NHS Choices

Ibuprofen 'barely better than placebo' at treating back pain

NHS Choices - Behind the Headlines -

"Widely used anti-inflammatory drugs such as ibuprofen have little more benefit than a placebo when it comes to treating back pain," reports the Guardian.

This is based on a study looking at more than 6,000 people with back pain, comparing non-steroidal anti-inflammatory drugs (NSAIDs) with a placebo ("dummy" medicine).

While NSAIDs were found to reduce pain and make moving and doing daily activities easier, the difference compared to a placebo was not large enough for the researchers to consider it important. Also, people taking NSAIDs were at greater risk of gastrointestinal problems compared with those taking a placebo.

This was a good piece of research that looked at a number of high-quality studies to reach the conclusion that, generally-speaking, NSAIDs aren't that effective for back pain.

However, this research doesn't mean NSAIDs don't work at all for back pain and shouldn't be used. It's possible that some people will still benefit from them, with the study suggesting that around one in six people taking NSAIDs – rather than placebo – experience a significant reduction in pain.

Back pain usually gets better by itself after a few weeks but it might be a good idea to seek help if your pain is continuing for longer than this, is getting worse, or is stopping you doing your daily activities. Discuss the treatment options with your doctor.

NSAIDs are currently recommended by the National Institute for Health and Care Excellence (NICE) as a treatment option for lower back pain, alongside other approaches such as staying active, group exercise classes and manual therapy, such as chiropractic.

Where did the story come from?

The study was carried out by researchers from the University of Sydney in Australia and was funded by the Department of Education and Training of Australia, the National Health and Medical Research Council Australia and Sydney Medical School. The authors declare no conflict of interest.

The study was published in the peer-reviewed journal, Annals of the Rheumatic Diseases. The abstract is available to read for free.

The media reporting of the story was rather exaggerated and misleading, with the Mail Online claiming "ibuprofen doesn't work for back pain", when in fact the study found the NSAIDs are effective in reducing pain, just that the amount of benefit people feel is not thought to be a clinically important reduction compared to a placebo.

The Mail also claimed that "adults taking the cheap pills are actually three times more likely to suffer from stomach ulcers". In fact the study found NSAIDs increased the likelihood of any gastrointestinal problem, not necessarily ulcers, by 2.5 times.

What kind of research was this?

This was a systematic literature review and meta-analysis of 35 randomised controlled trials (RCTs) investigating the effects and safety of NSAIDs for spinal pain, compared with a placebo.

Current UK guidance indicates that NSAIDs such as ibuprofen and high-dose aspirin could be considered for managing back pain, whereas paracetamol alone should not be used.

RCTs are one of the best ways of looking at the effect of a treatment, in this case the effect of NSAIDs for treating back pain.

But while a systematic literature review is useful in bringing together evidence on a specific topic, it can only be as good as the studies included. Any shortfalls of the included studies will be brought forward into the review.

What did the research involve?

Researchers carried out a systematic review of RCTs comparing effectiveness and safety of NSAIDs with a placebo.

The review included 35 RCTs, involving 6,065 participants with acute or chronic neck or lower back pain. Any class, formulation or route of administration (topical, oral or injection) of NSAIDs was included as well as any dose and frequency of NSAIDs intake.

A follow-up period of less than two weeks was defined as immediate-term and a follow up of between two weeks and three months as short-term.

Pain outcome measures reported in the trials were either visual analogue scales or numerical rating scales. These were converted to a common scale that ranged from 0-100, with 0 meaning no pain or disability and 100 meaning the worst possible pain or disability.

A difference of 10 points on the 0-100 scale between placebo and drug was considered the smallest worthwhile effect that patients perceive as important. A 10-point difference was therefore the minimum needed to be considered "clinically important".

The number needed to treat (NTT) – the number of patients who need to be treated with an NSAID rather than a placebo for one additional person to benefit – was also calculated to give a more understandable measure of the benefit.

What were the basic results?

NSAIDs were mostly administered orally, but some trials injected the drugs or used a gel, patch or cream. When considering pain:

  • NSAIDs reduced pain and disability compared with the placebo in the immediate term (mean difference (MD) -9.2, 95% confidence interval [CI] -11.1 to -7.3).
  • NSAIDs reduced pain and disability compared with the placebo in the short-term (MD -7.7, 95% CI -11.4 to -4.1).

