A decision analysis evaluating screening for kidney cancer using focused

BACKGROUND
Screening for renal cell carcinoma (RCC) has been identified as a key research priority; however, no randomised control trials have been performed. Value of information analysis can determine whether further research on this topic is of value.


OBJECTIVE
To determine (1) whether current evidence suggests that screening is potentially cost effective and, if so, (2) in which age/sex groups, (3) identify evidence gaps, and (4) estimate the value of further research to close those gaps.


DESIGN, SETTING, AND PARTICIPANTS
A decision model was developed evaluating screening in asymptomatic individuals in the UK. A National Health Service perspective was adopted.


INTERVENTION
A single focused renal ultrasound scan compared with standard of care (no screening).


OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS
Expected lifetime costs, quality-adjusted life years (QALYs), and incremental cost-effectiveness ratio (ICER), discounted at 3.5% per annum.


RESULTS AND LIMITATIONS
Given a prevalence of RCC of 0.34% (0.18-0.54%), screening 60-yr-old men resulted in an ICER of £18 092/QALY (€22 843/QALY). Given a prevalence of RCC of 0.16% (0.08-0.25%), screening 60-yr-old women resulted in an ICER of £37327/QALY (€47 129/QALY). In the one-way sensitivity analysis, the ICER was <£30000/QALY as long as the prevalence of RCC was ≥0.25% for men and ≥0.2% for women at age 60yr. Given the willingness to pay a threshold of £30000/QALY (€37 878/QALY), the population-expected values of perfect information were £194 million (€244 million) and £97 million (€123 million) for 60-yr-old men and women, respectively. The expected value of perfect parameter information suggests that the prevalence of RCC and stage shift associated with screening are key research priorities.


CONCLUSIONS
Current evidence suggests that one-off screening of 60-yr-old men is potentially cost effective and that further research into this topic would be of value to society.


PATIENT SUMMARY
Economic modelling suggests that screening 60-yr-old men for kidney cancer using ultrasound may be a good use of resources and that further research on this topic should be performed.

information analyses (VOI) of screening interventions have been undertaken using the 131 currently available evidence, prior to a large trial being undertaken, aiming to determine the 132 value of investing future funds into further research [1]. Indeed, VOI has been used to 133 examine uncertainty surrounding the optimal screening strategy for colorectal cancer and 134 therefore prioritise future research efforts [2]. 135 136 Screening for renal cell carcinoma (RCC) has repeatedly been identified as a research 137 priority [3][4][5][6]. Over a quarter of individuals diagnosed with RCC have metastases at 138 presentation. Five-year age standardized relative survival for these individuals is 6% 139 compared to 84% for those with stage I disease [7]. Ultrasound has been proposed as a 140 screening tool, as it is well tolerated, inexpensive and widely available [8]. National consists of a single focused renal ultrasound, delivered by technicians in the community, 161 similar to AAA screening [21]. If the ultrasound is reported as normal or as a simple cyst, the 162 patient is discharged. Any other abnormality is investigated with an outpatient urology clinic 163 ± CT as appropriate (Supplemental Figure 1). The primary outcomes are the incremental 164 costs (2016 £GBP), incremental quality adjusted life years (QALYs) and incremental cost-165 effectiveness ratio (ICER) comparing one-off screening with no screening. The ICER was 166 defined as the mean incremental costs divided by the mean incremental QALYs. A cycle 167 length of one year and a lifetime time horizon were adopted. Costs and QALYs were 168 discounted at 3.5%/annum. The UK willingness to pay threshold of £20,000-£30, 000

232
Determining the most cost-effective screening population

234
The point estimate ICER is <£30,000/QALY for 50-year-old men and <£20,000/QALY for 60-235 year-old men ( Table 2). The ICER is >£30,000/QALY for women of all ages, however the most 236 favourable ICER is observed for 60-year-old women. Therefore, age 60 years (males and 237 females) was chosen as the base case for all subsequent analyses. For 60-year-old males, there is a 62% probability that the ICER is <£20,000/QALY and a 66% 242 probability that the ICER is <£30,000/QALY. For 60-year-old females, there is a 44% 243 probability that the ICER is <£20,000/QALY and a 56% probability that the ICER is 244 <£30,000/QALY (Supplemental Figure 8). Cost-effectiveness improves as the prevalence increases and the cost of ultrasound 249 decreases (Table 3). Using £37[€47] as the cost of ultrasound, the ICER remains 250 <£30,000/QALY so long as the prevalence of RCC is ≥0.25% for men and ≥0.2% for women 251 aged 60 years. Using our current estimates for the prevalence of RCC for 60-year-old 252 women, the ICER is <£30,000/QALY if the cost of screening ultrasound was reduced from 253 For 60-year-old males, the ICER remains <£30,000/QALY so long as the disutility associated 255 with screening is 0.05 for one week (Supplemental Table 6). The ICER is <£30,000/QALY, if 256 the specificity of ultrasound is ≥85% (Supplemental Table 7). Furthermore, in the base case, 257 it was assumed that the combined prevalence of incidental benign conditions detected by 258 screening would be 2.7% [11,17,18]. The sensitivity analysis demonstrated that in 60-year-259 old men, the ICER remains <£30,000/QALY so long as the combined prevalence of other 260 incidentally detected renal conditions is 20% (Supplemental Table 8 The number of individuals aged 60 years eligible to receive screening in the UK is 362,766 266 men/annum and 374,008 women/annum. Assuming a time horizon for which additional 267 information is useful of ten years, this equates to a population that may benefit from Screening for RCC has the potential to improve survival outcomes [4,5]. However, as with 292 any screening program, there is also a potential for harm, including over-diagnosis, as well 293 as psychological and economic implications for patients and society. No RCTs of screening 294 for RCC have been undertaken [8]. We demonstrate that the population EVPI is £194 million 295 and £97 million for 60-year-old men and women respectively. This suggests further research 296 is likely to be of good value to the funder, and should be focused on estimating the 297 prevalence of RCC and the stage shift associated with screening. 298

299
Determinants of cost-effectiveness 300 301 Using current evidence, this decision model suggests screening may be cost-effective in 302 males but not females, due to lower prevalence of RCC in the latter [11,14]. The true 303 prevalence of RCC by age/sex in the UK is unknown. Sensitivity analysis suggests that 304 screening may be cost-effective if the prevalence is 0.25% for males and 0.2% for 305

females. A meta-analysis demonstrated the prevalence of RCC detected in middle-aged 306
Americans undergoing screening CT is 0.21% [24]. Once again, the prevalence was not 307 reported by age/sex, however it may indeed be above the threshold identified by our 308 sensitivity analysis. Although beyond the scope of the present analysis, risk-stratified 309 screening may increase cost-effectiveness by targeting screening towards individuals with a 310 higher prevalence. At present there is a lack of specific, validated models to predict the risk Evidence on the impact of screening for RCC on QoL is lacking [8,22]. In the base case, it was 331 assumed that undergoing screening ultrasound was not associated with a disutility, and this 332 may contribute to the results demonstrating that the EVPPI for utilities was £0. However, in 333 the sensitivity analysis, we showed that for 60-year-old men if the disutility associated with 334 screening renal ultrasound is ≥0.05 for one week, screening is no longer cost-effective. This 335 is because a small reduction in utility would be applied to such a large number of individuals 336 receiving screening that it would outweigh any benefit to the small minority of patients in 337 which RCC is detected. Therefore, it is essential that any future RCC screening studies 338 evaluate the impact of screening on QoL.