Economic Question

What is the cost effectiveness of immediate birth by planned induction of labour (henceforth ‘immediate birth’) compared with expectant management in women who have pre-eclampsia with mild or moderate hypertension of 34–37 weeks of gestation?

There are different resource implications and health consequences for mother and baby for these alternative policies. However, there is currently no evidence on the cost effectiveness of induction of labour in women who have pre-eclampsia with mild or moderate hypertension preterm compared with expectant management under regular monitoring. In view of this, the GDG requested a de novo economic analysis to help in its guideline recommendations.

Methods

The methods used are the same as those described for the term model (see Appendix I), except that this population consists of pregnant women who already have mild/moderate pre-eclampsia. In this population it has been recommended that there is no need to repeat quantification of proteinuria.

Model structure and assumptions

The model was developed in Microsoft Excel™ and in TreeAge Pro^®. The basic analytical approach is illustrated by the simple schematic in Figures J.1 and J.2 showing the decision tree for immediate birth (induction of labour) versus expectant management in women with mild to moderate gestational hypertension at 34–37 weeks of gestation. Pathways following assisted vaginal birth and emergency caesarean section are the same as those following spontaneous birth. Figure J.1 represents a sub-tree for spontaneous onset of labour and induction. Figure J.2 depicts the sub-tree for planned caesarean section.

Figure J.1

Spontaneous onset of labour and induction sub-tree for women with gestational hypertension.

Figure J.2

Planned caesarean section sub-tree for women with gestational hypertension.

Modelling effectiveness

There are no published effectiveness trials comparing immediate birth with expectant management in women with mild/moderate pre-eclampsia at 34–37 weeks of gestation. Two trials were found that compared the two policies before 34 weeks of gestation, which showed a clear association between immediate preterm birth and increased neonatal morbidity with no apparent decrease in maternal morbidity in women with severe pre-eclampsia.^145,138 Evidence from women with gestational hypertension at term, however, showed no difference in neonatal outcomes as all babies will have matured.¹²⁶

Owing to the lack of randomised trials in women with mild to moderate pre-eclampsia, for gestational age 34–37 weeks, data were taken from a retrospective case–control study in the USA by Habli et al.¹⁴⁵ The study was a secondary analysis of neonatal outcomes by week of delivery between 35 and 37 weeks of gestation. The neonatal outcomes included the percentage of babies requiring NICU admission, the mean duration of neonatal hospitalisation and the proportion of babies with neonatal complications. Neonatal outcomes for the immediate birth arm of the model were those reported at 35 weeks. The outcomes for expectant management were assumed to be those reported in weeks 36 and 37.

In the model, it is assumed that neonates who needed mechanical ventilation are managed in a high-dependency unit (HDU) and those who did not are managed in a special care baby unit (SCBU). Neonates with no complications are managed in the normal maternity ward.

Maternal morbidity (development of severe disease defined by the use of intravenous anticonvulsant medication) in women with pre-eclampsia was taken from Barton et al.⁹⁶ The GDG considered that severe morbidity was likely to be a rare event in women who are induced. However, they acknowledged that the disease can develop after giving birth and consequently estimated that about 1% of women develop severe disease in this group. Model probabilities are given in Table J.1.

Table J.1

Model probabilities used in the model by strategy in women with gestational hypertension.

Modelling costs

In accordance with NICE methods for clinical guidance,³⁸ a public sector, NHS and Personal Social Services (PSS) perspective was adopted.

The HYPITAT trial¹²⁶ showed that, on average, the immediate birth strategy had mothers delivering 1 week earlier than in the expectant management group at term. This meant that the expectant management group incurred an additional 1 week of usual monitoring costs as per the protocol. The average weekly costs per patient with mild to moderate pre-eclampsia were estimated to be £617. This was calculated assuming that, on average, women with moderate pre-eclampsia are hospitalised for at least 4 days and managed as outpatients for the rest of the week, while those with mild pre-eclampsia are admitted for at least 1 day. It was assumed that the women managed expectantly will deliver a week later than those who are induced immediately.