But neither of these results were clinically important according to the researchers as the difference between NSAIDs and placebo was less than their predefined threshold of 10 points on the 0-100 scale.

Taking into account what healthy people would consider an important reduction in pain, they calculated that six people (95% CI 4 to 10) would needed to be treated for two weeks with NSAIDs rather than placebo for one additional person to achieve clinically important pain reduction in the short-term.

When considering safety, a higher number of participants taking NSAIDs had gastrointestinal adverse reactions compared with placebo (relative risk [RR] 2.5, 95% CI 1.2 to 5.2). There was no difference between the NSAIDs group and placebo group in serious adverse events.

How did the researchers interpret the results?

The researchers conclude: "NSAIDs do not provide a clinically important effect on spinal pain, and six patients must be treated with NSAIDs for one patient to achieve a clinically important benefit in the short-term."

They add that "when this result is taken together with those from recent reviews on paracetamol and opioids, it is now clear that the three most widely used, and guideline-recommended medicines for spinal pain do not provide clinically important effects over placebo. There is an urgent need to develop new analgesics for spinal pain."


There was evidence that NSAIDs were effective in reducing pain and disability in patients with spinal pain, but treatment does not seem much more effective than a placebo and is not clinically important according to the researchers.

Moreover, for every six patients treated with NSAIDs rather than a placebo, only one additional patient would benefit in the short-term. People taking NSAIDs also have a higher risk of gastrointestinal adverse reactions. Patients might like to consider if this seems like a chance worth taking.

NSAIDs are currently recommended to treat back pain, but the authors suggest new, more effective drugs should be urgently developed.

The study does have some limitations:

  • The mode of treatment varied from oral intake to applying gel or cream. Some patients may feel better with a direct application compared to an oral drug, but it is difficult to say which is more effective as these were grouped together.
  • The dose also varied between studies and therefore it is difficult to know if NSAIDs were more effective at a higher dose.
  • The treatment period was on average only seven days and therefore it is hard to tell what long-term outcomes would have been if participants had continued to take NSAIDs.
  • The research focused on whether NSAIDs were effective for back pain as a whole, so it's difficult to know whether particular individuals or specific groups of patients might benefit from the treatment more than others.

This study was not set up to compare NSAIDs with other non-pharmacological treatments (such as exercise), some of which may be more effective than NSAIDs.

Links To The Headlines

Ibuprofen for people with back pain has no significant effect – new study. The Daily Telegraph, February 2 2017

Ibuprofen doesn't work for back pain: Only 1 in 6 feel any benefit from taking the drug (and exercise is the only effective treatment). Mail Online, February 2 2017

No common painkillers are effective for back pain – they actually cause nasty side effects, experts warn. The Sun, February 2 2017

Forget pills... Exercise is best way to fight misery of back pain. Daily Express, February 3 2017

Ibuprofen has little benefit in treating back pain and may cause harm – study. The Guardian, February 3 2017

Links To Science

Machado GC, Maher CG, Ferreira PH, et al. Non-steroidal anti-inflammatory drugs for spinal pain: a systematic review and meta-analysis. Annals of the Rheumatic Diseases. Published online February 2 2017

Poor sleep may affect good sex in later life

NHS Choices - Behind the Headlines -

"Good night's sleep boosts sex life for women over 50," reports the Mail Online.

US researchers asked more than 93,000 women aged 50 to 79 about their sleep patterns, difficulty sleeping, sexual activity and sexual satisfaction. They found women who sleep five or less hours a night, or who have insomnia, were less likely to have satisfying sex lives.

A higher proportion of women who reported they were satisfied were married or in an intimate relationship. However, women living without a partner who slept less than seven to eight hours were more likely to be sexually active but less likely to be sexually satisfied.

The research, part of an ongoing study of menopausal women in the US, took account of factors that could affect both sleep and sex, such as health problems, menopausal symptoms, age, and medicines use, including HRT. However, it only asked the questions at one point in time, so we don't know whether sleep problems happened before or after any sexual problems. This type of research can't tell us whether sleep is a cause of sexual problems.

Sleep is important for much of our wellbeing, however, including mental and physical health. It would not be surprising if lack of sleep also affected women's sex life.

Where did the story come from?

