Actuarial Outpost
Go Back   Actuarial Outpost > Actuarial Discussion Forum > Life
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions

Thread Tools Search this Thread Display Modes
Old 04-11-2018, 04:46 PM
Don Quijote Don Quijote is offline
Join Date: Sep 2001
Posts: 1,333

An earlier post from JMO referred to data mistakes as a potential cause for odd mortality spikes.

Some of the state’s top health experts released a report in the medical journal Obstetrics & Gynecology on Monday that used a new method for counting — and found that the number of women who died dropped from 147 to 56.


In 2016 Marian MacDorman, a professor at the University of Maryland Population Research Center, released a study in Obstetrics and Gynecology showing that in 2012, 148 Texas women died from pregnancy-related complications, including excessive bleeding, obesity-related heart problems and infection. Two years earlier, 72 women had died from those causes.

MacDorman wrote at the time that “in the absence of war, natural disaster, or severe economic upheaval,” such a rise seemed unlikely. The study made national and international news and raised questions over how Texas was addressing women’s health.

The state researchers addressed MacDorman’s findings in Monday’s study: “Given the significant reduction in the maternal mortality ratio when using confirmed maternal deaths, this high estimate reported was likely the result of data error.”

It seems that the death registries added a box to check for anytime that the deceased female was pregnant at time of death. Like any manual process, sometimes that box got checked in error. When measuring a small number of deaths, the error in the box checking was enough to overwhelm the actual results.
Reply With Quote
Old 05-15-2018, 09:40 AM
campbell's Avatar
campbell campbell is offline
Mary Pat Campbell
Join Date: Nov 2003
Location: NY
Studying for duolingo and coursera
Favorite beer: Murphy's Irish Stout
Posts: 82,934
Blog Entries: 6

Premature mortality and the long decline of hope in America
For the first time in this millennium, unemployment in the United States is below 4 percent.

Most parts of the economy are growing at a respectable rate, and market confidence is up. So why are Americans so despondent? The answer to this question matters not just for America but—as the political and policy shifts of the last two years have shown—for many other parts of the world.


Carol Graham
Leo Pasvolsky Senior Fellow and Research Director - Global Economy and Development
In part, the despair is due to the happenings in the “other America”: blue-collar workers who have experienced declining incomes even as new technologies and skills power success in thriving sectors. Those activities are so far removed from their daily lives they might as well be happening in another country. Many of these people live in the heartland, in hollowed out manufacturing towns and decaying cities. As these places have lost business, people who still live there have lost hope.

One marker of this lost hope is the 15 percent of prime-age males who have dropped out of the labor force. For the purpose of calculating the headline unemployment rate, they have ceased even to be a statistic. Another is the surprising amount of support for an antiestablishment, anti-immigrant, and xenophobic political agenda intended to incite anger in lieu of proposing realistic remedies. The starkest marker, though, is the rise of “deaths of despair” in the United States: preventable deaths due to suicide, opioid and other drug overdoses, and deaths related to poor health behaviors. The U.S. is the only rich country in the world where mortality rates are increasing rather than falling.

In recent work published in the Journal of Population Economics, Sergio Pinto and I explore the role of hope—or its absence—in explaining recent trends in premature mortality. We use metrics of well-being and ill-being from Gallup surveys and county-level Center for Disease Control data on deaths of despair. Our metrics of desperation, stress, and worry map closely with trends in deaths of despair at the level of the individual, race, and place. The dimension of well-being that is most closely associated with higher mortality rates is lack of hope, a marker that is starkest among whites who have not completed college. Conversely, individuals and places with higher levels of hope, which tend to be urban and more racially diverse, display much lower levels of deaths of despair.

In a new paper with Kelsey O’Connor, we take a historical look and ask how optimism was related to mortality before the rise in “deaths of despair” that began in the late 1990s. Using the U.S. Panel Study of Income Dynamics, we find evidence from as early as 1968 that more optimistic people live longer. The relationship depends on many factors, including gender, race, health, and education.

We explored these and other variables as determinants of individual optimism between 1968 and 1975. Greater education was associated with greater optimism; so was having wealthy parents. Back then, women and African Americans were less optimistic than the average. In recent years, this pattern has changed markedly. Now, on average, women and African Americans are more optimistic than their male/white counterparts.

We then predicted optimism for the same individuals in subsequent years, thereby generating our best guess as to how optimism changed for various demographic groups between 1976 and 1995. We found that people with less than a high school degree show the greatest declines in optimism, suggesting long-run links to premature mortality and deaths of despair. This is the same demographic with the lowest levels of hope today. Keep in mind that the pool of people with less than a high school degree over the same period shrank as high school completion rates increased, and the cohort with less than a high school diploma 40 years ago is comparable to the less than college-educated cohort today.

