*2.2. Participants*

Individuals aged 60 years and over were included in this analysis, but were excluded if (1) their weight exceeded the limit of the dual energy X-ray absorptiometry (DXA) scanners (≥120 kg), (2) a limb was affected by a prosthesis, plates or screws or had been amputated, (3) a full body scan, and/or a Timed-Up-and-Go (TUG) test was not performed, (4) a food

frequency questionnaire (FFQ) was not completed or (5) excessively high or low daily nutritional energy intakes were reported on the FFQ (i.e., <3360 or >16,800 kJ/day for men and <2100 or >14,700 kJ/day for women) [27].

#### *2.3. Outcome Measures*

#### 2.3.1. Muscle Mass and Muscle Function

As a surrogate for skeletal muscle mass, lean mass was measured by whole body DXA, which is the preferred method for assessing body composition in a research setting [28]. Appendicular lean mass (ALM, kg) was calculated as the sum of the lean mass measurements for arms and legs, expressed relative to height squared (ALM/height2, kg/m2).

A Lunar DPX-L (Lunar; Madison, WI, USA) was used to scan the first 544 men at baseline until an upgrade to a GE-Prodigy (Prodigy; GE Lunar, Madison, WI, USA). Crosscalibration was performed on 40 subjects aged 21 to 82 years to ensure comparability of the DXA scanners; no differences were detected in lumbar spine or femoral neck bone mineral density [26]. All scans for the women at 15-year assessment were performed on the GE Lunar Prodigy. The DXA scanner was calibrated three times per week with an anthropometric phantom (Hologic) to preserve the repeatability and accuracy of measures. Muscle function was assessed using a timed "Up-and-Go" (TUG) test, which measures the time taken to rise from a seated position in a chair with no arm rests, walk 3 metres, turn around, walk back and sit down [29].

In this study, a combination of low muscle mass and low muscle function was used as a representation of sarcopenia [28]. Low muscle mass was defined as ALM/height<sup>2</sup> < 7.87 kg/m<sup>2</sup> for men and <6.07 kg/m<sup>2</sup> for women (equal to T-score < −1) [30]. Cut points for ALM/height<sup>2</sup> were calculated using DXA from a sample of 374 men and 308 women aged 20–39 years from the GOS [30]. As suggested by the European Working Group on Sarcopenia in Older People (EWGSOP), low muscle function can be defined either as low muscle strength or low physical performance [28]. In this study, low muscle function was defined as TUG > 10 s for 3 metres [29]; the TUG is a recognised assessment tool for physical performance [28,31]. Measures of handgrip strength, used to assess low muscle strength, were not available for the recently updated definition of sarcopenia (EWGSOP2) [31].

#### 2.3.2. Exposure: Dietary Inflammatory Index (DII)

Dietary data were collected using the Dietary Questionnaire for Epidemiological Studies (DQES version 2), an FFQ created by Cancer Council Victoria, which was completed by participants at each assessment phase [32]. In this study, the baseline timepoint was used to assess diet from the FFQ for men and the 15-year timepoint for women. The FFQ DQES was designed for use in epidemiological studies and has been validated for the Australian population [33,34]; it captures usual eating habits over the past 12 months covering five types of dietary intake, incorporating 80 items: (1) cereal foods, sweets and snacks, (2) dairy products, meats and fish, (3) fruit, (4) vegetables, and (5) alcoholic beverages on a ten-point frequency scale. Portion sizes are based on dietary data collected on older Australian residents (mean age 61 years), which matches the sample used in our analyses [32]. Analysis of questionnaires for assessment of dietary intakes was undertaken by the Nutritional Assessment Office, Cancer Council Victoria. The output of the FFQ analysis provided estimated intakes of macronutrients and a range of micronutrients which were used to compute DII scores for all participants.

The DII is based upon up to 45 food parameters which have been scored based on reported pro-inflammatory or anti-inflammatory effects on specific inflammatory markers (IL-1β, IL-4, IL-6, IL-10, TNFα, and CRP) using 1943 peer-reviewed articles published through to December 2010. Details of the development of the DII have been reported elsewhere [16,35] and validation work using inflammatory biomarkers are also available [35–39]. Briefly, the scoring algorithm uses a global reference database (food consumption from eleven populations globally) and food parameter-specific inflammatory effect scores to create an overall DII score for an individual. The DII scores individuals' diets on a continuum from strongly anti-inflammatory ( −8.87) to strongly pro-inflammatory (+7.98).

To calculate DII scores for the participants in this study, dietary intake data were used to calculate an individuals' intake of food parameters which were then compared to the global reference database. A Z-score for each of the food parameters for each participant was calculated based on the global mean and standard deviation; this was achieved by subtracting the global mean from the amount reported and dividing this value by the standard deviation. The Z-scores were converted to a proportion to minimise the effects of outliers ("right-skewing"). The standardised dietary intake data (proportion) was centred by doubling and subtracting 1 and then multiplied by the inflammatory effect score of each food parameter and summed to obtain an overall DII score for every participant in the study. In this study, a total of 22 of 45 food parameters were available from the FFQ for computing the overall DII scores. These included energy, carbohydrate, protein, total fat, fibre, cholesterol, saturated fat, monounsaturated fat, polyunsaturated fat, omega-3 fatty acids, omega-6 fatty acids, niacin, thiamine, riboflavin, iron, magnesium, zinc, vitamin C, vitamin E, folic acid, beta-carotene and alcohol.
