Abstract
Purpose:
To determine whether covert infarcts and white matter hyperintensities (WMHs) are associated with increased risk of incident hip fracture.
Methods:
A prospective cohort of 3,373 community-dwelling adults aged ≥65 years enrolled in the Cardiovascular Health Study with a brain MRI (1992-93) was analyzed. Covert infarcts were categorized by number of infarcts and largest infarct size. WMH burden was assessed by radiologists and graded qualitatively from zero (no WMHs) to nine (extensive).
Results:
Participants had 465 incident hip fractures during a mean follow-up of 12.8 years. The demographic-adjusted hazard of incident hip fracture was 32% higher among participants with ≥ 1 covert infarct compared to those without infarcts (hazard ratio [HR] 1.32; 95%CI, 1.08-1.62). The hazard of incident hip fracture was similar after further adjustment for medications and medical history (HR=1.34;95%CI,1.08-1.65), but attenuated following additional adjustment for functional status, frailty, and falls (HR=1.25;95%CI,0.99-1.57). Fully-adjusted hazard of incident hip fracture per increase in infarct number was 1.10 (95%CI,0.98-1.23); risk in individuals whose largest infarct was ≥20mm versus 3 to <20mm were similar. Compared with WMH grades 0-1, the demographic-adjusted hazard of hip fracture was 1.34 (95%CI,1.09-1.66) and 1.83 (95%CI,1.37-2.46), respectively, for WMH grades 2-3 and 4-9. The hazard was similar following adjustment for medications and medical history (grade 2-3: HR=1.32;95%CI,1.05-1.64; grade 4-9: HR=1.69;95%CI,1.23-2.30), but attenuated following additional adjustment for functional status, frailty, and falls (grade 2-3: HR=1.24;95%CI,0.98-1.56; grade 4-9: HR=1.34;95%CI,0.95-1.90).
Conclusion:
Older, community-dwelling adults with covert infarcts or WMHs may be at increased risk of hip fracture.
Keywords: Osteoporosis, hip fracture, white matter hyperintensity, brain microvascular disease
MINI ABSTRACT
Covert brain infarcts and white matter hyperintensities (WMHs), incidental markers of brain microvascular disease commonly seen on brain MRIs in older adults, have been associated with falls and lower bone mineral density. We found covert infarcts and WMHs may also be associated with an increased risk of future hip fracture.
INTRODUCTION
Use of magnetic resonance imaging (MRI) for brain imaging is rising1. One consequence has been the incidental identification of markers of brain microvascular disease such as covert (i.e., clinically-silent) infarcts and white matter hyperintensities (WMHs)2,3. WMHs are considered a marker of incomplete infarction that suggest chronically reduced blood flow4. Pathologically, WMHs represent areas of gliosis, demyelination and axonal loss4,5. Up to a quarter of older adults have at least one covert infarct6; WMHs of some degree may be present in 80% or more of older adults3. While the prevalence of covert infarcts and WMHs increase with age, they are considered markers of small vessel disease, not benign markers of aging4.
Covert infarcts and WMHs are associated with incident symptomatic stroke and death3. WMHs are associated with gait impairment5–8 and an increased risk of falls5–11. Several studies have suggested a dose-response relationship, with a larger number of infarcts9 and greater density of WMHs more strongly associated with falls5–9. Covert infarcts12,13 and WMHs12 have also been associated with low bone mineral density (BMD). However, the association of covert infarcts and WMHs with incident hip fracture is less clear14.
A study of 820 older adults reported that diffuse WMHs, compared to no WMHs, were associated with over twice the hazard of incident hip fracture in participants aged 65 to 79 years, but not those aged 80 years or older15. Though a potentially clinically-important finding, analyses were post-hoc, there was wide uncertainty around the WMH risk estimates, and few participants were aged 80 years and older (n=161). This prior study also had no information on covert infarcts. To verify and potentially expand on these prior findings, we used brain MRI data from the Cardiovascular Health Study (CHS) to examine the association of covert infarcts and WMHs with risk of incident hip fracture.
METHODS
Participants
CHS is a prospective, longitudinal cohort study designed to track the onset and course of cardiovascular disease and stroke in older adults. A total of 5,201 non-wheelchair dependent community-dwelling adults aged 65 years or older were enrolled from four U.S. sites between 1989 and 1990. An additional 687 black participants were enrolled at the 1992-1993 study visit. The study protocol was approved by the institutional review boards at all participating centers. Written informed consent was obtained from all study participants. Further details of the CHS study design and recruitment are described elsewhere16,17.
All study participants were invited to undergo brain MRI scanning in years 1992-1994. The hip fracture analytic cohort included all enrollees who completed a brain MRI in years 1992-1994; did not have a confirmed, clinically-recognized history of stroke or transient ischemic attack (TIA) prior to the brain MRI; and had non-missing data regarding occurrence of incident hip fractures (n=3373; eFigure 1). Participants with previous hip fracture were not excluded. The BMD analytic cohort included the subset of participants who also completed dual x-ray absorptiometry (DXA) testing in 1994-1995 (n=1177).
