Matthew Edwards 

Matthew Edwards

Specializes in life insurance risk consulting and software.

Towers Watson, Reigate

Analysis of the basic drivers of longevity variation shows that the outlook for improvements may be easing, particularly in the U.K.

One of the most surprising developments in Western society over the last 20 to 30 years has been the dramatic increase in life expectancy. In the U.K., the life expectancy of a 65-year-old male increased from 13 years in 1981 to 18.3 years in 2011.* Similar improvements have been seen in many other countries, including the U.S. and nations in Continental Europe.

To what extent will such improvements continue? Most commentary about the future of longevity improvements paints a generally rosy picture, with a combination of medical advances and lifestyle changes leading to material longevity improvements every year and inexorably longer lives.

Longevity improvements have always been hard to predict. Simple extrapolation is particularly unsuitable, because future improvements may be driven by developments quite dissimilar from those that have driven recent increases. The best method to consider the extent of future mortality improvements, in our view, is to analyze the factors or drivers underlying improvements, and then assess how those factors are likely to develop over the medium to long term.

This article considers those factors, with particular reference to how they may indicate a change from the substantial improvements of the past.

Cardiovascular Disease

One of medicine's great successes in recent years has been in the treatment of cardiovascular disease (CVD). In the U.K., CVD-related mortality improvements between 2001 and 2011 have averaged 7% per year, increasing overall non-CVD improvements of approximately 1% per year into much greater overall improvements of 3.5% (ages 65 to 74), 3.1% (ages 75 to 84) and 1.8% (ages 85-plus). Absent the contribution from CVD, there would have been barely any overall improvement for ages 85-plus.

The cause-of-death table (Figure 1) shows how CVD improvements have contributed to the overall 3.5% average annual improvement for men age 65 to 74.

Figure 1. Male mortality (deaths per million), ages 65 to 74

The outlook for mortality improvements is therefore strongly linked to the outlook for improvements in cardiovascular mortality. To understand the prospects for cardiovascular mortality, we need to consider what has driven the extraordinary reductions to date. There are three main drivers:

  • Lifestyle changes
  • Surgical intervention
  • Pharmaceutical treatment


Lifestyle changes have been the principal driver of CVD-related mortality improvements. The major element is the reduction in smoking. Our own analysis indicates that smoking reductions have accounted for broadly half of improvements.

However, the reduction in smoking over the last 40 or so years in the U.K. has been so great that there is clearly much less to be gained from further reduction. Furthermore, looking at Figure 2, reductions among the 60-plus age range seem to have plateaued compared with changes across all ages.

Figure 2. Variation in smoking prevalence (% population)

There seems little chance of a material reduction in high-age CVD from future smoking reductions.

Surgical Intervention

Surgical intervention has been primarily in the area of angioplasty, with techniques such as keyhole surgery and stents allowing fast and effective repair to damaged arteries in a way that would have been unimaginable 50 years ago.

The application of these incredible innovations exploded during the 1990s and early 2000s, with the total number of interventions in the U.K. increasing from around 10,000 in 1991 to around 90,000 per year over the last three years. Almost all who need such an intervention and can be operated on in time now receive the required treatment. While we might expect further improvements in the mechanics of these techniques, there is no impending advance in sight that might come anywhere close to the revolution that these techniques themselves represented.

Pharmaceutical Treatment

Recent pharmaceutical treatment for CVD has centered on statins, a class of anti-inflammatory drugs designed to block a particular step in the body's biosynthetic pathway for the production of cholesterol. Prescriptions for statins have increased to around eight million people in the U.K., largely in older age ranges.

With so many now receiving statins, it is difficult to see further substantial long-term mortality reductions. Furthermore, there is increasing concern about their side effects, particularly diabetes. For instance, a study of almost half a million patients on statins by Dr. Aleesa Carter was summarized as follows:

The phenomenal scale of this "unintended adverse effect" (predictably, a particular hazard for older persons) is without precedent in the history of medicine.**

While Dr. John Abramson, writing in the British Medical Journal (October 2013), noted:

Statin therapy in low-risk people does not reduce all-cause mortality or serious illness and has about an 18% risk of causing side effects that range from minor and reversible to serious and irreversible.

