Researchers raise questions about prophylactic cesareandelivery to prevent Perinatal HIV transmission -

Stringer and colleagues argue that evidence is insufficientto supportthe use of cesarean delivery to prevent perinatal transmissionof human immunodeficiency virus (HIV) in most circumstances. (JAMA. 1999;281:1946)

Vol. 281, pp. 1946-1949, May 26, 1999 Prophylactic CesareanDelivery for the Prevention of Perinatal Human Immunodeficiency Virus Transmission

The Case for Restraint
Jeffrey S. A. Stringer, MD
Dwight J. Rouse, MD
Robert L. Goldenberg, M

The introduction of viral protease inhibitors into combinationantiretroviral regimens has dramatically changed human immunodeficiencyvirus (HIV) therapeutics. Capable of suppressing measurable viral loadsto undetectable levels, combination therapy has improved clinical status[1]and prolonged survival[2,3] for many patients in the developed world andis now firmly established as standard of care in the United States.[4] Recently,in an effort to afford these same benefits to HIV-infected pregnant women,the US Public Health Service has recommended that combination antiretroviraltherapy be offered during pregnancy as well.[5] As an associated effect,several centers that have implemented these recommendations have observedperinatal HIV transmission rates approaching zero among women receivingcombination agents[6-9]; by comparison, the HIV transmission risk in untreated,non­breast-feeding women in the developed world is 14% to 25%,[10] whichcan be reduced to as low as 5% with zidovudine monotherapy.[11,12]

Evidence on Use of Prophylactic Cesarean Delivery

Prophylactic cesarean delivery has been associated witha protective effect against vertical infection in some observational studies,[13-16]but not in others.[17-21] Recently, the International Perinatal HIV Group,[22]conducted a pooled analysis (meta-analysis), in which original data wereextracted from a number of these previously published observational studiesand all patients were reassigned to strictly defined mode-of-delivery strata.They hypothesized that some of the discrepancy among results of earlierstudies related to the inappropriate grouping of all cesarean deliveriesand that only those cesarean deliveries performed prior to the onset oflabor or membrane rupture were protective against vertical transmission.The meta-analysis described a decreased likelihood of perinatal HIV transmissionin both zidovudine-treated and zidovudine-untreated mother-infant pairswho underwent prophylactic (prelabor, premembrane rupture) cesarean delivery.

Because of the rapidly evolving standards for antiretroviraltherapy, few, if any, women in this meta-analysis[22] received maximallysuppressive combination drug regimens. Thus, the applicability of theseinteresting findings to the present-day population of HIV-infected pregnantwomen in the developed world is limited. Nonetheless, the presentation ofthe meta-analysis by Read at the 1998 World AIDS Conference and its prepublicationInternet release (, accessed 1/28/99) may have led someobstetrical care providers to believe that there is now an evidence-based,nonpharmacological intervention available for the reduction of perinatalHIV transmission and that this intervention is universally applicable forall HIV-infected pregnant women. However, the purpose of meta-analysis isto increase the power, precision, and generalizability of data from multiplesmall studies: the technique does not eliminate confounders or negate biaspresent in the original reports and therefore cannot substitute for a randomized,controlled trial.

Indeed, comparative studies[23] have demonstrated thatmeta-analysis of even randomized studies will predict an effect that willlater be discredited by large randomized trials as often as 35% of the time.Similar issues are addressed in a recent editorial by Bailar[24] on theshortcomings of meta-analysis. Two recent notable examples of this phenomenonin obstetrics include daily calcium supplementation and low-dose aspirintherapy for the prevention of preeclampsia. Both interventions showed markedbenefit in small randomized trials (and meta-analyses of those trials[25,26]) that failed subsequently to be borne out in larger definitive studies.[27,28]We emphasize that the analysis by Read is not a meta-analysis of treatmenttrials but rather of observational studies, and thus, while an excellenthypothesis-generating study, represents evidence of insufficient qualityby which to make broad clinical recommendations.[29]

To more rigorously investigate the protective effect ofcesarean delivery, a multicenter randomized trial has been conducted inEurope, and data from the first 370 infants delivered to study participantsare now available.[30] Overall, women randomized to elective cesarean delivery(n=188) in this trial had a significantly lower rate of vertical transmissionthan those randomized to vaginal delivery (n=220) (1.8% vs 10.5%, P<.001).However, in those women receiving zidovudine prophylaxis (n=236), the reductionin transmission risk was smaller and not statistically significant. Whenthe trial participants were stratified by intended mode of delivery, a transmissionrate of 4.3% was observed among zidovudine-treated women (n=117) assignedto vaginal delivery compared with 0.8% in those (n=119) assigned to cesareandelivery (odds ratio [OR], 0.2; 95% confidence interval [CI], 0.0-1.7).This difference was even less pronounced when data were analyzed by actualmode of delivery (OR, 0.6; 95% CI, 0.1-3.2). Thus, the only randomized trialdata presently available provide insufficient evidence that prophylacticcesarean delivery will substantially reduce perinatal HIV transmission inwomen treated with zidovudine monotherapy and do not address the role ofprophylactic cesarean delivery in women treated with combination agents.

