Skip to main content
Journal of the American Medical Informatics Association : JAMIA logoLink to Journal of the American Medical Informatics Association : JAMIA
. 2021 Oct 13;29(2):335–347. doi: 10.1093/jamia/ocab173

Gender-specific clinical risk scores incorporating blood pressure variability for predicting incident dementia

Jiandong Zhou 1, Sharen Lee 2, Wing Tak Wong 3, Khalid Bin Waleed 4, Keith Sai Kit Leung 5, Teddy Tai Loy Lee 5, Abraham Ka Chung Wai 5, Tong Liu 6, Carlin Chang 7, Bernard Man Yung Cheung 8, Qingpeng Zhang 1,, Gary Tse 6,9
PMCID: PMC8757295  PMID: 34643701

Abstract

Introduction

The present study examined the gender-specific prognostic value of blood pressure (BP) and its variability in the prediction of dementia risk and developed a score system for risk stratification.

Materials and Methods

This was a retrospective, observational population-based cohort study of patients admitted to government-funded family medicine clinics in Hong Kong between January 1, 2000 and March 31, 2002 with at least 3 blood pressure measurements. Gender-specific risk scores for dementia were developed and tested.

Results

The study consisted of 74 855 patients, of whom 3550 patients (incidence rate: 4.74%) developed dementia over a median follow-up of 112 months (IQR= [59.8–168]). Nonlinear associations between diastolic/systolic BP measurements and the time to dementia presentation were identified. Gender-specific dichotomized clinical scores were developed for males (age, hypertension, diastolic and systolic BP and their measures of variability) and females (age, prior cardiovascular, respiratory, gastrointestinal diseases, diabetes mellitus, hypertension, stroke, mean corpuscular volume, monocyte, neutrophil, urea, creatinine, diastolic and systolic BP and their measures of variability). They showed high predictive strengths for both male (hazard ratio [HR]: 12.83, 95% confidence interval [CI]: 11.15–14.33, P value < .0001) and female patients (HR: 26.56, 95% CI: 14.44–32.86, P value < .0001). The constructed gender-specific scores outperformed the simplified systems without considering BP variability (C-statistic: 0.91 vs 0.82), demonstrating the importance of BP variability in dementia development.

Conclusion

Gender-specific clinical risk scores incorporating BP variability can accurately predict incident dementia and can be applied clinically for early disease detection and optimized patient management.

Keywords: blood pressure variability; risk score; risk stratification, dementia; predictive model

INTRODUCTION

Dementia is a global health concern, particularly in the face of the ageing population and its burden upon healthcare systems. Therefore, predictors for dementia are warranted for early diagnosis and intervention to improve patient prognosis. An increase in both systolic and diastolic blood pressure, below the threshold of hypertension, has been reported to be associated with increased dementia risk.1–3 Moreover, over the past decade, studies have shown that increased blood pressure variability (BPV) was found to be associated with an increased risk of dementia.4–8 However, its clinical application in dementia risk stratification has yet been explored.

Furthermore, studies have reported apparent gender differences in the risk factors for dementia.9–11 Several hypotheses have been raised for the increased dementia risk among women, including the peri- and postmenopausal hormonal changes, difference in apolipoprotein E4 allele inheritance and stronger inflammatory dysregulation.12–15 In addition, gender affects the clinical presentation of dementia, such as a higher frequency of visual hallucination, depression, sarcopenia and frailty among female patients.16–18 However, there is a lack of research on the identification and application of gender-specific dementia risk factors. Therefore, the present study aims to explore the genetic-specific prognostic value of BP and BPV in the prediction of dementia risk and establish clinical risk scores for risk stratification.

MATERIALS AND METHODS

Research design and data

The present cohort consists of patients admitted to government-funded family medicine clinics between January 1, 2000 and March 31, 2002. The patients were identified from the Clinical Data Analysis and Reporting System, a territory-wide database that centralizes patient information from government-funded hospitals in Hong Kong to establish comprehensive medical data, including clinical characteristics, disease diagnosis, laboratory results, and medication prescription details. The system has been previously used by both our team and other teams in Hong Kong.19,20 Data were obtained regarding consecutive patients diagnosed with dementia, excluding those who died or were discharged within 24 hours after the first diastolic/systolic BP measurement and those with less than 3 diastolic/systolic BP measurements (study baseline). Mortality data were obtained from the Hong Kong Death Registry, a population-based official government registry with the registered death records of all Hong Kong citizens. Data on the clinical characteristics, disease diagnosis, laboratory results (including complete blood counts, biochemical tests, and diastolic/systolic BP measurements), and medication prescription details were extracted. Dementia was identified with codes from the International Classification of Disease, Ninth Edition (ICD-9): 331.82, 290.0, 290.1, 290.11, 290.12, 290.13, 290.2, 290.21, 290.3, 290.4, 290.41, 290.42, 290.43, 290.8, 290.9, 294.2, 294.1, 294.11, 294.21, 46.1, 42.0, 294.29. The ICD-9 codes for past comorbidities and historical medication prescriptions are detailed in Supplementary Tables 1 and 2.

Statistical analysis and primary outcomes

The primary outcome was the development of dementia from the study baseline in a time-to-event analysis. Patients were followed up from their admission date until December 31, 2019. We extracted the baseline/latest/maximum/minimum values of diastolic and systolic BP and calculated the temporal variability measures of diastolic and systolic BP:21,22 1) mean, 2) median, 3) standard deviation (SD), 4) root mean square (RMS) by first squaring all blood pressure values then performing square root of the mean of the squares, 5) coefficient of variation (CV) by dividing the BP SD by the mean BP then multiplying by 100, and 6) a variability score (from 0 [low] to 100 [high]) defined as the number of changes in BP of 5 mmHg or more, that is, 100*(number of absolute BP change of each 2 successive measurements > 5)/number of measurements.

Clinical characteristics were summarized using descriptive statistics. Continuous variables were presented as median (95% confidence interval [CI] or interquartile range [IQR]) while categorical variables were presented as count (%). The Mann-Whitney U test was used to compare continuous variables. The χ2 test with Yates’ correction was used for 2 × 2 contingency data, and Pearson’s χ2 test was used for contingency data for variables with more than 2 categories. Univariate Cox regression models were conducted based on male and female subgroups, respectively. Significant univariate predictors of demographics, prior comorbidities, clinical and biochemical tests, medication prescriptions, and BP variabilities were used as input of a multivariate Cox analysis model, adjusted by traditional factors and intercepts. Hazard ratios (HRs) with corresponding 95% CI and P values were reported. All statistical tests were 2-tailed and considered significant if P value <.001. Data analyses were performed using RStudio software (Version: 1.1.456) and Python (Version: 3.6).

RESULTS

Gender-specific cohort clinical characteristics

This retrospective cohort study included 74 855 patients (male = 39.2%). Over the course of follow-up, 3550 patients (incidence rate: 4.74%, including 1287 males and 2263 females) developed dementia after a median follow-up of 112 months (IQR= [59.8–168], max = 242) after initial BP measurement (Supplementary Figure 1). The baseline demographic, biochemical, and clinical parameters are summarized in Table 1 in a gender-specific way. The number of patients in the male cohort was smaller in all age intervals except for [0, 10], [60, 70] and [70, 80] years old. Males more frequently tended to have past comorbidities of cardiovascular diseases (38.81% vs 35.40%, P value < .0001), respiratory diseases (52.55% vs 43.32%, P value < .0001), and renal complications (25.93% vs 16.27%, P value < .0001), but tended less frequently to have diabetes mellitus (13.33% vs 14.48%, P value = .0001) and hypertension (59.05% vs 61.15%, P value = .0043) than females.

Table 1.

