## D.1. Introduction

The clinical evidence suggests that babies with Rhesus and ABO haemolytic disease receiving IVIG are less likely to require exchange transfusion, an expensive procedure with associated morbidity and mortality.^{241} However, IVIG is also a relatively expensive therapeutic intervention and there is a shortage of global supply. Therefore an economic evaluation was undertaken to help guide GDG recommendations. The analysis compared giving IVIG as an adjunct to phototherapy in babies with Rhesus haemolytic disease and ABO haemolytic disease where serum bilirubin is continuing to rise at more than 8.5 micromol/litre per hour against not giving IVIG to these babies.

## D.2. Method

A simple decision-analytic model was used to assess the cost-effectiveness of IVIG as an adjunct to multiple phototherapy in babies with haemolytic disease where bilirubin levels continue to rise. The structure of this model is shown in . Costs were taken from the perspective of the NHS and personal social services, which is in accordance with the NICE guidelines manual (www.nice.org.uk/guidelinesmanual).^{237} The cost of multiple phototherapy was not included in the analysis as it common to both treatment alternatives. However, the costs of exchange transfusion are important as the rationale for IVIG is that the rates of exchange transfusion will vary according to treatment. Naturally, the costs of IVIG are also an important cost input, as this is the treatment evaluated.

Decision tree for the IVIG cost-effectiveness model.

Health outcomes are measured in quality-adjusted life years (QALYs). Exchange transfusion is associated with mortality and morbidity^{241} and in this model the difference between the treatment alternatives in QALYs is assumed to only be a consequence of mortality arising from exchange transfusion. This was partly to simplify the analysis but also because any impact on QALYs from morbidity would be small relative to that from assuming causation between exchange transfusion and mortality. We assume that IVIG would have no adverse effects that would have important long-term morbidity.

## D.3. Model parameters

The treatment cost of IVIG was estimated using the unit costs in .

It takes 4 hours to administer IVIG and it was assumed that the drip was set up by a specialist registrar but that a nurse would supervise the treatment for the 4 hours. The treatment is provided as an inpatient procedure and therefore treatment also includes the resources involved in the occupation of a hospital bed. This was estimated by using the 2007/08 NHS Reference Costs^{242} excess bed-day cost for a non-elective inpatient with a major neonatal diagnosis.

The cost of an exchange transfusion was estimated using the 2007/08 NHS Reference Costs and the category of non-elective inpatient with a major neonatal diagnosis. The costs of treatment are summarised in .

The efficacy of treatment relates to the number needed to treat (NNT) with IVIG to avoid an exchange transfusion. This in turn influences the number of babies who avoid an exchange transfusion and its associated mortality. The clinical parameters used in the model are given in .

Finally, there is a range of cost–benefit inputs which reflect NICE methodology. These inputs are shown in . The QALY gain of an averted exchange transfusion death is an approximation of the discounted QALY gain from a life lived in perfect health and for an average life expectancy.

## D.4. Results

The results with base-case values are shown in and for Rhesus haemolytic disease and ABO haemolytic disease, respectively.

Cost-effectiveness of IVIG for babies with Rhesus haemolytic disease with base-case model values.

Cost-effectiveness of IVIG for babies with ABO haemolytic disease with base case model values.

With base-case values, IVIG appears to be cost-effective in babies with Rhesus and ABO haemolytic disease with incremental cost-effectiveness ratios (ICERs) of less than £20,000 per QALY. The treatment appears to be most cost-effective in babies with Rhesus haemolytic disease, which is because the data suggest a lower NNT to avoid an exchange transfusion. If more exchange transfusions are avoided, that has a beneficial effect on both ‘downstream’ costs and averted mortality.

## D.5. Sensitivity analysis

The results for the base-case analysis are only as reliable as the base-case inputs that produce them. There is important uncertainty in some of these inputs, especially with respect to exchange transfusion mortality and the NNT to avoid exchange transfusion. The 95% confidence intervals for the NNT were as follows:

In exploring this uncertainty, there is limited value in exploring scenarios where exchange transfusion mortality is higher than the base case, the NNT is lower than the base case and the cost differential between exchange transfusion and IVIG is bigger than the base case. That is not to say that such scenarios are implausible but rather they would simply reinforce the observed cost-effectiveness of treatment. Rather, we are more interested in subjecting the cost-effectiveness finding to scrutiny by observing the extent to which the cost-effectiveness would still hold with the least propitious but still plausible scenarios. Therefore, the sensitivity analyses take a ‘worst-case’ scenario with respect to clinical parameters and a threshold approach to costs.

### Analysis 1 – varying clinical parameters

In this analysis it was assumed that the mortality rate for an exchange transfusion was 3 per 1000.^{243} Furthermore, we took the upper limit of the 95% confidence intervals for the NNT. In this analysis the ICER for Rhesus haemolytic disease was £34,379. For ABO haemolytic disease the ICER was £243,381.

### Analysis 2 – varying treatment costs

In this sensitivity analysis the clinical parameters were maintained at their base-case value. The costs of IVIG were then increased to determine the value at which IVIG treatment would no longer be cost-effective. For ABO haemolytic disease IVIG treatment remained cost-effective for all IVIG treatment costs less than £2,421. For Rhesus haemolytic disease IVIG treatment would have to exceed £6,054 for it to no longer be considered cost-effective.

## D.6. Discussion

This analysis strongly suggests that IVIG is a cost-effective treatment in babies with Rhesus haemolytic disease as an adjunct to phototherapy where bilirubin levels are still rising. This finding seems reasonably robust with respect to uncertainty in model inputs. Even when the NNT was taken from the upper limit of the 95% confidence interval and much lower exchange transfusion mortality than the base case was assumed, the ICER was only just outside what would be considered cost-effective by NICE criteria. Also, with base-case clinical inputs the cost-effectiveness of IVIG in Rhesus disease babies was not sensitive to the costs of IVIG.

For patients with ABO haemolytic disease the cost-effectiveness of IVIG is less certain because of the higher NNT. The cost-effectiveness in this group is very sensitive to exchange transfusion mortality and the NNT within plausible ranges. While cost-effective in the base-case analysis, the ICER in the ‘worst-case’ scenario would not be considered to be a cost-effective use of scarce NHS resources.

While the sensitivity analysis did not suggest that the results were particularly sensitive to changes in IVIG cost, it should be remembered that if the true clinical inputs conferred a lower benefit with IVIG treatment then the importance of treatment costs as a determinant of cost-effectiveness would increase.

## D.7. Conclusion

The model seems to provide good evidence that IVIG treatment in babies with Rhesus haemolytic disease can be considered cost-effective and it therefore supports the GDG recommendation. IVIG treatment in babies with ABO haemolytic disease may also be cost-effective, as indicated by the base-case results. However, sensitivity analysis suggested this finding is subject to considerable uncertainty. Nevertheless, IVIG treatment in this group of patients is consistent with recent Department of Health guidance and is also likely to have a relatively small cost impact given the number of babies affected and therefore the GDG recommendation seems reasonable. Research to ascertain the cost-effectiveness of IVIG, especially in babies with ABO haemolytic disease, could be useful given the current evidence base.