The study was carried out by researchers from a number of US institutions: the Mayo Clinic, Harvard Medical School, Ohio State University, Georgetown University, Wake Forest School of Medicine, Stony Brook University, Veteran Affairs Palo Alto Health Care System, University of Texas and the University of California. It was funded by the US National Institutes of Health. The study was published in the peer-reviewed journal Menopause.

The Mail Online gave a reasonable overview of the study, although it did suggest that lack of sleep had been established as a cause of poor sexual satisfaction, rather than simply linked to it.

The Daily Telegraph's report approached the research from the perspective of a woman's "frustrated lover", advising readers to "Listen to your partner when she says she's too tired for sex" and saying that the research shows tiredness may not just be a "thin alibi… to avoid amorous relations". Its coverage suggests that their readers would otherwise ignore women's protests that they didn't feel like sex, which one hopes is not the case.

What kind of research was this?

This is a big cross-sectional observational study. Cross-sectional studies can show how people feel at one point in time, and make links between factors (in this case sleep and sex). However, they cannot show that one factor causes another.

What did the research involve?

Researchers analysed information given by 93,668 women aged 50 to 79, who took part in the Women's Health Initiative Observational Study, carried out from 1994 to 1998. After adjusting their figures to take account of potential confounding factors such as illness and medicines, they looked to see if there was a link between how women said they slept, and their sexual activity and satisfaction.

Most of the measures included in the study were self-reported. To measure sleep over the past four weeks, women were asked:

  • how many hours they slept at night
  • whether they had one of a number of factors suggesting insomnia (trouble getting to sleep, waking repeatedly, trouble getting back to sleep, waking too early, unrefreshing sleep)
  • whether they snored or fell asleep easily during quiet times in the day

To measure sexual function, they were asked:

  • whether they'd had sexual activity with a partner during the past year
  • how satisfied they were with their current sexual activity

Many women didn't answer the sexual questions (34% for sexual activity and 43% for sexual satisfaction), which could affect the reliability of the results.

The researchers took account of a wide range of possible confounding factors, including women's age, marital status, income, physical activity level, overall health, use of antidepressants, use of HRT, depression, weight and alcohol use.

What were the basic results?

Just over half (52%) of women who answered the question said they'd had sexual activity with a partner over the past year, and 57% said they were very or somewhat satisfied with their current sexual activities. About one third (31%) of women said they had symptoms of insomnia.

Women who slept for five or less hours a night, or who had insomnia, were less likely to feel satisfied with their sex lives than women who slept seven to eight hours and did not have insomnia:

  • Women who slept five or less hours were 12% less likely to feel satisfied than women who slept seven to eight hours (odds ratio [OR] 0.88, 95% confidence interval [CI] 0.81 to 0.95).
  • Women with insomnia were 8% less likely to feel satisfied than women without insomnia (OR 0.92, 95% CI 0.87 to 0.96).

Less sleep was linked to a 12% lower chance of having had sexual activity with a partner during the past year (OR 0.88, 95% CI 0.80 to 0.96). However, insomnia symptoms alone did not seem related to chances of having sex with a partner.

How did the researchers interpret the results?

The researchers said their results "suggest the potential importance of obtaining high-quality and sufficient sleep," for good sexual function.

They say future studies into sleep and sex in women after the menopause should be carried out over time, so that the changing relationship between sex and sleep can be clarified. 


These results show that women who sleep better are more satisfied with their sex lives, and more likely to be sexually active with a partner. However, the study can't tell us why this is. So many factors have the potential to affect both sleep and sexual satisfaction, that it's always going to be difficult to un-tangle the relationship between the two.

There are a few limitations to the study that make the results less reliable. Although it was a big study, a large proportion of the women chose not to answer the questions about sex. The questionnaire included the option to tick "prefer not to say". This means the results may not be representative of all the women in the study.

It's also important to note that the questions were asked only once, so we don't know how the relationship between sex and sleep changed over time. For example, it could be that some women's sexual satisfaction declined after they'd started having trouble sleeping, or that other women's satisfaction increased when their insomnia improved.

Conversely, women may have started having trouble with both sleep and sex after a life event such as bereavement, or because of a physical illness. A cross-sectional study can't help us unpick these possibilities. The study didn't ask about life events such as bereavement or divorce, although it did ask about whether women had a current sexual partner.

These caveats aside, it's well-known that sufficient sleep is important for health and general wellbeing. It would not be surprising if this extended to sexual wellbeing and satisfaction.