Although published information[5] regarding the effectof combination therapy on perinatal HIV transmission is presently limited,this is an area of active research. In recent months, 3 series, includinga total of 178 pregnant women treated with combination antiretroviral therapy,[7-9]have been presented; none have documented a single case of perinatal transmission.Based on the 95% confidence limits of the binomial probability distributions,the risk of vertical transmission in these patients is estimated at 2% orless. Thus, with the low baseline transmission risks suggested by thesestudies, it is difficult to determine whether prophylactic cesarean deliverywould be a protective adjunct to combination drug therapy. French investigatorsin the European trial[30] decided to withdraw from the study after calculatingthat, given the current low rate of perinatal transmission observed in theirpatients receiving zidovudine monotherapy,[11] demonstration of the postulated50% reduction in HIV transmission attributed to cesarean delivery wouldrequire a prohibitively large enrollment. Indeed, if the rate of transmissionamong patients receiving combination therapy were as low as 2% and couldbe reduced to 1% with prophylactic cesarean delivery, a clinical trial with80% power to demonstrate this difference would require randomization ofnearly 5000 patients.

Further, studies of thymic tissue from aborted fetuses[31]suggest that a small but finite percentage of vertical HIV infection mayoccur early in the first trimester of pregnancy. Antiretroviral therapyinitiated prior to conception or in early gestation may protect againstthis occurrence, but clearly prophylactic cesarean delivery would not. Indeed,it is possible that the presence of a small number of failures associatedwith combination regimens may represent these early infections, and thussignal the limit of therapeutic efficacy for any intervention institutedafter conception.

Morbidity Associated With Cesarean Delivery in HIV-InfectedWomen

Another important consideration is the operative morbiditythat may result from subjecting immunocompromised patients to major abdominalsurgery. Data on cesarean-related morbidity in HIV-infected patients arepresently limited. In a case-control study comparing 156 HIV-infected womenwith 156 uninfected women who had cesarean deliveries, Semprini et al[32]reported 6 cases of major postoperative complications-pneumonia (2 cases),pleural effusion (1 case), sepsis (1 case), and need for blood transfusion(2 cases)-vs 1 major complication (need for blood transfusion) among thecontrols (OR, 6.0; 95% CI, 0.9-38.5). The study also reported a 3-fold increasein minor (mostly febrile) complications (OR, 3.1; 95% CI, 1.3-7.6). In addition,women with low CD4 lymphocyte counts were at a statistically significantincreased risk of morbidity. These data are supported by 2 series reportedat the 6th Conference on Retroviruses and Opportunistic Infections, in whicha total of 91 women underwent elective cesarean deliveries vs 1518 womenwho underwent nonelective cesarean delivery or vaginal delivery. Each ofthese studies described a 2-fold increase in any complication assessed aswell as a 2-fold increase in infectious morbidity in those women who underwentelective cesarean delivery.[33,34]

Although still evolving, currently available data suggestthat cesarean delivery carries a substantially increased risk of perioperativecomplications in HIV-infected women.[32-34] The unfortunate irony is thatthe women whose infants theoretically will benefit most from cesarean delivery-womenwith high viral loads and low CD4 cell counts, who have the highest riskof transmission[35]-are also the women who are most likely to experienceoperative morbidity.

In the absence of more complete morbidity data, a usefulclinical measure for assessing the health consequences of prophylactic cesareandelivery is the number of operations required to avert a single case ofpediatric HIV infection. For instance, if the rate of transmission amongpatients receiving zidovudine monotherapy is 4.3%, and it could be reducedto 0.8% with prophylactic cesarean delivery,[30] the avoidance of a singlecase of mother-to-child transmission would require approximately 29 prophylacticcesarean deliveries. However, if the rate of perinatal HIV transmissionassociated with combination therapy is 2% and could be reduced to 1% withprophylactic cesarean delivery, the avoidance of a single case of mother-to-childtransmission would require approximately 100 prophylactic cesarean deliveries.However, it does not necessarily follow that because prophylactic cesareandelivery is able at relatively high baseline transmission rates to reduceby half the risk of vertical infection that this same degree of protectionwill occur at the lower baseline transmission rates attributed to combinationtherapy. In fact, given the above-described phenomenon of possible earlygestation transmission, it is likely that at progressively lower baselinetransmission rates, the protective efficacy of cesarean delivery would bemore modest, and thus commensurately more cesarean deliveries would be requiredto avert a single case of infant HIV.

