Concordance between Two Versions of World Health Organization/International Society of Hypertension Risk Prediction Chart and Framingham Risk Score among Postmenopausal Women in a Rural Area of Bangladesh

Lingkan Barua, Mithila Faruque, Palash Chandra Banik and Liaquat Ali

Indian J Public Health 2019;63:101-6. DOI: 10.4103/ijph.IJPH_178_18

http://www.ijph.in/article.asp?issn=0019-557X;year=2019;volume=63;issue=2;spage=101;epage=106;aulast=Barua

Dr Lingkan Barua, Bangladesh

1) Summarize your work in one sentence.

Here we evaluate the level of concordance among different risk prediction tools (WHO/ISH risk charts and Framingham risk score) to find out the cost-effective and suitable one that can be applied in a low-resource setting.

2) Summarize your findings in one sentence.

The “without” cholesterol version of WHO/ISH risk chart showed high concordance against the “with cholesterol” version and the Framingham risk score (FRS). Predictability of CVD risk positive (≥10%) cases was similar for both the versions of WHO/ISH risk charts. However, the “without” cholesterol version of WHO/ISH risk chart overestimated and underestimated less number of participants compared to “with” cholesterol version.

3) Which were the more important methods you used in this work? If it is not a traditional method you can briefly explain the concept of that methodology.

We used modified STEP-wise approach to Surveillance of Noncommunicable Diseases risk factors (STEPS) questionnaire of WHO to collect behavioral information. Cardiovascular risk was estimated using both versions (“with” and “without” cholesterol) of WHO/ISH risk charts and FRS.

Concordance between WHO/ISH risk charts and FRS was evaluated using Cohen’s Kappa (κ), Prevalence-Adjusted Bias-Adjusted Kappa (PABAK) and First-order Agreement Coefficient (AC1). In addition to Cohen’s Kappa, we used PABAK and AC1 as Kappa is highly influenced by prevalence and bias of the two tools. We also reported bias index (BI) and prevalence index (PI) to overcome the limitation of Kappa. Here PABAK adjusted the imbalances caused by differences in the prevalence and bias. On the other hand, AC1 overcome the phenomenon which is well-known as kappa paradoxes (low Kappa at high agreement and high kappa at unbalance marginal distribution).

4) What did you learn from this paper, what was your take-home message?

Findings of our study revealed that the “without” cholesterol version of WHO/ISH risk chart has the potentiality to use in a minimal resource setting to estimate CVD risk and thereby saving a greatest number of lives at lowest cost.