we know index records are present only at leaf levels in B+ tree no no. of index blocks are less hence we have to do more I/O(move the block from disk to RAM) if required index block is not in the tree(in RAM)
but in B tree we have more index blocks so I/O cost is less