Brain MRI
Brain MRI imaging was performed using a General Electric or Picker 1.5 Tesla MRI scanner at 3 of the 4 CHS sites and a 0.35 Tesla Toshiba MRI scanner at the fourth site18,19. MRI images were read by neuroradiologists trained per CHS protocol at a central reading center. The neuroradiologists had no information on the history of the individual. Each image was read by two neuroradiologists for consensus in identifying infarcts and white matter abnormalities.
Covert Infarcts and White Matter Hyperintensities
The primary predictors were covert infarcts and WMHs. Covert infarcts were defined as an area of abnormal signal intensity of ≥3mm in a vascular distribution and without mass effect22. Lesions <3mm were excluded from analyses due to poor pixel resolution and poor inter- and intra-reader reliability23. Infarcts were categorized by number (0, 1, 2, 3, 4, or ≥5) and size of the largest infarct (no infarcts, 3 to <20mm, or ≥20mm; categories chosen based on available data)22. Infarcts were analyzed as a categorical variable (none vs. ≥1 infarct) for the primary analysis and per increase in infarct number and by size of largest infarct in secondary analyses.
WMHs were graded based on total volume of periventricular and subcortical white matter signal abnormalities on spin-density weighted axial images. The volume of hyperintensities was compared to sets of reference scans with changes ranging from grade 0 (no white matter changes) to grade 9 (extensive and confluent)18,19. As reported in a prior study, the inter-reader reliability for agreement within one grade was 92.1%; the intra-reader agreement within one grade was 94.5% 25. WMH burden (graded 0-9) was analyzed as a categorical variable (0-1, 2-3, and 4-9); results of analyses of WMH as a continuous variable are not reported as WMH grade is not linear.
Incident hip fractures
Our primary outcome was incident hip fracture. Hip fractures were identified from hospital discharge International Classification of Diseases, Ninth Revision, codes of (ICD-9) 820.xx, excluding hip fractures due to motor vehicle accidents (E810.xx-E825.xx). Hospitalizations were ascertained from participant self-report every six months; hospital records were checked against Medicare claims data to identify any hospitalizations not reported by participants26.
Other measures
Covariates were conceptually organized according to a hypothesized causal pathway between covert infarcts, WMHs, and hip fractures (Figure 1)15. Covariate data were obtained from standardized questionnaires administered at the 1992-93 study visit, a comprehensive visit that corresponded to the first year brain MRIs were obtained.
Figure 1.
Hypothesized causal pathway between covert infarcts, white matter hyperintensities, and hip fractures.
Factors potentially associated with covert infarcts, WMHs, and/or hip fractures, but not in the causal pathway, included self-report of age at menopause (women only), smoking status (current, former, never), and current alcohol intake (<1 drink/week, 1-7 drinks/week, >7 drinks/week). Body mass index (kg/m2) was calculated based on weight and height obtained using standardized methods at the study visit. Medical history (diabetes, hypertension, coronary heart disease, and congestive heart failure) and medication use (loop and thiazide diuretics, thyroid medications, oral estrogen, oral corticosteroids, osteoporosis medications, and psychotropic medications defined as use of an antidepressant, antipsychotic, or benzodiazepine) were determined based a combination of self-report, clinical data collected at the study visit, and medical record review, as previously described27.
Factors potentially in the causal pathway between covert infarcts, WMHs, and hip fractures included modified Mini-Mental State Exam (3MS) score, limitations in activities of daily living (ADLs; self-report of limitations in at least one of walking around the home, getting out of bed, eating, dressing, bathing, using the toilet), limitations in instrumental activities of daily living (IADLs; self-report of limitations in at least one of heavy housework, light housework, shopping, preparing meals, paying bills, using the phone), tandem stand (yes/no able to stand in tandem during study exam), self-report of a fall in the last year, and vision problems (self-report of having difficulty watching TV and/or recognizing someone across the room even with glasses). Frailty was defined according to the Fried et al. CHS frailty criteria (weight loss, weakness, exhaustion, slowness, and inactivity)28. Participants were classified as non-frail if they met 0 criteria, prefrail if they met 1-2 criteria, and frail if they met ≥ 3 criteria.
Total hip areal BMD was measured at the 1994–1995 study visit at two of the four CHS sites, Pittsburgh, Pennsylvania and Sacramento, California, using QDR-2000 densitometers (Hologic, Bedford, MA). The coefficient of variation for the total hip BMD was <0.75%27.
Statistical Analysis
Baseline characteristics are shown as mean +/− standard deviation (SD) for continuous variables and as count (%) for categorical variables. Crude hip fracture incidence rates are presented for each brain MRI predictor, with follow-up time calculated as time from MRI until hip fracture, death, loss to follow-up, or until June 30, 2014 (date of last claims adjudication). Cox regression was used to evaluate the association of covert infarcts and WMHs with incident hip fractures. We sequentially added covariates not in our hypothesized causal pathway, then covariates in our hypothesized causal pathway, to the models as follows: Model (M)0 – demographics; M1 – M0 plus medical history, social history, and medications; M2 – M1 plus physical function, cognitive function, and falls; M3 – M2 plus BMD (BMD analytic cohort). In a secondary analysis, we used stratification to assess for effect modification by sex and age.15 We also evaluated the significance of separate cross product terms (i.e., interaction terms) including age and sex.
