Abstract: TH-PO1060
Identifying CKD of Unknown Etiology in Sri Lanka
Session Information
- CKD: Epidemiology, Risk Factors, Prevention - I
October 25, 2018 | Location: Exhibit Hall, San Diego Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: CKD (Non-Dialysis)
- 1901 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention
Authors
- Anand, Shuchi, Stanford University, Palo Alto, California, United States
- Montez-Rath, Maria E., Stanford University School of Medicine, Palo Alto, California, United States
- Adasooriya, Dinuka Madhushan, Teaching Hospital, Kandy, Bamunakotuwa, Sri Lanka
- Ratnatunga, Neelakanthi Vajira, Faculty of Medicine university of Peradeniya Sri Lanka, Kandy, Sri Lanka
- Kambham, Neeraja, Stanford University, Palo Alto, California, United States
- Badurdeen, Zeid, University of Peradeniya, Sri Lanka, Kandy, Sri Lanka
- Schensul, Stephen L., University of Connecticut School of Medicine, Farmington, Connecticut, United States
- Vlahos, Penny, University of Connecticut, Groton, Connecticut, United States
- Haider, Lali (Lalarukh), University of CT Health Center, West Hartford, Connecticut, United States
- Bhalla, Vivek, Stanford University, Palo Alto, California, United States
- Levin, Adeera, St. Paul's Hospital and University of British Columbia, Vancouver, British Columbia, Canada
- Chertow, Glenn Matthew, Stanford University School of Medicine, Palo Alto, California, United States
- Nanayakkara, Nishantha, Teaching Hospital, Kandy, Bamunakotuwa, Sri Lanka
Background
A kidney disease of unknown cause is common in Sri Lanka’s lowland region. No evidence exists to inform an approach to a non-invasive clinical diagnosis.
Methods
In a prospective study to determine whether non-invasive measures can identify CKDu, we surveyed 600 new patients coming to nephrology clinic in a hospital servicing endemic regions over one year. 87 underwent kidney biopsy; 43 (49%) had a pathology diagnosis of tubulointerstitial disease. Using logistic regression, we tested the association of nine pre-selected factors with likelihood of tubulointerstitial nephritis on biopsy. We used bootstrap validation to calculate the model validated AUC. We tested a Full model and five parsimonious models.
Results
AUC for the Full model to predict CKDu was 0.82. A parsimonious model with age, serum albumin, and urine dipstick for protein had an AUC of 0.84 and bootstrap calibration slope of 0.84 (Table 1); with PPV 82.9% and NPV 80.4%. Patients with diabetes or hypertension recommended for kidney biopsy did not experience lower odds of a CKDu diagnosis.
Conclusion
We developed a standardized approach relying on non-invasive measures to identify probable cases of CKDu in Sri Lanka. Such an approach can strengthen CKDu surveillance, geographic mapping and rigorous investigations into cause via case-control studies.
Predictors | Full model | Parsimonious model |
Urine dipstick negative for protein | * | * |
Age (cubic spline) | * | * |
HTN | * | |
Diabetes | * | |
Serum albumin | * | * |
Hematuria | * | |
No pyuria | * | |
Potassium | * | |
Woman | * | |
Urine dipstick negative * Serum albumin | * | * |
C statistic | 0.89 | 0.87 |
Sensitivity | 83.7 | 79.1 |
Specificity | 79.6 | 84.1 |
Positive predictive value | 80.0 | 82.9 |
Negative Value | 83.3 | 80.4 |
Bootstrap Validated C statistic | 0.82 | 0.84 |
Calibration slope | 0.59 | 0.84 |
Funding
- Clinical Revenue Support