REF-1139 · Bloch-Ibenfeldt et al., 2024 · BMJ Open Sport & Exercise Medicine · Paper · PubMed · Free full text
Heavy resistance training at retirement age: one supervised year of heavy lifting kept isometric leg strength at its baseline three years after the programme ended, while controls lost 7%
But leg power, grip strength and leg lean mass fell in every group alike
In the LISA trial, 451 older adults at retirement age were randomised to a year of supervised heavy resistance training (HRT: machines, three times a week, 3 sets of 6-12 reps at ~70-85% of 1RM), moderate-intensity training (MIT: body-weight and band circuits, 3 sets of 10-18 reps at ~50-60% 1RM, once a week at the hospital and twice at home) or a non-exercising control group (CON).
After the year nobody was coached; testing continued at 2 and 4 years from the start.
At year 4 (369 participants, mean age 71, 61% women), isometric quadriceps torque showed a group × time interaction (P < 0.001): HRT was unchanged from baseline (149.7 → 151.5 Nm, P = 1.00) while CON fell (144.4 → 134.5 Nm, P < 0.001; −7%, our calculation) and MIT fell non-significantly (147.6 → 142.9 Nm, P = 0.28); HRT's 4-year change was larger than both other groups'.
HRT also kept total lean body mass (47.5 → 47.3 kg) while MIT and CON lost a little, and visceral fat rose only in CON. But leg extensor power (the primary outcome), handgrip strength and lean LEG mass fell over 4 years in all three groups with no difference between them, and vastus lateralis cross-sectional area fell too (its group effect came from years 1-2).
The authors conclude that a year of heavy lifting at retirement age "may induce long-lasting beneficial effects by preserving muscle function".
Protocol
Participants: 451 home-dwelling adults at retirement age, not doing strenuous resistance training, stratified by sex, BMI and chair-rise performance and randomised to HRT (n = 149), MIT (n = 154) or CON (n = 148). Among the 369 tested at year 4: baseline age 66.4 ± 2.5 y (range 64-75 at year 4), 39% men, ~9 500 steps a day at baseline and at year 4.
HRT: a supervised full-body programme in a commercial gym three times a week for one year, with 6-8 weeks of habituation; periodised, machine-based, nine exercises, 3 sets of 6-12 reps at ~70-85% of 1RM (1RM estimated by the Brzycki equation).
MIT: circuit training with body weight and resistance bands, once a week at the hospital and twice at home, mimicking the HRT exercises at 3 sets of 10-18 reps, ~50-60% 1RM.
CON: asked to keep habitual activity, invited to cultural and social activities.
After year 1: no intervention and no health advice; participants knew the follow-up timeline.
Tests at baseline, year 1, year 2 and year 4: isometric quadriceps torque (Good Strength chair, best of ≥3 attempts per leg), leg extensor power (primary outcome), handgrip, DXA (lean mass, lean leg mass, body fat, visceral fat), MRI vastus lateralis CSA (blinded), 5-day accelerometer step count.
Statistics: two-way mixed-model ANOVA adjusted for age and sex across the four time points; one-way ANOVA for 4-year change; Bonferroni-corrected significance P < 0.006 (eight tests).
Key Results
| Outcome (baseline → year 4, mean ± SD) | HRT | MIT | CON | Group × time |
|---|---|---|---|---|
| Isometric leg strength (Nm) | 149.7 ± 51.5 → 151.5 ± 51.1 (P = 1.00) | 147.6 ± 54.9 → 142.9 ± 54.6 (P = 0.28) | 144.4 ± 43.3 → 134.5 ± 42.2 (P < 0.001) | F6,1049 = 8.607, P < 0.001, η² = 0.05; 4-year change HRT > MIT (P = 0.003) and > CON (P < 0.001) |
| Leg extensor power (W) — primary outcome | 195.3 ± 65.1 → 185.3 ± 60.2 | 193.2 ± 64.0 → 179.4 ± 64.1 | 187.7 ± 61.8 → 171.2 ± 57.2 | P = 0.39 (time P < 0.001: fell in all) |
| Handgrip (kg) | 35.6 → 34.3 | 34.3 → 32.7 | 34.9 → 32.8 | P = 0.77 (time P < 0.001) |
| Lean leg mass (kg) | 17.0 → 16.3 | 16.7 → 15.9 | 16.4 → 15.7 | P = 0.09 (time P < 0.001) |
| Lean body mass (kg) | 47.5 → 47.3 (P = 0.81) | 46.8 → 46.4 (P < 0.001) | 46.5 → 46.0 (P = 0.003) | F6,1085 = 5.353, P < 0.001, η² = 0.03 |
| Vastus lateralis CSA (mm²) | 1403.7 → 1306.2 | 1371.4 → 1284.3 | 1359.5 → 1243.1 | P = 0.001, driven by year-1 and year-2 changes |
| Visceral fat (g) | 1172.4 → 1188.5 (P = 1.00) | 1175.0 → 1210.1 (P = 0.95) | 1099.6 → 1184.4 (P = 0.04) | F6,1085 = 3.120, P = 0.005 |
Shape over time (Figure 1A): HRT's isometric strength rose at year 1 (P < 0.001) and was still above baseline at year 2 (P < 0.001), then came back to baseline by year 4; MIT rose at year 1 (P = 0.01); CON was below baseline at year 4. "Maintained" therefore means the gain was spent and baseline was held, against a falling control.
