This study detected no associations between fracture risk and ACEI or ARB use compared to non-users without hypertension in women. However, men using a lower dose of ARB medication had a higher risk of fracture than non-users without hypertension. Men with hypertension who did not use ARBs or ACEIs were also at an increased risk for fracture but this did not reach significance. This result is not unexpected, as previous literature suggests an increased risk of fracture in the presence of hypertension [4, 35]. No other associations were detected for men. Thiazide use did not alter the relationships observed in either men or women.

Previous studies have performed similar analyses with differing results. Several studies have utilised administrative databases such as a study by Choi et al. [10], which followed male and female participants over a 1.9 year period using medical claims data. The study reported that ARB use was not associated with an increased risk of fracture compared to non-users (HR, 95%CI 1.00, 0.95–1.05). However, ACEI users were at an elevated risk (1.68, 1.49–1.91). Models were adjusted for age, sex, diabetes and osteoporosis (diagnosis and treatment), however, data were not available for BMD or clinical measurements such as weight and height. In another study, Medicare data from the USA were used to follow new users of antihypertensive medications for 60–90 days [13]. Compared to calcium channel blocker use (referent group), men and women using ARB medications had a lower risk of fracture (HR, 95%CI 0.76, 0.68–0.86) whereas those using ACEI medications showed no difference (0.96, 0.90–1.04). The models in these analyses were adjusted for age, sex, ethnicity, osteoporosis (diagnosis and treatment), prior fracture, BMD, other medications, falls, hospitalisations and comorbidities. Importantly, height and weight were not included in these models. Additionally, the eligibility criteria for Medicare (age ≥ 65 years, USA citizen ≥ 5 years, younger person with a disability and people with end stage renal disease) could have resulted in a sample that was not generalisable to other populations. Kao et al. [15] used a health insurance database to follow participants with hypertension over a six year period. The results showed that both ACEI and ARB use were associated with a lower risk of fracture compared to non-users (HR, 95%CI 0.70, 0.62–0.79 and 0.58, 0.51–0.65, respectively). Potential confounders considered in the models included age, sex, comorbidities, socioeconomic variables and other medication use, however, weight and BMD were not available.

In a cross-sectional study, Rejnmark et al. [17] used data from computerised registers to investigate the risk of fracture in the year 2000, with the independent variable being ACEI use in the past five years. The study showed that compared to non-users, ACEI use was associated with a reduced risk of fracture (OR 95% CI, 0.93, 0.90–0.96). The models were adjusted for prior fracture, comorbidities, other medication use, hospitalisations, employment status and socioeconomic factors. Age and sex stratified analyses were also completed, which showed similar results. Another study examined the risk of fracture in new ACEI users compared to new ARB users using administrative databases [14]. The participants were aged > 65 years and time to fracture was similar for both ACEI and ARB users (mean ± SD; 1.7 ± 1.7 and 1.9 ± 1.8 years, respectively). The study reported that compared to ACEI users, ARB users did not have an increased risk of osteoporotic fractures (HR, 95%CI 0.90, 0.74–1.08). In this study, the models were adjusted for medication dose, though it is not clear whether the other available variables were included in the models.

Bokrantz et al. [11] investigated the association between hip fractures and ACEI or ARB use in hypertensive men and women aged ≥ 50 years using data from the Swedish Primary Care Cardiovascular Database. Participants were followed for a maximum of six years. The results showed that compared to non-users with hypertension, there were no differences in risk of hip fracture for either ACEI (HR, 95%CI 1.05, 0.95–1.15) or ARB use (0.98, 0.87–1.11). Models were adjusted for age, sex, BMI, smoking status, prior fracture, diabetes, other medications, comorbidities and socioeconomic variables. However, smoking and BMI data were missing for many participants in this study. The results were similar for men and women separately. An additional study by Ruths et al. [18] used data from the Norwegian Prescription Database and Norwegian Hip Fracture Registry to investigate the association between antihypertensive medications and hip fracture over a six year period. The authors reported an age effect, where individuals aged < 80 years using ACEI medication had an increased risk of hip fracture, while those aged ≥ 80 years had a reduced risk of hip fracture. Hip fracture risk was also reduced for individuals taking ARB medications.

