Practice Guidelines That Explicitly Consider Harms and Costs

Hematuria is prevalent in up to a third of the population.¹ Approximately 2–5% of patients with microscopic hematuria and 12% of patients with macroscopic or gross hematuria will be diagnosed with urinary tract cancer, the majority of which are superficial bladder tumors detected via cystoscopy.² Screening asymptomatic persons for bladder cancer with urinalysis to detect microscopic hematuria is not recommended by the U.S. Preventative Task Force due to insufficient evidence of benefits from treatment of screen-detected bladder cancer, as well as insufficient data on the harms of screening for and evaluating hematuria.³ There are also no acceptable screening tests for renal cell carcinoma, the predominant cancer detected via imaging during evaluation for hematuria. However, with the widespread use of urinalysis and high prevalence of hematuria, many patients are evaluated for incidental findings of hematuria. In this issue of JAMA Internal Medicine, Georgieva et al⁴ contribute to the growing evidence of potential harms caused by evaluations based on guidelines that recommend CT scanning and cystoscopy for all patients with hematuria rather than risk stratification and less aggressive evaluation of low-risk patients, such as those with microhematuria, women, younger men, and non-smokers, who can be imaged using renal ultrasound.

The authors report a patient-level microsimulation of 5 different guideline-based algorithms for the initial evaluation of hematuria, including both microscopic and gross, to estimate subsequent urinary tract cancer detection rates, harms, and costs. They consistently find that guidelines that include CT scanning for all patients with hematuria, such as those from the American Urologic Association¹, modestly improve urinary tract cancer detection, particularly renal cell carcinoma, but lead to significantly higher estimated rates of secondary cancer from imaging-associated radiation and cost approximately twice as much as alternative risk-based approaches that substitute renal ultrasound for CT scanning in low-risk patients. Notably, the number of secondary cancers estimated to occur due to imaging-associated radiation exposure was 5 times higher with a uniform CT scanning approach, and more than 10 times higher than the additional number of urinary tract cancers detected, compared to approaches that recommend renal ultrasound rather than CT scanning for low-risk patients. This study is an excellent example of how the development and comparison of guidelines is enhanced by an explicit and comprehensive evaluation of benefits, harms, and costs.

A simulation model is only useful if the model structure and inputs, which inform internal and external validity, are accurate and generalizable.⁵ Georgieva et al⁴ used multiple large, population-based estimates to model the prevalence of urinary tract cancer among patients with hematuria, distribution of risk factors, test characteristics, procedural complication rates, and Medicare costs. Radiation exposure estimates were adapted from real-world data on radiation from CT scans, and estimated risks of cancer or death due to radiation exposure were informed by reports from the National Academy of Sciences and the Agency for Healthcare Research and Quality. The authors compared model-predicted urinary tract cancer rates against those published in large meta-analyses and used probabilistic sensitivity analysis based on published error estimates to empirically incorporate uncertainty into their model parameters. These sensitivity analyses were robust to a range of assumptions and further support the authors conclusions: the harms of uniform CT scanning for hematuria, particularly microscopic hematuria, outweigh the benefits when compared to risk-based approaches that prioritize renal ultrasound for low-risk patients.

CT has modestly higher sensitivity compared to ultrasound for detecting renal cell carcinoma (0.96 versus 0.91, respectively), as well as higher sensitivity for upper tract urothelial carcinoma (0.95 versus 0.71, respectively). However, compared to ultrasound, CT scanning does not improve detection of bladder cancer, which accounts for 85% of cancers detected during hematuria evaluations. Thus, the majority of additional cancers detected with uniform CT scanning for hematuria are in upper urinary tract, primarily renal cell carcinoma. The potential benefit from detecting renal cell carcinoma during evaluation for hematuria compared to delayed diagnosis depends on whether current treatments significantly improve morbidity and mortality if cancer is detected earlier. Additional studies are needed to compare the stage, grade, and site of renal and upper urinary tract cancers detected using different hematuria evaluation algorithms. However, even under the liberal assumption of significant improvements in morbidity and mortality from early treatment of all additional detected upper urinary tract cancers, benefits likely do not outweigh harms because the number of additional radiation-induced cancers is significantly higher than the number of additional upper urinary tract cancers detected. Morbidity and mortality caused by these radiation-induced cancers cannot be ignored when developing and comparing hematuria evaluation guidelines.

While the benefits of uniform CT scanning for all patients with microscopic hematuria are likely overestimated in this study because the simulated population includes patients with a history of gross hematuria, and thus higher renal cell carcinoma risk, the benefits of CT scanning for all patients with gross hematuria are likely underestimated. The ability to compare guideline-based evaluations of gross hematuria is limited in this study because the authors did not separately compare evaluations within this higher risk subgroup of patients that are most likely to benefit from replacing ultrasound with CT as recommended in some, but not all, guidelines. Additional studies are needed to compare guidelines for the initial evaluation of patients with gross hematuria.

Although cost is less relevant when harms outweigh benefits, this study builds on prior literature suggesting that uniform CT scanning for hematuria is not cost effective. Among adult patients with asymptomatic microscopic hematuria, a model-based cost-effectiveness study found that replacing ultrasound with CT detected a single additional urinary tract cancer at an incremental cost per cancer detected of more than $6 million.⁶ Georgieva et al⁴ went a step further by including both microscopic and gross hematuria and found that uniform CT scanning for hematuria cost twice as much per person evaluated compared to other risk-based strategies with incremental costs of more than $1 million per urinary tract cancer detected. Using different sources of cost data and definitions of hematuria, multiples studies support the same conclusion: uniform CT scanning approaches lead to minimal increases in number of detected urinary tract cancers and much higher additional costs.

Certain guideline-based hematuria evaluations may prioritize decreasing morbidity and mortality due to urinary tract cancer, but disregard the harms and costs of the evaluation itself. This limited framework assumes that earlier detection and treatment will improve clinical outcomes, harms of the evaluation are minimal or do not exist, and resources are unlimited. However, potential harms of hematuria evaluation include false positives, procedural complications, radiation-induced cancers, incidental findings, over-treatment, and psychological distress, as well as time and financial burden to both the patient and the healthcare system. As Georgieva et al⁴ have demonstrated, it is possible to rigorously quantify and compare the benefits, harms, and costs of different guideline-based evaluations, which we hope will lead to the widespread use of the guidelines that take a more balanced approach to maximize the net benefit to patients and facilitate high-value care.

We applaud the authors thoughtful approach to comparing guidelines for initial hematuria evaluation and believe that this study sets a good example for future guideline development. Clinicians may not always be aware of the degree to which the harms of medical evaluations can outweigh the potential benefits of earlier diagnosis and treatment, and should look to evidence-based guidelines that appropriately balance these competing principles by maximizing net benefit. In the meantime, the best available evidence suggests that uniform use of CT for the initial evaluation of microscopic hematuria is ill advised and should be replaced with risk-based approaches and increased use of renal ultrasound. Furthermore, all guidelines should consider harms and costs as well as benefits.