Results
Literature search results
172 documents were obtained by searching the database through the
computer. After eliminating duplicates and irrelevant documents, 40
articles were included for further intensive reading of the full text,
and finally 13 articles18-30 were included in this
meta-analysis. The study selection process was indicated in Figure 1.
Figure 1 PRISMA flow diagram of study selection
The basic characteristics of the included literature and the
methodological quality evaluation
In the included 13 case-control studies, there were a total of 2754
patients with hypertension, including 1444 patients in the CI group and
1310 patients in the no CI group. The basic characteristics of the
included literature were shown in Table 1. We conducted methodological
quality evaluation based on the NOS evaluation scale, and the scores of
each literature were all above 5 points, and the quality of the articles
were good, see Table 2 for details.
Table 1 The characteristics of included studies
Table 2 The NOS quality score of included studies
Synthesized outcomes
24-hour blood pressure variation coefficient A total of 10
studies18,21-28,30 analyzed the coefficient of
variation of 24-hour systolic and diastolic blood pressure, a total of
2163 cases were reported, including 1127 cases in the CI group and 1036
cases in the no CI group. The heterogeneity analysis of the coefficient
of variation of systolic blood pressure showed that there was
heterogeneity among the studies (I2 =99%, P
<0.001,), so the random effects model was selected. The
results of meta analysis showed that the differences between the two
groups had statistical significance [MD= 3.54, 95% CI (2.48, 4.60), P
<0.001], indicating that the coefficient of variation of
systolic blood pressure increased significantly in the CI group(Figure
2A). The results of the analysis of the coefficient of variation of
diastolic blood pressure showed that there was heterogeneity among the
studies (I2 =98%, P <0.001), and the random
effects model was used. The results of meta-analysis showed that the
difference between the two groups was statistically significant
[MD=2.43, 95%CI (1.55, 3.31), P <0.001], indicating the
coefficient of variation of diastolic blood pressure in the CI group was
significantly increased(Figure 2B).
Standard deviation of blood pressure A total of four
studies18,23,24,29 included in this meta-analysis
analyzed the associations between blood pressure standard deviation and
cognitive function, with a total of 396 cases, including 185 cases in
the CI group and 211 cases in the no CI group. The meta-analysis of
systolic blood pressure standards showed that there was heterogeneity
among the studies (I2 = 81%, P = 0.001), so the
random effects model was chosen. Meta-analysis indicated the difference
between the two groups was statistically significant [MD=2.20, 95% CI
(0.27, 4.13), P = 0.03], indicating that the standard deviation of
systolic blood pressure increased in the CI group(Figure 2C). The
qualitative analysis of diastolic blood pressure standard deviations
showed that there was heterogeneity among the studies
(I2 = 61%, P = 0.05), so the random effects model was
used. The results of meta-analysis showed that the difference between
the two groups was statistically significant [MD=1.79, 95% CI (0.80,
2.79), P <0.001], and the standard deviation of diastolic
blood pressure of patients in the cognitive impairment group was
increased(Figure 2D)
Figure 2 The forest plots of synthesized outcomes
Mean blood pressure A total of 11
studies18-20,22-26,28-30 reported the mean blood
pressure with involvement of 1874 patients, including 821 cases in the
CI group and 1053 cases in the no CI group. There was heterogeneity
between the studies in the mean systolic blood pressure
(I2=74%, P <0.001), so the random effect
model was selected. The combined results showed that the difference
between the two groups was statistically significant [MD=3.73, 95% CI
(0.92, 6.53), P=0.009], indicating that the average systolic blood
pressure of the CI group was higher than that of the no CI group(Figure
3A). There was heterogeneity among the studies in the mean diastolic
blood pressure (I2 =97%, P <0.001).
Therefore, the random effects model was selected. The meta-analysis
results showed that the difference between the two groups was
statistically significant [MD = 5.41, 95% CI (0.42, 10.40), P =
0.03], indicating that the mean diastolic blood pressure of the CI
group was higher than that of the no CI group(Figure 3B).
Morning peak blood pressure drop A total of two
studies25,26 reported that morning peak blood pressure
with involvement of 357 patients, including 155 cases in the CI group
and 202 cases in the no CI group. There was significant heterogeneity in
the morning peak systolic(I2= 86%, P = 0.008) and
diastolic(I2= 93%, P< 0.001) blood pressure
drop, so random effects models were applied. The meta-analyses indicated
that there were no statistically significant differences in the
morning peak systolic
[MD=7.85,95% CI(-1.30,17.01), Figure 3C] and
diastolic[MD=4.44,95% CI(-6.00, 14.89), Figure 3D] blood pressure
drop between the CI group and no CI group(all P>0.05).
Figure 3 The forest plots of synthesized outcomes
Sensitivity analysis and publication bias
The sensitivity analysis of the included studies was carried out, each
included study was removed one by one to see the changes in overall
results. And we have found that no results were changed by removing any
single study. The funnel plots and Egger’s tests were performed for the
synthesized outcomes of the included study to detect publication bias,
and the results showed that there were
no publication bias (all
P> 0.05, Figure 4).
Figure 4 The funnel plots of synthesized outcomes