SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

Objective To assess the efficacy of corticosteroids in the treatment of severe hyperemesis gravidarum refractory to conventional management.

Design Multicentre, double-blind, randomised, placebo-controlled trial.

Setting Inpatient gynaecology wards in eight collaborating centres.

Population Twenty-five women with severe hyperemesis of which 24 completed the one- week study period.

Methods Twenty-five women were randomised to receive either 40 mg prednisolone daily in two divided oral doses, or equivalent placebo tablets. If, after three days, a woman was still vomiting, medication was changed to the equivalent intravenous alternative (hydrocortisone 100mg twice daily or normal saline injections)

Main outcome measures Frequency of vomiting and the dependence on intravenous fluid replacement therapy after one week of treatment.

Results There was a non-significant trend towards improved nausea and vomiting and reduced dependence on intravenous fluids. However, steroid therapy led to an improved sense of wellbeing (P= 0.021), improved appetite (P= 0.039) and increased weight gain (P= 0.025) compared with placebo. There was no difference in pregnancy outcome between the treatment and placebo groups.

Conclusions This study supports a beneficial role for steroids in severe hyperemesis, but did not validate the hypothesis that they lead to rapid and complete remission of symptoms. The study was not large enough to demonstrate a significant improvement in the primary outcome measures.


1Introduction

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

Nausea and vomiting in pregnancy are common, affecting 60%-70% of all pregnancies. Hyperemesis, defined as persistent nausea and vomiting of such severity that the woman is unable to maintain adequate hydration, electrolyte balance or nutritional status, affects about 1% of pregnancies1. Severe hyperemesis, refractory to conventional management with intravenous fluids and antiemetics is a rare and disabling condition, associated with multiple hospital admissions, time away from work and the family, and psychological morbidity2. In extreme cases, women may request, or their obstetricians recommend, termination of the pregnancy. If inadequately or inappropriately treated, hyperemesis may cause Wernicke's encephalopathy3, central pontine myelinolysis3, and maternal death4. Infants of mothers with severe hyperemesis have significantly lower birthweights and birthweight centiles5–7. We have previously reported a case series of four women with severe hyperemesis who all showed dramatic improvement following the use of steroids8. Following this, some workers reported similar success9–12 although others did not13. A definitive, randomised, double-blind, placebo-controlled trial was therefore needed to determine whether corticosteroids should be a therapeutic option for this miserable and disabling condition. Severe hyperemesis is not common and therefore a multicentre study was required. Based on our pilot studies8,9, we hypothesised that corticosteroids would lead to rapid and complete remission of the symptoms of nausea and vomiting in cases of severe hyperemesis gravidarum.

2Methods

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

The study population included pregnant women with severe or prolonged hyperemesis gravidarum, drawn from inpatient gynaecology wards in the eight collaborating centres. To be eligible the women met the following inclusion criteria: onset of nausea and vomiting before 12 weeks of gestation, dependent on intravenous fluids for at least one week (first admission for hyperemesis) or 24 hours (second or subsequent admission for hyperemesis), receiving regular treatment with at least one anti-emetic, ketonuria on admission, mid-stream urine specimen that did not indicate infection, normal random blood glucose (<6.5 mmol/L) unless known diabetic, vomiting at least twice a day or nausea so severe that they were unable to eat or drink, and receiving regular treatment with oral thiamine or a single dose of parenteral thiamine. Women were excluded if they had received treatment with oral steroids in the previous two months, proven peptic ulceration requiring treatment in the previous five years or a nonviable pregnancy. Biochemical hyperthyroidism was not an exclusion as it is recognized that this is common in severe hyperemesis, although women were excluded if there were overt clinical features of thyrotoxicosis.

