Time-lapse embryo monitoring did not raise IVF live-birth rates, a pooled analysis found

Time-lapse imaging is one of the most widely sold "add-ons" in fertility treatment. Instead of removing embryos from the incubator each day to check them under a microscope, a time-lapse system photographs them continuously inside a sealed, stable incubator. The pitch is twofold: embryos are never disturbed, and the stream of images yields morphokinetic data — the timing of each cell division — that software can use to pick the embryo most likely to implant. Clinics often charge extra for it. An individual-participant-data (IPD) meta-analysis published in Human Reproduction Update on October 29 asked whether it actually produces more babies [s1].

Why pool the raw data

Several randomised trials of time-lapse imaging already exist, and conventional meta-analyses that combine their published summaries have found the benefit uncertain [s1]. The problem is heterogeneity: trials enrolled different patients and used different selection algorithms, and summary-level analyses cannot adjust for that cleanly [s1]. An IPD meta-analysis obtains the original record for each participant, which allows harmonised adjustment for baseline risk, formal testing of whether the effect varies by patient or technique, and greater statistical power [s1].

The authors updated the searches behind the 2019 Cochrane review of time-lapse systems, drawing on MEDLINE, EMBASE, CINAHL and CENTRAL through mid-2025 and repeating the search in July 2026, then requested the raw data from trial authors [s1][s2]. They separated three questions: the overall effect of time-lapse systems, the effect of undisturbed culture alone, and the effect of morphokinetic-based embryo selection alone [s1]. The primary outcome was live birth after the first fresh or frozen embryo transfer [s1].

What it found

Eight trials contributed individual data, covering 5,642 women, with eight further trials (3,452 women) added as aggregate data for a sensitivity analysis [s1]. Risk-of-bias assessment raised no concerns with the trials that provided individual data [s1].

Across the board, the answer was no meaningful difference. For overall time-lapse versus standard incubation and selection, the adjusted odds ratio for live birth was 0.97 (95% confidence interval 0.81-1.15), an absolute risk difference of -0.7 percentage points (95% CI -4.9 to 3.4) across four trials and 2,913 women [s1]. Undisturbed versus disturbed culture gave an adjusted odds ratio of 1.02 (0.84-1.23), a risk difference of 0.5 points (-4.1 to 5.0), across five trials and 2,971 women [s1]. Morphokinetic-based selection versus standard selection gave 0.92 (0.78-1.07), a risk difference of -2.0 points (-5.7 to 1.6), across three trials and 2,937 women [s1].

The authors rated this as high-certainty evidence of no clinically important difference in live birth [s1]. Nor did the effect vary by a woman's age, the duration of undisturbed culture, the day of embryo transfer, or whether the transfer was fresh or frozen [s1]. Clinical pregnancy, miscarriage, biochemical pregnancy loss, birthweight and gestational age at birth were all similar between time-lapse and standard care across every comparison [s1].

Why "high certainty" is the key phrase

A null result can mean two very different things: that a treatment does not work, or that the study was too small to tell. The authors' judgement of high-certainty evidence is a claim that this is the first kind. The confidence intervals are narrow enough, and the individual-level data consistent enough, to say that time-lapse imaging does not meaningfully raise the chance of a live birth — not merely that a benefit was missed. That squares with the Cochrane review's long-standing verdict that the evidence does not support routine use [s2].

That does not make the technology worthless. Undisturbed culture may streamline a laboratory's workflow, and some patients value the images. But those are operational and experiential reasons, not a higher birth rate.

The distinction matters because fertility add-ons are a recurring concern for regulators. Patients paying for IVF are often willing to try anything that might improve their odds, and clinics face commercial pressure to offer the latest equipment. Several oversight bodies now publish ratings of IVF add-ons precisely because so many are sold ahead of the evidence. A high-certainty null result on the outcome patients actually care about — a baby — is the kind of finding those ratings are built to reflect, and it removes the usual defence that the trials were simply too small to settle the question.

The limits

The analysis measures reproductive outcomes, not laboratory convenience or cost, and it pools trials that used a range of selection algorithms — so it tests time-lapse selection as a category rather than any one proprietary system [s1]. A future algorithm could in principle outperform the ones tested here; the evidence speaks to what has been trialled.

What to watch

Whether fertility regulators and guideline bodies cite this analysis to downgrade time-lapse imaging as a paid add-on, and whether clinics continue to charge a premium for a step these data do not tie to more babies.

This article describes a meta-analysis of existing trials and is informational only. It is not medical advice, and decisions about fertility treatment belong with a specialist.