Research ArticleResearch

Chronic kidney disease in type 2 diabetes

Does an abnormal urine albumin-to-creatinine ratio need to be retested?

Divya Garg, Christopher Naugler, Vishal Bhella and Fahmida Yeasmin
Canadian Family Physician October 2018; 64 (10) e446-e452;

Abstract

Objective To determine the positive predictive value (PPV) of a single random abnormal urine albumin-to-creatinine ratio (ACR) compared with repeat test results in patients with type 2 diabetes to diagnose chronic kidney disease (CKD).

Design Retrospective, longitudinal secondary data analysis using Calgary Laboratory Services data.

Setting Calgary, Alta.

Participants Patients aged 21 and older with a new diagnosis of diabetes in the study period from January 2008 to December 2015 and with a first abnormal urine ACR followed by another ACR test completed within 120 days.

Main outcome measures The PPV of an abnormal urine ACR (2 to 20 mg/mmol) to diagnose CKD was calculated. A test result was considered a true positive if a subsequent positive test result (≥ 2 mg/mmol) was identified within 120 days of the first positive test result and a false positive if 2 subsequent negative test results were identified within the same time period. The relationship between the first and second urine ACR values to assess the probability of the second urine ACR being abnormal (≥ 2 mg/mmol) based on the values of the first abnormal urine ACR was also explored.

Results The PPV of the first abnormal urine ACR between 2 and 20 mg/mmol to diagnose CKD was calculated at 96.80% (95% CI 95.37% to 98.21%). Additionally, there was increased predictive probability of the second urine ACR being abnormal at higher values of the first urine ACR (2 to 20 mg/mmol). The data were further analyzed to exclude test results with a new or changed prescription of angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker medications around the time of the first urine ACR test to focus results on screening and not treatment response. With these exclusions, the PPV for first urine ACR between 2 and 20 mg/mmol to diagnose CKD was calculated as 96.23% (95% CI 94.13% to 98.32%).

Conclusion The first random abnormal urine ACR has a good PPV for the diagnosis of CKD in patients with type 2 diabetes, so multiple random urine ACR tests might not be necessary to diagnose patients with type 2 diabetes as having persistent microalbuminuria and CKD. A simpler diagnostic model for diagnosing renal disease might improve patient compliance, efficiency of testing, and implementation of health interventions. Reduced testing would also be expected to result in reduced cost from a health care expenditure perspective.

Chronic kidney disease (CKD) in diabetes has considerable implications in terms of morbidity and mortality and is associated with both reduced lifespan and lower quality of life.1,2 There is a higher risk of cardiovascular and all-cause mortality with declining renal function.3–5 Chronic kidney disease and cardiovascular disease also share common modifiable risk factors including hyperglycemia, hypertension, and hyperlipidemia.3

The Diabetes Canada clinical practice guidelines recommend screening for diabetic nephropathy, as early detection and intervention can delay or prevent loss of renal function and progression to end-stage renal disease.2,6 Diabetic nephropathy is defined as a progressive increase in proteinuria and decline in renal function in patients with diabetes.2 It is clinically diagnosed as the presence of microalbuminuria or overt nephropathy in patients with diabetes when not explained by other renal diseases.7

Microalbuminuria is the earliest clinically detectable stage of diabetic nephropathy defined as urinary albumin of 30 to 300 mg/d or a random urine albumin-tocreatinine ratio (ACR) between 2 and 20 mg/mmol.8 This level of albumin is below the detection threshold of a urine dipstick.7,9 Overt nephropathy is defined as a urinary albumin level of greater than 300 mg/d or a random urine ACR of greater than 20 mg/mmol.7,9

The Diabetes Canada guidelines recommend screening for CKD with random urine ACR and estimated glomerular filtration rate tests at the time of diagnosis of type 2 diabetes and annually thereafter.2,6 When the results of the first urine ACR test are between 2 and 20 mg/mmol, 2 out of 3 tests done over a 3-month period showing a persistent elevation in urinary albumin levels are needed to diagnose CKD. Multiple tests to confirm albuminuria are recommended to rule out false-positive results due to alternate causes of transient albuminuria, including considerable exercise, urinary tract infection, fever, congestive heart failure, menstruation, and acute severe elevation in blood glucose or blood pressure.2,6 Intraindividual day-to-day variability has also been reported.8 However, the recommendation to repeat an ACR test following abnormal test results between 2 to 20 mg/mmol is largely based on expert opinion, and further research is needed to determine if repeat testing improves diagnostic accuracy.10

