Screening and diagnostic tests Flashcards
Types of prevention
- Primary-eg limiting salt consumption to prevent hypertension
- Secondary-eg prescription of antihypertensives
- Tertiary to eliminate long term complications of disease-eg regular blood glucose checks for someone with type 2 diabetes
Screening is a major form of secondary prevention
What is screening?
- Screening purpose is to detect abnormality is asymptomatic or healthy people.
- Screening programmes are applied to whole population groups
- People who have no symptoms will not want an invasive test
- A positive screening test does not mean that a person has the disease, it just means that the person is more likely to have the disease
A positive screening test does not mean that a person has the disease, it just means that the person is more likely to have the disease:
Different types of screening:
- Case finding (technically not screening)-this is where an individual goes to see a clinician for one reason and at the same time that professional checks them for another condition eg go to Dr for back pain and they take BP and find it’s abnormally raised. This is not technically screening as there was nothing systematic about the way I was called or tested, it was purely opportunistic.
- Targeted screening-might identify a particular population who are more at risk of disease or more vulnerable to its impacts, eg TB in a prison population. We may want to systematically identify patients with TB, so that we can treat them and make sure TB has not passed around the prison. Targeted screening is about selecting a targeted population and adopting a systematic approach to inviting, testing and where appropriate treating the affected individuals.
- Mass screening is where we screen a huge population. There may be age, and or sex limitations on this eg we tend to only screen women for breast cancer.
Screening is not the same as diagnosis:
Before we explore screening programmes in more depth, it is important to highlight one of the key points from the video - Screening and diagnosis are not the same.
Screening gives an indication of the chance that somebody has a disease. Screening finds people at a higher chance of having the condition in question. For those deemed high risk, a diagnostic test will be performed. Diagnostic tests tell you whether the condition is actually present.
Take the cervical screening test for example:
- A smear test is a screening test available to all females in the UK to try and help prevent cervical cancer. In this test, a cervical brushing is taken which is tested for the presence of human papillomavirus (HPV) that can cause changes to the cervix.
- If the test is negative for HPV, there is a very low chance that the patient will have cervical cancer. The patient is informed the result is normal and will continue to have routine screening.
- If the test for HPV is positive, the sample will be then tested for cell abnormalities. If there are cell abnormalities which are deemed high risk then the patient will undergo biopsy or colposcopy. These are diagnostic tests.
The differences between screening and diagnostic tests are summarised in the table below:
What is a screening programme?
An effective screening programme must include the following:
7 steps of a screening pathway:
- Identify the people to invite (eg pregnant people are invited during their antinatal appointments, others may be invited by letter based on info taken from their healthcare records)
- Discuss and offer screening-may be done by leaflet or appointment
- Carry out screening test-eg smear
- Provide and discuss results/options-must be timely
- Carry out diagnostic/confirmatory test if required-if screening test is positive check if patient actually has disease
- Provide and discuss results/options
- Offer advice/treatment-ongoing advise
Consider the 7 steps of the screening pathway and the issues related should a step not be followed as may be the case in private “screening” programmes.
-
Patients are not identified and invited
- The population is self-selecting meaning the test may offered to an inappropriate person and there is inequitable access.
-
Screening is not discussed
- The patient does not understand the risks/benefits of screening, their questions/concerns are not addressed
-
Carrying or not carrying out screening test
- If previous steps are not followed then this may be an unnecessary intervention for the patient, which carries risk of physical/psychological harm . In the case that the screening test is not carried out then the patients may not have fully understood the risks/benefits.
-
Results/Options of screening test are not provided/discussed
- Patient unable to identify what to do with the results of the screening test, further treatment may or may not be chosen by patient, harm could come to the patient in either scenario
-
Diagnostic test not offered
- Patients with positive screening results are unable to find out if they actually have the disease
-
Results/Options of diagnostic test not provided/discussed
- Patient unable to identify what to do with the results of the test and what next steps to take, leading to physical/psychological harm
-
No treatment offered
- Screening has had no difference on patient’s health and may suffer psychological harm from result. Think about the ethical considerations here for possibly misleading someone to take a test that will have no benefit to them, having undergone physical/psychological risks along the way.
But why don’t we screen for everything?
We’ve seen from the previous videos that screening programmes allow for earlier identification and treatment of disease. So why don’t we screen for every condition?
The next video will introduce you to the Wilson-Jungner criteria - the criteria used to identify whether a disease is appropriate for a screening programme.
What are the 4 themes and 10 criteria?
Wilson-Jungner criteria-disease
Wilson-Jungner criteria-screening test
Wilson-Jungner criteria-diagnostic test and treatment
Wilson-Jungner criteria-overall screening programme
Define sensitivity, specificity, positive and negative predictive values.
Sensitivity = True positives/ (True positives + False negatives). Sensitivity is defined as the proportion of those with the disease who tested positive. This is also known as the true positive ‘rate’.
