Inflammatory Blood Scores Linked to Hospital Death Risk in Pulmonary Embolism

Patients who arrive at the emergency department with an acute pulmonary embolism face an urgent question that clinicians must answer within minutes: how likely is this person to die during this hospitalization? A retrospective study from Türkiye now adds two inexpensive blood-derived indices to the conversation about early risk stratification. Researchers at Elazığ Fethi Sekin City Hospital analyzed 399 adults with computed-tomography-confirmed pulmonary embolism admitted between January 2021 and December 2025 and found that a composite score combining C-reactive protein, albumin, and lymphocyte count — the CALLY index — and a simpler C-reactive protein-to-albumin ratio (CAR) were each independently associated with in-hospital mortality after adjustment for age, sex, and serum creatinine. The findings, published in Frontiers in Medicine, come with an important caveat from the authors themselves: the discriminative power of both markers was modest, and neither should be used alone to rule out high-risk patients.

The study population was drawn from the hospital’s electronic records system, with only initial presentations counted to avoid duplicate entries and chronic pulmonary embolism cases systematically excluded. Of the 399 patients included, 311 (78.0 percent) survived their hospitalization and 88 (22.0 percent) died. The overall median age was 72 years, and non-survivors were significantly older than survivors, with a median age of 79.5 years compared with 70 years. Female patients made up a larger share of the non-survivor group, and sex differences reached statistical significance in the comparison between outcome groups. Acute pulmonary embolism was confirmed in every included patient by computed tomography pulmonary angiography, and patients under 18, those without definitive imaging confirmation, and those with incomplete records or missing laboratory data were excluded from the final analysis.

The biological rationale behind the CALLY index rests on three distinct dimensions of a patient’s response to severe illness. C-reactive protein is a well-established acute-phase reactant that reflects systemic inflammatory stress and, in the context of pulmonary embolism, a hypercoagulable state. Albumin, which falls during prolonged inflammation, serves as a marker of nutritional reserve and the negative acute-phase response. The absolute lymphocyte count captures host immune competence, and lymphopenia in severe acute illness is thought to reflect stress-mediated immunosuppression driven by hypercortisolemia and accelerated lymphocyte apoptosis. Combining these three values into a single index is intended to capture an integrated picture of inflammation, nutrition, and immune function that a single biomarker cannot. The C-reactive protein-to-albumin ratio, by contrast, uses only two of these components and reflects the balance between inflammatory intensity and nutritional reserve.

The differences between survivors and non-survivors were statistically significant across all components of the indices. Non-survivors had markedly lower CALLY index values than survivors and correspondingly higher CAR values. Dissecting the components, the researchers found that patients who died had significantly reduced serum albumin levels and reduced absolute lymphocyte counts, alongside markedly elevated C-reactive protein concentrations. Other laboratory measures also differed between the groups, including urea, creatinine, lactate, and lactate dehydrogenase, all of which showed univariate associations with mortality. Because blood urea collinearizes with creatinine, and because lactate and lactate dehydrogenase were judged to reflect downstream acute hypoperfusion and shock rather than baseline constitutional risk, the authors deliberately restricted their primary multivariable adjustment to age, sex, and serum creatinine.

To quantify how well the two indices separated survivors from non-survivors, the team performed receiver operating characteristic analysis, determining optimal thresholds with the maximum Youden’s J statistic. The CALLY index yielded an area under the curve of 0.655, while the CAR achieved an AUC of 0.643. A pairwise comparison using DeLong’s test found no statistically significant difference between the two AUC values, with a p-value of 0.582. At the optimal CALLY cut-off of 1.17, the index demonstrated a sensitivity of 69.1 percent. The optimal CAR threshold was greater than 3.28, at which the ratio showed a sensitivity of 42.0 percent, a specificity of 83.2 percent, a positive predictive value of 40.7 percent, and a negative predictive value of 83.2 percent. In practical terms, the CALLY index offered higher sensitivity for initial screening while the CAR offered higher specificity, and both markers produced negative predictive values above 83 percent.

