Association of higher potency statin use with risk of osteoporosis and fractures in patients with stroke in a Korean nationwide cohort study

Abstract

This population-based cohort study aimed to evaluate the risk of osteoporosis and fractures associated with higher-potency statin use compared to lower-potency statin use in patients with stroke, using data from the Health Insurance and Review Assessment database of South Korea (2010–2019). Patients who received statin within 30 days after hospitalization for a new-onset stroke (n = 276,911) were divided into higher-potency (n = 212,215, 76.6%) or lower-potency (n = 64,696, 23.4%) statin initiation groups. The primary outcome was a composite of osteoporosis and osteoporotic fractures. Secondary outcomes were individual components of the primary outcome, including osteoporosis, vertebral fracture, hip fracture, and non-hip non-vertebral fracture. Cox proportional hazard models weighted by standardized morbidity ratios were used to estimate hazard ratios (HRs) with 95% confidence intervals (CIs). The risk of the composite outcome (HR 0.95, 95% CI 0.93–0.97), osteoporosis (0.93, 0.90–0.96), vertebral fracture (0.95, 0.91–0.99), and hip fracture (0.89, 0.84–0.95) were significantly lower in higher-potency statin users, while the risk for non-hip non-vertebral fracture was not significant (0.98, 0.95–1.02). The use of higher-potency statins compared to lower-potency statins was associated with a lower risk of osteoporosis, vertebral fracture, and hip fracture in patients with stroke.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-81628-z.

Introduction

Osteoporosis is a skeletal disease characterized by compromised bone strength and quality, leading to a higher risk of fragility fractures¹. Stroke is a well-recognized risk factor for osteoporosis and subsequent fractures due to loss of bone mineral density, gait disability, and a higher risk of falling after stroke^2–5. Previous studies have suggested an up to 7-fold increase in the risk of fractures in stroke survivors compared to the age- and sex-matched controls^6–9. On the other hand, a post-stroke fracture can lead to negative consequences for stroke survivors, including aggravating stroke injury, reducing functional recovery, and increasing the risk of morbidity and mortality^10–14. However, the preventive and treatment strategies for post-stroke osteoporosis remain an area of further investigation^14–16.

There is a long-standing interest in the potential beneficial effect of statin use on osteoporosis, based on several mechanisms that statins may stimulate osteogenesis and inhibit the activity of osteoclasts and bone loss^17,18. Statins also have cholesterol-independent pleiotropic effects, including inhibition of inflammation and thrombosis, beyond their lipid-lowering effect^19,20. Given that most stroke survivors receive statin therapy for secondary stroke prevention, evaluating the association between statin use and osteoporosis risk is clinically important^21,22. Several observational studies have suggested a lower risk of osteoporosis associated with statin use; however, the results are still conflicting and unconvincing^23–41. Moreover, only one study was conducted among patients with stroke, which reported a lower risk of post-stroke osteoporosis and fractures (hazard ratio [HR] 0.66, 95% confidence interval [CI] 0.58–0.76) in statin users versus non-users with an impressive dose-response relation²⁹. However, this study has several potential biases, such as a healthy user or sicker bias and immortal time bias, by comparing statin users with non-users^42,43. In particular, the reported dose-response relations may be exaggerated, as the long-term statin users are more likely to have a longer follow-up time for which the outcomes cannot occur (i.e., immortal time) compared with non-users^28–31,42.

To date, no study has been conducted among stroke survivors to evaluate the potential osteoprotective effect of higher-potency statins compared to lower-potency statins. Given the extensive use of statins among stroke survivors^21,22, and morbidity and mortality of post-stroke osteoporotic fractures^10–14, we conducted a large population-based cohort study to determine whether higher-potency statin use is associated with a lower incidence of osteoporosis and fractures compared to lower-potency statin use among patients with new-onset stroke.

Methods

Data source

We conducted a new-user, active comparator, population-based cohort study using the data from Korea’s Health Insurance and Review Assessment (HIRA) database (from January 1, 2010 to December 31, 2019). Under universal health coverage, this database covers the entire Korean population of > 50 million⁴⁴. This individual-level database provides comprehensive information, including patient demographics, medical procedures, hospital admission, diagnoses, and drug prescriptions, through inpatient, outpatient, and emergency visits of each unique patient. Drug prescription records include drug ingredients, dose, daily dosages, and dates of initiation and duration. Diagnoses are coded based on the International Classification of Diseases 10th Revision (ICD-10) system. The overall positive predictive value of the diagnosis codes recorded in the database was reported to be 82%⁴⁵. This study was approved by the Institutional Review Board (IRB) of Sungkyunkwan University (2020-12-002). Due to the retrospective nature of the study, the IRB of Sungkyunkwan University waived the need of obtaining informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline⁴⁶. All research was conducted in accordance with guidelines and regulations of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Participants and study design

