Error in reporting of afl injury surveillance data?

[hhh]

Hey, I've just had a look at the most recent AFL injury surveillance available which reports on data up to and including the 2003 season and compared it to the data present in Ochard and Seward 2002 (1997-2000 data). When comparing the injury incidence tables (new injuries per club per season), the figures presented for certain injuries are different. Has anyone else picked this up before, are you aware of it - have I messed up somehow? Which data is correct?

[injuryupdate]

Most of the data should be the same. However, from year to year certain diagnoses get changed from one category to another. E.g. posterior knee tendinitis might be considered a "hamstring injury" one year, and then down the track it is decided that it will be coded as "other knee condition". This is just an example.

Occasionally a major error will be uncovered and corrected in later years (e.g. a genuine miscode - a hamstring injury coded as a quadriceps strain and then recoded later back to the correct diagnosis of hamstring strain).

I would go with the later data when comparing year to year. Hopefully the overall injury prevalence and rates of the major categories change only marginally due to recoding or reanalysis at a later date.

[hhh]

So go with the latest as opposed to the medline index are you saying? One injury I noticed that had a massive change was lumbar and thoracic spine injuries - it is almost double in the BJSM paper. Lumbar injuries are probably the flavor of the week at the moment. Looking at the figures it is one of the more common injuries but as it isn't as big as the hamstrings, knees and groins it is probably an injury that has slipped under the radar. However the chagne in diagnosis may definitely effect this injury - where does low back finish and hamstring start in terms of injury. Having said this the drop off in lumabr injuries doesn;t correspond with an increase in hamstring.

How can the data change each year though? Isn't it punched in once, saved and then uncahnged for statistical evaluation?


Most of the data should be the same. However, from year to year certain diagnoses get changed from one category to another. E.g. posterior knee tendinitis might be considered a "hamstring injury" one year, and then down the track it is decided that it will be coded as "other knee condition". This is just an example.

Occasionally a major error will be uncovered and corrected in later years (e.g. a genuine miscode - a hamstring injury coded as a quadriceps strain and then recoded later back to the correct diagnosis of hamstring strain).

I would go with the later data when comparing year to year. Hopefully the overall injury prevalence and rates of the major categories change only marginally due to recoding or reanalysis at a later date.

[sydunisportsmed]

Strictly speaking, if an error became apparent in the data and a diagnosis had to be changed, authors should write back to the journal and let them know, and have an erratum published. If it was one diagnosis out of 600 injuries and it caused a category to change in incidence by 0.1 per season, it wouldn't be worth it (making the report).

Thanks for the tip though on lumbar spine category changing by so much. It is probably the case that a larger chunk of 'back-related' diagnoses have been switched, and it would be worth detailing this change, if it has happened, in the next report (currently being written up).

[hhh]

Thanks, I'd appreciate a mention in the surveillance (plus you get a free chance to plug injuryupdate.com.au). Out of interest how much do you get for doing this each year? Just give a ball park - 50-100K?

Thanks for the tip though on lumbar spine category changing by so much. It is probably the case that a larger chunk of 'back-related' diagnoses have been switched, and it would be worth detailing this change, if it has happened, in the next report (currently being written up).[/QUOTE]

[injuryupdate]

Ball park figure is somewhere between 0-50K (in the bands that you have mentioned).

What reference would you like mentioned in the next report?

(Can be non-peer-reviewed but something a bit more solid than a Forum post).

[hhh]

HHH, personal communication will suffice



Ball park figure is somewhere between 0-50K (in the bands that you have mentioned).

What reference would you like mentioned in the next report?

(Can be non-peer-reviewed but something a bit more solid than a Forum post).