Potential Role of Prophylactic Cesarean Delivery

According to the available data, the possibility of a substantialbeneficial effect from prophylactic cesarean delivery is most likely inthose women who are not taking antiretroviral medication.[22,30] Thus, thepatient with known HIV infection who presents late in gestation, or who,for whatever reason has not received antiretrovirals during pregnancy, mayprove to be a candidate for prophylactic cesarean delivery. However, analternative intervention for women not receiving antiretroviral therapyduring pregnancy, which has been proven efficacious in several randomizedtrials, would be the administration of short-course zidovudine using anyof several described protocols.[36-38] Likewise, the patient who is unableto adhere to the complex dosing schedule of multiple-agent therapy and whomay be unable to administer the recommended 6-week course of zidovudinesyrup to her newborn may prove an appropriate candidate as well. Althoughthe specific interplay of viral load and cesarean delivery on vertical infectionis not known, there is evidence to suggest an association between high viralloads and increased risk of vertical transmission.[35] Therefore, the uncommonpatient who continues to exhibit high levels of HIV RNA despite combinationdrug therapy may also eventually be deemed an appropriate candidate forprophylactic cesarean delivery.

Lessons From Electronic Fetal Monitoring

The climate evolving around the use of cesarean deliveryfor prevention of vertical HIV transmission is reminiscent of that whichoccurred 2 decades ago regarding the practice of electronic fetal monitoringto prevent birth asphyxia. During the early 1970s, reports of the beneficialeffects of continuous electronic fetal monitoring began to appear in theobstetrical literature. Based on reports of improved perinatal outcomesin these nonrandomized, retrospective analyses, electronic fetal monitoringwas quickly adopted as standard practice in many centers. By 1976, whenthe first randomized trial data were published, electronic monitoring hadbecome so interwoven into the clinical and legal fabric of obstetrics asto be ineradicable-despite the trial's findings that no improvement in fetaloutcome resulted from the practice, and its further demonstration of a substantiallyincreased probability of cesarean delivery in comparison with the alternativepractice of intermittent fetal heart tone auscultation.[39] Subsequent clinicaltrials with essentially the same findings[40] have done little to mitigatethe continued widespread use of this technology.

Historical precedent suggests that to act on the basisof incomplete data is to risk establishment as the standard of care a practicethat may not easily be changed, even in the face of contrary evidence. Indeed,if zidovudine monotherapy coupled with prophylactic cesarean delivery wereto become standard practice before adequate data regarding the effect ofcombination antiretrovirals on perinatal transmission are available, eliminatingcesarean delivery from this standard in the future could be extremely difficult.

Counseling Pregnant Women With HIV Infection

Thus, in light of recent publications on this matter, counselingHIV-infected women regarding mode of delivery is appropriate but shouldbe based on data available from women of similar disease status and treatmentregimen. Counseling should be nondirective and individualized and decisionsshould be made jointly between the patient and physician. Each patient shouldbe provided with all information needed to consider fully the relevant issues.We would counsel most HIV-infected pregnant women in the United States andother developed nations that if their HIV disease is well controlled andthe viral load is suppressed with combination therapy, there is no evidencethat prophylactic cesarean delivery would reduce the risk of vertical transmissionbeyond the already low rates associated with combination agents. For thosereceiving zidovudine monotherapy, we would counsel that the risk of transmissionis moderate (approximately 5%)[11,12] and that the data are only suggestivethat prophylactic cesarean delivery would be of protective benefit. Particularlyrelevant for women not receiving any antiretroviral therapy would be a discussionof the randomized trial evidence for a substantial reduction in transmissionrisk associated with prophylactic cesarean delivery. However, we would furthercounsel these women that cesarean delivery is associated with an increasedrisk of maternal morbidity and mortality in immunocompetent women[41] andthat this risk is likely potentiated by HIV infection. Thus, for women notreceiving antiretroviral medication, we would offer short-course antiretroviralsas an alternative intervention of proven efficacy. By extension, for womenin the developing world, where access to safe cesarean delivery is limited,we suggest that short-course antiretroviral regimens seem preferable fromthe perspective of both maternal morbidity and cost.


Caring for HIV-infected patients in the present climateof changing practice standards and rapidly evolving data is a challengefor all clinicians. When the additional consideration of protecting a developingfetus is included, decision making becomes even more complex. As patientslook for guidance, physicians must remember that many of them are in a positionof particular emotional vulnerability and may be willing to take on almostany risk to better, even in the slightest way, their odds against perinatalHIV transmission. This group of obstetrical patients may therefore be unlikelyto refuse cesarean delivery if it was offered. Adoption of prophylacticcesarean delivery as a matter of public health policy now may expose manyHIV-infected women to the cumulative morbidity and mortality of a procedurefor which benefit has been established only for a limited few. Thus, wesuggest restraint in the use of prophylactic cesarean delivery while weawait more data on the characterization of cesarean-associated operativemorbidity and the ability of combination drug therapy to prevent perinatalHIV transmission.