In total 979 studies were identified. After removing duplicates and checking for eligibility, 47 full text papers were retrieved (Fig. 1). Sixteen studies were included in the review.
3.1 Study type
Of the sixteen studies, seven compared the cost-effectiveness of new urinary or blood biomarkers compared to each other or to the standard of care (a PSA test) (Table 1). Another seven studies compared different approaches to prostate biopsy. Two studies compared follow-up strategies in men who have a negative initial biopsy result (16, 17). The studies were based in the US (n = 5), UK (n = 4), Netherlands (n = 4), Hong Kong (n = 1), Germany (n = 1) and Canada (n = 1). All but two studies (18, 19) carried out a cost-utility analysis where outcomes were measured in QALYs. The other two were cost-consequence analyses reporting the number of PSA tests and biopsies carried out and expected overall diagnostic costs (19) (18).
Table 1
Studies included following full-text screening
Author | Year | Country | Type of comparison | Tests compared |
Bouttell et al(18) | 2019 | Hong Kong | Biomarker | PHI, PSA |
NICE guideline(16) | 2019 | UK | Follow-up strategy | PSA, PSA velocity, PSA density, % free PSA, PSA doubling time, PSA density intransition zone, PCA3, PHI, mpMRI, no follow-up |
Sathianathen al(20) | 2018 | US | Biomarker | PSA, SelectMDx, PHI, EPI, 4Kscore |
Govers et al (21) | 2018 | US | Biomarker | PSA, SelectMDx |
Barnett et al (22) | 2018 | US | MRI | TRUS biopsy, MRI-guided fusion biopsy, combined biopsy |
Faria et al (23) | 2018 | UK | MRI | TRUS biopsy, mpMRI guided biopsy, template mapping biopsy |
Dijkstra et al(24) | 2017 | Holland | Biomarker | PSA, SelectMDx |
Pahwa et al (25) | 2017 | US | MRI | TRUS biopsy, MRI-guided fusion biopsy, cognitive-targeted biopsy, in-gantry biopsy |
Venderink et al (26) | 2017 | Holland | MRI | TRUS biopsy, MRI TRUS biopsy, in-gantry biopsy |
Heijnsdijk et al(27) | 2016 | Holland | Biomarker | PSA, PHI |
Cerantola et al (28) | 2016 | Canada | MRI | TRUS biopsy, MRI cognitive-targeted biopsy |
Nicholson et al (17) | 2015 | UK | Follow-up strategy | PSA, PCA3, PHI, mpMRI, TRUS biopsy, clinical assessment |
de Rooij et al(29) | 2013 | Holland | MRI | TRUS biopsy, MRI targeted biopsy |
Mowatt et al(30) | 2013 | UK | MRI | T2-MRI, MRS, DCE-MRI, T2-MRI or MRS, T2-MRI or DCE-MRI |
Schiffer et al (19) | 2012 | Germany | Biomarker | PSA, UPA-PC |
Nichol et al (31) | 2011 | US | Biomarker | PSA, PHI |
Legend: PHI – Prostate Health Index, PSA – Prostate-Specific Antigen, mpMRI – Multiparametric Magnetic Resonance Imaging, EPI - ExoDx® Prostate(IntelliScore), TRUS – Transrectal Ultrasound, MRI – Magnetic Resonance Imaging, MRS - Magnetic Resonance Spectroscopy, DCE-MRI - Dynamic contrast-enhanced magnetic resonance imaging, DW-MRI - Diffusion-weighted magnetic resonance imaging, UPA-PC - Urinary Proteome Analysis for PCa diagnosis |
3.1.1 Strategies compared – biomarkers
The novel diagnostic strategies that were compared to PSA-based testing alone included PSA plus Prostate Health Index (PHI) testing (18, 27, 31), SelectMDx testing(21, 24) and Urinary Proteome Analysis (UPA-PC) testing (19). The definitions of these biomarker tests are given in Appendix 3, Table 1. Most studies considered only one novel test, except Sathianathen et al who compared PHI, the 4Kscore, EPI and SelectMDx.(20) Of the seven studies comparing different biomarkers, six referred to TRUS-guided biopsy to confirm diagnosis and one did not report the biopsy method assumed(27). Only two studies (27, 31) modelled repeat PSA/biomarker testing, assuming annual and 4-yearly screening respectively.
