We question both the ethical and clinical justification for renal biopsies on normal pregnant women in the protocols of Strevens et al. (Br J Obstet Gynaecol 2003;110:825–836) and the propriety of the Journal's publication of their findings.
The authors note that the women were informed there might be bleeding and haematomas, leading to transfusion, but do not indicate whether the subjects were told their anticipated frequency, severity and potential for other more serious complications. Recent texts1 list complication rates for gross bleeding at 3–10%, intrarenal and perinephric haematomas at 60% and hypotension at 1–2%. Haemoglobin falls by 1 g/dL in half of cases and even nephrectomy has been reported. Other adverse events include persistent arteriovenous fistulae, infection and death. The use of automated techniques combined with ultrasound may decrease risk, but only by about half, and the more recent literature must be scrutinised carefully as the biopsies surveyed may include transplanted kidneys that are easily palpable and safer to biopsy. We wonder whether the University ethics committee was provided with appropriate reviews of all these risks, or only given selective optimistic reports, and a single local individual's limited record. We also wonder whether consideration was given to halting the study when a patient did in fact bleed and require transfusions.
Even if the risks were less than those stated above, this study would be unethical because it did not meet the requirement of ‘equipoise’, that is, that there be a balance between the risks and expected benefits of the different arms of the study2. Failure of the normal controls to meet this requirement is evident even if one only considers the incomplete list of risks cited by the authors.
We also question the biopsy of gravid hypertensives prepartum. Antenatal biopsies were performed 30 years ago, but clinicians rapidly realised that after gestational week 32 (now probably 30), biopsy does not affect management, producing too little acute and long term prognostic data to warrant puncture at a time when coagulation indices may change rapidly. The literature has emphasised, for decades now, that indications for prepartum renal biopsy be restricted to suspicion of certain treatable renal diseases, only to be performed remote from term3–5. One of us (MDL) discussed the potential use of carefully supervised ‘postpartum’ biopsies in 19756—when it was incorrectly believed their content might help counsel patients for subsequent pregnancies—but quickly concluded that even such use had little value. These facts dispute the authors' claim that inclusion of controls was necessary. They argue that if glomerular endotheliosis was present in this group's kidneys, the lesion would no longer be ‘pathognomonic’ and it would now be ‘unethical to use renal biopsy for diagnostic purposes …f. Actually, the term used by most is ‘characteristic’, not ‘pathognomonic’. Reports for over 25 years have documented minimal endotheliosis in patients who did not have hypertension and/or proteinuria. Proper interpretation of the Strevens et al.'s data still reveal ‘characteristic’ differences between the pre-eclamptics and controls, and there are virtually no indications to biopsy suspected pre-eclamptics, or anybody after week 30.
We also question the propriety of the Journal's publishing these articles, and hope that the editors will re-examine them. If they find the protocols grossly unethical, the Journal should acknowledge its error. This would avoid the appearance of acquiescence in ethically impermissible research.