The HIRA retrieved the data (2010–2019) for all patients with stroke who received at least one statin prescription between January 1, 2011 and December 31, 2019. Patients with stroke were defined as those who were hospitalized with a diagnosis of stroke (ICD-10: I63, I64 [ischemic stroke], I60, I61, I62 [hemorrhagic stroke]), which has been previously validated with a positive predictive value above 90%⁴⁷. Then, we excluded patients who met the following criteria: (a) history of stroke before the year 2011, (b) <19 years of age at new-onset stroke, (c) previous statin treatment before new-onset stroke, (d) history of osteoporosis, fracture or receiving osteoporosis medications before new-onset stroke, (e) history of osteomalacia, Paget’s disease, or vitamin D deficiency before new-onset stroke (f) no statin prescription within 30 days after new-onset stroke, and (g) < 30 days of follow-up after new-onset stroke (i.e., death, diagnosis of osteoporosis and fracture, or the end of study period within 30 days after new-onset stroke). Cohort entry was the date of new-onset stroke with hospitalization, and the follow-up start date was 30 days after cohort entry, while the 30 days served as an exposure assessment window. The overall study design is depicted in Supplemental Figure S1.

Exposure definition

We identified statin prescription records by using drug ingredient names, including simvastatin, lovastatin, pravastatin, fluvastatin, atorvastatin, rosuvastatin, and pitavastatin, which are all statins available in Korea. Higher-potency statin users were defined as patients who received at least one prescription of atorvastatin ≥ 20 mg/day, rosuvastatin ≥ 10 mg/day, or simvastatin ≥ 40 mg/day within 30 days after or on cohort entry, and the others were defined as lower-potency statin users. Statins with an LDL-C reduction potency of 24–36% were categorized as low potency and those with a reduction potency of 39–52% were categorized as high potency, based on the Korean guideline and the previous literature^48–51.

Study outcomes

The primary outcome was defined as a diagnosis of osteoporosis and osteoporotic fractures, and the secondary outcomes were defined as the individual components of the primary outcome, using ICD-10 codes, recorded by a physician from inpatient, outpatient and emergency visits, including primary and specialist care: osteoporosis (ICD-10: M80-M82), vertebral fracture (M48.4, M48.5, S22.0, S22.1, S32.0, S32.1), hip fracture (S72.0, S72.1, S72.2), and non-hip non-vertebral [NHNV] fracture (rib and thorax: S22.3, S22.4, S22.8, S22.9, lumbar spine and pelvis: S32.3, S32.4, S32.5, S32.7, S32.8, shoulder and upper arm: S42.2, S42.3, S42.8, forearm: S52.4, S52.5, S52.6, S52.8, S52.9, femoral neck: S72.3, S72.4, S72.8, S72.9, and patella: S82.0, S82.3, S82.5, S82.6)⁵². Using an on-treatment exposure approach, follow-up started 30 days after cohort entry until the earliest of outcome occurrence, treatment discontinuation or switching between the study drugs (i.e., switch from higher-potency to lower-potency or vice versa), death, or the end of study period, whichever came first. Continuous statin use was defined as the duration of one prescription overlapping with the start date of the next prescription, with a grace period of 50% of the days of supply for each prescription.

Statistical analysis

We used standardized mortality ratio weights (SMRW) to reweigh our study cohort and address differences in baseline characteristics between higher-potency and lower-potency statin users⁵³. Propensity scores (PS), the probability of receiving a higher-potency statin, were estimated using multivariable logistic regression, using the following covariates as potential confounders, all measured before or on cohort entry: demographics (age, sex, and health insurance type), stroke type (hemorrhagic and ischemic stroke), Charlson comorbidity index⁵⁴, comorbidities (including chronic kidney disease, cirrhosis, congestive heart failure, coronary artery disease, chronic obstructive pulmonary disease, dementia, depression, diabetes mellitus, epilepsy, hyperlipidemia, hypertension, hyperthyroidism, rheumatoid arthritis, Parkinsonism, alcohol problem, malignancy), and co-medication use (including calcium supplements, systemic corticosteroids, antiparkinson drugs, antipsychotics, antidepressants, antithyroid drugs, hormone replacement therapy). From these PS values, the study cohort was reweighed using SMRW, where higher-potency statin users were given a weight of 1, and lower-potency statin users were weighted by the odds of treatment probability⁵³. We conducted SMRW as the primary method to retain the entire study cohort, and other PS-based methods (inverse probability of treatment weighting and 1:1 PS matching) were performed as sensitivity analyses to assess the robustness of our primary results. The baseline characteristics of the study cohort were presented as number (percentage) and mean (standard deviation [SD]) for categorical and continuous variables, respectively. The differences between the two groups were assessed using a standardized mean difference (SMD), with the absolute value < 0.10 considered adequate.