RESULTS
The analytic cohort consisted of 3,373 adults with a mean age of 75 years (standard deviation [SD] 5.1) (Table 1). Participants with ≥1 covert infarct and/or WMHs grade ≥2 were, on average, older; more likely to have CHD, CHF, and hypertension; more likely to report IADL impairment and ≥1 fall in the past year; less likely to be able to stand in tandem; had a lower score on the 3MS and lower total hip BMD; and were more likely to meet criteria for frailty than participants without these markers of brain microvascular disease. Participants with WMHs grade ≥2 were also more likely than participants with WMHs grade 0-1 to report ADL impairment and vision problems.
Table 1.
Participant characteristics as a function of covert infarcts and WMHs.
| Overall (n=3373a) | Covert infarcta | White matter hyperintensitiesb | |||
|---|---|---|---|---|---|
| No (n=2433) | Yes (n=940) | No (n=1194) | Yes (n=2150) | ||
| Age (years), mean (SD) | 75.0 (5.1) | 74.6 (4.9) | 76.0 (5.4)* | 73.3 (4.3) | 76.0 (5.3)* |
| Male, n (%) | 1364 (40.4) | 947 (40.0) | 390 (41.5) | 513 (43.0) | 843 (39.2)* |
| Black race, n (%) | 521 (15.4) | 376 (15.5) | 145 (15.4) | 190 (15.9) | 325 (15.1) |
| Education, > 12 years, n (%) | 1568 (46.6) | 1146 (47.2) | 422 (45.1) | 596 (49.9) | 961 (44.9)* |
| Age at menopause (years),c mean (SD) | 47.2 (6.6) | 47.3 (6.5) | 46.7 (6.9) | 47.1 (6.6) | 47.3 (6.6) |
| Current smoker, n (%) | 309 (9.4) | 207 (8.7) | 102 (11.1)* | 108 (9.2) | 201 (9.6) |
| Current alcohol use, > 7 drinks per week, n (%) | 375 (11.2) | 284 (11.7) | 91 (9.7)* | 151 (12.7) | 224 (10.5)* |
| BMI (kg/m2), mean (SD) | 26.7 (4.5) | 26.7 (4.5) | 26.4 (4.4) | 26.9 (4.6) | 26.5 (4.4)* |
| Diabetes | 451 (14.0) | 325 (14.0) | 126 (14.0) | 148 (13.0) | 299 (14.6) |
| Coronary Heart Disease | 636 (18.9) | 416 (17.1) | 220 (23.4)* | 203 (17) | 430 (20.0)* |
| Congestive Heart Failure | 135 (4.0) | 82 (3.4) | 53 (5.6)* | 33 (2.8) | 100 (4.7)* |
| Hypertension | 1819 (54.4) | 1257 (52.1) | 562 (60.3)* | 550 (46.5) | 1250 (58.7)* |
| Medications, current use, n (%) | |||||
| Loop diuretic | 117 (3.5) | 73 (3.0) | 45 (4.8)* | 26 (2.2) | 91 (4.2)* |
| Thiazide diuretic | 384 (11.4) | 279 (11.5) | 105 (11.2) | 120 (10.1) | 263 (12.2) |
| Estrogen | 37 (1.1) | 23 (0.9) | 14 (1.5) | 12 (1) | 25 (1.2) |
| Thyroid medication | 333 (9.9) | 235 (9.7) | 98 (10.4) | 105 (8.8) | 227 (10.6) |
| Psychotropic medication | 366 (10.9) | 252 (10.4) | 114 (12.1) | 109 (9.1) | 253 (11.8)* |
| Oral corticosteroid | 80 (2.4) | 58 (2.4) | 22 (2.3) | 25 (2.1) | 53 (2.5) |
| Osteoporosis medication | 88 (2.6) | 58 (2.4) | 30 (3.2) | 31 (2.6) | 57 (2.7) |
| Frail, n (%) | 259 (8.3) | 157 (6.9) | 102 (11.8)* | 59 (5.2) | 197 (10)* |
| 3MS score, mean (SD) | 91.0 (8.3) | 91.5 (7.8) | 89.6 (9.4)* | 92.4 (6.7) | 90.23 (9.0)* |
| ADL impairment, n (%) | 282 (8.4) | 189 (7.8) | 93 (9.9) | 81 (6.8) | 200 (9.3)* |
| IADL impairment, n (%) | 781 (23.3) | 524 (21.6) | 257 (27.4)* | 233 (19.6) | 541 (25.3)* |
| Able to stand in tandem, n (%) | 2731 (83.3) | 406 (85.7) | 704 (77.1)* | 1028 (87.9) | 1683 (80.8)* |
| Vision problems, n(%)d | 72 (2.1) | 51 (2.1) | 21 (2.2) | 17 (1.4) | 54 (2.5)* |
| At least one fall all in last year, n (%) | 506 (15.1) | 340 (14.1) | 166 (17.8)* | 152 (12.8) | 351 (16.4)* |
| Total hip BMD (g/cm2),e mean (SD) | 1.03 (0.15) | 1.04 (0.16) | 1.01 (0.15)* | 1.04 (0.16) | 1.02 (0.15)* |
3MS, Modified Mini-Mental State Exam; ADL, activity of daily living; BMI, body mass index; BMD, bone mineral density; IADL, instrumental activity of daily living; WMH, white matter hyperintensities
p < 0.05 comparing participants with and without covert infarcts or WMHs
≥ 1 covert infarct
White matter hyperintensities grade ≥ 2
Female participants only (n=1875)