3.1.2 Strategies compared - biopsy methods
Men with a suspicion of prostate cancer indicated by a PSA test or other biomarker are generally referred for a TRUS-guided biopsy. The different biopsy methods the seven studies identified compared included MRI-targeted methods and template mapping biopsy (23). The definitions of biopsy methods are given in Appendix 3, Table 2. Different strategies were compared, including using MRI to decide whether a TRUS guided biopsy is necessary and to target biopsy, and strategies starting with TRUS guided biopsy and using MRI to decide whether a repeat biopsy is necessary. One study(30) compared each separate parameter of mpMRI to standard TRUS imaging, rather than comparing a pathway with MRI to one without. Only one study (22) modelled repeat screening, assuming that men would be screened every 2 years based on the 2013 American Urological Association (AUA) guideline.(32)
3.1.3 Strategies compared – follow-up strategies in men with negative biopsies
Two studies (16, 17) compared follow-up strategies for men with raised PSA, negative MRI and/or negative prostate biopsy. The strategies included various biomarkers (PSA, PSA velocity, PSA density, % free PSA, PSA doubling time, PSA density in transition zone, PCA3, PHI), and MRI imaging techniques.
3.1.4 Patient population
The majority of studies specified that their patient population was biopsy-naïve men with an elevated PSA and/or suspicious DRE (Table 2). Three studies stated that their population of interest was men in whom there is still a suspicion of prostate cancer following a negative initial biopsy result, due to clinical or pathological findings (16, 17, 30). Only two studies modelled screening strategies in a population of test-naïve men (22, 31). Only the NICE guideline justified the start and stop ages used, stating the committee advised an age of 75 to be a realistic upper threshold as the average man would be unlikely to be considered for radical therapy on diagnosis beyond this age. Half the studies did not report the age of the cohort modelled. The modelled prevalence of prostate cancer varied from 10.9% (18) to 66.7% (31), but was usually between 24–30%. Several studies reported the percentage of prostate cancers assumed to be high grade or significant (24) (33) (18, 21), ranging from 26%(18) to 51.2%(24).
Table 2
Assumed prevalence of prostate cancer in included studies
Author | Patient population | Age | Prevalence |
Bouttell et al(18) | Chinese men with normal DRE, PSA 4–10 ng/ml | NR | 10.9% |
Mowatt et al(30) | UK men with suspected PC with a prior negative/inconclusive biopsy, with indications for repeat biopsy (i.e. sustained suspicion of PC as a result of clinical and/or pathological findings) | 60 | 24% |
Schiffer et al(19) | German outpatients with PSA > 4 and/or suspicious DRE in a urological outpatient center setting | 66 | 24% |
Nicholson et al(17) | UK men who have been referred for a second biopsy because, following a negative initial biopsy result, clinicians still suspect that malignant PCa is present | NR | 24% |
Cerantola et al(28) | Canadian biopsy-naive men with clinical suspicion of PCa (based on DRE and PSA values 4–10 mg/) with a life expectancy of 20 y | 60–65 | 24% |
Venderink et al(26) | Dutch biopsy-naïve men with elevated PSA level or abnormal DRE | NR | 25% |
de Rooij et al(29) | Dutch men with an elevated PSA level (> 4 ng/ml) | 60 | 25% |
Nichol et al(31) | US men with a PSA 2–10 ng/ml | 50–75 | 25% |
Nichol et al(31) | US men with PSA 4–10 ng/ml | 50–75 | 25% |
Sathianathen et al(20) | US men with PSA > 3 | 50 | 29% |
Nichol et al(31) | US men with positive PHI at PSA 2–10 ng/ml | 50–75 | 29.6% |
Nichol et al(31) | US men with positive PHI at PSA 4–10 ng/ml | 50–75 | 30.3% |
Pahwa et al(25) | US biopsy-naive men who have been recommended for prostate biopsy on the basis of abnormal DRE results or elevated PSA levels | 41–50 | 37% |
Faria et al(23) | UK men at risk of PCa referred to secondary care for further investigation | NR | 38% |
Pahwa et al(25) | US biopsy-naive men who have been recommended for prostate biopsy on the basis of abnormal DRE or elevated PSA levels | 51–60 | 44% |
Dijkstra et al(24) | Dutch men with PSA > 3 ng/ml | NR | 44.4% |
Govers et al(21) | US men with elevated PSA or abnormal DRE who under current guidelines would undergo ultrasound guided biopsy | NR | 46.4% |
Pahwa et al(25) | US biopsy-naive men who have been recommended for prostate biopsy on the basis of abnormal DRE results or elevated PSA levels | 41–70 | 50% |
NICE guideline(16) | UK men who have a raised PSA, negative MRI and/or negative prostate biopsy | 66 | 58.2% |
Pahwa et al(25) | US biopsy-naive men who have been recommended for prostate biopsy on the basis of abnormal DRE results or elevated PSA levels | 61–70 | 65% |
Nichol et al(31) | US men with a PSA > 10 ng/ml | 50–75 | 66.70% |
3.1.5 Treatment types
Eight of sixteen studies reported the percentage of men allocated to each type of treatment (Table 3). The percentage of high-grade men allocated to a radical treatment (prostatectomy, radiotherapy, brachytherapy, hormone therapy or androgen deprivation therapy) varied from 65% (29) to 100% (22). The percentage of men with low grade cancer allocated to a radical treatment varied from 20% (24, 29) to 100% (25). Of the other eight studies, three did not include treatment in their timeframe (17–19), four stated that individual treatments were modelled but did not give the allocation ratio (20, 23, 27, 30), and one stated that they did not explicitly model individual treatments (31).
