blood sugar ran out and some participants
could not be tested during the second survey
data collection. Thus, the average of the blood
sugar level was just based on three measures,
which reduced the power of the test for
association between anthropometric factors
and diabetes. Last, some elderly could not
provide us information about their diseases,
so the data of their disease history, that was
previously diagnosed by a physician, could
not be collected completely, resulting in the
loss of relevant data for analysis.