Self-report of having difficulty watching TV and/or recognizing someone across the room even with glasses
In subset of 1177 participants who had bone mineral density measured at the 1994-1995 study visit
Participants were followed for up to 25 years with a mean of 12.8 years (SD 6.2; median 12.7) for a total of 43,326 person-years of follow-up. Participants experienced 465 incident hip fractures for an overall unadjusted incidence rate (IR) of 1.21 per 100 person-years (95% confidence interval [95% CI], 0.98-1.28). The IR of hip fracture for participants with ≥1 covert infarct was 1.41 per 100 person-years (95%CI, 1.08-1.83), compared to 1.03 (95%CI, 0.87-1.20) for participants without covert infarcts. The IR of hip fracture was 0.83 (95%CI, 0.66-1.03), 1.23 (95%CI, 1.02-1.50), and 1.81 (95%CI, 1.35-2.42), per 100 person-years, for participants with WMH grades 0-1, 2-3, and 4-9, respectively. Kaplan-Meier plots for time without hip fracture by presence or absence of covert infarcts and WMH grade are presented in Figure 2.
Figure 2.
Kaplan-Meier plots for time without hip fracture by (A) presence or absence of covert infarct(s) and (B) white matter hyperintensity grade.
Association of covert infarcts with incident hip fracture
After adjustment for age, sex, race, education, and study enrollment site, the hazard of incident hip fracture was 32% higher for participants with ≥1 covert infarct compared to participants without covert infarcts (hazard ratio [HR] 1.32; 95%CI, 1.08-1.62) (Table 2). This difference remained after adjusting for medical history, social history, and medication use (HR 1.34; 95%CI 1.08-1.65), but was attenuated following additional adjustment for functional status, frailty, and fall history (HR 1.25; 95%CI, 0.99-1.57). In the subset of 1,177 participants who completed a DXA in 1994-1995, the hazard of incident hip fracture was 56% higher for participants with covert infarcts compared to participants without covert infarcts (HR 1.56; 95%CI, 1.04-2.34) following additional adjustment for BMD (Table 3). Interaction terms for interaction by age and sex were non-significant at the 10% level.
Table 2.
Association of covert infarcts and WMHs with incident hip fracture (n=3,373).
| Model 1 HR (95% CI) |
Model 2 HR (95% CI) |
Model 3 HR (95% CI) |
|
|---|---|---|---|
| Any infarct1 | 1.32 (1.08, 1.62) | 1.34 (1.08, 1.65) | 1.25 (0.99, 1.57) |
| No infarcts | Ref | Ref | Ref |
| Per increase of 1 infarct | 1.13 (1.02, 1.25) | 1.12 (1.01, 1.25) | 1.10 (0.98, 1.23) |
| Size of largest infarct1 | |||
| ≥ 20mm | 1.35 (0.82, 2.23) | 1.35 (0.82, 2.28) | 1.22 (0.69, 2.16) |
| 3 to <20mm | 1.32 (1.07, 1.63) | 1.33 (1.07, 1.66) | 1.25 (0.99, 1.59) |
| No infarcts | Ref | Ref | Ref |
| WMH grade2 | |||
| 4-9 | 1.83 (1.37, 2.46) | 1.69 (1.23, 2.30) | 1.34 (0.95, 1.90) |
| 2-3 | 1.34 (1.09, 1.66) | 1.32 (1.05, 1.64) | 1.24 (0.98, 1.56) |
| 0-1 | Ref | Ref | Ref |
95%CI, 95% Confidence Interval; 3MS, Modified Mini-Mental State Exam; ADL, activity of daily living; BMD, Bone Mineral Density; HR, Hazard Ratio; IADL, instrumental activity of daily living; WMH, white matter hyperintensities
Model 1 adjusted for age, sex, race, education, and study enrollment site
Model 2 adjusted for covariates in Model 1 and smoking status, alcohol use, BMI, diabetes, coronary heart disease, congestive heart failure, hypertension, and medications (loop diuretics, thiazides, thyroid medications, oral estrogen, psychotropic medications, osteoporosis medications)
Model 3 adjusted for covariates in Model 2 and ADL impairment, iADL impairment, frailty, tandem stand, vision, 3MS score, and falls
Reference group is participants without covert infarcts
Reference group is participants with WMHs grade 0-1
Table 3.