We calculated crude incidence rates of each outcome per 1,000 person-years. SMRW-weighted Cox proportional hazard regression models were used to estimate HR with 95% CI for each outcome, and deaths were treated as censoring events. We also displayed weighted Kaplan-Meier curves with a log-rank test of each outcome comparing higher-potency statin users with lower-potency statin users.

Secondary and sensitivity analyses

We conducted two secondary analyses. First, we assessed a duration-response relationship. The cumulative duration was calculated by summing the duration of each statin prescription in a time-varying manner from cohort entry until the end of follow-up, using three predefined categories (≤1, > 1 to 3, > 3 years). Second, we assessed possible effect measure modification between exposure status and all covariates described above.

We performed six sensitivity analyses to demonstrate the robustness of our findings. For the first set of analyses, we repeated analyses using different PS-based methods of inverse probability of treatment weighting and 1:1 PS matching, respectively. Second, in addition to the on-treatment analysis, we conducted an intention-to-treat analysis to address potential informative censoring⁵⁵. Third, we repeated the analyses by varying the grace period to 25% and 75% of the prescription duration. Fourth, we repeated the analysis by including patients who were previously treated with statin before cohort entry (that is, prevalent users) to examine whether our main findings are consistent regardless of history of statin treatment and to address any potential impact of previous statin use on osteoporosis. Finally, we repeated the analyses by varying the exposure assessment window from 1 month to 6 and 12 months after cohort entry. All statistical analyses were performed using SAS 9.4 version (SAS Institute Inc., Cary, NC, USA). The PS was re-estimated in each secondary and sensitivity analysis. A 95% CI not overlapping 1.0 and 2-tailed p-value < 0.05 were considered statistically meaningful.

Results

Baseline characteristics

During the study period, we identified 276,911 patients hospitalized with new-onset stroke receiving at least one statin prescription within 1 month after the new-onset stroke (Fig. 1). Among the study cohort, 212,215 (76.6%) were prescribed higher-potency statins, with a mean (SD) age of 66.0 (13.6) years and 65.4% of men (Table 1). The median (IQR) follow-up durations were 0.6 (1.7) and 1.0 (2.4) years in higher-potency and lower-potency statin users, respectively. Before weighting, the two exposure groups were similar in most covariates except for stroke type (SMD = 0.35). After weighting, all the baseline characteristics were well-balanced, with all SMD below 0.1.

Fig. 1.

Flowchart for study cohort selection.

Table 1.

Baseline characteristics of higher-potency and lower-potency statin users before and after weighting.

