The policy suggestions presented in section 7.8 are unavoidably based on a num ber o f assumptions (table 4.1, 4.2, 4.3 pages 91-95) which might not be applicable to all districts. They are intended to provide a framework for rational decision making, rather than final solutions. Reassuringly however, the sensitivity analyses have shown that policy choice is generally robust to quite big changes in assumptions resulting from the extreme skew in the distribution o f the local prevalence o f fetal sickle cell disease that is evident in figure 7.1, page 129. Another limitation o f the policy framework is that district boundaries used were those from 1993, primarily because o f the availability o f national data on district ethnic compositons^^. By 1999 the 170 districts that existed in Great Britain at that time had merged into 120. On the other hand, although the commissioning unit is the health authority (and more recently also primary care groups), the unit within which the screening policy is implemented is the maternity hospital for antenatal screening and the community provider trust for neonatal screening. 1993 district boundaries are a convenient basis for the analysis o f local needs as they correspond more closely to provider catchment populations than the new districts, which might hide important local variations in ethnic composition. This is particularly true for districts with substanial cross boundary flow where maternity units and community provider trusts cannot be expected to deliver different screening services to residents o f different districts. In order to overcome these difficulties, we also presented approximate fetal sickle cell disease prevalence thresholds above which universal screening would be the preferred option, anticipating that more up-to-date local data might be available in some areas.
The framework gives the option o f constructing policy suggestions in favour o f universal screening for two reasons: firstly, because o f remaining uncertainty in parameter values it seemed appropriate to err by adopting universal screening in too many rather than too few districts. This was supported by the findings that adverse
screening effects in universal and selective programmes were very similar and
additional costs for universal screening were modest compared to total local health care budgets. Secondly, women living in areas with selective screening, and thus a higher failure to screen rate than for universal screening, are being asked to accept a lower standard o f service. This situation is in conflict with the aim o f the NHS to offer
an equitable service. However, the equity argument could be taken to an extreme to justify universal screening regardless of local demography. What was attempted here was a more balanced approach, which recognised both efficiency and equity issues.
The policy implications o f this study can be compared to those o f the SMAC report'^. SMAC recommended universal screening in districts where 15% or more o f the antenatal population are at risk of sickle cell disease. If this is interpreted as the proportion o f ethnic minorities, given our estimates o f local ethnic composition, 30 districts would qualify. This compares with the 7-15 districts that would adopt universal antenatal screening and 2-14 districts that would adopt universal neonatal screening under our baseline assumptions including a selective failure to screen rate o f 5.5%. The corresponding figures under assumptions favouring universal screening (‘best plausible case for universal screening’) but a similar selective failure to screen rate are 14-33 antenatally and 7-18 neonatally.
Cronin et al^^ suggested universal neonatal screening to be good value for money at a cost o f £25,000 to £100,000 for an additional case o f sickle cell disease detected. They calculated the corresponding fetal sickle cell disease prevalence thresholds to be about 10-30 per 10,000 births. Although the range o f our economic criteria differs (£10,000 to £50,000), approximate comparison using our baseline value o f £20,000, shows
accordance (table 7.33, page 159).
As crucial parameter values are likely to change over time the policy framework needs to be kept under review. Furthermore, the role o f Hb-pathy screening might
fundamentally change, for example because o f better treatment options (table 2.2,
page 18), or because antenatal screening might be replaced by preconceptional screening'*'^'^'^^^ Policies have to be adapted accordingly.
There are other conditions apart from Hb-pathies for which antenatal screening with PND is offered to give women choice over the outcome o f pregnancy. The main examples in the UK are screening for Down’s syndrome^^^ and Cystic Fibrosis^^"^. Unless screening policies are developed in an integrated way, rather than in isolation, unnecessary repetition of invasive prenatal sampling procedures and duplication o f pre-test information provision cannot be avoided.
The extent to which these research findings will be useful to local policy making cannot yet be evaluated". However, one can speculate about principal reasons for and
against their implementation, which are summarised in box 8.1. It was outside the
remit o f this study to examine operational details o f policy implementation.