Find out more about how to get a good night's sleep.


Links To The Headlines

Listen to your partner when she says 'I'm too tired tonight': new research shows good sleep leads to better sex. The Daily Telegraph, February 1 2017

Good night’s sleep boosts sex life for women over 50: Insomnia symptoms are linked  to decreased libido in post-menopause.  Mail Online, February 1 2017

BEST OF BOTH WORLDS Want to boost your sex life? Spend more time in bed… SLEEPING, scientists say so!  The Sun, February 1 2017

Links To Science

Kling JM, Manson JE, Naughton M, et al. Association of sleep disturbance and sexual function in postmenopausal women Menopause. Published online February 1 2017

'Breakthrough in communication for patients with severe MND', study claims

NHS Choices - Behind the Headlines -

"Mind-reading machine allows people with 'locked-in' syndrome to communicate," reports the Mail Online.

The report is based on a study that aimed to communicate with four patients unable to speak, move or blink due to a severe form of motor neurone disease (MND).

The patients were able to give "yes" or "no" answers to a series of questions via a computer, which interpreted their brain signals.

They were given statements such as "your husband's name is Joachim" or "Berlin is the capital of France" and told to think either "yes" or "no" in response.

They wore head caps fitted with sensors which measured changes to blood oxygen levels in the brain to work out if their answer was a "yes" or a "no".

Towards the end of the study, researchers asked open questions such as whether patients were in pain, and whether they felt positively about their quality of life. In line with previous studies of people who knew they would become completely paralysed and chose to be on ventilators, they said that they did feel positive.

Researchers say the system correctly relayed what the patients were thinking 70% of the time.

The patients, aged between 24 and 76, all had amyotrophic lateral sclerosis (ALS), the most common type of MND.

The average life expectancy of someone with ALS is from two to five years after symptoms first appear.

The patients were at different stages of Completely Locked-in State (CLIS), a condition where the patient can think and has emotions but is completely paralysed.

They had lost all eye movement and the ability to communicate with their families – some for several years. They were receiving round-the-clock care at home, with artificial breathing and feeding tubes.

This small experiment raises the possibility of meaningful communication for people with this type of condition.

However, it is a tiny study and the findings may not be applicable to people with other causes of CLIS, such as stroke.

Where did the story come from?

The study was carried out by researchers from the University of Tübingen and Central Institute of Mental Health in Germany, Shanghai Maritime University in China and the National Institute of Neurological Disorders and Stroke in the US.

It was funded by several organisations including the Deutsche Forschungsgemeinschaft, German Ministry of Education and Research, Eva and Horst Köhler-Stiftung, National Natural Science Foundation of China and an EU grant.

The study was published in the peer-reviewed journal PLOS Biology on an open-access basis and is free to read online.

The UK media gave broadly accurate coverage to the study. The Daily Telegraph and Mail Online both talked about the computer being able to "read people's thoughts" or being a "mind-reading machine", which is over-stating the reality.

At present the computer is only programmed to record brain responses to questions with yes/no answers, and it is not entirely accurate.

What kind of research was this?

This was an experimental study on a small number of people, with no control group. As such, it provides useful evidence in support of a theory that this type of technology can be used to communicate with people with locked-in syndrome, but the results need to be replicated to be sure they are reliable.

What did the research involve?

Four people with completely locked-in syndrome (meaning they are unable to move even their eye muscles, and are dependent on artificial breathing and feeding) were recruited to the study.

Researchers fitted them with caps which measured electrical activity and oxygenation. They were trained to answer "yes" or "no" to a series of known questions – questions the patient would find easy to answer.

A computer programme analysed changes to their brains during the sessions, and learned which responses typified a correct positive or negative response.

The people in the study had ALS, a motor neurone disease which progressively shuts down the body's ability to move muscles, even for automatic movements such as breathing or swallowing.

All the patients had moved past the stage at which they could communicate through blinking or eye movement.

Their families had completely lost the ability to communicate with them – one since 2010, two since August 2014 and the family of the youngest since January 2015. They were being cared for at home, with artificial breathing and feeding tubes.

The technology used to measure changes to brain oxygenation is called functional near-infrared spectroscopy (fNIRS).

The researchers also measured electroencephalogram (EEG) changes in the brain and activity in eye muscles, to see whether these could predict correct answers. The EEG results were also used to tell whether people were sleeping, or to identify times when their brains were inactive and less responsive to questions.