Clinical characteristics of male and female patients of the cohort

Males (N = 29 333, event: 1287) Females (N = 45 522, event: 2263) P value
Median (IQR); Max; N or Count (%) Median (IQR); Max; N or Count (%)
Demographics
Age of first BP test, years 64.6(51.5–73.2); 99.9; n = 29 333 62.3(49.0–72.8); 101.4; n = 45 522 .1201
 [0,10] 28(0.09%) 13(0.02%) .0003***
  [10,20] 287(0.97%) 322(0.70%) .0001***
  [20,30] 1307(4.45%) 1658(3.64%) <.0001***
  [30,40] 1303(4.44%) 2644(5.80%) <.0001***
  [40,50] 3643(12.41%) 7642(16.78%) <.0001***
  [50,60] 5133(17.49%) 8736(19.19%) <.0001***
  [60,70] 7395(25.21%) 9804(21.53%) <.0001***
  [70,80] 7653(26.09%) 10 275(22.57%) <.0001***
  [80,90] 1848(6.30%) 3070(6.74%) .026*
 90+ 338(1.15%) 621(1.36%) .0142*

Past comorbidities
Cardiovascular 11 387(38.81%) 16 115(35.40%) <.0001***
Respiratory 15 417(52.55%) 19 721(43.32%) <.0001***
Renal 7608(25.93%) 7408(16.27%) <.0001***
Endocrine 1312(4.47%) 1971(4.32%) .382
Diabetes mellitus 3912(13.33%) 6595(14.48%) .0001***
Hypertension 17 322(59.05%) 27 840(61.15%) .0043**
Gastrointestinal 11 433(38.97%) 17 909(39.34%) .5139
Stroke 54(0.18%) 85(0.18%) .9956

Medications
ACEI 5229(17.82%) 6353(13.95%) <.0001***
ARB 136(0.46%) 277(0.60%) .0109*
Calcium channel blockers 8675(29.57%) 11 657(25.60%) <.0001***
Beta blockers 7457(25.42%) 11 316(24.85%) .1819
Diuretics for heart failure 1478(5.03%) 1955(4.29%) <.0001***
Diuretics for hypertension 3505(11.94%) 6184(13.58%) <.0001***
Nitrates 3498(11.92%) 4708(10.34%) <.0001***
Antihypertensive drugs 5141(17.52%) 2804(6.15%) <.0001***
Anti-Diabetic drugs 3305(11.26%) 4893(10.74%) .0484*
Statins and fibrates 3518(11.99%) 5718(12.56%) .0427*

Complete blood count tests
Mean corpuscular volume, fL 90.8(87.5–94.0); 132.3; n = 11 927 89.5(85.9–92.5); 133.0; n = 18 776 .8711
Basophil, x10^9/L 0.02(0.01–0.03); 0.6; n = 5397 0.02(0.01–0.02); 0.5; n = 7871 .8921
Eosinophil, x10^9/L 0.1(0.1–0.295); 9.25; n = 6390 0.1(0.1–0.2); 8.8; n = 9454 .9324
Lymphocyte, x10^9/L 1.6(1.1–2.15); 137.94; n = 6462 1.8(1.3–2.3); 85.28; n = 9572 .4514
Metamyelocyte, x10^9/L 0.15(0.1–0.4); 3.0; n = 71 0.16(0.08–0.38); 3.0; n = 73 .831
Monocyte, x10^9/L 0.5(0.4–0.7); 3.7; n = 6434 0.5(0.36–0.6); 6.09; n = 9530 .9419
Neutrophil, x10^9/L 4.8(3.61–6.9); 72.38; n = 6431 4.4(3.3–6.2); 40.5; n = 9513 .3612
White blood count, x10^9/L 7.5(6.13–9.36); 145.2; n = 11 978 7.1(5.8–8.8); 6100.0; n = 18 851 .1782
Mean cell haemoglobin, pg 30.9(29.6–32.0); 44.1; n = 11 927 30.4(29.0–31.5); 46.6; n = 18 775 .9056
Myelocyte, x10^9/L 0.18(0.105–0.45); 1.62; n = 67 0.17(0.09–0.38); 3.95; n = 76 .8561
Platelet, x10^9/L 223.0(184.0–268.0); 1020.0; n = 11 977 244.0(203.0–290.5); 1745.0; n = 18 846 <.0001***
Reticulocyte, x10^9/L 55.08(35.3–80.5); 324.0; n = 429 55.4(39.2–80.8); 460.0; n = 639 .9122
Red blood count, x10^12/L 4.61(4.2–4.99); 7.95; n = 11 912 4.27(3.95–4.58); 7.08; n = 18 763 .8967
Hematocrit, L/L 0.41(0.38–0.44); 0.61; n = 10 669 0.38(0.35–0.4); 0.561; n = 17 300 .7671

Biochemical tests
K/Potassium, mmol/L 4.2(3.9–4.5); 10.0; n = 17 388 4.2(3.81–4.5); 13.3; n = 25 177 .9176
Urate, mmol/L 0.42(0.343–0.5); 1.12; n = 5009 0.35(0.28–0.431); 1.395; n = 6173 .5651
Albumin, g/L 41.5(39.0–44.0); 58.0; n = 14 593 41.2(39.0–43.6); 58.0; n = 21 232 .9165
Na/Sodium, mmol/L 140.0(138.2–142.0); 166.09; n = 17 431 141.0(139.0–142.0); 181.0; n = 25 240 .9249
Urea, mmol/L 6.0(5.0–7.3); 60.9; n = 17 411 5.5(4.5–6.8); 53.4; n = 25 207 .0145*
Protein, g/L 73.1(70.0–77.0); 112.0; n = 14 523 74.0(71.0–78.0); 147.0; n = 21 127 .8934
Creatinine, umol/L 99.0(88.0–113.0); 1957.0; n = 17 525 77.0(68.0–89.0); 1274.0; n = 25 396 <.0001***
Alkaline Phosphatase, U/L 78.0(65.0–95.0); 3275.0; n = 12 528 78.0(63.0–96.0); 4280.0; n = 18 090 .9123
Aspartate Transaminase, U/L 22.0(18.0–30.0); 5110.0; n = 3642 21.0(17.0–27.0); 2148.0; n = 5229 .4564
Alanine Transaminase, U/L 22.0(16.0–33.0); 3909.0; n = 10 498 18.0(13.0–26.0); 1576.0; n = 15 831 .0023**
Bilirubin, umol/L 10.2(7.9–14.0); 608.0; n = 12 667 9.0(6.6–12.0); 669.0; n = 18 274 .1562
Diabetes mellitus and lipid tests
Triglyceride, mmol/mol 1.44(1.0–2.08); 25.77; n = 8635 1.41(1.01–2.04); 30.3; n = 12 504 .8926
LDL, mmol/mol 3.2(2.6–3.8); 7.92; n = 6359 3.3(2.7–3.9); 9.42; n = 8969 .6721
HDL, mmol/mol 1.18(1.01–1.39); 4.14; n = 6652 1.37(1.16–1.63); 3.29; n = 9338 .0104*
HbA1c, g/dL 13.6(11.5–14.8); 19.5; n = 10 501 12.5(11.1–13.4); 18.1; n = 16 601 .1551
Cholesterol, mmol/L 5.13(4.5–5.8); 13.03; n = 8698 5.4(4.7–6.09); 13.84; n = 12 597 .8723
Glucose, mmol/L 6.0(5.2–7.6); 72.5; n = 12 819 5.8(5.1–7.5); 54.3; n = 18 668 .751

Diastolic blood pressure measures
Number of tests 7(5–12); 31; n = 29 333 7(6–11); 35; n = 45 522 .9012
Baseline, mm Hg 74(69–85); 140.0; n = 29 333 79(65–89); 137.0; n = 45522 .1923
Latest, mm Hg 73(66–81); 140.0; n = 29 333 70(63–79); 144.0; n = 45522 .8723
Maximum, mm Hg 82(78–94); 150.0; n = 29 333 89(75–98); 144.0; n = 45522 .0145*
Minimal, mm Hg 65(57–73); 140.0; n = 29 333 61(54–70); 128.0; n = 45522 .1261
Mean, mm Hg 75(69–81); 140.0; n = 29 333 72(66.3–78); 128.0; n = 45 522 .7862
Median, mm Hg 75(69–81); 140.0; n = 29 333 72(66–78); 128.0; n = 45522 .4523
Variance 53.8(31.62–84.52); 882.0; n = 23 964 56.6 (32.9–85.2); 1152.0; n = 37 682 .5621
SD 7.3(5.6–9.2); 29.7; n = 23 964 7.5 (5.7–9.2); 33.9; n = 37 682 .8723
RMS 75.4(69.4–81.3); 140.0; n = 29 333 72.4(66.7–78.3); 128.0; n = 45 522 .6778
CV 0.09(0.07–0.13); 0.33; n = 23 964 0.099(0.07–0.12); 0.4; n = 37 682 .9561
Variability score 55.2(45.5–66.7); 94.12; n = 23 964 56.25(47.76–66.67); 95.46; n = 37 682 .6241

