In medical qualitative research, medical researchers analyze historical patient data to verify known relationships and to discover unknown relationships among medical attributes. All the existing algorithms to solve this problem use measures which are asymmetric measure, so only one direction of the rule (P -> Q or Q->P) is taken into account. However, medical researchers are interested to find both asymmetric and symmetric relationship among medical attributes. We have developed pruning strategies and devised an efficient algorithm for the symmetric relationship problem. We propose measuring interestingness of known symmetric relationships and unknown symmetric relationships via the correlation measure of antecedent items and consequent items. We have demonstrated its effectiveness by testing it on real dataset.