EBM analysis BMJ EBM: first published as 10.1136/bmjebm-2020-111629 on 7 June 2021. Downloaded from ‘Diagnostic downshift’: clinical and system consequences of extrapolating secondary care testing tactics to primary care Imran Mohammed Sajid ,1,2 Kathleen Frost,3 Ash K Paul4 10.1136/bmjebm-2020-111629 ABSTRACT Several drivers push specialist diagnostic Numerous drivers push specialist diagnostic approaches down to the broader primary care layer of a health system, which we describe as ‘diag- 1NHS West London Clinical approaches down to primary care (‘diagnostic Commissioning Group, downshift’), intuitively welcomed by clinicians nostic downshift’. Aspirations for earlier disease London, UK and patients. However, primary care’s different detection or capacity pressures in specialist and 2University of Global Health population and processes result in under- cancer pathways underlie shifting of tests from Equity, Kigali, Rwanda recognised, unintended consequences. Testing high- cost hospital settings to primary care. Such 3 NHS Central London Clinical performs poorer in primary care, with indication assumptions have led to procurement and growth Commissioning Group, creep due to earlier, more undifferentiated of GP diagnostics, with unfettered direct access to London, UK presentation and reduced accuracy due to physiology tests, endoscopy, ultrasound, MRI, CT, 4NHS South West London spectrum bias and the ‘false-positive paradox’. and biochemical and immunological tests, often Health and Care Partnership only evaluated in secondary care. It is increasingly STP, London, UK In low- prevalence settings, tests without near- 100% specificity have their useful yield eclipsed expected, sometimes mandated, for GPs to perform by greater incidental or false- positive findings. secondary care- based testing strategies prior to Correspondence to: Ensuing cascades and multiplier effects can referral. Earlier diagnostics are presumed to accel- Dr Imran Mohammed Sajid, generate clinician workload, patient anxiety, erate patient journeys (eg, decision-making at NHS West London Clinical further low- value tests, referrals, treatments and first outpatient appointment) or reduce referrals (NHS). Protected by copyright. Commissioning Group, a potentially nocebic population ‘disease’ burden by empowering GPs. However, diagnostic growth, London, London, UK; 1 of unclear benefit. Increased diagnostics earlier in heavily cited for low- value overuse, has unin- imransajid@ nhs. net pathways can burden patients and stretch general tended consequences. In the COVID-19 pandemic practice (GP) workloads, inducing downstream context, reduced hospital access and increasing service utilisation and unintended ‘market virtual consultations may proliferate community failure’ effects. Evidence is tenuous for reducing testing. http://ebm.bmj.com/ secondary care referrals, providing patient Clinicians intuitively welcome tests, bolstering reassurance or meaningfully improving clinical autonomy and professional confidence, while outcomes. Subsequently, inflated investment in playing into patients’ biases. Medical risks should per capita testing, at a lower level in a healthcare be discussed, yet test inaccuracy and cascades system, may deliver diminishing or even negative go unrecognised,2 while clinicians poorly inter- economic returns. Test cost poorly represents pret results.3 4 Tests can guide management, but ‘value’, neglecting under- recognised downstream also generate anxiety, low- value disease labels, on June 14, 2021 at GPs and Practice Staff in London, Kent & Surrey consequences, which must be balanced against fear avoidance behaviours, further investigations, therapeutic yield. With lower positive predictive referrals and treatment cascades of little benefit.5–7 values, more tests are required per true diagnosis Consider vitamin D, recommended only in select and cost- effectiveness is rarely robust. With patients, now ubiquitously screened, with almost fixed secondary care capacity, novel primary 100- fold increases, reflecting massive costs, time care testing is an added cost pressure, rarely and prescribing of spurious value.8 reducing hospital activity. GP testing strategies require real-world evaluation, in primary care populations, of all downstream consequences. Testing in primary care is different from Test formularies should be scrutinised in view of secondary care the setting of care, with interventions to focus Tests are integral to primary care, with overlapping rational testing towards those with higher pretest symptoms between benign and serious conditions © Author(s) (or their probabilities, while improving interpretation and and uncertainty in up to 40% of consultations.9 employer(s)) 2021. No communication of results. However, secondary