Systolic blood pressure measures
Number of tests 7(5–12); 33; n = 29333 8(5–11); 34; n = 45522 .8923
Baseline, mm Hg 131(123–152); 244.0; n = 29 333 139(120–159); 251.0; n = 45 522 .0132*
Latest, mm Hg 133(121–146); 237.0; n = 29 333 135(120–146); 261.0; n = 45 522 .2173
Maximum, mm Hg 156(140–170); 249.0; n = 29 333 157(138–173); 274.0; n = 45 522 .7671
Minimal, mm Hg 117(106–130); 237.0; n = 29 333 114(104–128); 242.0; n = 45 522 .8921
Mean, mm Hg 135.96(126.5–145.5); 237.0; n = 29 333 135.4(125–145); 242.0; n = 45 522 .9016
Median, mm Hg 135.5(126–145.5); 237.0; n = 29 333 135(124–145); 242.0; n = 45 522 .9156
Variance 165.7(94.4–272.2); 4133.3; n = 23 964 167.7(97.0–271.4); 5618.0; n = 37 682 .8723
SD 12.9(9.7–16.5); 64.3; n = 23 964 12.95(9.9–16.5); 74.95; n = 37 682 .8912
RMS 136.5(127.0–146.1); 237.0; n = 29 333 136.0(125.3–145.6); 242.0; n = 45 522 .9015
CV 0.09(0.07–0.1); 0.3; n = 23 964 0.09(0.07–0.11); 0.35; n = 37 682 .9156
Variability score 69.2(55.7–77.8); 96.7; n = 23 964 70.0(57.1–77.8); 96.97; n = 37 682 .8954
*

P ≤ .05,

**

P ≤ .01,

***

P ≤ .001

In addition, males were more frequently prescribed for angiotensin-converting enzyme inhibitor (ACEI) (17.82% vs 13.95%, P value < .0001), calcium channel blockers (29.57% vs 25.60%, P value < .0001), diuretics for heart failure (5.03% vs 4.29%, P value < .0001), nitrates (11.92% vs 10.34%, P value < .0001), antihypertensive drugs (17.52% vs 6.15%, P value < .0001), and anti-diabetic drugs (11.26% vs 10.74%, P value = .0484), but were less frequently prescribed angiotensin receptor blocker (ARB) (0.46% vs 0.60%, P value = .0109), diuretics for hypertension (11.94% vs 13.58%, P value < .0001), and statins and fibrates (11.99% vs 12.56%, P value = .0427).

Males had lower platelet levels (median: 223 x10^9/L, IQR: 184.0–268, max: 1020 x10^9/L vs 244 x10^9/L, IQR: 203.0–290.5, max: 1745 x10^9/L, P value < .0001), high density lipoprotein (HDL) (median: 1.18 mmol/mol, IQR: 1.01–1.39, max: 4.14 vs median: 1.37 mmol/mol, IQR: 1.16–1.63, max: 3.29 mmol/mol, P value = .0104), maximum of diastolic BP (median: 82 mm Hg, IQR: 78–94, max: 150 mm Hg vs median: 89 mm Hg, IQR: 75–98, max: 144, P value = .0145), and baseline value of systolic BP (median: 131 mm Hg, IQR: 123–152, max: 244 vs median: 139 mm Hg, IQR: 120–159, max: 251 mm Hg, P value = .0132). However, male patients had higher urea levels (6 mmol/L, IQR: 5.0–7.3, max: 60.9 mmol/L vs 5.5 mmol/L, IQR: 4.5–6.8, max: 53.4 mmol/L, P value = .0145), creatinine (median: 99 umol/L, IQR: 88–113, max: 1957 vs 77 umol/L, IQR: 68.0–89, max: 1274 umol/L, P value < .0001), alanine transaminase (median: 22 U/L, IQR: 16.0–33, max: 3909 U/L vs 18 U/L, IQR: 13–26, max: 1576 U/L, P value = .0023),

Endocrine (median age= 73.9, IQR = [63.4–82.2]) and gastrointestinal (median age= 74.5, IQR = [63.6, 82.7]) comorbidities, in addition to diabetes mellitus (median age= 75.6, IQR = [66.4, 83.3]), were the 3 earliest comorbidities that occurred prior to dementia, with no significant gender differences (Supplementary Table 3). The incidence rates of female patients were significantly higher than those of male patients in the following age groups of [40, 50–90], and 90+ (Figure 1). The breakdown of incidences with respect to gender and age are shown in Supplementary Table 4, and the baseline characteristics of the dementia subgroup are shown in Supplementary Table 5. Kaplan-Meier curves for the nondementia patients are shown in Figure 2, while those for all-cause mortality are detailed in Supplementary Figure 2.

Figure 1.

Figure 1.

Age-specific incidence of dementia diseases between male patients and female patients.

Figure 2.

Figure 2.

Survival curves of dementia outcome in the overall cohort, male cohort, and female cohort.

Significant risk predictors of dementia and associations of BP measurements with time-to-dementia

Univariate predictors for incident dementia are summarized in Table 2, while those for mortality among those with dementia are detailed in the Supplementary Table 6. With identified significant univariate predictors as inputs, the following parameters were found to be significant multivariate predictors (Table 3): (1) age of first BP measurement: 40–50 (HR: 1.05, 95% CI: [1.01, 1.26], P < .001), 50–60 (HR: 1.17, 95% CI: [1.06, 1.45], P < .001), 60–70 (HR: 1.43, 95% CI: [1.20, 1.93], P: .001), 70–80 (HR: 1.45, 95% CI: [1.36, 1.93], P < .0001), 80–90 (HR: 1.47, 95% CI: [1.09, 3.06], P < .0001); (2) comorbidities: cardiovascular (HR: 1.10, 95% CI: [1.08, 1.55], P < .0001), respiratory (HR: 1.56, 95% CI: [1.05, 2.31], P: .028), hypertension (HR: 1.21, 95% CI: [1.09, 1.46], P < .0001), gastrointestinal (HR: 1.66, 95% CI: [1.23, 2.23], P: .001); (3) medication: calcium channel blockers (HR: 1.15, 95% CI: [1.04, 1.57], P < .0001), diuretics for hypertension (HR: 1.01, 95% CI: [1.01, 1.44], P < .0001); (4) laboratory parameters: eosinophil count (HR: 0.28, 95% CI: [0.10, 0.77], P: .014), neutrophil count (HR: 1.03, 95% CI: [1.08, 1.47], P < .0001), urate (HR: 0.14, 95% CI: [0.04, 0.53], P: .004), aspartate transaminase (HR: 0.99, 95% CI: [0.97, 1.00], P: .017); (5) diastolic BP: baseline (HR: 1.02, 95% CI: [1.01, 1.21], P < .0001), mean (HR: 1.25, 95% CI: [1.14, 1.57], P < .0001), variance (HR: 1.40, 95% CI: [1.04, 1.51], P < .0001), CV (HR: 1.31, 95% CI: [1.02, 1.65], P < .0001), variability score (HR: 1.22, 95% CI: [1.09, 2.11], P < .0001); (6) systolic BP: baseline (HR: 1.02, 95% CI: [1.01, 1.21] P < .0001), maximum (HR: 1.40, 95% CI: [1.18, 1.42], P < .0001), mean (HR: 1.27, 95% CI: [1.17, 1.61], P < .0001), SD (HR: 1.18, 95% CI: [1.01, 1.69], P < .0001), variability score (HR: 1.43, 95% CI: [1.18, 1.91], P < .0001). Nonlinear relationships between systolic or diastolic BP measurements and the time-to-dementia are shown in Supplementary Figures 3 and 4, respectively.

Table 2.

