Backcasting your fitness from 85: what’s evidence and what’s judgment

Attia’s framework starts at the end of life and works backwards: the things you want to be able to do at 100, what that requires at 80, at 60, and today. He groups the work into four pillars: stability, strength, aerobic performance and anaerobic output.

His example tasks include getting up off the floor with a single point of support, picking up a 30 lb child from a squat, lifting 30 lb overhead into a bin, getting out of a pool without a ladder, and walking up three flights of stairs with 10 lb of groceries in each hand.

He says it makes more sense to focus on the framework than on his own list, because people differ. In a 2023 podcast he defines "the marginal decade" as the last decade of your life: the decade the age-85 goals are meant to protect. The framework itself has not been tested, and his tasks and numbers are his own judgment.

The best support for the idea is about thresholds, not Attia’s list. A 1997 review by Young described ageing as eroding physiological safety margins, with lower-limb explosive power declining towards the thresholds for impaired mobility.

In the Honolulu Heart Program, 6,089 men aged 45–68 had their grip measured and were assessed again 25 years later. Those in the lowest third for grip had 2.87 times the odds of walking at 0.4 m/s or slower, compared with the highest third. The authors describe midlife strength as giving "a greater safety margin above the threshold of disability". It was men only, one ethnic group, and observational.

A study of 409 adults aged 60–96 found the link between leg strength and gait speed was not a straight line: in stronger people, strength was unrelated to walking speed; in weaker people, it was related. Small changes in capacity matter a lot near the threshold and little above it.

Together these make the case for a line rather than a pass mark: what matters is how far above the threshold you are when the decline of the last decade begins. None of these studies sets a numeric target for any test.

In 100 healthy people aged 65–89, differences between age groups equalled losses of about 3.5% a year in leg extensor power against 1–2% a year for strength. In a Copenhagen cohort of 1,305 adults, relative leg power began to decline above 40. Both are cross-sectional, comparing different people at different ages. The Baltimore Longitudinal Study of Aging, which did follow the same people, found arm strength and power declining from about 40, with power falling about 10% more than strength in men.

For backcasting, the point is that one rate of decline will not fit every test. That is why the app’s line uses each test’s own norms at each age rather than a single rate.

Simple physical tests carry information about survival. Pooling cohorts with an average age over 70, the slowest quarter on chair rises had about twice the mortality of the fastest (hazard ratio 1.96: dying at about twice the rate at any point during follow-up), the slowest walkers 2.87 times and the weakest grip 1.67 times. In 34,485 adults aged 65 and over, each 0.1 m/s faster walking speed went with a 12% lower hazard of death.

All of this is observational, and in older adults. It supports measuring what the body can do. It does not show that changing a score changes a lifespan.

You choose goals for 85 from a list of sixteen. Each maps to one of the ten tests and has a target at 85 for men and for women. Where two chosen goals use the same test, the more demanding one sets the line.

The app then finds where that target sits among 85-year-olds in the test’s norm table, for example the 60th percentile. The line at every earlier age is the value at that same percentile for that age. If your goal needs you at the 60th percentile at 85, the line asks you to be at the 60th percentile of your age group now. Zone 2 is the exception: its line is a flat weekly target. Sit-to-stand has no table under 60, so below 60 its line holds the value it has at 60.

Holding your rank is a modelling choice, not a finding. The age differences in the norm tables are cross-sectional, comparing different people of different ages, and we found no study that tracks sit-and-rise, eyes-closed balance or sit-to-stand in the same people from midlife to 85. If people naturally slip down the rankings with age, the line is too gentle; if healthy people tend to hold their rank, it is about right. We do not know which.

You can see the line for a goal and your age in the calculator. It is a planning tool, not a prediction, and not medical advice.

Of the sixteen targets, nine sit near a published number and seven do not. "Near a published number" means a threshold or reference is close by. It does not mean anyone has validated that number for the task.

• Get out of a low chair without hands: five sit-to-stands in 12 s (men) or 12.5 s (women), faster than the 14.8 s reference for people in their 80s and under the 15 s fall and sarcopenia cut-offs. The sex split has no source.
• Get up off the floor with one hand: a sit-and-rise score of 8, the bottom of the lowest-risk 2012 band. In the 2025 cohort an 8 still had three times the death rate of a 10.
• Get down on the floor to play and back up: a score of 6, the bottom of the 6–7.5 band.
• Open a new jar: grip 26 kg (men), exactly the FNIH weakness cut-off, or 18 kg (women), just above its 16 kg.
• Pick up a 15 kg child from a squat: grip 30 kg (men) or 20 kg (women), above the FNIH and EWGSOP2 cut-offs. The task is mostly legs and back; grip is a proxy.
• Hike 5 km with 300 m of climb (VO₂max 30 / 26 mL/kg/min) and walk 1 km briskly (24 / 21): both above the estimated minimum for independent living at 85 of about 18 (men) and 15 (women). The demand of each walk was not derived from data.
• Put on trousers standing up: 3 s on one leg, eyes closed, above Springer’s 1.9 s average at 80–99. Dressing is done with eyes open, so the mapping is judgment.
• Walk 30 minutes a day with grandchildren: 150 minutes a week, the lower end of the WHO range. Treating it as zone 2 is our choice.

Seven targets have no published anchor: catching a rail (dead hang), the two carries, the two push-up goals (overhead bin, climbing out of a pool), four loaded flights of stairs, and putting on socks on one leg (5 s, eyes closed). There are no peer-reviewed adult norms for the dead hang, and we found no study linking hang time, push-ups or carry distance to these tasks. The socks target is probably too demanding as an eyes-closed time, and because the more demanding goal sets the line, choosing it over the trousers goal changes your balance line a lot.

The honest summary: the method of working backwards rests on reasonable evidence about thresholds and decline. The numbers at the end of it are ours, and each test page says, goal by goal, which have an anchor.