Participants
Prevalence figures were calculated by the EWGSOP2 definition for sarcopenia, combined with any of three common definitions for obesity (see Table 1).
To define sarcopenia the updated EWGSOP2 definition was used; i.e. reduced chair-stand capacity (time to perform five repeated chair stands >15 seconds) or reduced hand grip strength (<16 kg for women and <27 kg for men), in combination with low muscle mass. Probable or severe sarcopenia (ref) was not considered for the SO definition.
In the H70 cohorts, all participants were born in 1930 and data on sarcopenia, obesity, mortality and related covariates for a total of 521 individuals (n=319 women and n=202 men) were collected from the examinations conducted in 2005 when participants were 75 years old (defined as baseline in this study). In the ULSAM cohort, data from 288 community-dwelling men aged 87 years were collected in 2008-2009 (defined as baseline in this study).
Definitions and cut-offs for sarcopenia, obesity and sarcopenic obesity
Prevalence figures were calculated by using the EWGSOP2 definition for sarcopenia, combined with three different definitions for obesity (Table 1).
To define sarcopenia the updated EWGSOP2 definition was used. This definition advocates the use of reduced chair-stand capacity (time to perform five repeated chair stands >15 seconds) or reduced hand grip strength (<16 kg for women and <27 kg for men) in combination with low muscle mass.
In the H70 Study, bioelectrical impedance spectroscopy (BIS, see below) was used to calculate skeletal muscle mass index (SMI). No cut-offs for SMI are proposed in the EWGSOP2, which is why we chose to use the cut-offs from Janssen et al. for H70 (as in EWGSOP [2] ≤5.75 kg/m2 for women and ≤8.5 kg/m2 for men [20]). In the ULSAM cohort, muscle mass was measured by Dual-energy X-ray absorptiometry (DXA) and the EWGSOP2 recommended cut-off for appendicular skeletal muscle index (ASMI) of <7 kg/m2 was used.
To define obesity, any of three measures of obesity was used, i.e., BMI ≥30 kg/m2, fat mass >42% (women) and >30% (men), or waist circumference ≥88 cm (women) and ≥102 cm (men) [6, 21, 22]. If any of the obesity criteria were fulfilled, the individual was defined as having obesity. Individuals defined as having sarcopenia according to EWGSOP2 and concurrent obesity, by any of the definitions, were considered having sarcopenic obesity. SO defined by sarcopenia and elevated fat mass only was used for sensitivity analyses (see below).
Measurements
Body composition: Body composition was measured by BIS using Xitron Hydra 4200 devices (Xitron technologies, San Diego, USA) in the H70 cohorts. Skeletal muscle mass (SMM) from BIS was estimated using the equation (Total Body Skeletal Muscle Mass, no Body weight (TBSMMnoBW) = -24.05 + (0.365*height)+(-0.005*Ri)+(-0.012*Re)+(-1.337*gender)( Ri and Re=Intra- and extracellular resistance)) developed and validated by Tengvall et al. [23]. Skeletal muscle index (SMI) was calculated as the ratio of SMM to height in meters squared.
In the ULSAM cohort, DXA (DPX Prodigy, Lunar corp., Madison, WI, USA) was performed and ASMI was calculated using total muscle mass from arms and legs divided by height in meters squared.
Strength and function: Grip strength were measured using a Jamar dynamometer in H70 and the Baseline hydraulic hand dynamometer in the ULSAM cohort. The highest value from the strongest hand was used in the analyses, and the thresholds were 16 kg and 27 kg for women and men, respectively. To measure leg strength, the participants were asked (both in H70 and in ULSAM) to perform five repeated chair stands without using their hands. The threshold value for reduced strength was >15 seconds [5]. Gait speed was measured for 30 meters indoors at a spontaneously chosen speed in H70. In ULSAM the course was 10 meters and the middle six meters were marked and registered.
Co-variates: In the regression analyses of body composition phenotypes as exposure for mortality various sets of co-variates were accessible for the two cohorts. In the analyses of the H70 women and men, adjustment was performed for comorbidities and smoking (number of cigarettes/day). Corresponding mortality analyses in the ULSAM male cohort were adjusted for age, comorbidities, education, exercise, living conditions (living alone: yes/no) and smoking (current smoker or non-smoker). When adjusting for co-morbidities, the un-weighted Charlson Comorbidity Index was used in both cohorts. The index was based on in-patient diagnoses (ICD9 - ICD10) in the patient register before the dates of the examinations [24, 25]. In the ULSAM cohort education was assessed by number of years in school divided into categories (7, 8 or 12 years), college education, or graduate exam. Regular exercise was defined as doing sports/heavy gardening more than three hours per week.
Statistical analyses
All values are presented as means± SD, median or percentage, as appropriate. In the survival analyses, the cohorts were divided into four groups based on body phenotype: sarcopenic obesity, sarcopenia (without obesity), obesity (without sarcopenia), and no sarcopenia or obesity (i.e. “normal” phenotype) as indicated above. In the analysis of the potential association between SO and all-cause mortality, we examined the 10-year survival in the 75 year old participants of the H70 cohorts (depending on date of examination, maximum years at risk was 9.7) and 4-year survival in the 87 year old participants of the ULSAM cohort (maximum years at risk 4.0). Ten- and four-year observation periods were chosen due to differences in expected survival time in the two cohorts. Analyses were executed using the log-rank test, the Kaplan-Meier survival curve and the Cox proportional hazard model. The Cox regression analyses were presented as hazard ratios with 95% confidence intervals. A p-value of <0.05 was considered statistically significant. Relevant multivariable co-variates for the associations of interest were included in the models. When finding the best fitting model, a likelihood ratio test was performed and a test for proportional hazard assumption including plots of Schoenfeld residuals. All analyses were conducted using STATA15 [26].
Sensitivity analyses: In sensitivity analyses (Cox regression for survival), we investigated if the results from the main analyses would remain, both in H70 and in ULSAM, when using only high body fat mass to define obesity (in combination with EWGSOP2 definition for sarcopenia to define SO). Furthermore, we performed sensitivity analyses where the mortality for the women with obesity (no sarcopenia) defined as BMI ≥30 kg/m2 was compared to the mortality for the group with no sarcopenia or obesity, and where women with obesity by any of the definitions (irrespective of sarcopenia) were compared to women without obesity.
Also, analyses were performed were individuals within the H70 cohort who had passed away within a year after the examination (2005-2006) were excluded.
The exercise-related co-variate in H70, “spare time activity during the last year”, was missing for almost half of the H70 sample. For this reason, complementary sensitivity analyses were performed by adding this co-variate in models that only included individuals with this data available. In the ULSAM cohort, mortality was also compared between the group with obesity (without sarcopenia) defined as waist circumference ≥102 cm and those with no sarcopenia or obesity, and between the group with obesity by any definition (irrespective of sarcopenia) and the group without obesity (irrespective of sarcopenia).