Lung cancer screening

I’ve written a little on this already

I know a thing or two about screening. It’s not like other healthcare, there’s an ethical issue of inviting people who think they are well……for a test that may have good and bad downstream consequences. It’s ethically not the same as offering someone with symptoms a test.

I also have personal experience. My old man died of lung cancer. It’s horrible.

My views about lung cancer (any) screening are well known.

If NSC haven’t recommended it, don’t screen. If they have only screening within programme parameters and with excellent attention to the QA process. The end. See screening 101 lesson plan. I’d encourage particular attention to ethics class. Screening is indeed special. See hereOne of the key ethical principles (justice) means “Screening programmes should be used only when all other primary preventive measures are in place (because primary prevention is likely to be more cost-effective than screening)”

The Manchester and Liverpool lung cancer screening programme has been well documented and publicised.

The data on initial roll out in Manchester is published, it’s an excellent and insightful paper.

Crosbie et al

summary of the numbers

  • Letters to 16.5k
  • Risk assessed 2541
  • Scanned (CT) 1.4k
  • 81 assessed in lung cancer clinic
  • Of the 81, 42 were confirmed to have lung cancer.
  • false positive rate was 48.1% (n=39/81)…… let that sink in 48%.

Lives will be saved, misery averted. This is good. Isn’t it? How many lives. Well I’m not sure.

Harm is appreciable

Accounting for the Harms of Lung Cancer Screening

Overdiagnosis is an often underappreciated harm of screening.

Causes and consequences of overdiagnosis set out

Estimation of Overdiagnosis of Lung Cancer in Low-Dose Computed Tomography Screening

NLST estimated that 18.5% of the cancers detected by CT are overdiagnosed

This study aimed to estimate overdiagnosis the Danish Trial

The are clear.

Benefits and harms of lung cancer screening

1 in 217 were helped (prevented death)

1 in 4 were harmed (false positive CT scan)

1 in 30 were harmed (unnecessary surgery)

1 in 161 were harmed (surgical complication)

Or flipped-

As a colleague of mine pointed out, if this were a drug trial, it would have been terminated early.

What’s the impact in a given population

Using the Canadian task force reference data & a simple cross reference with the with the Manchester data…..Letters to 16.5k Risk assessed 2541 Scanned (CT) 1.4k

The initial pilot was a practice population of 14 practices, I don’t know the population.

But using the Canadian taskforce information, especially the 1000 person tool

So if 1000 people are screened with LDCT

·         231 more people receive a positive test result

·         4–5 fewer late-stage lung cancers are found

·         8–9 more early-stage lung cancers are found

·         3 more people are diagnosed with lung cancer

·         3 fewer people die from lung cancer

(It’s worth saying this data a favourable estimate as it was derived from the NLST study and this didn’t track downstream complications)

So let’s say you have a pop if similar size to the GM pilot. I can’t see the population specified in the Crosbie paper but was 14 practices.

Current or ex smokers aged 55–74 years invited – 16.5k sent a letter – led to 1.4k scans

So let’s assume for the sake of a bit of fag packet (sic) epidemiology – 4 fewer die of lung cancer, what’s the impact of the lung cancer mortality rate…..

As I don’t know baseline pop, sort of hard to make comparative estimates, but I picked on

Calderdale CCG has about 20 practices

There are 125 or so lung cancer deaths there in a year

125 – 120 or so…….

4 less is not insignificant (esp if you’re one of them) but….. lung cancer screening isn’t going to have giant impact on lung cancer mortality……..

Cost effective?

I don’t know off hand the economics of lung cancer screening, on the basis of the above it’s seems quite difficult to prove a case that it passes muster value for money wise. I would encourage people to take a very close look at this study when published. This was the study commissioned by NSC. It’s a cost effectiveness  analysis of lung cancer screening in high risk groups using low dose CT using the data from the four trials.