The first-line induction drug was assumed to be prostaglandins. If labour did not begin, women were assumed to be given oxytocin. The cost of two tablets of prostaglandins was £27. For oxytocin, set-up costs of £20 and disposables costs of £7 were assumed. The cost of the drug itself was £3.30. Women in the immediate birth arm were given two doses of intravenous dexamethasone (steroids) of 12 mg each. One dose costs £14.64 and hence the two doses cost £29.28.

The costs of the various modes of birth were taken from NHS Reference Costs 2006/07.²⁴⁰ For the costs of ICU and HDU, the GDG assumed that three organs would need to be supported. For women who did not develop severe disease, it was assumed that they remained in the general maternity ward. Only those who developed severe disease were assumed to be referred to HDU or ICU. The total cost of a strategy was thus the sum of hospital stay, induction costs, and mode of birth, and pre-admission costs for the extra 1 week in the case of the expectant management strategy. Model costs are shown in Table J.2.

Table J.2

Health service costs incurred by women who have pre-eclampsia with mild or moderate hypertension, 2008–09.

Sensitivity analysis

One-way sensitivity analyses were undertaken to assess the impact of changing input parameter values on the base-case results. The parameters that were varied were those that the GDG felt could possibly change model conclusions, across ranges suggested by the GDG. These included the incidence of severe disease, quality of life estimates, neonatal admission rates and pre-admission monitoring costs.

Results

Table J.3 shows the total costs and QALYs of pregnancy for women with mild to moderate pre-eclampsia. For the immediate birth and the expectant management strategies, the average total costs are £4,301 and £4,114, respectively. Immediate birth generates 28.305 QALYs compared with 28.240 QALYs for the expectant management strategy. The incremental costs of immediate birth over expectant management are estimated to be £187. However, immediate birth generates 0.065 extra QALYs compared with expectant management. The estimated incremental cost-effectiveness ratio (ICER) is about £2,900 per QALY. The results suggest that the policy of immediate birth is cost effective when compared with expectant management in women with mild to moderate pre-eclampsia preterm.

Table J.3

Cost effectiveness of immediate birth compared with expectant management in women with mild to moderate pre-eclampsia preterm.

Sensitivity analysis

Varying the incidence of spontaneous onset of labour in the expectant management strategy

In this analysis, the spontaneous onset of labour rate with expectant management was varied between 20% and 80%. Immediate birth remained cost effective with favorable ICERs for immediate birth. At a rate of 20%, the ICER was £1,640 per QALY, and this only rose to £6,369 per QALY when a rate of 80% was assumed. This is not surprising, given that 75% of women who have spontaneous onset of labour give birth vaginally, which is cheaper than assisted birth or caesarean section.

Varying the incidence of severe disease in the expectant management strategy

The incidence of severe disease has an impact on costs and the QALYs since in the model only those who developed severe disease were hospitalised in the HDU and ICU, with additional costs of anticonvulsant medication. The incidence of severe disease using expectant management was varied between 2% (suggesting there was little difference compared with immediate birth) and 30%. Figure J.3 shows that the model was highly sensitive to changes in this parameter. If it is assumed that there is a small difference, i.e. 2%, in the incidence of severe disease between the strategies, immediate birth is dominated by expectant management. Even if the incidence of severe disease in the expectant management arm is 12%, immediate birth is not cost effective at a £20,000/QALY threshold. The immediate birth strategy becomes cost effective if the incidence of severe disease in the expectant management group is 13% and above. The bigger the difference in incidence of severe disease between the strategies, the more attractive it is to offer birth immediately in women with mild to moderate pre-eclampsia.

Figure J.3

Sensitivity analysis showing cost savings of immediate birth compared with expectant management in women with mild/moderate pre-eclampsia preterm, varying the incidence of severe disease in the expectant management strategy.

Varying the incidence of emergency caesarean section after spontaneous onset of labour in the expectant management strategy

Rates of emergency caesarean section after spontaneous labour were varied between 5% and 30%. In the immediate birth strategy it was assumed that there was no spontaneous onset of labour: women were either induced or had planned caesarean section. Immediate birth remained cost effective across the range of the values tested. Changing the incidence did not alter the base-case conclusion that immediate birth was cost effective when compared with expectant management in women with mild to moderate pre-eclampsia.

