Table 6 summarizes the significant covariates other than disease histories and variables that represent taking medications in Models A (HD model), B (CBD model), and C (KD model). As expected, Age is a very important variable affecting these diseases. As shown in Figure 2, the odds ratios (ORs) comparing individuals aged 50–70 years were 2.09 with a 95% confidence interval (CI) of 2.01–2.18 for HD, 2.27 with a 95% CI of 2.12–2.41 for CBD, and 1.34 with a 95% CI of 1.22–1.45 for KD. Although Nawata (23) evaluated the risk factors for HD, the ORs and CIs were not evaluated. Since the probabilities of having these diseases in the next year are small, OR and CI are approximately equal to the probability ratio (PR) and its CI (57). The risk of HD and CBD in individuals aged 70 years is more than twice as large as those of individuals aged 50 years. The increase in risk was relatively small for KD. Female is significant at the 1% level in Model A but not significant at the 5% level in other models.

Weight_1 (recent weight change) and U_Protein (urine protein level) were significant at the 1% level in all models and were considered important variables. Figure 3 shows the ORs and 95% CIs of these variables for the models. For the dummy variable, the OR was calculated by comparing 0 and 1. For U_Protein, the majority of the values were 1 or 2, so the OR was calculated by comparing U_Protein = 1 and 2. For HD, the ORs of U_Protein and Weight_1 (95% CI) were 1.10 (1.0–1.13) and 1.20 (1.15–1.24.) For CBD, the ORs were1.13 (1.08–1.17) and 1.24 (1.17–1.31). For KD, the ORs were 2.23 (2.20–2.31) and 1.21 (1.11–1.30). Weight_1 increased the risk by approximately 20% for all diseases. U_Protein increased the risk by about 10% for HD and CD; however, it doubled the risk for KD. U_Protein is a risk factor with obvious mechanisms assumed (58) and people with high urine protein levels should recognize this fact, and special attention is necessary.

Another interesting finding is that among BP levels (SBP and DBP), only DBP in the CBD model was significant, and the estimates were not significant in the other models. This may raise questions regarding the 2017ACC/AHA guideline, which lowered the hypertensive criterion.

Figure 4 shows the ORs and 95% CIs for the variables of a history of other diseases. In the HD model, the ORs were 1.72 (1.54–1.89) and 1.71 (1.35–2.06) for CBD_t and KD_t. In the CBD model, the ORs were 2.15 (1.93–2.36) and 2.17 (1.61–2.74) for HD_t and KD_t. In the KD model, the ORs were 1.57 (1.34–1.81) and 1.58 (1.27–1.89) for HD_t and CBD_t. The risk of having CBD for individuals with HD and KD histories was especially high, more than double of those without disease histories, and special care is necessary for them (57).

All medications (antihypertensive, antihyperglycemic, and cholesterol-lowering) increased the risk of all three diseases. Figures 5–7 show ORs and 95% CIs, respectively. People often take two or three types of medications, and seven different variables are separately shown in the figures. These variables were divided into two groups. One group was treated with antihypertensive medications (i.e., M_BP, M_BP&GL, M_BP&CH, and M_BP&GL&CH), and the other was without antihypertensive medications (i.e., M_GL, M_CH, and M_GL&CH). In the HD model, the ORs (95% CI) were 1.93 (1.83–2.02), 1.81 (1.59–2.04), 2.21 (2.06–2.36), and 2.19 (1.93–2.45) for M_BP, M_BP&GL, M_BP&CH, and M_BP&GL&CH, respectively. In contrast, the ORs were 1.26 (1.10–1.43), 1.41 (1.31–1.52), and 1.29 (1.07–1.51) for M_GL, M_CH, and M_GL&CH. In the former group, the risks associated with antihypertensive medications were almost double compared with those without medications. In the latter group, the increments were much smaller, approximately 25%–40%.

In the CBD model, the ORs (95% CI) were 1.89 (1.75–2.04), 2.15 (1.77–2.53), 2.20 (1.98–2.43), and 2.17 (1.79–2.55) for M_BP, M_BP&GL, M_BP&CH, and M_BP&GL&CH, respectively. In contrast, the ORs were 1.47 (1.18–1.86), 1.66 (1.49–1.84), and 1.62 (1.22–2.02) for M_GL, M_CH, and M_GL&CH. In the former group, the risks associated with medications were almost double compared to those without medication, as in the case of heart disease. In the latter group, the increments were about 50%–65%. The results of this study suggest that antihypertensive medications increase the risks of HD and CBD, and the statement of Muntner et al. (11) is not supported.

Concerning the KD model, the ORs (95% CI) were 2.75 (2.46–3.93), 3.22 (2.44–4.01), 3.69 (3.20–4.18), and 3.59 (2.77–4.41) for M_BP, M_BP&GL, M_BP&CH, and M_BP&GL&CH, respectively. In contrast, the ORs were 1.76 (1.22–2.31), 1.61 (1.35–1.88), and 1.93 (1.17–2.68) for M_GL, M_CH, and M_GL&CH. In the KD mode1, although the CIs were larger, the increments in the risks seem much higher than those of the previous two diseases. The risks triple (or increase more) in the first group. The increments were about 60%–70% in the second group. Especially in the first group, even the lower bounds of the 95% CIs, the risks become 2.4–3.2 times larger than those without medications. This may imply that the negative effects of antihypertensive medications are much more serious in patients with KD.