The study was approved by the local research ethics committees of each of the participating centres. Women eligible for randomisation were asked to give informed written consent to participate in the study. The intervention was a one-week course of either prednisolone, 20 mg (4 × 5mg tablets) orally 12 hourly, or equivalent placebo tablets. If, following 72 hours of the trial tablets, a woman was still vomiting or she was vomiting the tablets, and was still dependent on intravenous fluid and electrolyte replacement, the therapy was changed to an intravenous equivalent (i.e. hydrocortisone 100 mg 12 hourly) or normal saline as placebo. A woman discontinued the study if she was for any reason unwilling to continue, or if she developed gestational diabetes, infection or psychosis or other mental disturbance.

The women were interviewed by the local medical staff using a standardised patient questionnaire (see Appendix 1) at trial entry, 72 hours after and one week after commencing the trial medication. A standard data collection form was completed detailing the severity of nausea (0 to 10), wellbeing rating (0 to 10), the presence of ptyalism, the frequency of vomiting, and the amount of oral fluid and food taken over the preceding 24 hours. The primary outcome measures assessed were frequency of vomiting and the dependence on intravenous fluid replacement therapy after one week of treatment. The secondary outcome measures were length of hospital stay following randomisation, duration of intravenous fluid therapy following randomisation, severity of nausea, need for antiemetics, presence of ptyalism, wellbeing rating, intake of oral fluids and food, change in thyroid function tests and liver function tests and weight gain. Data regarding further pregnancy complications, delivery details and the neonates were collected following completion of pregnancy.

Our pilot study8 indicated a 100% reduction in the number of women still vomiting after one week of steroid therapy. Assuming only an 80% reduction in the steroid-treated group and a 30% reduction in the placebo control group, a sample size of 45 would provide a 90% power to detect such a difference at the 5% level. Because it was thought possible that steroids might have a marked benefit, an interim analysis was carried out after 24 patients were entered in November 1996 and reviewed by an independent data monitoring committee. If a convincing difference (of 3 standard deviations) was demonstrated in either of the principal endpoints between corticosteroids and placebo groups, consideration would have been given to premature termination of the study.

Data analysis was performed on an intention-to-treat basis. Proportions were compared using Fisher's exact χ2 test, and for other data the groups were compared using a nonparametric two-sample Wilcoxon rank-sum test (adjusted for ties).

The women were randomised individually using a computer generated allocation schedule. The code was held and the trial packs distributed by the pharmacy department at Queen Charlotte's hospital. The randomisation was stratified by centre. Each centre was allocated sequentially numbered trial packs held in pharmacy and each pharmacy department held a copy of the allocation schedule. The clinicians (assessors), nurses, midwives and subjects were blinded to the study medication. The prednisolone tablets were identical in appearance to the placebo tablets. The local pharmacists were blinded until the need for intravenous therapy. If a woman required intravenous therapy, the pharmacy department prepared injections of hydrocortisone (treatment arm) or normal saline (placebo arm). Other aspects of management including anti-emetic therapy remained under the direction of the local clinicians. The randomisation code was broken for each subject following completion of the one-week assessment. This was to allow for gradual weaning of the steroid dose in those women randomised to the treatment arm, and for the option of use of steroids in those allocated to the placebo arm.

3Results

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

A trial profile is given in Fig. 1. We did not collect data on the total number of eligible patients in each centre over the time period of the study. Between April 1995 and December 1996, 25 women were recruited. The numbers of patients allocated to each arm in each centre was similar.

image

Figure 1. Trial profile.

Download figure to PowerPoint

Entry of patients ceased abruptly after December 1996, prior to the planned recruitment of the 45 subjects required. Results of the interim analysis and meeting of the data monitoring committee were not disclosed to the coordinators or collaborators. The study continued for a further two years but no further subjects were entered. This was found to be due to a combination of different factors in different centres, including the departure of key members of staff, and the erroneous belief that steroids had had such a dramatic beneficial effect that continued randomisation of women was not justified. Of the 25 women recruited, one underwent termination of pregnancy on the day following randomisation and was therefore excluded from further analyses.