care has different populations, commercial re- use. See workflows and expertise; thus, diagnostic tactics rights and permissions. should not be blindly extrapolated. General prac- Published by BMJ. tice performs a technical but also wider psycho- To cite: Sajid IM, Frost K, social role for non-specific presentations, with a Paul AK. BMJ Evidence- Introduction more person- orientated, than pathophysiology- Based Medicine Epub ahead Are more tests, earlier in pathways, within primary orientated focus.10 While uncertainty management of print: [please include Day care helpful? While welcomed by clinicians, depends on psychological factors,11 12 there tends Month Year]. doi:10.1136/ patients and policymakers, we explore under- to be diminishing decision- making value from bmjebm-2020-111629 recognised consequences. additional tests.13 BMJ Evidence- Based Medicine Month 2021 | volume 0 | number 0 | 1 EBM analysis BMJ EBM: first published as 10.1136/bmjebm-2020-111629 on 7 June 2021. Downloaded from Diagnostic downshift’s pretest considerations (table 1) include patients.17–19 Systematic review shows little to no high- quality inflated (inappropriate) tests per capita, test indication creep and evidence of clinical or cost benefits to support increasing tests in inadvertent screening. Suspicion to trigger a test will typically be primary or community settings, with only low-quality evidence of lower than that for a referral. Thus, more patients are tested than reducing referrals, suggesting such diagnostic strategies may be otherwise referred. Post- test dynamics (table 2) include altered more politically motivated.20 performance (false- positive paradox and spectrum bias) with lower positive predictive values, greater false- positive rates, a burden of Earlier testing does not necessarily improve cancer incidental findings, misinterpretation problems (particularly for outcomes serial testing), limited reassurance and ‘multiplier effects’ of low- Rhetoric around cancer detection system delays often drives diag- value cascades. nostic expansion.21 22 However, there is a paucity of evidence that Diagnostic sensitivity trades off against specificity. Compared advanced GP testing improves outcomes (survival rate statistics with secondary care encounters, GPs do not require imme- are misleading due to lead time, length bias or overdiagnosis of diate high- sensitivity testing tactics, as patients can be referred indolent disease). Impact of ‘delayed diagnosis’ is mixed, including onwards for evaluation, as well as readily reattend for persisting the so-called ‘waiting- time- paradox’ (‘delay’ associated with or worsening symptoms. Test specificity is more critical to manage improved outcome for some cancers).23–25 Diagnostic strategy, referral appropriateness. In low-prevalence settings, without near- particularly for low- but- not- no- risk presentations, is complex. 100% specificity, benefit (diagnostic yield) is eclipsed by greater Systematic review of GP direct access testing suggests, although false- positives or incidental findings (see table 2). With pretest time- to- test may improve, there is no change in time- to- diagnosis probability <10% (common for primary care), even with 90% or outcomes.26 Novel pathway triage may enable prehospital specificity, Bayesian analysis shows greater false- positives than diagnostics, although it has its own drawbacks; in ‘straight-to- true- positives, with positive predictive value no better than a coin test’ pathways, alternate diagnostics may have been preferred toss. For example, carotid artery ultrasound screening, with 92% by specialists.27 Furthermore, pre- referral laboratory cancer tests specificity, across 100 000 patients, generates 7920 false- positives are broadly unreliable, including many biomarkers.28 Without versus only 940 true- positives.14 evidence, novel technologies should be cautioned, considering High-quality studies rarely demonstrate benefit from already pressured workloads. (NHS). Protected by copyright. advanced testing in primary care Outside of low- income countries, there is little evidence to suggest increased diagnostic direct access resolves the problem of Despite disseminated use in different populations, for wider indica- misdiagnosis in primary care, which is due to a myriad of factors, tions, traditionally specialist tests are rarely robustly evaluated in including cognitive reasoning errors.29 30 primary care, reliant on haphazard postmarket surveillance, such as audits. Referral reduction is often based on self- report without capturing downstream utilisation and typically lacks usual care Expansive testing in primary care creates a population comparator analysis. The few randomised controlled trials, such ‘disease’ burden of unclear