Early diagnosis of dehydration in hospitalized geriatric patients using clinical and laboratory criteria

Dehydration is a common yet underdiagnosed condition, which is associated with poor prognosis in older patients. The clinical and laboratory criteria for assessing dehydration are of variable or poor diagnostic value in this population and require further validation.

To test different clinical and laboratory criteria for the early diagnosis of dehydration in hospitalized older patients as compared to the standard diagnosis based on a body weight gain ≥ 3% during the first week of admission.

Prospective study using clinical and laboratory criteria for dehydration selected by an expert panel.

Acute geriatric unit.

Geriatric patients (n = 112; aged 83 ± 6 years) admitted to the unit.

Using selected criteria (skin fold, dry mouth, calf muscle consistency, systolic blood pressure < 90 mmHg, orthostatic blood pressure drop, postural dizziness, thirst, apathy or delirium, urea, creatinine, uric acid, proteins, hemoglobin), expert clinicians prospectively assessed the patients to make a clinical diagnosis of dehydration upon admission. Clinicians were asked to provide a yes/no answer for each criterion and a global yes/no answer for dehydration, along with its estimated probability. Body weight was measured at admission and at Day 7. Laboratory parameters were assessed at baseline and at Day 7. The clinical diagnosis made at admission was retrospectively compared to the standard diagnosis made at Day 7 (> 3% weight gain) to test the values of each criterion separately, the global diagnosis, and its estimated probability.

In total, 100 patients with complete data (mean age 83 ± 6 years) were considered for analysis. Dehydration was the principal reason for admission in three patients only but was clinically diagnosed as part of the study in 39 patients, whereas the standard diagnosis of dehydration was ascertained in 20. Inter-rater agreement (kappa) was fair for the clinical criteria and clinical diagnosis, moderate to near perfect for the laboratory criteria, and substantial for the estimated probability of dehydration. When matched with the standard diagnosis, the final clinical diagnosis of dehydration had 70% sensitivity and 69% specificity, with a 90% negative predictive value. Individually, both clinical and laboratory criteria had good specificity (65-90%), but poor sensitivity (< 55%).

Compared with the standard diagnosis, dehydration was overdiagnosed by systematic clinical assessment upon admission in this frail population. However, clinicians performed better at excluding dehydration. Clinical acumen seems better than any individual clinical symptom or sign.