The osteoporosis treatment gap in Switzerland between 1998 and 2018

Osteoporosis is a crippling disease with alterations in bone quantity and quality leading to an increased risk of fragility fractures most commonly located at the hip, spine, distal radius, and/or proximal humerus, also known as major osteoporotic fractures (MOF). The operational definition of osteoporosis proposed by the World Health Organization relies on a T score at or below 2.5 measured by dual-energy X-ray absorptiometry (DXA) at the femoral neck [1].

In Switzerland, approximately 1 in 2 women and 1 in 5 men will sustain a fragility fracture during their remaining lifetime after age 50 [2]. While increasing in number, the incidence and even more so the age-standardized incidence of hospitalizations for hip fractures have been shown to follow a long-lasting decreasing trend between 1998 and 2018 in both men and women [3]. During these 21 years of observation, the proportion of hospitalizations for hip fractures of all hospitalizations for MOF decreased from approximately 55% in both men and women to 44% and 42% in men and women, respectively. Among the most prescribed osteoporosis drugs in Switzerland, the aminobisphosphonates alendronate and zoledronate and the RANKL inhibitor denosumab were shown to significantly reduce the risk of hip fractures in fracture endpoint trials [4‐6] and their use may have contributed to the observed reduction in hip fracture incidence.

Since the launch of the first oral bisphosphonate approved for the treatment of osteoporosis in Switzerland (alendronate in 1996), an array of drug therapies aiming at reducing fracture risk in patients with osteoporosis has been made readily available to physicians and patients, allowing for multiple therapeutic options based on different mechanisms of action, efficacy and safety profiles, and administration routes and schemes. In Switzerland, approved and reimbursed osteoporosis pharmacological treatments include bisphosphonates (as daily, weekly, or monthly tablets and as quarterly or yearly intravenous infusions), denosumab as twice yearly subcutaneous injections, teriparatide as daily subcutaneous injections, and raloxifene as daily tablets. Of note, reimbursement restrictions generally based on a DXA T score threshold and/or the presence of one or more prevalent fractures, apply to all except alendronate (Supplemental table S1). Off-label prescriptions and prescriptions of non-reimbursed drugs are the exception rather than the rule and can be considered negligible.

In earlier publications, different groups have used different methodologies and definitions for estimating the osteoporosis treatment gap. While the common understanding is that treatment gap defines the proportion of patients who do not get treatment although eligible for therapy based on a given intervention threshold, the source databases and the inclusion criteria differed widely. Some authors estimated the treatment gap based either on national health databases and database linkage such as the National Danish Health Registries [7] or the National Medicare Database in the US [8] and others performed prospective studies such as the Austrian ICUROS [9] and the Belgian FRISBEE [10] studies. Typically, all these studies included patients who had experienced an index major osteoporotic fracture, either only women [8, 10] or both men and women [7, 9] with diverging criteria regarding age at inclusion. The gap was then estimated on these (sub-)populations with treatment generally defined as one of the bone active substances described above and generally excluding calcium, vitamin D, and estrogens, at the exception of the study by Malle et al. in which the latter were included [9]. By contrast, the SCOPE 2021 project, aiming at standardizing the approach for cross-country comparisons (EU27 + Switzerland + United Kingdom), followed a different methodology. SCOPE 2021 defined the patient population eligible for treatment as those with a 10-year probability of MOF exceeding that of a woman with a prior fragility fracture, estimated the number of patients treated based the IQVIA drug sales database (whereby calcium, vitamin D, and estrogens were not considered), and adjusted for adherence by using a point estimate derived from the Swedish Prescribed Drugs Register [11]. Thus, treatment gap studies and results should not be compared directly without the necessary caution with regard to the methodology applied.

The aim of the present analyses was to determine the number of patients treated with a pharmacologically active osteoporosis drug in Switzerland between 1998 and 2018, to evaluate the changes in treatment patterns following the introduction of new substances, to identify and quantify a potential osteoporosis treatment gap, and to explore the association between treated patients and changes in hip fracture incidence.

