When the meta-analysis finds no difference and you still have to recommend

1The decision

Single-operator cholangioscopy allows direct inspection of the bile duct. It costs more and is less widely available than endoscopic retrograde cholangiopancreatography, the standard technique. What ACED had to settle was whether it is worth using, and in which situations.

The decision sits with an association panel, and it has to hold up before its own members — the people who perform the technique — and before whoever pays for the equipment.

2Why it was hard

These were two different questions under one technology. Diagnosing an indeterminate biliary stricture and clearing a difficult stone have little in common: different patients, different outcomes, different comparator. The first is compared against brushing and biopsy; the second against large-balloon papillary dilation. Treating them as one would have produced a recommendation useful for neither.

And the panel had a composition problem. The people who know the technique best are the people who perform it and, in some cases, those who stand to gain if it is recommended. The panel was deliberately made multidisciplinary to reduce that bias, and the guideline says so.

3What we did

Universidad de los Andes and the Colombia GRADE Network ran the methodological process. For the difficult-stones question we carried out a systematic review with meta-analysis following the Cochrane Handbook and reported under PRISMA 2020.

We searched Medline and Embase from January 2014 — when the technique became available — to August 2025. Of 1,649 unique records, 1,362 went to title and abstract screening, 36 were read in full text, and three randomised trials with 224 patients in total met the criteria.

Three reviewers screened independently and extraction was done in duplicate. Certainty was rated with GRADE outcome by outcome, and the evidence profiles were published as supplementary material.

4What we found, including the uncomfortable part

None of the three outcomes showed a statistically significant difference.

• Successful stone clearance: RR 1.25 (95% CI 0.95 to 1.63). The point estimate favours cholangioscopy, but the interval crosses no effect. Moderate certainty.
• Procedure time: standardised mean difference 0.46 (95% CI −0.21 to 1.13), from only two of the three trials and 164 patients. Moderate certainty.
• Adverse events: RR 0.88 (95% CI 0.11 to 7.13). An interval that wide supports no conclusion at all. Low certainty.

All three trials were judged at high risk of bias, and none reported its funding source. With only three studies, publication bias could not be assessed. Two were run in the United States and one in Brazil, and one used a different size threshold from the panel's, which introduces indirectness.

The review also declares a limitation of its own: it was not prospectively registered in PROSPERO. Saying so costs something, and it makes everything else more believable.

5How you recommend on that

This is what makes the case worth reading. With no significant difference, the temptation is to recommend nothing and hand the problem back to the clinician. GRADE allows something else, because the direction and strength of a recommendation are not the same thing as statistical significance.

The panel issued two conditional recommendations, one per question. Conditional means what it says: most well-informed patients would choose this option, but many would not, and the clinician has to weigh the individual case. The certainty was low to very low, and it was declared rather than buried.

The guideline also put in writing what it lacks: economic estimates and knowledge-translation work to implement the technique locally. A guideline that names its gaps is more useful than one that pretends to have none.

6What can be checked

All three of us are authors on both papers.