The RIACE (Renal Insufficiency And Cardiovascular Events) cohort study showed that a single measurement of urinary albumin excretion can accurately predict the nephropathy stage.11 In the study, a subanalysis was conducted on 4062 of 15 773 patients who had 2 or 3 measures of urinary albumin excretion within a 3- to 6-month period. Among these patients, urinary albumin excretion measurements were obtained from 24-hour urine samples in 833 patients and first-void samples for ACR tests in 3229 patients. The intraindividual coefficient of variation was high at 32.5%. However, the concordance rate between single and the mean of multiple measurements was 94.6% for normoalbuminuria and 90.6% for albuminuria (micro or macro). In another study, a retrospective analysis was done on morning urine samples obtained from 95 patients with type 2 diabetes.12 In this study 246 triplicate urine ACR test results were analyzed, and the overall concordance rate for albuminuria categorization was 95%. It was determined that repeating a morning urine ACR sample was not necessary in order to categorize patients as having normoalbuminuria, microalbuminuria, or macroalbuminuria. The study did suggest, however, that repeat determinations are necessary for accurate quantification of albuminuria.

These studies suggest that multiple samples might not be necessary at the screening stage for categorizing a patient as having albuminuria or not. However, other studies have shown that multiple samples are still likely necessary to assess changes in urine ACR due to treatment or progression of disease.13

Further studies are needed in patients with type 2 diabetes in order to assess if results from the recent studies on reproducibility of a single value of morning urine ACR can be demonstrated with random urine ACR samples. The ability to make a diagnosis using a single sample would avoid the need for a patient to provide time-sensitive or multiple urine samples for screening purposes, and might improve patient compliance for testing and efficiency of implementing interventions in the primary care setting.

Although recent diabetes guidelines recommend initiation of angiotensin-converting enzyme inhibitor (ACEI) or angiotensin II receptor blocker (ARB) therapy independent of the diagnosis of renal disease in patients with diabetes (age ≥55 years), the diagnosis remains important, as the presence and progression of renal disease has important implications for pharmacotherapy options and as a cardiovascular risk predictor.6,14 Accordingly, simplification of the diagnosis of renal disease in the population with type 2 diabetes would have important clinical implications.

With more than 3 million Canadians currently diagnosed with diabetes and with a urine ACR test costing about $11, there could be an economic advantage from a health care utilization perspective of a simpler diagnostic model for CKD in diabetes.6,15

Our research objective was to calculate the positive predictive value (PPV) of a single random abnormal urine ACR compared with repeat test results in patients with type 2 diabetes to diagnose CKD. We also explored the relationship between the first and second urine ACR values (between 2 and 20 mg/mmol) when performed within 120 days.

METHODS

The study was conducted using secondary data from Calgary Laboratory Services, which is the only laboratory services provider in Calgary, and therefore these data capture all tests ordered within the region. De-identified data from the laboratory information system were extracted by Alberta Health Services employees and reported to the research team.

The data from January 2008 to December 2015 were assessed to identify patients aged 21 and older with hemoglobin A1c levels of 6.5% or higher or with 2 or more test results revealing fasting plasma glucose levels of 7 mmol/L or higher. To increase the probability of including patients with new-onset type 2 diabetes in the data set, patients were only included if they had a newly recorded hemoglobin A1c (≥6.5%) level or had 2 new recordings of abnormal fasting plasma glucose levels and no abnormal results for these tests recorded in the previous 3 years. To exclude those with gestational diabetes, patients with a previous elevated plasma glucose level when screened with a 50-g glucose tolerance test were not included. The inclusion group was further assessed for a first random abnormal urine ACR result (≥2 mg/mmol) followed by at least 1 other random urine ACR test over a 120-day period. These tests were done as part of routine care. Each patient was included in the data set only the first time he or she met the criteria.

