Real-World Osteoporosis Treatment Gap and Costs in Spain: Data from Women with a First Fragility Fracture or Diagnosis of Postmenopausal Osteoporosis

This retrospective, observational study used electronic medical records from the Spanish BIG-PAC^® administrative database (data source: secondary; proprietor: Atrys Health) [17]. This database includes medical records from seven integrated public health areas (primary care centres and hospitals) across seven autonomous communities covering 1.9 million patients and is representative of the national population [18, 19].

A study schema is shown in Fig. 1i. The study included women aged ≥ 50 with an index event of first FF or a PMO diagnosis between January 2014 and December 2018 (eligibility period) and at least 12 months of data available before their index event. Eligible women were observed for 2 years following their index event or until death. Women who experienced their first FF during the eligibility period, with or without a PMO diagnosis at the time of the FF or during the subsequent 2-year observation period, were assigned to cohort 1 (FF cohort). Women diagnosed with PMO with or without a subsequent FF were assigned to cohort 2 (PMO diagnosis cohort). Women with a PMO diagnosis before their index event of FF or with a FF before their index event of PMO diagnosis were excluded from cohorts 1 and 2, respectively. Women with a first FF and PMO diagnosis on the same date were assigned to cohort 1. Inclusion and exclusion criteria can be found in Supplementary Figure S1.

PMO was diagnosed and FF were identified using the International Classification of Diseases Clinical Modification (tenth edition; ICD-10-CM) (Supplementary Table S1). OP and concomitant treatments were identified using the ATC classification system (Supplementary Table S2) [20]. Fractures due to high or moderate trauma (e.g., an automobile accident) and other fractures considered unlikely to be related to OP were excluded (Supplementary Table S3).

In compliance with national legislation, all data were anonymised at the centre of origin prior to exporting to BIG-PAC^® [21]. All procedures performed in studies involving human participants were in accordance with the ethical standards of the research committee Comité de Ética de Investigación con Medicamentos (CEIm) Consorci Sanitari de Terrassa and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. For this type of study, formal consent is not required.

The primary study objective was to report the proportion of women not prescribed OP medication in the first 6 months after the index FF or PMO diagnosis (treatment gap). Secondary objectives were to describe the patients’ clinical outcomes, the proportion of patients diagnosed with PMO after an FF, healthcare resource use (HCRU), and direct medical costs over the 2-year observation period post-index event.

The primary study outcome was the OP treatment gap in each cohort; i.e. the percentage of women who were not prescribed OP treatment 6 months after their index event. Further analyses were performed on the population treated within 6 months after their index event (percentage of fracture site, patients who had received prior OP treatment, and percentage of patients with OP treatment according to their fracture site).

The following outcomes were summarised for patients who were prescribed OP treatment in the 24 months after the index event:

The following outcomes were summarised for every patient: patient demographics and clinical characteristics at the index event, specifically age, comorbidities (Supplementary Table S4), Charlson comorbidity index (CCI) score (Supplementary Table S5); concomitant treatments (Table S2); and polymedication (≥ 5 drugs per day). The site of the index fracture was summarised for cohort 1 only.

The following clinical outcomes were assessed during the 2-year observation period: rates of new FF (defined as occurring at a different site from the previous FF or more than 90 days after a previous FF at the same site) and mortality (both calculated as events per 1000 patient-years), annual mortality (events per 1000 patients for each year of the observation period), and HCRU and medical costs related to the clinical management of FF and PMO.

Direct medical costs were defined as costs related to healthcare services associated with managing PMO or FF and performed by healthcare professionals. Indirect or non-medical costs were calculated as the daily wage lost due to absence from work. Days of work loss were found within the BIG-PAC^® database. Total (direct and indirect) costs were calculated using the number of resources/services used and the unit price of each resource/service according to the diagnosis-related group (DRG) national tariff (Supplementary Table S6) and summarised as the mean cost per patient during the observation period (the overall follow-up for each patient was considered). Hospitalisation costs were estimated using the DRG national tariff; the cost of drugs and outpatient care were estimated using national reimbursement tariffs. Medication costs were calculated using the retail price per container when dispensing from price lists of the General Council of Associations of Official Pharmacists of Spain. The unit costs for each resource (i.e. hospitalisations, drug treatments, and diagnostic evaluations) were determined for the entire observation period using the information gathered by the National Statistical Institute (https://​www.​ine.​es). A summary of outcomes is shown in Fig. 1ii.

