The case series reported here suggests an ongoing concern with MHRA in cases of cardiotocography-associated fetal and neonatal death reported to the FDA between 2009 and 2019. The case series reported here is, to the best of our knowledge, larger than prior case reports and case series of perinatal mortality involving MHRA, most of which involve only one to two cases of fetal/neonatal death [2, 30,31,32,33]. This case series is also more contemporary, involving more CTG monitors with autocorrelation and signal ambiguity detection technologies, and is the first, to our knowledge, to bring to light the manufacturer’s response to such reports. We note that manufacturer recommendations (confirm fetal life by alternate means prior to CTG monitoring, utilize signal ambiguity technology, and/or better respond to alarms from signal ambiguity detection technologies) are not emphasized, if mentioned at all, in the intrapartum fetal monitoring/cardiotocography guidelines which we have reviewed [34,35,36,37,38,39,40,41].
We acknowledge that not all CTG monitors in clinical use include signal ambiguity technologies and that, even when these technologies are employed and utilized correctly, it is possible that they still may miss MHRA or Type 2 FHRA. Therefore, we believe that in addition to signal ambiguity detection technologies, healthcare provider education regarding the clinical features of potential CTG artefact is essential, as this may be critical when these technologies are not available, malfunction, or in cases which are not detected by those technologies (for example, MHRA with doubling of the maternal heart rate). The clinical warning signs for possible MHRA are outlined in various publications and should be emphasized in clinician training: sudden sustained improvement of a tracing to a normal pattern, when it was previously of poor technical quality or showed an abnormal FHR; a “normal” FHR showing accelerations with every contraction (the maternal response to pain or pushing) (See Fig. 2); the apparent presence of marked accelerations and decelerations; sudden change to a new baseline and /or fluctuation in the baseline or fluctuation in the character of the tracing (variation in appearance of the variability sometimes accompanied by intermittent loss of contact); doubling/halving of the rate (this may also be a feature of FHRA) [1,2,3,4, 42, 43]. These features are especially significant when they begin after a position change, gap in monitoring, epidural insertion, and especially in the second stage of labour with onset of pushing. They are less common with application of a fetal scalp electrode which will often give the true fetal heart rate signal (see Fig. 3), but may, in some circumstances still occur, particularly after fetal death with recording of the maternal heart rate [44]. For Type 2 FHRA, the key feature which should raise concern is a narrow or no difference between the presumably different FHR signals in multiple gestation.
To facilitate knowledge translation, we have created a supplementary file which highlights the key points for clinicians (Additional file 4).
It is also useful to consider the larger research context. MHRA may manifest in two distinct ways with different clinical implications. One, the MHRA may mask a normal fetal heart rate, potentially leading to unnecessary intervention or, two, the MHRA may mask an abnormal fetal heart rate, potentially leading to false reassurance. Pinto et al. have demonstrated that the former is more common, manifesting most often as periodic switching to the MHR giving the false appearance of fetal decelerations [45]. Our cases seem more likely to involve the latter and this is probably because we are analyzing the most extreme cases, those involving fetal/neonatal death. Of note, simultaneous computerized analysis of MHR and FHR is an ongoing area of research which shows great promise as a future direction [46, 47]. By demonstrating improvements in detection of fetal acidemia by computer analysis of both maternal and fetal heart rates, research groups working in this area have provided additional evidence to support continuously monitoring the maternal heart rate along with the fetal heart rate [46, 47].
Our study has several limitations. Our a priori study design has certain weaknesses, in particular the lack of a pre-published protocol. Admittedly our study methodology would have been stronger if all 117 reports retrieved had been reviewed by all authors, but we believe that by having all authors review the positive cases (defined as those involving probable MHRA and/or a manufacturer’s recommendation regarding MHRA), our methodology is sufficiently strong for the case series reported here. Of note, our a posteriori presentation of results is rigorous and accessible: by making all our positive data points transparent and compact, an interested reader can review all our positive reports in a short amount of time and draw their own conclusions. Further our search strategy can be easily and independently validated by searching the FDA MAUDE database for product code “HGM” along with event type “death” which retrieves over 90% of the reports we found. Several factors mitigate against bias: our endpoint, death, is objective and our exposure, MHRA, has often been effectively assessed a priori and independent of our study by the authors of the FDA reports. That being said, without a full review of the CTG trace and history in each case, it is, admittedly, difficult to state definitively whether MHRA is truly present and a causal factor. The data is inherently messy and several of the reports are incomplete.