Further analysis was done excluding results when ACEI or ARB therapy was started or adjusted within 3 months before or 4 months after the first urine ACR test to focus our results on screening and not treatment response. This step was completed by linking Calgary Laboratory Services data to Pharmaceutical Information Network data. Pharmaceutical Information Network data contain dispensing data from community and outpatient pharmacies and capture most prescriptions dispensed by Alberta pharmacies.

This project received ethics approval from the Conjoint Health Research Ethics Board at the University of Calgary.

Measures of interest

Logistic regression was used to assess the relationship between the first and second random urine ACR values in order to derive a predictive probability plot. A logistic regression model makes predictions on a log-odds scale, which can be converted to a probability scale. This enabled us to construct a predictive probability plot that showed the probability of the second urine ACR result being positive for a range of ACR values of the first urine ACR (predictor variable).

The PPV of the first urine ACR test result to diagnose CKD was also calculated. The first abnormal urine ACR (≥2 mg/mmol) was considered a true positive if a subsequent positive urine ACR result was obtained within 120 days and a false positive if 2 subsequent negative test results were identified within that time period. Correlation of true positives and false positives with age was assessed using a 2-sample t test and with sex using the Pearson Embedded Image2 test.

This analysis was repeated after excluding patients who had ACEI or ARB therapy started or adjusted around the time of the first ACR test.

Statistical analyses were done using R statistical software. A P value of less than .05 was considered statistically significant.

RESULTS

A total of 1243 cases were identified with the inclusion criteria (Figure 1); 206 cases in which urine ACR test results revealed values greater than 20 mg/mmol were excluded, as results that reveal macroalbuminuria do not need to be repeated as per Diabetes Canada guidelines.6 Table 1 presents the characteristics of the remaining 1037 patients. Analysis was done including all initial positive test results for microalbuminuria to assess how the first urine ACR (2 to 20 mg/mmol) predicts the results of the second urine ACR test. A predictive probability plot was derived from results of logistic regression. Figure 2 shows the predictive probability plot, which is a graphical representation of the predictive probability of the second urine ACR test being positive for a range of the first urine ACR values. There is increased probability of the second urine ACR being abnormal at higher values of the first urine ACR (2 to 20 mg/mmol), as illustrated in Figure 2. The probability of having positive results on the second ACR test is about 0.4 when the first urine ACR is between 2 and 4 mg/mmol and about 0.8 when the first urine ACR is between 6 and 8 mg/mmol.

Figure 1.
Figure 1.

Flowchart of study sample selection listing inclusion and exclusion criteria

ACEI—angiotensin-converting enzyme inhibitor, ACR—albumin-to-creatinine ratio, ARB—angiotensin II receptor blocker, HbA1c—hemoglobin A1c.

View this table:
Table 1.

Patient characteristics

Figure 2.
Figure 2.

The predictive probability of the second urine ACR result being positive for albuminuria (≥ 2 mg/mmol) based on a range of values of the first urine ACR result (2–20 mg/mmol) in patients with type 2 diabetes: The second urine ACR test was done within 120 days of the first urine ACR test.

ACR—albumin-to-creatinine ratio.

The PPV of the first urine ACR is calculated based on the Diabetes Canada recommendations that 2 out of 3 positive urine ACR test results in a 3-month period are required to confirm persistent microalbuminuria and diagnose CKD.6 To calculate the PPV of the first positive urine ACR result to diagnose CKD, it was considered a true positive if another positive test result was identified in the study time frame of 120 days and a false positive if 2 subsequent negative test results were identified in the same time frame. A total of 445 cases that had inadequate follow-up of the first abnormal test result to confirm or refute microalbuminuria were excluded. In the excluded cases, a third urine ACR value was not available when the first urine ACR result was positive but the second urine ACR result was negative.

From 592 remaining cases, 19 were identified as false positive and 573 were identified as true positive, resulting in a PPV of the first abnormal urine ACR (2 to 20 mg/mmol) to diagnose CKD calculated at 96.80% (95% CI 95.37% to 98.21%). There was no statistically significant association with sex (P = .58) and no difference in mean age (P = .51) between the false-positive and true-positive groups. The PPVs for discrete categories of the first urine ACR are presented in Table 2.

