Key Points

The annual incidence of CKD is relatively low (approximately 2%) even in groups that are considered at higher risk, such as people with diabetes or hypertension.A substantial number of people are tested at least annually for CKD despite being at very low risk, such as those with an annual incidence of 0.02%.Considering age together with other risk factors for CKD to determine the frequency of testing may increase the diagnostic yield for incident CKD.

Background

Guidelines recommend regular serum creatinine testing to detect CKD among people with diabetes or hypertension, but the ideal frequency of testing is unknown. We determined the diagnostic yield for incident CKD as defined by ≥2 measures of eGFR

Methods

We did a retrospective population-based cohort study of 3,515,163 adults older than 18 years with eGFR >60 ml/min per 1.73 m2 at baseline in Alberta, Canada. We assessed diagnostic yield overall and in categories defined by age, sex, comorbidity, albuminuria, or levels of a multivariable risk score for CKD.

Results

Assuming annual testing, the number of tests needed (NTN) to detect one new CKD case was >67-fold higher among those younger than 50 years (2149, [95% confidence interval (CI), 2103 to 2196]) as compared with older than 70 years (32, [95% CI, 32 to 32]). NTN for annual testing was 50 (95% CI, 49 to 50) among people with diabetes, 57 (95% CI, 57 to 58) in those with hypertension, and 20 (95% CI, 20 to 21) among people with heart failure. When stratified by CKD risk score, the NTN for annual testing ranged from 7 (95% CI, 7 to 8) at a score of 9 (highest risk) to 5708 (95% CI, 5494 to 5930) at a score of 0 (lowest risk). Testing people with diabetes every 3 years instead of every year would delay the diagnosis of CKD by a mean of 1.5 years for 2, 12, and 32 per 1000 people with diabetes aged 70 years, respectively. Corresponding delays associated with testing people with hypertension every 3 years instead of every year would affect 2, 9, and 27 per 1000 people aged 70 years, respectively. If applied to all adult Albertans, these two changes in testing frequency would potentially avert more than 5.9 million laboratory assays over the next decade.

Conclusions

Tailoring the frequency of serum creatinine testing according to age and the presence of other risk factors would decrease the NTN to detect cases of incident CKD.