The women randomised to receive steroids and placebo were well matched for centre, parity and singleton pregnancies, and severity of hyperemesis (Table 1). However, there was an unexplained discrepancy between the two groups in the number of women who fulfilled the inclusion criteria because it was their second or subsequent admission. Thus the number of women randomised during their first admission was five in the placebo group compared with one in the steroid group. The mean gestation at trial entry was higher in the steroid group (10.6 SD 2.1 weeks) than in the placebo group (8.3 SD 1.9 weeks). There did not appear to be any association between centre and gestation at randomisation or admission number, or any evidence of randomisation violation. We conclude that the discrepancy is due to chance alone (P<0.01).

Table 1.  Status at randomisation.
 Placebo (n= 12)Steroid (n= 12)
Primiparous women55
Singleton pregnancies1112
Triplets10
First admission (1 placebo and 1 steroid not known)51
Gestation [weeks]: mean [SD]8.3 [1.9]10.6 [2.1]
Weight [kg]: mean [SD61.8 [15.2]68.9 [19.8]
No. vomiting ≥ 5 times per day66
No. requiring > 1 antiemetic24
Ptyalism45
Biochemical hyperthyroidism47
Elevated liver function tests21
Blood glucose [mmol/l]: mean [SD]4.5 [0.8]4.8 [0.7]

Primary outcomes are shown in Table 2. After one week seven women in the placebo group and five women in the steroid group were still vomiting, of whom five in the placebo and two in the steroid group were vomiting 5 or more times daily.

Table 2.  Primary outcome measures. IVI = Intravenous infusion. *P= 0.26.
 Placebo (n= 12)Steroid (n= 12)RR (95% CI)
No. still vomiting at 1 week751.4 (0.6–3.2)
No. vomiting ≥ 5 times/day522.5 (0.6–10.5)
Reduction in vomiting score: median (range)*1.5 (−3.0 to 4.0)2.0 (−1.0 to 4.0) 
No. on IVI at 1 week331.0 (0.2–4.0)
Converted to IV Rx632.0 (0.6–6.2)

Severity of vomiting was coded 0 to 4, and the mean reduction in vomiting score for the placebo group was 1.2 (median 1.5) compared with 2.1 in the steroid group (median 2.0). After one week, three women in each group were still receiving intravenous fluid therapy. Six women in the placebo group and three in the steroid group converted to the intravenous form of intervention after 72 hours.

The results of the secondary outcomes are summarised in Table 3. Length of hospital stay after randomisation was similar in both groups. The median duration of intravenous fluids after randomisation was 96 hours in the placebo group (mean 96.7) and 72 hours in the steroid group (mean 90.5)(NS). There was a significantly greater improvement in the wellbeing rating (median 6.5 versus 3.5; mean 6.1 versus 2.9) in the steroid, compared with the placebo group (P= 0.021) and a significantly greater improvement in the amount of food taken (median point score rise 2.0 versus 1.5; mean 2.6 versus 1.2) in the steroid group (P= 0.039). The mean weight gain during the one-week study period was 1.56kg (median 1.25) in the steroid group and -0.13 kg (-1.0 kg) in the placebo group (P= 0.025). We found a non-significant greater reduction in the severity of nausea (mean 6.3 versus 3.4; median 6.5 versus 4.0) in the steroid, compared with the placebo group, and a greater increase in the amount of oral fluids taken. Six women in the steroid group compared with three women in the placebo group, were no longer requiring antiemetics at one week (relative risk 1.8 [0.8-4.33]).

Table 3.  Secondary outcome measures. Data are expressed as median (range). Relative risk (RR) and (95% confidence interval) are given for proportions.
 Placebo (n= 12)Steroid (n= 12)RR (95% CI)
Length of stay post randomisation (days)7.0 (2.0 – 26.0)7.0 (2.0 – 21.0)0.84
Duration of IV fluids (h)96.0 (24.0 – 168.0)72.0 (0 – 168.0)0.92
Wellbeing rating improvement3.5 (−2.0 to 8.0)6.5 (1.0 – 10.0)0.021
Food intake improvement1.5 (−2.0 to 4.0)2.0 (1.0 – 4.0)0.039
Weight change (kg) −1.0 (−2.0 to 4.5)1.25 (−0.5 to 5.0)0.025
Nausea score improvement4.0 (−5.0 to 9.0)6.5 (2.0 – 10.0)0.10
Drinking score improvement2.0 (−1.0 to 4.0)3.0 (1.0 – 3.0)0.54
No. on no antiemetics at one week3/126/111.8 (0.8–4.33)
Ptyalism at one week2/43/5 
No. readmitted for hyperemesis851.6 (0.7–3.5)