Box 8.1 Principal reasons for and against implementation of the research
findings
R easons f o r implementation
• C ontinuing uncertainty am ongst policy m akers about the basis for the m ost appropriate
local policy.
• S uggested changes from a selective to universal policy do not require radical changes b u t expansion o f already existing services.
• A dditional resources required to change from selective to universal screening are sm all com pared to a district’s overall health care budget.
• Im proving the health o f B lack and m inority ethnic groups has been recognised as an im portant issue on the national public health agenda and H b-pathy screening is one effective intervention which can be offered.
Reasons against implementation
• D evelopm ent o f the decision model w as rem oved from local decision m akers w hich
m ight com prom ise ow nership o f the findings.
• Local circum stances, including priorities for service developm ent, availability o f
resources and political ‘b eliefs’ m ight prevent im plem entation o f the findings.
8.4
Conclusions
8.4.1 The model
The decision model is potentially a powerful planning tool: it made it possible to compare the cost-effectiveness o f universal and selective antenatal and neonatal Hb- pathy screening strategies for any given ethnic distribution o f the antenatal population, to identify parameters most influential on policy choice and to provide an evidence- based policy framework as guide for local decision makers. W hether policy decision will in practice be influenced remains to be seen.
“ Publication o f the health technology assessm ent on which this thesis is based is expected in the Autumn 1999.
The major limitation of the method was the lack o f quality data to support the model, particularly data about the fetal prevalence o f sickle cell disease and the differential coverage between universal and selective programmes.
Decision analysis is no substitute for obtaining reliable empirical data and it is important to consider that the method itself consumes substantial resources.
8.4.2 Results
For most districts universal compared to selective screening is not cost-effective but selective compared to no screening is cost-effective in all districts. Currently it is unlikely that antenatal screening will render neonatal screening redundant. There is therefore a need for all districts to adopt an explicit policy o f either selective or universal antenatal and neonatal screening. Cost-effectiveness depends mainly on the local prevalence o f fetal sickle cell disease, the coverage that can be achieved in a selective compared to a universal programme, laboratory costs, and the value placed on reproductive choice and late diagnosis o f sickle cell disease prevented. These
p a r a m e t e r s can be used to develop local criteria for policy choice. The coexistence o f
universal and selective programmes is inequitable as long as there is differential coverage between them. Therefore the failure to screen rates in selective programmes need to be minimised. This requires development and implementation o f standardised selection procedures based on ethnicity, setting o f minimum standards for coverage and monitoring o f coverage. To facilitate audit on screening perform ance improvement in routine information systems is needed.
Finally, antenatal and neonatal screening are part o f broader preventive and treatment services for haemoglobinopathies and should be fully integrated.
Main areas for research and development, where new findings might lead to change in universal versus selective policy decisions have been identified throughout this study and are summarised in box 8.2.
Box 8.2 Areas for research and development
A u d it
• To support quality control o f screening program m es and im prove data particularly on local
ethnic com position, fetal sickle cell disease prevalence and coverage.
• Preferably should include the screening process, adverse outcom es follow ing failure to screen
and m anagem ent o f affected children.
Im provem ent o f information systems
• To facilitate audit.
• R equires integration between maternity, neonatal and child health inform ation system s; and
• developm ent o f inform ation protocols and definition o f com puters system s
Guidance on laboratory tests fo r haem oglobinopathy screening
• To ensure consistent use throughout the U K o f m ost appropriate laboratory tests/equipm ent
for antenatal and neonatal screening (including sam ple collection for neonatal screening).
• C onsideration needs to be given to ensure effective im plem entation o f guidance.
P re-test information
• To determ ine how much and in w hat form at inform ation about a screening test needs to be
given before inform ed consent can be considered to have been obtained.
P ren atal diagnosis
• To establish the relationship between the various factors possibly influencing uptake o f
prenatal diagnosis, in particular tim ing o f the offer in relation to gestation, ethnicity, counselling.
Note; only areas for research and developm ent are listed where new findings might lead to change in universal versus selective policy choice.