EBP Lec 8: Clinical Decision Analysis and Screenings

Learning Objectives

• Recall considerations of diagnostic testing, including different types of cost and maximising patient benefit

• Apply the concepts of Clinical Decisions Analysis in the context of screening (such as the UNHS and adult ABR)

• Recall outcomes of screening (hit, correct rejection, false alarm, miss) and applicability of sensitivity and specificity in a diagnostic context

• Correctly outline what the current UNHS screening process is in Victoria

• Recall the role of ABR in adult diagnostic testing, including when to refer for ABR and management pathways after ABR testing

Clinical decision analysis framework

• Diagnostic testing in audiology must balance: time to perform tests, patient time, test costs, availability, and test accuracy (sensitivity, specificity).

• Core question: Will the test outcome change patient management?

  • If yes, perform the test; if no, avoid it to save costs and time.

• Example discussed: testing reflexes at 4 kHz is possible but often does not change management because it is time-consuming and uncomfortable, so it’s not routinely done.

Costs in healthcare (types of costs to balance with patient outcomes)

• Financial for diagnostics (healthcare and personal): cost to healthcare system, out-of-pocket costs for patients

• Loss of productivity

• Financial for disease burden

• Emotional costs (anxiety, distress for families, newborns during screening)

• Cost of time (time off work, waiting, etc.)

• Cost of not doing anything (i.e. costs of untreated conditions)

And

• Transportation costs and access (getting to appointments)

• Waste and sustainability (consumables, environmental impact)

• Side effects or adverse events from tests

• Examples in newborn screening context: emotional cost of referral for a baby who is later found not to have a condition; cost of missed diagnosis with potential long-term impacts if a condition is not treated early.

Clinical Decision Analysis

• Two components of CDA:
  Measure and Maximise
  → Evaluate usefulness of a test and ensure best outcome for patients

Examples:

• Screenings for bowel cancer, hearing loss in newborns

• Drug trial (cost of drug vs benefit)

• Home vs hospital-based nursing program

We want the most efficient and cost-effective test that will detect the most pathology

Screenings

• identifies individuals at risk of a target condition (what we’re screening for), then go on to have diagnostic testing for the condition

• only have positive or negative outcome (“Pass” or “Refer”)

• Diagnostic tests investigate further:

  • determine presence/absence

  • may determine additional information (e.g. hearing loss severity)

• Outcomes (two-by-two table in newborn hearing screening example):

  • True Positive (TP): baby has hearing loss and screen is positive (referral).

  • False Positive (FP): baby does not have hearing loss but screen is positive (referral). *Error

  • True Negative (TN): baby does not have hearing loss and screen is negative (pass).

  • False Negative (FN): baby has hearing loss but screen is negative (pass). *Error

• Consequences of errors:

  • False Positive: emotional cost, time and resource use for unnecessary diagnostic testing, possible less trusting.

  • False Negative: delayed diagnosis and management, potential higher downstream costs (e.g., speech-language delays, education costs and lower productivity in workforce).

Sensitivity and Specificity

• Sensitivity & Specificity → accuracy of screen

• Sensitivity: a percentage of the test picking up target condition (detect it)

• Specificity: a percentage of the test rejecting target condition (identify it’s not here)

• Ideal screening would be 100% sensitive and 100% specific:

  • only fail screening if HL is present (100% sensitive)

  • only pass screening if no HL is present (100% specific)

Finding the balance

• Ideally maximise true positives (i.e., high sensitivity to pick up all hearing-impaired babies)

• however more sensitive a test is, false positives will also increase

• consider the cost of missing a HL compared to referring lots of babies for hearing assessment that do not need it

Examples: fire alarm - prioritise sensitivity because cost (life & property) is so high

Newborn hearing screening (UNHS): overview and components

• Target condition for universal newborn hearing screening (UNHS) generally defined as bilateral moderate or greater sensorineural hearing loss (SNHL).

• Two main technologies:

  • Otoacoustic emissions (OAEs): fast, easy, cheap; measures outer hair cell function; does not detect auditory neuropathy; higher referral rate; quick screening.