Yearly sales of counting units (i.e., the smallest available single dosage of a preparation, e.g., a tablet or a vial) of all pharmacologically active osteoporosis drugs including generics in all formulations, dosage strengths, and presentations available in Switzerland since 1998 and up to year 2018, were obtained from IQVIA Switzerland (IQVIA RDS Switzerland Sàrl, Saint-Prex, Switzerland). Pharmacologically active osteoporosis drugs were defined as oral bisphosphonates (alendronate 10 mg daily or 70 mg weekly, risedronate 5 mg daily or 35 mg weekly, ibandronate 150 mg QM, including generics), intravenous bisphosphonates (ibandronate 3 mg Q3M and zoledronate 5 mg Q12M), denosumab 60 mg Q6M, raloxifene 60 mg daily, and teriparatide 20 mcg daily. Not included were abaloparatide and strontium ranelate (not available in Switzerland) and basedoxifene (available but not reimbursed). Early bisphosphonates (including etidronate, clodronate, and pamidronate) were neither available in Switzerland nor approved for the treatment of osteoporosis and thus not included in the analysis. Counting units were converted into treatment-years by applying the annualized recommended dose for the treatment of osteoporosis published in the Swiss prescribing information (https://​compendium.​ch). Of note, prescription data were not available by sex, such that only the total number of patients treated could be calculated.

In the absence of Swiss data, persistence defined as the duration of time from initiation to discontinuation of therapy was calculated based on previously published findings in women with US Medicare fee-for-service coverage [12]. The annual persistence rates of women with records covering three years or more were averaged leading to values of 0.596, 0.476, and 0.312 for denosumab, intravenous bisphosphonates, and oral bisphosphonates, respectively. The number of patients on treatment with a pharmacologically active osteoporosis drug in a given year was obtained by multiplying the number of treatment-days for each substance with the corresponding persistence values. It was assumed that persistence did not change between 1998 and 2018.

The total number of patients eligible for treatment (patient potential) was estimated based on four scenarios. For all, the annual Swiss population statistics by sex and 5-year age groups after age 45 published by the Swiss Federal Office of Statistics were used as a starting point. The most conservative scenario considered that the patient potential was restricted to men and women with a T score at or below − 2.5 at the femoral neck or the lumbar spine. The latter was estimated based on the published US osteoporosis prevalence rates of 3.9% and 15.8% in non-Hispanic White men and women [13]. In this study, reference groups used for the femoral neck and the lumbar spine were 20–29 year-old non-Hispanic White females from NHANES III and 30-year old White females from the DXA manufacturer database, respectively [13]. The three other scenarios for estimating the patient potential were defined based on FRAX risk thresholds by 5-year age groups for MOF in men and women. The FRAX algorithm allows for the country-specific calculation of the individual 10-year absolute risk of either hip fracture or MOF based on clinical risk factors with or without BMD. For the purpose of this analysis, the number of men and women aged 45 years or older with a FRAX risk for MOF exceeding 15% or 20% or exceeding the risk of a Swiss woman with a prior fracture was calculated based on previously published prevalence data used for the Swiss-specific FRAX model (https://​www.​sheffield.​ac.​uk/​FRAX/​) calibrated for Swiss-specific fracture risk and life expectancy [2, 14, 15]. It was assumed that fracture risk in the age group 45–49 years was identical with that in the age group 50–54 years. The thresholds of 15% and 20% were chosen because osteoporosis treatment had previously been shown to be cost-effective in the Swiss healthcare setting with a 10-year probability for a MOF at or above 15.1% (range 9.9 to 19.9%) and 13.8% (range 10.8 to 15.0%) in men and women, respectively [14]. The prior fracture risk equivalent threshold, i.e., the 10-year fracture probability for MOF exceeding that of a woman with a prior fragility fracture according to the FRAX algorithm calibrated for Switzerland, was chosen because of its general acceptance in the Swiss health economic context (almost all drugs are reimbursed in the presence of a positive fracture history) and in the latest recommendations for osteoporosis treatment from the Swiss Association Against Osteoporosis (SVGO/ASCO) [16]. Furthermore, earlier findings suggested that osteoporosis treatment was cost-effective or cost-saving in the Swiss setting after the age of 55 years in men and 60 in women who had previously sustained a fragility fracture [14, 15]. For each FRAX-based scenario, the age-specific proportion of men and women qualifying for the predefined fracture risk categories was applied to the annual structure of the Swiss population from year 1998 through year 2018.

The annual osteoporosis treatment gap was defined as the absolute or relative difference between the patient potential and the number of treated patients between 1998 and 2018.

The number of hospitalizations for hip fractures in men and women aged 45 years and older were extracted for the years 1998 through 2018 from the Swiss Federal Office of Statistics health database using the ICD-10 codes S72.0 Fracture of the femoral neck, S72.1 Pertrochanteric fracture, and S72.2 Subtrochanteric fracture) [3]. The crude incidence of hip fractures in men and women pooled was calculated per 100,000 person-years. In the absence of available covariates, the degree of association between the number of patients treated by year and the earlier reported decrease in hip fracture incidence [3] was explored by univariate regression analyses. It was considered that all hip fractures were hospitalized.

