next up previous contents index
Next: Results Up: Established Complications II Previous: Specification

Tests

Due to the fact that this statistical function only counts the number of database  fields for certain periods of time that have a value  of y or of n, there are not a lot of good tests which can be done. So basically the same steps as described above in the patient's section were carried out again.

Once more, this statistic is also produced for each quarter and each year. So it is not possible that more patients have been medicated in a certain year than in the sum of the year's four quarters. Nevertheless, this happens here, and so a bug has to be in this module. To locate this fault, the input was simplified, and several tests were made with data for two years only. The results of one of those two years was always correct, but the result of the second year was too high. A more detail inspection made obvious that the second year's result was increased by the first one's. If a third year was added, the same happened: One year was correct, one's result was increased by the outcome of the first, and the last year was the sum of all three.

Next, a boundary value analysis  was done which led to a success when only empty fields were entered for one period of time into the   table neuropat. In this case, no output should be produced, since empty fields had -- according to the specification -- to be ignored, and all fields were empty. The function still produced an output for that period of time with the number of patients set to zero. DPV  has a function to display the statistics  graphically. This feature is not able to handle database tables which have a not positive number of  patients correctlygif.


next up previous contents index
Next: Results Up: Established Complications II Previous: Specification

Ingo Melzer
Mon Aug 5 15:12:01 MET DST 1996