While the loss of hope in America does not apply across the board—and poor minorities in particular are far more optimistic than poor whites—the deep despair in parts of the country is having negative consequences for our society, political and civic discourse, economy, and longevity. Better understanding its causes and consequences—and its historical roots—is an important part of finding a solution. Tracking well-being metrics regularly, meanwhile, as many countries are already doing, is an efficient way to monitor social health along the way. Local and community level initiatives that seek to enhance community well-being—as in the City of Santa Monica and in the What Works Well-Being initiative in the U.K.—can provide another part of the solution, particularly in places where economic solutions are unrealistic or insufficient.


LinkedIn Profile
Reply With Quote
Old 07-09-2018, 02:52 PM
campbell's Avatar
campbell campbell is offline
Mary Pat Campbell
Join Date: Nov 2003
Location: NY
Studying for duolingo and coursera
Favorite beer: Murphy's Irish Stout
Posts: 82,934
Blog Entries: 6

Some interesting graphs from PartnerRe

Mortality Rate Improvements – End of an Era, For Now…
Improvements in mortality rates characterized the twentieth century – the trend of living ever longer and healthier lives seemed assured.

Recent mortality rates, however, point to a change. The positive trend has slowed. But why? Are all ages affected? And surely medical advances will anyway have us back on track asap?

To answer those questions, we look at high-level developments in the main causes of mortality and their relevant risk factors. The recommendation is for caution: The future will remain less rosy, for now at least…

A slowing improvement in mortality rates
Figure 1 shows the overall trends in mortality rates and life expectancy at birth for the US population. US data is predominantly used throughout this paper as detailed mortality data is available for the US.

The graph shows that the positive trends of the past began to level off in 2010. Comparing 2000-2010 to 2010-2016, mortality rate improvements deteriorated across all ages, with the worst deterioration in males aged 35-441. Similar trends are mirrored in other developed markets: e.g. in the UK, male mortality improvements averaged 3.1% per year from 2000-2011, falling to 0.7% a year from 2011-20162.

Figure 1: Age-adjusted mortality rates (blue line) and life expectancy at birth (orange line), both sexes, all races, US population, 1900-2015. Mortality rates show steady improvement (reduction) over the period, but in fact began to level off in 2010. Life expectancy at birth has likewise increased, but shows a similar levelling off in recent years. Looking far back, the impact of the 1918 flu pandemic is clearly visible, as is the bounce back in rates and life expectancy after this event. Source: CDC3.

We now break down this data by cause-of-death (leading causes, US); coronary heart disease, stroke and cancer, see figure 2. For each cause, we consider which, if any, age groups are most impacted by slowing mortality rate improvements, and how the evolving medical expectations and relevant risk factors for each cause might impact the future trends for these diseases.

Figure 2: Age-adjusted mortality rates by major cause of death, US population, both sexes, all races, 1960-2015. Most notably, heart disease rates (green line) fell significantly since the 1960s, being the major contributor to overall population mortality improvements during that time period, but flattened out in the last five years. Cancer rates (red line) continue to steadily fall. ‘Accidents’ refers to ‘unintentional injuries’, i.e. excludes suicide. After these main causes of death, Alzeimer’s disease and diabetes were the next two leading causes in the US in 2015-164. Source: CDC3.

The trends for the main natural causes of death
Coronary heart disease and stroke (28% of deaths5): Five decades of mortality rate improvement from these diseases in the US across all age groups was a major contributor to falling overall population mortality rates. In the UK, for example, 70% of all improvements from 1968-2010 were due to the decline in deaths from circulatory diseases2.

“In the UK, 70% of all improvements from 1968-2010 were due to the decline in deaths from circulatory diseases.”

However, over the last few years, rate improvements from coronary heart disease and stroke have reduced (figure 2); for heart disease (US population, age-adjusted, both sexes, all races), the average annual mortality improvement rate for the period 1999-2016 was 2.7%, whereas this fell to 0.9% for the more recent five-year period 2011-2016. This is important given that this is the leading cause of death. The only age groups not seeing a deterioration were ages 1-4 and 25-34. Ages 65-74 were worst affected, with improvements of 3.4% for 1999-2016, versus just 0.3% for 2011-20165.

“Over the last few years, mortality rate improvements from coronary heart disease and stroke have reduced.”

The earlier improvements can be attributed to lifestyle changes (especially reduced smoking) and medical advances, including bypass surgery and pacemakers in the 1970s, followed by coronary stents and stroke units in the 1990s.

