Critical appraisal of Prognostic studies
Are the results of the study valid? (Internal Validity)
1.Was the defined representative sample of patients assembled at a common
(usually early) point in the course of their disease)?
What is best?
Where do I find the information?
It is preferable if study patients are enrolled at
a uniformly early time in the disease usually
when disease first becomes manifest. Such
groups of patients are called an ‘inception
cohort’. Patients should also be representative
of the underlying population. Patients from
tertiary referral centres may have more
advanced disease and poorer prognoses than
patients form primary care.
The Methods section should describe the stage
at which patients entered the study (e.g., at the
time of first myocardial infarction; Stage 3 breast
cancer). The Methods section should also
provide information about patient recruitment,
whether patients were recruited from primary
care or tertiary referral centres.
This paper: Yes No Unclear
Comment:
2. Was patient follow-up sufficiently long and complete?
What is best?
Where do I find the information?
Length of follow-up should be long enough to
detect the outcome of interest. This will vary
depending on the outcome (e.g., for pregnancy
outcomes, nine months; for cancer, many
years). All patients should be followed from the
beginning of the study until the outcome of
interest or death occurs. Reasons for non
follow-up should be provided along with
comparison of the demographic and clinical
characteristics of the patients who were
unavailable and those in whom follow-up was
complete.
The Results section should state the median or
mean length of follow-up.
The Results section should also provide the
number of and the reasons for patients being
unavailable for follow-up. A comparison of the
two groups (those available and those
unavailable) may be presented in table form or
the authors may simply state in the text whether
or not there were differences.
This paper: Yes No Unclear
Comment:
3. Were outcome criteria either objective or applied in a ‘blind’ fashion?
What is best?
Where do I find the information?
A clear definition of all outcomes should be
provided. It is ideal if less objective outcomes
are assessed blindly, that is, the individual
determining the outcome does not know
whether the patient has a potential prognostic
factor.
The Methods section should provide a clear
definition or explicit criteria for each outcome
and whether determination is blinded to
prognostic factors will be found in either the
Methods or Results sections.
This paper: Yes No Unclear
Comment:
4. If subgroups with different prognoses are identified, did adjustment for important prognostic
factors take place?
What is best?
Where do I find the information?
A prognostic factor is a patient characteristic
(e.g., age, stage of disease) that predicts the
patient’s eventual outcome. The study should
adjust for known prognostic factors in the
analysis so that results are not distorted.
The Results section should identify any
prognostic factors and whether or not these
have been adjusted for in the analysis. Also look
at the tables and figures for evidence of this
(e.g., there may be separate survival curves for
patients at different stages of disease or for
different age groups).