Given the already high intake of CVD-related pharmaceuticals such as statins, and given increased concern about side effects, we do not expect to see pharma-related improvements in CVD mortality anywhere near what has been seen over the last 10 to 20 years.

Across the three drivers of CVD improvements, there is nothing to suggest any material reductions likely to compare with those of the last 10 years.

Diabetes and Obesity Trends

Diagnosed diabetes mellitus has approximately tripled for men in the U.K. between 1991 and 2011, and increased by a factor of around 2.5 for women. These increases have been broadly similar across age ranges. Many other Western countries have also experienced large increases in diabetes, to the point that it is often called a "disease of Western civilization."

Diabetes is one of the major factors in a number of conditions, particularly CVD. Analysis by Towers Watson shows that the mortality of diabetics in the 70 – 79 age range was 175% that of equivalent members of the population with no history of diabetes.

There are no convincing and generally accepted explanatory models for the recent epidemic of diabetes. Interestingly, there is some evidence that current nutritional guidelines around low-fat, high-carbohydrate diets may be partly responsible. Whatever the underlying causes, there have been no public health policy pronouncements on the subject that differ from those of the past two decades (e.g., eat according to official guidelines, exercise more). Therefore, there is no reason to expect the situation to improve.

We expect diabetes mellitus to become increasingly prevalent in older age ranges and also expect related mortality to deteriorate. The outlook is similar for the prevalence of obesity and its related impact on longevity.

Health Care Budgets

A significant part of the U.K.'s historical mortality improvements is attributable to improved National Health Service (NHS) interventions, in particular, screening (and hence early diagnosis), pharmaceutical preventive strategies and surgical treatment post-incident. We should therefore look at how NHS expenditure has increased in the past, and what is expected in the near future.

The NHS's expenditure over the period of 1951 to 2011 rose in real terms by a factor of 10.3. Allowing for population change over that period, per capita expenditure rose in real terms by a factor of 8.3, an annual increase of 3.6% (or 2.8% per capita for people over age 65).

Given the combination of government budget pressures, at best, the NHS's long-term expenditures can be expected to "flatline" in real terms, and because the population is increasing (driven by an increasing proportion of elderly), our estimated trend from 2011 to 2031 is a –0.6% annual decrease per capita, or –1.6% per capita for people over age 65.

This clearly poses a major challenge, especially in a sector where costs tend to grow substantially more than cost price inflation. The NHS's budgetary outlook implies that its per capita expenditure will not produce any material mortality improvements.

Bacterial Resistance and the Failure of Antibiotics

For some time now there has been heightened concern in the medical sector about the growing resistance of bacteria to antibiotics. Some bacteria have evolved by selective mutation (facilitated by their short life span and some capacity to transfer genetic information among bacterial "siblings") to become resistant to the standard range of antibiotics, a real challenge since there have been almost no promising results in the development of new antibiotics.

Infectious diseases aren't the main killers in the 65-plus age group, but the possibility of death from a routine infection raises the particular threat of large-scale deaths post-operation or while patients' natural defense systems are weakened by anticancer treatments — an outcome that would clearly have great consequences for mortality in this higher age range.

U.K. Chief Medical Officer, Professor Dame Sally C. Davies, summarizes this possibility in three statements:***

  • We are losing the battle against infectious diseases.
  • Bacteria are fighting back and are becoming resistant to modern medicine.
  • In short, the drugs don't work.

All other things being equal, we should reasonably expect mortality related to infectious diseases in the developed world to deteriorate, not improve.

Cancer and the Pipeline for Medical Advances

While recent mortality improvements have been driven primarily by improvements in CVD mortality, cancer now represents the major cause of death, as shown in Figure 1. Lifestyle changes, particularly smoking reductions, represent a large part of the cancer improvements (1.8% per year). We see no prospect for large-scale improvements from further smoking reductions, while many other lifestyle longevity measures seem to be deteriorating.