Association of brain infarcts and WMHs with incident hip fracture in the subset of participants with data on BMD (n=1,177).
| Model 1 HR (95% CI) |
Model 2 HR (95% CI) |
Model 3 HR (95% CI) |
Model 4 HR (95% CI) |
|
|---|---|---|---|---|
| Any infarct | 1.30 (0.92, 1.84) | 1.34 (0.93, 1.93) | 1.38 (0.93, 2.06) | 1.56 (1.04, 2.34) |
| No infarcts | Ref | Ref | Ref | Ref |
| Per increase of 1 infarct | 1.19 (1.0, 1.41) | 1.19 (0.99, 1.42) | 1.20 (1.0, 1.45) | 1.28 (1.06, 1.55) |
| Size of largest infarct | ||||
| ≥ 20mm | 1.47 (0.46, 4.67) | 1.31 (0.41, 4.24) | 1.36 (0.40, 4.60) | 1.82 (0.54, 6.14) |
| 3 to <20mm | 1.29 (0.92, 1.84) | 1.34 (0.92, 1.95) | 1.39 (0.92, 2.09) | 1.54 (1.02, 2.33) |
| No infarcts | Ref | Ref | Ref | Ref |
| WMH grade | ||||
| 4-9 | 2.27 (1.37, 3.78) | 2.19 (1.26, 3.79) | 1.94 (1.08, 3.50) | 1.70 (0.94, 3.09) |
| 2-3 | 1.22 (0.84, 1.76) | 1.38 (0.93, 2.03) | 1.22 (0.80, 1.85) | 1.19 (0.78, 1.81) |
| 0-1 | Ref | Ref | Ref | Ref |
95%CI, 95% Confidence Interval; 3MS, Modified Mini-Mental State Exam; ADL, activity of daily living; BMD, Bone Mineral Density; HR, Hazard Ratio; iADL, instrumental activity of daily living; WMH, white matter hyperintensities
Model 1 adjusted for age, sex, race, education, and study enrollment site
Model 2 adjusted for covariates in Model 1 and smoking status, alcohol use, BMI, diabetes, coronary heart disease, congestive heart failure, hypertension, and medications (loop diuretics, thiazides, thyroid medications, oral estrogen, psychotropic medications, osteoporosis medications)
Model 3 adjusted for covariates in Model 2 and ADL impairment, iADL impairment, frailty, tandem stand, vision, 3MS score, and falls
Model 4 adjusted for covariates in Model 3 and BMD
Following stratification by sex, the adjusted hazard of incident hip fracture for participants with ≥1 covert infarct compared to participants without covert infarcts was similar for men (1.37; 95%CI, 0.89-2.11) and for women (1.22; 95%CI, 0.93-1.61) (eTable 1). Following additional adjustment for BMD in the BMD cohort, the hazard of hip fracture for women with covert infarcts was 92% higher than the hazard for women without covert infarcts (HR 1.92; 95%CI, 1.18-3.11), but appeared similar for men with and without covert infarcts (HR 1.08; 95%CI, 0.44-2.62). In individuals younger than age 80, the adjusted hazard of hip fracture was 35% greater for those with covert infarcts compared to those without covert infarcts (HR 1.35; 95%CI 1.05-1.76). In contrast, among participants aged 80 years or older, risk of hip fracture appeared similar between those with and without covert infarcts (HR 1.02; 95%CI, 0.62-1.67). This pattern persisted following additional adjustment for BMD in the BMD cohort.
In secondary analyses looking at number of infarcts and size of largest infarct, the adjusted hazard of incident hip fracture was 10% higher per additional covert infarct (highest category ≥5 infarcts; HR 1.10; 95%CI, 0.98-1.23) in the overall cohort (Table 2), and 20% higher per additional covert infarct (HR 1.20; 0.98-1.45) in the BMD cohort. Following additional adjustment for BMD in the BMD cohort, the hazard of hip fracture was 28% higher (HR 1.28; 95%CI 1.06-1.55) per additional infarct. Size of largest infarct was not a significant predictor (Table 3).
Association of WMHs with incidence hip fracture
After adjustment for demographics, compared to participants with WMHs grades 0-1, the hazard of incident hip fracture was 34% higher (HR 1.34; 95%CI, 1.09-1.66) and 83% higher (HR 1.83; 95%CI, 1.37-2.46) for participants with WMH grades 2-3 and 4-9, respectively (Table 2). Results remained significantly different following adjustment for medical history, social history, and medication use, but were attenuated by further adjustment for physical function, cognitive function, and fall history (HR 2-3 vs 0-1, 1.24; 95%CI 0.98-1.56; HR 4-9 vs 0-1, 1.34; 95%CI, 0.95-1.90). Adjustment for BMD in the BMD cohort further attenuated the association (Table 3). Interaction for age ≥80 was significant in the model adjusted for demographics only; otherwise, interaction terms for interaction by age and sex were non-significant at the 10% level.
Adjusting for demographics, WMH grade was significantly associated with hazard of hip fracture for women (WMH grade 4-9 vs 0-1: HR 1.98; 95%CI 1.40-2.79), but no association was found for men (HR 1.53; 95%CI, 0.88, 2.67) (eTable 2). The results were attenuated and no longer significant following additional adjustment. In contrast to the results for covert infarcts, WMHs were associated with hazard of hip fracture for the oldest participants (aged ≥80), but there was no significant association in younger participants.
