REF-1289 · Nuuttila et al., 2022 · Medicine & Science in Sports & Exercise · Paper · PubMed · Free full text
Does setting the number of hard sessions by recovery beat a fixed plan?
In 29 recreational runners over 12 weeks, a recovery-guided plan cut 10-km time 6.2% vs 2.9% fixed; volume was also adjusted. B+, Polar part-funded.
Main caveat: What was individualised is a package. Volume in the first block and interval count in the second were both guided.
Nuuttila et al. asked whether training adjusted to recovery and training status beats a fixed plan. Forty recreational runners (20 men, 20 women) did a 3-week preparatory period, were matched in pairs on sex, performance and training volume, and randomised to a predefined plan (PD) or an individualised plan (IND); 30 were analysed (PD 14, IND 16).
Both did a 6-week volume period (easy running growing in 2-up-1-down steps) and then a 6-week interval period built around 6 × 3 min at maximal sustainable effort with 2-min recoveries. PD did a fixed 3-3-1 pattern of interval sessions per 3 weeks (13-14 in all).
IND was re-evaluated every Monday and Thursday: the next 3-4-day block was increased, kept or cut depending on whether nocturnal HRV (4-week rolling average ±0.5 SD), fatigue and muscle soreness (above 5 on a 1-7 scale = negative) and an HR-running-speed index (worse by more than 0.50) were in range; in the volume period this changed session length, in the interval period the number of interval sessions (10 to 25 in 6 weeks; mean 15.8, not different from PD).
Both arms improved; 10-km time improved more in IND (−6.2 ± 2.8% vs −2.9 ± 2.4%, p = 0.002, ES 1.23), while maximal treadmill speed (+4.0 vs +3.0%) and VO2max did not differ between arms. More IND runners were "high responders" by a typical-error yardstick (10 km: 81% vs 23%).
Our reading: with REF-438 (40 recreational runners, HRV deciding each morning whether to run the planned hard session), one of two randomised trials in recreational runners in which recovery markers steered the hard sessions; here the guided arm's interval count did not differ significantly from the fixed arm's (15.8 vs 13.6, a "slight tendency" to more in the authors' words), so that timing rather than number did the work is inference; but the 10-km gain built up across both periods (neither period alone significant between arms), volume was also individualised, and the fixed arm's last week was a scheduled lighter week before the test, so "a recovery-guided HIT count" is only part of what was tested.
Protocol
Design: after a 3-week preparatory period (PREP), pairs matched on sex, maximal treadmill speed, 10-km time and training hours were randomised to PD or IND; 6-week volume period (VOL) then 6-week interval period (INT); tests at T0 (before PREP), T1 (after PREP), T2 (after VOL) and T3 (after INT). Field training, mainly outdoors, late spring to autumn (mean daily peak 17.3 °C). Ethics: University of Jyväskylä. Not blinded.
Participants: 40 recruited (20 men, 20 women), healthy, running at least 4 times a week, ECG checked by a cardiologist. Seven dropouts (illness 3, lower-limb injury 3, personal 1) and three exclusions (adherence < 90% of main sessions, a training break > 2 weeks, an illness > 2 weeks before testing): n = 30 analysed (PD 7 men + 7 women; IND 8 + 8); n = 13 in PD for the 10-km test (one ill). Age 34 ± 7 (PD) and 37 ± 7 y (IND); VO2max 45.7 ± 4.0 and 46.5 ± 6.7 mL/kg/min.
Volume period (both arms): easy running (below LT1) 30-90 min, one long run > 90 min and one 30-min moderate session (between LT1 and LT2) a week; PD volume 1.1, 1.2, 0.75×, 1.3, 1.4, 0.75× of baseline by week. IND started at 1.10 × baseline duration; +5% on "increase", −25% on "decrease".
Interval period: main session 6 × 3 min at maximal sustainable effort, 2-min recoveries (speed between LT2 and maximal test speed); other running easy. PD: three interval sessions in weeks 1, 2, 4, 5 and one in weeks 3 and 6 (lighter weeks at 0.75 × volume).IND: each 3-4-day block started from one interval session; "increase" added one, up to two or three per block; "decrease" cut volume 25% and removed interval sessions; after the last evaluation day at most one interval session before the tests.
Decision markers (IND): nocturnal HR and RMSSD (Polar Vantage V2, 4-h window from 30 min after sleep onset, LnRMSSD, range = 4-week rolling mean ±0.5 SD); fatigue and lower-limb soreness on a 1-7 scale each morning (Coach4Pro app; > 5 = negative); HR-running-speed index from minutes 5-10 of each run on flat ground (a decrement > 0.50 vs the previous 2 weeks, about 3-4 beats/min, = negative).Load adjustments in IND: 55 ± 12% maintained, 35 ± 10% increased, 10 ± 8% decreased.