View this table:
Table 2.

The PPV of the first urine ACR to diagnose microalbuminuria, based on urine ACR range

Additional analysis was done after excluding cases in which ACEI or ARB therapy was started or adjusted within 3 months before or 4 months after the first urine ACR test. With these exclusions 318 cases were identified, of which 12 were false positive, resulting in a PPV of first abnormal urine ACR (2 to 20 mg/mmol) of 96.23% (95% CI 94.13% to 98.32%). There was no statistically significant association with sex (P = .84) or difference in mean age (P = .37) between the false-positive and the true-positive groups.

DISCUSSION

Some Canadian studies have previously reported the rate of screening for microalbuminuria in patients with type 2 diabetes in primary care to be below 30%.16,17 Our study indicates that there is a gap between recommended practice guidelines and clinical practice in follow-up of abnormal urine ACR to screen for CKD in diabetes.

Diabetes Canada recommends that 2 out of 3 abnormal urine ACR results over a 3-month period are required to diagnose CKD.6 In our analysis, there is a strong PPV for the first abnormal urine ACR (between 2 and 20 mg/mmol) to diagnose CKD at 96.80%. The PPV with exclusion of cases with ACEI or ARB therapy started or adjusted around the first ACR test is similar at 96.23%. We also excluded 445 results in which there were insufficient follow-up data to make a determination if the first abnormal urine ACR was a true positive or a false positive. The number of results with inadequate follow-up indicates a gap between guidelines and actual practice. In Table 2, it is noteworthy that most of the excluded results occur in the urine ACR range of 2 to 6 mg/mmol and that there is a reduced proportion of excluded results for urine ACR values of 6 to 20 mg/mmol.

We performed an additional analysis looking strictly at the relationship between the first and second urine ACR (irrespective of whether a third value was available) to derive a predictive probability plot. From this analysis, there is greater probability of the second urine ACR result being positive (≥2 mg/mmol) when the first urine ACR is greater than 6 mg/mmol.

Our results suggest that a single random urine ACR might be used to diagnose microalbuminuria or CKD in a patient with type 2 diabetes. Having multiple steps in the diagnostic pathway can negatively affect early detection and screening of diabetic nephropathy in primary care. Reduced steps in screening imply fewer patient visits to the laboratory or to health care providers, which might improve patient compliance and decrease health care expenditure.

Limitations

As previously stated, inadequate follow-up of abnormal urine ACR results, as per guideline recommendations, limited the number of results available to calculate the PPV of the first urine ACR. We also did not assess the relationship between true-positive and false-positive results of first urine ACR and renal function or glycemic control. Our data are limited as this is a retrospective analysis with selection criteria designed to increase the probability of including cases with a new diagnosis of diabetes and a subsequent incident case of abnormal urine ACR. The smaller number of cases obtained secondary to these factors could affect how well the data represent the population with type 2 diabetes. Additional limitations include that we only assessed repeating the urine ACR test for screening. As previous studies indicate, repeat tests are still important to assess progression of disease or treatment response. In our study outcomes related to CKD screening in diabetes were also not examined.

Conclusion

Based on our results, multiple random urine ACR tests might not be necessary to diagnose patients with type 2 diabetes as having persistent microalbuminuria and CKD. A simpler diagnostic model for diagnosing renal disease might improve patient compliance, efficiency of testing, and implementation of health interventions. Reduced testing would also be expected to result in reduced cost from a health care expenditure perspective.

Further studies are required with a greater proportion of triplicate tests available when the second urine ACR result is negative to further assess and confirm the PPV of the first positive urine ACR result.

Notes

Editor’s key points

  • ▸ With more than 3 million Canadians currently diagnosed with diabetes and with a urine albumin-to-creatinine ratio (ACR) test costing about $11, a simpler diagnostic model for chronic kidney disease (CKD) in diabetes might be economically advantageous.

