Healthcare and Medicine Reference
184.108.40.206.2 How to Recognize Well-Constructed Algorithms
The construction of clinical algorithms should follow these clearly defined
1. The clinical problem to be solved by the algorithm must be well for-
mulated and defined37 37 (diagnostic, occurrence, target population, users,
expected results, etc.). Current practice must be described.
2. The expected results of algorithm use must be specified (gains for
practice, economy, patient, physician workload, etc.). “Discriminators,” 38
that is, points in clinical workup leading to a suggested action, are well
defined and described in explicit terms.
3. Clear indications are given as to when, where, and by whom the algo-
rithm is supposed to be used and to what kind of patients should it be
- Patients eligible for the algorithmic approach must be well defined
(target population for use of the algorithm).
- Situations in which algorithms should be used must be defined.
- Sites and settings are identified: emergency rooms, operating the-
aters, family practice offices, others?
- Users of algorithms must be identified.
These four specifications help in making pragmatic decisions (where,
when, and in whom an algorithm might and should be used).
4. The clinical situation statement → diagnostic statement → therapeutic
and/or diagnostic options must be clearly defined in a realistic time-
5. Enough data and information must be retrieved from the literature,
original studies, data analysis, clinical trials, consensus studies, medical
audits, or personal and other clinical experience, or simple “gut feel-
ings” or “lair” to justify each step of the algorithm. It must be clear,
from the description of the proposed algorithm, which of the above-
mentioned elements was used in its construction and justification; the
best available evidence for the purpose must support the algorithms'
6. Each node indicating action to follow must be supported not only by
the best evidence , but also by its justification through an argumenta-
tive process and its claims and conclusions in which the best evidence
is used. We may reason poorly even when we have the best evidence
7. Entry and exit points of the algorithm need to be defined.