Hypothesis: Selected elements of a simple physical examination algorithm accurately predict categories of shock.
Setting: A 350-bed community teaching hospital.
Methods: Resident trainees who manage all critically ill and medically unstable patients were instructed to document capillary refill, (palpated) pulse volume, skin temperature, jugular venous pressure (JVP) and lung examination in all patients with prolonged (>30 minutes hypotension <90 mmHg). Treatment was determined by house officers guided by attending physicians of record. All cases were retrospectively reviewed by a senior clinician who applied consensus criteria/definitions to categorize shock as septic, cardiogenic or hypovolemic. Operating characteristics of examination findings for predicting categories of shock were computed.
Results: A total of 68 patients, averaging 71 ± 16 years, were studied. A total of 37 patients were diagnosed with septic shock, 18 with cardiogenic shock of and 13 with hypovolemic shock. Capillary refill and skin temperature predicted septic shock with sensitivity of 89%, specificity of 68%, positive predictive value (PPV) of 77%, negative predictive value (NPV) of 84%, and overall accuracy of 79%. Presence of JVP >7 cmH(2)O was more accurate than bilateral pulmonary crackles (>1/3 from bases) in predicting cardiogenic shock for low-output patients with sensitivity of 82%, specificity of 79%, PPV of 75%, NPV of 85%, and overall accuracy of 80%. Using just skin temperature and JVP, the bedside approach correctly diagnosed 52/68 cases (overall accuracy = 76%).
Conclusions: Simple bedside clinical examination findings correctly predict categories of shock in a majority of cases.