The main part of the study looked to see how often the computer could read an accurate "yes" or "no" response to a known question, in up to 46 sessions spread over several weeks.

They were asked 20 questions each session, with an equal mix of true and false statements being presented in the same format (for example, "Paris is the capital of France" and "Paris is the capital of Germany").

In some sessions, people were asked open questions, such as whether they were in pain. Only three people were asked open questions in the study.

Researchers worried that the youngest (age 23), whose disease had progressed very fast over two years, might be emotionally unable to give reliable responses to open questions. Her brain pattern responses to yes and no were less distinct from each other than the other patients.

What were the basic results?

The correct response rate of the four people in the study for questions with known answers was over 70%, averaged across the several weeks of the study. This is higher than the level you would expect from chance alone.

Three people answered open questions and were given feedback about their perceived answers. The "correct" rate was estimated at 78.6%, 78.8% and 75.8% for these three people.

The researchers judged that they could be sufficiently certain of the answer if people gave the same answer to an open question seven out of 10 times, when questions were repeated over several weeks.

These patients answered open questions containing quality of life estimation repeatedly with "yes" response, according to the researchers. They say this indicated a positive attitude to their situation and to life in general.

How did the researchers interpret the results?

The researchers say their results are "potentially the first step towards abolition of completely locked-in states, at least for patients with ALS".

They say the results need to be confirmed in other studies over a longer period of time, because of the importance of getting them right.

They also acknowledge that they cannot explain why blood oxygen levels in the brain were different when the response was "yes" compared with "no". They added that any theories would be "highly speculative".


It's hard to imagine the situation of being alert, aware of what's happening around you, but unable to move, respond or communicate with the outside world.

So it is comforting, then, to hear that people with complete locked-in syndrome may be able to communicate – and may be relatively content with their situation.

However, it's important to remember the limitations of this study.

It's very small. Only four people took part, and full results are available for only three of them.

The results may only apply to people with this very specific type of neurodegenerative disease, not to people with other types of paralysis or locked-in syndromes like those caused by a stroke or head injury.

People in the study were all being given intense nursing care in their homes, looked after by family members. They had all chosen to have artificial respiration – in other words, had chosen to live with locked-in syndrome rather than to allow nature to take its course. That might affect how they answer questions about quality of life.

It's hard to know how accurate the results of the study are. We can't directly test them, so we have to rely on likelihood and chance of people repeatedly giving the same answers, and the computer reading the patterns correctly.

As the authors note, we do not know why the oxygenation results would be different for "yes" and "no" answers. There was also no clear pattern in the responses between patients, which would be expected if there was truly a physiological reason for the results.

Links To The Headlines

Completely 'locked-in' patients can communicate. BBC News, February 1 2017.

Ground breaking system allows locked-in syndrome patients to communicate. The Guardian, January 31 2017.

Locked-in patients tell doctors they are 'happy' after computer reads thoughts. The Daily Telegraph, January 31 2017.

Mind-reading machine allows people with 'locked-in' syndrome to communicate - and all four patients said it has made them HAPPY. Mail Online, January 31 2017.

Links To Science

Chaudhary U, Bin X, Silvoni S, et al. Brain-Computer Interface-Based Communication in the Completely Locked-In State. PLOS Biology. Published online January 31 2017

Diabetes could be a warning sign of pancreatic cancer

NHS Choices - Behind the Headlines -

"Experts have revealed the onset of diabetes, or existing diabetes getting much worse could be a sign of hidden pancreatic cancer," reports The Daily Express.

The media reports follow a press release of a study presented at the European Cancer Congress (ECCO) yesterday. The research analysed nearly a million people with type 2 diabetes in Belgium and Italy, some of whom went on to be diagnosed with pancreatic cancer.

The recent onset of diabetes appeared to be a possible warning sign of pancreatic cancer, with 25% of cases in Belgium and 18% in Italy being diagnosed within three months of a diabetes diagnosis. Faster progression of diabetes (where patients needed insulin or other more intensive treatments sooner) was also associated with a greater chance of being diagnosed with pancreatic cancer.

Pancreatic cancer is rare and often has a poor outcome, partly because it is difficult to detect at an early stage.

