Development of a predictive score to discriminate community acquired pneumonia with underlying lung cancer: A retrospective case – control study

Abstract

Background

A pneumonic infiltrate might hide an occult lung cancer (LC). This awareness depends on each clinician personal experience, turning definitive LC diagnosis challenging and possibly delayed. In this study we aimed to develop a clinical score to better identify those cases.

Materials and methods

We conducted a retrospective case–control study, including previously undiagnosed LC patients admitted in our institution, with a presumptive suspicious of community acquired pneumonia (CAP). Cases were compared with random CAP inpatient controls, using a matched 2:1 ratio. Demographic, clinical, and laboratorial variables were assessed for a possible association with the presence of a CAP with underlying LC (CAP–uLC).

Results

Among 535 hospitalized LC patients, 43 cases had a presentation compatible with CAP and were compared with 86 CAP controls. A scoring system was built using 6 independent variables, which positively correlated with CAP–uLC: smoking history (OR: 8.3 [1.9–36.2]; p = 0.005); absence of fever (6.5 [2.0–21.5]; p = 0.002); sputum with blood (5.9 [1.2–29.9]; p = 0.033); platelet count ≥ 232x10³/μL (5.8 [1.6–20.6]; p = 0.006); putative alternative diagnosis than CAP (4.6 [1.5–14.7]; p = 0.009); and duration of symptoms ≥ 10 days (3.7 [1.1–13.0]; p = 0.037). Our score presented an AUC of 0.910 (95 % CI, 0.852–0.967; p < 0.001), a sensitivity of 88.1 % and specificity of 84.7 %, in predicting the risk of presenting a CAP–uLC, when set to a cutoff of 18.

Conclusion

We propose a novel risk score aimed to aid clinicians identifying patients with CAP–uLC in the acute setting, possibly prompting early LC diagnosis.