My guess, having kept a close eye on the primary studies as they are published, is that hazard ratio for overall mortality crosses one, ie screening DOESN’T “save lives” in this case and I’d bet a weeks pay screening for Lung cancer the available evidence to date suggests unlikely cost effective – close to, at or over the threshold, and definitely somewhere at top or over the top of the diminishing marginal return chart and the jury is definitely out re harm / benefit ratio.

We await NELSON economic analysis.

NSC consultation may follow this, but it’s seems hard to justify any lung cancer screening outside formal research.

But……”curative intent” treatment was given to 41. it saves lives, so we must roll it out

the curative intent in 41 is significant, the staging data is definitely worth thinking hard about.

Curative intent doesn’t equate to cure. I don’t know the 5yr survival of those with lung cancer treated with curative intent, compared to later stage treatment.

Seems a lot better than standard routine clinical care. Obviously excellent news for those people. Maybe we should check actual outcomes before we roll out?

Oh…….I’m told there’s a £4m roll out across the whole of GM. 400k population and 105 practices.

So sticking with my fag packet, that’s going from 14 to 105 practices or c8 times increase compared to the pilot (assuming practices similar size).

So let’s assume 1,400*8 = 11,200 CT scans.

For the 11k scans, one might (if the Canadian Taskforce is to be believed reduce lung cancer deaths by 11*3 = 33 people.

There are about 310 lung cancer deaths in Manchester (city council boundary) a year.

33 less is not trivial, but at what cost, and what opportunity cost.

In the round the evidence doesn’t support it that well. Cochrane review here. I would encourage people to think quite critically about this. I know it’s difficult, but……

  • is is effective. Cochrane review here ( admittedly a high bar)
  • Is the proposition to screen for lung cancer cost effective. Evidence to date suggests not.
  • does it satisfy the Wilson Jungner criteria.
  • What is the NSC position. We should not be screening for anything outside research till NSC publish their position and QA criteria published (I’d encourage you to think again about the false positive rate, higher than in NLST)
  • There are higher value investments. There are better value opportunities to reduce lung cancer mortality. See for example here and here, especially table 2 for estimates of Numbers Needed to Treat to prevent death for commonly used interventions. What indeed is the opportunity cost of all that scarce radiologist time. What things won’t they be doing as they are doing this.
  • we should factor in the Harm in the equation. If you believe the Canadian stuff CT screening 11k people will lead to appreciable harm – see the infographic.
  • And the Opportunity cost. £4m crowds out potentially more valuable investments…..children’s mental health care, or social care. £4m on something that doesn’t not seem objectively cost effective (lung cancer screening), and £0 new resource (tell me if I’m wrong) on smoking cessation is basically the headline figure for allocative efficiency for the people of Manchester.

Others can improve the maths no doubt. I’m just doing simple sums on an iPad.

“Screening saves lives” vs “screening (often only marginally) changes cause specific death rates

One of these statements you often see in the Daily Mail. One you don’t………..

Even if you believe it saves lives and thus is a good thing to do, those who have care denied by role of opportunity cost of investment in non cost effective programmes may not agree with you. You ought to ask those people as well,

In the meantime we are pushing that any project to “do” lung cancer “screening” does so in an explicitly research framework

If you are pressing on regardless, please try to minimise harm. Here are some suggestions


hammer home message re smoking cessation at every opportunity

Far more valuable


Make sure epidemiologically accurate evaluation


Ensure mega focus on QA

Minimise the harm


Maybe seek to tighten the net and ensure higher risk people only sent letter – maximise yield and

Ilkley minimise net harm


Ensure high quality shared decisions in place

The Canadian tool is good

There are others

Ensure people are trained to, and do use it as a method

Changes the decisions people make (see Cochrane review)

of note we are poor at implementing shared decisions. This series on lung cancer screening and shared decision making is excellent

Failing Grade for Shared Decision Making for Lung Cancer Screening

Sadly now paywall

Worth trying to get


No punches pulled

Evaluating Shared Decision Making for Lung Cancer Screening

See me after class

“Despite recommendations, shared decision making for lung cancer screening in practice may be far from what is intended by guidelines”


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s