MIT: its 4-year fall was not significant on the trajectory post hoc (P = 0.28) but was on the baseline-vs-year-4 comparison (P = 0.01); either way its change was smaller than HRT's.
Attendance: 82% at year 4. The 82 who left were heavier at baseline, but their year-1 responses did not differ from those who stayed.
Mechanisms
Measured: the strength difference held while lean LEG mass and thigh CSA fell in all groups, so the retained strength is not explained by retained muscle size.
Authors' interpretation: neural adaptations to heavy training may persist and carry strength even as the muscle shrinks; they cite work linking lifelong training to elevated acetylcholine receptors. Nothing neural was measured.
Authors' interpretation: heavy loads, not just any resistance training, may be needed for long-lasting effects — MIT improved lean mass and function at year 1 but left no muscular effect at year 4. MIT also differed in supervision and setting (mostly at home), so load is not the only difference (inference).
Inference: the result is a residue of a single training year followed by no training, not a maintenance effect of continued lifting; the base's data on what continued low-dose lifting keeps are in REF-587.
Practical Application
**Can a 65-year-old get stronger, and does it last?A year of heavy lifting at retirement age left leg strength at baseline 3 years after coaching ended while controls lost 7% — but leg power, grip and leg muscle fell in all groups.** In 451 Danes, supervised heavy lifting (3×/week, 3 × 6-12 reps at ~70-85% 1RM) raised isometric leg strength at year 1; at year 4 it was 149.7 → 151.5 Nm against 144.4 → 134.5 Nm in controls (−7%, our calculation).So yes to "you can get stronger and some of it sticks"; no to "a year of lifting stops ageing". The 4-year result is a secondary outcome; the primary one, leg power, showed no group difference.
Prefer heavy over light for a lasting effect (heavy was also supervised in a gym, light mostly done at home). Body-weight and band circuits at ~50-60% 1RM gave a year-1 gain but no measurable advantage over control at year 4. That matches the base's dose evidence in adults over 50 (REF-1086: ≥70% 1RM built more leg strength, no difference in falls or adverse events).
Keep lifting rather than banking a year. Strength drifted back to baseline once coaching ended; holding it above baseline needs continued work (inference). In older general populations the base's maintenance floor is about 2 sessions a week and 2-3 sets at maintained load (REF-587), and losses after stopping are larger over 65 (REF-375).
Train power separately. Leg power fell ~5% in HRT and ~9% in CON over 4 years (our calculation) with no group difference, so heavy lifting (tempo not reported) did not protect it here. The base's evidence for power in adults over 50 favours fast-intent, high-velocity work (REF-597) — not from this paper.
Limitations
Interim analysis of the 4-year follow-up; the 7- and 10-year follow-ups are planned. Isometric leg strength is a secondary outcome; the primary outcome (leg extensor power) did not differ between groups, and handgrip, lean leg mass and power fell alike.
Isometric quadriceps torque in a test chair is not a functional or sport outcome; no falls, mobility or performance outcomes are reported at year 4.
18% lost to follow-up (82 of 451), heavier at baseline; the year-1 responses of leavers did not differ.
The groups differed in supervision and setting as well as load (HRT supervised in a gym; MIT mostly at home with bands), so "heavy" is confounded with supervision.
Activity after year 1 was not controlled beyond step counts; whether some HRT participants kept lifting is not reported.
A well-functioning, active sample (~9 500 steps a day) at retirement age; the authors note it is healthier than the average ageing population.
Funding: Lundbeck Foundation and Nordea Foundation; no competing interests declared.
How this entry was checked and changed
Why this level
large registered RCT (n=451 randomised, 369 tested at year 4), but the headline 4-year result is a SECONDARY outcome (isometric leg strength) measured three years after training stopped, the primary outcome (leg extensor power) showed no group difference, and 18% were lost to follow-up; read from the full text
Full text (checked 2026-10-15)
Randomised: HRT, n=149; MIT, n=154; CON, n=148; year 4: HRT, n=128; MIT, n=126; CON, n=115; study attendance 82% at year 4 (full text).
HRT: three times per week, 6–8 weeks of habituation, 3 sets of 6–12 repetitions at ~70%–85% of 1 RM; MIT: 3 sets of 10–18 repetitions at ~50%–60% of 1 RM (full text).
Age 66.4±2.5 at baseline and 70.5±2.5 at year 4 (range: 64–75 years); 61% women; 9548±3446 steps/day at baseline (full text).
Isometric leg strength interaction F6,1049 = 8.607, p<0.001, η2 = 0.05; HRT 149.7±51.5 → 151.5±51.1 Nm, p=1.00; CON 144.4±43.3 → 134.5±42.2 Nm, p<0.001; MIT 147.6±54.9 → 142.9±54.6 Nm, p=0.28 (full text).
Δ over 4 years: HRT>MIT, t(350) = 3.273, p=0.003; HRT>CON, t(350) = 3.655, p<0.001 (full text).
Power: HRT 195.3±65.1 → 185.3±60.2 W; MIT 193.2±64.0 → 179.4±64.1; CON 187.7±61.8 → 171.2±57.2; F6,1067=1.054, p=0.39 (full text).
Lean body mass F6,1085 = 5.353, p<0.001: HRT 47.5 → 47.3 kg (p=0.81), MIT 46.8 → 46.4 (p<0.001), CON 46.5 → 46.0 (p=0.003) (full text).
Significance p<0.006 after Bonferroni correction for eight tests; ≈80% had at least one chronic medical disease (full text).