hta_ knee intro.qxp - Ministero della Salute

The data loss due to a narrow study question, inclusion criteria, 5 year survival cut-off, and
focused outcomes is another potential limit of our study (see figure 9.1 that depicts the various
levels and causes of data loss through the study). For example the 50 year limit entailed the exclusion
of some large studies which reported ages in aggregate or as ranges (i.e. 18-75). Some of
these studies probably reported data which were relevant to our study, but had to be excluded to
avoid confounding and preserving our focus on OA.
In addition, our analyses are based on costs and effectiveness estimates which may not be
generalisable across nations or even regions of the same system. This is shown by the difference
in some effectiveness profiles for the same model of prostheses between the two registers. The
availability of more individual data from registers would have allowed more sophisticated comparisons
to be made, adjusted for example by risk factors such as level of physical activity, age and
co-morbidities. Finally, one of the manufacturers also told us that our weighted costs estimates
grossly overestimated the costs of their products, probably unfavourably impacting on their ICERs.
While this may be so, we had to rely on what respondents told us. Their cost data in turn reflected
local purchase prices, i.e. reality.
Figure 9.1: Flow of data loss through the study.
Objecttive
Reeport’s
Survey Systematic
review
DATA INPUT
Arthroplasty
register
Economic
evaluation
Data collection
Questionnaire not evaluable
(outcome, ( ppopulation, p follow-up) p)
(incorrectly compiled)
Reporting
Inclusion Criteria
(format and content)
(prostheses not in use in
Italy Italy, follow - up < 5 years years,
Objecti
population, clinical
Data sharing
indication)
ive
Partial information
(effectiveness or price)
CONCLUSIONS AND RECOMENDATIONS
Report’s
O
61