Urine dipsticks in screening for diabetes mellitus
Report by Ooi Yau Wei, Medical Officer
Search checked by Stewart Teece, Specialist Registrar
Singapore Armed Force
A short cut review was carried out to establish whether urine dipstick testing for glycosuria was useful as a screening tool in asymptomatic patients. 178 papers were found using the reported searches, of which five presented the best evidence to answer the clinical question. The author, date and country of publication, patient group studied, study type, relevant outcomes, results and study weaknesses of these best papers are tabulated. It is concluded that dipstick testing for urinary glucose is insufficiently sensitive to be used as a screening tool for diabetes mellitus.
Clinical Scenario
An asymptomatic conscript walks into the Medical Centre and provides a urine sample for testing for glycosuria as part of a screening procedure for Diabetes Mellitus. You wonder how useful a urine dipstick is in screening for this condition.
Three‐part question
In [asymptomatic individuals] is [urine dipstick testing for glucose] sensitive enough to be used as [a screening tool for diabetes mellitus]
Search Strategy
Ovid MEDLINE 1966 to October Week 4 2005
[{ exp diabetes mellitus/OR diabetes mellitus.mp OR diabet$.mp OR IDDM.mp OR NIDDM.mp} AND { exp urine/OR urin$.mp} AND {exp reagent strips/OR exp glycosuria/OR exp “indicators and reagents/” OR stix.mp OR glucostix.mp OR stick.mp} AND { screen$.mp OR diagnos$.mp}] LIMIT to human and English language.
Embase 1980 to 2005 week 47. [exp diabetes mellitus/OR diabetes mellitus.mp. OR diabet$.mp. OR IDDM.mp. OR NIDDM.mp.] AND [exp urine/OR urin$.mp.] AND [exp test strip/OR exp glucosuria/OR stix.mp. OR glucostix.mp. OR stick.mp.] AND [screen$.mp. OR diagnos$] LIMIT to human and English Language
The Cochrane Library Issue 4 2005. Diabetes mellitus [MeSH] AND (Reagent strips [MeSH] OR reagent kits, diagnostic [MeSH] 20 hits
Outcome
178 papers found, of which 173 either did not answer the three part question or were review articles with no original data. The remaining five are shown below.
Table 2.
| Author, country, date | Patient group | Study type | Outcomes | Key results | Study weaknesses |
|---|---|---|---|---|---|
| West KM & Kalbfleisch JM 1971 USA | 4262 subjects (including 50 known diabetics) in 10 countries | Prospective diagnostic study | +ve urine clinistix test. DM diagnosed by blood glucose 2 h after glucose load | Sensitivity 35.3% (fasting) 38.9% (2–4 h after food) 67.0% (after 1 g/kg oral glucose) Specificity 99.7% (fasting) 97.6% (2–4 h after food) 94.4% (after oral glucose) | Diagnostic blood glucose lower than current WHO criteria |
| False negative rate not extrapolated to whole population in sensitivity calculation | |||||
| Statistical significance/95% C.I. not assessed | |||||
| Subject selection not discussed | |||||
| Davidson JK et al, 1978, USA | 1952 patients aged 16–76 not known to have DM presenting to medical clinic | Prospective diagnostic study | +ve urine clinitest. DM diagnosed in 20 of 25 +ve patients by fasting glucose or OGTT. 25 normoglucosuric patients underwent OGTT | Sensitivity 74% Specificity 86% (p<0.05) | Diagnostic blood glucose lower than current WHO criteria |
| False negative rate not extrapolated to whole population in sensitivity calculation | |||||
| Anderson DKG et al, 1992, Sweden | 3201 patients aged 35–79 not known to have DM presenting to GP | Prospective diagnostic study | +ve redia urine test. DM diagnosed by fasting glucose or OGTT. All patients 3 year follow up | Sensitivity 18.1% Specificity 99.1% | 3 year follow up may overestimate false negatives |
| Statistical significance/95% C.I. not assessed | |||||
| Davies MJ et al, 1993, UK | 442 consecutive patients aged 40–65 not known to be diabetic in larger postal study | Prospective diagnostic study | +ve urine test. All had OGTT | Sensitivity 44% Specificity 98% | Statistical significance/95% C.I. not assessed for sensitivity |
| Selection criteria not clear | |||||
| Friderichsen B & Maunsbach M, 1997, Denmark | 1530 patients aged 45–76 responding to postal survey of 3041. | Prospective diagnostic study | +ve urine glukotest R. DM diagnosed by fasting blood glucose | Sensitivity 20.80% (95% C.I. 8.14–52.81%) Specificity 99.14% (95% 98.54–99.59%) |
Comments
If the sensitivity of the first two studies are recalculated, extrapolating the documented prevalence of diabetes in each group to the patients who were considered true negative with no further tests, then the resultant sensitivities are 34.1% and 50.4% respectively. This is a reflection of the chances of finding glycosuria in a diabetic rather than the quality of urine dipsticks. Specificities are high, however low sensitivities preclude their use as a screening tool.
Clinical bottom line
Urine dipsticks are of insufficient sensitivity to be used as a screening tool in an asymptomatic population.
References
- West K M, Kalbfleisch J M. Sensitivity and specificity of five screening tests for diabetes in ten countries. Diabetes 1971;20(5):289-96. [DOI] [PubMed] [Google Scholar]
- Davidson J K, Reuben D, Sternberg J C.et al. Diabetes screening using a quantitative urine glucose method. Diabetes 1978;27(8):810-6. [DOI] [PubMed] [Google Scholar]
- Andersson D K, Lundblad E, Svardsudd K. A model for early diagnosis of type 2 diabetes mellitus in primary health care. Diabetic Medicine 1993;10(2):167-73. [DOI] [PubMed] [Google Scholar]
- Davies M J, Williams D R, Metcalfe J.et al. Community screening for non-insulin-dependent diabetes mellitus: self-testing for post-prandial glycosuria. Quarterly Journal of Medicine 1993;86(10):677-84. [DOI] [PubMed] [Google Scholar]
- Friderichsen B, Maunsbach M. Glycosuric tests should not be employed in population screenings for NIDDM. J Public Health Med 1997;19(1):55-60. [DOI] [PubMed] [Google Scholar]