Introduction

Fragility fractures impose a massive burden on health care systems and the global community [1]. Such fractures are the hallmark of osteoporosis, causing high morbidity, loss of independence, and negatively affecting the quality of life [2]. People who have experienced a fragility fracture (i.e., spontaneous or low-traumatic) have a greater fracture risk immediately after the event [3]. However, antiosteoporotic therapy administered soon after a fragility fracture may mitigate this risk [4]. Unfortunately, temporary or permanent medication discontinuation are frequent (> 50–80%), especially in secondary prevention [5, 6].

The detection of a major fragility fracture (i.e., spontaneous fracture or fracture resulting from a low-impact trauma/fall from standing height or less, occurring at the vertebral bodies, proximal hip, wrist, or humerus) is crucial to identify patients at high risk of subsequent fractures, evaluate bone fragility, and prescribe antiosteoporotic medication [7]. Following an initial fracture, several strategies may be adopted, although secondary preventive measures might not be promptly used. In the last decade, several initiatives (at the various levels—local, regional, national, and international) have been undertaken to improve secondary fracture prevention; these include fracture liaison services (FLS). FLS models were originally introduced in the orthopedic departments of tertiary referral centers as multidisciplinary teams, including coordinators, orthopedic surgeons, bone nurses, bone doctors (internists, endocrinologists, orthogeriatrics, rheumatologists), radiologists, and physiatrists, centered on the fractured patient. At present, these programs also involve primary care in the form of the general practitioner, which is fundamental to promote and support short- and long-term adherence to the antiosteoporotic treatments [8]. These models have proven to be effective in different settings and clinical pathways [9]. Indeed, FLS programs have been demonstrated to reduce fracture-related morbidity and mortality as well as decrease healthcare costs for the secondary prevention of fragility fractures [10].

This systematic review and meta-analysis aims to provide recommendations based on the best available evidence on the efficacy and effectiveness of clinical governance models. The findings may support decision-makers to minimize the cost and social burden associated with fragility fractures.

Methods

We conducted a systematic review to support the Panel of the Italian Fragility Fracture Guidelines (published on the platform of the Italian National Institute of Health) in formulating recommendations. Adopting the GRADE-ADOLOPMENT methodology [11] and the standards defined by the Sistema Nazionale Linee Guida (SNLG [12]), the multidisciplinary panel updated the clinical question of the Scottish guidelines (SIGN, Scottish Intercollegiate Guidelines Network [13]): “Is the use of clinical governance models, such as the so-called fracture liaison services, suitable for the post-fracture patient’s management?”

Inclusion and exclusion criteria

Randomized clinical trials (RCTs) and/or observational studies were selected if they met the following criteria: (1) population: patients who experienced a fragility fracture; (2) intervention: clinical governance models, such as case manager interventions or FLS; (3) comparison: standard care; (4) outcome: (i) primary outcome measures, specifically bone mineral density (BMD) testing rate, antiosteoporotic therapy initiation, adherence to antiosteoporotic medications, subsequent fracture, and mortality risk, and (ii) secondary outcomes were quality of life and physical performance.

Studies were excluded if they (i) were not published in the English language, (ii) did not report original findings (i.e., letters, case report), (iii) did not identify patients affected by a fragility fracture, or (iv) were not before and after studies on the clinical governance model implementation or did not consider standard treatment/non-attenders/another model as a comparator.

Data source and search strategy

We performed a PubMed, Embase, and Cochrane Library search to update the search of the SIGN guidelines, from 2013 up to 17 December 2020, and identified publications on clinical governance models for patients who have sustained a fragility fracture. A systematic review of the available literature was carried out according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [14] (Supplemental Material, Table S1). The search strategy (Supplemental Material, Table S2) included specific keywords and/or corresponding MeSH terms related to “fragility fracture” AND “integrated models of care.” We checked the reference lists of the studies and the systematic reviews identified during the search process.

Study selection and data extraction

Three independent authors (AB, GP, RR) screened titles and abstracts according to the search strategy and then assessed the full text of the potentially relevant studies. Discrepancies between reviewers were resolved by a consensus meeting. For each included publication, the following information was extracted: (i) first author, year, and country of publication, (ii) study setting, (iii) type of population, (iv) intervention and comparator, and (v) follow-up period.