Tests: incremental treadmill test (3-min stages, +1 km/h) for VO2max, maximal speed (vMax), LT1 and LT2; 10-km run on a flat 1.6-km asphalt loop in small groups, at least 48 h later; no interval or long session in the 2 days before tests.
Responder classes: change T1-T3 against the coefficient of variation between T0 and T1 (high > 2 × CV, moderate 0.5-2 × CV, trivial < 0.5 × CV).
Analysis: repeated-measures ANOVA with Bonferroni post hoc tests; between-group ANCOVA with T0 as covariate; Cohen's d with 95% CI.
Funding and interests: "supported by Polar Electro Oy (heart rate monitors and partial funding for the study) and by the grant received from the Foundation of Sports Institute"; the authors declare no conflicts of interest. Nummela is at the Finnish Institute of High Performance Sport KIHU.
Key Results
| Outcome, T1 → T3 | Predefined (n = 14) | Individualised (n = 16) | Between arms |
|---|---|---|---|
| 10-km time | −2.9 ± 2.4% (p = 0.004; d 0.20) | −6.2 ± 2.8% (p < 0.001; d 0.46) | p = 0.002; ES 1.23 (0.42 to 2.02); interaction p = 0.006 |
| 10 km, volume period (T1 → T2) | −0.8 ± 2.1% (p = 0.534) | −2.6 ± 3.1% (p = 0.001) | p = 0.125; ES 0.64 (−0.12 to 1.38) |
| 10 km, interval period (T2 → T3) | −2.0 ± 3.3% (p = 0.051) | −3.7 ± 2.2 (p < 0.001) | p = 0.087; ES 0.61 (−0.15 to 1.35) |
| Maximal treadmill speed (vMax) | +3.0 ± 2.4% | +4.0 ± 1.9% | p = 0.322; ES 0.46 (−0.27 to 1.18) |
| VO2max, mL/kg/min (T1 → T3) | 46.7 ± 3.9 → 50.7 ± 6.1 | 47.3 ± 7.2 → 50.3 ± 7.6 | no group effect reported |
| Interval sessions in the 6-week interval period | 13.6 ± 0.5 (range 13-14) | 15.8 ± 4.3 (range 10-25) | not different |
| Weekly hours: PREP / VOL / INT | 4.6 / 5.7 / 3.8 | 4.3 / 5.3 / 3.8 | not different |
Responder distribution (T1 → T3, against the T0-T1 CV): vMax high/moderate/trivial 29/50/21% (PD) vs 50/50/0% (IND); 10 km high/moderate/trivial/moderate-negative 23/54/15/8% (PD) vs 81/6/13/0% (IND).
Spread of change: the SD of the 10-km change was 2.4% (PD) and 2.8% (IND), so guiding the plan shifted the mean without narrowing the spread; there was no untrained control, so the trial cannot show how much of the spread is measurement noise.
Interval speed: rose in IND from 14.4 ± 1.6 km/h in week 1 to 14.9 ± 1.8 in week 6 (p = 0.023); no change in PD (14.6 → 14.7-14.9).
Markers in range: IND spent more time with the HR-speed index (82 ± 6% vs 75 ± 7%, p = 0.015) and LnRMSSD (52 ± 5% vs 45 ± 5%, p = 0.046) inside the normal range; fatigue and soreness did not differ.
High vs trivial responders (pooled): no difference in age, baseline vMax, training volume (56.1 vs 56.6 h) or perceived stress; high responders had more "normal" HR-speed-index values (82 ± 6% vs 73 ± 8%, p = 0.03).
Tolerance: serum free testosterone, cortisol, creatine kinase and countermovement jump did not change in either arm.
Mechanisms
Measured: the guided arm held its HR-speed index and nocturnal HRV in range more often and ran its interval sessions faster as the block went on, while the fixed arm's interval speed stayed flat.
Authors' interpretation: the fixed arm's three-a-week interval weeks may have left some runners in a compromised state ("probably because of too high interval frequency"); adjusting the load to the individual allowed sufficient recovery and also sufficient loading, which reduced low responses.
Authors' interpretation: part of the 10-km difference may be test timing — IND had no scheduled lighter week before the test week and the 10-km test came on the second test day, when adaptations may have been better realised.
Inference: the guided arm's mean interval count (15.8) was close to the fixed arm's (13.6); the effect, if real, came from when the sessions were done and from the volume adjustments, not from doing more.
Practical Application
Does setting the number of hard sessions by recovery beat a fixed plan? In 29 recreational runners over 12 weeks, a recovery-guided plan cut 10-km time 6.2% vs 2.9% fixed; volume was also adjusted. B+, Polar part-funded.
