Predictive model to estimate ionized calcium from routine serum biochemical profiles in dogs

Danner, Julie Ann

Permalink

https://hdl.handle.net/2142/97685 Copy

Description

Title

Predictive model to estimate ionized calcium from routine serum biochemical profiles in dogs

Author(s)

Danner, Julie Ann

Issue Date

2017-04-12

Director of Research (if dissertation) or Advisor (if thesis)

Ridgway, Marcella D.

Committee Member(s)

• Rubin, Stanley I.
• Solter, Philip

Department of Study

Vet Clinical Medicine

Discipline

VMS-Veterinary Clinical Medcne

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Calcium
• Hypercalcemia
• Ionized
• Veterinary

Abstract

Ionized calcium is the gold standard to assess calcium status in dogs, but measurement is not readily available in private veterinary practices. The objectives of this study were to (1) predict ionized calcium concentration from serum biochemical values and (2) compare the diagnostic performance of predicted ionized calcium (piCa) to those of total calcium (tCa) and two corrected tCa formulae; and (3) study the relationship between biochemical values and variation of measured ionized calcium (miCa). This was a cross-sectional study. Records from 1,200 dogs who were patients at the University of Illinois Veterinary Teaching Hospital were randomly selected from a population of 1,719 dogs with mical and biochemical profile performed within 24 hours for the creation of a multivariate adaptive regression splines (MARS) model, with the final model being determined by backward elimination. Accuract and diagnostic performance of piCal and its prediction interval (PI) were tested on 519 dogs via Bland-Altman analysis, Pearson’s R, and receiver operator characteristic (ROC) curves. The final model included creatinine, albumin, tCa, phosphorus, sodium, potassium, chloride, alkaline phosphatase, triglycerides, and age, with tCa, albumin, and chloride having the highest impact on miCa variation. Predicted ionized calcium was better correlated to miCa than tCa and corrected tCa, and its overall diagnostic accuracy was significantly higher to diagnose hypocalcemia and improved for hypercalcemia. The average difference between the piCal and miCal was 0.002 +/- 0.080 mmol/L. The PI included miCal 94% of the time. For hypercalcemia, piCa was as sensitive (64%) but more specific (99.6%) than tCa and corrected tCa. For hypocalcemia, piCa was more sensitive (21.8%) than tCa, and more specific (98.4%) than corrected tCa formulae. Positive predictive values of piCa were high for both hypercalcemia (90%) and hypocalcemia (70.8%). Predicted ionized calcium can be obtained from readily available biochemical and patient variables, and seems more useful than tCa and corrected tCa to approach calcium disorders in dogs when miCa is not available. A webpage has been designed for piCa calculation (http://vetmed.illinois.edu/study/mars-model/VetMed.php).

Graduation Semester

2017-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/97685

Copyright and License Information

Copyright 2017 Julie Danner

Owning Collections