The 28uA that is mentioned in the AppNotes is the average value of current consumption when the advertising interval is 1 second.
In your case, you are measuring the deep-sleep current of 2uA which is correct. But I guess you are not be able to measure the current when the advertisement takes place, i.e. when the RF transmission takes place. The current will be much higher(in the range of mA) during this state. This duration is very small(around 8 milli-seconds) and this current might not be picked up by your multimeter because of the small duration.
When you average out these two currents, i.e. the 2uA deep sleep current and the current during advertisement, you will be seeing a ball park value of 28uA.
If you use a mutltimeter such as Keysight 34410A, where you can change the aperture, you will be able to see the average current value.
Also note that, the 28uA is the average current observed for the example projects in the AppNote. The current consumption will vary based on your application, components used, low power code used.
This solved my doubt, thanks for your reply yssu.
However, there is another thing that I would like to confirm. I am trying to measure the duration of the BLE advertising. I used the firmware timer to time the time interval between the CPU wake-up from deep-sleep and CPU entering deep-sleep. The result was sent through UART to my pc terminal and the result updated exactly every 1 second which matches the advertising interval. But, the value of the time interval I obtained was around 0.260ms which is quite small. Not sure whether this is a correct way to measure the duration of the BLE advertising.