Comparison of the sensitivity and specificity of two cardiovascular risk scales in patients with idiopathic peripheral facial paralysis: Globorisk and Lindavista Score

Abstract

Introduction: Recently, it has been proposed to measure cardiovascular risk (CVR) in Mexico using the Lindavista Score (LS) adapted to the specific characteristics of our population, in contrast with the international Globorisk scale (GR), which appears to underestimate CVR by not considering these characteristics. To our knowledge, LS has not been applied to Mexican patients with idiopathic peripheral facial paralysis (IPFP), and it has not been compared with GR.

 

Objective: To estimate the CVR rates obtained by both scales and compare their results in a sample of Mexican patients with IPFP classified using the House-Brackmann (H-B) scale.

 

Hypothesis: We hypothesized that LS would be more sensitive than GR in identifying cases with higher CVR depending on the gender of the patients.

 

Material and Methods: With written informed consent (protocols 8619 and 9022) approved by the research committee of INR LGII, 138 patients with IPFP were studied. CVR was measured with both scales, and the triglyceride/HDL-c ratio (TG/HDL-c) was taken as the reference criterion, considered the most sensitive prognostic factor to date. Both the reference criterion and the scales were dichotomized to reduce data dispersion, clarifying that the GB scale only identified cases of medium and low risk and none of high risk. The concordance between the scales was evaluated using the Kappa coefficient. The intensity of associations was calculated using OR (CI 95%), and these were compared using the Mantel-Haenszel stratified analysis (MyH). A p<0.05 was considered significant.

 

Results: Of the total cases, 54.3% were female, with a global average age of 49.2 +/- 14.2 years. The prevalence of CVR estimated with the TG/HDL-c ratio >3.0 was 61.6%, with an average ratio of 4.2 times higher TG compared to HDL-c. With GB, the prevalence of medium CVR would be 15.9%, and with LS, the CVR levels 3-4 would be 65.2%. The Kappa concordance coefficient of GB concerning the TG/HDL-c reference criterion was 0.06 (p = 0.24), and that of LS was 0.206 (p = 0.01). The concordance between GB and LS had a Kappa of 0.136 (p = 0.003). However, GB had a sensitivity of 18.8% for detecting true positives and a specificity of 88.7%; in contrast, LS had a sensitivity of 72.9% with a specificity of 47.2%. The LS was stratified by gender: in females, sensitivity increased to 74.4% and specificity to 52.8%; in males, it was 71.7% and 35.3%, respectively. The OR for females was 3.2 (1.2-4.9, p = 0.01) and for males 1.3 (0.4-4.5, p = 0.58), with a chi-square of MyH 4.1 (p = 0.04).

 

Conclusions: The Lindavista Score (LS) is much more sensitive than Globorisk (GB), but GB is more specific. As screening tests for CVR in Mexican patients with IPFP, our results suggest a sequential use: first applying LS and then GB. Although ideally, the Lindavista Score should be improved to balance sensitivity and specificity in the correct identification of CVR, which is increasingly prevalent.