Archery injuries: Konda 2023 study

A query of the United States emergency department surveillance database, NEISS, for archery-related injuries between 2013 and 2022: 43,083 over ten years, an average of about 4,308 a year. Of these, 35,367 concern men and 8,038 women.

The peak for men falls in the 10 to 19 band, highest at fifteen, and for women in the 30 to 39 band, highest at thirty-six; in both sexes the count rises again after fifty. About 78 per cent of the patients are white, 2 per cent Black, 4 per cent report more than one background.

In men the site is the finger in 66.2 per cent of cases (23,342), the hand in 17.1 (6,012), the face in 6.2 (2,122), the shoulder in 5.4 (1,909), the foot in 5.1 (1,803). In women the database reports the finger in 100 per cent of cases. The commonest diagnosis in men is a cut, 58.1 per cent, then fracture 12.7, contusions and abrasions 7.5, foreign body 5.9, puncture 4.9, strains and sprains 4.9, and 6 per cent in a residual category. In women cuts cover 100 per cent. No patient was admitted: all were treated, examined and discharged.

From these numbers the abstract concludes that men are 4.4 times more likely to be injured. The ratio between the two counts, 35,367 over 8,038, comes to exactly 4.4: it is a ratio of cases, not of risks, and the discussion of the same article acknowledges this in a sentence the abstract leaves out, where it reads that it cannot be concluded that men are more susceptible, because archery is a sport predominantly played by males.

The conclusions propose reinforced gloves, finger tabs and forearm braces, and as future research a comparison between archers using different combinations of protection.

NEISS is built by the United States Consumer Product Safety Commission and gathers trauma cases arriving at emergency departments that can be attributed to consumer products, from about a hundred hospitals across the country. The authors picked archery within the sports and recreation equipment group, code 1235, which covers equipment, apparel and activity, and queried the years 2013 to 2022 without filtering by age, sex or background.

The search included every body site, listed one by one with its code: head, face, eyeball, mouth, neck and ear for the upper region; shoulder, elbow, lower arm, wrist, upper arm, hand and finger for the upper limb; pubic region and lower trunk; knee, lower leg, ankle, upper leg, foot and toe for the lower limb. On the diagnosis side every code was queried, with those carrying numerical data then used for the analysis: contusions, foreign bodies, fractures, lacerations, punctures, strains and sprains, and a residual category. Dispositions were analysed as well, to separate those treated and released, transferred, admitted, or who left without observation.

For a secondary analysis the sites were grouped into six categories: head and neck region, upper extremities, lower trunk and pubic region, lower extremities, a quarter or more of the body surface, upper trunk. The processing was done with JMP Pro 15.

In the section given to statistical analysis the authors state that they examined sex, anatomical locations, age bands and diagnoses. No test follows: the article contains no p-values, no confidence intervals and no models, and every result is a frequency or a percentage. It is not reported how many emergency department records were actually read, nor whether the 43,083 are records or national estimates derived from the database's sample weights.

The central sentence of the abstract is a ratio between counts presented as a ratio between risks. 35,367 male cases over 8,038 female ones come to 4.4, but without knowing how many men and how many women shoot, that number says nothing about who gets injured more. The denominator sits in the discussion of the same article, where the authors report that participants in the United States are male by three to one, and a few lines later write that it cannot be concluded that men are more susceptible. The claim is therefore made and withdrawn in the same paragraph, and only the first half survives into the abstract. Dividing their own two numbers, 4.4 cases against 3 participants, gives about 1.5, which is as much as their data support.

What those numbers are is never stated. NEISS collects from about a hundred hospitals and provides national estimates through sample weights: 43,083 over ten years has the order of magnitude of a weighted estimate, not of a count of records, while the article calls them recorded injuries and computes percentages and ratios on them with no confidence interval anywhere. Either reading creates a problem. If they are records, they cannot be set against the 3.2 million national emergency visits for sports injuries, as the discussion does. If they are estimates, the uncertainties are missing, and with them any way of knowing which cells hold.

The figure for women shows which of the two readings is the right one and how thin the cells are: 100 per cent of injuries at the finger and 100 per cent lacerations. The authors place it among their limits, writing that it does not account for the variety of injuries that probably affect women, and meanwhile the same figure appears in the abstract as a result.

The arithmetic does not close, at four points. 35,367 plus 8,038 make 43,405, while the stated total is 43,083. The dispositions, 33,769 men and 7,692 women, add to 41,461 against a total for which it is claimed that everyone was treated and discharged. The five sites listed for men sum to exactly 100.0 per cent although the search covered some twenty regions, and their cases add to 35,188 against a male total of 35,367. And the claim that this amounts to about one sports injury in six hundred and seventy does not follow from the article's own numbers: 3.2 million divided by 4,308 a year gives about one in seven hundred and forty.

In the list of diagnosis codes, contusions and foreign bodies carry the same code 53. It is the list the database was queried with, so anyone wanting to repeat the search cannot do so from what is written.

The title promises risk factors and no risk factor is examined: nothing relates equipment, experience, bow type or anything else to injury, and the section on statistical analysis is followed by no inferential analysis at all.

The population is not the one the rest of the registry deals with. The product code covers archery equipment, apparel and activity, so recreational shooting and hunting are inside the count, and the shape of the data says as much: finger lacerations as the leading diagnosis, a peak at fifteen and a second one after fifty. The conclusions, addressed to athletes, are drawn largely from people who are not competitive archers.

Only what reached an emergency department enters, which is the authors' first declared limit, and it cuts exactly where archery research looks: a shoulder overload or a tendon problem goes to a physiotherapist or a sports physician, not to an emergency room. The 5.4 per cent for the shoulder in this series does not compare with case series gathered in clinics. Severity, finally, is not in the database, so the fact that everyone was discharged describes the coding rather than the extent of the damage.

Three citations remain that do not support what they are attached to. The sentence about over twenty-three million American adults who shoot is credited to a 1999 archaeology article on the adoption of the bow and arrow in central Arkansas, while the figure comes from the industry report cited elsewhere. The sentence on higher male attendance points to a page from a job listings website about the occupation of archer. And the claim that earlier epidemiology had found higher rates in men points to a case series of crossbow injuries to the thumb.

What one takes away from this work is an order of magnitude. In a country where more than twenty million people shoot, emergency departments see about four thousand three hundred archery-related cases a year, and none of those cases, across ten years of data, required admission. That is little, and it fits the idea of archery as a low-risk sport, without ever becoming a rate.

The second thing is where people who end up in hospital get hurt, and it is not where an archer would expect: fingers and hands first of all, with two thirds of male cases at the finger, and cuts as the commonest diagnosis. The shoulder is a twentieth of the total. This is the profile of someone handling arrows, not of someone training: recreational shooting and hunting are in the count, where sharp points are handled, and the repeated-load problems are not visible, because they are treated elsewhere.

For an archer, the useful thing these numbers point at is the handling of arrows. Pulling, nocking, carrying them in the hand: in the data that is what sends people to hospital, more than any act of shooting. On protective gear this card draws no recommendation, because the study compares no item of protection against any other, and the authors themselves name that as research still to be done.

What is not here is everything else: no probability of being injured, because the denominator is missing; nothing on overload, because overload does not reach an emergency room; nothing specific to competitive archers. The registry keeps this work for its order of magnitude and for the distribution of acute injuries in a mixed population, and not as a measure of risk.