  • Automated Auditory Brainstem Response (AABR or ABR): more expensive and slower but provides information about neural pathways up to the brainstem; can detect auditory neuropathy in some contexts; used as the primary screening tool in Victoria (AABR is automated and not waveform-interpreted by screeners).

• CHIRP stimulus (time-delayed chirp) vs broadband click:

  • CHIRP presents low frequencies first and high frequencies later; results in more synchronous basilar membrane response and shorter screening times without changing intensity: ext{Stimulus intensity} = 35 ext{ dB NHL} is maintained.

  • CHIRP may miss some auditory neuropathy cases because synchrony is less critical for those patterns; some programs are adapting to avoid missing neuropathy.

• Stimulus and target used in Victoria (Australia):

  • Stimulus intensity: 35 ext{ dB NHL}

  • Target condition for screening: bilateral moderate or greater SNHL; not all hearing losses are targeted by the screen.

• Two-stage ABR process (ABR1 and ABR2):

  • ABR1: first screening attempt; most babies pass or proceed to ABR2.

  • ABR2: second screening attempt for those who did not pass ABR1.

  • If ABR2 still not a pass in both ears, the baby is referred to audiology for further testing.

  • A “flip-flop” phenomenon: passing in one ear on ABR1 and failing on ABR2 cannot be summed across ears to declare normal hearing.

• Flow and outcomes:

  • Green path: clear response (pass) in ABR1 or ABR2; move to community surveillance.

  • Brown path: refer to audiology if ABR is not clearly pass; may still result in a diagnosis that is different from the target condition.

• Key terms:

  • “Pass” = screen indicates hearing is adequate for early speech and language development; does not prove normal hearing.

  • “Refer” = screen indicates risk; requires diagnostic audiology to determine presence/severity and management.

  • “Target condition” = condition the screening aims to identify (here, bilateral moderate or greater SNHL).

• Two primary screening options and outcomes:

  • OAEs approach: faster, cheaper; higher referral rate; less likely to detect auditory neuropathy.

  • ABR approach: slower and more costly but provides more information; used in Victoria as the primary newborn screen (AABR).

• Epidemiology and rollout:

  • The VicNIC trial demonstrated that the technology could balance sensitivity and specificity; rollout began in 2005; system evolved to aim for universal coverage with pre-discharge screening.

  • Population coverage around 99.7% (in Victoria in 2019), with some gaps (refusals or births outside hospital).

  • General population prevalence approx. rac{1}{1000} for bilateral moderate or greater SNHL; NICU population has higher risk (~ rac{1}{100}).

• Population-level outcomes and costs (2019 data):

  • Target condition among those referred for diagnostic audiology around 30\%

  • Unilateral hearing loss of any degree comprised about 25 ext{%} of diagnostic results.

  • Permanent bilateral mild hearing loss about 12 ext{%}.

• Practical implications:

  • Newborn hearing screening has dramatically reduced median age of diagnosis from over a year (historical with risk-factor screening) to around 1 month with UNHS.

  • Early identification enables earlier intervention and potentially better language outcomes, though benefits vary by degree of hearing loss and other factors (as shown by later data like the LONG-TERM LOCKY study).

• Historical development in Victoria (to illustrate program evolution):

  • Pre-1992: no universal screening; risk-factor screening only.

  • 1992–2005: risk-factor referral plus maternal/child health nursing behavioral screening (BOA-like approaches) with high resource use and poor reliability.

  • VICNIC trial (early 2000s) validated ABR/OAEs approach; 2005 rollout began, expanding to universal coverage.

  • 2005 onward: pre-discharge ABR-based screening and statewide expansion; ongoing monitoring and data collection by ViSP.

• Guthrie card and metabolic screening context in newborns:

  • Guthrie card: heel-prick blood spot collected at birth; used for screening for metabolic disorders with very high sensitivity and specificity; simple yes/no results.

  • Distinction: metabolic screening (Guthrie card) is a laboratory test measuring a defined disease state; there is a clear “has/has not” result.

  • In contrast, newborn hearing screening detects a risk for a sensory/retrocochlear issue and is not a definitive diagnostic test.

