What's Actually Going On
The fifth metacarpal is the outer bone of the hand, running from the wrist to the little-finger knuckle. The "neck" is the narrow section just behind the knuckle head. A punch drives force down the length of the bone into that narrow section, and it buckles.
Because the strong structures on the palm side pull the head forward while the shaft stays put, the break almost always bends with the knuckle head tipping into the palm. That is the flattened knuckle.
The reason the bend is tolerated is the joint at the other end. The fifth ray is the most mobile of the four fingers where it meets the wrist, with roughly 20 to 30 degrees of movement available there. That mobility compensates for what the neck lost. It is the anatomical reason the evidence keeps finding that the angle does not predict function, and it is also why the same deformity in the index or middle finger bone, which are rigidly fixed, would be a genuine problem.
What the bend costs is the visible contour of the knuckle, plus, at extremes, a theoretical loss of grip efficiency from the bone head sitting prominently in the palm. What it does not cost, across eighteen prospective comparative studies, is measurable hand function.
Twisting is the different problem. A bend happens in the plane you can see on a side-on X-ray, and it is either tolerated or partly remodelled. A twist happens around the long axis of the bone, does not remodel, and makes the little finger cross under or over its neighbour when the hand closes. Every conservative trial in this evidence base excluded twisted fractures, so every reassuring number on this page depends on that one check having been done.
How to Identify It
An unusual and important gap, stated plainly: across 55 retrieved papers there is no published accuracy data for any clinical test for this fracture, and no diagnostic-accuracy study at all. For the most common metacarpal fracture there is, that is a real hole. The examination below rests on anatomy and convention rather than on measured test performance, and the badges say so rather than carrying invented numbers.
- Finger cascade on making a fist (the test that actually changes management) no published data — fingertips should converge without any finger crossing. Compare with the other hand.
- Active straightening at the knuckle (for a lag or clawing) no published data — full active extension, compared side to side.
- Loading along the little finger no published data — gentle lengthwise compression reproduces pain at the neck.
- X-ray, front, angled and true side-on views no published data — confirms the pattern, rules out a break at the base or into the joint, and catches a second broken bone. The angle it measures is the number the evidence says not to act on.
Look: compare knuckle contours end-on with both fists closed. Inspect the skin over the knuckles with the fist closed as well as open, because a small wound that looks trivial on a relaxed hand can sit directly over an open joint when the fist is made.
Feel: the neck, then work back along the bone to the base and the joint at the wrist. Check the fourth bone too, since multiple fractures change the plan.
Track: total finger movement and grip as a percentage of the uninjured hand. After surgical fixation, grip runs about 68 to 71% of the other hand at 6 weeks and 91 to 93% at 12 weeks. The big knuckle joint is the slowest part to recover.
What else it could be
The Debate
There is no clinical practice guideline for this condition as of July 2026. No NICE, APTA or equivalent guideline specific to fifth metacarpal neck fractures was found, and no recent practice-changing trial. So the conflicts here are between long-standing teaching and trial evidence rather than between a guideline and a newer study. There are five, and every one runs the same way.
1. How much bend is acceptable
Long-standing teaching, from cadaver studies
Acceptable bend is about 30° maximum. Beyond that, refer for surgery.
vs
Boeckstyns 2021, systematic review of 18 prospective comparative studies
No study found any link between the bend, initial or final, and clinical result. 90% do well up to 70° in a functional brace with no straightening, with disability scores under 10 reported uniformly.
Follow the trial evidence. The threshold governing referral is a laboratory number that was never validated against a patient outcome, and cadavers cannot report how a hand works. The reviewer notes he changed his own practice a decade before publishing.
2. Whether to straighten the bone
Standard practice
Straighten it under local anaesthetic, then hold the correction.
vs
Martinez-Catalan 2020 and Kaynak 2019, randomized and prospective
The correction is lost by three weeks, with the final bend equal to people never straightened at all. A better initial correction drifted from 16° back to 21° within a month.
Follow the trial evidence. It is a procedure with a real cost whose radiographic gain has usually disappeared within three weeks, and whose functional gain was never detectable.
3. Cast or tape
Standard practice
Ulnar gutter cast for 3 to 6 weeks.
vs
Pellatt 2019, van Aaken 2016, Martinez-Catalan 2020, three randomized trials
Median disability score 0 in both arms at 12 weeks with a confidence interval of 0 to 0. More complications with casting. Between 11 and 28 days less time off work with tape.
Follow the trial evidence. Tape or a soft wrap is first-line for a closed, untwisted break up to 70°.
4. Whether follow-up is needed
Standard practice
Fracture clinic review with repeat X-rays.
vs
Bansal 2007, and Luciani 2024 across 611 fractures
Strapping plus an information sheet and no follow-up beat plaster-with-follow-up on satisfaction, with return to work at 2.7 versus 5 weeks and no difference in hand function. Zero conservatively treated cases across 611 fractures later needed surgery for failed or crooked healing.
