REF-1146 · van Ark et al., 2016 · Journal of Science and Medicine in Sport · Paper · PubMed

In-season jumping athletes with patellar tendinopathy who kept training and competing: 4 weeks of isometric or isotonic leg exercise, 4 sessions a week, both reduced decline-squat pain, with no difference between them (median change 2.5 vs 3.0 of 10, p = 0.208); n=29 randomised, about 19 analysed

Blimited evidenceMeasured on trained to highly trained athletes

Van Ark and colleagues tested whether a home-style exercise programme can reduce patellar tendon pain in jumping athletes without cutting their training and competition. Twenty-nine athletes playing at least three times a week were randomised to isometric or isotonic exercise, 4 sessions a week for 4 weeks.

Pain during a single-leg decline squat (SLDS, 0-10) fell in both groups (isometric p = 0.012, isotonic p = 0.003, effect size r = −0.63 in each). The between-group difference was not significant: median improvement 2.5 (IQR 1-4.5) after isometrics vs 3.0 (2-6) after isotonics (p = 0.208).

The authors call it the first study to reduce patellar tendon pain without modifying training and competition load, and recommend either programme. The trial is small and unblinded with no untreated group, so the pain fall cannot be separated from time or expectation. It does not support the idea that isometrics beat isotonic work over weeks.

Protocol

Participants (from the abstract): jumping athletes with patellar tendinopathy, playing at least three times a week; 29 included and randomised to an isometric or an isotonic group. Clifford 2020 tabulates 13 isometric and 16 isotonic, ages 16-32 (mean 23), 7% women, symptoms 1-120 months (mean 35.8), diagnosed clinically and on imaging.
Programmes: four isometric or isotonic exercise sessions a week for four weeks, alongside unchanged training and competition. Clifford 2020 describes both as unsupervised with identical time under tension.
Exercise detail is not in the abstract: the type of isometric (holding a load or pushing against a fixed resistance), %MVIC, hold duration, joint angle and sets are not given. Rio 2017 (REF-1145), from the same group in the same competitions, used leg-extension holds at 60 degrees of knee flexion at 80% of MVIC against leg extensions at 80% of 8RM; that this trial used the same dose is likely but unconfirmed.
Outcome: SLDS pain on a 0-10 numeric rating scale at baseline and at 4 weeks. Clifford 2020 lists VISA-P and global rating of change as further outcomes (not in the abstract).

Key Results

OutcomeIsometricIsotonicStatistics
Randomised13 (per Clifford 2020)16 (per Clifford 2020)29 in total (abstract)
Analysed (our calculation from the reported Z, U and r)about 8about 11about 19 of 29
Within-group change in SLDS pain over 4 weeksimproved: Z = −2.527, p = 0.012, r = −0.63improved: Z = −2.952, p = 0.003, r = −0.63Wilcoxon-type within-group tests
Median (IQR) improvement in SLDS pain2.5 (1-4.5)3.0 (2-6)U = 29.0, p = 0.208, r = 0.29
SLDS pain baseline → 4 weeks, medians as tabulated by Clifford 2020 (REF-1149, Table 7)6.3 → 45.5 → 2treatment effect −0.5 (−2.6 to 1.6) as reported by the authors, per Clifford
VISA-P, global rating of changeboth improved, no between-group difference (Clifford 2020's summary)not in the abstract
Analysed numbers: with r = Z/√N, r = −0.63 gives N ≈ 16 observations (8 athletes) for Z = −2.527 and N ≈ 22 (11 athletes) for Z = −2.952; U = 29.0 with r = 0.29 also fits about 19 athletes. Our calculation, consistent with Clifford's 62% follow-up; the paper's own flow numbers were not seen.
Is "no difference" evidence of equivalence? No. With about 8 and 11 athletes and IQRs of 3.5 and 4 points, the trial could miss a between-group difference of a point or more; our inference.

Mechanisms

Not measured: the abstract reports no mechanism outcome.
Clifford 2020's interpretation (not in van Ark's abstract): with time under tension matched, contraction type may matter less than the load itself.
Inference: without an untreated group, part of the pain fall could be time, regression from a painful baseline, or expectation.

Practical Application

Can a jumper keep playing while treating patellar tendon pain? In 29 in-season athletes, 4 weeks of isometric or isotonic leg exercise both cut pain (median 2.5 vs 3 of 10, p = 0.208), no winner; only about 19 were analysed.
Pick the form the athlete will do: neither contraction type beat the other over 4 weeks, though with no untreated group the trial cannot show how much of the fall was the exercise; Clifford's meta-analysis (REF-1149) reached the same conclusion across tendons.
Playing on was compatible with improvement here: pain fell while training and competition continued unchanged. Not from this paper; the base's own practice: keep the sport under the pain-monitoring rule in SYNTH-73, whose tested source is REF-628 (Achilles).
Not from this paper; the base's own practice: for runners and cyclists with front-of-knee tendon pain this is a transfer; heavy, progressive loading of any kind is the base's default for tendon capacity (REF-626, REF-517).

Limitations

Abstract only: exercise dose, supervision, adherence, dropouts and secondary outcomes could not be checked.
Small, with heavy attrition: about 19 of 29 analysed by our calculation (Clifford: 62% follow-up); completers only as far as the abstract shows.
No untreated control; participants and, per Clifford, assessors were not blinded; self-rated pain.
Clifford 2020 rated it poor quality: high risk for blinding, attrition, external validity (two sports only) and precision (no power calculation); acute and chronic cases mixed (symptoms 1-120 months).
Possible overlap with Rio 2017 (REF-1145): same group, same competitions, same two programmes; Clifford tabulates both with this trial's 29 athletes. They should not be counted as two independent trials.
Funding and conflicts of interest are not stated in the abstract.
How this entry was checked and changed
Why this level

(from the abstract) — parallel RCT with 29 randomised but only about 19 analysed (rubric B: parallel RCT <20 analysed): the reported test statistics imply that only about 19 were analysed (our calculation below), non-parametric median comparisons with no confidence interval for the between-group difference, no power calculation reported, and Clifford 2020 (REF-1149) rated it poor quality (unblinded assessors, 62% follow-up, unclear adherence)