Does combining lifting and endurance work after 50?
Against doing nothing, yes: strength SMD 0.74, power 0.50, aerobic fitness 0.48, in 15 RCTs of healthy adults, status unknown. No single-mode arm, so it cannot show interference.
Main caveat: Passive controls only. It answers whether concurrent training beats nothing. The authors themselves name comparison with single-mode training as a gap.
Markov and colleagues pooled randomised trials of concurrent strength and endurance training against passive controls in healthy adults aged 50 and over, searched to February 2022. Fifteen trials and 566 participants qualified. Strength rose moderately (SMD 0.74), power moderately (0.50), cardiorespiratory endurance a little (0.48), and balance not significantly (0.33).
Single-factor analyses suggested larger strength effects with 12 weeks, 3 sessions a week, sessions of more than 30 to 60 min and strength before endurance, and larger endurance effects with 21 weeks, 4 sessions a week, sessions of more than 60 to 90 min and separate days.
Those lists look like a prescription and are not one: meta-regression found no training variable predicted the strength effect, most subgroup differences were not significant, the moderators were analysed one at a time, and several rows are the same few trials under different labels.
Our reading: solid for "doing both beats doing nothing in healthy adults over 50"; silent on interference; too thin for dose.
Protocol
Design: systematic review with meta-analysis; PRISMA; PROSPERO CRD42020188618, registered 5 July 2020.
Search: PubMed, Web of Science Core Collection, MEDLINE, Google Scholar, to February 2022; English, peer-reviewed RCTs.
Eligibility: healthy adults aged 50 and over, any sex or activity level, excluding disease (diabetes, sarcopenia, asthma, hypertension); concurrent training vs a passive control; active controls excluded.
Who: n = 566 in 15 trials; mean age from 50 to 73.5 (mean 61.0 ± 5.9 years); where sex was reported 188 women and 363 men; 332 in training arms and 219 controls.
Training: 8 to 21 weeks, 2 to 4 sessions a week, 30 to 90 min per session. Strength intensity was rated moderate-to-hard in almost all trials, so load cannot be compared here.
Outcomes: strength (knee-extensor isokinetic, isometric or 1RM), power (knee-extensor power, rate of force development, jumps), cardiorespiratory endurance (VO2peak or VO2max, maximal aerobic workload), balance.
Quality: PEDro median 5 (range 3-6); only two trials reached 6; Egger's test gave no evidence of funnel-plot asymmetry.
Funding and interests: no funding for the review (open-access fee covered by Projekt DEAL); the three authors declare no relevant conflicts. Data and R code on the Open Science Framework.
Key Results
| Outcome vs passive control | SMD [95% CI] | p | I² | Studies (groups); n intervention (control) |
|---|---|---|---|---|
| Muscle strength | 0.74 [0.37 to 1.11] | < 0.001 | 49.4% | 12 (19); 235 (158) |
| Muscle power | 0.50 [0.04 to 0.96] | 0.037 | 0.0% | 5 (6); 102 (66) |
| Cardiorespiratory endurance | 0.48 [0.31 to 0.64] | < 0.001 | 0.0% | 9 (14); 184 (114) |
| Balance | 0.33 [−0.31 to 0.97] | 0.221 | 16.5% | 4 (5); 85 (65) |
Subgroups and single factors (Table 5; all indirect, across trials, none within one trial):
| Factor | Strength SMD | Endurance SMD | Difference |
|---|---|---|---|
| Age ≤ 65 vs > 65 | 0.60 vs 1.04 | 0.52 vs 0.39 (not significant) | not significant |
| Men vs women | 0.38 vs 1.05 | 0.45 vs 0.55 | strength p = 0.06, endurance p = 0.71 |
| 12 vs 21 weeks | 0.87 vs 0.47 | 0.40 vs 0.62 | not significant |
| 3 vs 4 sessions a week | 0.91 vs 0.55 | not significant at 3 vs 0.58 | not significant |
| More than 30-60 vs more than 60-90 min | 0.99 vs 0.40 | 0.34 vs 0.61 | not significant |
| Strength first / endurance first / separate days | 1.00 / 0.65 (not significant) / 0.55 | not significant at 0.10 / 0.49 / 0.58 | not significant |
The one significant training factor, power: 12 weeks (SMD 0.88) vs 21 weeks (0.29), p = 0.016, and more than 30 to 60 min (0.88) vs more than 60 to 90 min (0.29), p = 0.016. The two contrasts print identical figures: 2 studies against 3 (Table 5).
