Limitations
The major limitation of this study was the recall bias, as in all case–
control studies. However, an effortwas given to minimize this limitation
by choosing newly diagnosed consecutive patients and collecting
all necessary subjects in a small period of time. Regarding the selection
of the controls, the “healthy” volunteer effect may exist. Moreover,
selection bias may also exist, but, as mentioned in the Material and
methods section, an effort was given to reduce it by performing, when
possible, random selection of the controls, enrolling controls based
on their willingness to participate and not on any of the exposure
factors' distribution that would substantially alter the results (only
age-matching was performed), using the same catchment area as the
patients, and avoiding any “specific” sampling rule in the allocation of
the participants. All these procedures were applied to ensure that the
control group reflected the distribution of the exposure characteristics
of the referent population of the cases. Furthermore, an effort was
given to account for the most common confounders concerning the
association studied; nevertheless, residual confounding may always be
a limitation in this work. Moreover, people who collected the data
were commonly and well trained before the beginning of the study,
limiting the intra-investigators bias. Additionally, the choice of the
investigators aswell as the face-to-face interviewswith the participants
helped so as to limit, when possible, the difficulties that usually occur
from the use of Food Frequency Questionnaires: measurement errors
concerning the fact that many details of dietary intake are notmeasured
and the quantification of intake is not as accurate as with other dietary
assessment methods, the difficulty of the respondents to evaluate the
serving size of foods consumed as well as their “limited” sometimes
ability to recall their dietary intake during the last year. The effect size
measures used in case–control studies (i.e., the odds ratios) tend to
overestimate the actual effect of the cause on effect usually observed
in prospective studies; and thus, the findings should be interpreted
with caution.
LimitationsThe major limitation of this study was the recall bias, as in all case–control studies. However, an effortwas given to minimize this limitationby choosing newly diagnosed consecutive patients and collectingall necessary subjects in a small period of time. Regarding the selectionof the controls, the “healthy” volunteer effect may exist. Moreover,selection bias may also exist, but, as mentioned in the Material andmethods section, an effort was given to reduce it by performing, whenpossible, random selection of the controls, enrolling controls basedon their willingness to participate and not on any of the exposurefactors' distribution that would substantially alter the results (onlyage-matching was performed), using the same catchment area as thepatients, and avoiding any “specific” sampling rule in the allocation ofthe participants. All these procedures were applied to ensure that thecontrol group reflected the distribution of the exposure characteristicsof the referent population of the cases. Furthermore, an effort wasgiven to account for the most common confounders concerning theassociation studied; nevertheless, residual confounding may always bea limitation in this work. Moreover, people who collected the datawere commonly and well trained before the beginning of the study,limiting the intra-investigators bias. Additionally, the choice of theinvestigators aswell as the face-to-face interviewswith the participantshelped so as to limit, when possible, the difficulties that usually occur
from the use of Food Frequency Questionnaires: measurement errors
concerning the fact that many details of dietary intake are notmeasured
and the quantification of intake is not as accurate as with other dietary
assessment methods, the difficulty of the respondents to evaluate the
serving size of foods consumed as well as their “limited” sometimes
ability to recall their dietary intake during the last year. The effect size
measures used in case–control studies (i.e., the odds ratios) tend to
overestimate the actual effect of the cause on effect usually observed
in prospective studies; and thus, the findings should be interpreted
with caution.
การแปล กรุณารอสักครู่..