Table 3
Treatment allocation assumed (%)
Study | Dijkstra et al(24) | Govers et al(21) | Barnett et al(22) | Pahwa et al(25) | Venderink et al(26) | Cerantola et al(28) | de Rooij et al(29) | NICE guideline intermediate risk (high risk in brackets)(16) |
High Grade/Clinically significant | |
RP | 70 | 54 | 100 | 32 | 70 | 30 | 40 | 16 (12) |
RT | 25 | 40 | - | 18 | 25 | 30 | 25 | 35 (35) |
BY | - | - | - | 8 | - | - | - | 3 (1) |
BY + EBRT | - | - | - | - | - | 10 | - | - |
ADT | - | - | - | 33 | - | - | - | - |
RT + ADT | - | - | - | - | - | 30 | - | - |
HT | - | - | - | - | - | - | - | 22 (48) |
WW | 5 | 6 | - | 2 | - | - | 18 | - |
AS | - | - | - | 2 | 5 | - | 18 | 25 (5) |
Low Grade/Clinically insignificant | |
RP | 10 | 50 | 50 | 57 | 40 | 35 | 10 | 18 |
RT | 10 | 30 | - | 7 | 10 | 35 | - | 20 |
BY | - | - | - | 16 | - | 15 | 10 | 7 |
ADT | - | - | - | 8 | - | - | - | - |
HT | - | - | - | - | - | - | - | 9 |
WW | - | - | - | 5 | - | - | 40 | - |
AS | 80 | 20 | 50 | 5 | 50 | 15 | 40 | 47 |
Source | (29) expert opinion | (34) | (35) | (36) | expert opinion | (30) | (36), (37), expert opinion | (38) |
Legend: RP – Radical Prostatectomy, RT – Radiotherapy, BY – Brachytherapy, EBRT – External Beam Radiotherapy, ADT – Androgen Deprivation Therapy, WW – Watchful Waiting, AS – Active Surveillance, HT – Hormone Therapy |
3.2 Model inputs
3.2.1 Accuracy data
All but three studies (17, 28, 31) explicitly reported the sensitivity and specificity of the tests. The assumed sensitivity of a standard biopsy ranged from 0.9 based on ERSPC data (27, 39) to 0.46 based on de Rooij et al (20, 25, 29). The biomarkers were generally either particularly sensitive i.e. good at correctly identifying those with the disease, or particularly specific i.e. good at correctly identifying those without the disease. The SelectMDx test, for example, had the highest sensitivity (0.957, but specificity 0.336) (21, 24, 33) and PHI had the highest specificity (0.974, but sensitivity 0.129) (18, 40). The MRI-targeted biopsy methods generally had a better balance of sensitivity and specificity, ranging from a sensitivity of 0.965 (specificity 0.597) for MRI using a Prostate Imaging–Reporting and Data System (PI-RADS) threshold ≥ 3 (22, 41) to 0.770 (specificity 0.68) using fusion biopsy (20, 26, 42). Appendix 3, Table 3 details the accuracy estimates used along with their evidence sources.