Specificity = True negatives/ (False positives + True Negatives). Specificity is defined as the proportion of those without the disease who tested negative. This is also known as the true negative rate.
In sensitivity and specificity the disease state is the denominator.
For example, the denominator in sensitivity is ‘True Positives + False Negatives’ which is equal to the ‘Total number who are disease positive.’ Similarly, for specificity the denominator is ‘disease negative’ – the sum of false positives + true negatives.
Sensitivity and specificity are fixed. These are properties of the test itself and do not change based on the population. This is important to remember when we start to think about negative and positive predictive values.
Calculating sensitivity and specificty
The denominator for both is the disease state ie having the disease or not having the disease
2) How many false negative tests were there?
3) What is the sensitivity of the test? (to 1 decimal place)
4) What is the specificity of the test? (to the nearest whole number)
5) What do you think about the reliability of this test?
1)= 33
Working
178-145 = 33
2)=21
Working:
322- 301 = 21
3)=145/166 x 100 = 87.3%
4)= 301/334 x 100 = 90%
5) Reasonably good – but even with relatively high sensitivity/specificity we still see significant number of FP/ FN in a small population size.
- Do you think this test would be acceptable to be rolled out as a screening test for breast cancer in the UK?
- FP = 132,000 and FN = 84,000
- When making this decision we would have to think it about whether it would be possible to feasibly give a diagnostic test to all 132,000 false positives in order to rule out breast cancer. Would it be acceptable to miss the cancer in 84,000 people given false negatives? Or should be identifying a more reliable screening test with fewer false negatives?
Positive and Negative Predictive Values
We have now learnt how to understand and calculate specificity and sensitivity of a test and you can now start to see that tests are not perfect.
But what does this mean in a clinical context? When we are working clinically, we need to be able interpret positive and negative test results. That is to understand – how likely is my patient to have the disease if they have tested positive?
Now watch this video about positive and negative predictive values:
To recap a couple of key points from the video:
Positive predictive value is the proportion of positive tests that correctly identified those with the disease. In other words, the probability that following a positive test result, an individual will truly have that disease.
PPV = True positive/ (True positive + false positive). Here, the denominator is the positive test result. ie true positive/ALL positive
Conversely, negative predictive value is the proportion of negative tests that correctly identified those without the disease. In other words, the probability that following a negative test result, an individual will truly not have the disease.
NPV = True negative/ (True negative + false negative). Here, the denominator is the negative test result
These values are particularly valuable clinically – without understanding what the positive predictive value, or negative predictive value of a test is, we can’t really interpret test results.
Importantly, predictive values are impacted by the prevalence of a disease.
This can be difficult to get your head round at first. But think of this in practice: If you had a positive test result in a patient for a disease that is incredibly rare (very low prevalence), you are less likely to accept this as a true result as if you had a positive test for a very common (high prevalence) condition.
PPV of 50% means that half of the tests are wrong, which is pretty terrible. But the same test when applied to a population where the disease prevalence is 10%, yields a PPV of 99%
Well’s score:
- What is the positive predictive value of d-dimer in this population? (Give answer to 1 decimal place)
- What is the negative predictive value of d-dimer in this population? (give answer to 1 decimal place)
- How do you interpret these values?
- PPV is 11.4%
49/429 x 100
- NPV is 99.8%
570/571 x 100
3.The positive predictive value in this case is very low – only 11% - i.e. only an 11% chance that this positive test result means the patient will actually have a DVT.
However, the negative predictive value is 99.8% This means that if a d-dimer is negative in this patient, there is almost ‘no chance’ he has a DVT.
What we can infer from these results is that a negative d-dimer blood test is very good at ruling out DVT, when the prevalence of DVT is low.
However, a positive d-dimer blood test is not very good at ruling in a DVT when prevalence is low. This is partly because there are a range of other conditions which can cause this to be positive – for example, other blood clots in the body, infections, inflammatory diseases…the list goes on.
2) What is the negative predictive value of d-dimer in this population?
3) What is the positive predictive value?
4) How have these changed with increasing prevalence of DVT in this population?
5) How has this increasing prevalence changed your clinical interpretation of the same test result?
6) Would d-dimer be a good screening test for DVT?
2) NPV is 97.1%
3) PPV is 68.5%
4) With increasing prevalence, the PPV has increased and the NPV has decreased
5)
- In this example, the Well’s score is higher, and therefore your prevalence (or your pre-test probability) is higher
- A higher PPV means that it is more likely that a positive d-dimer test represents an underlying DVT
- Therefore as a clinician you are more likely to ‘believe’ this positive test.
- Equally, as the prevalence of DVT in patients with a Well’s score of 5 is 47%, you are less likely to ‘believe’ a negative test when compared to the previous example.
- This is reflected by the decrease in the NPV (though this still remains high!)
6) No – screening tests are done in asymptomatic well people. In well people, prevalence of DVT is extremely low (<1%). Therefore, your positive predictive value would be very low which would mean the test would be very difficult to interpret.