Univariate predictors of dementia diseases for all patients, males, and females

All patients P value Males P value Females P value
HR [95% CI] HR [95% CI] HR [95% CI]
Demographics
Male gender 0.88[0.83,0.95] .0005***
Age, years
 [30,40] 0.02[0.01, 0.05] <.0001*** 0.016[0.002, 0.12] <.0001*** 0.014[0.004, 0.06] <.0001***
  [40,50] 0.07[0.05,0.091] <.0001*** 0.08[0.05, 0.14] <.0001*** 0.06[0.04, 0.09] <.0001***
  [50,60] 0.199[0.17, 0.23] <.0001*** 0.27[0.21, 0.34] <.0001*** 0.17[0.1, 0.2] <.0001***
  [60,70] 1.2[1.1, 1.8] <.0001*** 1.15[1.02, 1.3] <.0001*** 1.4[1.1, 2.1] <.0001***
  [70,80] 2.6[2.4, 2.8] <.0001*** 2.3[2.02, 2.5] <.0001*** 3.3[2.6, 4.2] <.0001***
  [80,90] 3.3[3.1, 3.6] <.0001*** 2.9[2.5, 3.4] <.0001*** 4.6[3.2, 5.0] <.0001***
 90+ 2.1[1.7, 2.59] <.0001*** 1.7[1.1, 2.5] .0169* 3.1[1.8, 4.1] <.0001***

Comorbidities
Cardiovascular 1.9[1.8,2.1] <.0001*** 1.5[1.4,1.7] <.0001*** 2.2[2.0,2.4] <.0001***
Respiratory 3.8[3.5,4.1] <.0001*** 4.2[3.7,4.9] <.0001*** 3.8[3.4,4.1] <.0001***
Renal 1.6[1.5,1.7] <.0001*** 1.4[1.2,1.6] <.0001*** 1.8[1.6,2.0] <.0001***
Endocrine 0.7[0.6,0.8] .0002*** 0.6[0.5, 0.9] .005** 0.7[0.6,0.9] .0116*
Diabetes mellitus 1.3[1.2,1.4] <.0001*** 1.1[0.9,1.2] .452 1.4[1.2,1.5] <.0001***
Hypertension 1.7[1.6,1.9] <.0001*** 1.6[1.4,1.8] <.0001*** 1.8[1.7,2.2] <.0001***
Gastrointestinal 1.6[1.5,1.8] <.0001*** 1.5[1.4,1.7] <.0001*** 1.7[1.6,1.9] <.0001***
Stroke 1.9[1.8,2.0] <.0001*** 1.6[1.4,1.8] <.0001*** 2.2[2.0,2.3] <.0001***

Medications
ACEI 1.3[1.2,1.5] <.0001*** 1.1[0.9,1.2] .362 1.6[1.4,1.7] <.0001***
ARB 1.1[0.7,1.7] .687 1.3[0.7,2.7] .427 1.0[0.6,1.7] .934
Calcium channel blockers 1.4[1.3,1.5] <.0001*** 1.2[1.1,1.3] .004** 1.6[1.4,1.7] <.0001***
Beta blockers 1.1[1.0, 1.2] .0036** 1.0[0.9,1.13] 0.94 1.2[1.1,1.3] .0002***
Diuretics for heart failure 1.9[1.7, 2.1] <.0001*** 1.7[1.4,2.1] <.0001*** 2.0[1.7,2.3] <.0001***
Diuretics for hypertension 1.3[1.2,1.4] <.0001*** 1.08[0.9,1.3] .335 1.4[1.2,1.5] <.0001***
Nitrates 1.7[1.5,1.8] <.0001*** 1.4[1.2,1.6] <.0001*** 1.9[1.7,2.1] <.0001***
Antihypertensive drugs 1.6[1.5,1.7] <.0001*** 1.8[1.6,2.0] <.0001*** 1.6[1.4,1.8] <.0001***
Antidiabetic drugs 1.2[1.1,1.4] <.0001*** 1.1[0.9,1.3] .377 1.3[1.2,1.5] <.0001***
Statins and fibrates 1.1[0.99,1.2] .0516. 0.9[0.8,1.1] .294 1.2[1.1,1.4] .002**

Complete blood count tests
Mean corpuscular volume, fL 1.02[1.01,1.03] <.0001*** 1.01[0.99,1.02] .0983. 1.03[1.02,1.03] <.0001***
Basophil, x10^9/L 0.4[0.1,1.6] .186 0.26[0.02,2.82] .266 0.49[0.07,3.44] .47
Eosinophil, x10^9/L 0.5[0.4,0.8] .0008*** 0.5[0.3,0.8] .0092** 0.6[0.4,1.01] .0546.
Lymphocyte, x10^9/L 0.77[0.7,0.8] <.0001*** 0.8[0.7,0.9] .0001*** 0.75[0.7,0.8] <.0001***
Metamyelocyte, x10^9/L 0.9[0.2,4.1] .919 2.2[0.5,10.1] .324 0.3[0.01,7.74] .474
Monocyte, x10^9/L 1.5[1.2,1.7] <.0001*** 1.1[0.8,1.5] .545 1.7[1.5,2.1] <.0001***
Neutrophil, x10^9/L 1.04[1.03,1.06] <.0001*** 1.02[1.0,1.04] .0828. 1.06[1.04,1.1] <.0001***
White blood count, x10^9/L 1.0[0.999,1.001] .904 1.006[0.99,1.03] .559 1.00[0.99,1.001] .929
Mean cell haemoglobin, pg 1.04[1.02,1.06] <.0001*** 1.02[0.99,1.05] .11 1.05[1.03,1.07] <.0001***
Myelocyte, x10^9/L 0.6[0.1,4.7] 0.662 0.001[0.001,12.5] .564 0.8[0.2,3.7] .731
Platelet, x10^9/L 0.998[0.997,0.999] <.0001*** 0.998[0.997,0.999] .0014** 0.998[0.997,0.999] .0008***
Reticulocyte, x10^9/L 0.998[0.99,1.004] .522 0.99[0.98,1.001] .094. 1.002[0.99,1.01] .585
Red blood count, x10^12/L 0.65[0.6,0.69] <.0001*** 0.67[0.6,0.74] <.0001*** 0.62[0.56,0.68] <.0001***
Hematocrit, L/L 0.02[0.01,0.04] <.0001*** 0.007[0.002,0.03] <.0001*** 0.03[0.007,0.1] <.0001***

Biochemical tests
K/Potassium, mmol/L 0.78[0.73,0.85] <.0001*** 0.78[0.68,0.89] .0002*** 0.8[0.72,0.88] <.0001***
Urate, mmol/L 0.4[0.2,0.8] .007** 0.14[0.04,0.44] .0009*** 1.1[0.46,2.52] .859
Albumin, g/L 0.94[0.93,0.95] <.0001*** 0.94[0.92,0.95] <.0001*** 0.94[0.93,0.96] <.0001***
Na/Sodium, mmol/L 0.986[0.97,0.998] .0194* 0.97[0.95,0.99] .0015** 0.99[0.98,1.01] .389
Urea, mmol/L 1.05[1.04,1.06] <.0001*** 1.02[1.01,1.04] .0122* 1.06[1.05,1.08] <.0001***
Protein, g/L 0.97[0.97,0.98] <.0001*** 0.97[0.963,0.985] <.0001*** 0.97[0.96,0.98] <.0001***
Creatinine, umol/L 1.001[1.001,1.002] <.0001*** 1.001[0.9995,1.002] .273 1.003[1.002,1.003] <.0001***
Alkaline phosphatase, U/L 1.001[1,1.001] .0183* 1[0.9985,1.001] .964 1.001[1,1.001] .003**
Aspartate transaminase, U/L 0.999[0.99,1.001] .852 0.999[0.997,1.001] .53 1.001[0.999,1.002] .389
Alanine transaminase, U/L 0.987[0.98,0.99] <.0001*** 0.98[0.97,0.98] <.0001*** 0.99[0.99,1.00] .0012**
Bilirubin, umol/L 1.001[0.997,1.01] .735 1.001[0.99,1.01] .774 1.002[0.997,1.01] .474
Diabetes mellitus and lipid tests
Triglyceride, mmol/mol 0.95[0.9,1.004] .0687. 0.82[0.73,0.92] .0007*** 1.012[0.95,1.07] .685
LDL, mmol/mol 1.04[0.96,1.13] .322 0.9[0.8,1.05] .185 1.11[1.01,1.23] .039*
HDL, mmol/mol 1.2[1.02,1.5] .0336* 1.7[1.2,2.3] .002** 0.95[0.74,1.2] .661
HbA1c, mmol/mol 0.99[0.98,0.99] .002** 0.99[0.97,0.999] .0371* 0.98[0.97,0.998] .03*
Cholesterol, mmol/L 1.02[0.96,1.07] .58 0.92[0.84,1.01] .0702. 1.05[0.99,1.12] .126
Glucose, mmol/L 1.03[1.02,1.05] <.0001*** 1.02[1.002,1.05] .0322* 1.04[1.03,1.06] <.0001***

