Abstract :
I did read the paper "The knowledge, attitude and behavior of HIV/AIDS patient's family toward their
patients before and after counseling" in a recent issue of IJMS.1 There are, however, some concerns
that I would like to mention here.
In the paper's Abstract, it was better to mention in detail the family members who participated in the
study. Also a description should have been provided on the type of the intervention. For example, it
had to be made clear wether the intervention was done on individual subjects or was performed on a
group on subjects. In Materials and Methods section, the type of the study, which was a one-group
semi-experimental pretest/post-test design,2 was not mentioned. Moreover, in the Acknowledgements
section, instead of using “investigation”, or “study”, the word “survey” was used, which is a different
type of study by itself.
No description was provided on the method of counseling. It is not known if the counseling were
conducted for single individuals or for a group of individuals. The theoretical basis of intervention,
which should have been thoroughly explained in introduction section as well as time, number and frequency
of sessions were not clear as well. Moreover, it was not clear if the counseling sessions for
children, mothers, fathers, siblings and spouse were the same.
Reliability coefficients for subscales of the questionnaire were not mentioned. Also, the paper could
have used more accurate ways to validate the questionnaire rather than using face and content
validity. As it can be seen in the attitude section of the questionnaire (table 2), instead of evaluating
the families' attitudes toward their patients, items number 2, 3, and 4 evaluated the participant’s
attitude toward the show that was presented at Behavioral Counseling Center. It would have been
better if the attitude dimension of the questionnaire had been measured by rating scales using
appropriate scoring system such as the Likert scale. The age range of the participants was 27-53
years; therefore, children as the patients' family members could not have been included in the study
as participants. In the results section, the frequency and percentage of spouses, which were a sizeable
portion of the participants, were not mentioned.
Since there are no appropriate non-parametric tests for analyzing most of the findings in behavioral
sciences research, it is recommended to use parametric tests with strong statistical power, even when
the data are discrete (items 8 and 9 in table 1) and do not meet the criteria of normality of distribution
and homogeneity of variance. 3 In the section of discussion, there was no an explanation as to why the
study was better than similar studies, which had used more accurate designs. Moreover, it was not
possible to conclude from the study that counseling had an explanatory role, although such a role had
been confirmed in many controlled randomized trials.
The paper indicated that family members were assumed as one individual in the processes. However,
it would have been better to evaluate each family member in a separate group, because each
group’s responses to intervention and assessment could be different. Therefore, it might not be possible
to know that the presented results were applicable to which group of family members. For instance,
if children were infants or adolescents, then it remains speculative as how could they help the
patient by referring to Behavioral Counseling Center. Therefore, it would have been better to select a
more detailed sample so that the results could be generalized to all groups of family members.
In conclusion, although the statistical tests employed were appropriately selected for the type of research
design used in the paper, the selected research design was not appropriate for a number of
reasons such as the effect of confounding variables including simultaneous events, impact of pretest
findings on post-test ones, and statistical regression. Such confounding variables could be a source
for extraneous variances, which may prevent the understanding of the relation between changes and
independent variable. 4-6 One-group designs are appropriate when we try to change a confirmed or
resisting characteristic, when we are able to remove disturbing factors with a high degree of confi-
