As described above, it appears that elevated blood pressure is positively correlated with cognitive impairment in middle-aged individuals but that this positive association declines with increasing age. It is unclear whether this association is applicable across the entire range of blood pressures. To address this issue, SBP, DBP and MABP were transformed into categorical data (SBP<140 mmHg, 140 mmHg? SBP<160 mmHg, and SBP?160 mmHg; DBP<90 mmHg, 90 mmHg?DBP<100 mmHg, and DBP?100 mmHg; MABP<100 mmHg, 100 mmHg?MABP<110 mmHg, and MABP?110 mmHg). Model 6 was established using dummy variables, which were defined based on the blood pressure parameters (the SBP<140 mmHg, DBP<90 mmHg, and MABP1) than subjects with SBP<140 mmHg among middle-aged subjects but that ORs reduced as the age increased; among individuals older than 60 years, the ORs were lower than 1.0 (Fig 4A). For subjects with 140 mmHg ?SBP<160 mmHg, the trend of ORs was similar but less prominent than that observed for individuals with SBP?160 mmHg (Fig 4A). The results for MABP are similar to those for SBP (Fig 4C). For DBP, the DBP?100 mmHg and 90 mmHg ?DBP1) for middle-aged subjects, whereas ORs reduced as age increased (Fig 4B). The rate of change may be lower for DBP than SBP, and the ORs for DBP were lower than 1.0 only when age?70 years. Detailed information is presented in Fig 4.
When we transformed new continued parameters to the binary analysis, new pattern of relationship is actually undamaged, while the importance is actually reduced but nevertheless present (design 4, Table 3)
SBP, DBP and MABP were transformed into categorical data (SBP<140 mmHg, 140 mmHg?SBP<160 mmHg, and SBP?160 mmHg; DBP<90 mmHg, 90 mmHg?DBP<100 mmHg, and DBP?100 mmHg; MABP<100 mmHg, 100 mmHg?MABP<110 mmHg, and MABP?110 mmHg). The SBP<140 mmHg, DBP<90 mmHg, and MABP<100 mmHg groups were established as the reference groups. The confounding variables considered in model 6 were the same as those considered in model 5.
Our study showed that elevated blood pressure was positively correlated with cognitive impairment in middle-aged subjects, but that this positive association declined with increasing age and tended to become negative for elderly subjects. These changes in the relationships between the blood pressure parameters and cognitive impairment with age were prominent when the blood pressure parameters were considered as continuous variables (model 3, Table 3; model 5, Fig 3). The reduction of the significance was plausible due to the information loss and reduction of statistical amolatina recenzja power when continuous variables were transformed to binary data. For example, the relationship between cognitive impairment and SBP can be expressed as OR = 1.130?0.989 (age-55.5) per 10mmHg for 40?age?85. Further analysis indicated that this age-dependent association was particularly prominent in the SBP?160 mmHg, DBP?90 mmHg and MABP?110 mmHg groups (model 6, Fig 4). In model 6, the changes in OR for DBP were not exactly as predicted; in particular, OR was larger in the 60–69 age group than in the 40–49 and 50–59 age groups (Fig 4). On the one hand, the relationship between DBP and cognitive impairment may invert from positive (OR>1) to negative (OR<1) with an older age compared with the relationship between SBP and cognitive impairment, on the other hand, it ple size leading to stratified errors. In combination with the other analyses (model 3–5), model 6 appears to validate the age-dependent relationship between DBP and cognitive impairment.