Two screening modalities and target conditions beyond newborns

• Absence of a perfect test for all hearing loss levels means screening will identify some non-target conditions or milder losses; this is acceptable so long as follow-up and management pathways are efficient and accurate.

• Retrocochlear pathology considerations (ABR as screening tool; MRI as confirmatory diagnostic tool):

  • ABR is less sensitive than MRI for detecting retrocochlear pathology (e.g., acoustic neuroma) but is cheaper, faster, and useful for triage.

  • Estimated false positive rate for ABR screening of about 11\% in some Danish studies; MRI false positive rate is under 1\% for acoustic neuroma.

  • The ABR-guided pathway: abnormal ABR → refer to ENT for MRI; normal ABR → still monitor and consider ENT if symptoms persist; not all GPs will refer, so ABR can strengthen referral rationale.

• Cost and access considerations in triage:

  • MRI is more accurate but more expensive; ABR helps triage who should get MRI, reducing unnecessary MRI requests while not delaying care for those who need it.

  • In some settings, MRI cost equivalence to ABR changes the calculus; in others, ABR remains a cost-effective triage step.

Retroc Cochlear pathology: acoustic neuroma/vestibular schwannoma

• Anatomy and presentation:

  • Acoustic neuroma (vestibular schwannoma) is a benign tumor at the cerebellopontine angle (CPA) where pons, medulla, and cerebellum meet.

  • Common presentation: unilateral or asymmetric SNHL, tinnitus, balance disturbances; facial weakness or other cranial nerve symptoms can appear as the tumor grows.

• Diagnostic pathway and management decisions:

  • Diagnosis: usually MRI; ENT involvement for treatment planning and monitoring.

  • Audiological indicators that raise suspicion: unilateral SNHL, asymmetrical SNHL (sensorineural asymmetry), absent reflexes, and speech rollover.

• Speech rollover in testing:

  • Definition: Speech score decreases when presenting at higher levels, indicating possible retrocochlear pathology.

  • Practical threshold: a drop of 20 percentage points or more from the maximum score is considered significant rollover.

  • Example: maximum score 75%; at higher level score 50% → rollover significant (
    75% − 50% = 25% drop).

  • Important caveats: rollover alone is not diagnostic; used to strengthen the case for further retrocochlear investigation.

• Sensorineural asymmetry criteria for retrocochlear concern:

  • A significant sensorineural asymmetry is a difference in air- or bone-conduction thresholds of at least 30\text{ dB} across adjacent frequencies.

  • Change from octave-only to include non-octave (interpolated) frequencies increases sensitivity but also false positives; current practice uses adjacent/interpolated frequencies (not just octaves).

  • If an asymmetry is detected, ABR and ENT/MRI referrals are considered; asymmetries are considered sensorineural in origin (bone conduction used for comparison).

• Reflex patterns and signs:

  • Absent or abnormal acoustic reflexes (tone burst; tone decay) can indicate retrocochlear pathology when combined with other signs.

• Management considerations for amplification:

  • In retrocochlear cases, amplification (hearing aids) must be used cautiously due to potential distortion and rollover; rewards depend on location and severity of pathology.

• Referral pathway and monitoring:

  • Suspected retrocochlear pathology → referral to ENT with ABR as part of the audiological triage; MRI for confirmation; ongoing monitoring (e.g., yearly) if pathology is slow-growing.

ABR in screening vs retrocochlear diagnosis; practical considerations

• ABR screening advantages:

  • Provides information beyond a pure cochlear check; helps identify issues along the auditory pathway up to the brainstem.

  • Two-stage ABR improves accuracy and reduces missed cases while keeping turnaround time manageable.

• ABR limitations and triage role:

  • ABR is less sensitive than MRI for detecting retrocochlear pathology; false positives exist (e.g., ~11\% in some reports).

  • MRI is more accurate but costly; ABR can triage who should proceed to MRI, balancing sensitivity, specificity, cost, and access.

• Diagnostic pathway after ABR:

  • Abnormal ABR: report to GP, recommend ENT referral for MRI; provide patient-facing language about the test and the need for specialist follow-up.