However, it's important to put these findings in context. Diabetes has previously been linked with pancreatic cancer, though it is unclear why. It could be that diabetes increases the risk of pancreatic cancer. What is probably more likely is that rapid onset or progression of diabetes could be a symptom of the cancer itself.

Diabetes is fairly common in the UK, with around 4 million cases, while pancreatic cancer remains very rare. Just because you have diabetes does not mean you will go on to get pancreatic cancer.

However, if you are concerned that you may have diabetes or that your diabetes is poorly controlled, you should talk to your GP.

There are also steps you can take to reduce your risk of developing diabetes.

Where did the story come from?

The study was carried out by researchers from the International Prevention Research Institute in Lyon, France. The study has not yet been published in a journal but was presented at the European Cancer Congress held in Amsterdam. The findings come from the press release.

Funding was provided by Sanofi, a French pharmaceutical company. The authors declare the sponsor had no influence on the study design, conduct, analysis and reporting.

This has been reported widely in the UK media, though not always accurately. The Mail Online claims the researchers "analysed nearly a million type 2 diabetics in Italy and Belgium who had been told they had pancreatic cancer" however this was the number of people in the database with diabetes. Only 2,757 people had been diagnosed with pancreatic cancer.

Moreover, The Daily Telegraph reports "50 per cent of patients diagnosed with pancreatic cancer had been diagnosed with type 2 diabetes the previous year". This is rather misleading and suggests 50% of all people with pancreatic cancer also have diabetes.

But this study only looked at people with diabetes. Of those who developed pancreatic cancer, half had received their diabetes diagnosis in the past year. The overall proportion of all people with pancreatic cancer who also have diabetes in the population is unknown.

What kind of research was this?

This was a retrospective cohort study looking at the association between type 2 diabetes and the diagnosis of pancreatic cancer.

The study is currently only available as a published abstract and was presented at the European Cancer Congress with an accompanying press release. A full study publication is not available so we can't fully critique the methods and analysis.

Pancreatic cancer has a notoriously poor prognosis as it is often hard to diagnose at an early stage due to a lack of symptoms or non-specific symptoms. Individual outcomes vary, but generally only 1% of all people diagnosed with pancreatic cancer live for more than 10 years after their diagnosis.

Diabetes has already been linked as a possible risk factor for pancreatic cancer, but in what context is uncertain. However, onset of diabetes or rapid deterioration of current diabetes could be a possible marker for early pancreatic cancer so could potentially aid earlier diagnosis.

What did the research involve?

The researchers used a prescription database (the Inter Mutualist Agency AIM-IMA) to identify 368,377 people receiving treatment for type 2 diabetes in Belgium between 2008 and 2013. They also identified 456,311 being treated in Lombardy, Italy, between 2008 and 2012.

These data were linked to pancreatic cancer data from the Belgium Cancer Registry and hospital discharge databases in Lombardy.

The rates of pancreatic cancer were analysed in association with time of first prescription of diabetes drugs, and use of different diabetes treatments.

What were the basic results?

In Belgium, 885 of 368,377 people with diabetes had pancreatic cancer. In Lombardy, 1,872 of 456,311 people with diabetes had pancreatic cancer.

Among all those with pancreatic cancer in the two regions, 50% had been diagnosed within one year of being diagnosed with type 2 diabetes.

In Belgium, 25% of pancreatic cancer cases were diagnosed within 90 days and in Lombardy 18% were diagnosed within 90 days.

When considering treatment, the researchers generally found that switching to more intensive diabetes treatments was also linked with a greater risk of pancreatic cancer diagnosis:

  • People who switched from oral diabetes drugs to more intensive treatment of incretin-based therapy (injected drugs that help the body produce more insulin) had 3.3 times the risk (95% confidence interval [CI] 2.0 to 5.5) of cancer diagnosis in the following three months.
  • This decreased to around a two-fold risk for 3 to 6 months after the first prescription of incretin drugs (hazard ratio [HR] 2.3, 95% CI 1.2 to 4.7) and again for 6 to 12 months after the first prescription (HR 2.1, 95% CI 1.2 to 3.9).
  • Switch from oral diabetes drugs or incretin to insulin injections was also linked with increased risk of pancreatic cancer (HR 11.9, 95% CI 10.4 to 13.6).
  • When comparing those who developed pancreatic cancer with those who remained cancer-free, switching from oral diabetes drugs to incretin or insulin injections happened sooner after diabetes diagnosis in those who developed cancer: median 372 days to switch to incretins and 315 days to switch to insulin in those who developed cancer versus median 594 days to switch to incretins and 437 days to switch to insulin.
How did the researchers interpret the results?

