Stretch mark creams: the larger trial missed its primary outcome — and then one subgroup differed

A 2000 Cochrane review, whose objective was to assess the effects of topical treatments to prevent the development of stretch marks.

• Searched the Cochrane Pregnancy and Childbirth Group trials register and CENTRAL
• Date of last search: April 1999
• Selection criteria: randomised trials comparing active creams with placebo in pregnant women
• Two reviewers independently assessed trial quality and extracted data

The result.

One study involving 100 women was included. Compared to placebo, treatment with a cream containing Centella asiatica extract, alpha tocopherol and collagen-elastin hydrolysates was associated with fewer women developing stretch marks.

What to notice in this item is not the result but the denominator.

A Cochrane review gathers every randomised trial in a field. And on this topic, one trial met the criteria. Compare that with the LED phototherapy page on this site, where Cochrane gathered 71.

And the last search was April 1999. That is material more than 25 years old. The record notes an update in 2012, but what we verified at source is this version.

Look again at what was in the cream. Centella asiatica extract, alpha tocopherol, collagen-elastin hydrolysates — three things. The result belongs to that combination, not to any one of them. The ferulic acid page records a problem of the same shape.

So this item sits at 'moderate'. It is a placebo-controlled randomised trial vetted by Cochrane, but it is one trial, of 100 women, of a combination product.

A randomised, double-blind, placebo-controlled trial conducted in Spain between November 2009 and April 2011.

• Treated group: a moisturising cream containing hydroxyprolisilane-C, rosehip oil, Centella asiatica triterpenes and vitamin E
• Control group: the same cream without the active ingredients

The design is good. Using the base with only the actives removed as the comparator is the condition we keep looking for on this site — the gap we wrote about on the EGCG and l-carnitine pages, where there was no plain-base arm, is filled here.

And the primary outcome came out like this.

• Overall incidence of stretch marks during pregnancy: 33.3% control, 37.6% treated — not significant

The cream side was numerically higher. Since the difference is not significant, it must not be read as 'the cream caused more stretch marks' — but it cannot be read as 'it reduced them' either.

Then other results appeared.

• Severity of previous stretch marks: increased significantly in the control group (17.8%, P = .001), not significantly in the treated group (6.3%)
• Among women who developed new stretch marks, the difference in severity (baseline to maximum): control 0.47, treated 0.14, P = .031
• Among women without previous striae, incidence was 5.6% treated against 35% control, P = .031

That last line stands out most. 35% against 5.6% is a large difference.

But how should it be read?

This is a subgroup analysis. There was no difference across the trial as a whole, and a difference appeared in one slice of the participants.

Subgroup results need care because there are many ways to slice. By age, by weight gain, by family history, by whether there were previous marks — slice enough times and the chance that one of them shows a significant difference by luck goes up.

Whether this subgroup was specified in advance is not in the abstract. A pre-specified subgroup and a subgroup found after looking at the results carry entirely different weight.

So this item is filed as 'not established'. What is not established is whether the cream reduces the incidence of stretch marks. The primary outcome did not show it, and a subgroup result cannot carry the conclusion on its own.

One thing the trial did well should also be recorded. The primary outcome was negative, and they put it at the front of the abstract. They could have buried it and led with the subgroup.

Both trials above are about preventing stretch marks from forming in pregnant women. Neither is about fading marks that already exist.

Why that distinction matters.

A stretch mark is a tear in the dermis. As the stretch marks page sets out, it is not a problem of the stratum corneum or of pigment but of broken structure. That a topical cream passes through the stratum corneum and knits that broken structure back together is a claim at an entirely different level from prevention.

We did not verify, for this page, any randomised placebo-controlled trial of a cream applied to stretch marks that already exist.

On that side, what we verified at source is procedures.

• Vascular laser and carboxytherapy — they tied with each other on red stretch marks
• Neither has a trial against doing nothing at all

The contribution of moisturising itself also has to be left open. Applying cream to a pregnant belly comes with massage. In the trials above the control group applied the same cream (minus the actives) the same way, so that contribution fell equally on both sides — the design handled that point well.

Separating what is and is not established:

• In the one trial Cochrane included, a three-ingredient cream beat placebo — established
• In the 2013 trial there was no difference in overall incidence — established
• There was a difference among women with no previous marks — established, but as a subgroup
• Which ingredient it belongs to — not established
• Whether applying it to existing marks fades them — not established