	Before weighting			After weighting*
	Higher-potency statin (n = 212,215)	Lower-potency statin (n = 64,696)	SMD*	Higher-potency statin (n = 212,208)	Lower-potency statin (n = 212,100)	SMD*
Duration of follow-up, years
Median (IQR)	0.6 (1.7)	1.0 (2.4)	-	0.6 (1.7)	1.0 (2.4)	-
Mean (SD)	1.3 (1.7)	1.9 (2.1)	-	1.3 (1.7)	1.9 (2.1)	-
Age, year, means (SD)	66.0 (13.6)	65.6 (13.8)	0.03	66.0 (13.6)	66.1 (24.5)	0.00
Age group, n (%)
19–39	6,317 (3.0)	1,972 (3.0)	0.00	6,317 (3.0)	5,480 (2.6)	0.02
40–49	19,420 (9.2)	6,603 (10.2)	-0.04	19,420 (9.2)	19,626 (9.3)	0.00
50–59	43,386 (20.4)	13,902 (21.5)	-0.03	43,385 (20.4)	44,557 (21.0)	-0.01
60–69	49,679 (23.4)	14,350 (22.2)	0.03	49,676 (23.4)	48,542 (22.9)	0.01
≥ 70	93,414 (44.0)	27,869 (43.1)	0.02	93,410 (44.0)	93,895 (44.3)	-0.01
Sex, n (%)
Male	138,825 (65.4)	39,790 (61.5)	0.08	138,818 (65.4)	138,579 (65.3)	0.00
Female	73,390 (34.6)	24,906 (38.5)	-0.08	73,390 (34.6)	73,521 (34.7)	0.00
Medical aid recipients	12,174 (5.7)	4345 (6.7)	-0.04	12,174 (5.7)	12,284 (5.8)	0.00
Stroke type, n (%)
Hemorrhagic stroke	8,696 (4.1)	8,930 (13.8)	-0.35	8,696 (4.1)	8,734 (4.1)	0.00
Ischemic stroke	203,519 (95.9)	55,766 (86.2)	0.35	203,512 (95.9)	203,366 (95.9)	0.00
CCI, mean (SD)	0.6 (1.1)	0.6 (1.0)	0.00	0.6 (1.1)	0.6 (1.9)	0.00
CCI = 0, n (%)	135,107 (63.7)	40,971 (63.3)	0.01	135,107 (63.7)	134,659 (63.5)	0.00
CCI = 1, n (%)	37,087 (17.5)	11,536 (17.8)	-0.01	37,087 (17.5)	37,165 (17.5)	0.00
CCI = 2, n (%)	26,068 (12.3)	8004 (12.4)	0.00	26,064 (12.3)	26,262 (12.4)	0.00
CCI ≥ 3, n (%)	13,953 (6.6)	4185 (6.5)	0.00	13,950 (6.6)	14,014 (6.6)	0.00
Comorbidities, n (%)
Chronic kidney disease	2,730 (1.3)	821 (1.3)	0.00	2,727 (1.3)	2,751 (1.3)	0.00
Cirrhosis	1,067 (0.5)	353 (0.5)	-0.01	1,067 (0.5)	1,092 (0.5)	0.00
Congestive heart failure	8,411 (4.0)	2,604 (4.0)	0.00	8,410 (4.0)	8,509 (4.0)	0.00
Coronary artery disease	525 (0.2)	147 (0.2)	0.00	524 (0.2)	528 (0.2)	0.00
COPD	9,794 (4.6)	3,110 (4.8)	-0.01	9,794 (4.6)	9,826 (4.6)	0.00
Dementia	5,096 (2.4)	1,746 (2.7)	-0.02	5,096 (2.4)	5,142 (2.4)	0.00
Depression	10,357 (4.9)	3,352 (5.2)	-0.01	10,357 (4.9)	10,480 (4.9)	0.00
Diabetes Mellitus	41,614 (19.6)	12,337 (19.1)	0.01	41,613 (19.6)	41,753 (19.7)	0.00
Epilepsy	1,962 (0.9)	735 (1.1)	-0.02	1,962 (0.9)	1,968 (0.9)	0.00
Hyperlipidemia	13,290 (6.3)	3,578 (5.5)	0.03	13,283 (6.3)	13,239 (6.2)	0.00
Hypertension	90,117 (42.5)	27,749 (42.9)	-0.01	90,115 (42.5)	90,332 (42.6)	0.00
Hyperthyroidism	1,143 (0.5)	327 (0.5)	0.01	1,142 (0.5)	1,138 (0.5)	0.00
Rheumatoid arthritis	2,512 (1.2)	826 (1.3)	-0.01	2,512 (1.2)	2,543 (1.2)	0.00
Parkinsonism	2,118 (1.0)	680 (1.1)	-0.01	2,116 (1.0)	2,145 (1.0)	0.00
Alcohol problem	1,562 (0.7)	569 (0.9)	-0.02	1,562 (0.7)	1,581 (0.7)	0.00
Malignancy	9,083 (4.3)	2,345 (3.6)	0.03	9,076 (4.3)	9,093 (4.3)	0.00
Co-medications, n (%)
Calcium supplements	3,169 (1.5)	982 (1.5)	0.00	3,169 (1.5)	3,189 (1.5)	0.00
Systemic corticosteroids	89,159 (42.0)	27,188 (42.0)	0.00	89,154 (42.0)	89,093 (42.0)	0.00
Antiparkinson drugs	11,670 (5.5)	3,997 (6.2)	-0.03	11,670 (5.5)	11,728 (5.5)	0.00
Antipsychotics	6,848 (3.2)	2,304 (3.6)	-0.02	6,848 (3.2)	6,925 (3.3)	0.00
Antidepressants	23,250 (11.0)	7,689 (11.9)	-0.03	23,250 (11.0)	23,487 (11.1)	0.00
Antithyroid drugs	785 (0.4)	212 (0.3)	0.01	783 (0.4)	790 (0.4)	0.00
Hormone replacement therapy	1,851 (0.9)	700 (1.1)	-0.02	1,851 (0.9)	1,859 (0.9)	0.00

^*Weighted pseudo-population created by applying standardised morbidity ratio weights from propensity scores.

CCI, Charlson comorbidity index; COPD, chronic obstructive pulmonary disease; SRMW, standardized mortality ratio weights; SD, standard deviation; SMD, standardized mean difference.