A few cohort studies have also been used to investigate the risk of fracture in users of ACEI and ARB medications. A study by Kwok et al. [12] compared ACEI and ARB use to non-users with hypertension using data from the Osteoporotic Fractures in Men Study (MrOS), which followed 2573 men aged ≥ 65 years for a mean of 6.8 years. The results showed that compared to non-users, men taking ACEI or ARB medications had a lower risk of fracture. The authors also reported that a higher duration of ARB use was associated with a larger reduction in fracture risk, however, the same was not observed for ACEI use. Models were adjusted for age, other medications, prior fracture, inability to complete a narrow walking trial, falls, BMD and depressed mood. This study did have data on weight, but this was not included in the final models. Another study followed women over a median 6.5 year period and showed that ACEI use was not associated with fracture risk (HR, 95%CI 0.78, 0.47–1.29) and neither was ARB use (1.16, 0.69–1.98) [36]. However, the authors reported that there was a dose effect observed, where longer term use (> 3 years) was associated with a lower risk of fracture, whereas the opposite was true for shorter term use (≤ 3 years). The models were adjusted for a large number of potential confounding variables including age, BMI, ethnicity, smoking status, alcohol consumption, prior fracture (self and parents), comorbidities and other medication use.

Other studies have investigated the risk of fracture considering ACEIs and ARBs combined, rather than separately. For example, Chen et al. [37] used a health insurance database to investigate the risk of fracture for men and women with hypertension who were aged ≥ 40 years. The results showed that individuals using a RAAS blocker (which included ACEIs, ARBs and mineralocorticoid receptor antagonists) had a lower risk of fracture compared to non-users (HR, 95%CI 0.66, 0.59–0.75). This effect was observed for both men and women, as well as across multiple age groups. The models were adjusted for age, sex, comorbidities, other medications and socioeconomic factors, however, weight and BMD were not available. Another study by Shea et al. [16] included men and women aged ≥ 65 years and determined the risk of hip fracture for users of ACEIs or ARBs (combined). Compared to non-users, those who used ACEI/ARB medications had a lower risk of hip fracture (HR 95%CI, 0.707, 0.585–0.853). The models were adjusted for age, sex, socioeconomic variables, prior fracture and other medication use. Kunutsor et al. [38] conducted both a meta-analysis and a cohort study in their publication, the latter of which followed men and women for a median of 14.8 (12.8–15.8) years. The authors reported that ACEI or ARB use was not associated with fractures compared to non-users (HR, 95%CI 1.00, 0.59–1.69). Their models included age, sex, BMI, smoking status, alcohol consumption, diabetes, blood pressure, other medication use, socioeconomic status and physical activity. In these three above studies, it is not clear if the associations are driven by ACEI use, ARB use, or both, as other studies have reported that the two different classes of medication have differing associations with fracture.

Overall, there have been several studies that have examined associations between ACEI and/or ARB medication use and fracture risk. However, the results have been inconsistent and this may be due to heterogeneity among studies, making meaningful comparisons difficult. We have described examples of this heterogeneity above, highlighting the major differences in studies that make it challenging to compare our data with previous work. For example, many studies did not have data for potential confounding variables such as weight and BMD, some had short or unspecified follow-up duration, some examined hip fractures only, some include men and women whereas others only include one sex, some have combined ACEIs and ARBs whereas others have not, different studies have used different control groups, some studies have examined new users of ACEIs and ARBs whereas others included longer duration users and age groups differ across the studies. Our study has some strengths in that we have examined ACEI and ARB use separately, both men and women were included, adjustments for other confounding variables were made (particularly weight and BMD), follow-up time was long and all fractures were considered (except high trauma, face/skull, fingers and toes).