Diastolic blood pressure measurements
Number of tests 1.07[0.13,1.23] .8511 0.65[0.23,1.42] .0611 1.03[0.54,1.22] .1801
Baseline, mm Hg 1.15[1.11,2.34] <.0001*** 1.43[1.01,1.76] <.0001*** 1.24[1.01,1.93] <.0001***
Latest, mm Hg 1.03[1.01,1.12] <.0001*** 1.09[1.02,1.13] .0045** 0.99[0.8,0.99] .234
Maximum, mm Hg 1.21[1.1,1.83] <.0001*** 0.98[0.90,0.99] .2834 1.34[1.03,2.12] <.0001***
Minimal, mm Hg 0.98[0.94,0.983] .6523 0.97[0.92,0.98] .0823 0.98[0.91,0.99] .831
Mean, mm Hg 1.31[1.11,1.85] <.0001*** 1.13[1.03,1.45] <.0001*** 1.43[1.01,1.76] <.0001***
Median, mm Hg 1.53[1.24,3.13] <.0001*** 1.23[1.11,2.1] <.0001*** 1.13[1.01,1.4] <.0001***
Variance 1.003[1.003,1.003] <.0001*** 1.002[1.001,1.003] <.0001*** 1.003[1.003,1.004] <.0001***
SD 1.074[1.062,1.085] <.0001*** 1.052[1.034,1.071] <.0001*** 1.086[1.072,1.1] <.0001***
RMS 0.97[0.92,0.98] .035* 0.96[0.93,0.99] .2341 0.99[0.98,0.991] .8734
CV 58.7[69.7,194.2] <.0001*** 11.5[5.16,19.6] <.0001*** 13.8[4.1,39.3] <.0001***
Variability score 1.008[1.006,1.01] <.0001*** 14.5[6.13,17.9] <.0001*** 13.9[4.4,32.1] <.0001***

Systolic blood pressure measurements
Number of tests 0.87[0.13,1.23] .2315 0.95[0.63,1.02] .1956 0.73[0.34,1.51] .8523
Baseline, mm Hg 1.011[1.01,1.012] <.0001*** 1.006[1.003,1.009] <.0001*** 1.014[1.012,1.015] <.0001***
Latest, mm Hg 1.008[1.006,1.01] <.0001*** 1.003[1.001,1.006] .0157* 1.01[1.008,1.012] <.0001***
Maximum, mm Hg 1.011[1.01,1.013] <.0001*** 1.008[1.006,1.01] <.0001*** 1.013[1.011,1.015] <.0001***
Minimal, mm Hg 1.005[1.003,1.007] <.0001*** 1.001[0.9978,1.004] .615 1.008[1.005,1.01] <.0001***
Mean, mm Hg 1.016[1.014,1.018] <.0001*** 1.009[1.006,1.013] <.0001*** 1.02[1.018,1.022] <.0001***
Median, mm Hg 1.016[1.014,1.018] <.0001*** 1.009[1.005,1.012] <.0001*** 1.02[1.017,1.022] <.0001***
Variance 1.001[1.001,1.001] <.0001*** 1.001[1.001,1.001] <.0001*** 1.001[1.001,1.001] <.0001***
SD 1.052[1.047,1.057] <.0001*** 1.042[1.034,1.051] <.0001*** 1.057[1.051,1.063] <.0001***
RMS 1.017[1.015,1.019] <0.0001*** 1.01[1.006,1.013] <.0001*** 1.021[1.018,1.023] <.0001***
CV 44.4[18.6,105.9] <.0001*** 10.5[2.5,44.8] <.0001*** 10.3[3.5,30.8] <.0001***
Variability score 1.009[1.007,1.012] <.0001*** 1.009[1.005,1.013] <.0001*** 1.01[1.007,1.012] <.0001***
*

P ≤ .05,

**

P ≤ 0.01,

***

P ≤ .001

Table 3.

Multivariate predictors of dementia diseases for all patients, males, and females

All patients P value Males P value Females P value
HR [95% CI] HR [95% CI] HR [95% CI]
Demographics
Male gender 0.88[0.62, 1.27] .5051
Age
  [30,40] 1.05[1.01,1.37] .0035**
  [40,50] 1.05[1.01, 1.26] .0003 *** 1.03[1.01,1.12] <.0001***
  [50,60] 1.17[1.06, 1.45] .0004 *** 1.09[1.04,1.20] <.0001***
  [60,70] 1.43[1.20, 1.93] .0011 ** 1.23[1.04,1.31] <.0001*** 1.42[1.24,1.72] <.0001***
  [70,80] 1.45[1.36, 1.93] <.0001*** 1.28[1.11,1.81] <.0001***
  [80,90] 1.47[1.09, 3.06] <.0001*** 1.18[1.01,1.52] <.0001*** 1.27[1.06,2.11] <.0001***
 90+ 1.21[0.41, 3.57] .7316 1.67[1.15,3.28] <.0001***

Comorbidities
Cardiovascular 1.10[1.08, 1.55] <.0001*** 1.03[0.32, 3.32] .9629 1.07[1.04,1.37] <.0001***
Respiratory 1.56[1.05, 2.31] .0275 * 1.71[0.48, 6.10] .4095 1.59[1.22,2.07] .0006***
Renal 0.84[0.60, 1.18] .3239 2.08[0.76, 5.69] .156 0.79[0.62,1.02] .0688.
Endocrine 1.26[1.09, 1.71] .0089 **
Diabetes mellitus 1.27[0.86, 1.87] .2260 1.48[1.13,1.94] .0049**
Hypertension 1.21[1.09, 1.46] <.0001*** 1.05[1.03, 4.76] <.0001*** 1.24[1.15,1.61] <.0001***
Gastrointestinal 1.66[1.23, 2.23] .0009 *** 2.80[0.99, 7.89] .0513. 1.36[1.10,1.67] .0043**
Stroke 0.95[0.69, 1.31] .7431 1.35[0.44, 4.14] .6017 1.13[1.02,1.43] <.0001***

Medications
ACEI 0.93[0.66, 1.30] .6554 0.84[0.65,1.07] .1618
Calcium channel blockers 1.15[1.04, 1.57] <.0001*** 0.86[0.30, 2.45] .7769 1.21[1.05,1.41] <.0001***
Beta blockers 1.06[0.77, 1.46] .7140 1.04[0.83,1.31] .7449
Diuretics for heart failure 0.84[0.54, 1.33] .4671 0.77[0.13, 4.66] .7772 1.23[1.05,1.61] <.0001***
Diuretics for hypertension 1.01[1.01, 1.44] <.0001*** 1.18[1.02,1.55] <0.0001***
Nitrates 0.75[0.51, 1.11] .1499 0.74[0.24, 2.30] .6071 1.24[1.17,1.45] <0.0001***
Antihypertensive drugs 1.06[0.74, 1.50] .7566 2.20[0.76, 6.32] .1439 0.92[0.66,1.29] .6351
Antidiabetic drugs 0.94[0.64, 1.39] .7661 0.96[0.73,1.27] .7745
Statins and fibrates 0.86[0.66,1.13] .2883