Quality of studies

The systematic reviews were evaluated using the AMSTAR-2 checklist [15]. The quality of each included publication, derived by our search, was assessed using the Cochrane Risk of Bias (RoB) tool for RCTs [16] and the Newcastle-Ottawa scales [17] for observational studies. The following domains of the Cochrane RoB tool were appraised: selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other bias (such as funding bias). Each domain was classified as “high,” “low,” or “unclear” RoB to assess to what extent the publication did not provide sufficient information. In the Newcastle-Ottawa scales, the following domains were evaluated: selection, comparability, and outcome. The threshold for identifying high-quality studies was more than five points.

Quality of evidence

The quality of evidence of each outcome was judged by evaluating five dimensions (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) using the Grading of Recommendations Assessment Development and Evaluation (GRADE) approach [18]. The evidence was downgraded from “high quality” by one level if serious limitations were found for each of the five dimensions, or by two levels if very serious limitations were found.

Statistical analysis

The intervention effect was estimated using the dichotomized measure of risk ratio (RR) to evaluate the effect of clinical governance models. Where possible, we adopted the adjusted RR and pooled adjusted estimates from the original studies. Estimates were summarized if at least three studies reported the association of interest.

Heterogeneity between study-specific estimates was tested using X2 statistics [19] and measured with the I2 index (a measure of the percentage variation across the studies) [20]. Meta-analyses were conducted to combine the outcome data using the DerSimonian random effects model [21], which takes into account both the sampling variance within the studies and the variation in the underlying effect across studies, such as sample characteristics. Furthermore, subgroup analyses according to RCTs were carried out. A publication bias was tested using Egger’s regression and funnel plot visual analysis [22].

All tests were considered statistically significant for p-values less than 0.05. The analyses and the correspondent graphical visualization of forest and funnel plots were respectively performed by using RevMan V.5.4 (Nordic Cochrane Center) and STATA Software Program V.16.1 (STATA).

Results

Study selection

The objective of this study was to evaluate the efficacy of clinical governance implementation. A systematic literature review was carried out using the Embase, Medline, and Cochrane Central databases to update the clinical question elaborated by the SIGN Guideline [13]. As shown in Fig. 1, we identified 10,781 records.

Fig. 1
figure 1

Flowchart of study selection

Full size image

We excluded 10,461 studies because they were unrelated to the issue based on the title and/or abstract. Among the remaining 320 publications assessed for full-text review, we excluded the studies that (i) considered the wrong population (5), intervention (1), comparison (5), or outcome (10); (ii) were study protocol (3) or abstract (57); (iii) had a wrong study design such as letter or case report (14); (iv) were out of scope (182) or not published in the English language (1). Further, the full text of five studies was not available. The remaining 36 publications were considered for the analysis, respectively: 30 primary studies [23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52] and 6 systematic reviews [53,54,55,56,57,58], from which 47 studies [59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105] were extracted (Table 1; Supplemental Material, Table S3).

Table 1 Characteristics of included studies
Full size table

Study characteristics

The majority of the studies were conducted in Australia (n = 10 [27, 36, 43, 59, 87,88,89,90, 94, 95]), Canada (n = 17 [25, 28, 34, 35, 38, 60,61,62,63,64,65,66,67,68,69,70, 102]), USA (n = 18 [23, 29, 31, 32, 37, 46, 77,78,79,80,81,82,83,84,85,86, 92, 93]), and European Union (n = 19 [24, 26, 33, 40, 42, 44, 45, 48,49,50, 76, 91, 96,97,98, 100, 101, 103, 105]). Five studies were carried out in the Asian continent (Israel, Japan, Thailand, and Lebanon), four publications were from the UK, three studies were performed in Ireland and one in New Zealand.

Fifteen studies were RCTs [28, 60,61,62,63,64,65,66, 72, 77, 92, 94, 100, 102, 105], while the remaining papers were observational studies.

Average follow-up was 9 and 17 months respectively for RCT and observational studies, although eight of them did not specify it [23, 24, 26, 32, 75, 83, 93, 104]. Studies were conducted using information from hospital or community-based hospitals or general practitioners [23,24,25,26, 28,29,30,31,32,33,34, 39,40,41,42,43,44,45, 48,49,50, 52, 59,60,61, 63, 66,67,68,69,70,71, 73, 75,76,77,78,79, 81,82,83, 86, 91, 93,94,95, 97,98,99,100,101, 103,104,105], tertiary referral hospital or centers [27, 36, 47, 88,89,90], or both [87], community pharmacies [64], administrative data [35, 37, 46, 51, 65, 70, 84, 85, 96] or specialized clinics or centers [62, 72, 74, 80, 92, 102]. In general, patients had low trauma fracture, specifically hip [23, 24, 27, 31, 32, 34, 35, 40,41,42, 44, 45, 47, 49, 51, 52], upper extremity [28], and vertebral [29] fracture.