All analyses were descriptive in nature; study outcomes were described using summary statistics. Categorical variables were summarised using count, percentage and 95% confidence interval (95% CI). Quantitative variables were summarised using the mean and standard deviation (SD), the median and interquartile range (IQR; 25th [Q1] and 75th [Q3] percentiles), the minimum, and the maximum. Missing data were not imputed. Event rates and incidence rates were calculated by taking the sum of events that each woman had during their follow-up (variable in each patient) and dividing it by the sum of days of follow-up. This was then divided by 365 days and multiplied by 1000. For the HCRU cost correction, an analysis of covariance (ANCOVA; generalised linear model) was conducted with gender, age, and CCI as covariates, and the 95% CI was calculated. Data management and statistical analyses were performed using Microsoft Access, SQL software, and SPSS version 25.

From 2,255,224 patients in the BIG-PAC^® database, a total of 22,142 women met the study eligibility criteria: 3190 with an index event of a first FF (cohort 1) and 18,952 with an index event of PMO diagnosis (cohort 2). The attrition table is presented in Fig. 2.

Patient characteristics for both cohorts are presented in Table 1 and Supplementary Table S7. Women in cohort 1 were older than those in cohort 2 (mean [SD] age, 75.4 [10.2] vs 71.8 [10.8] years) and had a higher comorbidity burden (mean [SD] CCI score 1 [1.3] vs 0.6 [1.1]) (Table 1). Across both cohorts, other inflammatory arthropathies (than rheumatoid arthritis and osteoarthritis) and hypertension were the most common comorbidities (Supplementary Table S7).

In cohort 1 (index event of a first FF), the most common index FF was vertebral, followed by hip/femur and forearm/wrist/radius fractures (30.8%, 21.2%, and 16.1%, respectively) (Table 1). Over half (59.2%) of the women in cohort 1 were subsequently diagnosed with PMO. Diagnosis frequencies increased with age (from 37.4% in women aged 50‒59 years to 74.4% in women aged ≥ 80 years) (Fig. 3i). They were highest among women with vertebral fractures (83.9%), followed by those with hip/femur and forearm/wrist/radius fractures (78.1% and 62.4%, respectively) (Fig. 3ii). The percentage of fractures was similar across age groups (Fig. 3iii). Traumatologists, followed by general practitioners, diagnosed FF most frequently (Fig. 3iv).

Most (2821/3190 [88.4%]) women in cohort 1 experienced at least one subsequent FF in the 2 years after the index event, compared with only 6.5% (1238/18,952) of those in cohort 2. Fracture incidence increased with age in both cohorts (Supplementary Fig. S3i), with vertebral fractures being the most common fracture type (27.9%), followed by hip/femur fractures (24.6%) (Supplementary Fig. S3ii). All-cause mortality incidence was higher in cohort 1 vs cohort 2 (4.9% vs 2.3%) (Supplementary Table S8).

Fracture and mortality event rates for the 2-year observation period are shown in Fig. 6. Overall, fracture rates were higher in women not prescribed OP treatment in the 6 months after the index date and those who were non-persistent with and non-adherent to treatment at 24 and 12 months, respectively, vs women who were prescribed OP treatment in the 6 months post-index, and those who were persistent with and adherent to treatment. Mortality rates were higher in non-persistent vs persistent patients and were similar among treated and untreated and adherent and non-adherent patients. Annual fractures and mortality rates increased with age in both cohorts (Fig. 7). Overall, vertebral fractures had the highest rate; rates of forearm/wrist/radius fractures were higher in cohort 1 vs cohort 2, and rates of hip/femur fractures were higher in cohort 2 vs cohort 1 (Fig. 7).

Table 3 summarises HCRU, sick leave, and associated direct and indirect costs during the 2-year observation period. HCRU was higher in cohort 1 (index event of a first FF) vs cohort 2 (index event of PMO diagnosis), with the largest difference observed for hospital admissions (57.7% of women in cohort 1 had at least one hospital admission vs 0.8% in cohort 2). The percentage of working-age patients taking at least 1 day of sick leave was also higher in cohort 1 (11.0%) vs cohort 2 (4.9%), with a mean (SD) of 6.8 (24.9) vs 2.3 (17.1) days off work, respectively. The mean (SD) total cost per patient was €10,601 (13,952) in cohort 1 and €1659 (2436) in cohort 2. After adjusting for age, gender, and CCI, the mean (95% CI) cost per patient was €10,446 (10,245–10,646) in cohort 1 and €1685 (1603–1766) in cohort 2.

Costs increased with age and were higher in cohort 1 vs cohort 2 across all age groups. Hip fractures had the highest mean (SD) cost per patient (€15,833 [14,666]), followed by vertebral fractures (€10,593 [14,326]) (Supplementary Table S9). Across both cohorts, costs were higher for non-adherent and non-persistent patients and switchers vs adherent and persistent patients and those not switching treatment (Supplementary Table S9).