REF-941 · Feely et al., 2023 · Frontiers in Sports and Active Living, Vol. 4, article · Paper · PubMed · Free full text

Estimating the cost of training disruptions on marathon performance

B+moderate evidenceMeasured on recreational to trained athletesWritten before October 2026, being re-checked

Using more than 15 million activities from 300,000 recreational runners who completed marathons during 2014-2017, the authors identified periods of complete cessation of training (training disruptions) of varying durations up to 16 weeks before the marathon.

They then took runners who had both a disrupted marathon (a long disruption of ≥ 7 days) and an undisrupted one, and expressed the cost as the percentage difference between those two finish times.

Over 50% of runners had short disruptions up to and including 6 days (abstract), while the Results report that about 58% of male and 55% of female runners had a disruption of at least 7 days (corrected 2026-10-12, full text); longer ones were increasingly less frequent among those who reached race day.

Longer disruptions (≥ 7 days) carried a finish-time cost of 5-8% compared with when the same runners had only short disruptions (<7 days). The cost was greater for males than females, for faster than slower runners, and for disruptions late in training than early.

Participants

300,000 recreational runners who completed marathons during 2014-2017 (more than 15 million logged activities). The cost analysis used runners with multiple marathons: at least one disrupted and at least one undisrupted. Ages and the male-female split are not given in the abstract.

Protocol

Training disruption = a period of complete cessation of training, of varying duration, in the up to 16 weeks before the marathon.
Long disruption: ≥ 7 days. Short disruption: <7 days.
Cost = percentage difference between a runner's disrupted and undisrupted marathon times.
Frequency and cost compared by sex, age, ability, and early vs late timing of the disruption; two sample t-tests.

Key Results

Runners with longer training disruptions (≥ 7 days) had a finish-time cost of 5-8% compared with when the same runners had only short disruptions (<7 days).
Over 50% of runners experienced short training disruptions up to and including 6 days (abstract); the Results report a disruption of at least 7 days in about 58% of male and 55% of female runners, and the Discussion says disruptions of at least 7 days are experienced by more than 50% of runners (corrected 2026-10-12, full text); longer disruptions were increasingly less frequent among those who made it to race day.
Likelihood of disruption differed little (<5%) by sex, age, ability or timing; the cost of disruption differed significantly (10-15%) among these groups.
Sex: greater finish-time cost for males (5%) than females (3.5%).
Ability: faster runners (5.4%) vs slower runners (2.6%).
Timing: disruptions late in training, close to race day (5.2%) vs earlier in training (4.4%).

Practical Application

"I missed a few days of my marathon build — how much did it cost me?" In 300,000 recreational marathoners, short breaks (up to and including 6 days) were the comparison baseline (the abstract says over 50% of runners had them; the Results say about 55-58% had a stop of at least 7 days (corrected 2026-10-12, full text)); a complete stop of ≥ 7 days went with a 5-8% slower finish than the same runner's undisrupted marathon. A gap of a few days is normal; a full week or more off is where a measurable cost shows up.
After a long disruption, re-set the race goal instead of chasing the lost training. Not from this paper; the base's own practice: restart from the calendar and do not stack sessions to catch up (SYNTH-72).
The cost was larger when the break came late in training (5.2% vs 4.4%) and in faster runners (5.4% vs 2.6%). Inference, not tested here: a late, long break in a runner with a fast goal is the case that most needs a revised pacing plan.
The reason for the break is not separated from the break: illness or injury may slow a marathon beyond the lost training. Returning after a break is in SYNTH-36.

Limitations

Observational: disruptions were not assigned, and their cause is unknown from the logs; illness or injury may contribute to the cost.
A disruption is a gap in logged activities; unlogged training would be counted as a stop.
Only runners who reached race day are included; those who dropped out after a long disruption are not.
Cost of a single missed session or of reduced training is not addressed.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

very large observational cohort with each runner as their own comparison; disruptions were not assigned, so their cause (illness, injury, motivation) travels with them and the cost is an association

Full text (checked 2026-10-12)
Correction: the entry says short breaks "up to and including 6 days" happened to "over 50% of runners"; the results report that approximately 58% of male runners and 55% of female runners experience a training disruption of at least 7 days, and the discussion states disruptions of at least 7 days are experienced by more than 50% of runners (full text).
The dataset was 292,323 Strava runners (80% male and 20% female) with 509,979 marathons, mean age 38.51 (female) and 40.22 (male), training 3.12 and 3.07 days per week (full text, Table 1).
The cost analysis used 43,933 runners with 56,735 disrupted/undisrupted pairs whose two marathons were at least 6 months and no more than 2 years apart (full text).
A marathon counted as disrupted if the longest gap 3–12 weeks before race-day was at least 7 days; late meant 3–7 weeks and early 8–12 weeks from race-day, and the final three weeks (taper) were excluded (full text).
A 7–13 day longest disruption was associated with a 4.25% increase in finish time, and cost rose with disruption length (full text).
With a 21–27 day disruption, males slowed close to 8% versus just over 4% for females, and a 21–27 day disruption 3–7 weeks before race-day cost almost 10% versus 6% when it occurred earlier (full text).
Fast runners were defined by a PB less than 240 min, and their disruption cost was 2–3x as great as that of slower runners (full text).