Table 1.
Descriptions of the studies with main findings and limitations.
| Authors/Year | Agent Investigated | Study Type | n | Main Results | Limitations |
|---|---|---|---|---|---|
| Mortality | |||||
| Behanova et al., 2019 [13] | oBP iBP DMAB |
retrospective | 54,145 | - 17% lower risk of dying in treated women (HR 0.83, 0.71–0.98, p = 0.023) statistically significant only for intravenous bisphosphonates | - lack of criteria associated with treatment prescription - no data regarding compliance with treatment |
| Bondo et al., 2013 [14] | BP | cohort study | 42,076 | - lower 3-month mortality in BP-treated patients before fracture (OR 0.68, 0.59–0.77, OR 0.73, 0.61–0.88) for treatment initiation after the fracture | - possible confounders: lower mortality even in patients who filled only one prescription (OR = 0.84, CI 95% 0.73–0.95) |
| Bergman et al., 2019 [15] | BP | retrospective, cohort | 49,765 | - 15% lower mortality (HR 0.85, 0.79–0.91) in BP-treated patients | -possible confounders: the risk was lower starting from day 6 of treatment |
| Sambrook et al., 2011 [16] | BP | randomized prospective |
220 | - BP treatment was associated with reduced mortality after hip fracture (aHR of 0.92 per month treated) 8% relative reduction per month and 63% per year of treatment |
-relatively young and healthy hip fracture cohort -small number of patients |
| Brozek et al., 2016 [17] |
BP | retrospective | 31,668 | - lower risk of mortality with BP treatment before and after the hip fracture (lowest HR of 0.43 for treatment started after hip fracture) | -no data regarding compliance to treatment |
| Peng et al., 2016 [18] | BP | metanalysis of 4 trials (2 randomized and 2 prospective matched controlled studies) | 3088 | - lower mortality risk in BP-treated patients (OR 0.66, 0.52–0.85, p = 0.001) | - few studies, mixed randomized and non-randomized studies no data regarding the type of BP, dose, duration - no statistical heterogeneity (I2 = 35%) -dominated results by the HORIZON Recurrent Fracture Trial [12] |
| Cobden et al., 2019 [19] | BP | retrospective multicenter study | 562 | - 5-year survival rate was 16% for BP-treated patients compared to 5% for non-treated ones (p = 0.002); | - retrospective study no predictive risk factors for mortality included in the analysis - only patients after hemiarthroplasty were investigated -small number of patients |
| van Geel et al., 2018 [20] | BP | prospective cohort study | 5011 (2534 on BP) |
- lower mortality rate for BP treated patients (HR 0.79, 0.64–0.97) in 8 years of follow-up | - no data regarding adherence to treatment -patients included in the Fracture Liaison Services with more medical interventions included than anti-osteoporosis treatment -all major fracture included |
| Wang et al., 2019 [21] | BP | retrospective | 690 | - lower mortality for BP and non-BP osteoporosis medication (HR 0.35, 0.19–0.64) and (HR 0.49, 0.34–0.69) | - retrospective study -only 57% of the fractures were followed -small number of patients |
| Abtani et al., 2020 [22] | BP | population-based cohort study | 163,273 | - 28% lower mortality after hip fracture in current BP-treated patients - 42% lower mortality after hip fracture in patients with past BP exposure |
- no data regarding adherence to treatment |
| Boonen et al., 2011 [23] | Zoledronic Acid | randomized, placebo-controlled, double-blind trial | 508 men (248/260) 1619 women |
- lower mortality after 5 mg of Zoledronic (HR 0.71, 0.46–1.31 in men) and (HR 0.74, 0.54–1.02 in women) | - short median follow-up of 1.9 years |
| Prieto-Alhambra et al., 2014 [24] | Zoledronic Acid | secondary analysis from HORIZON randomized controlled trial | 4093 1966/2127 |
- lower mortality rates in the treatment arm (6.2% compared to 10.5%, in the placebo arm, p < 0.001) - smaller difference in mortality rates between the two groups in cognitive impaired patients (23.2% compared to 26.9%) |
- study design with main focus on cognitive impaired patients |
| Sing et al., 2018 [25] |
BP | retrospective cohort study | 4594/13,568 | - lower cardiovascular mortality rate (HR 0.33, 0.17–0.65) for alendronate and (HR 0.35, 0.2–0.63) for other BPs at 1 year -the lower realtive risk is mantained up to ten years of follow-up for alendronate (HR 0.59 0.44–0.79, p) or other BPs (HR 0.58, 0.44–0.75) |
- no clear criteria for treatment recommendation - only one prescription needed for inclusion - possible confounding cardiovascular risk factor not included |
| Nordstrom et al., 2017 [26] |
BP | restropective cohort study | 5845/15,518 | - decreased risk of death adjusted for all covariates (HR 0.79, 0.73–0.85) | -observational study -high mean time between the hip fracture and initiation of BP (331 days, rage of 1 to 2770 |
| Center et al., 2011 [27] | BP | prospective cohort Dubbo study |
1223 women/819 men (429 with fractures) |
