Twenty-one RCTs (n=893) were included.
Half of the studies described adequate randomisation methods and in four the outcome assessment was blinded. The majority of the trials were small: a median of 37 women were enrolled (range: 6 to 200).
There was no difference in the impact on persistent severe hypertension between hydralazine and all other antihypertensives combined (RR 1.08, 95% CI: 0.78, 1.49; 14 trials). Statistically significant heterogeneity was found. When a subgroup analysis was performed, there was a higher rate of persistent severe hypertension with hydralazine than with nifedipine or isradipine and a trend towards a lower rate of persistent severe hypertension with hydralazine than with labetalol. Again, there was statistically significant heterogeneity.
Hydralazine was associated with higher rates of maternal hypotension than the other antihypertensives (RR 3.29, 95% CI: 1.50, 7.23; 13 trials). There was statistically significant heterogeneity. There was also a statistically significant higher risk of Caesarean section, placental abruption and oliguria with hydralazine than with the other antihypertensives.
When compared with all the other antihypertensives, there were more maternal side-effects of any kind with hydralazine (RR 1.50, 95% CI: 1.16, 1.94; 12 trials). There was statistically significant heterogeneity. When a subgroup analysis was performed, there were more headaches, palpitations and maternal tachycardia with hydralazine than with labetalol or ketanserin.
Foetal heart rate.
There were more adverse effects on foetal heart rate with hydralazine than with the other antihypertensives (RR 2.04, 95% CI: 1.32, 3.16; 12 trials). There was statistically significant heterogeneity.
There were more low Apgar scores at one minute with hydralazine than with the other antihypertensives (RR 2.70, 95% CI: 1.27, 5.88; 3 trials). There was a trend towards an increase in stillbirths with hydralazine. When a subgroup analysis was performed, hydralazine was associated with less bradycardia than labetalol.
The authors noted that all outcomes for which the RR was increased without statistically significant heterogeneity showed heterogeneity when the analysis was repeated using the risk difference.