Rate improvements are now reducing and further significant rate improvements from medical treatment are not anticipated for this disease group: most therapeutic innovations are already widely implemented, clinical trials for heart drugs significantly lag behind those for cancer6, and although new drugs offer hope within the next two decades, these are primarily for smaller subgroups of heart patients.

Overall, and before any meaningful implementation of next generation medicine, the period of strong mortality rate improvements for coronary heart disease and stroke would appear to be behind us.

2.7% 0.9%
Fall in US mortality rate improvements for heart disease,
1999-2016 cf. 2011-2016, Worst affected ages: 65-74

Cancer (21% of deaths5): After many decades of gradually increasing mortality rates, cancer deaths began to fall in around 1990 and have continued to steadily decline (figure 2) at a relatively consistent 1.5% (US population, age-adjusted, both sexes, all races) since 1999. The highest average annual improvement 2015-2016, 3.2%, was seen in the 45-54 age group, while ages 25-34 and 35-44 recorded slight deteriorations, respectively 1.1% and 0.2%5, something to watch.

Steady US mortality rate improvement
for cancer since 1999

As for coronary heart disease and stroke, improved lifestyle has been a contributor to the positive trend. From a medical perspective, improvements from radiation therapy and chemotherapy in the last century are expected to be succeeded by future applications in cancer genomics, personalized medicine such as immunotherapy and earlier detection from liquid biopsy over the next 10-15 years. The high number of compounds in clinical development for cancer adds to the positive future outlook6.

External (c.f. natural) causes of death increasing in significance in the US
Now the third highest cause of US mortality (6% of deaths5).

Mortality rates from external causes (e.g. from traffic accidents, homicide and self-harm, including suicide and poisonings (mainly drug/opioid addiction)) have been slowly increasing in the US since 1999 (1.8% over the period 1999-2016, with variations by cause and age group).

Rates accelerated upward in 2015-2016. From 2014-2016, age groups 25-34 and 35-44 respectively experienced substantial 16.1% and 14.4% increases in accident mortality rates5. In 2016, the leading cause of death for ages 25 to 44 was poisonings, followed by suicide and then traffic accidents7.

Opioids are a significant contributor to the upward trend in the US, impacting all ages (above 15 years) and social classes, but with higher mortality rates observed for lower socio-economic groups8. Canada9 and the UK10 are also affected, but to a lesser extent.

Increase in US opioid mortality rates 2015 to 2016
Worst affected ages: 15-44

From 2015 to 2016, US opioid mortality (all ages) rose by a staggering 27.4% (figure 3; ages 15 to 74 all experienced over 20% mortality rate increases, ages 15-44 being the worst affected at over 30%5. US overdose deaths (all drugs) rose to 64,000 in 2016, a 20% increase on 201511.

Figure 3: Age-adjusted opioid mortality rates, 1999-2016, US population, all races. The upward trend followed by a sharp increase in 2015 is apparent. Source: SOA5.

Despite potential future improvements for some external causes of death from developments such as driverless cars and stricter weapons controls, the overall outlook for the US remains negative. If the statistics for these causes of death continue to worsen, the impact on future mortality rates will be meaningful.

AMR – the potential to undo future steps forward
With increased understanding of infectious diseases and how they are spread, combined with the power of antibiotics, mortality rates from infectious diseases have been in strong decline since the beginning of the twentieth century.

Future rates, however, are threatened by increasing antimicrobial resistance (AMR), which is rising at an accelerating rate, and by the fact that there is a lack of investment in new antibiotics12; only 1.6% of drugs in clinical development by the world’s 15 largest pharmaceutical companies were antibiotics13.

Next generation medicine could get things back on track – but not yet
Next generation medicine represents a sea change in our capabilities to improve mortality rates. But when will this happen? Digital health, for example, is already upon us and developing fast (e.g. artificial intelligence, eHealth, wearables, electronic health records, telemedicine and health apps). Genomics, the key to a new level of disease understanding, innovations in disease prevention, new drug targets14 and better drug efficiency, is also developing fast but still has many challenges to overcome, most likely requiring at least another two decades. The two combined (e.g. for simulations of an individual’s likelihood of disease and targeted, preventative surveillance) offer even greater future potential through precision, individualized medicine.

Impact of socio-economic factors
While overall mortality rates in the US are falling across all socio-economic groups and ages, global studies observe variations linked to socio-economic factors such as wealth, marital status, level of education and race15,16,17. For example, as previously noted for opioids, higher social class can be a general proxy for lower mortality. How socio-economic inequalities in mortality are changing is complex and varies, for example, by age18: inequalities have decreased for younger ages (0-20), notably increased for those aged over 50, remained steady for women aged 20-50, and decreased for men aged 20-50 (closing the gap between men and women in this age group).