However, it is our opinion that our study design was uniquely suited to the problem we are trying to address and allowed us to thread the needle to an important clinical finding which continues to be under-recognized. Prospective clinical study of maternal heart rate artefact is likely to be hampered by the Hawthorne effect as relatively modest healthcare provider training and education in this topic is likely to greatly reduce the risks of adverse outcome. Further, a large number of cases would need to be studied. Neilson et al. reported that in approximately 10,000 deliveries, they observed 5 cases of unanticipated adverse perinatal outcome due to signal ambiguity [2]. Further, we anticipate that proposals for large healthcare-system based retrospective studies of fetal and neonatal deaths with an aim of determining if they were due to failure to detect MHRA and, potentially, failure of healthcare providers to follow manufacturer’s instructions for use, is likely to face significant barriers to approval due to medico-legal implications. Reviewing the FDA MAUDE database allowed us to sample a significant number of severe cases while circumventing the barriers listed above. Manufacturers have an obvious incentive to conclude that their CTG monitor was not at fault. However, we consider it significant that they so often concluded that the true problem was failure of the healthcare team to detect MHRA rather than failure to simply interpret the CTG tracing in general.
The FDA MAUDE database most likely suffers from important levels of under-reporting, potentially leading to an underestimation of the magnitude of the problem. The reasons to not report adverse outcomes are well known to clinicians: medico-legal concerns, potential for loss of reputation, medical license, and/or employment, lack of institutional support, lack of clear protocols for reporting, a belief that reporting will be futile, hierarchy and its associated problems, and the second victim syndrome [48] which can demoralize the healthcare provider in the aftermath of an adverse outcome. Further, there is an additional reason not to report to the FDA: uncertainty as to whether the CTG monitor or the healthcare provider was truly at fault. We would expect only those cases where the healthcare provider felt reasonably certain the CTG monitor was at fault to be reported to the FDA. However, if there is under-reporting, our conclusions are potentially more important than suggested by the number of cases reported here. In fact, they may only be the tip of the iceberg. Even with a rate considerably lower than 5 in 10,000 for unexpected adverse perinatal outcomes due to signal ambiguity, as estimated in 2008 by Nielson et al. [2], the clinical implications may be important due to the potentially preventable nature of the adverse events, the large number of births worldwide (estimated at more than 130 million) [49] and the low baseline risk of neonatal encephalopathy associated with intrapartum events (for example, estimated at 16 per 10,000 births with an associated death rate of 10% or 1.6 per 10,000 in for “level 1” countries with neonatal mortality rates less than 5 per 1000 live births) [50]. There is one caveat. While the number of cases reported to the FDA of fetal/neonatal death involving MHRA is most likely a gross underestimate of the number of such cases which occur, it also seems likely that the percentage of fetal/neonatal deaths which involve MHRA reported to the FDA MAUDE database (40% in our study) is an overestimate of the percentage in reality. The reason is that, while under-reporting leads to an underestimate of the magnitude of the problem of MHRA, knowledge gaps regarding MHRA on the part of healthcare providers and the subsequent confusion when there is an adverse outcome likely make such cases somewhat more likely to be reported.
The FDA MAUDE database website states that the reports should not be used to estimate rates of events [14]. That being said, our quantitative results are less important than the answer to the simple qualitative question: is there enough evidence here regarding the need for increased healthcare provider training in CTG artefact detection and management, in particular signal ambiguity detection technology, that the subject merits greater attention in CTG clinical practice guidelines? We believe that the answer to that question is yes. To quote a patient safety expert from another field, anesthesiologist David Gaba wrote that “no industry in which human lives depend on the skilled performance of responsible operators has waited for unequivocal proof of the benefits of simulation before embracing it.” [51] An FDA MAUDE database study such as ours can never provide unequivocal proof that increased healthcare provider awareness and training about MHRA will prevent adverse outcomes. However, ours is not a purely academic pursuit. With potentially preventable intrapartum injury and death to babies that might otherwise be born healthy in the balance, we believe that urgent dissemination of the results presented here is essential.