Page 266 - 20211214_OHKF_Health_Finance_Research_Report_E (1)

APPENDIX S: SCREENING TEST PERFORMANCE AND
ACCURACY CONCERNS

The administration of either FPG or HbA1c is the recommended protocol as the first line for
screening tests in Hong Kong. Yet, more recent evidence suggests there are accuracy
concerns relating to HbA1c as well as FPG (Barry et al., 2017; Heianza et al., 2011). The
implications of inaccurate testing are two-fold: i) screening may result in an incorrect
diagnosis (false positive) and intervention is offered to those that do not require it; or ii)
screening may result in no diagnosis and timely intervention may not be offered to those in
need (false negative). It is therefore imperative that available screening tests attain high
sensitivity (measures probability of true positives) and specificity (measure probability of true
negatives) to reduce the number of false positives and false negatives.

In a leading study by Barry and colleagues (2017), a systematic review and meta-analysis of
empirical studies evaluating accuracy of tests for identification of pre-diabetes was
presented. The study reviewed 148 international publications and extracted data
from 46 papers (inclusive of studies with WHO and ADA criteria) specifically to construct
the diagnostic accuracy for a meta-analysis and presented findings on sensitivity and
specificity for FBG and HbA1c tests. One of the key principal findings concluded that the
diagnostic accuracy of tests used to detect pre-diabetes in screening programme is low,
specifically that HbA1c is neither sensitive 0.47 (95% CI; 0.37 to 0.58) nor specific 0.81 (95%
confidence interval 0.74 to 0.86); while FPG is specific 0.95 (95% confidence interval 0.93 to
0.97) but not sensitive 0.24 (95% confidence interval 0.17 to 0.32) (Barry et al., 2017). In sum,
the findings illustrate that both tests are low in sensitivity rendering a higher
number of false negatives, therefore screened individuals are at greater risk of being
falsely reassured and not given timely intervention. In terms of specificity however, FPG
should perform better while HbAc1 may lead to a high number of false positives, therefore
the chance of being misdiagnosed and given intervention is a higher probability for the latter
test.

The Hong Kong Reference Framework for Diabetes Care for Adults in Primary Care Settings
suggests that either FPG or HbA1c tests can be sufficient in the two rounds of screening
test. While no screening test is 100% accurate, there is no universal evidence as to
which test is the most accurate including the trade-offs between accuracy, costs
and feasibility. In view of this, our prospective screening programme confronts this challenge
by aligning with the recommendation stipulated under our local guidelines with the added
component of combining FPG+HbA1c tests. Therefore, the screening programme consists
of two rounds of screening, with at least one round of combined
FPG+HbA1c tests, followed by either a repeat of both, or both tests in
addition to OGTT.

In the case of Hong Kong, we recommend the added component of combining FPG+HbA1c
tests as a feasible alternative to partly overcome the associated inaccuracies
presented within each screening test. For example, taking the case of an
asymptomatic individual that receives an abnormal value for fasting plasma glucose or HbA1c
value, a repeat of either test is required. However, in the case of an individual that does not
present an abnormal value, the performance of screening test based on FPG and HbA1c
results together versus on the basis of a single FPG or Hba1c value alone should perform
better. Therefore, presented with the challenge of a low sensitivity (many false negatives)
observed in both tests, the approach to combine the two tests is a method to reduce the
inaccuracies associated with a case in which false negative may arise.

264