Study Population
This study utilizes data from the Health 2000 survey, a comprehensive health examination survey conducted across Finland in 2000-2001. A random sample of 10,000 individuals aged 18 and above was drawn from the national population register, employing a stratified two-stage cluster sampling method. The survey included both community-dwelling and institutionalized individuals residing in mainland Finland. Detailed descriptions of the survey's methodology have been published previously [13]. In total, 8,028 participants took part in the survey, with 3,439 of them being 55 years or older. For this study, we focused on participants aged 55 and above who had available data on sarcopenia and osteoporosis, resulting in a sample size of 2,142 (62.3%).
Measurement of probable sarcopenia, osteoporosis and osteosarcopenia
Grip strength was assessed using an electronic device (Good Strength, IGS01, Metitur Oy, Finland) with participants seated and their elbows resting on a table while holding the device's handle. [13, 14] The measurement was conducted twice, with a 30-second interval between attempts. If the difference between the two measurements exceeded 10%, a third measurement was taken. The highest value recorded was used for analysis. Probable sarcopenia was identified based on the EWGSOP2 criteria, with low grip strength defined as less than 27 kg for men and less than 16 kg for women. [4]
Bone mineral density was evaluated using a calcaneal ultrasound device (Sahara Clinical Bone Sonometer, Hologic, Waltham, Massachusetts). The Quantitative Ultrasound Index (QUI), provided by the manufacturer, served as the indicator of bone mineral density. QUI was calculated from the speed of sound (SOS) and broadband ultrasound attenuation (BUA) using formula:

Osteoporosis was defined as a bone density measurement with a T-score less than -2.5, based on ultrasound results. The reference group for the T-score consisted of 30–35-year-old women without chronic illness or disability (n=300). [1] Additionally, participants reporting a prior diagnosis of osteoporosis confirmed by DXA (dual-energy X-ray absorptiometry) were classified as having osteoporosis, regardless of their ultrasound-based bone density measurement. In total, 163 participants (8%) self-reported a prior osteoporosis diagnosis.
Osteosarcopenia was defined as having both probable sarcopenia and osteoporosis. Study participants were assigned into four groups: no sarcopenia and no osteoporosis, probable sarcopenia only, osteoporosis only, and osteosarcopenia.
Measurement of bone fractures
The follow-up information about bone fractures was obtained from the National Hospital Discharge Register by using the national identification numbers assigned to each Finnish resident.
Fractures were identified by the ICD-10 codes corresponding to the care event and three different fracture outcomes were defined: any fracture, major osteoporotic fractures and hip fractures. The following ICD-10 codes and all their sub-codes were included in the ‘any fracture’ outcome: S02, S12, S22, S32, S42, S52, S62, S72, S82, and S92. Major osteoporotic fractures, as defined by the WHO, included hip (S72), clinical spine (S12, S22, S32), shoulder (S42), and wrist fractures (S52). To comply with the WHO’s definition of Major osteoporotic fracture, exclusions were made for non-vertebral thoracic and pelvic fractures (S22.2-S22.9, S32.1-S32.8), non-wrist and non-shoulder upper extremity fractures (S52.0-S52.4, S52.7-S52.9) and diaphyseal and distal femur fractures (S72.3, S72.4, and S72.7-S72.9). [15] Hip fractures included events coded as S72.0, S72.1 and S72.2. Because high-energy impacts may cause fractures regardless of bone health or sarcopenia status, we excluded all fractures resulting from a high-energy impact (traffic accidents, falls down stairs or ladders, injuries from motorized machines).
To exclude participants with earlier bone fractures for sensitivity analysis purposes, information on fractures preceding the enrollment for the study was also gathered from the National Hospital Discharge Register. In total, the data on fractures spanned from November 21st, 1994, to December 31st, 2019.
The follow-up for each fracture outcome (any fracture, major osteoporotic fracture, and hip fracture) continued until the occurrence of the specific fracture type being analyzed, the date of death, or the end of the study period (December 31st, 2019), whichever came first. Participants experiencing a different fracture type remained in the study for the outcome of interest.
Mortality ascertainment
Mortality was followed until the date of death or end of follow-up, i.e. 31st December 2019. The Health 2000 dataset was linked to Statistics Finland’s Causes of Death Register, which includes information on the date and cause of death, by utilizing the personal identity codes assigned to each Finnish resident.
Demographic and lifestyle-related covariates
Information on age and sex were obtained from the population register. Education, smoking, physical activity and mobility limitations were obtained from survey questionnaires. Education was categorized as basic, secondary or higher. Smoking was categorized dichotomously as current and ex-smokers, or never smokers. Physical activity was categorized either exercise training (“Leisure time includes strenuous physical exercise at least 3 hours per week”), active (“Leisure time includes walking, bicycling and other forms of physical activity at least 4 hours per week”) or inactive. (“Leisure time consists of reading, television, or activities not involving physical activity”)[16] To measure mobility, subjects were asked “Are you able to walk about half a kilometre without resting?” and “Are you able to climb one flight of stairs without resting?”. Any difficulty in either task indicated a mobility limitation.
Statistical analysis
Comparisons between osteosarcopenia groups at the baseline were performed using Student’s t test for continuous variables and chi-square test for categorical variables.
To examine the association between osteosarcopenia groups and fracture risk and mortality, we used three different analytic approaches. First, we conducted unadjusted survival analyses using Kaplan-Meier estimates with 95% confidence intervals (PROC LIFETEST in SAS 9.4). Second, we performed multivariable survival analyses using the Cox proportional hazards model (PROC PHREG in SAS 9.4) to estimate adjusted hazard ratios (HR) with 95 % confidence intervals (CI). Initially we adjusted the analyses for age and sex (Model 1). In Model 2 we additionally adjusted for education, smoking, and physical activity; and in Model 3 we additionally adjusted for mobility limitation. Finally, in Model 4, to further account for death as a competing risk for fractures, we utilized the Fine-Gray method incorporating covariates from Model 3. [17] We tested the proportional hazards assumption using Schoenfeld residuals.
We also conducted several sensitivity analysis to test robustness of our findings. First, to eliminate the risk of previous fractures being identified as new incidents, we excluded all subjects with previous fracture of the respective type (e.g. excluded all those with a previous hip fracture from the hip fracture hazard analysis). Second, we included only fractures resulting from low-energy impacts (falls on the same level, falls from bed, falls due to ice and snow) to capture only fractures related to fragility. Finally, we limited the follow-up to December 31st, 2010, i.e. about 10 years, to minimize the bias related to changing health status and especially changes in the exposure variables. For all sensitivity analyses we used the Fine-Gray method for fractures, and Cox Proportional Hazards analysis for mortality, both with adjustment Model 3.
All analyses were conducted using SAS software version 9.4 (SAS Institute Inc., Cary, North Carolina, United States). The analysis code used for this study is available in the supplementary materials.