3.2.2 Quality of Life
As detailed in Table 4, all but two studies assigned disutilities to various aspects associated with testing including screening attendance, the biopsy procedure, diagnosis of cancer, treatment, active surveillance, advanced or metastatic cancer, post-treatment or recovery, adverse events associated with biopsy and treatment and palliative therapy. Five of the fourteen studies (17, 21, 22, 24, 26, 27) sourced all utility estimates used in their model from Heijnsdijk et al(6) who in turn obtained their utility estimates from the Cost-Effectiveness Analysis Registry and various additional studies (43–55). The other studies sourced their utility estimates from various unrelated publications, also in different countries and settings. None of the included studies provided Ara et al’s recommended level of detail on health state utility values which are sourced from the literature i.e. detail on searches, inclusion/exclusion criteria, the quality and relevance of included studies, and a justification for the utility values chosen.(56) Only three studies fully reported the uncertainty in the disutility estimates used.(22, 25, 29)
Table 4
Disutility estimates used for prostate cancer states, tests and treatments in the identified economic models (annual values)
Study | Biopsy | Diagnosis | RP | RT | AS | Advanced cancer | Post-treatment | AEs | Other | Source | Report uncertainty |
Barnett et al 2018(22) | 0.006 | 0.017 | 0.247 | | 0.03 | 0.3 | 0.05 | 0.0161 (post-biopsy infection) | 0.0002 (PSA screening) 0.00077 (MRI) 0.60 (Palliative therapy) | (6) (75) | Yes |
Cerantola et al 2016(28) | | | | | | | 0.08 | | 0.22 (relapse) | (76) | No |
de Rooij et al 2014(29) | | | 0.33 | 0.27 | 0.16 | | | | | (46) | Yes |
Dijkstra et al 2017(24) | 0.006 | 0.017 | 0.228 | 0.247 | 0.03 | | 0.05 | | | (6) | No |
Faria et a 2018(23) | 0.007 (TPM biopsy) | | | | | 0.137 | | | | (43), PROMIS IPD (4), (77) | Only for TPM biopsy |
Govers et al 2018(21) | 0.006 | 0.017 | 0.228 | 0.247 | 0.03 | | 0.05 | | | (6) | No |
Heijnsdijk et al 2016(27) | 0.006 | 0.017 | 0.247 | 0.228 | 0.03 | 0.3 | 0.05 | | 0.0002 (Screening attendance) 0.60 (Palliative therapy) | (6) | No |
Mowatt et al 2013(30) | | | | | | 0.365 | | 0.16 (urinary incontinence) 0.17 (bowel problems 0.12 (erectile Dysfunction) | 0.11 (Localised (undiagnosed)) 0.1 (Localised (diagnosed)) 0.19 (Locally advanced (undiagnosed)) 0.18 (Locally advanced (diagnosed)) | (78) (79) (80) | Only for cancer states |
NICE guideline 2019(16) | 0.004, 0.007 (Template mapping biopsy) | | | | | 0.137 | | | 0.027 (low-risk) 0.029 (intermediate-risk) 0.027 (high-risk) | (6, 30, 75, 77, 81, 82) | No |
Nichol et al 2012(31) | 0.027 | | | | | | | | 0.2 (PCa) | (83) (84) (85) | Only for PCa |
Nicholson et al 2015(17) | 0.006 | | | | | | | | | (6) | No |
Pahwa et al 2017(25) | 0.027 | Only lifetime QALYs reported | | (83) | Yes |
Sathianathen et al 2018(20) | 0.004 | | 0.14 | | 0.03 | 0.42 | 0.05 | | | (6, 86) (83) | Yes |
Venderink et al 2017 (26) | 0.006 | 0.02 | 0.25 | 0.23 | 0.03 | 0.55 | 0.05 | | | (6) | No |
Legend - RP: radical prostatectomy, RT: radiotherapy, AS: active surveillance, AEs: adverse events. |
Table 4. Disutility estimates used for prostate cancer states, tests and treatments in the identified economic models (annual values)
3.2.3 Resource use
The majority of studies took a healthcare provider perspective for the analysis (only including costs incurred to the provider rather than any wider patient or societal costs) (16, 17, 19–24, 26, 28–30, 57). Two studies stated that a societal perspective was taken but did not refer to the societal costs that were included (27, 31). Only one study included productivity costs in terms of missed days of work when a patient undergoes a biopsy or MRI (25). No study gave a justification for the perspective taken. The main costs included were the cost of testing, biopsy and subsequent management strategy. Ten studies included costs of complications arising from biopsy (16–18, 21–26, 30). Only six studies explicitly stated that costs associated with complications arising from treatment were included (16, 21, 23, 24, 26, 30).
