Research · Growth & Maturation

Two Thirteen-Year-Olds, Years Apart: What Maturity Assessment Can and Cannot Tell You

By Muhammad Dzulhisham · MSc Sports Coaching, NSCA-CSCS

Put twenty thirteen-year-olds in a room and you are not looking at one developmental stage. You are looking at a spread that can cover several years of biological maturity — some children well past their growth spurt, some not yet in it, all carrying the same number on a registration form.

This is why maturity assessment has moved from research labs into academies, and increasingly into schools. The logic is sound: if training load, injury risk and selection judgement all track biological development rather than birthdays, then knowing where a child sits in their own growth curve is more useful than knowing their age. The complication is that every practical method of finding that out carries error, and the size of that error is routinely understated.

What is actually being measured

There are two different questions, and they get confused constantly.

Maturity status asks how far through the process a child currently is — usually expressed as a percentage of predicted adult height already attained, or as a skeletal age from an X-ray, DXA or MRI of the hand and wrist.

Maturity timing asks when the growth spurt happens — usually expressed as years before or after peak height velocity (PHV), the point of fastest growth.

A narrative systematic review by Sullivan and colleagues examined the methods used to answer both questions in male adolescent soccer players, covering invasive approaches (skeletal age via X-ray, DXA and MRI; Tanner-Whitehouse methods; the Fels method) and non-invasive ones (the Khamis-Roche and Bayley-Pinneau adult-height predictions, and the Mirwald, Moore and Fransen equations for maturity offset).¹

How accurate each one really is

The review's numbers are worth sitting with, because they set the limit on what any of these results can honestly be used for.

Two findings there deserve emphasis. The first is that the Mirwald maturity offset equation — by far the most commonly used in applied settings, because it needs only height, sitting height, mass and age — does not err evenly. It overestimates the timing of PHV in earlier-maturing players and underestimates it in later maturers.¹ In other words, its error is largest exactly where the decision matters most: at the extremes, where a child is being flagged as early or late.

The second is the concordance figure. Invasive and non-invasive methods agreed on classification between 55% and 75% of the time. The review's conclusion is unambiguous: they "should not be used interchangeably for assessing maturational status and timing in academy soccer players."¹ A player recorded as a late maturer on one method and an on-time maturer on another has not changed; the measurement has.

The population problem — and why it matters here

These equations were developed, in the main, on adolescents of European ancestry drawn from general populations.¹ Two separate consequences follow for anyone using them in Singapore.

The first is ancestry. The reference samples do not reflect the populations most Singapore squads are drawn from, and the review explicitly calls for existing equations to be validated, or new ones developed, in culturally and ethnically diverse environments.¹

The second is sampling. Academy players tend to mature earlier than the general population after about thirteen years of age¹ — so an equation calibrated on general-population adolescents is being applied to a group that has already been filtered by the very characteristic being estimated.

There is also a purely practical limitation. The Khamis-Roche method, one of the better performers, requires both parents' heights. Where those are unavailable and national averages are substituted, the standard error inflates.¹ A method is only as good as the inputs a coach can actually obtain on a Tuesday evening.

What the assessment is genuinely good for

None of this argues against measuring. A 2026 review in the Scandinavian Journal of Medicine & Science in Sports weighing the benefits and risks of growth and maturation assessment in youth sport identifies three defensible uses.²

Injury risk windows. The period during and for roughly twelve months after peak height velocity is a particularly vulnerable window.² Knowing which athletes are inside it changes how much jumping, sprinting and load you prescribe — and that is a training decision, not a selection decision.

Individualised programming. Calibrating training to biological rather than chronological age is the most straightforward application, and the one least likely to cause harm.

Fairer talent identification. Bio-banding — grouping by maturity rather than age — is described in that review as a promising counterweight to the systematic exclusion of late maturers.² Promising is the right word: a 2026 systematic review of 13 bio-banding studies found the evidence still rests on short-term match responses rather than any demonstrated improvement in selection accuracy.³

The part that gets handled badly

The same 2026 review is direct about the risks: body image disturbance, potential contribution to disordered eating, and psychological harm from careless communication or labelling. Its judgement is that these risks sit not in the act of measuring a child but in how the information is handled afterwards.² It proposes five conditions, and they read as a minimum standard rather than best practice:

Everyone involved — athlete, parent, coach — should understand that growth-related change is a normal and positive sign of healthy development. Results stay confidential, with no public or in-team disclosure. Participation is voluntary, with opt-out available without penalty. Assessment is carried out only by people with documented competence in both the technical and the psychological side. And the data is integrated with training load, injury monitoring and nutritional guidance rather than used in isolation.²

The review's warning about misclassification is the one to hold onto: it can lead to inappropriate training modification, inaccurate reassurance, or unintended labelling.² Given a method that disagrees with the gold standard a quarter to nearly half of the time, misclassification is not a hypothetical.

A workable position

Measure, but hold the result loosely. Use a maturity estimate to decide how much plyometric volume a fourteen-year-old gets this block, and be slower to use it to decide whether that fourteen-year-old has a future. Record the method used and stay with it, because switching equations mid-season generates apparent changes in maturity that are artefacts of measurement. Repeat measurements over time rather than trusting a single snapshot — the review's own recommendation is that all predictions carry individual error and that individual growth patterns must be tracked.¹

And never hand a child a number without a sentence around it. "You are a late maturer" is a label. "Your growth spurt has not happened yet, which is completely normal, and here is what we are changing in your training because of it" is coaching.

Where this evidence stops

The methodological review is restricted to male adolescent soccer players, so the accuracy figures quoted above cannot simply be transferred to female athletes or to other sports. The bio-banding evidence is 99.8% male and almost entirely European. And none of the reviewed work validates these equations in Singapore populations — which is precisely the gap that should make a local practitioner more cautious with the output, not less.