  • ▸ In this study there was a strong positive predictive value for the first abnormal urine ACR (between 2 and 20 mg/mmol) to diagnose CKD at 96.80%. An evaluation of the relationship between the first and second urine ACR results (irrespective of whether a third test result was available) to derive a predictive probability plot revealed that there was greater probability of the second urine ACR being abnormal (≥ 2 mg/mmol) when the first urine ACR was greater than 6 mg/mmol.

  • ▸ Several random urine ACR tests might not be necessary to diagnose patients with type 2 diabetes as having persistent microalbuminuria and CKD. Having multiple steps in the diagnostic pathway can negatively affect early detection and screening of diabetic nephropathy in primary care.

Points de repère du rédacteur

  • ▸ Puisqu’il y a plus de 3 millions de diabétiques au Canada et que la mesure du rapport albumine/créatinine (RAC) dans l’urine coûte environ 11 $, il pourrait être économiquement avantageux d’utiliser un moyen plus simple pour établir un diagnostic de maladie rénale chronique (MRC) chez un diabétique.

  • ▸ Cette étude a révélé qu’une première mesure du RAC (entre 2 et 20 mg/mmol) possède une excellente valeur prédictive positive pour diagnostiquer une MRC dans 96,80 % des cas. Une évaluation de la relation entre le premier et le deuxième résultat du RAC pour établir une valeur prédictive probable pour une MRC a révélé qu’il était plus probable que la deuxième valeur du RAC soit anormale (≥ 2 mg/mmol) quand le premier résultat était supérieur à 6 mg/mmol, et ce, peu importe qu’il y ait ou non un troisième résultat disponible.

  • ▸ Il pourrait ne pas être nécessaire de faire plusieurs mesures du RAC pour diagnostiquer une MRC chez un diabétique de type 2 qui présente une microalbuminurie persistante et une MRC. Dans un contexte de soins primaires, le fait de faire plusieurs mesures peut retarder le diagnostic et le dépistage d’une néphropathie diabétique.

Footnotes

  • Contributors

    Drs Garg, Naugler, and Bhella contributed to the concept and design of the study; data gathering, analysis, and interpretation; and preparing the manuscript for submission. Ms Yeasmin contributed to statistical analysis and interpretation, as well as preparing the manuscript for submission.

  • Competing interests

    None declared

  • This article has been peer reviewed.

  • Cet article a fait l’objet d’une révision par des pairs.