The authors are explicit that these negative predictive values largely reflect the baseline survival prevalence in the cohort and should not be interpreted as sufficient to definitively rule out high-risk patients when either marker is used in isolation. With area under the curve values in the mid-0.6s, both indices sit firmly in the range of modest rather than strong discrimination, well below the performance generally expected of established clinical tools. The pulmonary embolism severity index and its simplified version remain the reference standards for mortality risk stratification, and this study could not test whether the new biomarkers add incremental value to those scores because the clinical variables needed to calculate them were not consistently available in the retrospective dataset.

The regression analysis strengthened the association beyond simple group comparison. In multivariable Model 1, which evaluated the CALLY index adjusted for age, sex, and serum creatinine, a low CALLY index below 1.17 was associated with roughly a 2.5-fold increase in adjusted odds of in-hospital mortality, with an adjusted odds ratio of 2.54 and a 95 percent confidence interval of 1.46 to 4.50. The continuous log-transformed CALLY variable showed a similar direction of association. In Model 2, a high CAR above 3.28 carried an adjusted odds ratio of 3.02 with a 95 percent confidence interval of 1.66 to 5.50, and the continuous log-transformed CAR was also independently associated with mortality. To prevent collinearity between the closely related composite metrics, the two indices were analyzed in separate models. Age and creatinine remained significant predictors in both models, while sex showed a borderline association that approached but did not clearly cross the threshold for statistical significance.

The study’s methods section details a careful approach to missing data. A complete-case framework was used without imputation: patients lacking essential clinical records or the admission laboratory tests required for biomarker calculation were excluded at the cohort level, yielding the final 399 patients. Within that cohort, individual laboratory parameters were analyzed using an available-case approach, with the exact number of analyzed patients reported for each variable. The multivariable models were complete-case analyses restricted to patients with observations across all included covariates, with sample sizes of 367 for the categorical CALLY model, 366 for the continuous log-CALLY model, 368 for the categorical CAR model, and 366 for the continuous log-CAR model. The authors note that minor variations in sample sizes across specific parameters arose from missing values in the electronic health database.

Context from prior literature supports the plausibility of the findings. Low CALLY indices have been linked to higher mortality in patients with decompensated heart failure admitted to intensive care and in other heart failure populations, and C-reactive protein-to-albumin ratios have been identified as predictors of short-term mortality in elderly hospitalized patients, community-acquired pneumonia, and sepsis, with reported thresholds varying by clinical setting. Recent multicenter data in acute pulmonary embolism reported a similarly modest discriminatory performance for the CAR, with an area under the curve of approximately 0.64, which the present single-center results corroborate. The authors suggest that the value of these indices lies not in replacing established tools but in potentially supporting early risk assessment when integrated with clinical severity scores.

The limitations deserve as much attention as the headline associations. The retrospective, single-center design introduces potential selection bias and limits generalizability to other hospitals and health systems. Echocardiographic right ventricular parameters such as the right-to-left ventricular ratio and TAPSE were not consistently available and could not enter the multivariable models, so the study cannot establish that CALLY or CAR outperforms or adds incremental value to established clinical risk scores or imaging parameters. Biomarkers were measured only at emergency department presentation, precluding any assessment of how the indices change during hospitalization and whether such changes carry prognostic information. The proposed CALLY cut-off of 1.17 requires prospective, multicenter validation before it could be adopted in practice, and it remains unknown whether incorporating these indices into existing risk stratification approaches actually improves clinical decision-making or patient outcomes. What the study does establish is an independent statistical association, in a single cohort, between two cheap and widely available laboratory indices and in-hospital death — a signal that now awaits confirmation in better-designed prospective studies.

Subject of Research: Blood-based inflammatory and nutritional indices as predictors of in-hospital mortality in acute pulmonary embolism

Article Title: Prognostic value of C-reactive protein-albumin-lymphocyte (CALLY) score and CRP-albumin ratio on mortality in patients diagnosed with pulmonary thromboembolism

Article References: Aydemir, A. M. (2026). Prognostic value of C-reactive protein-albumin-lymphocyte (CALLY) score and CRP-albumin ratio on mortality in patients diagnosed with pulmonary thromboembolism. Frontiers in Medicine, 13, Article 1944670. https://doi.org/10.3389/fmed.2026.1944670

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1944670

Keywords: pulmonary embolism, CALLY index, C-reactive protein-to-albumin ratio, in-hospital mortality, risk stratification, emergency department, biomarkers, retrospective study, inflammation, albumin, lymphocyte count, prognosis