Our results for ACEI use do agree with several studies reporting no association with fracture risk [11, 13, 36]. However, there are also studies reporting an elevated fracture risk [10] or lower fracture risk [12, 15, 17]. Different associations have also been reported by age, where fracture risk was both elevated and reduced depending upon age (Ruths et al. [18]). In several studies, no associations between ACEI use were observed when younger age groups were considered (e.g. ≥ 50 years) [11, 36] (and this study), whereas two studies including an older group of participants (e.g. ≥ 65 years) have reported differences in fracture risk [12, 17]. Overall, for ACEI use, previous literature, as well as our own study generally indicate a lower or unaffected risk of subsequent fracture. No previous studies have reported that ARB use was associated with a higher risk of fracture in men. Previous work has generally indicated a reduction in fracture risk with ARB use, irrespective of age group studied [12, 13, 15, 18]. It is unclear why our study differs from the literature, but may be related to ARB medications not being available in Australia until 1997, whereas ACEI medications were available much earlier [27]. Of the men in this study who were currently taking an ARB medication, 29 reported taking another antihypertensive medication previously. Of these, 10 were previously taking an ACEI medication, five were taking an ARB medication and 14 were taking a different type of antihypertensive medication (such as a calcium channel blocker). For men currently taking an ACEI medication, 19 had a previous antihypertensive medication, where 12 of these were on previous ACEI medication, while the other seven were neither ACEI nor ARB medications. None of the current ACEI users had previously taken an ARB medication. It is possible that some individuals currently taking ARB medication may have changed from a previous antihypertensive medication due to insufficient reduction in blood pressure, perhaps with a background of other comorbidities, or alternatively experienced adverse drug events such as persistent cough. In this study, men taking a lower dose of ARB medications had a higher number of comorbidities compared to those taking a higher dose, which may have affected fracture risk. One comorbidity, diabetes, which is known to affect fracture risk [39,40,41], was more common among those using low dose (17.1%) compared to high dose (8.3%) ARB medications. Men taking a lower dose of ARB medication also had lower weight compared to those taking a higher dose and this is important because BMD increases with increasing BMI [42]. Femoral neck BMD appeared to be lower for men taking lower compared to higher ARB doses, though the difference was not significant (p = 0.699). Another possible reason for the results observed in this study is that individuals taking a lower dose of ARB medications may not have sufficient blockage of the angiotensin II receptor type 1. Since angiotensin II is detrimental to bone by promoting bone resorption and inhibiting bone formation [8, 9], it is possible that insufficient blockade of the angiotensin II receptor may lead to reduced bone quantity and quality leading to an increased fracture risk. Finally, the men currently using an ARB medication may also have had specific unknown characteristics that placed them at a higher risk for fracture that were not accounted for in the analyses.

In this study, we also reported different results for men and women, which may be related to differences in response to RAAS inhibition. It has been reported that female rats had a greater change in blood pressure compared to males following administration of the ACEI, enalapril [24]. It has also been reported that overall, males have greater expression levels and responses to angiotensin II, angiotensin II type 1 receptor and the angiotensin converting enzyme compared to females [22]. Additionally, androgens can promote the synthesis of angiotensinogen, which leads to higher levels of angiotensin II [25]. Therefore, it is possible that men in this study showed a greater response to angiotensin II levels, leading to different observations compared to women, who may have had a greater response to RAAS inhibition. Additionally, participants in this study were aged ≥ 50 years, where the fracture risk for women is higher than in men, following menopause [43]. Thus, it is also possible that women in this study were at an increased risk for fracture, such that RAAS inhibition was not a significant risk factor for fracture, while for men this was not the case.

While the assessment of other anti-hypertensive agents and their association with fracture risk was outside the scope of this study, we note that previous literature shows mixed associations between other anti-hypertensives (such as loop diuretics, thiazides and beta blockers) and fracture risk [11, 13]. This may be an avenue for future research.

There were several strengths and limitations to this study. The participants were randomly selected and population-based. Some self-reported data were used in this study, however some variables including fracture endpoints were objectively measured. However, since radiological reports were used to determine fracture endpoints, there was minimal loss to follow-up, even if a participant did not attend a subsequent follow-up visit. In determining fractures, we have taken an inclusive approach and utilised ICD-9 codes for cause of injury which aim to capture all minimal trauma fractures, and acknowledge that we cannot guarantee that all higher trauma fractures were excluded from the study. National Deaths Index data were also used to determine date of death, which further minimised bias from loss to follow-up. Non-users without hypertension were younger than their counterparts with hypertension, therefore, models were adjusted for age, but additional confounding from other factors associated with age may not have been captured in these models. Additionally, data were available for other potential confounders including weight, other medications and comorbidities. However, these confounders and medication use were assessed at a single time point prior to assessment of fracture and changes may have occurred in these variables in the intervening time period. There may also be residual confounding that had not been taken into account. Additionally, since medication use was only assessed at a single time point, and this was earlier for men than women, it is possible that changes in the prescription of ARB medications in Australia over time could have influenced the observed differences between the sexes. It is also possible that this study had insufficient power to detect differences between the groups, particularly for women. An association between ACEI use and fracture risk was observed in unadjusted analyses, though this was attenuated following adjustment for other variables. However, the point estimate remained elevated in adjusted analyses and it is possible that there was insufficient power to detect any differences. There was also an insufficient number of participants using different types of ACEI and ARB medications (e.g. irbesartan and candesartan for ARBs) and it was not possible to assess differences between these different agents. In addition, compliance with medication dose was not available. It is also possible that some participants changed medication dose or type during the follow-up period and this could have affected the results.