Risk of osteoporosis and fractures

The incidence rates of any diagnosis of osteoporotic outcomes were 50.5 and 52.3 per 1,000 person-years in higher-potency and lower-potency statin users, respectively (Table 2). The use of higher-potency statins was associated with a lower risk of any osteoporotic outcomes (weighted HR 0.95, 95% CI 0.93–0.97) compared to lower-potency statins. For individual outcomes, the risks of osteoporosis (HR 0.93, 95% CI 0.90–0.96), vertebral fracture (HR 0.95, 95% CI 0.91–0.99), and hip fracture (HR 0.89, 95% CI 0.84–0.95) were significantly lower in higher-potency statin users than the reference. For any fractures, the composite of vertebral, hip, and NHNV fractures, the incidence rates were 32.0 and 32.7 per 1,000 person-year in higher and lower-potency statin users, respectively, and the weighted HR was 0.96 (95% CI 0.93–0.99). The cumulative incidence of any osteoporotic diagnoses was lower in higher-potency statin users than lower-potency statin users, with a slight divergence of the curves during the over 8-year follow-up time (log-rank p = 0.0007), and similar trends were shown in individual outcomes (Fig. 2).

Table 2.

Risk of osteoporosis and fractures comparing higher-potency statins with lower-potency statins in patients with stroke.

Outcomes	Higher-potency statin (n = 212,215)			Lower-potency statin (n = 64,696)			Hazard ratio
	No. of events	Person-years	Incidence rate*	No. of events	Person-years	Incidence rate*	Crude (95% CI)	Weighted†(95% CI)
Any outcomes†	14,286	283,003	50.5	6,350	121,328	52.3	0.92 (0.90–0.95)	0.95 (0.93–0.97)
Osteoporosis	7,164	293,972	24.4	3,318	126,966	26.1	0.90 (0.86–0.93)	0.93 (0.90–0.96)
Vertebral fracture	3,226	301,286	10.7	1,475	131,197	11.2	0.93 (0.88–0.99)	0.95 (0.91–0.99)
Hip fracture	1,647	304,206	5.4	781	132,683	5.9	0.89 (0.82–0.97)	0.89 (0.84–0.95)
NHNV fracture	5,445	297,468	18.3	2,379	129,467	18.4	0.97 (0.92–1.02)	0.98 (0.95–1.02)
Any fractures‡	9,327	291,790	32.0	4,130	126,254	32.7	0.95 (0.91–0.98)	0.96 (0.93–0.99)

^*Crude incidence rate per 1,000 person-year. ^† The models were weighted using standardized mortality ratio weights.

^†Any outcomes include osteoporosis and vertebral, hip, and NHNV fractures.

^‡Any fractures include vertebral, hip, and NHNV fractures.

CI, confidence interval; NHNV, non-hip non-vertebral.

Fig. 2.

Kaplan-Meier curves for (a) any osteoporotic outcome*, (b) osteoporosis, (c) vertebral fracture, (d) hip fracture, and (e) NHNV fracture in the higher-potency and lower-potency statin users after new-onset stroke. *Composite outcome including osteoporosis and vertebral, hip, and NHNV fractures. The inset shows the same data on an expanded y-axis. P-value < 0.05 in the log-rank test indicates a statistically significant difference in survival between the two groups. Abbreviations: NHNV, non-hip non-vertebral.

Secondary and sensitivity analyses

In secondary analyses, there was no clear duration-response relationship by the cumulative duration of statin use, with HRs ranging from 0.86 to 1.12 for primary and secondary outcomes (Supplemental Table S1). While there were lower risks of osteoporosis and fractures among patients ≥ 70 years old (HR 0.94, 95% CI 0.92–0.97) and females (HR 0.93, 95% CI 0.90–0.95), p-for-interaction values were not statistically significant (p = 0.7881 and p = 0.3109, respectively), which suggests no effect measure modification by age and sex (Table 3). The sensitivity analyses generated highly consistent results, with HRs ranging from 0.91 to 0.97 for the primary outcome (Table 3; Supplemental Tables S2-S7).

Table 3.

Subgroup and sensitivity analyses for risks of overall osteoporosis and fractures comparing higher-potency statins with lower-potency statins in patients with stroke.

