2 1 5 in vivo 6 7 9 10 12 7 10 12 14 15 Materials and Methods Setting http://www.gprd.com 16 International Classification of Diseases 17 18 http://www.pharmo.nl 19 Definition of Cases and Controls GPRD 20 21 PHARMO RLS Cases were patients aged 18 years and older with a first admission for a hip/femur fracture between January 1, 1991, and December 31, 2002. The date of the hospital admission was the index date. Up to four control patients, who did not have a history of any type of fracture, were matched to each case by year of birth, gender, region, and calendar time. Exposure Assessment 22 21 Statistical Analysis 23 In our analysis, we controlled for a wide range of clinical variables that have been associated with risk of falls or fractures. In the GPRD study, we included the following variables in the final model: history of diabetes mellitus, rheumatoid arthritis, hyperthyroidism, congestive heart failure, seizures, anemia, dementia, depression, psychotic disorder, cerebrovascular accident, chronic obstructive pulmonary disease, osteoporosis, and a record of back pain or falls in the year before the index date. Furthermore, prescriptions, in the 6 months prior to the index date, for anticonvulsants, nonsteroidal anti-inflammatory drugs (NSAIDs), methotrexate, hormone replacement therapy, other hypertensive drugs (low-ceiling diuretics, renin-angiotensin-aldosterone system [RAAS] inhibitors, calcium channel blockers), anxiolytics/hypnotics, antipsychotics, antidepressants, anti-Parkinson drugs, oral and inhaled glucocorticoids, and bronchodilators as well as information on body mass index (BMI) were retained in the model. In the PHARMO RLS study, variables included in the final model were a dispensing of benzodiazepines in the 3 months prior to the index date, or, in the 6 months prior to the index date, a dispensing of oral glucocorticoids, inhaled glucocorticoids, bronchodilators, statins, BMD-modifying drugs, hormone replacement therapy, antipsychotics, antidepressants, opioids, antiepileptics, other hypertensive drugs (low-ceiling diuretics, RAAS inhibitors, calcium channel blockers), antidiabetics, laxants, disease-modifying antirheumatic drugs, NSAIDs, or metoclopramide. A history of hospital admission for cerebrovascular disease, cancer, endocrine disorders, inflammatory bowel disease and other bowel diseases, obstructive airway disease, musculoskeletal and connective tissue diseases, anemia, and skin diseases prior to the index date were also included in the final model. Results 1 Table 1. Characteristics of hip/femur fracture cases and controls in the GPRD and PHARMO RLS Characteristic GPRD PHARMO RLS n n n n Gender   Women 75.8% 75.8% 72.9% 72.7% Age (years)   <65 13.9% 13.9% 15.8% 16.2%   65–79 30.8% 32.2% 36.6% 37.3%   ≥80 55.2% 53.9% 47.6% 46.5% a 22.1% 20.6% No data available  b   20–25 46.6% 42.4%     <20 18.0% 9.7% No data available    >25 35.4% 47.9%   Antidepressants 13.0% 7.2% 9.5% 5.1% Oral glucocorticoids 7.2% 4.4% 5.4% 3.5% Thiazides 11.9% 12.9% 12.1% 11.4% Nitrates 6.9% 7.4% 9.4% 9.1% Hormone replacement 0.6% 1.2% 1.1% 1.3% a b 2 1 Table 2. Use of β-blockers and risk of hip/femur fracture in the GPRD and PHARMO RLS β-blocker exposure GPRD PHARMO RLS Cases (%) Controls (%) Crude OR (95% CI) a Cases (%) Controls (%) Crude OR (95% CI) a Timing of use   Current use 4.5 5.9 0.70 (0.64–0.77) 0.82 (0.74–0.91) 12.4 13.7 0.91 (0.83–0.98) 0.87 (0.80–0.95)   Recent use 0.6 0.6 0.89 (0.70–1.14) 0.88 (0.68–1.15) 1.7 1.7 1.01 (0.82–1.24) 0.93 (0.75–1.15)   Past use 0.7 0.6 1.10 (0.86–1.40) 0.99 (0.76–1.28) 1.4 1.3 1.11 (0.88–1.40) 0.99 (0.78–1.26)   Distant past use 3.7 2.8 1.10 (0.98–1.22) 1.06 (0.94–1.20) 6.8 6.3 1.07 (0.96–1.20) 0.97 (0.87–1.09) Among current users of  -blockers Selectivity   Low 1.0 1.2 0.75 (0.63–0.90) 0.86 (0.71–1.05) 3.5 3.1 1.12 (0.97–1.30) 1.04 (0.89–1.21)   Medium 0.1 0.2 0.79 (0.48–1.32) 0.81 (0.47–1.41) 0.3 0.5 0.40 (0.20–0.84) 0.38 (0.18–0.79)   High 3.3 4.4 0.69 (0.62–0.76) 0.77 (0.69–0.87) 8.7 10.1 0.86 (0.79–0.95) 0.84 (0.76–0.93) Lipophilicity   Hydrophile 3.2 4.2 0.69 (0.62–0.77) 