* In clinical practice, d-dimer is a test used on symptomatic individuals to rule out venous thromboembolism. A positive d-dimer result needs further diagnostic testing (ultrasound of deep veins and/or CTPA) with a higher specificity, in order to confirm presence/absence of venous thromboembolism.*
By now you should understand the following implications of prevalence on screening:
- If a condition is very rare, even with a very sensitive and specific test, the test will have low PPV
- In practice in screening this often means that target populations are used – e.g. AAA has negligible prevalence in younger patients, but >65 years it is estimated to be about 1-2%. Therefore only this population are screened.
Bias in screening
Watch the following video on self-selection bias, lead time bias and length time bias. It is important that you understand these concepts.
Please be aware that this video uses the term “resilience” to describe populations. Resilience is a public health term that refers to the general capacity of that population to adapt and thrive in the face of adversity. Populations that already face more adversity (higher levels of disease and disability) than others have less reserve to adapt and thrive when faced with further adversity.
Self-selection bias-where healthier, more health conscious people come forward and take part in screening. So we can incorrectly assume our screening has been beneficial in keeping people healthier, when actually this population is likely to be healthier than the unscreened population anyway as they are more health conscious people, nothing to do with the screening test.
Measures to improve participation:
- leaflets in multiple languages
- outreach work using influencers
- Mobile screening units to geographically isolated communities
2 other types of bias that can lead us to overestimate the benefits of screening programmes: lead time bias and length time bias
- Lead time bias
Let’s say i am destined to die in 5yrs time. In scenario A I get symptoms in 4 years time and I die a year later. In scenario B I’m screened tomorrow and diagnosed next week and i die about 5 years after that. In both circumstances, I die in 5 years time. Only in scenario B i was diagnosed earlier having undergone screening.
If my measurement of screening programme effectiveness was survival, then my survival in scenario A was one year and in scenario B it was 4 years. So it looks as is screening programme caused additional 3 year survival so screening is brilliant. No actually it wasn’t as still died in 5 years.
- length time bias
Let’s take 12 patients with cancer onsets at different stages over 5 years. Half of them have high grade or quick cancer, and the other half have low grade or slow cancer and we screen them. Patients with a slower cancer will have longer periods of time in the earlier stages of disease. They are therefore likely to be detected by screening. So the screening programme has detected 5 of the 6 with low grade cancers, but only 3 of the 6 with high grade cancers. Now we return to survival as the measure used to evaluate our screening programme, low grade cancers kill more slowly, therefore because our screened cancer patients are more likely to have a low grade cancer, the survival of the group on average will be higher than we’d observe in an unscreened cohort. So once again, we incorrectly surmise that our screening programme is fantastic, but all we’ve done is identify low grade cancers and infer that the survival advantage is due to them being screened.
Section 4 of 5: NHS Screening programmes (20 minutes)
Screening programmes (10 minutes)
Almost all of you will have been part of a screening programme. It is important you are aware of the NHS screening programmes so that you can appropriately signpost patients to these in your clinical practice.
In Population Health we are not expecting you to memorise the exact schedule – but we do want you to have a broad awareness of the screening programmes offered throughout the life-course. Be aware that the NHS screening programmes are not static, there will likely be differences by the time you qualify. The UK National Screening Committee (NSC) constantly review current programmes and any potential new programmes.
Reading: Take a look at page 1 of this document which summarises the population screening timeline in the NHS.
Videos : Watch the following videos which summarise the male, female and pregnancy pathways:
Remember - Screening is a choice
As we have already covered step 2 of the screening pathway is:
- Discuss and offer screening
Taking part in screening must be an empowered, personal choice. Can you think of any reasons why someone may choose not to take part in a screening?
For example consider screening for Abdominal Aortic Aneurysm – this is an ultrasound scan offered males at aged 65. If their aorta shows a large aneurysm they are offered surgical management to prevent rupture. However, surgery to stent an aortic aneursym is not without risk. Someone who would not want surgery, or knew they were not an appropriate candidate for surgery may choose to forgo screening. Knowing that they have a large aneurysm that may eventually kill them may cause psychological harm if they are not going to treat it.
Another example is screening in pregnancy. Although all pregnant people are offered screening for some genetic conditions, some choose not to have this done. Reasons for this could be that it would not affect their decision to continue the pregnancy or they may rather not know this information pre-emptively.
It is important not to cast judgement on people’s decisions to engage or not engage in the screening process, everyone is entitled to make the right decision for them. It is often your role as a clinician to help guide them through this decision making process.
PPV and NPV change on the population your testing!
As increase prevalence, PPV increases, NPV decreases
7 main steps in screening programme:
As increase prevalence, PPV increases, NPV decreases
Work out the highlighted box:
Sensitivity=TP/(TP+FN)
Which is the same as sensitivity=TP/total number with disease
so therefore TP=sensitivity x total number with disease
=88.9% x 1
=1 rounded
