Diagnostic and Screening Tests (Respiratory)

Question 1

What is a diagnostic test?

Answer 1

• used to confirm the presence of disease or otherwise (establish diagnosis)
• used only when there is a high pre-test probability of disease
  
  • clinical suspicion of disease presence
• results are mostly definitive

Question 2

What is a screening test?

Answer 2

• used to identify patients who may have a disease or identify patients with certain levels of risk factors that make them more susceptible to disease
  
  • allows for early interention
• results are preliminary and must be confirmed with a definitive diagnostic test
• applied to individuals where there is no clinical suspicion of disease
  
  • ​ie lower pre-test probability of disease
  • have certain risk factors for the disease
    
    • targeted to high-risk individuals

Question 3

How is validity of diagnostic and screening tests assessed?

Answer 3

• sensitivity and specifcity
  
  • inherent to the test, constant
• postive and negative predictive value
  
  • depend on sensitivity and specificity of the test
  • depend on underlying prevalence of the disease
    
    • ​a highly sensitive and specific test applied to a population with low prevalence will have a small PPV and large NPV = little confidence in test
    • a highly sensitive and specific test applied to a population with a high prevalence will have a large PPV and a large NPV = more confidence in test

Question 4

What is sensitivity?

Answer 4

• of all people that have the disease (denominator is sum of true positives and false negatives), what proportion does the test pick up as being positive?
  
  • sensitivity = TP/TP+FN

= % of people with the disease that test positive

Question 5

What is specificity?

Answer 5

• of the people who do not have the disease (denominator is sum of true negatives and false positives), what proportion will the test pick up as not having the disease
  
  • specificity = TN/TN+FP

= % of people without the disease that test negative

Question 6

What is positive predictive value?

Answer 6

• of all tests that are positive (denominator is sum of true positives and false positives), what proportion are truly positive?
  
  • positive predictive value = TP/TP+FP

= % of positive tests that are truly positive

• positively correlated with underlying prevalence of disease

Question 7

What is negative predictive value?

Answer 7

• of all the tests that are negative (denominator is true negatives plus false negatives), what proportion will the test pick up as truly negative
  
  • negative predictive value = TN/TN+FN

= % of negative tests that is truly negative

• negatively correlated with underlying prevalence

Question 8

What is the utility of diagnostic and screening tests dependent on?

Answer 8

underlying prevalence of the disease

i.e. use the right test for the right people

Question 9

What is a likelihood ratio?

Answer 9

• likelihood that a given test result would be expected in a patient **with **the disease compared to a patient without the disease
  
  • LR of a positive test = sensitivity/(1-specificity)
  • LR of a negative test = (1-sensitivity)/specificity

Question 10

How are sensitivity and specificty related to actual disease state?

Answer 10

• most test results are expressed on a continous scale with arbitrary thresholds that define presence of disease
  
  • eg low FEV1 = COPD, high troponin = MI
  • these states are often not absolute
    
    • ie high FEV1 w/COPD, low FEV1 but no COPD
• if thresholds are set low, get increased sensitivity but decreased specificity
  
  • pick up all people who have disease (100% sensitivity) but pick up half of those without disease as having the disease (50% specificity)
• if thresholds are set high, get decreased sensitivity but increased specificity
  
  • pick up all people without the disease as -ve (100% specificity) but classify 50% of those with the disease as -ve (50% sensitivity)

Question 11

What is a Receiver Operator Characteristic (ROC) curve?

Answer 11

• graphical representation of the trade-off between sensitivity and specificity in tests
• summarizes the capacity of a particular test to distinguish people with the disease from people without the disease
• plot of 1-specificity vs. sensitivity for various thresholds (cut-offs) for a test (red dots)
  
  • sensitivity = probability that people with the disease will test +ve (TP)
  • 1-specificty = probability that people without the disease will test +ve (FP)
    
    • if these two values are equal, the test is worthless (diagonal line)
    • if 100% of people with disease test positive, and 0% without test positive, the test is ideal
      
      • the discriminating ability of the test is measured by the area under the curve between the test and the ‘worthless’ diagonal test
        
        • greater AUC = greater discriminating ability

Question 12

What is the rationale for and use of screeining tests?

Answer 12

• key preventative strategy
• early detection will allow for better outcomes
• assessment of the population to identify:
  
  • risk factors (for primary prevention)
  • early disease (for secondary prevention, to more severe disease)
• commonly undertaken on healthy people

Question 13

What are the WHO criteria for screening tests?

Answer 13

• important health problem
• natural history well understood
• detectable early stage
• early treatment is beneficial
• suitable test for early disease
• acceptable test
• intervals for testing determined
• adequate healthcare provision for extra workload
• risks (including psychological) less than benefits
• costs balanced against benefits
  
  • may require diagnostic follow up to confirm disease presence

Question 14

What are the limitations of screening tests?

Answer 14

• may be inaccurate (low sensitivity, low specificity)
• may not be cost-effective
• adverse physical and psychological side effects (especially FP who have to undergo further, diagnostic testing)
• biases in measurement of effectiveness (ST perceived better than it is):
  
  • selection - screeining more likely to be of healthy people
    
    • more aware of health, more motivated
  • lead-time - early detection does not prolong survival
  • length-time bias - detection of non-aggressive disease or those with long periods of early stage; will have different prognostic outcomes compared to aggressive diseases with shorter early stages (selects for slower-progressing diseases that enable detection)

Question 15

What is lead-time bias?

Answer 15

• perception that survival time is longer with screening, although time of death is the same as without
  
  • ie the disease is picked up earlier, so they live longer with it (survival time) than if it were picked up later
    
    • survival time with screening is longer than survival time without
  • screening does not change the prognosis - still die at same age