We report on the diagnostic strength of pedal edema and JVP for volume assessment, using BCM as the criterion standard, in an outpatient peritoneal dialysis population without evidence of cardiac disease. For the prediction of euvolemia, the absence of edema (−LR 0.44, 95 % CI 0.23–0.83) was a better sign than a reduced JVP, except possibly when the JVP was 0 cm above the sternal angle (excellent average staff negative LR but poor precision). However, the negative likelihood ratio for edema is modest and does not reduce the probability of volume overload by much. For example, given the prevalence of volume overload in our population of 48 %, the absence of pedal edema results in a post-test probability of 27 %, which is still quite a significant proportion.
As for the prediction of volume overload, no conclusions can be drawn for the diagnostic strength of an elevated (>3 cm) JVP due to our study’s small sample size and resultant wide confidence intervals. On the other hand, statistical significance was reached for the presence of edema to rule in volume overload (+LR 16, 95 % CI 1.02–260), and while the precision is also poor, it is probably an excellent sign with an average LR of over 10. In addition, there was no added diagnostic benefit of adding the JVP to the assessment for pedal edema as a tool to assess overall volume status. As no previous studies document the sensitivity and specificity of these physical exam findings, the available evidence suggests that the presence of edema is a more useful sign than an elevated JVP for volume overload.
Our 48 % prevalence of volume overload is similar to the previously reported value of 53 % by Van Biesen et al. in their large European PD bioimpedance study [9]. We were unable to find any previous studies that assessed the power of the JVP as a clinical tool to assess volume status. Review of the literature found that the central venous pressure (CVP), for which JVP is a physical exam marker, has been studied, but only in an intensive care unit population. A recent systematic review of these studies found poor correlation between CVP and measured blood volume (r = 0.2) [10]. Unlike our analysis, these studies did not exclude patients with cardiac dysfunction, which would be expected to interfere with the purported association between intravascular volume and central venous pressure.
Agarwal et al. evaluated the association between pedal edema and volume status in dialysis patients [11]. Contrary to our results, these authors found no association between edema and their criterion standards for volume overload. However, differing from our study, they used inferior vena cava diameter, blood volume monitoring, and brain natriuretic peptide (BNP) as their standards, which would all be expected to be less accurate than better validated measurements of body water, such as BCM. Indeed, Imaz et al. found only moderate correlation between BNP (r = 0.6) and inferior vena cava diameter (r = 0.5) with BCM in their study of stage 3 or greater CKD patients not undergoing dialysis [12].
Patients who were volume overloaded tended to be older and have lower mean diastolic pressures and a higher BMI. The higher BMI likely reflects the added weight from volume excess. We are unable to explain the increased incidence of volume overload with age, especially since both groups seemed to have similar residual renal function. The reason behind the observed lower diastolic pressure in the volume-overloaded group is equally obscure, although this could just be a chance occurrence.
Our study has several limitations. As stated, the sample size is small, limiting the precision of some of our findings. This was partly due to the fact that we had to exclude nearly half of potential subjects due to co-existing heart disease. Second, we are unable to entirely rule out the co-existence of heart disease in the population studied as some did not have a recent echocardiogram on record. Third, examiners were not blinded to the presence or absence of edema before assessing the JVP. As the JVP is one of the more subjective physical exam findings, this could have influenced our results. Furthermore, while both the examiners were trained and experienced in the JVP technique, different individuals may have obtained more accurate results. Fourth, our results are limited to dialysis patients without heart disease. As heart disease is quite prevalent among dialysis populations, this limits the more general application of our results. Fifth, while BCM has been shown to correlate very well with traditional reference standards of body water in hemodialysis patients, there is still some variation, with one study finding an error of −0.9 ± 1.4 L for ECW when compared to bromide dilution [4]. Thus, our criterion standard is close to, but does not correlate completely with, the gold standard, which could have confounded our results. We note that were unable to study intravascular volume, as BCM is unable to measure this quantity. Sixth, only one of the examiners evaluated for the presence of pedal edema.
Finally, it can be argued that increased abdominal pressure from abdominal fluid in our patients (most of whom had an ongoing dwell during assessment) could increase the JVP independent of total volume status. However, when a prolonged abdominal pressure is applied onto healthy patients without cardiac dysfunction in multiple studies of the abdominojugular reflux, the JVP is found to only transiently elevate [13]. As we excluded patients with significant cardiac dysfunction from our study population, we believe this behavior should also apply to them. In addition, ongoing dwell may affect BCM measurements, even though it is thought to be an “invisible” body compartment with respect to the machine. Arroyo et al. studied the effect of abdominal dwell on the BCM measurements and found a small (~1 %) difference in OH/ECW values between the filled and un-filled states [14]. However, we do not believe this error is large enough to have unduly affected our results.