Hi, from previous literature , i can obtain the data in the form of sensitivity, specificity, negative and positive predictive values. However, i would require exact numbers of false positive, false negative, true positive and true negative cases, can someone suggest formulas?
Number of Patients/100% = Number of True Positives/Sensitivity [%]
1. Multiply Number of Patients with Sensitivity [%]
2. Divide by 100%
3. Result is number of patients with true positive results (True Positives)
4. Number of Patients minus True Positives = False Negatives
Same with Specificity:
Number of Healthy Controls/100% = Number of True Negatives/Specificity [%]
1. Multiply Number of Healthy Controls with Specificity [%]
2. Divide by 100%
3. Result is number of true negative results (True Negatives)
4. Number of Healthy Controls minus True Negatives = False Positives
Use Excel and the Solver add in
Set 4 cells to contain your 4 different patient groups and 4 cells to calculate sensitivity/specificity/PPV/NPV
Add constraints in Solver for 3 of the answers (eg set PPV, NPV & sens to your known values) and a constraint that your 4 patient groups are all integers. If you know your total patient numbers, add this as a constraint as well.
Finally, solve for specificity using your known value.
However, be warned that if the quoted percentages are rounded then the values you get may not work well. You can select the options to adjust the precision of your constraints. It may also give you a fraction of you numbers, so for example it may give you the results for 150 patients when in fact there were 300
Try that and see how you get on
Richard
Dear Torsten and Richard, Thanks a lot for your suggestions and solutions. It really worked out.
Hi, from 2x2 table, you can obtain values like sensitivity, specificity, +ve PV, -ve PV, accuracy of the test, PLR& NLR. All these indicators are calculated based on the observations of how a test behaves in the presence/ absence of the phenomenon under study eg. the presence or absence of a particular disease.
So, if that test gives you a positive result in the presence of the phenomenon (or a disease), then this is the "true positives", while all diseased subjects will be considered as "false negatives" when the test result is negative. You can imagine what will be then 'false positives" and "true negatives".
So, you can use the following equations to find what is related to your queries:
(please find the attached files)/ also you can refer to may attached published paper about the performance of "serum fructosamine" in Dx. of T2DM
Good Luck
- Badges
- Science topic