Five women in the steroid group and four in the placebo group complained of ptyalism, of whom three and two respectively, still had this symptom after one week. Seven women in the steroid group and 4 in the placebo group had biochemical hyperthyroidism (arbitrarily defined as free T4 > 20 or TSH <0.1) at trial entry. Of these, one in the steroid group and two in the placebo group had reverted to normal after one week. Two women in the placebo group and one in the steroid group had mildly elevated transaminases at trial entry. These reverted to normal in all women after one week. Blood glucose increased by 1.36 mmol/l between trial entry and one week in the steroid group compared with 0.43 mmol/l in the placebo group (data not shown). At trial entry none of the women had random blood glucose > 6.0 mmol/l. At one week four women in the steroid group, compared with one in the placebo group, had random blood glucose > 6.0 mmol/l, with one > 7.5 mmol/l (7.8 mmol/l).

Eight women in the placebo, compared with five in the steroid group, were readmitted later in the pregnancy for hyperemesis. In the placebo group, seven were “crossed-over” and received steroids to treat hyperemesis at some stage after the trial week.

Pregnancy outcome is summarised in Table 4. One woman in the steroid group was lost to follow up. One woman in the placebo group underwent termination of pregnancy one week following completion of the one-week study period. There was one stillbirth at term in the placebo group and one triplet pregnancy (also in the placebo group), resulting in three premature live births. There were three neonatal deaths, one in each group, both related to premature delivery and chorioamnionitis, and one eight week old triplet in the placebo group. There were four premature deliveries before 37 weeks in the placebo group; one in a triplet pregnancy, one due to chorioamnionitis and two others. In the steroid group, there were two premature deliveries; one due to chorioamnionitis and one other. There was no difference in mean or median birthweights or gestational age at delivery (excluding the triplet pregnancy) between the two groups and no difference in the numbers of babies born with birthweights less than the 5th centile.

Table 4.  Pregnancy outcome in 24 women completing one-week study period. Data are expressed as median (range) or absolute numbers. LB = livebirth; SB = stillbirth; TOP = termination of pregnancy; NND = neonatal death.
 Placebo (n= 12)Steroids (n= 12)
  1. aExcluding triplets.

Birthweight (kg)a3.14 (0.46 – 4.6)3.1 (0.59 – 3.98)
Gestation (weeks)38.5 (22 – 41)40 (24 – 41)
Birthweight < 5th centile11
Delivery < 37 weeks42
Outcome1 SB (37 wks)1 NND (24 wks, chorioamnionitis)
 1 TOP (8 wks) 
 2 NND (1 at 22 wks, chorioamnionitis; 1 triplet 8 wks old) 
 1 triplet (3 × LB)10 singleton LB
 8 singleton LB1 lost to follow up

4Discussion

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

Hyperemesis is a heart-sink problem for gynaecologists, possibly because of the frustration at not being able to cure such distressed women, and also because prolonged or repeated admissions lead to blocking of cold surgery gynaecology beds. In its severe, but fortunately rare, form substantial maternal morbidity and occasionally mortality may result. Since corticosteroids are effective in the treatment of vomiting induced by chemotherapy, it was thought that they might also be of use in other conditions, such as hyperemesis in which vomiting is thought to have a central origin involving the chemoreceptor trigger zone. Steroids were first used successfully to treat hyperemesis almost 50 years ago14.

