Page 40 - PATIENT REGISTRY DATA FOR RESEARCH: A Basic Practical Guide
P. 40

to the sixth and seventh digits' number) by referring to the 12-digit number on the person's

               identification card.


                       Although it is always possible to have erred by determining someone's ethnicity by

               basing on his/her name alone, the above-mentioned method is considered a more objective


               way to determine a person's ethnicity, which makes it more relevant for the Malaysian

               patients. Apart from this, it is usually possible to decipher a person's gender based on his/her


               name alone. However, the researcher must also have to accept the fact it is also possible for a

               deviation from the truth to occur if he/she is imputing a 'missing' gender by basing on the


               person's name alone. Thus, it is necessary to declare such limitations in the study report or

               manuscript in order to make them transparent to all the readers.


                       On the other hand, imputation is a technique which replaces the missing data with a

               probable value which has been estimated by mathematical computations. There are many

               imputation techniques which can be applied in patient registry databases, such as the single


               imputation methods (including mean imputation, median imputation, conditional mean

               imputation) and the multiple imputation methods (including hot decking method and multiple


               imputation by chained equations). However, a full discussion of all these imputation

               techniques is beyond the scope of this paper, and so the reader is advised to look up other


               relevant sources for further information about these techniques.

                       The decision to apply imputation techniques to rectify the error caused by missing


               data also depends on the objective of the study. In some technical reports, it is necessary to

               report missing data in order to evaluate the competency of the registry for collecting


               information. Hence, imputation will not be necessary in such instances. When imputation is

               deemed necessary in other circumstances, then the researcher shall have to select an

               appropriate imputation technique and to provide a valid justification for it.
   35   36   37   38   39   40   41   42   43   44   45