Introduction

Globally there are 430,000 new diagnoses of bladder cancer and over 165,000 deaths every year [1]. It is a deadly disease with only 50% of those diagnosed surviving at 10 years [2].

The most common presentation for bladder cancer is haematuria. Patients with suspected bladder cancer are referred to secondary care where a urologist takes focussed history and examination, carries out a cystoscopy [3] and requests imaging, urine and blood tests (Fig. 1). Cystoscopy is the gold standard test used to diagnose bladder cancer. The imaging test is used to assess the upper tracts for malignant (renal or upper tract urothelial cancer) or benign causes of haematuria (e.g. urinary stones) [4].

The bladder cancer rate in patients referred with visible haematuria is 18.9% but only 4.8% in those referred with non-visible haematuria [5]. Upper tract urothelial cancer (UTUC) is rare with an estimated annual incidence of 2 cases per 100,000 inhabitants [2]. Less than 1% of patients referred to secondary care with haematuria have UTUC [6]. A large proportion of patients investigated therefore will have unnecessary, invasive procedures. Cystoscopy causes discomfort and can carry risks such as infection and bleeding and the radiation from a CT scan is associated with secondary malignancy [7]. Equivocal tests can lead to further invasive testing, e.g. biopsy under general anaesthetic, to confirm a negative finding. Annual costs of these investigations are significant. In the UK, patients with normal results from haematuria investigations cost the National Health Service £33.5 million annually [8].

International guidelines for suspected cancer referral pathways vary greatly. Established risk factors are not featured in referral criteria. This may be due to a lack of high-quality evidence. The AUA (American Urological Association) and UK NICE (National Institute for Clinical Excellence) guidelines differentiate between visible and non-visible haematuria, but the evidence behind this is low [9], [10]. Both give different age thresholds for recommended investigation of each type of haematuria. These arbitrary thresholds are derived from observation of cancer detection rates for wide age-group categories. In the UK, a National Institute for Health Research (NIHR) Health Technology Assessment systematic review highlighted the uncertainty in the optimal diagnostic pathway for haematuria and specified that future studies should address this [11].

Rationale

There is currently no published data describing the variation of current diagnostic strategies and respective cancer yield. Some alternative strategies have been proposed, e.g. the use of CT urography as a triage test to avoid performing 17% of flexible cystoscopies [12]. However, a predictive modelling study suggested CT urography in patients with a low risk of urinary tract malignancy may cause more cancers than it diagnoses due to a high radiation dose [13].

Considering the rarity of upper tract cancers, there is a need for a large-scale study in current clinical practice to investigate the optimal strategy for urological cancer diagnosis in patients with haematuria.

IDENTIFY aims to be the largest contemporary study of haematuria investigation. It will provide data on the utility of diagnostic pathways for bladder and upper urinary tract cancer in patients presenting with haematuria, a contemporary evaluation of these pathways and an assessment of the prevalence of urinary tract cancer (bladder cancer, upper tract urothelial cancer, renal cancer and prostate cancer).

We will assess the predictive ability of recognised risk factors for urinary tract cancer and explore novel risk factors that could improve the selection of patients for referral of suspected cancer. By evaluating the patient factors that predict urinary tract cancer and subsequently classifying them in to high, medium and low risk, we will allow clinicians to select the appropriate intensity of investigations. This risk stratification strategy will prompt earlier diagnosis and treatment of more aggressive disease by allocating resources to patients that need them most. It will also optimise use of the most suitable tests to the patient’s risk of urothelial cancer. The ultimate hope is that risk stratification will reduce the number of invasive procedures and imaging tests performed, reduce the proportion of negative investigations, and increase cancer diagnostic yield.

Since we do not know the best diagnostic strategies to use and which risk factors are relevant, it was not possible to design an interventional study comparing different strategies. Therefore we designed a prospective observational study with a specifically designed data collection tool.

IDENTIFY pilot study

The IDENTIFY pilot study collected data on the incidence of urological tract cancer in 824 patients referred with haematuria in 2016, from 7 hospitals in the South of England. It confirmed the feasibility of the project, the processes required to collect data, design of data collection instruments and design of the project management tools for the IDENTIFY study (Appendix A).

Study design

The IDENTIFY study is a prospective international multi-centre observational cohort study carried out in urological secondary care centres.

Aims and objectives

The IDENTIFY study aims to develop risk-based diagnostic strategies for patients referred to secondary care with suspected urinary tract cancer.

Primary Objective: To determine the prevalence of urinary tract cancer in patients referred to secondary care with suspected urinary tract cancer.

