DISCUSSION

The cost-effectiveness of Doppler ultrasonography in singleton high-risk pregnancies is highly dependent on the assumptions made in terms of cost and clinical effectiveness and on the economic model used. First, fetal death is rare, which means that the effect in terms of absolute risk reduction of fetal death is low and can only be estimated with wide confidence intervals. Second, the cost of providing Doppler ultrasonography is fairly high if the Doppler equipment is not used to full capacity. Third, as Doppler examination has to be provided to a large number of pregnant women to avoid one perinatal death, a small change in the unit cost of Doppler ultrasound can have a large impact on the total cost.

The economic model that was used in the present study has some limitations. It included only immediate health care costs and not costs related to the consequences of obstetric interventions. For example, a reduced cesarean section rate may influence the course and costs of future pregnancies and deliveries. In addition, patients' costs were considered only in relation to the Doppler ultrasound examination and not to other obstetric interventions or admittance of the infant to NICU. If these calculations had been included in the model, the incremental costs of saving an infant life would have been reduced.

In the cost calculations, the cost of the Doppler ultrasound device was calculated as the annual equivalent cost under the assumptions of a ten-year lifetime, an annual interest rate of 5%, and the need to buy a new Doppler device at each obstetric department. The use of alternative assumptions would alter the cost estimates. For example, if the Doppler equipment were also used as a diagnostic tool for other health problems, the cost per test would be lower (because the fixed cost would be shared across a greater number of tests). It would be too optimistic to assume, however, that the marginal cost of the equipment was zero.

In a Dutch study from 1995, Nienhuis showed that the implementation of Doppler ultrasonography in pregnancies with suspected IUGR could result in net savings of 15 million Dutch guilders merely through a reduction in antenatal admissions (8). However, the study did not assess total health care costs or the savings introduced by the Doppler program, nor did it assess cost-effectiveness.

Recommendations for issues to consider in evaluations of ultrasound were published in 1994 (1) and, since then, many studies have reported on the use of ultrasound during pregnancy, both as a screening program and for medical indications. However, a recent HTA showed that there are very few good quality economic evaluations and primary cost studies of the use of ultrasound in pregnancy (3). While routine scanning in the second trimester was found to be relatively cost-effective, the costs to the women who were attending for ultrasound examination were significant in comparison to the costs to the National Health Service (3). Others have discussed the considerable uncertainty about the relative economic efficiency of alternative programs for ultrasound screening in pregnancy and the weakness of the available evidence (9).

In the present study, the baseline data (the decentralized model) suggested that a program that offered Doppler ultrasonography to pregnant women with fetal IUGR would cost 1 million DKK per avoided fetal death. If a county model was used instead, the incremental cost per avoided perinatal death could be reduced to 563,000 DKK, whereas if Doppler ultrasonography was organized in a centralized model, the result would be a net saving. In the sensitivity analysis, the use of lower confidence intervals (i.e., more optimistic assumptions about the effects of Doppler examination) suggested that both the centralized model and the county model would give net savings. The implication of a net saving is that the additional cost of Doppler ultrasonography is offset by the savings from fewer interventions and more appropriate antenatal care.

The calculations illustrate, not surprisingly, that a decentralized service organization is more costly than a centralized organization, and thus highlight the tradeoffs that decision makers have to consider between the costs of providing services and the proximity and specialization of services. It is noted that, in the decentralized model, only a few Doppler examinations were performed per device, thereby making the program more expensive. However, the information gathered through the questionnaires that were sent to obstetric departments showed that all the thirty-two obstetric departments in Denmark used their Doppler equipment, and in fact there were seventy-eight Doppler devices in total (10;11). Furthermore, in the centralized model, there were also more Doppler devices than needed, as the calculations showed that 5,000 Doppler examinations could be done with two devices if all obstetric Doppler examinations were centralized to one or two departments.

The cost-effectiveness calculations were based on simplified assumptions that require validation before the results can be used for decision making. In the cost calculations, no adjustments were made to take account of productivity gains that would arise from economics of scale, nor have the benefits of clinical specialization been considered explicitly. In the event of a large expansion of the service, however, it is unlikely that the costs and effects would remain unchanged from the baseline.

It should also be noted that the effect data from the meta-analyses date back to the early 1990s (13). Clinical practice may have changed since then and may result in different effects than those suggested by the meta-analysis. For example, as a result of the use of Doppler ultrasonography, there are now practically no hospital admissions for pregnant women with fetal IUGR. Furthermore, the introduction of strict guidelines for the handling of IUGR pregnancies may result in even greater reductions in obstetric interventions (6). The increased use of Doppler technology has also resulted in the assessment of other high risk conditions such as twin pregnancies and gestational diabetes, as well as the examination of other blood vessels than the umbilical artery (such as the uterine arteries and other fetal vessels). The costs of obstetric Doppler ultrasonography, therefore, may increase even further without a proven benefit.

As long as the Doppler ultrasonography program dominates the no-Doppler alternative, then implementation is recommended. However, if the incremental cost of the Doppler program exceeds that of the alternative, the decision to implement or not should depend on the preferences of the decision makers and their willingness to pay for health improvement. It is likely that a cost-effectiveness rate of, say, 1 million DKK per avoided death would be acceptable, but if the figure is instead around 12 million DKK per avoided death, then some decision makers may think that the opportunity cost is too high and, thus, may be unwilling to support the program.