Categories	Higher-potency statin				Lower-potency statin				Hazard ratio
	No. of patients	No. of events	Person-years	IR*	No. of Patients	No. of Events	Person-years	IR*	Crude (95% CI)	Weighted†(95% CI)
Subgroup analyses
Age group, year
19–39	6,317	82	9,509	8.6	1,972	37	4,273	8.7	1.01 (0.68–1.49)	0.99 (0.74–1.32)
40–49	19,420	434	29,251	14.8	6,603	230	13,530	17.0	0.88 (0.75–1.04)	0.97 (0.86–1.10)
50–59	43,386	1,708	62,063	27.5	13,902	864	28,017	30.8	0.89 (0.82–0.96)	0.97 (0.91–1.03)
60–69	49,679	3,262	68,850	47.4	14,350	1,402	27,616	50.8	0.92 (0.86–0.98)	0.96 (0.92–1.08)
≥70	93,413	8,800	113,331	77.6	27,869	3,817	47,891	79.7	0.92 (0.89–0.96)	0.94 (0.92–0.97)
Sex
Male	138,825	5,895	190,553	30.9	39,790	2,409	78,212	30.8	0.97 (0.93–1.02)	0.97 (0.94–1.003)
Female	73,390	8,391	92,450	90.8	24,906	3,941	43,115	91.4	0.95 (0.92–0.99)	0.93 (0.90–0.95)
Sensitivity analyses
Main analysis	212,215	14,286	283,003	50.5	64,696	6,350	121,328	52.3	0.92 (0.90–0.95)	0.95 (0.93–0.97)
IPTW	212,215	14,286	283,003	50.5	64,696	6,350	121,328	52.3	0.92 (0.90–0.95)	0.95 (0.93–0.97)
1:1 PS matching‡	212,215	14,286	283,003	50.5	64,696	6,350	121,328	52.3	0.92 (0.90–0.95)	0.94 (0.90–0.98)
Intention-to-treat definition	212,215	31,981	653,210	49.0	64,696	12,495	251,115	49.8	0.95 (0.93–0.97)	0.97 (0.95–0.98)
25% grace period	212,215	12,636	252,629	50.0	64,696	5,504	105,950	51.9	0.92 (0.89–0.95)	0.94 (0.92–0.96)
75% grace period	212,215	15,031	297,929	50.5	64,696	6,844	129,011	53.0	0.91 (0.89–0.94)	0.93 (0.92–0.96)
Including prevalent statin users	305,784	21,418	390,124	54.9	114,455	12,068	197,473	61.1	0.87 (0.85–0.89)	0.94 (0.92–0.96)
6-month exposure window	199,414	13,073	285,750	45.7	65,524	6,732	138,792	48.5	0.91 (0.89–0.94)	0.93 (0.91–0.95)
12-month exposure window	182,330	11,330	268,630	42.2	63,105	6,395	139,840	45.7	0.90 (0.87–0.92)	0.91 (0.89–0.94)

^*Crude incidence rate per 1,000 person-year. ^† The models were weighted using standardized mortality ratio weights except for IPTW.

^‡After 1:1 PS matching, 64,457 patients were included in each exposure group. Abbreviations: CI, confidence interval; IR, incidence rate; IPTW, inverse probability of treatment weighting; PS, propensity score.

Discussion

Principal finding

In this large population-based cohort study, we found that higher-potency statin use was associated with a modestly lower risk of osteoporosis and fractures than lower-potency statin use among patients with new-onset stroke (HR 0.95, 95% CI 0.93–0.97). For individual outcomes, the risks of osteoporosis (HR 0.93, 95% CI 0.90–0.96), vertebral fracture (HR 0.95, 95% CI 0.91–0.99), and hip fracture (HR 0.89, 95% CI 0.84–0.95) were significantly lower, while NHNV fracture was not (HR 0.98, 95% 0.95–1.02) in higher-potency statin users than lower-potency statin users. There was no clear duration-response relationship or effect modification by age and sex. The results were highly consistent across several sensitivity analyses, strengthening the robustness of our main findings.

Potential mechanisms

Although the biological mechanism to explain the osteoprotective effects of statins has not been fully understood, numerous pre-clinical studies have identified several potential pathways that statin may affect bone anabolism^17,24,56,57. First, statins have been shown to stimulate bone formation by upregulating the bone morphogenetic protein-2 (BMP-2), which plays a key role in osteoblast differentiation¹⁸, through the RAS/PI3K/AKT/MAPK signal pathway in a dose-dependent manner⁵⁶, which subsequently induces runt-related transcription factor 2 gene and stimulates osteoblast differentiation^58,59. Second, statins enhance osteoblast differentiation and activity by mobilizing osteoprogenitor cells and increasing osteogenic gene expression (e.g., Runx2, osteocalcin) and stimulating the Wnt/β-catenin signaling pathway and inhibiting adipogenic differentiation, redirecting mesenchymal stem cells toward osteoblast formation^60–69. Third, statins may indirectly affect bone deposition by their immunomodulatory effects, reducing systemic inflammation, which is known to contribute to bone loss by inhibiting bone formation^70–72. Fourth, statins inhibit the synthesis of intermediates such as farnesyl pyrophosphate (FPP) and geranylgeranyl-PP (GGPP) in the mevalonate pathway, which is key for osteoblast differentiation^73,74. Fifth, statin may dose-dependently reduce osteoclastogenesis via the OPG/RANKL/RANK pathway^75,76. Sixth, statins activate the TGF-β/Smad3 signaling pathway, which preserves osteoblasts from apoptosis and thereby supports bone deposition and maintenance^77,78. Lastly, satins can facilitate bone-protective estrogen signaling by inducing estrogen receptor α (ERα), which may help inhibit osteoclastogenesis⁷⁹. Thus, these mechanisms can stimulate bone formation and resorption and inhibit bone loss, contributing to the reduced risk of osteoporosis and fractures.

