocardial infarction and ischemic stroke in COPD patients without acute exacerbations were significantly lower than those in patients with hospitalized acute exacerbations (33.9% vs. 44.6% and 13.8% vs. 20.3%; all p < 0.001) . Given the high morbidity and mortality, it is necessary to establish valid tools for the risk stratification of patients with AECOPD and CVDs.
In the multivariable analysis, we found that independent risk factors for in-hospital mortality in patients with AECOPD and CVDs were age > 75 years, pulse > 109 beats per minute, DBP ≤ 60 mmHg, altered mental status, BUN > 7 mmol/L and consolidation. These factors have been established as being associated with outcome in AECOPD/CVDs in previous studies [19, 26,27,28,29,30]. For instance, a study by Byrd et al. showed that a higher pulse rate was more linearly related to increases in all-cause mortality and cardiovascular events in patients with COPD . Similar to this former study, we also observed a significant reduction in DBP in patients who died during hospitalization. Lower DBP on admission was reported to be associated with increased mortality and excess cardiovascular events in previous studies [27, 29]. Serum BUN > 7 mmol/L was also proven to be an independent risk factor for in-hospital mortality in patients with AECOPD and CVDs, and the same results were shown in our study as well [26, 28, 30].
As previously highlighted, previous studies have proposed several predictive tools for mortality in patients with AECOPD, such as the BAP-65 score , the CURB-65 score , the DECAF score [19, 20], and the NIVO score. BAP-65 is a disease-specific severity-of-illness score for AECOPD, which was designed to use only information that is generally available to physicians at the time of patient presentation. CURB-65 is simple to apply but was designed for use in patients with pneumonia. DECAF is a clinical prediction tool, incorporating indices routinely available at the time of hospital admission and can stratify patients hospitalized with AECOPD into clinically relevant risk. The NIVO score allows for accurate risk stratification of patients admitted to the hospital with AECOPD complicated by acidaemia and acute hypercapnic respiratory failure who required assisted ventilation. All these scoring systems were derived from large cohorts and showed good predictive values. However, the performance of the scores in predicting in-hospital mortality of patients with AECOPD and CVDs has not been commonly reported. In our study, we found that the discriminative power of the predictive risk scores in predicting in-hospital mortality of patients with AECOPD and CVDs was acceptable but dissatisfactory after ROC curve analysis (AUC, 0.619–0.775).
In our study, ROC curve analysis showed an excellent discriminate power of our ABCDMP model (AUC = 0.847, 95% CI, 0.805–0.890; P < 0.001), after which a validation of our ABCDMP model was performed using a multicenter cohort, revealing the validated AUC of our ABCDMP of 0.811. No in-hospital mortality occurred among patients with AECOPD and CVDs who scored < 2 with our ABCDMP, while the in-hospital mortality of patients who scored > 4 with our ABCDMP was 12.84% during hospitalization. Moreover, the AUC of our ABCDMP score was statistically greater than those of the BAP-65, CURB-65, DECAF, and NIVO scores (AUC BAP-65, 0.742; 95% CI, 0.659–0.825; P < 0.001 vs. ABCDMP; AUC CURB-65, 0.775; 95% CI, 0.700–0.850; P < 0.001 vs. ABCDMP; AUC DECAF, 0.629; 95% CI, 0.538–0.720; P < 0.001 vs. ABCDMP; AUC NIVO, 0.619; 95% CI, 0.533–0.706; P < 0.001 vs. ABCDMP) after the Z test, which suggested that the discriminatory value of our risk score for predicting in-hospital mortality in patients with AECOPD and CVDs was significantly better. Although this new risk score was originally developed for the short-term prediction of mortality, the score also showed moderate predictive performance for 3-year mortality in patients with AECOPD and CVDs (AUC: 0.703, 95% CI, 0.670–0.735; p < 0.001). It can be used to stratify patients admitted to the hospital with AECOPD and CVDs into different management groups (P < 0.001).
The ABCDMP score might also be helpful for clinical decision-making regarding the selection of management strategies. Patients at low risk may receive outpatient treatment or discharge them early, which could ease their financial burden of hospitalization and reduce the unnecessary waste of medical resources. Those at high risk can be considered for early interventions and escalation of care. Finally, patients with medium-risk scores can be managed with regular reassessment of risk. However, the calculation of a risk score cannot be the only variable determining such far-reaching decisions, which must take into account many other individual aspects, such as the patient’s wishes, comorbidities, and economic situation. Nevertheless, the ABCDMP score might provide valuable assistance.
Strengths and limitations
This study is the first attempt to derive and validate a clinical prognostic score among inpatients with AECOPD and CVDs. It has several strengths, including the enrollment of patients from diverse hospitals and a near complete prospective data collection. In addition, the ABCDMP score is easy to calculate and apply in clinical practice and allows for good identification of patients at risk for in-hospital mortality. Nevertheless, our study had limitations. First, this was a secondary analysis of a prospective cohort of patients with AECOPD and CVDs, and we could not reach all baseline characteristics. Fortunately, the proportion of excluded patients was small, and the impact on our results can be neglected. Second, our sample included inpatients with AECOPD and CVDs; thus, our results may not be generalizable to outpatients or those with hospital at home.