We are pleased that Drs. Colindres and Kshirsagar acknowledge the methodologic strengths of our meta-analysis and the concordance of our findings with the large clinical trial reported by Maschio and colleagues. We agree that these factors make our findings more robust.

Drs. Colindres and Kshirsagar also question our interpretation that "Angiotensin-converting enzyme inhibitors are more effective than other antihypertensive agents in reducing the development of end-stage nondiabetic renal disease" because 5 of the 10 studies compared ACE inhibitors with placebo rather than other antihypertensive agents. Table 3 of our article notes that studies using placebo in the control group, as well as studies using antihypertensive agents in the control group, used concomitant therapy with antihypertensive agents (other than ACE inhibitors) to achieve the defined blood pressure targets. Concomitant therapy with antihypertensive agents was also used in the ACE inhibitor groups. Therefore, the pooled analysis tests the hypothesis that an antihypertensive regimen using an ACE inhibitor and other agents as necessary to reach a defined blood pressure goal is preferable to regimens not using ACE inhibitors. The lower risk for end-stage renal disease in the ACE inhibitor group indicates that regimens including an ACE inhibitor are more effective than regimens that do not include ACE inhibitors. Practically, we believe that this interpretation is equivalent to the interpretation above. We agree, however, that we could not precisely determine whether this beneficial effect is due to greater decline in blood pressure in the ACE inhibitor group or to other effects of ACE inhibition. In addition, we acknowledge that our conclusion is not based on a direct comparison of each ACE inhibitor with each alternative antihypertensive agent. Nonetheless, we believe that the message should be clear to practitioners. Angiotensin-converting enzyme inhibitors should be the first choice of antihypertensive agents in patients with nondiabetic renal disease.

Finally, we agree that there may be undetected clinical heterogeneity among the clinical studies included in our analysis. Detecting this would require pooling individual patient data, and we are currently planning to do this.