  • Normal ABR: still may warrant ENT follow-up in some cases; ABR does not completely rule out retrocochlear pathology.

• Considerations for tele-audiology and access:

  • Digital and remote screening/tracking can impact costs and timeliness; planning for access in remote populations is part of optimizing the system.

CMV screening and emerging screening approaches

• CMV (cytomegalovirus) as a contributor to congenital SNHL:

  • CMV infection can cause sensorineural hearing loss and other neurodevelopmental issues if untreated.

• Screening approaches:

  • Guthrie card (birth blood spot) has poor sensitivity for CMV; retrospective testing is possible but unreliable for early intervention.

  • Saliva or urine PCR testing within the first 21 days of life has higher sensitivity and specificity for CMV, enabling earlier antiviral interventions when needed.

  • Guidelines suggest offering salivary CMV testing when a child fails the newborn hearing screen within the first 21 days to enable timely management.

• Practical implication:

  • Incorporating CMV screening with newborn hearing screening could improve outcomes by enabling earlier antiviral treatment and better developmental trajectories if CMV is the underlying cause.

Population-level benchmarks and key performance indicators (KPIs)

• Joint Committee on Infant Hearing (JCIH) benchmarks for resource-rich environments:

  • Population coverage for UNHS: high (near universal access).

  • Referral rate should be low enough to not overwhelm diagnostic services while retaining sensitivity.

  • Follow-up rate to diagnostic testing after referral should be high.

  • Diagnostic accuracy and enrollment in early intervention programs should be achieved by defined ages: aim for screening by 1 month, diagnostic assessment by 3 months, and early intervention by 6 months (the 1-3-6 rule).

• Evolution of KPIs and timelines:

  • In the Victoria program, the aim has been to achieve rapid progression through the pathway with a pre-discharge (in-hospital) ABR screening and subsequent diagnostic processes.

  • 2019 ViSP data show breakdowns by what the diagnostic audiology findings were (Target Condition, unilateral loss, permanent bilateral mild loss, etc.).

Practical data and examples from the Victorian program

• Population coverage and workflow (2019 ViSP data):

  • Coverage: around 99.7\% population coverage; slight gaps due to refusals or out-of-hospital births.

  • Diagnostic follow-through: strong capacity for diagnostic audiology services in the state; differences exist across states (e.g., NSW) in cross-border patient flows.

  • Diagnostic results by screening: about 30\% had the target condition; unilateral loss of any degree ~25\%; permanent bilateral mild loss ~12\%.

• Testing intensity and pacing:

  • The stimulus used in screening: 35\ ext{dB NHL}; the CHIRP approach reduces time to screen but must balance sensitivity for neuropathies.

  • The proportion of infants exiting after ABR screening to the surveillance process is substantial, illustrating the efficiency of the two-stage ABR model.

• Turnaround times and milestones:

  • Time from screen to diagnostic audiology and then to hearing aid fitting (where applicable) is tracked with green boxes showing median days; the system aims to minimize time to diagnosis and intervention.

• The LOCKY study (Longitudinal Outcomes of Children with Hearing Impairment):

  • A multi-state study comparing early versus late identification and amplification.

  • Used BE4FA (Better Ear Four-Frequency Average: average of thresholds at 500, 1000, 2000, 4000 Hz) to assess language outcomes at age 5 across three severity bands (30 dB HL, 50 dB HL, 70 dB HL).

  • Findings: Early identification and amplification yield greater language outcomes for more severe hearing losses (70 dB); for mild losses, the difference is smaller; early amplification may not always translate into large gains for mild losses due to several interacting factors.

• Clinical interpretation from LOCKY data:

  • Early detection and amplification are most clearly beneficial for more significant losses; policy and clinical decisions about screening emphasis should consider this differential impact.

Practical implications for clinical decision-making and patient care

• The clinical decision analysis framework informs which tests to perform in order to maximize patient benefit while minimizing costs and patient burden.

• Screenings are designed to efficiently triage who needs full diagnostic workups; the cost-benefit balance is central to program design and ongoing policy decisions.

• In newborn screening, high catchment and adherence to follow-up are essential for real-world effectiveness; trust and communication with families are crucial to ensure adherence to referral pathways.