DISCUSSION
In this cohort of community-dwelling older adults, presence of covert infarcts and greater WMH burden, incidental markers of brain microvascular disease, were associated with a greater risk of incident hip fracture. Consistent with our proposed causal pathway, the association between covert infarcts, WMHs, and incident hip fracture was not attenuated by adjustment for medical history, social history, and medication use, but was attenuated by further adjustment for physical function, cognitive function, and fall history. The association between covert infarcts, WMHs, and incident hip fracture remained nearly significant (p<0.1) following complete adjustment. Our results suggest there may be additional shared risk factors and causal pathways beyond the comprehensive risk factors and clinical characteristics considered in this analysis.
To our knowledge, our analysis is the first to evaluate the association of covert infarcts with risk of incident hip fracture. We are aware of one prior study that evaluated the association of WMHs with risk of incident hip fracture15. Consistent with our findings, this longitudinal study of 820 adults aged 65 years and older in Italy found double the incidence rate of hip fracture for participants with diffuse WMHs compared to participants without WMHs15. Our results, with far more incident hip fractures (465 vs 51), confirm this association. Further, we were able to analyze WMHs by radiographically-graded burden, the absence of which was a significant limitation of the Corti et al. analysis15.
Similar to Corti et al., our results suggested possible effect modification by age, though interaction terms were not statistically-significant and the direction of the association is different. Corti et al. reported no association between WMHs and incident hip fracture in participants aged ≥80, whereas our results suggest a stronger association among participants aged ≥80 compared to participants aged <80. This difference could be seen if the cohorts differed in important baseline characteristics. We do not have the requisite data to directly compare age-stratified baseline characteristics between the cohorts, but the age-stratified incidence rates of hip fracture per 1000 person-years were different (Age ≥80: CHS cohort, 22.8 [95%CI, 17.7, 29.3] vs Corti et al., 25.0; Age <80: CHS cohort, 9.7 [95%CI 8.3, 11.2] vs Corti et al., 7.7). The risk estimates in Corti et al. were also derived from a small sample size (161 participants aged ≥80, 20 hip fractures), with 95% confidence intervals that contained our point estimates. The association between WMHs and hip fracture was slightly stronger among female participants compared to male participants. This may relate to sex differences in the pathways for fracture development; a 2018 CHS analysis also found a slightly stronger association between subclinical cardiovascular disease (CVD) and hip fracture in women compared to men30.
There is a well-established connection between CVD and osteoporosis, with hypothesized causal pathways including shared risk factors and common pathophysiologic pathways, such as cytokines and growth factors associated with both vascular calcification and bone turnover29,30. The incomplete attenuation of the association between covert infarcts, WMHs, and incident hip fracture may, in part, be explained by inability to control for these factors in our analysis. Future studies are needed to identify additional shared risk factors and to formally test our hypothesized causal pathway and further elucidate linkages between brain microvascular disease and bone disease.
This study had several strengths, including its prospective design, large sample size relative to the previous study on WMHs and hip fracture15, and consideration of medical history, functional status, and frailty. Now commonly-used treatments for osteoporosis, such as bisphosphonates, were not available for use in the early 1990s and thus did not confound our analysis. This study also had limitations. CHS participants who agreed to brain MRI imaging were, on average, younger and healthier than participants who declined brain MRI imaging, which limits the generalizability of our results18. We utilized MRI at a single time point, and thus cannot discuss the impact of incident covert infarcts or incident increases in WMH grade on risk of hip fracture. CHS did not collect data on history of hip fracture prior to study enrollment; thus, we were unable to consider history of fracture in this analysis. Location of covert infarcts33 and WMHs10 has been shown to be important for risk of falls. CHS contains data on infarct location; however, approximately 80% of identified infarcts are in the basal ganglia33. Due to concerns for limited sample size (estimated <300 infarcts outside of the basal ganglia), we did not consider infarct location in this analysis. Our results are most applicable to periventricular WMHs as the majority of WMHs in our sample were periventricular18. Our analyses were post hoc and hypothesis-generating only. We informally looked at mediation by factors hypothesized to be on the causal pathway, but did not complete a formal mediation analysis and thus cannot definitively comment on mediation.
In conclusion, community-dwelling older adults with covert infarcts and WMHs, markers of brain microvascular disease, are at an increased risk of incident hip fracture. The association between covert infarcts, WMHs, and hip fracture was not attenuated by controlling for medical history, social history, and medication use, but was attenuated by controlling for physical function, cognitive function, and self-report of falls, likely mediators in the pathway from covert infarcts and WMHs to hip fracture. The attenuation was incomplete, suggesting there may be additional shared risk factors, residual confounding, and causal pathways in addition to the comprehensive risk factors and clinical characteristics considered in this analysis. While beyond the scope of this analysis, some have recommended fall prevention measures when markers of cerebral microvascular disease are identified on brain imaging17. Future research, including prospective cohorts with a pre-specified focus on brain and bone health, is needed to further elucidate linkages between brain microvascular disease and bone disease. Future research could also examine whether accounting for brain microvascular disease markers improves the performance of validated fracture prediction tools. Interventional studies to examine the effects of fall prevention measures on skeletal outcomes in older adults with markers of cerebrovascular disease also may be valuable.