Comparison with previous studies

Although the results remain inconsistent and conflicting, there are several studies to investigate the association between statin use and the risk of osteoporosis and fractures^23–38. The post-hoc analyses of randomized clinical studies, including JUPITER and LIPD trials, have shown no association of statin use with fracture risk compared to placebo^35,36. However, those trials assessed only a fixed dose of statins among patients who were relatively healthy under specific conditions (e.g., patients without a prior history of cardiovascular disease and diabetes mellitus and with low-density lipoprotein cholesterol level < 130 mg/dL were eligible in the JUPITER trial)³⁵, which may limit the generalizability of the results to broader populations as well as the stroke population in real-world clinical settings. In a recent drug-targeted Mendelian randomization study, reductions in HMGCR-mediated non-HDL cholesterol, as achieved through statin use, are associated with increased heel bone mineral density, indicating a protective effect against osteoporosis⁴⁰.

On the other hand, several observational studies have suggested a lower risk of osteoporosis or fracture associated with statin use in the general or diabetic population^{23–28,30,31,39,41}, while only one study was conducted in patients with stroke²⁹. A recent meta-analysis reported that statin use was associated with a reduced risk of all fractures (relative risk 0.80, 95% CI 0.72–0.88)²³. However, all the prior studies compared statin users with non-users^{23–28,30,31,39,41}; thus, the results cannot rule out the impact of potential biases, such as a healthy user bias, confounding by indication, and detection bias. Moreover, in most prior studies, the definition of follow-up start date was unclear^27,30,32 and not equally applied between statin users and non-users^28,29, which can cause potential time-related bias and exaggerate the magnitude of risk estimates^42,43. In the present study, we conducted among stroke survivors, most of whom require statin therapy, and compared higher-potency statins with lower-potency statins, which allowed us to guarantee more comparability between the two groups and address potential biases that the prior studies could not. Given that most stroke survivors are under statin therapy, our results comparing different statin potencies may provide clinically meaningful information to manage osteoporosis risk in patients with stroke. Our results showed that higher-potency statin use was associated with a slightly lower risk of osteoporotic outcomes. Nonetheless, the potential benefit of higher-potency statins should be weighed against their potential risks, as several adverse outcomes, such as acute kidney injury or type 2 diabetes, have been suggested to be associated with higher-potency statins^80,81.

Our study found no duration-response relation of higher-potency statins with the risk of osteoporosis and fractures compared to lower-potency statins. On the contrary, several prior studies reported a remarkable dose-response relation between osteoporosis or fractures and statin use versus non-use^28–31. In a prior study conducted among stroke survivors, the HRs were 0.96, 0.86, and 0.34 for patients who received 1–90, 91–365, and > 365 cumulative defined daily doses of statins, respectively. However, they did not apply identical follow-up time criteria between statin users and non-users, which can induce selection bias and spurious dose-response relations. For instance, if we restrict patients to those who received > 365 cumulative daily doses of statin, most of them are likely to be followed over 1 year, and patients with outcomes within 1 year would have been excluded from the group, whereas those patients were not excluded in the non-user group. This can cause immortal time (the time for which the outcomes cannot occur) in statin users, especially in long-term users, which can cause an exaggerated protective effect. In our study, we applied identical categories of durations of statin use for both groups to address this issue. Also, as our study compared different statin potencies, the result can differ from that of comparing with non-users. Conversely, another study reported that low-dose statins were associated with a lower risk, whereas high-dose statins were associated with a higher risk of osteoporosis compared to non-use²⁷. However, this dose-dependent result may be affected by confounding by indication, whereby sicker patients who received higher statin doses are more likely to be diagnosed with osteoporosis³⁸.