The present analysis represents a pragmatic and easily reproducible approach to the evaluation of the treatment gap. Key findings include the observation that the introduction of intravenous bisphosphonates (ibandronate and zoledronate) and of the RANKL inhibitor denosumab has not resulted in a further increase of the total number of patients treated. Of mirrored concern, the treatment gap, which had been constantly decreasing since 1998, reached its nadir in 2008 and steadily increased thereafter. Keeping in mind that the secular decrease in hip fracture incidence had already started in 1998 to 2001, i.e., at a time where the number of treated patients was low, there was a strong association, which does not imply causation, between the decrease in hip fracture incidence and the number of patients treated with a pharmacologically active osteoporosis drug.

Using a more comprehensive and therefore complex modeling approach aimed at estimating the clinical and economic burden of osteoporotic fractures in year 2010, the osteoporosis treatment gaps in men and women in Switzerland were estimated at 36 and 58%, respectively [17]. In the present analysis, the calculated treatment gap for both sexes taken together varied from 27 to 66% in year 2010 depending on the fracture risk threshold for intervention scenario considered as acceptable. Importantly, while all of these scenarios were shown to be cost-effective or even cost-sparing in the Swiss setting [14], none exactly reflects the current requirements for reimbursement of osteoporosis drugs in Switzerland. The latter stipulate cum grano salis that reimbursement is warranted in the presence of a T score at or below − 2.5 and/or a prevalent (fragility) fracture only. From a clinical perspective, treating patients having experienced one or more fragility fractures and thus all individuals with a 10-year fracture probability exceeding that of a woman with a prior fragility fracture may be more acceptable to physicians than a cost-effective intervention threshold expressed as percent risk [11, 16, 17]. However, such an approach may lead to treating a minority of patients at highest risk, while a majority of patients at lower but still increased risk may not be granted reimbursed access to treatment. In terms of total number of fractures, the end result would be that a large number of fractures would still occur in unprotected patients at risk and undermine achievable better public health results at the global population level. This triggers and further supports recent developments suggesting that a population-based screening for fracture risk in postmenopausal women based on FRAX should be considered, in addition to the currently implemented high-risk case finding and treatment approach, for incorporation in many healthcare systems to reduce the burden of fractures [18, 19].

The calculation of the number of patients on therapy should be considered with the caution relative to the underlying assumptions. The drug-specific adjustments for persistence were taken from a subset of Medicare patients with fee for service coverage who had stayed on therapy for at least three years [12]. Whether similar persistence rates apply to Switzerland is likely but not established. Whether these rates vary over time is also unknown. However, considering that persistence is generally better in patients treated with denosumab than with IV bisphosphonates and worst with oral bisphosphonates has been repeatedly shown in the literature [20‐23]. More important than the absolute values, the changes over time allow for internal consistency and acceptable comparability. It is remarkable that the introduction of intravenous bisphosphonates and denosumab has not contributed to a market expansion in terms of number of patients treated, although the pool of eligible patients has increased over time due to the rapid aging of the population. A possible explanation is that at the time of launch of the early bisphosphonates all efforts were concentrated towards case finding, i.e., identifying new patients eligible for treatment aimed at reducing fracture risk within the boundaries set by reimbursement restrictions. Ten years later, when ibandronate (oral in 2005, intravenous in 2006), zoledronate (2007), and denosumab (2010) were launched, an established market of almost 250,000 patients, already identified and treated with an oral bisphosphonate, existed. In such a situation, recommended marketing strategies aim at “picking the low hanging fruits” for fast market penetration, i.e., at switching patients on therapy to the new substance. Another explanation could be that the market was already saturated, i.e., that all eligible and accessible patients had already been treated which seems unlikely in the context of a growing patient potential. Finally, the number of newly identified patients starting on an osteoporosis drug and of those already diagnosed who resume treatment after an interruption may be equal to the number of patients stopping treatment. The concept of drug holiday in low-risk patients treated during 3 to 5 years with a bisphosphonate was introduced in 2011 based on a FDA recommendation [24] triggered by considerations related to bisphosphonate accumulation in bone [25], potentially sufficient residual effects on fracture risk after treatment discontinuation in low risk patients [26‐28], and safety concerns regarding rare reported of uncommon adverse events (osteonecrosis of the jaw and atypical femoral fractures) [29]. While holiday implies that treatment will be reinitiated and recommendations insisted on the importance of regular reassessments, many patients may have been lost to follow-up.