Same trials under different labels (our reading of Table 5 and Table 3): "4 sessions a week" and "separate days" give identical estimates for strength (0.55; 3 studies, 5 groups) and endurance (0.58; 3 studies, 5 groups), and every 21-week arm comes from the Holviala and Karavirta papers (Table 3). Duration, frequency, session length and separation cannot be told apart in these data.
Meta-regression: no training variable predicted the strength effect (R² 0-3.76%, p > 0.05); run for strength only.
Women vs men: the key points say larger effects in women; the difference was not significant, no trial compared the sexes directly, and the authors ask for caution.
Mechanisms
Measured: none. The authors state that the review has no mechanistic outcomes.
Authors' interpretation: strength and power adapt on a shorter clock than cardiorespiratory endurance, and residual fatigue after endurance work may hold back strength gains when endurance comes first.
Inference: the time-course reading rests on the 12-versus-21-week split, which is confounded with frequency, session length and research group, so it does not stand on these data alone.
Practical Application
Does combining lifting and endurance work after 50? Against doing nothing, yes: strength SMD 0.74, power 0.50, aerobic fitness 0.48, in 15 RCTs of healthy adults, status unknown. No single-mode arm, so it cannot show interference.
Balance needs its own work (authors' reading): concurrent training did not raise balance significantly, on 4 studies; the authors suggest adding balance exercises.
For a masters endurance athlete (inference): expect smaller effects than these, because the samples were not selected as athletes and faced passive controls; the base's transfer rule says lower-tier effects are probably overstated (REF-529).
Limitations
Passive controls only. It answers whether concurrent training beats nothing. The authors themselves name comparison with single-mode training as a gap.
Few, small, low-scoring trials. Median PEDro 5; four of 15 trials blinded assessors; subgroup analyses underpowered by the authors' own statement.
Moderators analysed one at a time and across trials, ignoring interactions; the Holviala and Karavirta papers fill the 21-week rows and, with one 12-week trial (Libardi), the 4-session rows.
Reporting. The abstract says no subgroup differed, and Table 5 prints p = 0.049 for age on power and p = 0.03 in other cells while the text reports no difference between subgroups; the discussion calls concurrent training effective for balance and in the same paragraph not effective. We report the numbers, not those summaries.
Who. Healthy volunteers without the listed diseases; training status not established; no athletes selected. Body composition was not an outcome.
Interests. No funding for the review; no conflicts declared.
How this entry was checked and changed
Why this level
read from the full text (open access, CC BY 4.0). An SR/MA of controlled trials is A by design; held at A- because the trials are small, median PEDro 5 (range 3-6, the authors read most as below their cut-off of 6), and the moderator analyses rest on few studies
Full text (checked 2026-10-31)
Fifteen studies were eligible with a total sample of 566 participants, mean age from 50 to 73.5 years (mean 61.0 ± 5.9 years) (full text)
Concurrent training vs passive control: strength SMD = 0.74 [95% CI 0.37–1.11], power SMD = 0.50 [95% CI 0.04–0.96], cardiorespiratory endurance SMD = 0.48 [95% CI 0.31–0.64], balance SMD = 0.33, p = 0.221 (full text)
Median PEDro score 5, range 3–6, with only two studies reaching 6 (full text)
Meta-regression: none of the training variables predicted the strength effect (R2 = 0–3.76%; p > 0.05) (full text)
Power: 12 weeks SMD = 0.88 vs 21 weeks not effective, difference p = 0.016 (full text)
Training status could not be established consistently because the included studies lacked those details (full text)
No funding was used for the review and the three authors declare no conflicts of interest relevant to it (full text)