Our earlier studies8,9 and those of others10,11 suggest a beneficial role for corticosteroids in the treatment of severe hyperemesis gravidarum. We used 40-60 mg of oral prednisolone or intravenous hydrocortisone 100mg twice daily in 10 patients with severe hyperemesis, including some who had had previous terminations of pregnancy for hyperemesis and one on total parenteral nutrition. The response to steroids was dramatic, rapid and complete in all patients8,9. Taylor10 used intravenous hydrocortisone 50 mg twice daily in seven patients and found that vomiting stopped within 3 hours of the first dose in all women. Safari et al.11 used 48 mg of oral methylprednisolone (equivalent to 60 mg prednisolone) in 18 women with intractable hyperemesis including six on total parenteral nutrition. A complete response was noted in 94% within three days. None of these studies were controlled. Safari et al.12 have also reported the successful use of oral methylprednisolone in a randomised, double-blind study of 40 patients with failed outpatient therapy of hyperemesis. The control group received oral promethazine, not placebo. There was no difference in the numbers of women in each group who stopped vomiting within two days, but no patients in the steroid group compared with 5 in the promethazine group required readmission within two weeks of discharge12. However, steroids have failed in at least one patient13, and intramuscular adrenocorticotrophic hormone (ACTH) has been shown to be no more effective than placebo in the treatment of hyperemesis15. This randomised placebo-controlled study was designed to answer the clinical question “Do steroids improve the symptoms of severe hyperemesis?”.

The study was terminated prematurely due to lack of recruitment from all participating centres. The main reasons for this were the rarity of severe hyperemesis and key personnel leaving the collaborating centres. Only half the estimated numbers of women required to demonstrate a significant beneficial effect of steroids were randomised. Furthermore, the lack of matching with regard to number of previous admissions may have biased our results. We cannot explain this difference. The randomisation procedure discussed above was rigorous. It seems unlikely that patients were admitted to the trial on the basis of knowledge of therapy to be allocated since no single centre accounted for the effect, and in any case randomisation was central. We have to conclude that this was a true random effect. If one assumes that those women admitted on more occasions at later gestations have a more resistant form of hyperemesis, the bias would be against a beneficial steroid effect. But if one assumes that those women for whom it was the second or subsequent admission at the time of randomisation were less sick, since the inclusion criteria stipulated that they should have been on intravenous fluids and not improving for only 24 hours, as opposed to one week as for women in their first admission, then the bias would be in favour of steroids. Furthermore, women in the steroid group were on average two weeks more advanced in gestation and it could be argued that they were more likely to improve anyway (i.e. a bias in favour of steroids).

Although we specified the severity of vomiting necessary for admission to the trial, vomiting in pregnancy may have a variety of causes. Thus some women may have been vomiting purely because of the hormonal effects of pregnancy but in others, emotional factors may have been dominant. It is possible that because of the pressure to enter patients into the trial, more of the latter group may have been recruited; and patients who are vomiting for emotional reasons may be resistant to steroids and other antimemetics.

Notwithstanding the above limitations, we have demonstrated that treatment of women with severe hyperemesis with steroids produces a significant improvement in the sense of wellbeing, the amount of food taken and the weight gain after one week, compared with a matched placebo group. In addition there was a non-significant reduction in the severity of nausea and amount of vomiting after one week, the duration of intravenous fluid therapy after randomisation, and an improvement in the oral fluids taken in the steroid compared with the placebo group. With larger numbers, these findings may have reached statistical significance. We did not find a significant difference in the primary outcomes of frequency of vomiting and the dependence on intravenous fluid replacement therapy at the end of one week of treatment.

As expected from earlier reports16,17, we found a substantial proportion of these women with severe hyperemesis to have abnormal thyroid (11/24, 46%) and/or liver function (3/24, 13%). Steroid treatment had no significant effect on these abnormalities. Steroid therapy affected glucose levels, with an increase in the mean random blood glucose of 1.36 mmol/l after one week of steroid therapy. One woman had a random blood glucose > 7.5 mmol/l after one week's treatment with 40 mg of prednisolone.