It was also intriguing to observe that the SERM raloxifene (reimbursed in Switzerland if the T score is at or below − 1) and the bone anabolic substance teriparatide (reimbursed as a second-line therapy after a new fracture occurring under treatment with a bisphosphonate or denosumab) were virtually inexistent in terms of patients treated, both representing less than 2% of the treatment-days in year 2018. According to T score-based US prevalence data, 15.8% of all White non-Hispanic women should have a T score at or below − 2.5 and 52.6% a T score at or below − 1 which corresponds to more than a tripling of the patient potential. Obviously, osteoporosis treatment is reserved to those at highest risk and little to no pharmacological intervention is the rule in patients with osteopenia. By contrast and consistent with the available published evidence, many patients with highest fracture risk (e.g., patients with multiple vertebral fractures) should have been eligible for treatment with the only bone anabolic substance available during the period of observation under scrutiny, namely, teriparatide [30]. Here again, available data suggest important underuse. Overall, these findings plead in favor of an urgent need for drug prescription data with higher granularity which would allow evaluating the treatment gap by risk categories and by sex.

Overall, the present findings suggest that the key challenges for overcoming the treatment gap published in a 2017 narrative global perspective on strategies for the prevention of fragility fractures need more urgent attention than ever [31]. Case finding and management of individuals at high risk of fracture should be revitalized, raising public awareness about osteoporosis and fragility fractures should be amplified or reinitiated, improving reimbursement and health system policies should ensure access to the most appropriate treatments to those patients expected to benefit most, and optimizing our epidemiological understanding of osteoporosis and fractures should include tools for regular progress assessment. For that, the newly proposed scorecard within the SCOPE project should be considered as an excellent starting point to be tailored to individual countries’ specific needs [11].

Among the strengths of this study is the straightforward and reproducible approach to the treatment gap calculation ensuring internal data consistency and allowing for monitoring changes over time. It also overcomes the challenges met when trying to estimate treated patient numbers based on DDDs (defined daily dosages). The latter are mandated by the WHO in order to ensure worldwide data comparability but sometimes suffer from important deviations when put into perspective with the approved dosing schemes. Limitations include that the IQVIA drug sales database does not report drug use by indication. While some of the drugs used for the treatment of osteoporosis have additional indications (such as the prevention of bone loss under antihormonal treatments or Paget’s disease of bone), the impact on the number treated patients can be considered as limited. The latter is further warranted by the existence of specific formulations for additional indications unrelated to osteoporosis, such as zoledronate 4 mg or denosumab 120 mg monthly, both for oncological indications, which were excluded from the present analysis. The IQVIA database does not report drug use by sex, such that an overall approach pooling men and women was chosen at the expense of a loss in data granularity. The IQVIA database does not either collect clinical characteristics which also preempted a more detailed analysis by gradients of fracture risk. Assumptions regarding persistence and DXA-based osteoporosis prevalence where taken from US publications which may differ from the Swiss reality, highlighting the urgent need for further epidemiological research at the individual country level. In the absence of better data, it was assumed that persistence, which addresses the question of how long a patient stays on therapy after treatment initiation, did not change between 1998 and 2018. Whether and to what extent prescription behavior and persistence have been impacted by real or suspected emerging safety concerns with antiresorptives, including the scientific discussions around atypical femoral fractures, osteonecrosis of the jaw, and atrial fibrillation events, remains unknown and deserves further research. On the other hand and complementary to that, adherence to the prescription, i.e., the extent to which a patient acts in accordance with the prescribed interval and dose of a dosing regimen, would be important to know and adjust for, especially for oral drugs. Both adherence and persistence have been shown important for optimizing fracture outcomes [32, 33]. Finally, we did not retain hormone replacement therapy (HRT) in our selection of bone active substances, albeit it was shown to significantly reduce hip fracture risk by approximately one-third in the Women’s Health Initiative (WHI) trials [34, 35]. The WHI trials also showed a significant increase in breast cancers, cardiovascular diseases, and all-cause mortality which led to drastic reductions in systemic HRT use since 2002, now being mainly reserved to younger postmenopausal women with vasomotor symptoms [36, 37]. As the effect of HRT on hip fracture risk was shown having vanished three years after discontinuation [38], we consider that omitting HRT from our analyses should not be expected to alter the present observations and conclusions. Whether and to what extent changes in the HRT prescription patterns due to safety concerns that emerged during the observation period covered by the present analyses may have contributed to the observed slowing of the secular decrease in hip fracture incidence deserves further research.

Overall, the present study suggests that although multiple treatment options were made available over the past two decades, the treatment gap remains large and has even increased over the past decade. Case finding strategies which may have been neglected in the more recent past should be reassessed. Future research should focus on establishing instruments for measuring progresses made, including on more detailed datasets reporting drug usage by sex and indication.