Only one [46] of the observational studies extracted through the search had an NOS score lower than 6 and was therefore assigned to the category of low-quality study. Generally, “Comparability of cohorts on the basis of the design or analysis” in the comparability section was the domain for which problems were encountered in the most studies (14 [23, 24, 26, 30,31,32, 34, 35, 38, 39, 45,46,47, 52]), followed by the domain “Adequacy of follow-up of cohorts” in the outcome section (5 studies [25, 26, 34, 44, 46]) and “Demonstration that outcome of interest was not present at start of study” in the selection section (3 studies [25, 44, 46]).

Two systematic reviews [57, 58] were assessed as low quality, while the remaining were of very low quality (Supplemental Material, Table S4).

Studies considered the following comparisons: (a) after vs before the implementation of a specialized [23, 24, 26, 31, 32, 34, 38, 40, 44, 48, 49, 59, 67, 69, 76, 80, 82, 83, 85, 87, 88, 91, 97, 104] or a FLS [25, 29, 30, 35, 37, 39, 41, 43, 47, 51, 71, 86, 101] model, (b) a specialized model [27, 45, 46, 73, 78, 81, 84, 92, 93, 102, 103] or FLS [28, 33, 74, 75, 94] vs a comparator model, (c) a specialized [52, 60,61,62,63,64,65,66, 68, 70, 72, 77, 79, 89, 100, 105] or FLS [36, 42, 98, 99] model vs standard care or a specialized [90, 96] or FLS [50, 95] model vs non-attenders.

Primary outcomes

As shown in Fig. 2a, increased BMD testing rate was detected after the implementation of a specialized model or FLS group compared their pre-implementation, respectively 10,946 and 5059. Overall, 20 studies detected a statistically significant RR of 1.92 (95% CI, 1.44 to 2.55) with a high heterogeneity between groups (I2 = 98%) and without publication bias (p = 0.29; Supplemental Material, Figure S1).

Fig. 2
figure 2

Evaluation of BMD testing rate a after vs before the specialized or fracture liaison service (FLS) model implementation, b in the specialized or FLS model vs comparator model, c in the specialized model vs standard care. Squares represent study-specific relative risk estimates (size of the square reflects the study-specific statistical weight, that is, the inverse of the variance); horizontal lines represent 95% CIs; diamonds represent summary relative risk estimates with corresponding 95% CIs; p values are from testing for heterogeneity between study-specific estimates. Asterisk indicates randomized controlled studies. Abbreviations: CI confidence interval, RR relative risk

Full size image

Then, higher BMD testing rate was found in the specialized or FLS model respect to comparator model (Fig. 2b) RR 2.31 (95% CI, 1.40 to 3.82), or standard care (Fig. 2c) RR 2.45 (95% CI, 1.86 to 3.23), with a high heterogeneity among groups (I2 = 97% and 88%). Evaluation of antiosteoporotic therapy showed increased initiation after the specialized or FLS model implementation (RR 1.91, 95% CI 1.58 to 2.29; 18 studies, Fig. 3a) or compared to a standard care/non-attenders (RR 1.87, 95% CI 1.50 to 2.32; 15 studies, Fig. 3c). Furthermore, improved adherence to treatment was detected after the implementation of FLS or specialized model (RR 1.54, 95% CI 1.03–2.31; 5 studies, Fig. 4a) or compared to a standard care (RR 1.31, 95% CI 1.01 to 1.26; 2 studies, Fig. 4c). Both analyses were characterized by high heterogeneity among studies and absence of publication bias (Supplemental Material, Figure S1).

