J. Lord et al., Secondary analysis of economic data: a review of cost-benefit studies of neonatal screening for phenylketonuria, J EPIDEM C, 53(3), 1999, pp. 179-186
Citations number
42
Categorie Soggetti
Envirnomentale Medicine & Public Health","Medical Research General Topics
Study objective-To estimate the net financial benefit of neonatal screening
for phenylketonuria (PKU): by a simple pooling of cost data from the liter
ature; and by a more complex modelling approach.
Design-A systematic literature review was conducted to identify papers cont
aining data on the monetary costs and benefits of neonatal screening for PK
U. The methodological quality of the studies was appraised, and data were e
xtracted on resource use and expenditure. Monetary data were converted to c
ommon currency units, and standardised to UK incidence rates. Net benefits
were calculated for median, best case and worst case scenarios, and the eff
ect of excluding poor quality studies and data was tested. The net benefit
was also estimated from a model based on data from the literature and assum
ptions appropriate for the current UK situation. Extensive sensitivity anal
ysis was conducted.
Main results-The direct net benefit of screening based on the median costs
and benefits from the 13 studies identified was;pound 143 400 per case dete
cted and treated (pound 39 000 and pound 241 800 for worst case and best ca
se scenarios respectively). The direct net benefit obtained by the modellin
g approach was lower at pound 93 400 per case detected and treated. Screeni
ng remained cost saving under sensitivity analysis, except with low residen
tial care costs (less than pound 12 300 per annum), or very low incidence r
ates (less than 1 in 27 000).
Conclusions-The economic literature on PKU screening is of variable quality
. The two methods of secondary analysis lead to the same conclusion: that n
eonatal PKU screening is worthwhile in financial terms alone in the UK, and
that it justifies the infrastructure for collecting and testing neonatal b
lood samples. This result cannot necessarily be extrapolated to other count
ries.