-lower risk of mortality in women (HR 0.33, 0.16–0.66) in the multivariate analysis | -no additional data regarding fracture type or the number of hip fractures included -observational study - lack of treatment recommendation criteria |
| Degli Esposti et al., 2012 [28] | Mainly BP |
retrospective cohort study | 5636 (187 pre-fracture/651 post fracture) |
- 62.3% lower risk of death (−73.4%–−46.6%) in treated patients, (HR 0.377, 0.266–0.534, p < 0.001) | -retrospective study with small number of patients -common analysis for all anti-osteoporosis agents -all hip fractures regardless of traumatic event |
| Han et al., 2020 [29] | Teriparatide | metanalysis of 6 studies (2 randomized and 4 retraospective studies) | 607 (269/338) |
- lower mortality in the teriparatide group in the fixed model (OR 0.34, 0.13–0.88, p = 0.03) but the random-effect was not significant (OR 0.37, 0.12–1.09, p = 0.07) | - no heterogeneity (I2 of 4%) -observational studies that can exaggerate the effects -inconstant treatment duration -confounding factors not assessed |
| Hsu et al., 2020 [30] | BP/Raloxifene/Teriparatide/Denosumab | retrospective cohort study | 946 (210/736) |
- lower risk of mortality in the persistence group (HR 0.83, 0.62–1.11), p = 0.21 for all anti-osteoporosis treatment - higher risk of mortality in patients aged 70–79 (aHR = 1.29, 1.08–4.86, p = 0.031) and >80 (aHR = 3.11, 1.49–6.5, p = 0.003) in persistence group |
- small number of patients - no separate analysis for the non-BP agents - certain confounding factors not assessed |
| Lebanon et al., 2020 [31] | Zoledronic acid (46.3%)/Denosumab (34.1%)/Teriparatide | prospective cohort study | 253 (85/168) |
-mortality rate in treated patients was 5.1% at one year compared to 26.3% in naive patients (p < 0.001) | - possible effects of calcium and vitamin D - small number of patients |
| Gonzalez-Quevedo et al., 2020 [32] | any treatment | prospective cohort study | 724 | -all-cause mortality was lower in the FLS-treated group (HR 0.66, 0.47–0.94) compared to all patients before FLS - treatment after the implementation of the FLS associated higher mortality risk compared to treatment initiated before (1.75, 0.54–5.49) |
- small number of patients - no cause-and-effect relationtioship between mortality and FLS - small number of patients - patients included in the FLS with more medical interventions included than anti-osteoporosis treatment |
| Rotman-Pikielny et al., 2018 [33] | any treatment | prospective cohort study | 218/219 | - lower mortality rates in treated patients (4.3% versus 21.8%) with a 53% decrease in mortality risk in female sex (HR 0.47, 0.30-0.72, p < 0.001) - |
- small number of patients - outcomes observed in a multidiscliplinary team with probability of secondary unrelated effects in compliant patients - confounding factors not assessed |
| Second Fracture | |||||
| Shen et al., 2014 [34] | BP | nationwide population-based longitudinal observational study | 87,415 | - BP treatment after hip fracture had a negative risk association with second fracture (20.8% versus 32.3%, p = 0.023), aOR = 2.24, 1.38–2.90, p = 0.017 | - register-based study |
| Palacios et al., 2015 [35] | DMAB | randomized post hoc analysis FREEDOM |
7808 | - 39% relative risk reduction for a secondary fragility fracture in denosumab treated patients (10.5% compared to 17.3% in the non-treated group, p < 0.0001) | - post hoc analysis |
| Behanova et al., 2019 [13] | oBP iBP DMAB |
retrospective | 54,145 1919 oBP 1870 iBP 555 DMAB 42,795 untreated |
- higher risk of subsequent hip fracture in patients with antiresorptive treatment: men with oBP (HR 2.89, 1.58–5.30), women on DMAB (HR 1.77, 1.08–2.91), and iBP (HR 1.81, 1.35–2.41) |
- lack of data regarding adherence to treatment - short follow-up period - data regarding criteria for treatment recommendation |
| Sambrook et al., 2011 [16] | BP | randomized prospective | 220 | - additional fractures were observed in 53% without treatment and 26% BP treatment before the event | - relatively young and healthy hip fracture cohort -small number of patients |
| Brozek et al., 2016 [17] |
BP | retrospective | 31,668 | - higher hip refracture rates in BP treated patients, before or after the index hip fracture, regardless of age - OR 1.87, 1.32–2.66 of second fracture in BP users age 70–84, after 1-year post fracture |
- no data regarding compliance to treatment |
| Peng et al., 2016 [18] | BP | metanalysis of 4 trials (2 randomized and 2 prospective matched controlled studies) |
3088 | - second hip fracture different rate between the BP group and the control group (mean difference of 0.6, 0.39–0.93, p = 0.02) | - few studies, mixed randomized and non-randomized studies no data regarding the type of BP, dose, duration - no statistical heterogeneity (I2 = 35%) |
| Cobden et al., 2019 [19] | any treatment | retrospective multicenter study | 562 | - no significant differences in second fracture risk | -small number of patients - only patients after hemiarthroplasty were investigated |