Complete blood count tests
Mean corpuscular volume, fL 0.98[0.89, 1.07] .5890 1.21[1.04,1.67] <.0001***
Eosinophil, x10^9/L 1.28[1.10, 1.77] .0138 * 0.61[0.06, 6.42] .6803
Lymphocyte, x10^9/L 1.03[0.96, 1.11] .3769 1.28[0.59, 2.78] .5312 0.99[0.87,1.12] .8179
Monocyte, x10^9/L 0.66[0.36, 1.21] .1808 1.11[1.07,1.59] <.0001***
Neutrophil, x10^9/L 1.03[1.08, 1.47] <.0001*** 1.22[1.09,1.53] <.0001***
Mean cell haemoglobin, pg 0.99[0.82, 1.20] .9334 0.96[0.84,1.10] .5891
Platelet, x10^9/L 1.00[1.00, 1.00] .3739 1.00[0.99, 1.01] .5299 1.00[1.00,1.00] .6566
Red blood count, x10^12/L 0.53[0.18, 1.57] .2488 0.82[0.22, 3.05] .7632 0.66[0.25,1.69] .3824
Hematocrit, L/L 35.71[0.00,191.00] .5199

Biochemical tests
K/Potassium, mmol/L 0.84[0.65, 1.08] .1703 0.58[0.55, 1.24] .2612 0.96[0.79,1.15] .6261
Urate, mmol/L 1.14[1.04, 1.53] .0037 ** 0.60[0.01, 35.17] .8035
Albumin, g/L 0.99[0.95, 1.03] .6981 0.91[0.77, 1.07] .2497 1.03[0.99,1.06] .1187
Urea, mmol/L 0.98[0.91, 1.05] .4914 1.17[1.03,1.72] <.0001***
Na/Sodium, mmol/L 1.10[0.94, 1.30] .2317
Protein, g/L 1.03[1.00, 1.06] .0543. 1.13[1.01, 1.26] .0623. 0.99[0.97,1.01] .4858
Creatinine, umol/L 1.00[0.99, 1.01] .9421 1.00[1.00,1.01] <.0001***
Aspartate transaminase, U/L 0.99[0.97, 1.00] .0166* 0.96[0.91, 1.01] .0833. 1.00[1.00,1.00] .7237
Alanine transaminase, U/L 1.00[1.00,1.00] .9126
Diabetes mellitus and lipid tests
Triglyceride, mmol/mol 1.22[0.66, 2.26] .5324
HDL, mmol/mol 2.73[0.78, 9.52] .1161
Glucose, mmol/L 1.02[0.99, 1.06] .2476 1.01[0.97,1.04] .6423

Diastolic blood pressure measurements
Baseline, mm Hg 1.14[1.07, 1.52] <.0001*** 1.15[1.08, 1.43] <.0001*** 1.21[1.02,1.21] <.0001***
Latest, mm Hg 1.00[0.98, 1.02] .1834 1.06[0.99, 1.12] .0816.
Maximum, mm Hg 1.01[0.96, 1.05] .8364 1.19[1.06,1.92] <.0001***
Mean, mm Hg 1.25[1.14, 1.57] <.0001*** 0.87[0.65, 1.17] .366 1.32[1.12,1.79] <.0001***
Median, mm Hg 1.04[0.96, 1.13] .3311 1.23[1.01, 1.32] <0.0001*** 1.02[0.96,1.08] 0.4816
Variance 1.4[1.04, 1.51] <.0001*** 1.3[1.07, 1.94] <.0001*** 1.11[1.01,1.32] <.0001***
SD 1.29[0.97, 1.72] .0800 1.52[0.37, 6.16] .5582 0.96[0.79,1.17] .6825
CV 1.31[1.02, 1.65] <.0001*** 0.00[0.00, 12.00] .5327
Variability score 1.22[1.09, 2.11] <.0001*** 1.19[1.05, 1.83] <.0001*** 1.22[1.12,2.41] <.0001***

Systolic blood pressure measurements
Baseline, mm Hg 1.02[1.01, 1.21] <.0001*** 1.03[0.99, 1.07] .1789 1.31[1.09,2.34] <.0001***
Latest, mm Hg 1.01[1.00, 1.02] .1029 1.00[1.00,1.01] .3566
Maximum, mm Hg 1.40[1.18, 1.42] <.0001*** 1.00[0.94, 1.06] .9429 1.02[1.01,1.03] <.0001***
Minimal, mm Hg 1.02[0.99, 1.05] .2281 1.02[1.00,1.05] .0322*
Mean, mm Hg 1.27[1.17, 1.61] <.0001*** 0.00[0.00, 10.33] .1196 0.71[0.27,1.84] .4817
Median, mm Hg 1.00[0.95, 1.04] .9105 0.94[0.82, 1.08] .399 1.03[1.00,1.07] <.0001***
Variance 1.00[0.99, 1.00] .1617 0.98[0.94, 1.02] .251 1.15[1.01,1.42] <.0001***
SD 1.18[1.01, 1.69] <.0001*** 1.86[0.48, 7.22] .3698 0.98[0.85,1.12] .7148
RMS 3.69[0.98, 13.81] .0529. 1.31[1.11,3.35] <.0001***
CV 0.00[0.00, 12.00] .2279
Variability score 1.43[1.18, 1.91] <.0001*** 1.03[0.98, 1.08] .2033 1.32[1.09,2.11] <.0001***
*

≤ .05,

**

P ≤ .01,

***

P ≤ .001

Gender-specific clinical risk score to predict incident dementia

Based on the findings of multivariate Cox regression and cutoff values of significant predictors, excluding predictive post-hoc medication variables, we developed a clinical risk score for early prediction of dementia in male and female patients in Table 4. For both genders, the following common variables were used: age, prior hypertension, baseline, median, variance, and variability score of diastolic blood pressure and systolic blood pressure. For female patients, the following additional variables were included: prior cardiovascular, respiratory and gastrointestinal diseases, hypertension and stroke, and laboratory examinations.

Table 4.

Clinical risk scores for early prediction of dementia diseases in male (left) and female (right) patients

Clinical Risk Score for Males
Clinical Risk Score for Females
Risk factors Score Cutoff Risk factors Score Cutoff
Age Age of first BP
  [60,70] 1.23 Present [30,40] 1.05 Present
  [80,90] 1.18 Present [40,50] 1.03 Present
Prior hypertension 1.05 Present [50,60] 1.09 Present
High diastolic BP baseline, mm Hg 1.15 75.5 mm Hg [60,70] 1.42 Present
High diastolic BP median, mm Hg 1.23 73.2 mm Hg [70,80] 1.28 Present
High diastolic BP variance 1.3 67.4 [80,90] 1.27 Present
High diastolic BP variability score 1.19 59.2 90+ 1.67 Present
High systolic BP median, mm Hg 1.01 141.5 mm Hg Prior cardiovascular 1.07 Present
High systolic BP variance 1.01 235.4 Prior respiratory 1.59 Present
Prior diabetes mellitus 1.48 Present
Prior hypertension 1.24 Present
Prior gastrointestinal 1.36 Present
Prior stroke 1.13 Present
High mean corpuscular volume, fL 1.21 92.4 fL
High monocyte, x10^9/L 1.11 0.53 x10^9/L
High neutrophil, x10^9/L 1.22 5.3 x10^9/L
High urea, mmol/L 1.17 6.5 mmol/L
High creatinine, umol/L 1.00 102.4 umol/L
High diastolic BP baseline, mm Hg 1.21 77.2 mm Hg
High diastolic BP maximum, mm Hg 1.19 79.1 mm Hg
High diastolic BP mean, mm Hg 1.32 75.5 mm Hg
High diastolic BP variance 1.11 69.8
High diastolic BP variability score 1.22 68.5
High systolic BP baseline, mm Hg 1.31 145.2 mm Hg
High systolic BP maximum, mm Hg 1.01 169.3 mm Hg
High systolic BP median, mm Hg 1.03 149.5 mm Hg
High systolic BP variance 1.15 245.1
High systolic BP RMS 1.31 149.23
High systolic BP variability score 1.32 0.13

Furthermore, the details of the score for male and female patients with/without dementia are summarized in Supplementary Table 7. Comparing within the gender subgroups, both male (median: 4.22, IQR: 2.36,5.56, max: 9.17 vs median: 3.5, IQR: 2.31,4.77, max: 5.47, P value < .0001) and female (median: 11.58, IQR: 8.82,14.7, max: 26.56 vs median: 8.96, IQR: 6.05,12.22, max: 15.81, P value < .0001) with dementia had a higher score than their nondemented counterparts. The discrimination performance of the scores is shown in Figure 3. For females, the score had a cutoff value of 11.13 and is also able to significantly predict the initial presentation of dementia (HR: 1.13, 95% CI: 1.12–1.24, P value < .0001), and the dichotomized score system shows much more predictive ability (HR: 26.56, 95% CI: 14.44–32.86, P value < .0001).