Fig. 3
figure 3

Evaluation of antiosteoporotic initiation a after vs before the specialized or fracture liaison service (FLS) model implementation, b in the specialized or FLS model vs comparator model, c in the specialized model or FLS vs standard care/non-attenders. Squares represent study-specific relative risk estimates (size of the square reflects the study-specific statistical weight, that is, the inverse of the variance); horizontal lines represent 95% CIs; diamonds represent summary relative risk estimates with corresponding 95% CIs; p values are from testing for heterogeneity between study-specific estimates. Asterisk indicates randomized controlled studies. Abbreviations: CI confidence interval, RR relative risk

Full size image
Fig. 4
figure 4

Evaluation of antiosteoporotic adherence a after vs before the specialized or fracture liaison service (FLS) model implementation, b in the specialized or FLS model vs comparator model, c in the specialized model vs standard care. Squares represent study-specific relative risk estimates (size of the square reflects the study-specific statistical weight, that is, the inverse of the variance); horizontal lines represent 95% CIs; diamonds represent summary relative risk estimates with corresponding 95% CIs; p values are from testing for heterogeneity between study-specific estimates. Asterisk indicates randomized controlled studies. Abbreviations: CI confidence interval, RR relative risk

Full size image

Thus, a significant decreased risk of subsequent fracture and a reduction of the mortality rate was found after the specialized or FLS group implementation (subsequent fracture: RR 0.65, 95% CI 0.53 to 0.79; 2 studies; Fig. 5a; mortality: RR: 0.72, 95% CI 0.54 to 0.95; 12 studies, Fig. 6a) or respect to standard care/non-attenders (subsequent fracture: RR 0.57, 95% CI 0.37 to 0.87; 7 studies; Fig. 5c; mortality: RR 0.68, 95% CI 0.48-0.96; 9 studies; Fig. 6c). Both analyses were characterized by high heterogeneity between studies and no existence of publication bias (Supplemental Material, Figure S1).

Fig. 5
figure 5

Evaluation of the risk of subsequent fracture a after vs before the specialized or fracture liaison service (FLS) model implementation, b in the FLS model vs comparator model, c in the specialized or FLS model vs standard care/non-attenders. Squares represent study-specific relative risk estimates (size of the square reflects the study-specific statistical weight, that is, the inverse of the variance); horizontal lines represent 95% CIs; diamonds represent summary relative risk estimates with corresponding 95% CIs; p values are from testing for heterogeneity between study-specific estimates. Asterisk indicates randomized controlled studies. Abbreviations: CI confidence interval, RR relative risk

Full size image
Fig. 6
figure 6

Evaluation of the risk of mortality a after vs before the specialized or fracture liaison service (FLS) model implementation, b in the specialized or FLS model vs comparator model, c in the specialized or FLS model vs standard care/non-attenders. Squares represent study-specific relative risk estimates (size of the square reflects the study-specific statistical weight, that is, the inverse of the variance); horizontal lines represent 95% CIs; diamonds represent summary relative risk estimates with corresponding 95% CIs; p values are from testing for heterogeneity between study-specific estimates. Asterisk indicates randomized controlled studies. Abbreviations: CI confidence interval, RR relative risk

Full size image

For all of the aforementioned outcomes, the certainty of the evidence was downgraded from low to very low due to serious inconsistency and study design (Supplemental Material, Table S5).

All the above mentioned results are summarized in Supplemental Material, Table S6.

Subgroup analyses

Previous findings regarding the BMD testing rate and antiosteoporotic initiation were confirmed based only on RCTs, specifically for the specialized or FLS model implementation compared to standard care/non-attenders (Supplemental Material, Figures S2-3). Moreover, an increased antiosteoporotic initiation was found for the specialized or FLS model implementation respect to a comparator model (Supplemental Material, Figure S3). Conversely, the summary estimate of the RCTs showed a non-significant reduction in the adherence to antiosteoporotic treatment, risk of subsequent fracture or mortality (Supplemental Material, Figure S2-6).

Secondary outcomes

A systematic review [58] evaluated the effect of clinical care pathways that enrolled patients of over 50 years of age who had sustained a hip fracture. Twenty-two studies evaluated these secondary preventive measures compared to usual care. Twelve studies measured the health-related quality of life (HRQoL) between 3 and 12 months, which improved compared with usual care patients following hip fracture. Moreover, 19 studies estimated the physical function between 3 and 12 months that increased with respect to standard treatment. When the meta-analyses were stratified by length of follow-up, a greater HRQoL measure and physical function were found compared to usual care between 3 and 12 months.

Discussion

This systematic review evaluated a clinical question of the Italian Guideline [106] and a panel of experts formulated recommendations through a structured and transparent process. Specifically, we conducted a systematic review and meta-analysis on the efficacy of clinical governance models (i.e., FLS, structured service delivery models, nurse-led clinics) versus the pre-implementation, a comparator model or standard care/non-attenders in low-income and developed countries. These results highlighted that implementation of clinical governance significantly improved BMD testing rate, antiosteoporotic therapy initiation, adherence as well as reduced the risk of subsequent fracture or mortality compared to the standard care/non-attenders. Moreover, a higher BMD testing rate, number of patients who initiated antiosteoporotic therapy and adherence to the medications was found after the FLS or specialized model implementation respect to their pre-implementation.