| van Geel et al., 2018 [20] | BP | prospective cohort study | 5011 (2534 on BP) |
- lower risk for subsequent fractures (HR 0.60, 0.49–0.73) | - no data regarding adherence to treatment -patients included in the Fracture Liaison Services with more medical interventions included than anti-osteoporosis treatment -all major fracture included |
| Osaki et al., 2012 [36] | Risedronate | prospective matched cohort study | 529 (173/356) |
-HR 0.31, 0.12–0.79 in univariate and 0.218, 0.074–0.63 in multivariate analysis for subsequent fracture risk in risedronate treated patients | -no randomization -history of BP treatment in the control group -small number of patients |
| Liu et al., 2019 [37] | Zoledronic Acid | randomized controlled trial | 482 (353/129) | - lower refracture rate in the treatment group (5.9% compared to 8.5%, p < 0.01) | -short observation time -small number of patients |
| Prieto-Alhambra et al., 2014 [24] | Zoledronic Acid | secondary analysis from HORIZON randomized controlled trial | 4093 1966/2127 |
- no significant correlation between treatment arm and placebo arm regarding re-fracture (p > 0.05) | - study design with main focus on cognitively impaired patients |
| Lee et al., 2013 [38] | BP | retrospective | 59,782 | - lower incidence of second hip fracture in compliant patients compared to non-compliant (0.8% versus 2.3%, p < 0.001) - lower incidence of second hip fracture in persistent users compared to non-persistent (0.9% versus 2.4%, p < 0.001) |
-retrospective study -all hip fractures included regardless of the traumatic event |
| Boonen et al., 2011 [23] | Zoledronic Acid | randomized, placebo-controlled, double-blind trial | 508 men (248/260) 1619 women |
- 8.9% rate of new fractures in zoledronic acid treated compared to 15.6% in placebo-treated women | - short median follow-up of 1.9 years |
| Nordstrom et al., 2017 [26] |
BP | restropective cohort study | 5845/15,518 | - after BP initiation, the risk of hip fracture was lower (HR 0.76, 0.65–0.90) - BP users of more than 90 days had a lower risk of new hip fracture (HR 0.69, 0.54–0.87) - effect on second hip fracture risk decrease was seen even in later BP users with OR of 2.22 compared to 2.63 in never users - a decrease for any subsequent fracture was seen (HR 0.90, 0.78–1.04) |
-observational study - new fracture data based on the national registry entries -high mean time between the hip fracture and the initiation of BP (331 days, rage of 1–2770) |
| Han et al., 2020 [29] | Teriparatide | metanalysis of 6 studies (2 randomized and 4 retraospective studies) | 607 (269/338) |
- no significant difference in subsequent fracture risk between the two groups (OR 0.60, 0.30–1.18, p = 0.14) | - no heterogeneity (I2 of 0%) -observational studies that can exaggerate the effects -inconstant treatment duration -confounding factors not assessed |
| Degli Esposti et al., 2012 [28] | Mainly BP |
retrospective cohort study | 5636 (187 pre-fracture/651 post fracture) |
- lower risk of re-fracture of -53.3% (−67.3% – −33.2%), p < 0.001 | -retrospective study with small number of patients -common analysis for all anti-osteoporosis agents -all hip fractures regardless of the traumatic event |
| Hsu et al., 2020 [30] | BP/Raloxifene/Teriparatide/Denosumab | retrospective cohort study | 946 (210/736) |
- lower rate of recurrent fractures in the persistence group for all agents (aHR 0.64, 0.49–0.99, p = 0.043) -persistence BP users with lower recurrent fracture rates (aHR 0.54, 0.32–0.90, p = 0.018) |
- small number of patients -no separate analysis for the non-BP agent- certain confounding factors not assessed |
| Gonzalez-Quevedo et al., 2020 [32] | any treatment | prospective cohort study | 724 | - no statistically significant difference between groups | - small number of patients;- no cause-and-effect relationtioship between mortality and FLS - small number of patients - patients included in the FLS with more medical interventions included than anti-osteoporosis treatment |
| Chen et al., 2020 [39] |
Alendronate | population-based cohort study | 88,320 (9278/79,042 |
-lowest risk of second fracture in the MPR 75–100% (aHR 0.61, 0.47–0.78, p < 0.001) | - certain confounding factors not assessed - incomplete treatment adherence data - lack of criteria choice for aledronate treatment |
| Makras et al., 2020 [40] |
any treatment | multicenter prospective study | 392 | - significant increase in new fractures in patients not receiving anti-osteoporosis treatment, p < 0.001 | -small number of patients -small period of treatment- effects possibly related to the FLS interventions besides treatment |
n—number of patients, BP—bisphosphonates; oBP—oral bisphosphonates; iBP—intravenous bisphosphonates; DMAB—denosumab; HR—hazard ratio; aHR—adjusted hazard ratio; OR—odds ratio, aOR—adjusted odds ratio; FLS—Fracture Liaison Service; MPR—medication possession ratio.