Supplementary Material
Funding:
This research was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, 75N92021D00006, N01HC15103 and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. The funding sources had no involvement in the conduct of this research or preparation of the article.
This manuscript is also the result of work supported with resources and use of facilities of the Minneapolis VA Health Care System. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the US Department of Veterans Affairs or the US government.
Conflict of interest:
KS: No conflict of interest.
PB: No conflict of interest.
ZC has received research grants from the Lovell Foundation.
LC: No conflict of interest.
JC: No conflict of interest.
JB has received research grants from Eli Lilly and owns stock in Pfizer and Merck as part of an IRA.
JS: No conflict of interest.
LM: No conflict of interest.
HF: No conflict of interest.
References
- 1.Vernooij MW, Ikram MA, Tanghe HL, et al. Incidental findings on brain MRI in the general population. N Engl J Med. Nov 01 2007;357(18):1821–8. doi: 10.1056/NEJMoa070972 [DOI] [PubMed] [Google Scholar]
- 2.Gouw AA, Seewann A, van der Flier WM, et al. Heterogeneity of small vessel disease: a systematic review of MRI and histopathology correlations. J Neurol Neurosurg Psychiatry. Feb 2011;82(2):126–35. doi: 10.1136/jnnp.2009.204685 [DOI] [PubMed] [Google Scholar]
- 3.Debette S, Schilling S, Duperron MG, Larsson SC, Markus HS. Clinical Significance of Magnetic Resonance Imaging Markers of Vascular Brain Injury: A Systematic Review and Meta-analysis. JAMA Neurol. 01 2019;76(1):81–94. doi: 10.1001/jamaneurol.2018.3122 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Vermeer SE, Longstreth WT, Koudstaal PJ. Silent brain infarcts: a systematic review. Lancet Neurol. Jul 2007;6(7):611–9. doi: 10.1016/S1474-4422(07)70170-9 [DOI] [PubMed] [Google Scholar]
- 5.Taylor ME, Lord SR, Delbaere K, et al. White matter hyperintensities are associated with falls in older people with dementia. Brain Imaging Behav. Oct 2019;13(5):1265–1272. doi: 10.1007/s11682-018-9943-8 [DOI] [PubMed] [Google Scholar]
- 6.Zheng JJ, Delbaere K, Close JC, Sachdev PS, Lord SR. Impact of white matter lesions on physical functioning and fall risk in older people: a systematic review. Stroke. Jul 2011;42(7):2086–90. doi: 10.1161/STROKEAHA.110.610360 [DOI] [PubMed] [Google Scholar]
- 7.Zheng JJ, Lord SR, Close JC, et al. Brain white matter hyperintensities, executive dysfunction, instability, and falls in older people: a prospective cohort study. J Gerontol A Biol Sci Med Sci. Oct 2012;67(10):1085–91. doi: 10.1093/gerona/gls063 [DOI] [PubMed] [Google Scholar]
- 8.Whitman GT, Tang Y, Lin A, Baloh RW, Tang T. A prospective study of cerebral white matter abnormalities in older people with gait dysfunction. Neurology. Sep 2001;57(6):990–4. doi: 10.1212/wnl.57.6.990 [DOI] [PubMed] [Google Scholar]
- 9.Callisaya ML, Srikanth VK, Lord SR, et al. Sub-cortical infarcts and the risk of falls in older people: combined results of TASCOG and Sydney MAS studies. Int J Stroke. Oct 2014;9 Suppl A100:55–60. doi: 10.1111/ijs.12279 [DOI] [PubMed] [Google Scholar]
- 10.Blahak C, Baezner H, Pantoni L, et al. Deep frontal and periventricular age related white matter changes but not basal ganglia and infratentorial hyperintensities are associated with falls: cross sectional results from the LADIS study. J Neurol Neurosurg Psychiatry. Jun 2009;80(6):608–13. doi: 10.1136/jnnp.2008.154633 [DOI] [PubMed] [Google Scholar]
- 11.Del Brutto OH, Peinado CD, Mera RM, Del Brutto VJ, Sedler MJ. Neuroimaging signatures of cerebral small vessel disease and risk of falls in stroke-free older adults living in rural Ecuador. The Atahualpa Project. J Neurol Sci. Jul 2019;402:133–135. doi: 10.1016/j.jns.2019.05.019 [DOI] [PubMed] [Google Scholar]
- 12.Barzilay JI, Bůžková P, Fink HA, et al. Systemic markers of microvascular disease and bone mineral density in older adults : The cardiovascular health study. Osteoporos Int. 11 2016;27(11):3217–3225. doi: 10.1007/s00198-016-3649-9 [DOI] [PubMed] [Google Scholar]
- 13.Minn YK, Suk SH, Do SY. Osteoporosis as an independent risk factor for silent brain infarction and white matter changes in men and women: the PRESENT project. Osteoporos Int. Oct 2014;25(10):2465–9. doi: 10.1007/s00198-014-2785-3 [DOI] [PubMed] [Google Scholar]
