Scores versus clinical profiles in therapeutic decisions: a positive example from the Italian Medicines Agency (AIFA) decisions in the field of osteoporosis
Clinical Rheumatology
Scores versus clinical profiles in therapeutic decisions: a positive example from the Italian Medicines Agency (AIFA) decisions in the field of osteoporosis
Antonio Del Puente 0 1
Antonella Esposito 0 1
Luisa Costa 0 1
Francesco Caso 0 1
Raffaele Scarpa 0 1
0 Rheumatology Unit, Department of Clinical Medicine and Surgery, School of Medicine and Surgery, University of Naples Federico II , Via Sergio Pansini 5, 80131 Naples , Italy
1 Antonio Del Puente
Introduction
The attempt to mathematicise the cognitive approach to reality
can be traced back to ancient Greek culture. Mathematics is a
most powerful cognitive tool, whose potential is evoked by
the name itself, as mathesis means Blearning reality^. Actually,
and surprisingly enough, reality reveals a peculiar
correspondence between this approach and its intrinsic constitutive
mechanisms, and the development of a favorable cultural
context has disclosed the operative potential of this approach.
The link between science and praxis has changed completely
our relationship with reality. Mathematization has been a
driving element also in the field of medical care, generating levels
of progress that have recently involved also the strict clinical
milieu.
An example of this is the contribution to the definition of
risk profile for a given patient to develop a given outcome. In
Rheumatology, for instance, the use of diagnostic scores is
employed to assess the activity profile of the disease in
patients with Rheumatoid Arthritis (RA) using the 28-joint
Disease Activity Score (DAS28) [
1
], or to calculate the risk
of fragility fractures, using the fracture risk assessment tool
(FRAX) [
2
]. The importance of these tools, however, has
generated a tendency deserving careful analysis, as it goes
beyond a specific medical field, and that, in our opinion,
requires a more general evaluation for the possible problems
that may ensue. Indeed, this issue evolves in a totally different
way and with very specific implications when the approach is
*
automatically extrapolated to the domain of therapeutic
decision-making.
Scores in therapeutic decisions
Using mathematical scores to define a therapeutic approach
raises issues that cannot be underrated. Here are the three main
issues.
1. Overestimating the conditions for the indications to
treatment. As said, in RA, DAS28 is obtained combining four
numeric variables: tender joint count (TJC), swollen joint
count (SJC), visual analogue scale (VAS), erythrocyte
sedimentation rate (ESR) [
1
]. DAS28 is used to evaluate
the activity of arthritis in clinical trials, but this score is
being increasingly proposed as guide to treatment
decisions [
3
]. However, several other rheumatologic
conditions may cause overestimation of DAS28 thus
confounding its relationship with the clinical profile of arthritis.
VAS and TJC, for example, can also be affected by other
causes of pain not generated by arthritis, as for example
fibromyalgia. In addition, osteoarthritis, for example, can
generate pain in the hand and knee joints—conditions also
included in the DAS28. The score can be high, and
therefore suggestive of active arthritis, for other causes, making
it less reliable when dealing with questions as relevant as a
therapeutic decision [
4
].
2. Scores may emphasize statistically significant conditions
that are not relevant from a clinical point of view. A recent
article analyzed the results of trials comparing the
radiological progression of arthritic damage (assessed using the
Sharp/van der Heijde score), in patients treated with
biotechnological drugs or with methotrexate (MTX) [
5
]. For
instance, the TEMPO study on patients with early RA
showed a radiological progression of 3.3 units in two
years of treatment in patients randomized to MTX,
compared to 1.1 units in patients on etanercept, and −0.56 units
in patients treated with a combination of MTX and
etanercept [
6
]. The difference was statistically significant,
although, a 3.3 units difference is 0.7% of the total
number of units in the Sharp/van der Heijde score, where the
total score is 448 units. One objection could be that rarely
patients develop damage as high as 50% over the highest
value, suggesting that a value of 224 be considered as the
real value, i.e., 50% of the highest value. Here again, a
3.3 units variation in a two-year period is 1.5% of this
value. This change would not be clinically detectable in
the single patient, whereas a clinically relevant change (5
to 15 units) would be observable only after several years
and would not be of any practical use for patient
management [
7
]. Similar results in patients with early RA have
been reported with other biotechnological agents, with
statistically significant but not clinically relevant
differences vs. MTX in the individual patient [
8
].
Underrated role of the physician. Tendentially, the risk
with the use of scores in therapeutic decision-making is
that the professional role (...truncated)