Figure 3.

Figure 3.

Discrimination performance of clinical risk scores for male (top) and female (bottom) patients.

The performance of the scores were compared in Supplementary Table 8 to predict the initial presentation of dementia. For males, the score had a cutoff of 4.48 and can significantly predict initial presentation of dementia (HR: 1.08, 95% CI: 1.05–1.11, P value < .0001), while the dichotomized score system demonstrated even more predictive strength (HR: 12.83, IQR: 11.15–14.33, P value < .0001).

To explore further a simpler score that can be used at baseline (rather than incorporating subsequent results which would not be available at that juncture) (Table 5). In this simplified score, only baseline blood pressure was included. However, the performance metrics (Table 6) showed that there was a reduction in the c-statistic by 0.088 and 0.096 for male and female patients, respectively, indicating the importance of incorporating successive measurements for blood pressure on follow-up to improve risk stratification.

Table 5.

Simplified clinical risk scores for early prediction of dementia diseases in male (left) and female (right) patients after excluding BP variability measures

Clinical Risk Score for Males
Clinical Risk Score for Females
Risk factors Score Cutoff Risk factors Score Cutoff
Age Age of first BP
  [60,70] 1.33 Present [30,40] 1.04 Present
  [80,90] 1.28 Present [40,50] 1.07 Present
Prior hypertension 1.05 Present [50,60] 1.06 Present
Lower alanine transaminase, U/L 0.96 23.2 U/L [60,70] 1.42 Present
Hematocrit, L/L 0.23 0.45 L/L [70,80] 1.31 Present
High diastolic BP baseline, mm Hg 1.21 75.4mm Hg [80,90] 1.25 Present
90+ 2.15 Present
Prior cardiovascular 1.06 Present
Prior respiratory 1.61 Present
Prior diabetes mellitus 1.52 Present
Prior hypertension 1.43 Present
Prior stroke 1.82 Present
High mean corpuscular volume, fL 1.23 94.1 fL
High monocyte, x10^9/L 1.19 0.53 x10^9/L
High neutrophil, x10^9/L 1.24 5.2 x10^9/L
High urea, mmol/L 1.21 6.6 mmol/L
High diastolic BP baseline, mm Hg 1.32 77.5 mm Hg
High systolic BP baseline, mm Hg 1.28 143.2 mm Hg

Table 6.

Five-fold cross validation for the comparisons between gender-specific clinical risk scores with BP variabilities and simplified clinical risk scores without BP variabilities for early prediction of dementia diseases

Systems for males Cutoff C-index
Scoring system considering BP variabilities 4.48 0.9082
Simplified scoring system without BP variabilities 4.32 0.8201
Systems for females C-index Cutoff
Scoring system considering BP variabilities 11.12 0.9123
Simplified scoring system without BP variabilities 17.23 0.8161

DISCUSSION

The main findings of this study include the following:

  1. A combination of clinical, biochemical and systolic/diastolic BP value and variability can be used to predict the onset of dementia;

  2. There are nonlinear associations between diastolic/systolic BP value and variability and the time to dementia manifestation;

  3. A gender-specific, easy-to-use clinical risk score for early prediction of dementia has been constructed and found to be of high predictive strength;

  4. The constructed gender-specific clinical risk scores outperformed the simplified scores that excluded BP variability, demonstrating the importance of the latter in dementia development.

The nonlinear associations between diastolic and systolic BP value and variability reported by the present study support findings from existing studies.7,23–25 There are several hypotheses proposed for the underlying mechanisms of the nonlinear relationship observed. Previous studies propose that the apolipoprotein E4 allele upholds a modulatory role in the effects of BP on cognitive function.26,27 Furthermore, patients with chronic hypertension have been shown to have increased Tau phosphorylation under BP reduction, suggesting that chronic hypertension may increase one’s susceptibility to dementia particularly under extreme BP changes.28,29 Moreover, in a recent study by Walker et al, a pattern of midlife hypertension and late-life hypotension was reported to precede cognitive decline, which suggests a potential early neurological change underlying both the BPV and the cognitive decline. The age-dependent BP change and its associated dementia risk can also be attributed to the nonlinear relationship between BP value and the risk of dementia.

Although it remains controversial whether females have a higher risk for dementia, the presence of gender-specific risk factors has been continuously explored.30,31 First of all, the menopausal transition in middle-aged females was reported to induce a hypometabolic state and can increase brain beta-amyloid deposition thus increasing dementia risk, which is supported by the drastic increase in the HR among the peri- and postmenopausal age groups.32,33 The loss of cardioprotective effect by estrogen among postmenopausal females and resulting BP instability, as reflected by the predictiveness of BPV among female patients, may also underlie their higher risk for vascular dementia.34 In addition, it has been reported that a selective survival of males less susceptible to cardiovascular conditions after mid-life can explain the lower dementia risk among males, which coincides with the presence of cardiovascular comorbidities as a female-only predictor in the present cohort.35 While screening assessments, such as The Montreal Cognitive Assessment, are available for identifying patients with cognitive impairment, carrying out such tests is very time-consuming, and simple clinical scores that can be used to predict longer term dementia development, not just early cognitive impairment, would be helpful for clinicians to manage the patients accordingly.

The plethora of factors underlying the gender differences in dementia risk demonstrates the importance of a gender-specific risk-stratification score system to increase the chances of early disease detection and optimize patient care.

Limitations

Several limitations should be noted for the present study. Given its retrospective and observational nature, this study is prone to selection bias and susceptible to errors due to undercoding and coding errors. Moreover, due to local data availability, only visit-to-visit BP records could be obtained for the analysis of long-term BPV, whereas short-term BPV data were not available. Other important risk factors for dementia, such as the family history of dementia, apolipoprotein E4 allele status, body mass index, and smoking status were not routinely coded into structured data. We have indirectly accounted for the influence of cardiovascular risk factors by examining the prognostic value of cardiovascular comorbidities. In addition, the age distribution for male and female dementia patients were different. For example, the age distribution for female dementia patients was wider. This could potentially explain the need for additional BP measurements for the model development. These scores will be validated in the future when additional data become available.

CONCLUSION

Gender-specific clinical risk scores incorporating BP variability can accurately predict incident dementia and can be applied clinically for early disease detection and optimized patient management.

FUNDING

This work was supported by the National Natural Science Foundation of China (NSFC): 71972164; Health and Medical Research Fund of the Food and Health Bureau of Hong Kong: 16171991; Innovation and Technology Fund of Innovation and Technology Commission of Hong Kong: MHP/081/19; National Key Research and Development Program of China, Ministry of Science and Technology of China: 2019YFE0198600.

AUTHOR CONTRIBUTIONS

JZ, SL: data analysis, data interpretation, statistical analysis, manuscript drafting, critical revision of manuscript

WTW, KBW, KSKL, TTLL, AKCW, TL, CC: project planning, data acquisition, data interpretation, critical revision of manuscript

BMYC: study supervision, data interpretation, statistical analysis, critical revision of manuscript

QZ, GT: study conception, study supervision, project planning, data interpretation, statistical analysis, manuscript drafting, critical revision of manuscript

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.

DATA AVAILABILITY STATEMENT

The dataset for this study can be obtained by contacting the corresponding author(s) upon reasonable request for research purposes.

CONFLICT OF INTEREST STATEMENT

None declared.