The benefits of the abovementioned results were more evident considering the RCT that underlined the effectiveness of the integrated structure of care versus standard treatment, specifically for the BMD testing rate and the antiosteoporotic initiation, or versus a comparator model, specifically for the antiosteoporotic initiation. These findings are consistent with studies that evaluated the implementation of an FLS, which similarly to our study underlined the effectiveness in reducing the bone fragility evaluation and treatment gaps, and subsequent fractures and mortality rates [57, 58, 107]. The results of this meta-analysis enabled us to recommend the management of patients with fragility fractures through multidisciplinary care systems (e.g., FLS) which ensures patients' transition to out-hospital services.

The primary objective of an FLS is the prevention of subsequent fragility fractures, associated with indirect and direct costs attributable to the antiosteoporotic treatment, which should be administered for prolonged periods to maintain therapy in subjects at high risk of fracture [5, 6]. Recently, the scientific community has focused on the impact of fragility fractures and their clinical consequences. Structures such as the multidisciplinary FLS are becoming increasingly popular in medical communities around the world. In the last decade, these programs have been promoted and supported by international scientific organizations, such as the International Osteoporosis Foundation (IOF), the American Society for Bone and Mineral Research (ASBMR) and the European League Against Rheumatism (EULAR) together with the European Federation of National Associations of Orthopaedics and Traumatology (EFORT) [9, 108,109,110,111]. International scientific societies have largely endorsed and promoted the establishment of coordinated, multidisciplinary clinical care governance for the management of patients with recent major fragility fractures in various parts of the world [112,113,114,115,116].

Regarding secondary outcomes, the establishment of clinical care pathways compared to usual care was demonstrated to improve HRQoL and physical performance in a meta-analysis that included patients over 50 years. This acquires particular importance in older patients with comorbidities and potentially improves the cost-effectiveness of these systems in clinical practice, given the various comorbidities displayed by these subjects.

Limitations and strengths

Some limitations must be acknowledged. First, we considered different models of clinical governance, which may reduce the reliability of our findings. Moreover, the majority of studies were conducted in Europe or America, which may limit the generalizability of the results. Second, we have some concerns regarding heterogeneous multidisciplinary programs, characteristics of patients, fracture site at baseline, and length of follow-up. Third, the certainty of the evidence for the assessed outcomes was judged as “very low” or “low” due to the inconsistency of the estimates and the inclusion of observational studies with a modest sample size. Fourth, the majority of the included studies did not account for competing risks of death, which could have affected the results. Fifth, although falls may influence and increase the risk of subsequent fracture, this determinant was not an outcome of interest of the present meta-analysis. However, the role of falls will be investigated in a clinical question of the Italian Guideline and will be converted into a scientific article.

Despite the above limitations, this study presents points of strength. The exhaustive search strategy identified an overview of studies on the implementation of clinical governance programs. Then, the internal validity of the included studies was assessed using the Newcastle-Ottawa Scale for observational studies and the RoB tool for RCTs. Finally, preliminary performance indicators of FLS efficacy might be represented by BMD testing rate and initiation of treatment [109].

Perspectives

Rigorous RCT testing the efficacy and effectiveness of models of clinical governance in secondary fracture prevention (i.e., FLS) against “standard care” will not likely be furtherly pursued in the future, mainly for ethical reasons. Therefore, longitudinal, large “real-world” studies, preferably designed and homogenized for including specific Key Performance Indicators of the efficacy of FLS, as advised by the international initiative IOF Capture The Fracture initiative-Best Practice Framework [117], are expected to be included in future systematic analyses in this field to reinforce the results. With this respect, also results coming from the surveys carried out within National Registries, which are now at an advanced stage of development worldwide [118,119,120,121,122,123], will be likely incorporated in these future assessments.

Conclusion

This systematic review and meta-analysis indicate that the implementation of structured and integrated models of care increased the BMD testing rate, antiosteoporotic initiation and adherence to medication as well as reduced the risk of subsequent fracture and mortality and improved HRQoL and the physical function of patients experiencing a fragility fracture. The task force formulated recommendations on the introduction of these programs, although our systematic review judged outcomes affected by “very low” to “low” quality evidence.