Five-year 'death test' provides few answers


A “death test” that supposedly predicts the chance of a healthy person dying in the next five years has been developed by scientists, The Daily Telegraph reports. The test, which was based on a study sampling 17,000 people, measured a total of 106 biomarkers.

A “death test” that supposedly predicts the chance of a healthy person dying in the next five years has been developed by scientists, The Daily Telegraph reports.

The test, which was based on a study sampling 17,000 people, measured a total of 106 biomarkers; these included substances in the blood such as alpha-1-acid glycoprotein – a protein associated with inflammation.

Based on the levels of each biomarker, the test provided a summary score, which was then used to predict the risk of dying from any cause within the following five years.

The study found that four biological markers (biomarkers) in particular predicted the participants’ risk of dying from an illness within five years.

Researchers found that people in the top 20% of the summary score range were 19 times more at risk of dying in the next five years than people in the lowest 20%.

However, the implications of such a test are unclear. As this was an observational study and can only show an association between the biomarkers and risk of death, it was not able to predict what the underlying cause of death would be. Therefore, it does not provide an opportunity for targeted prevention or treatment strategies.

There are already a number of fairly reliable health outcome tests that are based on common sense factors, such as smoking history and body mass index (BMI).

Where did the story come from?

The study was carried out by researchers from several universities in Estonia and Finland, hospitals in Massachusetts, the Wellcome Trust Sanger Institute and Bristol University. It was funded by the European Commission, Estonian Research Council, Estonian Ministry of Education and Research, University of Tartu, Estonian Science Foundation, Academy of Finland, Academy of Finland Center of Excellence in Complex Disease Genetics, Finnish Funding Agency for Technology and Innovation, European Foundation for the Study of Diabetes, Jenny and Antti Wihuri Foundation, Novo Nordisk Foundation, Sigrid Juselius Foundation, Finnish Foundation for Cardiovascular Research, UK Medical Research Council, Wellcome Trust UK, Strategic Research Funding from the University of Oulu (Finland) and University of Bristol (UK).

The study was published in the peer-reviewed medical journal PLOS Medicine. All PLOS publications are free, as PLOS Medicine is an open access journal. Read the study for free here.

Four of the researchers are shareholders of start-up company Brainshake Ltd, which offers nuclear magnetic resonance spectroscopymetabolite profiling – the blood profiling technique used in the research study.

Overall, the media reported this study accurately, but generally took the findings at face value and did not discuss the limitations of the research. They also tended to overstate the immediate impact of the test, which is likely to be minimal. As the researchers themselves state: “Additional studies are, however, still required” to discover what conditions the biomarkers were uncovering and how these could be treated or prevented.

The media has raised concerns about what would happen if insurance companies were to obtain the results of a test such as this. However, no one is currently proposing to use it for any purpose in mainstream medicine, meaning this is a purely speculative viewpoint.

What kind of research was this?

This was an observational study of two large groups of people from Estonia and Finland, which aimed to see if biomarkers detected in people’s blood could predict death from any medical cause over the following five years. Due to its observational nature, it can only show an association, rather than causation, thereby limiting its potential impact.

What did the research involve?

The researchers took blood samples from over 17,000 people and measured the levels of 106 biomarkers (such as cholesterol). They recorded all causes of death over the next five years and looked to see if there was an association between death and any of the biomarkers.

Between October 2002 and February 2011, they recruited 50,715 volunteers from Estonia’s general population, with no restriction on health status or age (they sampled citizens aged 18-103). They then randomly selected 9,842 volunteers and performed a blood test using Nuclear Magnetic Resonance (NMR) Spectrometry.

The researchers looked at the cause of all death of these participants over subsequent years (median 5.4 years, range 2.4-10.7 years).