The lead researcher commented: "There is currently no good, non-invasive method for detecting pancreatic cancer that is not yet showing any visible signs or symptoms. We hope that our results will encourage the search for blood markers indicating the presence of pancreatic cancer, which could guide decisions to perform a confirmation examination like endoscopy."


This study uses a large prescription database to investigate the link between diabetes and pancreatic cancer, looking at the timing of first diabetes prescription and change in drugs prescribed.

Among people with type 2 diabetes, diagnosis of pancreatic cancer was linked with recent onset of diabetes or rapidly deteriorating diabetes. This suggests these could both be potential warning signs of hidden pancreatic cancer and indicate the need for more investigations.

While diabetes has previously been linked with pancreatic cancer, the nature of the cause and effect relationship remains unclear. It could be that diabetes increases risk of the cancer, or it could be that recent onset or deterioration of diabetes is a symptom of the cancer.

It had also previously been thought that incretin therapies could promote pancreatic cancer. However, it could be that incretin therapies and insulin therapies are often prescribed sooner in patients who have undiagnosed pancreatic cancer.

As the authors make clear, it is probably pancreatic cancer that causes deterioration of diabetes.

A limitation of this study is that it was carried out in two specific areas in Europe. Sociodemographic variations in diabetes or cancer prevalence, medical care or risk factors may mean the results are not fully applicable to the UK.

The findings are also based on a prescription database, so only look at raw data on numbers. The researchers haven't delved further into the nature of the individual diabetes and cancer diagnoses, investigations and treatment.

These are early findings presented at a conference. A full, published study is not available so it is not possible to analyse the methods and possible implications further.

It's not possible to say whether the findings could lead to more in-depth investigation of people with newly diagnosed or rapidly progressing diabetes, or whether this could make earlier pancreatic cancer diagnosis and improved survival rates possible.

Links To The Headlines

Pancreatic cancer symptoms: Diabetes could be a warning sign for deadly disease. The Daily Express, January 30 2017

Having diabetes is a warning sign of one of the deadliest forms of CANCER, shocking study finds. Mail Online, January 30 2017

Diabetes could be a warning sign of cancer, new study suggests. The Daily Telegraph, January 30 2017

Breath test shows potential for detecting cancer

NHS Choices - Behind the Headlines -

"Breath test could save lives by diagnosing deadly cancers earlier," reports The Daily Telegraph. The story is based on new research into whether it is possible to detect cancers of the stomach and oesophagus (gullet) using a breath test.

A possible "chemical signature" composed of five substances was tested against the breath samples of more than 300 people who previously had an endoscopy to investigate upper digestive tract symptoms.

The researchers found that four of these chemicals were expressed differently in the breath samples from those diagnosed with cancer, compared to those where no cancer had been found.

The breath test was able to correctly indicate cancer in around 80% of patients who had cancer, and similarly able to correctly exclude cancer in around 80% who did not have cancer.

These were early findings from a conference presentation. While they show promise, it is not possible to say from the available information whether the test could have a future role in practice.

In most of these people with cancer who took part in the research, the cancer had spread to the lymph nodes. It is unclear if the breath test would be able to detect less advanced cases.

Both oesophageal and stomach cancer tend to be diagnosed late because in the early stages they either cause no symptoms – in the case of oesophageal cancer – or symptoms that are vague and easy to mistake for other less serious conditions – in the case of stomach cancer.

A breath test sensitive enough to identify a "chemical signature" of cancer and allow earlier diagnosis would be ideal. However, the test is not completely reliable and larger studies are needed to confirm these early findings. 

Where did the story come from?

The study was carried out by researchers from Imperial College London and the Karolinska Institutet in Sweden. Funding was provided by the National Institute for Health Research. The study has not yet been published in a journal but was presented at the European Cancer Congress held in Amsterdam.

This has been reported widely and mostly accurately in the UK media with a number of quotes from the research team.

What kind of research was this?

This was a case-control study that aimed to see whether a breath test could be used to detect stomach and oesophageal cancers (OGC).

The study is currently only available as a published protocol and poster presentation with accompanying press release. A full study publication is not available so we can't fully critique the methods and analysis.

