Korean J Nephrol.
2005 May;24(3):390-398.
The Mathematical Model for Predicting the Probability of Minimal Change Nephrotic Syndrome Using Clinical Parameters
- Affiliations
-
- 1Department of Internal Medicine, Chungbuk National University College of Medicine, Cheongju, Korea. hyekim@chungbuk.ac.kr
- 2Department of Preventive Medicine, Chungbuk National University College of Medicine, Cheongju, Korea.
- 3Dr. Earm's Medical Clinic, Cheongju, Korea.
Abstract
- PURPOSE
We retrospectively investigated to find out the equation of calculating the probability of minimal change nephrotic syndrome (MCNS) using clinical parameters. We prospectively investigated to determine the usefulness of the mathematical model. METHODS: We retrospectively examined 56 patients with nephrotic syndrome (NS) (30 MCNS and 26 non-MCNS) diagnosed by kidney biopsy. A mathematical model for calculating the probability of MCNS was obtained through multiple logistic analysis in SAS statistics package. In addition, we prospectively studied 28 patients with NS. Clinical MCNS and non-MCNS were classified according to the probability of 85% in the mathematical model. Kidney biopsy was performed, and serum albumin and urinalysis were measured after 2 weeks of steroid treatment. RESULTS: In the retrospective study, the mathematical model was P=ea/(1+ea), a=17.2507 - 5.5777xON - 4.2256xALB-0.000579x24PROT - 1.2569xUBL+2.1703xUAL. The mode of onset (ON), 24 hours urine protein (24PROT), serum albumin concentration (ALB), the grade of hematuria (UBL) and proteinuria (UAL) were included as clinical parameters. At the probability of 85%, the sensitivity and specificity for predicting MCNS was 73.3% and 100% respectively. In the prospective study, the result of kidney biopsy was consistent with clinical MCNS and non-MCNS according to a mathematical model. All clinical MCNS showed negative proteinuria on urinalysis and a significant increase in serum albumin after 2 weeks treatment (1.85+/-0.30 g/dL to 2.88+/-0.26 g/dL, p<0.05). CONCLUSION: We conclude that the mathematical model for predicting the probability of MCNS may be useful in diagnosis of the MCNS.