Like Dr. Scissors, our preference is for clinical practice to be based on randomized, controlled trials. We also recognize that until the required trials are performed, a sensible approach to distilling existing data and informing practice should be used. However, we need to emphasize that our inclusion criteria were not limited to stage V chronic kidney disease and that we searched for trials of any vitamin D compound, including generic over-the-counter products. No trial data were available on the effects of these products, and very few biochemical data were available on any product in stage III or IV chronic kidney disease. Until these trials are performed, we would be more circumspect than Dr. Scissors and suggest that use of these products for this indication may not be safe.

Dr. Coyne raised several specific issues with our meta-analysis, but all concern biochemical or surrogate outcomes. He presumably concurs with our major conclusion that the effects of vitamin D compounds on clinically important outcomes are largely unknown. We agree that the intervention in most studies was fixed-dose vitamin D rather than target PTH level. We also agree that there was an improvement in bone histomorphometry, but we are unaware of trial data to demonstrate a causal relationship between bone histomorphometry and clinical outcomes. Trials of established vitamin D compounds used several different assay types for measuring PTH levels; however, there was no statistically significant effect on outcomes when different PTH assay types used in the trials were included as a covariate in our meta-regression analysis.

Contrary to Dr. Coyne's assertion, we did not exclude a 266-patient trial of calcitriol versus paricalcitol (see Appendix Table 2 of our article). This trial is not included in the Forest plots because we used the outcome of end-of-treatment PTH levels, as prospectively defined in our a priori–based study protocol, and because this was the PTH level outcome reported by most trials (1). The choice of the primary outcome of greater than 50% reduction in PTH level used in this trial, even if it occurred more frequently in the analogue group, does not seem to be validated with hard end points any more than other PTH level outcomes. According to Dr. Coyne, we excluded a placebo-controlled trial of paricalcitol, which was the basis for U.S. Food and Drug Administration approval of paricalcitol (2), but this is incorrect. The study is included, but end-of-treatment PTH levels were not available. Dr. Coyne notes that we chose hyperphosphatemia and hypercalcemia as markers of cardiovascular and mortality outcomes. We describe these end points as surrogate outcomes. We agree most strongly that predicting patient-relative outcomes, such as cardiovascular end points from treatment changes in calcium and phosphorus, remain speculative and require randomized trials that show concordance between these outcomes. Dr. Coyne states that we combine newer vitamin D products despite evidence of differences between compounds. Data are so scarce on any intervention that all data are included, but differences in interventions are also clearly described in text, table, and graphical form. Grouping by the same intervention exactly would not change our primary conclusion.

Finally, Dr. Coyne notes that vitamin D is associated with improved survival on dialysis (in cohort studies). We agree, but suggest that this is not strong evidence that vitamin D intake improves outcomes because of residual confounding and other well-established biases, which are typical of observational studies. Recent examples of discrepant results between trials and cohort studies have been well publicized (3, 4). Although trials for vitamin D compounds are ongoing, none is designed to examine clinical outcomes.

For decades, we have used vitamin D compounds, a potentially beneficial intervention. The perils of accepting interventions in practice have now changed, and adequately powered trials need to be performed for us to consider a drug to have "proven" efficacy. Our patients deserve this and more.