The adjustable count did not differ significantly from the fixed one. The guided runners did 10-25 interval sessions in 6 weeks (mean 15.8, about 2.6 a week) against 13-14 (about 2.3 a week), a range of about 1.7-4.2 a week (our arithmetic); the authors saw a slight, non-significant tendency to more. On about four runs a week the averages sat between two and three interval sessions a week (our arithmetic).
Decision rule as tested (B+): every Monday and Thursday, look at the last days' nocturnal HRV against a 4-week rolling mean ±0.5 SD, fatigue and leg soreness (above 5 on a 1-7 scale is a flag) and the heart rate at a given easy speed in the first minutes of a run; the next 3-4 days were increased (one more interval session), kept, or cut (volume −25%, interval sessions dropped) by a combination rule shown in the paper's Figure 1, not spelled out in its text; one marker out of range did not by itself force a cut (our inference: nocturnal HRV was in range on only 52% of data points, yet 10% of decisions were cuts).
Agrees with the base: HRV-guided arms did equal or better with similar or fewer hard sessions (REF-994, A-; REF-438, B+; REF-362, B), and self-report tracks load better than resting markers (SYNTH-30, REF-368).Not from this paper; the base's own practice: the quality-session count comes from SYNTH-26's decision table (usually two on four to seven sessions a week), "a recovery-budget judgement, not a measured optimum", and "Recovery decides the actual count" (SYNTH-26); this trial is the nearest test of letting recovery set that count.
Do not label a runner a non-responder from one block (inference): guiding the plan cut the share of trivial 10-km responders from 15% to 13% and moved most runners into the high class, but the spread of change was no narrower, and the responder yardstick came from a 3-week period in which the runners were also training.
Limitations
What was individualised is a package. Volume in the first block and interval count in the second were both guided; the between-arm 10-km difference was significant only over the whole 12 weeks (p = 0.002), not within the volume block (p = 0.125) or the interval block (p = 0.087).
Test timing differed by design. The fixed arm's last interval week was a scheduled lighter week (0.75 × volume, one interval session) and the guided arm did at most one interval session after its last evaluation; the authors themselves offer timing as a possible explanation.
Not blinded; outdoor 10 km. Runners knew their arm (though not the exact model), and the 10-km test was run outdoors in small groups.
Losses. 7 dropouts and 3 exclusions of 40 (injury 3, illness 3 plus 1 prolonged illness); the 10-km PD arm has 13.
Who and how long. Recreational runners on about four sessions and 4-6 h a week; men and women pooled; 12 weeks after a 3-week preparation. No cyclists, triathletes, masters-specific or competitive athletes.
Responder classes are relative to a training period. The CV used as the typical error came from T0-T1, across 3 weeks of preparatory training, not a repeated-test reliability study.
Reporting. The Results give the guided arm 15.8 ± 4.3 interval sessions; the Discussion says 15.6.
Funding and interests: Polar Electro Oy supplied the heart-rate monitors and part of the funding, and the decision rule ran on a Polar wrist device's nocturnal HRV; also a Foundation of Sports Institute grant. No conflicts declared.
How this entry was checked and changed
Why this level
parallel RCT, n = 30 analysed of 40 recruited, read from the full text (open access, CC BY). The rubric's 20-39 band; the primary performance outcome (10 km) differed between arms. Held at B+: not blinded, the 10-km test was run outdoors, 10 of 40 were lost or excluded, the effect cannot be assigned to the interval-count rule alone, and the funder made the HRV device that drove the decisions
Full text (checked 2026-10-27)
Recreational runners were divided into predefined (PD; n = 14) or individualized (IND; n = 16) training groups; 40 recreationally endurance-trained males (20) and females (20) were recruited (full text)
The change in the 10-km time was greater in IND compared with PD (−6.2% ± 2.8% vs −2.9% ± 2.4%, P = 0.002); between-group ES = 1.23; 0.42 to 2.02 (full text)
Volume period 10 km: IND −2.6% ± 3.1%, PD −0.8% ± 2.1%, between groups P = 0.125; interval period between groups P = 0.087 (full text)
Maximal treadmill performance T1 to T3: PD, 3.0% ± 2.4%; IND, 4.0% ± 1.9%; between-group P = 0.322 (full text)
HIT sessions during INT: PD, 13.6 ± 0.5 sessions; IND, 15.8 ± 4.3 sessions; range PD, 13–14 sessions; IND, 10–25 sessions (full text)
10-km responders high/moderate/trivial/moderate negative: 23%/54%/15%/8% for PD and 81%/6%/13%/0% for IND (full text)
IND adjustments: 55% ± 12% maintained, 35% ± 10% increased, and 10% ± 8% decreased the training load (full text)
The study was supported by Polar Electro Oy (heart rate monitors and partial funding for the study) and by the grant received from the Foundation of Sports Institute (full text)