Follow the trial evidence, with one condition. The twist-and-skin check has to have been done properly at the first visit, because that visit is now the only one. Where swelling made that check unreliable, one review at 3 to 7 days is the single genuinely useful appointment in this pathway.
5. Whether a severe bend needs surgery
Standard practice
Severe bends must be fixed operatively.
vs
Zong 2016 network meta-analysis, Zawam 2024 randomized trial, Berube 2021 low-value-care review
Conservative treatment ranked best of four options for complications, with transverse pinning worst. No difference in healing or function between a slab and pinning at one year, all patients back to pre-injury work and activity. Casting a confirmed fracture is formally classified as low-value care.
Follow the trial evidence. Surgery is for a twist, an open injury, a knuckle that cannot straighten and exceptional demand. Not for a number on an X-ray.
Honest Limitations
1. The trials are top-tier by label and unsatisfactory by method
What the research shows: six randomized trials underpin everything on this page.
The gap: a 2023 methodological appraisal scored all six. Both reviewers rated every trial as top-tier evidence and every trial as unsatisfactory on a formal methodology score, with complete agreement. The named weaknesses were randomization, blinding, group comparability, how clinical effect was measured, and allocation. Separately, the Cochrane review set validated hand function as its primary outcome and found no included trial had measured it. The field randomized patients for decades while measuring the X-ray angle, which is exactly the variable later shown not to matter.
What to do with that: trust the direction, which is consistent across a Cochrane review, a network meta-analysis, a low-value-care review and six trials. Do not treat the precise numbers as precise.
2. Every reassuring number depends on a check the trial patients had already passed
What the research shows: tape is as good as a cast, and bends up to 70° do not affect function.
The gap: all three of those trials excluded twisted, open and old fractures. The trial populations were pre-filtered by precisely the hands-on examination that is easiest to skip on a swollen hand, in a busy department, in someone who may be distressed or intoxicated and cannot cooperate with making a proper fist. A swollen hand at 24 hours is also the hand where a twist is hardest to see.
What to do with that: the twist-and-skin check is the load-bearing part of this encounter, not the X-ray. If the hand is too swollen to judge the finger cascade reliably, that is a reason to look again in a few days rather than to assume.
3. The rehabilitation evidence is thin and partly borrowed
What the research shows: a 2026 review found structured rehabilitation improves hand function, grip and movement.
The gap: it pools only five mixed-design studies across quite different treatments, and its combined grip and movement estimates are implausibly precise for that base, so it supports the direction and not the size of the effect. It also covers the second through fifth knuckle bones as a group rather than this fracture specifically. And no trial in the set has shown formal supervised sessions outperforming a home programme on the main functional measures, with the one exception of total finger movement, where a handout alone did not keep up.
What to do with that: the value here is getting early movement to actually happen, and reserving in-person sessions for the patient who is genuinely stiffening, which is where that one exception points. No trial in 55 papers tests a dosed exercise programme for this fracture, which is why the doses above are labelled as convention.
4. The evidence has been clear for twenty years and practice did not follow, and that gap was measured
What the research shows: tape is non-inferior to plaster, published in 2019.
The gap: a 2024 study audited the emergency department that ran that very trial, before and after its own publication. 69% of staff knew the findings and 57% said they had adopted them, but the proportion of patients actually receiving tape went from 6% to 28%. Roughly seven in ten still got plaster in the building that produced the evidence. The recorded barriers were social rather than clinical: fear of reprisal, needing permission, the opinions of senior decision makers, assumptions about what patients want, and a tendency to play it safe. A separate 611-fracture review found the surgery rate unchanged across eight years, and that a patient's insurance status independently raised the odds of an operation alongside the clinical findings.
What to do with that: if you are the patient, this is the page to bring with you. Knowing that tape is an evidence-based choice rather than a lesser one is most of what you need.
The Nuance
Surgery versus conservative treatment, with the actual numbers.
Beyond 70°, honestly. About 11% of these fractures heal with more than 70° of bend. The only evidence on that group is one retrospective series of 15 patients at an average of 32 months, in which all 15 rated their hand function "very good" and 80% reported zero disability, in a group where nearly half were manual labourers and 87% had injured their dominant hand. That is a weak study pointing the same way as a strong one. It is why the recommendation there is "reasonable and discuss it" rather than "proven".
The honest bottom line. Almost nobody with this fracture needs an operation, and the version most often done through the skin is the one with the highest complication rate. What surgery reliably buys is the shape of the knuckle. Boeckstyns's review put it plainly, finding operative treatment showed no benefit over non-straightened conservative treatment "other than aesthetic issues", and van Aaken states the other half of the same trade, that a patient treated with a soft wrap must be willing to accept losing the knuckle contour. That is the real conversation, and it is a cosmetic one. It deserves to be had openly rather than smuggled in as a clinical necessity. A young patient who genuinely minds how their hand looks is entitled to know the options exist and are time-limited. A patient who does not mind is entitled not to spend six weeks in a cast over it.
