Sinert and Spektor analyze a systematic review published by McGee and coworkers evaluating the utility of the physical exam in determining if a patient is hypovolemic. McGee and co-workers reviewed studies from 1966 to 1997 which assessed the operating characteristics of physical exam findings in hypovolemia. Two types of studies were included in the review: studies of healthy volunteers who lost a standardized amount of blood via phlebotomy and patients presenting to the emergency department with suspected non-traumatic hypovolemia.
Studies showed postural vital signs may be abnormal in a significant portion of normovolemic subjects. Postural hypotension, defined as a greater than a 20 mm Hg decrease in systolic blood pressure, was seen in 10% of normovolemic subjects less than 65 years of age and between 11–30%of subjects older than 65. The studies of phlebotomized healthy volunteers showed that postural vital signs are highly specific but insensitive (Table 1). In patients with moderate blood loss (450 to 630 ml) no vital sign measurement achieved a sensitivity greater than 27%. In patients with large blood loss (630 to 1150 ml) no measurement exceeded a sensitivity of 33% with the exception of the presence of severe dizziness or increase in heart rate of more than 30 beats per minute, which had a sensitivity of 97%. Delayed capillary refill was highly specific (95%) but insensitive (34%).
In patients presenting to the emergency room with suspected hypovolemia from nontraumatic causes no individual physical exam finding appeared to be useful (postural hypotension, pulse increase >30, dry axilla or mucous membranes or tongue, longitudinal furrows on the tongue, sunken eyes, confusion, upper and lower extremity weakness, garbled speech, and prolonged capillary refill). No exam finding yielded a sensitivity greater than 85% (the presence of dry mucous membranes and longitudinal furrows on the tongue had sensitivities of 85%) and the remainder of findings are far less sensitive. Positive likelihood ratios for changes in postural vital signs were only 1.7 for hypotension and 1.5 for pulse increase greater than 30 beats per minute, making these findings unhelpful in diagnosing hypovolemia. It should be noted that these studies used elevated BUN to creatinine ratios, an unproven gold standard. Studies show physical exam findings may be more helpful when used in combination. Elevated BUN to creatinine ratios correlated with an average of four to six physical exam findings, whereas the BUN to creatinine ratio was not elevated when only one finding was present
This LLSA article focuses on the role of likelihood ratios in evaluating diagnostic tests. When used in combination with the pretest probability, the likelihood ratio allows one to arrive at the statistical probability of the presence or absence of disease, the posttest probability.
A positive likelihood ratio represents the ratio of true positives to false positives provided by a given test. Mathematically speaking, the positive likelihood ratio is represented as sensitivity divided by 1-specificity. For example, if a test provides a true positive 90% of the time and a false positive 10% of the time the positive likelihood ratio would be 0.9/0.1, or 9.0. This means that when a positive result is found it is 9 times more likely to be a true positive than a false positive. A negative likelihood ratio represents the ratio of false negatives to true negatives. For example, if a test provided a false negative test 9% of the time and a true negative 91% of the time, the negative likelihood ratio would be 9:91, approximately 0.1. Thus, a negative result suggests that it is 10 times more likely that disease is absent. A smaller negative likelihood ratio implies that a negative result is more likely to be a true negative. It is generally accepted that a positive test result using a test with a positive likelihood ratio of 10 or greater is almost conclusive of disease and a negative test result using a test with a likelihood ratio of 0.1 or less nearly excludes disease. Likelihood ratios of 1 imply that the test is not useful in ruling in or ruling out disease.
This article suggests that vitals signs, both traditional and postural, are not helpful by themselves in determining if a patient is hypovolemic. Traditional vital signs are insensitive for blood loss. Postural vital signs are similarly unhelpful (i.e. have low likelihood ratios) except in the setting of otherwise healthy individuals with large (greater than 1 liter) blood loss. The reviewers suggest that measurement of serum lactate and base deficit may be a more valuable indicator of hypoperfusion and therefore a better predictor of mortality. While abnormal vital signs are highly specific for hypovolemia, this does not translate well to the emergency room where there are many explanations for tachycardia and hypotension and ruling out emergent conditions is the priority. Instead of vital sign measurements and specific physical exam findings, an organized history, physical exam, and clinical judgment continue to be the best marker for determining volume status.