Acute Physiological Predictors of Invasive Ventilation and Mortality in Geriatric Intensive Care Unit (ICU) Patients: A Retrospective Cohort Study From India.

Background  Older adults constitute an increasing proportion of ICU admissions in low- and middle-income countries (LMICs), yet predictors of invasive mechanical ventilation (IMV) and in-hospital mortality remain poorly characterized in dedicated geriatric ICUs. We examined whether acute physiological parameters measured at admission carry greater prognostic value for these two outcomes than baseline comorbidity burden in critically ill older adults. Methods  We conducted a single-center retrospective cohort study of 291 consecutive patients aged ≥60 years admitted to the dedicated Geriatric Medical ICU at AIIMS New Delhi between June and December 2020. Admission demographics, comorbidities, functional status, and physiological and laboratory parameters were extracted from electronic records. Independent predictors of IMV were identified by multivariable logistic regression and of in-hospital mortality by Cox proportional-hazards modeling. Model discrimination was summarized by the area under the ROC curve (AUC) for IMV and Harrell's concordance index (C-index) for mortality. Results  The median age was 76 years (IQR: 69-80); 176 patients (60.5%) were male. IMV was required in 124 patients (42.6%), and in-hospital mortality was 47.1% (137 patients). Independent predictors of IMV were respiratory rate (adjusted OR [aOR] 1.05, 95% CI 1.01-1.10), oxygen saturation (aOR 0.94, 0.91-0.98), Glasgow Coma Scale (aOR 0.84, 0.78-0.92), and PaCO₂ (aOR 1.02, 1.00-1.04). Model AUC 0.73 (0.67-0.79). In the primary mortality model (n=265, which excluded lactate owing to 34% missingness), the independent predictors were diastolic blood pressure (aHR 0.98, 0.97-0.99), bicarbonate (aHR 0.96, 0.93-0.98), and Glasgow Coma Scale (aHR 0.95, 0.91-1.00); age was not independently associated (aHR 1.02, p=0.08). C-index 0.70. In a secondary model restricted to patients with a measured admission lactate (n=186), lactate was independently associated with mortality (aHR 1.22, 1.10-1.36). Baseline multimorbidity was not independently associated with either outcome. Kaplan-Meier analysis demonstrated significantly worse survival among patients aged ≥75 years (log-rank p=0.002) and among those with admission lactate ≥2 mmol/L (log-rank p<0.0001). Conclusions  In this dedicated geriatric ICU cohort, the degree of acute physiological derangement at admission rather than chronological age or comorbidity burden was the dominant predictor of both invasive mechanical ventilation and in-hospital mortality. Because these parameters are captured routinely at the bedside, they provide a practical basis for early risk stratification and escalation-of-care decisions in resource-limited geriatric critical care.