The response rate is not given anywhere in the paper, but it was low - 43 units are represented and there are about 200 units in the UK. Not quoting your response rate for a survey is odd to say the least and I'm suprised the reviewers let that one go. The authors do assure us in the discussion section that "It is unlikely that the remaining units would have changed the results very much..." Well that's OK then - I've no idea what they base this pretty massive yet fantastically vague claim on!. (You can look for others yourself if interested, but theres a paper here that looks at the problem of missing data.).
Figure 1 show that units with more beds have more admissions, but doesn't look at the relationship between the two any further which is a shame but ICNARC data is there for that.
The beds per. consultant data is interesting - if the data is correct, in one unit one consultant is looking after 21 beds (we have no idea whether these are L2 or L3 because the authors haven't looked at that) which sounds horrendous. It looks like they're working with 2 trainees during the day, but 1.3 trainees overnight (one of many examples where the data doesn't add up - 21 beds, 10.5 per trainee during the day, 16 per trainee overnight).
Another massive problem is the series of data looking at PAs. In case you don't know, PAs are units of time and money. Consultants each have a job plan which is essentially a timetable of a standard week. Between 7am and 7pm, each 4 hours of work is 1PA (outside of that it's 3 hours). So if you do a 4 hour list on a Monday morning, that's easy - it's 1PA. It gets a bit muddy when you're talking about SPAs, which is the time you're paid for to do appraisal, audit, journal clubs, basically anything non-clinical. Different trusts and people within trusts can be paid different numbers of SPAs, and it doesn't always bear relationship to what people actually do. For example, if a consultant gets 2 SPAs per week, it doesn't follow they do exactly 8 hours on non-clinical work each week. Naturally, most people feel thay do more, and so it's very contencious who gets what, and what's fair etc. (I know this sounds childish and very much a first world problem until you get a consultant job!). It's similar for on-call - you get a representative number of PAs for the hours you work, but the relationship between this number and the reality is not always 100%. Because this is all linked to money, people get very tetchy about it.
So when a survey comes round asking how many PAs etc people get it gets backs up; the survey is essentially asking how much you are being paid. I took the view (rightly or wrongly) that the pay of my colleagues was none of their (or my) business. It looks like I wasn't the only one, figures 6 and 8 look like the data for the northern region has come from one unit.
So all the results to do with PAs may be incomplete, untrue, and most importantly measure the wrong thing. If only they had asked about hours of work and not money!
It does look, however, that you will be working above the standard contract of 10PAs if you do ICM.
The reporting of qualitative data in the paper is not standard, reporting unquoted anecdote. It doesn't look like there has been any thematic analysis.
The discussion is speculative, which is fair enough.
This survey was always going to be difficult as it was trying to sample a complex system - think how easy it is to explain the trainee rotas just in this region let alone nationally. Was this a reasonable attempt? - I'll leave that to you to answer. Did it reach the threshold for publication? Does it tell us anything we didn't already know?
As a final comment though, I would like to remind you that the value of a survey doesn't come from it being national, it comes from seeking the right data and presenting the results in a clear and useful way to answer the question that was defined at the very start.
Now my ranting is over, please leave a comment and we'll aim to get some discussion going. Maybe you want to post about problems you have had doing surveys, or surveys that went well? Perhaps we can share some 'tip tips'!