Supplementary Material

ocab173_Supplementary_Appendix

REFERENCES

  • 1. Abell JG, Kivimaki M, Dugravot A, et al.  Association between systolic blood pressure and dementia in the Whitehall II cohort study: role of age, duration, and threshold used to define hypertension. Eur Heart J  2018; 39 (33): 3119–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Gregson J, Qizilbash N, Iwagami M, et al.  Blood pressure and risk of dementia and its subtypes: a historical cohort study with long-term follow-up in 2.6 million people. Eur J Neurol  2019; 26 (12): 1479–86. [DOI] [PubMed] [Google Scholar]
  • 3. Ding J, Davis-Plourde KL, Sedaghat S, et al.  Antihypertensive medications and risk for incident dementia and Alzheimer's disease: a meta-analysis of individual participant data from prospective cohort studies. Lancet Neurol  2020; 19 (1): 61–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Oishi E, Ohara T, Sakata S, et al.  Day-to-day blood pressure variability and risk of dementia in a general Japanese elderly population: the Hisayama study. Circulation  2017; 136 (6): 516–25. [published Online First: Epub Date]|. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Nagai M, Hoshide S, Ishikawa J, Shimada K, Kario K.  Visit-to-visit blood pressure variations: new independent determinants for cognitive function in the elderly at high risk of cardiovascular disease. J Hypertens  2012; 30 (8): 1556–63. [DOI] [PubMed] [Google Scholar]
  • 6. Yano Y, Ning H, Allen N, et al.  Long-term blood pressure variability throughout young adulthood and cognitive function in midlife: the Coronary Artery Risk Development in Young Adults (CARDIA) study. Hypertension  2014; 64 (5): 983–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Jain S, Kuriakose D, Edelstein I, et al.  Right atrial phasic function in heart failure with preserved and reduced ejection fraction. JACC Cardiovasc Imaging  2019; 12 (8 Pt 1): 1460–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. de Heus RAA, Olde Rikkert MGM, Tully PJ, Lawlor BA, Claassen J, Group NS; NILVAD Study Group. Blood pressure variability and progression of clinical Alzheimer disease. Hypertension  2019; 74 (5): 1172–80. [DOI] [PubMed] [Google Scholar]
  • 9. Kim S, Kim MJ, Kim S, et al.  Gender differences in risk factors for transition from mild cognitive impairment to Alzheimer's disease: a CREDOS study. Compr Psychiatry  2015; 62: 114–22. [DOI] [PubMed] [Google Scholar]
  • 10. Choi J, Kwon LN, Lim H, Chun HW.  Gender-based analysis of risk factors for dementia using senior cohort. Int J Environ Res Public Health  2020; 17 (19): 7274. doi: 10.3390/ijerph17197274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Paul KC, Debes F, Eliasen E, Weihe P, Petersen MS.  Incidence, gender influence, and neuropsychological predictors of all cause dementia in the Faroe Islands-the Faroese Septuagenarian cohort. Aging Clin Exp Res  2021; 33 (1): 105–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Altmann A, Tian L, Henderson VW, Greicius MD, Alzheimer's Disease Neuroimaging Initiative Investigators. Sex modifies the APOE-related risk of developing Alzheimer disease. Ann Neurol  2014; 75 (4): 563–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Podcasy JL, Epperson CN.  Considering sex and gender in Alzheimer disease and other dementias. Dialogues Clin Neurosci  2016; 18 (4): 437–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Hall JR, Wiechmann AR, Johnson LA, et al.  Biomarkers of vascular risk, systemic inflammation, and microvascular pathology and neuropsychiatric symptoms in Alzheimer's disease. J Alzheimers Dis  2013; 35 (2): 363–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Mosconi L, Berti V, Quinn C, et al.  Correction: Perimenopause and emergence of an Alzheimer's bioenergetic phenotype in brain and periphery. PLoS One  2018; 13 (2): e0193314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Chiu PY, Teng PR, Wei CY, Wang CW, Tsai CT.  Gender difference in the association and presentation of visual hallucinations in dementia with Lewy bodies: a cross-sectional study. Int J Geriatr Psychiatry  2018; 33 (1): 193–9. [DOI] [PubMed] [Google Scholar]
  • 17. Lee J, Lee KJ, Kim H.  Gender differences in behavioral and psychological symptoms of patients with Alzheimer's disease. Asian J Psychiatr  2017; 26: 124–8. [DOI] [PubMed] [Google Scholar]
  • 18. Ohta Y, Nomura E, Hatanaka N, et al.  Female dominant association of sarcopenia and physical frailty in mild cognitive impairment and Alzheimer's disease. J Clin Neurosci  2019; 70: 96–101. [DOI] [PubMed] [Google Scholar]
  • 19. Li CK, Xu Z, Ho J, et al.  Association of NPAC score with survival after acute myocardial infarction. Atherosclerosis  2020; 301: 30–6. [DOI] [PubMed] [Google Scholar]
  • 20. Ju C, Lai RWC, Li KHC, et al.  Comparative cardiovascular risk in users versus non-users of xanthine oxidase inhibitors and febuxostat versus allopurinol users. Rheumatology (Oxford)  2019; 59 (9): 2340–9. [DOI] [PubMed] [Google Scholar]
  • 21. Zhou J, Li H, Chang C, et al.  The association between blood pressure variability and hip or vertebral fracture risk: a population-based study. Bone  2021; 150: 116015. [DOI] [PubMed] [Google Scholar]
  • 22. Zhou J, Lee S, Wong WT, et al.  Gender- and age-specific associations of visit-to-visit blood pressure variability with anxiety. Front Cardiovasc Med  2021; 8: 650852. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Wang ZT, Xu W, Wang HF, et al.  Blood pressure and the risk of dementia: a dose-response meta-analysis of prospective studies. CNR  2019; 15 (4): 345–58. [DOI] [PubMed] [Google Scholar]
  • 24. Rajan KB, Barnes LL, Wilson RS, Weuve J, McAninch EA, Evans DA.  Blood pressure and risk of incident Alzheimer's disease dementia by antihypertensive medications and APOE epsilon4 allele. Ann Neurol  2018; 83 (5): 935–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Walker KA, Sharrett AR, Wu A, et al.  Association of midlife to late-life blood pressure patterns with incident dementia. JAMA  2019; 322 (6): 535–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Haan MN, Shemanski L, Jagust WJ, Manolio TA, Kuller L.  The role of APOE epsilon4 in modulating effects of other risk factors for cognitive decline in elderly persons. JAMA  1999; 282 (1): 40–6. [DOI] [PubMed] [Google Scholar]
  • 27. Hofman A, Ott A, Breteler MM, et al.  Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimer's disease in the Rotterdam Study. Lancet  1997; 349 (9046): 151–4. [DOI] [PubMed] [Google Scholar]
  • 28. Glodzik L, Rusinek H, Pirraglia E, et al.  Blood pressure decrease correlates with tau pathology and memory decline in hypertensive elderly. Neurobiol Aging  2014; 35 (1): 64–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Power MC, Tchetgen EJ, Sparrow D, Schwartz J, Weisskopf MG.  Blood pressure and cognition: factors that may account for their inconsistent association. Epidemiology  2013; 24 (6): 886–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Andersen K, Launer LJ, Dewey ME, et al.  Gender differences in the incidence of AD and vascular dementia: the EURODEM studies. EURODEM Incidence Research Group. Neurology  1999; 53 (9): 1992–7. [DOI] [PubMed] [Google Scholar]
  • 31. Roberts RO, Geda YE, Knopman DS, et al.  The incidence of MCI differs by subtype and is higher in men: the Mayo Clinic study of aging. Neurology  2012; 78 (5): 342–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Mosconi L, Berti V, Quinn C, et al.  Sex differences in Alzheimer risk: brain imaging of endocrine vs chronologic aging. Neurology  2017; 89 (13): 1382–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Brinton RD, Yao J, Yin F, Mack WJ, Cadenas E.  Perimenopause as a neurological transition state. Nat Rev Endocrinol  2015; 11 (7): 393–405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Dufouil C, Seshadri S, Chene G.  Cardiovascular risk profile in women and dementia. J Alzheimers Dis  2014; 42 (Suppl 4): S353–63. [DOI] [PubMed] [Google Scholar]
  • 35. Chene G, Beiser A, Au R, et al.  Gender and incidence of dementia in the Framingham Heart Study from mid-adult life. Alzheimers Dement  2015; 11 (3): 310–20. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ocab173_Supplementary_Appendix

Data Availability Statement

The dataset for this study can be obtained by contacting the corresponding author(s) upon reasonable request for research purposes.


Articles from Journal of the American Medical Informatics Association : JAMIA are provided here courtesy of Oxford University Press

RESOURCES