Predicting the Risk of Adverse Events in Pregnant Women With Congenital Heart Disease
Chu R, Chen W, Song G, Yao S, Xie L, Song L, Zhang Y, Chen L, Zhang X, Ma Y, Luo X, Liu Y, Sun P, Zhang S, Fang Y, Dong T, Zhang Q, Peng J, Zhang L, Wei Y, Zhang W, Su X, Qiao X, Song K, Yang X, Kong B.
J Am Heart Assoc. 2020 Jul 21;9(14):e016371. doi: 10.1161/JAHA.120.016371. Epub 2020 Jul 14.
Commentary from Dr. Helen Parry (Leeds UK), section editor of ACHD Journal Watch:
The aim of this paper was to produce a model to improve the accuracy of risk assessment in pregnant women with congenital heart disease.
A total of 318 pregnancies in women with congenital heart disease were included in the study. These were divided into a development cohort, used to create the risk model (n=213) and validation cohort, used to test whether the models were accurate (n=105). Models were created to assess both maternal risk and neonatal risk to the baby. Patients were excluded if maternal death occurred prior to delivery or if spontaneous abortion occurred prior to 28 weeks gestation.
A regression analysis model was used to identify independent risk factors for poor outcome. Variables were regarded as significant if the p-value was <0.10. Once the model was established, this was tested on the validation cohort and assess for positive predictive value, negative predictive value, sensitivity and specificity.
Ten per cent of patients gave birth vaginally. Forty-one (12.9%) of the women experienced adverse maternal events whilst 93 (29.3%) of neonates experienced adverse events. A number of risk factors for worse outcomes for the mother or neonate were identified including pulmonary hypertension, Eisenmenger’s syndrome, higher NYHA functional class, left heart obstruction, significant impairment in left ventricular function, oxygen saturations and sinus tachycardia.
There were 13 maternal deaths, 2 of which were due to heart failure developing after Caesarean section to remove the fetus following intrauterine death.
The risk models will help guide clinicians when looking after patients with congenital heart disease.
- N=318 is a reasonable sample size, especially in a study examining patients with congenital heart disease, which are notorious for small samples.
- There was an indication of sinus tachycardia being a risk factor for adverse outcome, which I don’t believe is quite so widely recognised as the other risk factors identified.
- Although the overall sample size is reasonable, there are actually very small numbers of patients in some of the subgroups. For example, there were only 8 patients with Eisenmenger’s syndrome in the validation cohort and it is difficult to conclude that the model is valid for this group with a sample of this size. It is recognised that studying large numbers of patients with Eisenmenger’s syndrome who become pregnant is incredibly challenging given that these patients would generally be advised not to conceive as they are too high risk. It raises the possibility that this is actually a self-selecting group of patients with relatively good physiology within the Eisenmenger’s group; otherwise, they would be unlikely to carry the pregnancy beyond the 28 week gestation period required for the inclusion criteria.
- The lack of inclusion of pregnancies that did not continue to the 28 week stage means the data cannot be used when counselling patients regarding risk of early miscarriage and is likely to exclude some of the highest risk mothers. The same is true of exclusion if maternal death occurred before delivery.
- The cut off value for statistical significance was p<0.10. This is unusually high; most papers do not regard a p-value of >0.05 as statistically significant.
- It is unclear whether the conclusions actually add anything to the assessment and management of patients with congenital heart disease who become pregnant. The risk factors identified are in keeping with those categorised under the modified WHO categorization and there are other obstetric scoring systems that already take these factors into account.
- It is difficult to see how management improves as a result of the risk scores generated as we already knew about the risk factors and the paper provides no information on whether intervention reduces risk. Does beta-blockade in patients with sinus tachycardia reduce their risk or is it detrimental as it impedes an appropriate response to reduced cardiac output?
Of note, it is interesting there is such a low proportion of vaginal deliveries. In our centre, vaginal delivery is preferred if at all possible.