Worldwide, OGC cancers account for around 1.4 million diagnoses a year but diagnosis tends to be late and therefore survival rates are low.

At the moment these cancers can only be diagnosed using endoscopy, which involves a camera attached to a flexible tube being passed down the throat. The procedure can be uncomfortable and is costly to the NHS.

A breath test that is able to identify the "chemical signature" of a cancer could be an ideal way to indicate a cancer diagnosis and help decide whether further invasive investigations are needed. It would hopefully enable more patients to be diagnosed at an earlier stage of the disease.

What did the research involve?

The researchers included two groups of patients, those diagnosed with an OGC and those found to be cancer-free (the control group).

All participants were over 18 years of age and had already had an endoscopy to investigate upper gastrointestinal symptoms.

Only people with non-metastatic cancer (cancer that hadn't spread to other organs) were included in the OGC group. Potential participants were excluded if they had an active infection, known liver failure, and if they were unable to provide informed consent or unable to provide a 500ml breath sample.

Breath samples from both groups were collected in steel breath bags from three hospitals. Before sample collection the participants were instructed to fast for at least six hours and rest in the same area for at least 20 minutes. All breath samples were sent to a central laboratory for analysis.

A previous systematic review carried out by the research group had identified significant differences in the volatile organic compound profiles from exhaled breath of people with OGC cancer.

Based on these findings, the chemicals of interest in the breath samples were:

  • butyric acid
  • pentanoic acid
  • hexanoic acid
  • butanal
  • decanal

These five substances were considered a "chemical signature" for OGC cancer.

What were the basic results?

The analysis included 335 patients (163 with OGC, 172 controls). More than two thirds of the OGC group (69%) had cancer that had spread to nearby lymph nodes.

Of the five chemicals of interest, four were expressed differently in the OGC group compared to the control group. This association remained after adjustments were made for possible confounders including patient age, other medical conditions and medications.

The test correctly detected 80% of cancer cases and 81% of non-cancer cases. 

How did the researchers interpret the results?

The researchers conclude: "This study shows the potential of breath analysis in non-invasive diagnosis of OGC. The potential benefits of this technology to patients may be early diagnosis and improved chance of survival. If placed as an endoscopy triage test, the benefits to the healthcare system may include cost-saving through reducing the number of negative endoscopies. However these findings must be further validated in an un-enriched larger population of patients undergoing diagnostic endoscopy and in false negative patients the value of repeat testing should be established."


This study aimed to see whether a breath test could be used to detect stomach and oesophageal cancers.

The researchers found that the breath test was fairly accurate in distinguishing between those with and without cancer.

The lead researcher, Dr Markar, said: "Because cancer cells are different to healthy ones, they produce a different mixture of chemicals. This study suggests that we may be able detect these differences and use a breath test to indicate which patients are likely to have cancer of the oesophagus and stomach, and which do not."

However he went on to say that the findings of this study would need to be validated in a larger sample of patients before being put into practice.

These are early findings presented at a conference. While they show promise, it is not possible to say from the available information whether the test could have a future role in practice. In most of the patients involved in the research the cancer had spread to the lymph nodes. It is unclear if the breath test would be able to detect cancers at an earlier stage.

As the researchers suggest, this test is likely to be most useful as a possible indicator for when endoscopy, a more invasive test, is needed in people who present with gastrointestinal symptoms.

A breath test sensitive enough to identify a "chemical signature" of cancer could be an ideal way to diagnose more patients at an earlier stage of the disease. However, at only around 80% accurate, the test is not infallible. The consequences of not carrying out further tests for the 20% with cancer who would test negative need to be considered.

It is also not possible to say whether this test could have an impact on survival outcomes. We need further larger studies to validate these findings and weigh up the risks and benefits before considering using this test to screen for cancer.

Read more about stomach cancer and oesophageal cancer, including symptoms and diagnosis.

Links To The Headlines

Simple breath test could catch deadly cancers at early stage. The Daily Telegraph, January 30 2017

Breath test could be used to detect deadly cancers. Daily Mirror, January 29 2017

TOP CHECK FOR CANCER New breath test could save thousands of lives a year by spotting deadly cancers with 85 per cent accuracy. The Sun, January 30 2017



Subscribe to East Cambridgeshire District Council aggregator - NHS Choices