Expected developments in surgical treatment techniques and in pharmaceuticals offer the potential for longevity improvements, although the current outlook is, in our view, uninspiring across the main cause-of-death categories.

The great strides made over the last 50 years, incredible developments that we now take for granted, are part of the challenge. Nothing in recent years compares with organ transplants, hip replacement, kidney dialysis, radiotherapy, keyhole surgery and angioplasty, and in their wake, we face diminishing returns. Every "eureka" of the past makes the residual post-eureka mortality more impervious to further breakthroughs — and more costly.

The pharmaceutical sector faces dual challenges: increased regulatory pressure and the increasingly challenging problem of finding any form of quantum breakthrough to mirror the giant advances made 50 years ago. And both manpower and R&D expenditure are shrinking.

More positively, human genomics offers the possibility of personalized treatments subsequent to individual genotyping. Although this is likely to lead to more effective cancer treatment for individuals with the "right" genes, it is still a long stretch before there is a cancer cure, and trials to date show modest effects. For instance, one of the leading targeted breast cancer drugs seems to increase life expectancy by about half a year to around 30 months — clearly beneficial, but in no way revolutionary.

While major medical and pharmaceutical breakthroughs need to be considered by insurers setting their one-in-200 stresses for economic capital, at the level of best-estimate assumptions, there is little prospect for developments to rival those of the past.

Recent Mortality Statistics

Our rationale has been to consider the long-term outlook for longevity improvements by looking at the major drivers of improvements. In our view, this is a better approach than simply considering overall mortality statistics and extrapolating. That said, however, it is instructive to look at the U.K.'s more recent mortality results for older ages. These indicate a material reduction in the longevity improvements seen in previous years at higher ages and, as such, support the points made above.

For instance, recent 2013 mortality figures for England and Wales published by the Office for National Statistics show mortality at higher ages deteriorating, with the number of male deaths at ages 75 and over increasing by 2.6% from 2012, while population numbers were broadly stable. A harsh 2012/2013 winter only partially explains the increase and doesn't even address the possibility of more frequent future climate swings.

Changes in the population mortality of England and Wales in earlier years are reflected in the two most recent versions of the U.K. actuarial profession's mortality projection model, with the 2012 and 2013 versions producing reduced life expectancy results, assuming a constant long-term 1.5% increase for the sake of comparison across model versions (Figure 3).

Figure 3. 2012 and 2013 life expectancy results

While extrapolating from two years of statistics would be nonsensical, these results do lend some support to our earlier points regarding changes in longevity drivers.

Basis Risk

Although there seems to be a case for some softening of long-term improvements across the population, the situation may be different for annuitants and pensioners. There is an increasing awareness of the possibility of basis risk arising in longevity improvements, as discussed in our article, "Longevity Strides Raise Basis Risk Questions" (Emphasis 2013/4).

The link between socioeconomic profile and sources of mortality improvement should also be considered. For instance, one of the main areas of possible material improvement lies in genotyping and personalized medicine, but health budgetary pressures may well lead to advances enjoyed only by those who can afford private care.

Solving the Longevity Jigsaw

Figure 4 illustrates factors to consider in setting appropriate long-term longevity improvement assumptions for an annuitant portfolio.

Figure 4. Considerations for long-term longevity improvement assumptions

The foundation is a proper consideration of the underlying drivers of longevity improvements and expert medical opinion on the likely prospects of improvements in primary, secondary and tertiary care for the main causes of death.

In our view, a disease-based mortality model is the best way to bring together a range of views that illuminates how expected improvements across a variety of diseases and conditions could translate into one overall long-term improvement assumption, and could improve an understanding of the implications of a range of scenarios.

Finally, these considerations need to be adjusted as necessary for basis risk, i.e., taking account of profile differences between the identified annuitants and the general population.

For comments or questions, call or email
Matthew Edwards at +44 1737 284771,


* Period-of-life basis, Office for National Statistics

** Daily Telegraph, Dr. James Le Fanu. June 2013

*** The Drugs Don't Work, Professor Dame Sally C. Davies, Dr. Jonathan Grant, Professor Mike Catchpole, Penguin. 2013