3.3 Modelling methods
3.3.1 Model type
Six combined decision tree/Markov cohort models were identified. In half of these, the decision tree reflected the diagnostic process and the Markov model reflected treatment (21, 23, 29). In the other half, the decision tree captured both diagnosis and treatment and the Markov model was used for post-treatment states (20, 24, 26). Six cohort Markov models (16, 19, 22, 28, 30, 31), 1 continuous time discrete-event microsimulation model (the MISCAN model) (27), and three decision tree models (17, 18, 25) were also identified. No study provided a justification for choosing one model type over another.
The decision trees generally used data on disease prevalence and accuracy of the tests to categorise men into true positives, false positives, true negatives and false negatives(18, 24, 26) with some also incorporating clinical significance of cancer.(21, 23–25, 29) The Markov models captured cancer progression and survival. All but two studies developed a de novo model.(22, 27)
3.3.2 Cycle length in cohort Markov models
Seven studies had a one-year cycle length(21, 22, 24, 26, 28, 29, 31), two assumed a cycle length of 3 months(16, 30) and one had a cycle length of 6 months(20). The other two studies did not report the cycle length assumed.(23, 58) Only one study justified the cycle length stating that a cycle length of 3 months is sufficient to reflect possible clinical events associated with prostate cancer.(16)
3.3.3 Time horizon
The time horizon of the models varied from when patients reached the treatment stage (18, 19) to their entire lifetime (20–23, 25, 27, 31). Two studies had a time horizon of 18 years(24, 26) as this was the median follow-up time of survival data for patients with prostate cancer, described in the Scandinavian Prostate Cancer Group-4 (SPCG-4) study. (59) One compared results using a 5, 10, 15 and 20 year time horizon (28), one used a 10 year time horizon because ‘after this period no differences were expected between the strategies’ (29), and one used a 30 year time horizon as ‘by this stage the majority of the modelled cohorts were dead and the additional QALYs per cycle had fallen to < 0.001’ (30).
3.3.4 Sensitivity Analyses
All studies conducted a deterministic sensitivity analysis where input parameters or sets of parameters were varied to see the impact on results. Half of the studies (8/16) also carried out a probabilistic sensitivity analysis where repeated simulations sampled all parameters from their respective distributions to observe the impact on results.(16, 17, 20, 23, 25, 29–31) No study carried out a Value of Information analysis to determine the value of further research in prostate cancer screening.(60)
3.3.5 Model Structure
The structure of a model relates to how different health states are categorised and how patients move between health states. Related to this, the natural history of a disease refers to how a disease progresses in a person over time in the absence of treatment.(61) Only five of the included models (16, 22, 23, 27, 62) took account of how prostate cancer progresses, and how the introduction of a new test might impact on this, and all of these captured this progression differently. Five studies modelled survival only (24) (21) (26) (29) (28). Three did not model beyond diagnosis (15–17). In addition, the definition of clinically significant cancer varied across studies (Table 5). Four of the models did not consider stages or grade of cancer, only presence or absence of cancer(20, 21, 26, 31).
Table 5
Study | Model type | Progression modelled | Low Risk | Intermediate Risk | High Risk | Time horizon | Cycle length | DSA | PSA |
Dijkstra et al(24) | Decision tree/Markov | No | G > 6 | - | G ≥ 7 | 18 years | 1 year | Yes | No |
Sathianathen et al(20) | Decision tree/Markov | No | - | - | - | Lifetime | 6 months | Yes | Yes |
Govers et al(21) | Decision tree/Markov | No | G > 6 | - | G ≥ 7 | Lifetime | 1 year | Yes | No |
Faria et al(23) | Decision tree/Markov | Yes | PSA < 10, G < 6 | PSA 10–15 or G7 | G > 8 | Lifetime | Not reported | Yes | Yes |
Venderink et al(26) | Decision tree/Markov | No | - | - | - | 18 years | 1 year | Yes | No |
de Rooij et al(29) | Decision tree/Markov | No | G3 + 3 or small-size 3 + 4 | | large tumours with a G3 + 3 or ≥ 3 + 4 | 10 years | 1 year | Yes | No |
NICE guideline(16) | Decision tree/Markov | Yes | G ≤ 6, PSA ≤ 10 | G = 7 or 10 ≤ PSA < 20 | G ≥ 8 and PSA > 20 | Lifetime | 3 months | Yes | Yes |
Nichol et al(31) | Markov cohort | No | - | - | - | Lifetime | 1 year | Yes | Yes |
Schiffer et al(19) | Markov cohort | No | - | - | - | Up to treatment | Not reported | Yes | Yes |
