Clinical Prediction Rules and Diagnostic Tests in Physiotherapy

This set of flashcards covers critical concepts from three articles regarding Clinical Prediction Rules, diagnostic test interpretation, and the diagnosis of cervical radiculopathy, including statistical formulas and validation hierarchies.

Who are the authors of Article 1 and what is its type?

Beattie and Nelson; Conceptual review paper.

What is the purpose of Beattie and Nelson's paper?

Describe the potential role of clinical prediction rules in physiotherapy practice and suggest strategies for determining their appropriate use in clinical settings.

How do Beattie and Nelson define Clinical Prediction Rules (CPRs)?

Research-based tools that combine relevant clinical findings to calculate a numeric probability of the presence of a specific disorder or likelihood of an outcome.

What are the three primary uses of CPRs according to Beattie and Nelson?

1: Screening, 2: Prognosis, 3: Classification.

What is the specific definition of screening in the context of CPR uses?

Predicting the likelihood of presence/absence of a condition.

What is the specific definition of prognosis in the context of CPR uses?

Prediction of likelihood of a specific outcome.

What is the specific definition of classification in the context of CPR uses?

Identifying patients most likely to benefit from specific treatment.

When are CPRs most needed and valuable?

In areas of clinical uncertainty where misinference may result in increased risk of adverse events, unnecessary cost, or failure to identify under-diagnosed conditions.

Define Criterion-Referenced validity for a Clinical Prediction Rule.

The CPR score correlates with the gold standard of screening.

Define Predictive validity for a Clinical Prediction Rule.

The initial score is strongly associated with change in patient status over time for prognosis.

Define Prescriptive validity for a Clinical Prediction Rule.

Patients with similar scores are randomized to different treatments, and one treatment shows superior outcomes for treatment classification.

Describe Level 4 of the McGinn Validation Hierarchy.

Preliminary validation on a limited, well-defined population.

Describe Level 3 of the McGinn Validation Hierarchy.

Validated on a prospective, similar sample; results can be applied to a similar population.

Describe Level 2 of the McGinn Validation Hierarchy.

Validated on a variety of patients/clinicians; results can be applied in a variety of settings.

Describe Level 1 of the McGinn Validation Hierarchy.

Impact analysis demonstrates improved outcomes/cost-effectiveness; results can be applied with confidence in most all circumstances.

What are the definitions of sensitivity and specificity as per Article 1?

Sensitivity is the true positive rate; specificity is the true negative rate.

What clinical significance is attributed to Likelihood Ratios (LRs) larger than $10$ or less than $0.1$?

Large changes.

What clinical significance is attributed to LRs of $5-10$ or $0.1-0.2$?

Moderate changes.

What clinical significance is attributed to LRs of $2-5$ or $0.2-0.5$?

Small but important changes.

What clinical significance is attributed to LRs of $0.5-2.0$?

Rarely important.

What was the finding regarding functional capacity evaluations as a limitation of CPRs?

They were found to be weakly predictive of return to work and inversely related to sustained recovery.

What is the example provided for the most validated CPR?

Ottawa Ankle Rules.

Who are the authors of Article 2 and what is its purpose?

Fritz and Wainner; Present an evidence-based perspective on the diagnostic process in physical therapy.

What are the two aspects of 'Diagnosis' in physical therapy as defined by Fritz and Wainner?

1. The process of evaluating data from the examination. 2. The end result (classification label).

What serves as the link between examination findings and interventions in physical therapy?

Diagnosis.

What are the two purposes of diagnostic tests summarized by Fritz and Wainner?

1. Screening/Focusing (detect conditions not appropriate for PT or refine focus). 2. Classification (identify most effective intervention).

What is the optimal study design for diagnostic tests?

Prospective, blind comparison of the diagnostic test and reference standard in a consecutive series of patients from a relevant clinical population.

Why is a Randomized Controlled Trial (RCT) not the optimal design for diagnostic tests?

RCTs are optimal for treatment, but not for diagnosis.

What criteria must a Reference Standard meet according to Article 2?

Valid, reproducible, consistent with test purpose, applied consistently to all subjects, independent of diagnostic test, and judged by blinded examiners.

What are the three requirements for a Diagnostic Test Description?

1. Intended Use, 2. Physical Performance (replicable), 3. Scoring Criteria.

Define Spectrum Bias in the context of diagnostic study populations.

A bias occurring when study subjects are unrepresentative, such as using only asymptomatic controls instead of a relevant clinical population.

How does Spectrum Bias affect diagnostic accuracy?

It artificially inflates diagnostic accuracy.

Define Sensitivity and Specificity in terminology presented by Fritz and Wainner.

Sensitivity is the ability to detect a condition when present; Specificity is the ability to detect absence when absent.

What is the relationship between Predictive Values (PPV/NPV) and condition prevalence?

Positive Predictive Value and Negative Predictive Value are highly dependent on condition prevalence.

What is the relationship between Sensitivity/Specificity and condition prevalence?

Sensitivity and Specificity remain constant across prevalence levels.

What is the formula for a Positive Likelihood Ratio ($+LR$)?

$$+LR = \frac{\text{Sensitivity}}{1 - \text{Specificity}}$$

What is the formula for a Negative Likelihood Ratio ($-LR$)?

$$-LR = \frac{1 - \text{Sensitivity}}{\text{Specificity}}$$

What is the main limitation of using the Chi-Square statistic for diagnostic test interpretation?