Women with severe hyperemesis appear to be a high risk group for adverse pregnancy outcome regardless of treatment. The premature delivery rate was 25% and perinatal mortality rate was 14% (3/21;excluding triplets). There was one case of chorioamnionitis in each group.

There has been much anxiety in the past18, and even recently19, concerning a possible association between steroid use in the first trimester and oral facial clefts. However, a very large case–control study failed to show any association between the rate of different congenital abnormalities and corticosteroid treatment in the second and third months of gestation20. Studies examining the use of corticosteroids for asthma in pregnancy similarly have failed to show any congenital malformations or adverse fetal effects attributable to maternal steroid therapy21. A notable common feature from reports of successful use of steroids for refractory hyperemesis8–11 is the frequent need to extend, gradually taper, or re-institute corticosteroid therapy following initial remission of symptoms. This raises issues concerning possible deleterious fetal effects of steroids later in gestation on preterm delivery22, fetal growth and lung and neuronal development23, and hypertension24. Similarly, there may be maternal adverse effects from prolonged steroid therapy in pregnancy, notably the increased risk of infections and reduced glucose tolerance and increase in gestational diabetes. Problems such as osteoporosis are unlikely since the dose of prednisolone can usually be reduced to 5-10 mg by 20 weeks of gestation9.

5Conclusion

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

Our experience highlights the difficulties recruiting subjects to therapeutic trials in obstetrics. This multicentre randomised double-blind placebo-controlled study did not have the power to support or refute the hypothesis that steroids lead to a rapid and complete remission of the symptoms of nausea and vomiting in cases of severe hyperemesis gravidarum refractory to conventional management. Nevertheless, we believe that the results are important for comparison with previous and future studies. We have demonstrated a beneficial effect of steroids on appetite, weight gain and wellbeing, and suggest that steroids may lead to improvement of nausea, vomiting and oral fluid intake and reduce the duration of dependence on intravenous fluids. A larger study will be needed to determine whether these findings are attributable to chance or whether steroids significantly improved nausea and vomiting.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

The authors would like to thank all the women who agreed to enter the study and the collaborators in the participating centres, including Dr C. Bottomley, Dr K. Harding, Professor P. Rubin, Dr K. Brackley, Dr J. Rutherford, Dr S. Crawshaw, Professor J. Walker, Dr J. Shillitto, Professor R. Taylor, Ms S. Jones, Dr J. Haley and Dr A. Mc Elligott. The collaborating centres included Queen's Medical Centre, Nottingham; Princess Mary Maternity Hospital, Newcastle-upon-Tyne; Bradford Hospital; Queen Charlotte's Hospital, London; University College Hospital, London; Homerton Hospital, London, St. James University Hospital, Leeds; St Thomas’ Hospital, London. The Medical Research Council provided a project grant to fund the study, which is gratefully acknowledged.

References

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix

Appendix

  1. Top of page
  2. Abstract
  3. 1Introduction
  4. 2Methods
  5. 3Results
  6. 4Discussion
  7. 5Conclusion
  8. Acknowledgements
  9. References
  10. Appendix
Table Appendix1..  Patient questionnaire administered at trial entry, 72 hours, and 1 week.
1. How would you rate your nausea on a scale from 0 to 10, with 10 being the most severe degree imaginable, and 0 being no nausea in the last 24 hours?
012345678910
2. How many times have you vomited in the last 24 hours?
0 12–4 5–7 >7
3. Are you able to swallow your own saliva?
     Yes No   
4. Have you felt like drinking over the last 24 hours?
Are you drinkingNormally
   Nearly normally
   Less than normal
   Virtually nothing
   Nothing?
5. Have you felt like eating over the last 24 hours?
Are you eatingNormally
   Nearly normally
   Less than normal
   Virtually nothing
   Nothing?
6. How would you rate your overall sense of wellbeing over the last 24 hours on a scale from 0 to 10, with 10 being awful, and 0 being wonderful?
012345678910