Strengths and limitations of this study

This is the first study comparing higher-potency statin users with lower-potency statin users among stroke survivors. By comparing different statin potencies, we guaranteed comparability between the two groups, addressing several biases of the previous studies that can occur when non-users were the comparators. Also, this is the largest population-based cohort study evaluating the effect of statin on osteoporosis among patients with stroke. We generated generalizable real-world evidence using a nationwide database of 276,911 stroke survivors. Our study also has several limitations. First, misclassification of exposure and outcome is possible. However, the misclassification may be minimal as all statins are only available with prescriptions in Korea. Also, a previous validation study found a high positive predictive value for hip fracture with 93.1%.⁴⁵ Second, regarding outcome measures of osteoporosis, it does not necessarily indicate an incident event of osteoporosis as some diagnoses can result from a delay in diagnosis of pre-existing osteoporosis that has not been confirmed yet, which should be cautious when interpreting the results. However, the bias from the delayed diagnosis would not be significant because the delays may occur non-differentially between the two groups (Appendix S1). Also, we employed an active-comparator design, where both groups would have a consistent healthcare encounter and comparable chance of being diagnosed with osteoporosis. Moreover, the fractures, the outcome of which can be considered as incident events, were associated with the higher-potency statin use compared to the lower-potency, supporting our findings. Second, even though we compare among statin users, confounding by indication of higher-potency statins is still possible in both ways – i.e., sicker patients can receive higher potency statins, or they can receive lower potency statins to avoid its adverse effects. Third, due to the nature of observational studies, unmeasured confounding is possible as the HIRA does not collect information on laboratory data (e.g., creatinine, lipids, blood glucose, glycated hemoglobin), physical exercise and diet, body mass index, and supplements such as vitamin D, which are factors that may be associated with osteoporosis. Thus, to minimize unmeasured and residual confounding, we restricted our study cohort to patients with stroke-induced hospitalization, used lower-potency statin users as an active comparator, and considered the enormous number of covariates as potential confounders and their surrogates in PS-based models, achieving excellent balance in our study cohorts. Fourth, while ischemic and hemorrhagic strokes may differentially influence osteoporosis risk, we were unable to examine effect modification by stroke subtype in this study. Future research with access to subtype-specific data could provide valuable insights into the osteoprotective effects of statin use across distinct stroke populations. Lastly, while we assessed possible effect measure modification by the type of stroke (hemorrhagic and ischemic strokes), we did not consider more subdivisions of strokes (e.g., intracerebral hemorrhage and subarachnoid hemorrhage, thrombotic and embolic strokes) in the present study.

Conclusions

The findings of our study suggest that the use of higher-potency statins compared to lower-potency statins was associated with a slightly lower risk of osteoporosis, vertebral fracture, and hip fracture in patients with stroke. However, given the nature of the observational study, further studies, particularly randomized controlled trials and prospective studies, will be needed to verify this association and establish causality. Nonetheless, our findings may provide valuable insights for healthcare providers to determine the prevention and treatment strategy of osteoporosis and related fractures in stroke survivors with a high risk of osteoporosis, while the findings should be further confirmed through other prospective studies or clinical trials.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Drs Noh and Shin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. JSJ, YN, and JYS designed the research; all authors contributed to the acquisition, analysis, or interpretation of data; JSJ and YN drafted the manuscript; all authors contributed to the critical revision of the manuscript for important intellectual content; statistical analysis: YN; JYS obtained the funding; JSJ, YN, and JYS provided administrative, technical, or material support. All authors read and approved the final manuscript.

Funding

This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: RS-2024-00438443). This study was supported by a grant (RS-2024-00393167) from the Ministry of Food and Drug Safety, Korea, in 2024-2028. This research was supported by a grant (RS-2024-00332632) from the Ministry of Food and Drug Safety in 2024–2028. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data availability

No additional data available. Data generated and/or analyzed during the current study cannot be shared publicly due to the data-sharing policy of the Health Insurance Review and Assessment Service (HIRA) of Korea, governed by Article 18 of the Personal Information Protection Act(“Limitation to Out-of-Purpose Use and Provision of Personal Information” available at (https://elaw.klri.re.kr/kor_service/lawView.do?hseq=53044&lang=ENG). However, the data are available from the HIRA on reasonable request for researchers who meet the criteria for access to confidential data (https://www.data.go.kr/en/tcs/eds/selectCoreDataView.do?coreDataInsttCode=B551182&coreDataSn=1&searchCondition2=coreDataNmEn&searchKeyword2=).

Declarations

Competing interests

JYS received grants from the Ministry of Food and Drug Safety, the Ministry of Health and Welfare, the National Research Foundation of Korea, the Government-wide R&D Fund for Infectious Disease Research, and pharmaceutical companies, including Daiichi Sankyo, GSK, and Pfizer, outside the submitted work. YN received grants from the Ministry of Health and Welfare, outside the submitted work. SWC is a CEO/founder of Cellus, inc., outside the submitted work. HK has received research funds from JW Pharmaceutical, outside the submitted work. JSJ, CYH, and YC declare no competing interests. No other relationships or activities that could have influenced the submitted work.

Informed consent

Due to the retrospective nature of the study, the Institutional Review Board of Sungkyunkwan University waived the need of obtaining informed consent.

Ethical approval

This study was approved by the Institutional Review Board of Sungkyunkwan University (No. SKKU 2020-12-002), which waived the requirement for informed consent as only de-identified data was used.