It tests association but not the strength or direction.

What is the relationship between reliability and validity according to Fritz and Wainner?

Reliability should not be a prerequisite for validity studies; it is a complement, not a precursor.

Describe the Probabilistic Diagnostic Process sequence.

Pretest Probability $\rightarrow$ Test Result (likelihood ratio) $\rightarrow$ Posttest Probability.

Define 'Treatment Threshold'.

The level of certainty needed to act.

Which tests most efficiently move a clinician from pretest uncertainty to an action threshold?

Tests with the largest Likelihood Ratios (LRs).

What are the two methods mentioned to calculate posttest probability?

1. Fagan Nomogram. 2. Converting Pretest Probability to Odds, multiplying by LR, and converting back to Probability.

In the clinical example for Lumbar Spinal Stenosis, what finding resulted in a positive LR of $6.6$?

Absence of pain when seated.

In the clinical example for Lumbar Spinal Stenosis, what was most useful to rule out the condition?

Ranking sitting/standing/walking, with a LR of $0.33$.

Which test for subacromial impingement is labeled most sensitive for ruling out?

Hawkins-Kennedy Test.

Which test for subacromial impingement is labeled most specific for ruling in?

Drop arm test.

What are the consequences of using non-evidence based diagnosis systems like NDT and McKenzie?

Clinical trials using these systems have shown negative outcomes, leading to conclusions that PT may not be effective.

What did Article 3 (Sleijser-Koehorst et al.) investigate?

The diagnostic accuracy of patient interview items and clinical tests for diagnosing cervical radiculopathy.

What was the study design for Article 3?

Prospective diagnostic accuracy study using consecutive patients recruited from a multidisciplinary clinic.

What was the reference (gold) standard used in Sleijser-Koehorst et al.?

Neurosurgeon diagnosis based on clinical presentation (radicular pain and/or neurological deficit) and positive MRI.

What were the characteristics of the 134 patients in Article 3?

$66$ with cervical radiculopathy, $68$ without; mean age $49.9$ years; 49\text{%} female; median symptom duration $26$ weeks.

What was the prevalence of cervical radiculopathy in the Article 3 sample?

49%

Which patient interview items were identified as best to rule in cervical radiculopathy?

Arm pain worse than neck pain, provocation when ironing, and reduction of symptoms by walking with hand in pocket.

Which patient interview items were identified as best to rule out cervical radiculopathy?

Paresthesia and numbness.

Which clinical tests in Article 3 were best to rule in cervical radiculopathy?

Spurling’s and Reduced Reflexes.

Which clinical tests in Article 3 were best to rule out cervical radiculopathy?

Upper Limb Neurodynamic Test 1 (ULNT1) and Shoulder abduction relief test.

What was the key finding regarding Likelihood Ratio thresholds in Article 3?

None of the interview items or clinical tests reached moderate or high likelihood ratio thresholds.

What were the most common nerve roots affected in the Sleijser-Koehorst et al. study?

$C6$ ($27$ patients) and $C7$ ($24$ patients).

How many patients in Article 3 had both $C6$ and $C7$ involvement together?

$10$ patients.

One strength of the Article 3 study was its design. Give two details.

Prospective design and blinded assessments.

What is a major limitation mentioned for cervical radiculopathy research in Article 3?

There is no universal gold standard for cervical radiculopathy.

What factor, other than the tests themselves, may have influenced findings in Article 3?

Medication use.

What is the clinical application of a positive Spurling's test based on Article 3?

It increases the likelihood of cervical radiculopathy.

What is the clinical application of the absence of paresthesia or numbness based on Article 3?

It decreases the likelihood of cervical radiculopathy.

What does Fritz and Wainner suggest the clinician must understand regarding study design?

Understand study design pitfalls, interpret sensitivity/specificity, and apply LRs to shift pretest probability.

In Article 1, how is sensitivity defined?

The true positive rate.

In Article 1, how is specificity defined?

The true negative rate.

What is the significance of an LR between $0.5$ and $2.0$?

Rarely important.

What numeric probability estimate do CPRs provide?

Numeric probability of the presence of a specific disorder or likelihood of an outcome.

In Article 3, what was the median symptom duration?

$26$ weeks.

What does Article 2 state about 'classification' as a purpose of diagnostic tests?

To identify which intervention will be most effective.

According to Article 2, why must reference standards be independent of the diagnostic test?

To ensure the validity of the comparison and avoid bias.

In the diagnostic process, what follows the Pretest Probability?

The Test Result (expressed as a Likelihood Ratio).

What is the result of multiplying Pretest Odds by the Likelihood Ratio?

Posttest Odds (which can be converted back to Posttest Probability).

What is the Hawkin Test specifically useful for in subacromial impingement?

Ruling out the condition (most sensitive).

In Article 3, provocation by which activity helped rule in cervical radiculopathy?

Ironing.

In Article 3, walking with a hand in what location reduced symptoms?

Hand in pocket.

What did Sleijser-Koehorst et al. conclude about using tests in isolation?

Clinicians must use combinations of findings rather than relying on any single test.

What percentage of the sample in Article 3 was female?

49%