I want to assess county differences in medication rates in Norway to detect outliers. For this purpose I want to use funnel plots as recommended by Spiegelhalter (see e.g. The Art of Statistics). I am trying this on publically available Norwegian data. However, I have worked on this for some time now and I am not sure whether I am specifying the funnel plot correctly, and I wonder if any of you have some input on this.
Specifically, I am having some trouble with (i) specifying the correct weights (the plot command only works with the -nowei- option), and (ii) the confidence intervals seem to narrow, although I suspect this may be due to sample size, not my specification of the funnel plot.
It's county data for a year so it's a small sample with 18 observations. The medication rate is by 1000 individuals per county. I've gathered population data per county, and made a population rate in 1000's so it's scaled to the medication rate.
I obtain the standard deviation of the medication rate by
Code:
egen sd_medrate=sd(medrate1000)
Code:
funnelcompar medrate1000 poprate1000 county sd_medrate, cont nowei
Array
Without -nowei- I get the following error message:
Code:
may not use noninteger frequency weights
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input long county float(county_pop medrate1000 poprate1000 sd_medrate) 4 72856 6.37 72.856 1.6318192 12 107080 4.68 107.08 1.6318192 2 108499 9.68 108.499 1.6318192 14 156494 4.76 156.494 1.6318192 13 168231 7.46 168.231 1.6318192 16 170377 6.4 170.377 1.6318192 9 185216 5.56 185.216 1.6318192 5 190709 8.24 190.709 1.6318192 17 231286 7.83 231.286 1.6318192 8 236271 9.19 236.271 1.6318192 7 251262 5.87 251.262 1.6318192 3 257673 6.28 257.673 1.6318192 18 271662 7.59 271.662 1.6318192 15 422102 5.61 422.102 1.6318192 11 427947 6.51 427.947 1.6318192 6 477175 4.79 477.175 1.6318192 1 536499 5.32 536.499 1.6318192 10 586860 3.62 586.86 1.6318192 end label values county fylke label def fylke 1 "Akershus", modify label def fylke 2 "Aust-Agder", modify label def fylke 3 "Buskerud", modify label def fylke 4 "Finnmark", modify label def fylke 5 "Hedmark", modify label def fylke 6 "Hordaland", modify label def fylke 7 "Møre og Romsdal", modify label def fylke 8 "Nordland", modify label def fylke 9 "Oppland", modify label def fylke 10 "Oslo", modify label def fylke 11 "Rogaland", modify label def fylke 12 "Sogn og Fjordane", modify label def fylke 13 "Telemark", modify label def fylke 14 "Troms", modify label def fylke 15 "Trøndelag", modify label def fylke 16 "Vest-Agder", modify label def fylke 17 "Vestfold", modify label def fylke 18 "Østfold", modify
0 Response to Specification of funnel plot with -funnelcompar-
Post a Comment