The secondary objectives are to determine:

• •  the prevalence of urinary tract cancers in key clinical subgroups Visible haematuria (VH), Non-visible haematuria (NVH), no haematuria (NH).
• •  the prevalence of urinary tract cancers by age group and sex.
• •  current practices in urinary tract cancer diagnosis across different healthcare settings and the effect on prevalence.
• •  differences in the prevalence of urinary tract cancers in secondary care in different countries.
• •  risk factors that predict urinary tract cancers and define clinically useful risk stratification groups.
• •  the diagnostic accuracy of US, CT and urine cytology for bladder, renal and UTUC.
• •  the diagnostic accuracy of flexible cystoscopy for bladder cancer.
• •  the cost-effectiveness of different diagnostic strategies for suspected urinary tract cancer

Study setting

Patient data will be collected from secondary care centres that evaluate patients with suspected urinary tract cancer and have the ability to perform cystoscopy. A list of anticipated participating countries from pre-registered interest in the study can be found in Table 1.

Eligibility

Patients are included if they are over 16 years of age and were referred to secondary care with haematuria (visible or non-visible), or without haematuria but with suspicious symptoms suggestive of urinary tract cancer.

Exclusion criteria: Patients with a previous or known diagnosis of primary urological cancer, patients referred for suspected recurrence of primary urological cancer, or patients undergoing cystoscopy for a reason unrelated to ruling out urothelial cancer. Patient records lacking sufficient data to determine the primary outcome will be excluded.

Sample size

The sample size was determined pragmatically based on anticipated accrual. Based on a minimum of 50 patient records collected per site and 100 sites, we plan for a minimum sample size of 5000 patients. Based on the prevalence of urinary tract cancer from the IDENTIFY pilot study of 12%, a confidence level of 95% and a sample size of 5000, the precision for the estimate of urinary tract cancer prevalence will be +/− 0.01%.

Data collection

Participating collaborators will complete a registration survey describing their normal protocol for the investigation of haematuria at their institution. Data will be collected on consecutive patients seen for assessment, with a minimum of 50 patients per centre. Data collected is routine information recorded as part of clinical assessment. Some patients may undergo further investigations following their initial tests. These can include biopsies, definitive cancer surgery or transurethral resection of bladder tumour (TURBT). This data will also be collected, and patients will be followed up until histopathology is available (if applicable) or until the outcome of their haematuria investigations is complete, whichever is later. It is not anticipated that follow up for any patient will exceed 3 months. Where no cancer is found, patients discharged from secondary care by a urologist will be determined to have met the clinical threshold for a negative cancer outcome.

Non-identifiable patient data will be collected by individual investigators using REDCap electronic data capture tool [14], [15].

Data will be collected on:

• Hospital protocol (if any) for the investigation of patients with haematuria
• Reason for referral
• Baseline demographic information
• Details of clinical history
• Examination findings
• Bedside urinalysis
• Urine microscopy and cytology results
• Blood test results
• Ultrasound, CT and other imaging results
• Flexible and Rigid cystoscopy results
• Histopathology from any biopsies or surgery
• Further pathology results within the time frame of the study.

Quality control

All submitted data will be quality checked by a dedicated independent quality control team. Data will be checked for completion, outliers and adequacy. Queries will be posed to investigators who will have an opportunity to address any deficiencies.

Missing data: All records will be included in the analysis if there is sufficient data to determine the primary outcome. The primary outcome will not be imputed in any case. Missing data will not be imputed for univariable analysis. Missing data may be imputed in multivariable analysis.

Missing results (tests ordered but not completed) and uninterpretable results (tests performed but inadequate for assessment) will be reported, and reasons examined, but omitted from the main diagnostic test analysis.

Data analysis plan

We will create a formal statistical analysis plan prior to data analysis.

Prevalence will be calculated as follows:

$\frac{\text{Total}\text{number}\text{of}\text{patients}\text{with}\text{target}\text{condition}}{\text{Total}\text{number}\text{of}\text{patients}\text{at}\text{risk}}$

The defined population at risk are all patients in the study who met the eligibility criteria.

We will also calculate prevalence separately for the following groups:

• ·  Visible haematuria
• ·  Non-visible haematuria
• ·  No haematuria

Risk factor analysis

We will use multivariate analysis to assess the association of well-established patient risk factors with individual urinary tract cancers. We will also assess less established and controversial risk factors. Together these will be used to develop risk categories for urinary tract cancers.

Diagnostic test evaluation

Collaborators will score test results on a three-point scale: Normal, Equivocal (defined as a valid test with inconclusive findings) and Positive. Test adequacy will also be recorded [16]. This includes any intended tests that were not performed, tests that were attempted but deemed inadequate, and tests that were completed and were adequate. This will allow for a more accurate diagnostic test evaluation. Sensitivity, specificity, negative and positive predictive values will be reported based on adequately completed tests. Index tests will be compared against specified reference standards for the diagnosis of the relevant cancer type.

Cost effectiveness

Following analysis of results, diagnostic test strategies will be proposed. The estimated costs of these will be estimated by multiplying standard unit costs by key resource use. Where possible, standard unit costs will be derived from sources such as NHS Reference costs in the UK. A cost analysis will be performed on the various diagnostic strategies proposed. The effect of test inadequacy or equivocal test results will also be taken into account for the cost analysis.