Rodney H. Breau, Department of Surgery, Division of Urology, 501 Smyth Road, Box 222, Ottawa, Ontario, Canada K1H 8L6. e-mail: firstname.lastname@example.org
What's known on the subject? and What does the study add?
Urine cytology is frequently used by urologists to evaluate patients with microscopic or gross haematuria. The results of urine cytology can be used as impetus to perform or triage further diagnostic studies, e.g. cystoscopy.
The impact of urine cytology results on patient care warrants clarifying. This evidence-based medicine article explores how positive or negative urine cytology will impact the probability that a patient has urothelial carcinoma of the bladder before cystoscopy.
You just finished a long day in the operating room and sit down in your office to review paperwork. The first item you review is a referral from a GP for a 65-year-old female with microscopic haematuria (>3 red blood cells per high power field) found on two microscopic urine analyses. Her past medical history is negative for urological disease, including UTIs and urolithiasis, and she has never smoked. Her GP also forwarded results of an abdominal/pelvic ultrasound and urine culture that were normal. Urine cytology has been sent, and the results will be forwarded to your office in 2–3 days. You have a long waiting list for elective cystoscopy and wonder to yourself if the results of the cytology should influence how you triage the consultation. You elect to review the relevant literature on urine cytology diagnostic performance.
For patients with microscopic haematuria, how does urine cytology affect the probability of diagnosing bladder cancer?
FINDING THE BEST EVIDENCE
You begin your search by reviewing an article from the Users' Guide to the Urological Literature titled ‘How to Use an Article about a Diagnostic Test’. When reading the article you identify the critical components to appraise a study evaluating a diagnostic test . First, you identify the baseline prevalence (pre-test probability) of bladder cancer in your population of interest. In this case, the prevalence represents the baseline probability that someone with asymptomatic microscopic haematuria has bladder cancer. Then, you verify that the diagnostic test study is likely to be valid, and if so, calculate the positive and negative likelihood ratios for urine cytological testing. Together, the baseline probability and likelihood ratios (based on the results of urine cytology) allow you to calculate the probability of bladder cancer in this patient (post-test probability) and determine how urgently to see her.
You review the AUA guideline for the management of non-muscle-invasive bladder cancer, which reports that population-based studies of patients with microscopic haematuria and no additional risk factors aside from age estimate the prevalence of bladder cancer to be 1.3% (range 0.4–6.5%) [2–5].
SENSITIVITY AND SPECIFICITY
To calculate the likelihood ratios associated with urine cytology you require information about the sensitivity and specificity of the test. You perform a PubMed search using the terms ‘urine cytology’ AND ‘bladder cancer’, yielding >2400 articles. You think that a comprehensive review of reported studies is likely to be more informative and applicable for your patient than a single study alone and therefore elect to limit your search to systematic reviews. This yields five articles, one of which examines tumour markers for the diagnosis of primary bladder cancer . You elect to download the full version of this article for further review.
This systematic review included studies examining bladder tumour markers, including urine cytology, and compared them with a reference standard (cystoscopy or histopathology) for bladder cancer diagnosis. In all, 42 observational studies were included in the meta-analysis, 26 of which included data on urine cytology. A total of 3444 patients were included with urine cytology data. This analysis found that urine cytology had an overall sensitivity of 55% (95% CI 48–62) and specificity of 94% (95% CI 90–96) .
You conclude that although this meta-analysis is limited by a relatively heterogeneous group of studies, and the authors of the review deemed the quality of the individual studies were generally weak, these data are probably the best available, and confirms your background knowledge of urine cytology performance. Furthermore, because this analysis included a large and diverse selection of patients, the results are likely to be generalizable to your clinical question. You also know that sensitivity, specificity, and likelihood ratios are properties inherent to a diagnostic test and do not change based on the population prevalence of disease. Therefore, you have more confidence that the results can be applied to your patient.
LIKELIHOOD RATIOS (LRS)
LRs represent how a diagnostic test increases or decreases the odds of your patient having the disease of interest . In this example, LRs signify how positive or negative urine cytology changes the odds that your patient with microscopic haematuria has bladder cancer. You calculate the positive LR (LR+) using the equations in the Appendix. The LR+ is the likelihood of a positive test in a patient with bladder cancer compared with a positive test in a patient without bladder cancer. The LR+ for urine cytology is 9.2. Next you calculate the negative LR (LR–) using the equation in the Appendix. The LR−is the likelihood of a negative test in a patient with bladder cancer compared with a negative test in a patient without bladder cancer. The LR−for urine cytology is 0.48.
CASE RESOLUTION – APPLYING THE RESULTS TO THE CARE OF YOUR PATIENT
You now possess enough information to make an informed assessment of the 65-year-old woman referred to you with microscopic haematuria. Using the pre-test probability that she has bladder cancer (1.3%) and the LR+ (9.2) and LR−(0.48), you refer to a Fagan nomogram to determine the probability of bladder cancer when she has urine cytology (Fig. 1) . To use a Fagan nomogram, you identify the pre-test probability of your patient having the outcome of interest along the left side of the nomogram and apply the LR+ and LR− to determine the post-test probabilities by drawing a straight line that traverses the LR. Using the Fagan nomogram you realise that her probability of bladder cancer is 10% if she has positive urine cytology and 0.6% if she has negative urine cytology.
Pleased with your progress and increased understanding of the use of urine cytology, you elect to also explore how urine cytology would impact a patient referred to you for gross haematuria. A brief search reveals the pre-test probability of bladder cancer in patients with macroscopic haematuria to be ≈15% [8–11]. Applying the LRs in the same fashion as above, you calculate the post-test probability after a positive and negative test to be 62% and 8%, respectively (Fig. 1) .
A clear understanding of how the results of a diagnostic test (positive or negative) will impact the management of your patients is imperative for optimal practice of evidence-based urology. The above example and calculations show that a positive urine cytology test dramatically increases the probability that a patient referred to you with haematuria has bladder cancer. With this information, you commit yourself to reviewing a microscopic haematuria patient's cytology and arranging earlier cystoscopy if the cytology is positive.
CONFLICT OF INTEREST
APPENDIX: CALCULATION OF LRS FOR URINE CYTOLOGY AND UROTHELIAL CARCINOMA