Varying the pre-admission monitoring costs in the expectant management strategy

The average weekly cost of monitoring prior to admission in the expectant management strategy was varied between £100 and £650. At a monitoring cost as low as £100 per week, the ICER rose to about £11,000/QALY, still suggesting that immediate birth was cost effective. The conclusions are not sensitive to changes in monitoring costs (Figure J.4).

Figure J.4

Sensitivity analysis showing ICERs of immediate birth compared with expectant management in women with mild/moderate pre-eclampsia preterm, varying the pre-admission monitoring costs in the expectant management strategy.

Varying the neonatal admission rate in the expectant management strategy

In the base model, admission rates in the immediate birth strategy were about 57% compared with 33% in the expectant management strategy.

Neonatal admissions have an impact on costs since the cost of SCBU and ICU is more expensive compared with the general ward, and has an impact on the quality of life of mothers who are separated from their babies. Neonatal admission rates in the expectant management strategy were varied between 20% and 50%. At a neonatal admission rate of 20%, the ICER is approximately £8,000 per QALY. At admission rates of greater than 42%, expectant management is dominated by immediate birth.

Varying the NICU costs

In the base model, the NICU costs were taken from the NHS reference costs²⁴⁰ and were £1,423 per day (2008/09 prices). In the model, severe disease was approximated by rate of cerebral palsy. Only those neonates who had cerebral palsy were admitted into NICU. The costs of NICU stay were varied between £1,000 and £5,000 per day. This result shows that the model is not sensitive to changes in assumptions about NICU admission costs.

Varying the short-term utility loss due to pregnancy

In the base model, the quality of life weights used were obtained from Sonnenberg et al.²⁴³ Short-term utility loss was assumed to be about 0.03 over the 9 months. In this analysis, the health-related quality of life loss (‘utility’) was varied between 0.1 and 0 (see Figure J.5). With a loss of utility of 0.1, the ICER was approximately £3,400 per QALY. If it was assumed that there were no utility loss from pregnancy, the ICERs fell to £2,700 per QALY, suggesting that immediate birth is cost effective.

Figure J.5

Sensitivity analysis showing ICERs of immediate birth compared with expectant management in women with mild/moderate pre-eclampsia preterm, varying short-term utility loss due to pregnancy.

Varying the short-term utility loss due to development of severe maternal disease

In this analysis, the short-term utility loss due to the development of severe maternal disease was varied between 0.6 and 0.95. The ICER remained below £3,000 per QALY across this range, suggesting that the model results are not sensitive to changes in the quality of life assumptions surrounding development of severe maternal disease.

Discussion

The model demonstrated that the immediate birth strategy compared with the expectant management strategy in women with mild to moderate pre-eclampsia preterm is cost effective, with an estimated ICER of around £2,900/QALY. However, this finding is highly sensitive to the incidence of severe disease used in the model.

The risk of developing severe disease is considerably higher in the expectant management group. The HYPITAT trial,¹²⁶ which compared the two strategies in women with gestational hypertension at term, demonstrated that severe disease was reduced by half when women with mild/moderate gestational hypertension were offered immediate birth. This could be an important finding as admission to HDU and ICU due to development of severe disease has significant cost implications and adversely affects the quality of life of the women.

Effectiveness data were taken from observational studies^145;192;242 and a Cochrane review comparing vaginal prostaglandins used for third-trimester cervical ripening or labour induction with placebo/no treatment in unselected pregnant women.²⁴⁸ In the absence of published comparative data comparing the two policies, however, the GDG used expert judgement and observational data to populate the model. The GDG is aware of a continuing trial comparing immediate birth with expectant management in women with mild/moderate pre-eclampsia.

It is acknowledged that quality of life weightings (utility) data, derived from pregnant women without gestational hypertension and those hospitalised in ICU for any other reason, may not accurately approximate those for women who have pre-eclampsia with mild or moderate hypertension. However, given the lack of published quality of life data in women with pre-eclampsia, the GDG felt that this was the best estimate available for the quality of life for women with pre-eclampsia. Sensitivity analysis using different quality of life weights did not alter the cost-effectiveness outcome.

Conclusion

The model shows that the immediate birth strategy is cost effective compared with the expectant management strategy in women who have pre-eclampsia with mild or moderate hypertension at 34–37 weeks of gestation. However, the results need to be interpreted with caution in the absence of head-to-head trials comparing the two alternatives.