Risk score for first-screening of prevalent undiagnosed chronic kidney disease in Peru: the CRONICAS-CKD risk score

Table 3 Associated factors with undiagnosed CKD: Beta Coefficients and Odds Ratios using CRONICAS database (N = 2368)

	Univariable Model		Multivariable Model		Risk Score (points)
	Coefficient (SE)	OR (95%CI, P-value)	Coefficient (SE)	OR (95%CI, P-value)	Risk Score (points)
Sex
Women	1	1
Men	0.18 (0.23)	1.20 (0.76-1.87, 0.45)
Age
< 50	1	1	1	1	0
50-69	1.82 (0.61)	6.17 (1.87-20.32, < 0.01)	1.54 (0.61)	4.66 (1.40-15.53, 0.01)	1
≥ 70	3.48 (0.60)	32.30 (10.00-104.40, < 0.01)	2.59 (0.61)	13.38 (4.01-44.62, < 0.01)	2
Personal History of any CVD
No	1	1
Yes	1.59 (0.63)	4.91 (1.42-17.09, 0.01)
Personal History of Infarction
No	1	1
Yes	−0.72 (1.04)	0.49 (0.06-3.75, 0.49)
Personal History of Stroke
No	1	1
Yes	−3.38 (1.01)	0.03 (0.01-0.25, < 0.01)
Personal History of Heart Failure
No	1	1
Yes	−2.67 (1.23)	0.07 (0.01-0.77, 0.03)
Smoking
No	1	1
Yes	−0.74 (0.47)	0.48 (0.19-1.19, 0.11)
Hypertension
No	1	1	1	1	0
Yes	2.15 (0.25)	8.60 (5.25-14.09, < 0.01)	1.54 (0.27)	4.65 (2.76-7.83, < 0.01)	1
Diabetes
No	1	1
Yes	0.80 (0.29)	2.22 (1.26-3.90, < 0.01)
BMI
Normal	1	1
Overweight	−0.22 (0.27)	0.80 (0.47-1.37, 0.42)
Obesity	−0.20 (0.30)	0.82 (0.45-1.48, 0.50)
Central Obesity
No	1	1
Yes	0.29 (0.30)	1.34 (0.75-2.40, 0.33)
Parents w/Infraction <60y
No	1	1
Yes	−0.74 (0.72)	0.48 (0.12-1.96. 0.30)
Anemia
No	1	1	1	1	0
Yes	1.82 (0.25)	6.16 (3.76-10.09, < 0.01)	1.27 (0.28)	3.56 (2.07-6. 21, < 0.01)	1
Total Cholesterol
Desirable	1
Borderline High/High	0.31 (0.23)	1.36 (0.87-2.13, 0.18)
HDL-Cholesterol
Low	1	1
High	−0.08 (0.23)	0.92 (0.59-1.44, 0.71)
LDL-Cholesterol
Optimal	1	1
Above Optimal/High	−0.16 (0.25)	0.85 (0.53-1.38, 0.52)
Triglycerides
≤ 150	1	1
> 150	0.28 (0.23)	1.32 (0.85-2.07, 0.22)

The multivariable model was created following a backward elimination approach: all variables with p < 0.20 in the univariable model were fitted and one by one removed starting with the least significant until the final model only included significant variables (p < 0.05). Although personal history of stroke or personal history of heart failure could have been included in the multivariable model, these were not included because their estimates were very small and personal history of any CVD, which included the aforementioned conditions, had a higher and significant estimate. For the final multivariable model: Hosmer-Lemeshow X² test: 4.13 with a p-value of 0.53. Area under de ROC curve of the final model was 0.8445. Each category of the risk factors included in the final model is given the corresponding point (see last column), and add up. For both, the complete and laboratory-free risk scores, the optimal cut-off point was 2 points: if a subject reaches 2+ points he/she would have greater odds of having CKD at that time; for example a 75-year old men with hypertension but without anemia would score 3 (2 points of age and 1 for hypertension)

ISSN: 1471-2369