Urinary Tract Infection (UTI) is the most common bacterial infection among elderly residents of nursing homes [1
] and often results in antibiotic treatment [2
]. Thus, a correct diagnosis is important for minimizing unnecessary antibiotic treatment. Dipstick urinalysis is often the first measure for detecting bacteriuria [3
]. The diagnostic value of dipstick urinalysis is most often evaluated for children and working age adults, preferably women which may lead to different results depending on age group and patient criteria. Thus, the clinical value of dipstick urinalysis could be quite different for elderly patients at nursing homes compared to younger patients whereby elderly patients have a higher prevalence of bacteriuria [1
Numerous errors can occur during the testing procedure of urine dipsticks [6
]. Timing and misalignment errors as well as subjectivity can be reduced by using a urine chemistry analyzer and thus achieve better precision [6
]. Other studies showed only minor improved reproducibility [9
] and no improvement in speed of analysis [10
] by using mechanized methods. Furthermore, when urine tests are performed under daily conditions results can be considerably lower, even for simple tests such as nitrite, than for optimal and standardized conditions achieved in most studies of the validity of urine tests [11
]. Thus, the importance of analyzer readings compared to visual readings of dipsticks in nursing homes for elderly remains to be clarified.
It should be noted that while sensitivity and specificity are of major interest for manufacturers of dipsticks these measures are of no interest to the physician making a clinical decision in one case. The positive predictive value (PPV) and the negative predictive value (NPV), however, are of the utmost clinical importance to the physician. These values are affected by the prevalence of bacteriuria [12
When estimating sensitivity and specificity it is appropriate to present an interval estimate [13
]. This is rarely done in studies evaluating diagnostic tests [13
]. The precision of predictive values, as with sensitivity and specificity, is dependent on the sample size [13
]. It is therefore also appropriate to use some kind of interval estimate for predictive values.
Unfortunately, only one previously published study evaluating dipstick urinalysis in elderly has presented confidence intervals for PPV and NPV [15
]. Other studies evaluating dipstick urinalysis of the elderly present confidence intervals only for sensitivity and specificity [16
] or no confidence intervals at all [8
Furthermore, as Yule-Simpson's statistical paradox predicts, the outcome of analysing a single bacterium might differ from analysing "any bacteria" [24
]. In such cases, results from analysing a single bacterium are more appropriate while results of analysing "any bacteria" are inappropriate. All previously published studies evaluating dipstick urinalysis of the elderly combine different bacterium to "any bacteria" when calculating sensitivity, specificity, PPV or NPV.
The primary aim of this study was to document the sensitivity, specificity, PPV and NPV with 95% confidence intervals for detection of bacteriuria among males and females in nursing homes for the elderly by dipstick urinalysis performed by non-laboratory personnel. The secondary aim was to compare manual readings of dipstick urinalysis with a urine chemistry analyzer.