The research group analysed the 106 biomarkers to see if any were linked to subsequent death and then adjusted the results for known predictors of mortality:

  • high-density lipoprotein (HDL) cholesterol (“good” cholesterol)
  • smoking status
  • diabetes
  • cardiovascular disease
  • cancer

They also analysed the results, looking at:

  • age and sex
  • body mass index (BMI)
  • systolic blood pressure
  • total cholesterol
  • triglycerides
  • creatinine (a marker for kidney function)
  • cigarettes smoked a day
  • years of cigarette smoking
  • alcohol consumption

They repeated the study on a second group from Finland whose blood had been taken for a different study back in 1997 and stored in the laboratory. The researchers used the NMR Spectrometry test on 7,503 samples and used the Finnish registry to determine causes of any of their deaths from 1997 to 2002. They were also from the general population and their ages ranged from 24-74 years old.

The original testing using the Estonian sample looked for links between the biomarkers and death. Once found, they used the second Finish sample to test if the same links were found in a different group of people. This was a way of validating their initial findings in different groups, increasing the reliability of their results.

What were the basic results?

There were 508 deaths in the Estonian sample and 176 deaths in the Finnish sample.

Four biomarkers were identified that predicted the risk of all-cause mortality, after adjusting for HDL cholesterol, smoking status and whether they had any diagnosed conditons:

  • increased levels of Alpha-1-acid glycoprotein (a protein that is raised during infection and inflammation)
  • reduced levels of albumin (a protein that carries vital nutrients, hormones and proteins in the bloodstream)
  • reduced levels of very-low-density lipoprotein (VLDL) particle size (usually known for being “very bad” cholesterol)
  • increased levels of citrate (a compound that is an essential part of the body’s metabolism)

These biomarkers were also predictors of death from “cardiovascular causes”, “cancer” and “other causes”.

When all four levels were added together to get a biomarker summary score, 15.3% of people in the top 20% of the sample died within five years, compared to 0.8% in the bottom 20%. This means those in the top 20% had a relative risk of dying that was 19 times higher than those in the bottom 20%.

There were no notable differences between men and women in terms of the results.

How did the researchers interpret the results?

The researchers concluded that: “the biomarkers … may potentially aid the identification of high-risk individuals in need of medical intervention”. However, they stated that the clinical implications “remain unclear”, as a link between the studied biomarkers and the reasons behind increased mortality risk was “disparate” and could not be identified. The researchers were also unable to discover any prevention strategies.


This large population-based study was able to show which people were at increased risk of dying from cardiovascular, cancer or other causes over a five-year period. However, the researchers could not predict which illness a person may be at higher risk of getting or provide an opportunity for targeted prevention or treatment strategies.

Strengths of the study include the large sample size and the fact participants were taken from the general population. The results also remained statistically significant after adjusting for age, sex, current disease and many other recognised indicators of chronic disease.

However, the implications of such a test are unclear. As this was an observational study, it can only show an association between the biomarkers and risk of death. It does not predict what the underlying cause of death would be for an individual and does not therefore provide an answer in terms of treatment.

At best, this type of test could encourage people to adapt a healthier lifestyle; at worst, it could lead to higher anxiety, higher risk-taking and a sense of fatalism.

There is also the danger that it could lull people into a false sense of security if they were deemed to be at lower risk and make them less likely to live a healthy lifestyle. The media have also raised concerns about the possible implications if insurance companies were to make use of this type of test. However, these are purely speculative at this stage.

In summary, this study does not alter the general prevention and health promotion strategies to reduce the risk of death.

Predicting what is likely to kill you, barring accident, is not rocket science.

The biggest risk factors for potentially fatal conditions such as cancer, heart disease, stroke and diabetes are already well documented and include:

  • smoking
  • obesity
  • excessive alcohol consumption
  • lack of regular exercise 
  • a poor diet, lacking in fruit and vegetables

It is also important that you attend NHS Health Check screening appointments when invited.

Article Metadata Date Published: Wed, 22 Nov 2017
Author: Zana Technologies GmbH
NHS Choices