ࡱ > Q P \ p hiltya7b B a = = hY;[{*8 X@ " 1 6[SO1 6[SO1 6[SO1 6[SO1 6[SO1 6[SO1 6[SO1 $ 6[SO1 6[SO1 6[SO1 6[SO1 6[SO1 6A r i a l 1 6A r i a l 1 6A r i a l 1 6A r i a l 1 6A r i a l 1 6A r i a l 1 6[SO1 . 6T i m e s N e w R o m a n + " " # , # # 0 ; " " \ - # , # # 0 5 " " # , # # 0 ; [ R e d ] " " \ - # , # # 0 7 " " # , # # 0 . 0 0 ; " " \ - # , # # 0 . 0 0 A " " # , # # 0 . 0 0 ; [ R e d ] " " \ - # , # # 0 . 0 0 i * 2 _ " " * # , # # 0 _ ; _ " " * \ - # , # # 0 _ ; _ " " * " - " _ ; _ @ _ . ) ) _ * #,##0_ ;_ * \-#,##0_ ;_ * "-"_ ;_ @_ y , : _ " " * # , # # 0 . 0 0 _ ; _ " " * \ - # , # # 0 . 0 0 _ ; _ " " * " - " ? ? _ ; _ @ _ 6 + 1 _ * #,##0.00_ ;_ * \-#,##0.00_ ;_ * "-"??_ ;_ @_ \$#,##0_);\(\$#,##0\) \$#,##0_);[Red]\(\$#,##0\) \$#,##0.00_);\(\$#,##0.00\)% \$#,##0.00_);[Red]\(\$#,##0.00\) "Yes";"Yes";"No" "True";"True";"False" "On";"On";"Off"] , [ $ - 2 ] \ # , # # 0 . 0 0 _ ) ; [ R e d ] \ ( [ $ - 2 ] \ # , # # 0 . 0 0 \ ) 0.0000 0.000 , * + ) @ + @ @ # T D D ! @ 3 D D - D D - ` + Sheet1 X Sheet2 g Sheet3 V V " 2 Time interval months (t) 7 Sci(ts) 7 Rci(ts) 7 Cci(t) 7 Rci(t) 7 Dci(t) 7 Sri(ts) 7 Rri(ts) 7 Cri(t) 7 Rri(t) 7 Dri(t) 7 ln(HRi(t)) 7 var[ln(HRi(t))] 7 1/var[ln(HRi(t))] 7 ln(HRi(t))/var[ln(HRi(t))] 7 ln(HR) var(ln(HR)) se(ln(HR)) HR Low CI Upper CI Hazard Ratio Meta-analysis Spreadsheet - Developed by Hans Messersmith using the methods in Parmar et al, Statistics in Medicine 1998;17:2815-2934j How the spreadsheet works: Each line in the table below represents a different method presented by Parmar et al to determine the hazard ratio given a set of data from a report of a randomized controlled trial. For all methods, the boxes in light yellow are descriptive only, and play no part in the calculations. The boxes in darker yellow represent the necessary data elements for the method represented by that line of the spreadsheet. For example, the first method (Eqn 7 from Parmar), requires the hazard ratio and the lower and upper confidence intervals. The calculated ln(HR) and its variance and standard error can be found in the blue highlighted section. The green highlighted section is confirmatory; the values here are back calculated from the ln(HR) and its standard error and allow you to check the face validity of those numbers. The data in the table below are examples only. How to use the spreadsheet: First, save a copy of this spreadsheet to another directory with a name appropriate to the systematic review/meta-analysis you are conducting. Then, for each trial, select one of the lines in the table below for which you have all the necessary data in the report (i.e. you can fill all the dark yellow boxes for that line). If you could use more than one method, given the reported data, select the line nearest the top; methods nearer the top of the list are likely to provide more accurate results compared to lines lower on the list. Copy that line to the Data Area below the Base Table. In the copied line in the Data Area, replace the example data in the yellow boxes with the correct data from your trial. Once the correct data is entered you can read off from the blue area the ln(HR) and its standard error for use in Review Manager or other meta-analysis software, and check the validity of your calculations by comparing your entered data tob the values in the green area. Repeat this procedure for each trial. Make sure to save your work.H BASE TABLE - MAKE NO CHANGES HERE - COPY APPROPRIATE LINES TO AREA BELOW Trial Arms Event p Events Total Rand Trmt Rand Control Events Trmt Events Control Parmar Eqns Albain 2004 GT vs. T OS Example 2 6, 12 6, 11 4, 12, 14 4, 11, 13O NOTE: the method highlighted in orange is not appropriate if the reported HR isS NOTE: methods used in the last two lines (purple box) do not indicate the DIRECTION8 exactly one. In that case, you must use another method.S of the hazard ratio, only its magnitude. You will need to establish its direction [ by some other method (e.g. inspection of the survival curves). The direction is determined by the sign of the ln(HR). CHINV 7 # P value 7 e g f r - 7 egfr(+) 7 0,100 0.63745,100 1.2749,100 1.43426,100 2.39044,100 2.55024,99.7214 3.50598,100 3.02833,99.7214 4.30279,100 3.03144,97.7716 5.41833,100 3.03543,95.2646 5.89641,93.8719 3.03899,93.0362 3.20234,90.5292 3.68221,89.415 7 6.05578,99.1643 4.16163,88.5794 4.48213,87.4652 6.21514,82.7298 4.80396,85.5153 \ N*Nepvpe