Reply

Abstract

We thank Dr Knight for her comments on our article. She raises 2 issues: (1) that our study may overestimate the incidence of amniotic fluid embolism (AFE), and (2) that the UKOSS registry may yield data closer to the “true” population incidence. With respect to overreporting of AFE cases (ie, false positives), Dr Knight suggests that we did not undertake any validation beyond confirmation of pregnancy. Yet, we clearly state in both the Methods and Comments sections of our article that each of our 234 cases was reviewed for consistency in diagnosis (beyond pregnancy ascertainment) to reduce overestimation caused by hemorrhage, DIC, transfusions, etc. Moreover, unlike population-based cohort studies, registries rely heavily on voluntary reporting and are therefore prone to underreporting (ie, false negatives). Appropriate validation will reduce false-positive cases but will do nothing to uncover false-negative cases.Unlike a registry, population-based cohort studies do not select for diagnoses or depend on reporting of cases and thus should be far less prone to underreporting. To suggest that a registry may more accurately measure disease incidence than a population-based cohort study seems naïve. For rare diseases such as AFE, population based-cohorts and registries both play an important role. Rather than question whether the estimates should be 2 or 7 in 100,000, we hope that Dr Knight and her team use the UKOSS registry for its strengths and attempt to further our understanding of effective management approaches for this serious and often fatal disease. We thank Dr Knight for her comments on our article. She raises 2 issues: (1) that our study may overestimate the incidence of amniotic fluid embolism (AFE), and (2) that the UKOSS registry may yield data closer to the “true” population incidence. With respect to overreporting of AFE cases (ie, false positives), Dr Knight suggests that we did not undertake any validation beyond confirmation of pregnancy. Yet, we clearly state in both the Methods and Comments sections of our article that each of our 234 cases was reviewed for consistency in diagnosis (beyond pregnancy ascertainment) to reduce overestimation caused by hemorrhage, DIC, transfusions, etc. Moreover, unlike population-based cohort studies, registries rely heavily on voluntary reporting and are therefore prone to underreporting (ie, false negatives). Appropriate validation will reduce false-positive cases but will do nothing to uncover false-negative cases. Unlike a registry, population-based cohort studies do not select for diagnoses or depend on reporting of cases and thus should be far less prone to underreporting. To suggest that a registry may more accurately measure disease incidence than a population-based cohort study seems naïve. For rare diseases such as AFE, population based-cohorts and registries both play an important role. Rather than question whether the estimates should be 2 or 7 in 100,000, we hope that Dr Knight and her team use the UKOSS registry for its strengths and attempt to further our understanding of effective management approaches for this serious and often fatal disease. Amniotic fluid embolism: active surveillance versus retrospective database reviewAmerican Journal of Obstetrics & GynecologyVol. 199Issue 4PreviewAbenhaim et al1 reported an incidence of amniotic fluid embolism (AFE) of 7.7 per 100,000 births in a retrospective, population-based cohort study. Both this, and other population-based studies reported in the literature,2 ascertained cases through interrogation of a coded hospital admissions database. As other authors2 point out, diagnosis of amniotic fluid embolism is difficult, and coded data may be flawed,3 thus limiting the accuracy of incidence estimates obtained from retrospective database review. Full-Text PDF