The interactions between medications are important issues when individuals use two or more different types of medications. In the models without antihypertensive medications, the 95% CIs of M_GL and M_CH overlapped with the CI of M_GL&CH, and no significant interaction effect was observed. However, when antihypertensive and cholesterol medications were used simultaneously (M_ BP & CH), the risks were significantly higher (i.e., the 95% CIs did not overlap) than those using only antihypertensive medications (M_BP) in the HD and KD models.

Not only are the values of estimates of antihypertensive medications much higher than those of the other medications but the use of antihypertensive medications is also higher than that of other medications (antihypertensive medications: 11.18%, antihyperglycemic medications: 3.21%, and cholesterol medications: 7.46%). Here, we mainly discuss antihypertensive medications. For possible interpretations of the positive estimates, see Nawata (23).

Although there are some alternative therapies such as renal denervation (59), prescribing antihypertensive medications is a major therapy for hypertension. Over 200 drugs are available globally (60). Jackson and Bellamy (61) classified these medications into two group. One group is being those which directly or indirectly block the renin–angiotensin system (RAS) and the other group works through non-RAS pathways. The first group includes angiotensin-converting enzyme inhibitors (ACEIs), angiotensin-converting enzyme inhibitors, angiotensin receptor antagonists and direct renin inhibitors. The second group consists of adrenoceptor antagonists ($\beta$-blockers, $\alpha$-blockers), calcium channel blockers, diuretics (thiazides, loop, potassium sparing/aldosterone antagonist), vasodilators, centrally acting agents, and ganglion block.

Laurent (62) also classified the medications into $\beta$-blockers, diuretics, ACEIs, angiotensin II receptor blockers (ARBs), calcium-channel blockers, and other classes. Although the information of individual prescription records is not available, ACIs and ARAs are most widely used antihypertensive medications in Japan (63). RAS is a regulator of blood volume, electrolyte balance, and systemic vascular resistance. For summaries of RAS in the management of hypertension, see Fountain et al. (64) and Lange-Jacobs et al. (65). RAS is an extreme complex system in playing hypertension, and the understanding of RAS has expanded tremendously over the last few decades (65). Crifciler and Haznedarouglu (66) outlined intersections of circulating and local angiotensin systems in the vascular pathobiological microenvironment of central nervous system.

Since antihypertensive medications are widely used, their mechanisms of action, effectiveness, proper usage methods, and side effects have been widely studied (67–76). Lithell (67) reported that “both $\beta$-blockers and thiazide diuretics worsened insulin resistance and deteriorated lipoprotein metabolism. …These data may explain the unexpectedly high incidence of diabetes among hypertensive patients and the poor effect on risk for coronary heart disease in intervention trials.” Santos et al. (68) did systematic reviews to identify the possible pharmacogenetic implications for RAS-blocker medications in the hypertension-CKD scenario. They described that studies focusing on CKD were scarce. The CDC (77) and International Diabetes Foundation (78) described KD as a serious diabetes complication.

Marcum and Fried (79) mentioned that antihypertensive medications are frequently used in older adults with CKD, and the most common adverse drug events (ADEs) with antihypertensive use include acute kidney injury and antihypertensives, which may lead to medication errors and ADEs. Even in the Systolic Blood Pressure Intervention Trial (SPRINT) (16), which was heavily weighted in the 2017ACC/AHA guideline, it was admitted that “Rates of serious adverse events of hypotension, syncope, electrolyte abnormalities, and acute kidney injury or failure, …, were higher in the intensive-treatment group than in the standard-treatment group.” They also reported the percentages of the serious adverse events in both groups. These findings are consistent with the results of the present study. Although Ptinopoulou et al. (80) wrote that “it is now common knowledge that adequate blood pressure control is the most important factor for the preservation of renal function, so every drug that effectively lowers hypertension is believed to be renoprotective,” the findings of the paper do not support this statement. The risks of hypertensive medications were three times or more higher than those without medications for KD. Any medication may have negative side effects. In particular, the risk of KD is much higher for individuals taking antihypertensive medications. These medications must be prescribed with caution to minimize any negative side effect, and further studies are needed on the risks of antihypertensive medications.

We obtained slightly different results from those of previous studies. One possible reason might be a sample selection bias. In this study, we analyzed the general population. In other studies, special individuals with a high risk of diseases were analyzed. The rate of having heart diseases in the next year is about 0.5%. If the sample size is 10 thousand, the expected number of new patients is just 50 per year and it is sometimes difficult to get statistically significant results. In the SPRINT (16), they selected 9,361patients (4,678: intensive treatment; 4,683: standard treatment) at high risk for cardiovascular events. They admitted that “The lack of generalizability to populations … is a limitation.” Over 2.8 million observations were used in the study and the generalizability problem might be improved. For the theoretical details, see Nawata and Kimura (19).