• For retrocochlear pathology, ABR serves as a screening/triage tool rather than a definitive diagnosis; MRI remains the gold standard for confirmation but is more resource-intensive.

• Ethical and practical implications:

  • False positives can cause parental distress and unnecessary medical testing; false negatives can delay critical early intervention.

  • System design should minimize harm from false results while ensuring high sensitivity for treatable conditions.

  • Access and equity considerations are central: remote populations, tele-audiology, and cross-border service access all affect outcomes.

• Next steps in education and practice:

  • Understand the differences between OAEs and ABR, including suitability for different target conditions and the implications for referral rates.

  • Be comfortable with the ABR screening workflow, including two-stage ABR, CHIRP vs broadband click stimuli, and interpretation caveats (e.g., flip-flop, rollover).

  • Be able to discuss with families what a referral means and what steps will occur next in the diagnostic pathway.

  • Keep in mind the evolving guidelines (JCIH 1-3-6; later pushes toward even earlier intervention) and how they apply to local health service realities.

Quick reference: key terms and formulas (LaTeX)

• Target condition for UNHS (typical): bilateral moderate or greater SNHL.

• Two-by-two outcomes (screening): TP, FP, TN, FN.

• Sensitivity and Specificity:

  • ext{Sensitivity} = rac{TP}{TP + FN}

  • ext{Specificity} = rac{TN}{TN + FP}

• BE4FA (Better Ear Four-Frequency Average):

  • ext{BE4FA} = rac{TH(500) + TH(1000) + TH(2000) + TH(4000)}{4}

• Targeted hearing loss definitions (illustrative):

  • Bilateral moderate or greater SNHL: thresholds in both ears at or above the moderate level across key frequencies.

• CHIRP stimulus and intensity in UNHS:

  • 35\text{ dB NHL} (intensity used in many ABR screenings)

• Time targets in JCIH 1-3-6 framework (common baseline):

  • Screen by age 1\text{ month}

  • Diagnostic assessment by age 3\text{ months}

  • Early intervention by age 6\text{ months}

• Population and prevalence (illustrative Victorian data):

  • Population coverage: 0.997 (99.7%)

  • General prevalence: rac{1}{1000} for bilateral moderate or greater SNHL

  • NICU prevalence: rac{1}{100}

• ABR false positives (in some reports): ext{False positives rate} \approx 0.11

• ABR screening intensity and outcomes:

  • ABR1 pass rate and ABR2 pass rate determine the majority going to green path (surveillance) vs refer-to-audiology paths.

Clinical Decision Analysis Framework

Purpose

To balance:

• Diagnostic accuracy and sensitivity,

• Time and cost efficiency, and

• Impact on patient management and outcomes.

Key Considerations

• Time (clinician + patient)

• Cost (financial + emotional + societal)

• Test sensitivity and specificity

• Availability of equipment/personnel

• Whether the test result will change patient management

Example

• Reflex testing at 4 kHz: technically possible, but time-consuming, uncomfortable, and rarely changes management → typically unnecessary.

3. Types of Costs in Healthcare

4. Screening vs. Diagnostic Testing

Clinical Decision Analysis Goal

• Maximise benefit (catch true cases early)

• Minimise cost (limit unnecessary diagnostics)

5. Screening Outcome Matrix

Cost of Errors

• False Positive: emotional distress, wasted time, loss of trust, extra costs.

• False Negative: more severe → missed diagnosis, delayed speech/language development, missed cochlear implant window, lifelong educational and economic consequences.

6. Sensitivity & Specificity

• High sensitivity → fewer missed cases but more false alarms.

• High specificity → fewer false alarms but risk missing true cases.

• Must balance both depending on cost of errors.

Analogy

7. ABR as a Screening Tool

Two Major Uses

1. Newborn Hearing Screening

2. Retrocochlear Pathology Management (adults)

8. Screening Example – Newborn Hearing Screening

Laboratory vs. Functional Screening

• Metabolic (e.g., Guthrie card): clear yes/no lab result.

• Hearing: subjective threshold definition; no single “blood test”.