0.78 (0.70–0.88) 5.1 5.2 0.98 (0.86–1.11) 0.94 (0.83–1.06)   Intermediate 0.2 0.2 0.77 (0.50–1.20) 0.88 (0.56–1.39) 0.5 0.4 1.15 (0.77–1.73) 1.04 (0.69–1.58)   Lipophile 1.1 1.4 0.72 (0.60–0.85) 0.81 (0.68–0.98) 6.9 8.0 0.85 (0.77–0.95) 0.83 (0.74–0.92) First prescription   Yes 0.2 0.2 0.93 (0.58–1.48) 1.18 (0.69–1.99) 0.4 0.6 0.63 (0.42–0.95) 0.62 (0.41–0.94) b   <0.67 0.9 1.2 0.72 (0.59–0.87) 0.81 (0.65–1.00) 8.7 9.8 0.89 (0.81–0.98) 0.87 (0.79–0.96)   0.67–1.33 2.0 2.6 0.73 (0.64–0.83) 0.85 (0.74–0.99) 3.2 3.4 0.95 (0.82–1.11) 0.90 (0.77–1.06)   >1.33 1.4 2.0 0.64 (0.55–0.75) 0.81 (0.69–0.97) 0.4 0.4 0.95 (0.61–1.49) 0.85 (0.54–1.35) a b Fig. 1. dashed line solid circles solid line open circles  The most frequently prescribed β-blocker in the GPRD was atenolol (3.0% among cases vs. 4.0% among controls, adjusted OR = 0.80, 95% CI 0.71–0.90). Current use of other β-blockers was infrequent (propranolol 0.4% vs. 0.5%, metoprolol 0.2% vs. 0.2%) and not associated with a decreased risk of hip/femur fracture (adjusted OR = 0.90, 95% CI 0.76–1.06). In the PHARMO RLS, the most frequently used β-blockers at the index date were metoprolol (4.6% among cases vs. 5.6% among controls), atenolol (3.2% vs. 3.6%), sotalol (1.9% vs. 1.5%), and propranolol (0.9% vs. 0.9%). The adjusted ORs for current use of metoprolol and atenolol were 0.79 (95% CI 0.69–0.90) and 0.89 (95% CI 0.77–1.04), respectively. Current use of propranolol (adjusted OR = 0.98, 95 CI 0.74–1.21) and sotalol (adjusted OR = 1.15, 95% CI 0.93–1.42) were not associated with a protective effect of hip/femur fracture. 2 3 Table 3. Current use of β-blockers and risk of hip/femur fractures in patient subgroups Current use of β-blockers GPRD PHARMO RLS Cases  Controls (%) Crude OR (95% CI) a Cases (%) Controls (%) Crude OR (95% CI) a Gender   Men 3.4 4.5 0.68 (0.55–0.84) 0.77 (0.60–0.98) 10.1 12.0 0.83 (0.70–0.98) 0.77 (0.64–0.93)   Women 4.8 6.3 0.71 (0.64–0.78) 0.83 (0.74–0.93) 13.3 14.4 0.93 (0.95–1.02) 0.90 (0.82–1.00) Age (years)   <65 2.9 3.1 0.93 (0.68–1.26) 0.91 (0.62–0.91) 5.0 7.2 1.04 (0.80–1.35) 0.94 (0.70–1.27)   65–80 7.0 9.1 0.71 (0.62–0.80) 0.84 (0.73–0.98) 14.5 16.8 0.84 (0.58–1.12) 0.80 (0.70–0.91)   80+ 3.5 4.6 0.65 (0.57–0.75) 0.77 (0.66–0.91) 12.4 13.5 0.94 (0.84–1.06) 0.94 (0.83–1.07) History of any use of other antihypertensive drugs   No 2.2 2.4 0.89 (0.77–1.04) 0.97 (0.82–1.14) 8.3 8.1 1.02 (0.85–1.23) 1.01 (0.90–1.14)   Yes 9.7 14.7 0.60 (0.53–0.67) 0.73 (0.64–0.83) 20.9 26.9 0.73 (0.65–0.83) 0.76 (0.67–0.86) a 3 P 2 Fig. 2. A B dashed line solid circles solid line open circles 2  Discussion In both the GPRD and PHARMO RLS data sets, current use of β-blockers was associated with a decreased risk of hip/femur fractures. However, there was no reduced risk of hip/femur fractures among patients who did not have a history of using other antihypertensive drugs: a protective effect of β-blockers was only observed for patients with current or prior use of other antihypertensive agents. Even within this group of patients, no dose-response relationship with β-blocker use was found. The effect was constant with cumulative dose and the OR was below 1.0 even among patients who had just started treatment with β-blockers. As the mechanism by which β-blockers could influence BMD is likely to need some time to exert a clinically relevant effect, this finding suggests that the association between β-blockers and fracture risk is not causal. in vivo in vitro 3 24 25 6 7 12 8 26 10 27 30 31 32 33 34 36 37 38 A potential limitation is that we confined our study to hip/femur fractures and did not evaluate other type of fractures. It is possible that potential beneficial effects of β-blockers are present only at sites other than the hip/femur, but there is no evidence to support this. In conclusion, the reduction in hip/femur fracture risk was not related to cumulative dose of β-blockers and was only present in patients using β-blockers with a history of using other antihypertensive drugs as well. This suggests that the effect of β-blockers on hip/femur fracture is not causal.