- 14.Dyer SM, Crotty M, Fairhall N, et al. A critical review of the long-term disability outcomes following hip fracture. BMC Geriatr. 09 2016;16:158. doi: 10.1186/s12877-016-0332-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Corti MC, Baggio G, Sartori L, et al. White matter lesions and the risk of incident hip fracture in older persons: results from the progetto veneto anziani study. Arch Intern Med. Sep 2007;167(16):1745–51. doi: 10.1001/archinte.167.16.1745 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fried LP, Borhani NO, Enright P, et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. Feb 1991;1(3):263–76. doi: 10.1016/1047-2797(91)90005-w [DOI] [PubMed] [Google Scholar]
- 17.Tell GS, Fried LP, Hermanson B, Manolio TA, Newman AB, Borhani NO. Recruitment of adults 65 years and older as participants in the Cardiovascular Health Study. Ann Epidemiol. Jul 1993;3(4):358–66. doi: 10.1016/1047-2797(93)90062-9 [DOI] [PubMed] [Google Scholar]
- 18.Longstreth WT, Manolio TA, Arnold A, et al. Clinical correlates of white matter findings on cranial magnetic resonance imaging of 3301 elderly people. The Cardiovascular Health Study. Stroke. Aug 1996;27(8):1274–82. doi: 10.1161/01.str.27.8.1274 [DOI] [PubMed] [Google Scholar]
- 19.Manolio TA, Kronmal RA, Burke GL, et al. Magnetic resonance abnormalities and cardiovascular disease in older adults. The Cardiovascular Health Study. Stroke. Feb 1994;25(2):318–27. doi: 10.1161/01.str.25.2.318 [DOI] [PubMed] [Google Scholar]
- 20.Longstreth WT, Bernick C, Manolio TA, Bryan N, Jungreis CA, Price TR. Lacunar infarcts defined by magnetic resonance imaging of 3660 elderly people: the Cardiovascular Health Study. Arch Neurol. Sep 1998;55(9):1217–25. doi: 10.1001/archneur.55.9.1217 [DOI] [PubMed] [Google Scholar]
- 21.Bernick C, Kuller L, Dulberg C, et al. Silent MRI infarcts and the risk of future stroke: the cardiovascular health study. Neurology. Oct 2001;57(7):1222–9. doi: 10.1212/wnl.57.7.1222 [DOI] [PubMed] [Google Scholar]
- 22.Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol. Aug 2013;12(8):822–38. doi: 10.1016/S1474-4422(13)70124-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kuller LH, Longstreth WT, Arnold AM, et al. White matter hyperintensity on cranial magnetic resonance imaging: a predictor of stroke. Stroke. Aug 2004;35(8):1821–5. doi: 10.1161/01.STR.0000132193.35955.69 [DOI] [PubMed] [Google Scholar]
- 24.Bůžková P, Barzilay JI, Fink HA, Robbins JA, Cauley JA, Fitzpatrick AL. Ratio of urine albumin to creatinine attenuates the association of dementia with hip fracture risk. J Clin Endocrinol Metab. Nov 2014;99(11):4116–23. doi: 10.1210/jc.2014-2409 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Carbone LD, Bůžková P, Fink HA, et al. Association of Dietary Niacin Intake With Incident Hip Fracture, BMD, and Body Composition: The Cardiovascular Health Study. J Bone Miner Res. 04 2019;34(4):643–652. doi: 10.1002/jbmr.3639 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. Mar 2001;56(3):M146–56. doi: 10.1093/gerona/56.3.m146 [DOI] [PubMed] [Google Scholar]
- 27.Robbins J, Hirsch C, Whitmer R, Cauley J, Harris T. The association of bone mineral density and depression in an older population. J Am Geriatr Soc. Jun 2001;49(6):732–6. doi: 10.1046/j.1532-5415.2001.49149.x [DOI] [PubMed] [Google Scholar]
- 28.Barzilay JI, Buzkova P, Cauley JA, Robbins JA, Fink HA, Mukamal KJ. The associations of subclinical atherosclerotic cardiovascular disease with hip fracture risk and bone mineral density in elderly adults. Osteoporos Int. Oct 2018;29(10):2219–2230. doi: 10.1007/s00198-018-4611-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.den Uyl D, Nurmohamed MT, van Tuyl LH, Raterman HG, Lems WF. (Sub)clinical cardiovascular disease is associated with increased bone loss and fracture risk; a systematic review of the association between cardiovascular disease and osteoporosis. Arthritis Res Ther. Jan 17 2011;13(1):R5. doi: 10.1186/ar3224 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Farhat GN, Cauley JA. The link between osteoporosis and cardiovascular disease. Clin Cases Miner Bone Metab. Jan 2008;5(1):19–34. [PMC free article] [PubMed] [Google Scholar]
- 31.Rosano C, Kuller LH, Chung H, Arnold AM, Longstreth WT, Newman AB. Subclinical brain magnetic resonance imaging abnormalities predict physical functional decline in high-functioning older adults. J Am Geriatr Soc. Apr 2005;53(4):649–54. doi: 10.1111/j.1532-5415.2005.53214.x [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.