Barnett et al(22) | Markov cohort | Yes | G < 7 | G = 7 | G > 7 | Until death | 1 year | Yes | No |
Cerantola et al(28) | Markov cohort | No | - | - | - | 5, 10, 15, and 20 years | 1 year | Yes | No |
Mowatt et al(30) | Markov cohort | Yes | G ≤ 6, PSA ≤ 10, ≤T1a | G ≤ 7, PSA ≤ 20, ≤T2b | G > 7, PSA > 20,>T2b | 30 years | 3 months | Yes | Yes |
Pahwa et al(25) | Decision tree | No | G ≤ 6 | - | G ≥ 7 | Until death | - | Yes | No |
Nicholson et al(17) | Decision tree | No | - | - | - | 3 years | - | Yes | Yes |
Boutell et al(18) | Decision tree | No | - | - | - | Up to biopsy | - | Yes | Yes |
Heijnsdijk et al(27) | Microsimulation | Yes | 18 stages (combination of T1, T2 and T3+, G < 7, = 7 and > 7) | Lifetime | - | Yes | No |
Legend: G = Gleason grade, DSA = Deterministic Sensitivity Analysis, PSA = Probabilistic Sensitivity Analysis, - = not included in model |
3.3.6. Data Sources to Inform Progression
Several studies used data from the SPCG-4 study to inform progression risks in diagnosed and undiagnosed men and relative treatment effects in terms of survival (16, 21, 24, 26, 30). SPCG-4 randomly assigned 695 men with localized prostate cancer to watchful waiting or radical prostatectomy from October 1989 through February 1999 and collected follow-up data up to 2017.(63) The studies therefore used data from the watchful waiting arm of SPCG-4 to inform progression in the undiagnosed cases and data from those receiving radical treatments to inform progression and survival in diagnosed cases. Other studies used calibration to prostate cancer specific survival estimates from various studies (64, 65) to estimate the probability of transition through health states (16, 23). Two studies based their transitions through health states on data from the ERSPC(22, 27).
3.4 Cost-effectiveness results
3.4.1 Biomarkers
Of the seven studies that compared PSA testing to testing with PSA plus a new biomarker, four studies found that introducing the new biomarker saves costs and increases QALYs (20, 21, 24, 31). Two did not measure QALYs but found that diagnostic costs were reduced(18, 19). Only Heijndisk et al(27) considered progression through stages or grades of cancer. This study found that PSA + PHI testing saves costs compared to PSA testing and results in the same QALYs (27). The results from all studies were generally driven by a decrease in negative biopsies.
3.4.2 Biopsy methods
Six of the seven studies that compared MRI guided biopsy strategies to each other and to TRUS biopsy found at least one MRI guided strategy to be cost effective (increased costs but also increased QALYs). The exception was Cerantola et al (28) who found that MRI-guided biopsy dominated TRUS guided biopsy (reduced costs and increased QALYs). Incremental Cost Effectiveness Ratios (ICERs) for MRI-guided biopsy methods compared to standard methods ranged from €323 per QALY(29) to $23,483 per QALY (22), indicating cost-effectiveness according to the £20–30 k threshold recommended by NICE(66). The increased QALYs and reduced costs were generally due to an avoidance of the adverse effects and resource use associated with overdiagnosis.
3.4.3 Follow-up strategies
Neither of the studies comparing follow-up strategies in men with a previous negative biopsy identified a clear indication of cost-effectiveness for any strategy. The NICE guideline (16) concluded that PSA velocity, density and %free PSA may be the best indicators to trigger further diagnostics in higher risk populations however the “no screening” strategy appeared optimal for the lowest-risk subpopulation who had MRI Likert scores of 1 or 2 (very unlikely/unlikely that the patient has prostate cancer that needs to be treated) and 2 previous negative biopsies. Nicholson et al (17) found no strategy to be cost-effective at the threshold recommended by NICE.
3.4.4 Assessing uncertainty in cost-effectiveness results
Four studies found that the results were sensitive to the potential of the tests to identify cancer, particularly clinically significant cancer (21, 23–25, 29). Three studies found results to be sensitive to the assumed prevalence of cancer and significant cancer.(25, 26, 29) The cost of the tests was also stated as an important factor in four of the studies, although in these cases the estimated costs would have to change substantially to impact the results (18, 20, 23, 26). Furthermore, studies found results to be sensitive to probabilities of cancer progression in undiagnosed cases (16, 22), survival rates (16, 26), and quality of life values used for diagnosed cancer states (22, 30).