References

  1. 1.↵
    1. Fox CS,
    2. Matsushita K,
    3. Woodward M,
    4. Bilo HJ,
    5. Chalmers J,
    6. Heerspink HJ,
    7. et al
    . Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without diabetes: a meta-analysis. Lancet 2012;380(9854):1662-73. Erratum in: Lancet 2013;381(9864):374.
    OpenUrlCrossRefPubMed
  2. 2.↵
    1. Canadian Diabetes Association Clinical Practice Guidelines Expert Committee,
    2. McFarlane P,
    3. Gilbert RE,
    4. MacCallum L,
    5. Senior P
    . Chronic kidney disease in diabetes. Can J Diabetes 2013;37(Suppl 1):S129-36. Epub 2013 Mar 26.
    OpenUrl
  3. 3.↵
    1. Suckling R,
    2. Gallagher H
    . Chronic kidney disease, diabetes mellitus and cardiovascular disease: risks and commonalities. J Ren Care 2012;38(Suppl 1):4-11.
    OpenUrlCrossRefPubMed
  4. 4.
    1. Go AS,
    2. Chertow GM,
    3. Fan D,
    4. McCulloch CE,
    5. Hsu C
    . Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004;351(13):1296-305. Erratum in: N Engl J Med 2008;18(4):4.
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Chronic Kidney Disease Prognosis Consortium
    . Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: a collaborative meta-analysis. Lancet 2010;375(9731):2073-81. Epub 2010 May 17.
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Diabetes Canada Clinical Practice Guidelines Expert Committee,
    2. Matsushita K,
    3. van der Velde M,
    4. Astor BC,
    5. Woodward M,
    6. Levey AS,
    7. et al
    . Diabetes Canada 2018 clinical practice guidelines for the prevention and management of diabetes in Canada. Can J Diabetes 2018;42(Suppl 1):S1-325.
    OpenUrlCrossRefPubMed
  7. 7.↵
    Nephropathy. Can J Diabetes 2003;27(2):S66-71.
    OpenUrl
  8. 8.↵
    1. Miller WG,
    2. Bruns DE,
    3. Hortin GL,
    4. Sandberg S,
    5. Aakre KM,
    6. McQueen MJ,
    7. et al
    . Current issues in measurement and reporting of urinary albumin excretion. Clin Chem 2009;55(1):24-38. Epub 2008 Nov 21.
    OpenUrlAbstract/FREE Full Text
  9. 9.↵
    1. Lemley KV,
    2. Abdullah I,
    3. Myers BD,
    4. Meyer TW,
    5. Blouch K,
    6. Smith WE,
    7. et al
    . Evolution of incipient nephropathy in type 2 diabetes mellitus. Kidney Int 2000;58(3):1228-37. Erratum in: Kidney Int 2000;58(5):2257.
    OpenUrlCrossRefPubMed
  10. 10.↵
    1. Scheid DC,
    2. McCarthy LH,
    3. Lawler FH,
    4. Hamm RM,
    5. Reilly K
    . Screening for microalbuminuria to prevent nephropathy in patients with diabetes: a systematic review of the evidence. J Fam Pract 2001;50(8):661-8.
    OpenUrlPubMed
  11. 11.↵
    1. Pugliese G,
    2. Solini A,
    3. Fondelli C,
    4. Trevisan R,
    5. Vedovato M,
    6. Nicolucci A,
    7. et al
    . Reproducibility of albuminuria in type 2 diabetic subjects. Findings from the Renal Insufficiency And Cardiovascular Events (RIACE) study. Nephrol Dial Transplant 2011;26(12):3950-4. Epub 2011 Mar 25.
    OpenUrlCrossRefPubMed
  12. 12.↵
    1. Deeb A,
    2. Zaoui P,
    3. Le Penven S,
    4. Tartry D,
    5. Lantelme P,
    6. Ducher M,
    7. et al
    . Are triplicate urine samples necessary to assess albuminuria? Nephron Clin Pract 2012;122(3–4):80-5. Epub 2013 Apr 12.
    OpenUrl
  13. 13.↵
    1. Naresh CN,
    2. Hayen A,
    3. Weening A,
    4. Craig JC,
    5. Chadban SJ
    . Day-to-day variability in spot urine albumin-creatinine ratio. Am J Kidney Dis 2013;62(6):1095-101. Epub 2013 Aug 16.
    OpenUrlCrossRefPubMed
  14. 14.↵
    1. Canadian Diabetes Association Clinical Practice Guidelines Expert Committee,
    2. Stone JA,
    3. Fitchett D,
    4. Grover S,
    5. Lewanczuk R,
    6. Lin P
    . Vascular protection in people with diabetes. Can J Diabetes 2013;37(Suppl 1):S100-4. Epub 2013 Mar 26.
    OpenUrlPubMed
  15. 15.↵
    1. Hale I
    . Add to cart? Can Fam Physician 2015;61:937-9. (Eng), 941–4 (Fr).
    OpenUrlFREE Full Text
  16. 16.↵
    1. Harris SB,
    2. Worrall G,
    3. Macaulay A,
    4. Norton P,
    5. Webster-Bogaert S,
    6. Donner A,
    7. et al
    . Diabetes management in Canada: baseline results of the group practice diabetes management study. Can J Diabetes 2006;30(2):1-7.
    OpenUrl
  17. 17.↵
    1. Harris SB,
    2. Stewart M,
    3. Brown JB,
    4. Wetmore S,
    5. Faulds C,
    6. Webster-Bogaert S,
    7. et al
    . Type 2 diabetes in family practice. Room for improvement. Can Fam Physician 2003;49:778-85.
    OpenUrlAbstract/FREE Full Text
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Respond to this article
Chronic kidney disease in type 2 diabetes
Divya Garg, Christopher Naugler, Vishal Bhella, Fahmida Yeasmin
Canadian Family Physician Oct 2018, 64 (10) e446-e452;
Twitter logo Facebook logo Mendeley logo

Jump to section

Subjects