Cost Justification

• Late-identified profound loss ≈ $1 million educational support per child.

• ~70–80 babies/year in Victoria with bilateral, ≥ moderate SNHL → $70–80 million impact vs. far lower screening cost.

9. History of Victorian Newborn Hearing Screening

• NICU risk: 1 in 100 babies with ≥ moderate SNHL (vs 1 in 1000 general).

• Screening before discharge ensures maximum coverage.

10. AABR Technology

• AABR = Automated Auditory Brainstem Response.

• Uses algorithmic waveform template → screener doesn’t interpret waveforms.

• Detects presence/absence of neural response to sound.

11. Stimulus and Procedure Details

• Stimulus type: initially click, now CE-CHIRP (since 2021)

  • Low → high frequency sequencing improves synchrony.

  • Faster test time; still 35 dB nHL intensity.

• Target condition: Bilateral, ≥ moderate permanent SNHL.

• CHIRP may slightly reduce detection of auditory neuropathy (fewer failures due to better synchrony) → clinical monitoring under revision.

12. Newborn Hearing Screening Pathway

Terminology

• Pass = auditory nerve responding at 35 dB nHL → adequate for speech/language development.

• Refer = needs diagnostic follow-up (“ticket to audiology”).

  • Not “fail” → family-friendly language.

Two-Stage AABR Protocol

1. AABR 1: Initial screen.

   • Must obtain clear response in both ears in one session.

   • If not → AABR 2.

2. AABR 2: Repeat screen (separate day).

   • If still not clear both ears → referral to diagnostic audiology.

Flow Outcomes

• Most babies → AABR 1 pass → discharge.

• Some → AABR 2 pass → community surveillance.

• Few → refer → diagnostic ABR.

13. JCIH Benchmarks (“1-3-6” & “1-2-3” Rules)

• Australia meets/exceeds benchmarks due to strong infrastructure.

• Emphasis: earlier = better → stronger speech/language outcomes.

14. Victorian Performance (2019)

• Population coverage: 99.7%.

• Median timeline:

  • Diagnostic completion ≈ < 60 days.

  • HA fitting ≈ < 70 days.

• Follow-up: High (due to strong public network).

• Outcomes: Among diagnostic cases:

  • ~30% = target condition.

  • ~25% = unilateral losses.

  • ~12% = bilateral mild loss.

• Confirms expected false-refer rate (intentional safety margin).

15. Impact of Universal Screening

Median Age of Diagnosis

→ Enables timely fitting and improved outcomes.

16. The LOCHI Study (Longitudinal Outcomes of Children with Hearing Impairment)

• Conducted by NAL (Hearing Australia).

• Compared states with vs. without newborn screening.

Design

• Children with bilateral ≥ moderate SNHL.

• Compared age at first fitting (months) vs global language score at 5 years.

Findings

17. Transition to Retrocochlear Pathology Section

Definition

• Retrocochlear = beyond the cochlea (8th nerve → brainstem → temporal lobe).

• Most cases → vestibular schwannoma / acoustic neuroma.

Pathophysiology

• Benign (non-cancerous) tumour in cerebellopontine angle (CPA).

• Compresses adjacent nerves → auditory + vestibular symptoms.

Common Presentation

• Middle-aged adults.

• Unilateral or asymmetrical:

  • Hearing loss,

  • Tinnitus (differs between ears),

  • Aural fullness,

  • Distortion,

  • Mild imbalance.

• Severe/late → facial weakness, swallowing difficulty, headaches.

Diagnosis

• MRI (gold standard).

• Audiologists → identify red flags → refer for medical imaging.

18. Audiological Red Flags / Findings

Speech Rollover

• Normally, louder = better clarity.

• In retrocochlear loss, louder = poorer clarity (distortion).

• Criterion: ≥ 20 % absolute drop from max score.

• Suggests 8th-nerve pathology → supports referral for ABR / MRI.

19. ABR in Retrocochlear Screening

• Detects delayed or absent Wave V on affected side.

• Compared interaural Wave V latency (> 0.4 ms difference = abnormal).

• Used as triage tool before MRI.

20. Key Clinical Reasoning Principles

✅ Summary Table