Results:
In total 383 obese adults (123M, 260F) aged 36.0 to 55.0 years old
(Table 3) assigned to eight hypocaloric diets (five high-protein, and
three non-high-protein diets) were included across the studies. The
initial mean BMI and primary outcomes measured are listed in table 2,
they include changes in pre- and post- diet mean BMI, LDL-C, HDL-C, TAG,
and TC levels. Studies were 4 weeks to 9 months in length. Retention
rates ranged from 88.9%-100%. All studies reported statistical powers
as compared to baseline values as well as between diet groups, all
p-values are stated in table 2. Mean BMI change was significant from
-21.3% to -60.9% of a greater decrease in the high-protein diet group
as compared to the control diet groups in the two out of three studies
which had both a control and high protein diet group (Abete et
al. , 2009; D.A. et al. , 2015). In one study both diet groups
were high in protein and significant mean BMI changes were present in
both diet groups as compared to baseline (Johnstone et al. ,
2011). Significant mean BMI changes were present therefore in three
studies (Abete et al. , 2009; D.A. et al. , 2015; Johnstoneet al. , 2011), the high-protein group and control diets had a
mean BMI change of -2.28 kg/m2 vs -1.85
kg/m2 respectively. There were no significant changes
in mean LDL-C, HDL-C, TAG, and TC levels among diet groups in the three
studies which had both a control and high protein diet group (Abeteet al. , 2009; D.A. et al. , 2015; Petrisko et al. ,
2020). The one study where participants withdrew was before
randomization, and therefore whether or not an intention-to-treat
analysis would have been applied to that or any of the other studies
with 100% retention rates was unknown. The risk of bias assessment
(Figure 2) noted a high risk of bias in three out of four studies (Abeteet al. , 2009; D.A. et al. , 2015; Petrisko et al. ,
2020) The high risks of bias in the “deviations from the intended
interventions” domain in these three studies was due to a likely lack
of compliance to diets as participants may have eaten more than the
prescribed diet as subjects were not resident at a dietary unit. The
study by Johnstone et al. had complete control over dietary input
as participants were resident at a Human Nutrition Unit over the course
of the study, nonetheless some concerns arose in this study as
participants left the unit to attend their workplace, again indicating
possible calorie intake outside of the prescribed diet. However ethical
concerns arise in the nature of such dietary studies when complete
surveillance of participants is present, and therefore a dietary control
design as presented in the study by Johnstone et al . has
attempted to decrease bias in terms of dietary control to it’s best
abilities whilst remaining within ethical guidelines of such studies, as
such the relevance of the bias score in this domain is lessened in this
study. All the studies had not stated available pre-specified planned
outcome analyses (as seen in registered studies on ISRCTN or an
equivalent register of randomized control trials) and therefore leading
all of the studies to having some concerns in the “selection of the
reported result” domain. Finally the study by D.A. et al. had stated
that subjects were allocated “alternatively” to diets, which arises
concern in whether or not the subjects assigned to the diets were
randomly allocated.