And the overall prognosis?
Mortality rate improvements, largely driven by wins from healthier lifestyles (smoking reduction) and advances in diagnostics and the treatment of common diseases, have slowed in the last decade.

For heart disease, US ages 65-74 experienced the worst slowing. Cancer rates are still showing some promise, but the cohort effect (see below, ‘Background insights’) that helped to boost the past in some regions is fading, external causes of death (accidents and suicide) are increasing in significance in some regions (in the US, especially since 2015 and in ages 25-44; for opioids, ages 15-74), and AMR is on the rise. Next generation medical progress, digital health and genomics in particular, may begin to claw back some of that downward trend in coming years, but any meaningful impact from these areas will need more time.

We find ourselves in a dynamic, interim phase of mortality rate improvement. Caution is needed. Slowing will remain in place in some regions, at least for a while. Thereafter, the new mortality landscape will be drawn out by the interaction and timing of next generation medical advances and by the specific progression of AMR, lifestyle and socio-economic risk factors.

“We find ourselves in a dynamic, interim phase of mortality rate improvement. Caution is needed.”

Your partner for Life risks
Our experts – a closely collaborative team of Life actuaries, underwriters, market specialists and medical experts – continually analyze the trends in and determinants of mortality data to ensure best practice for supporting our clients’ Life term and annuity portfolios.

We are a forward-looking discussion partner for our clients for all Life risks. Please contact us to find out more about our Life risk solutions or to discuss how the trends summarized in this paper could impact your portfolio.

Background insights
The cohort effect – past driver of mortality rate gains now losing its impact

The cohort effect refers to the observation that those born in a particular period, for example in the UK the cohort comprising those born between 1925 and 1944 (centered on 1931), experienced better mortality improvements than other generations19. The cohort in question had a very significant and positive impact on overall historic population mortality rate improvements. The contribution of this cohort to mortality rate improvements, however, now won’t repeat, an effect that is contributing to the observed slowdown in mortality rate improvements.

The effect has been documented in the UK, US20 and Canada21, but is most pronounced in the UK. There is no clear documentation of it in other countries. The specified cohort experienced the depression, war, smoked, quit smoking in their masses, and later benefitted from major medical advances in the 1960s and 70s. It experienced materially improved mortality compared to the preceding cohort and subsequent cohorts have not improved as much.

1 WillisTowersWatson ‘Insights’, 2016.
2‘Mortality Improvements in Decline’, The Actuary, August 2017.
5US Population Mortality Observations, Updated with 2016 Experience, Society of Actuaries (2018).
7Center for Disease Control:
8e.g. – Study “found a difference in mortality of 29.22 overdose deaths per 100 drug users in the lowest socioeconomic group compared to the most advantaged group”.
11e.g. .
14Identifying and proving that DNA, RNA or a protein molecule is directly involved in a disease process and can be a suitable target for the development of a new therapeutic drug.
15e.g. –
16 micpositionbasedonthenationalstatisticssocioeconom icclassificationenglandandwales/2015-10-21
17Rogot, Eugene, Paul D. Sorlie et al .1992. “Life Expectancy by Employment Status, Income, and Education in the National Longitudinal Mortality Study.” Public Health Reports 107(4):457–61.
19e.g. ‘The Cohort Effect: Insights and Explanations’, R.C.Willets, 2004.
20e.g. ‘Mortality Improvement Scale MP-2016’, Society of Actuaries, October 2016.

LinkedIn Profile
Reply With Quote
Old 07-10-2018, 09:41 AM
DES DES is offline
Join Date: Nov 2003
Posts: 73

Any idea what the right y-axis is in the graph? Seems silly anyway since it's almost the exact same scale as the left axis.
Reply With Quote
Old 07-10-2018, 10:13 AM
Kalium Kalium is offline
Join Date: Dec 2014
Posts: 30

Originally Posted by DES View Post
Any idea what the right y-axis is in the graph?
The original data on the CDC site doesn't appear to have a right axis.
Reply With Quote
Old 07-10-2018, 10:20 AM
campbell's Avatar
campbell campbell is offline
Mary Pat Campbell
Join Date: Nov 2003
Location: NY
Studying for duolingo and coursera
Favorite beer: Murphy's Irish Stout
Posts: 82,934
Blog Entries: 6

Yeah, that looks like an error. There was an earlier graph with a secondary axis that made sense, and maybe they just copied over the graphs and dumped in new data without realizing.

I'll email them about that.

and I found where they grabbed their data (as they linked to it):

LinkedIn Profile
Reply With Quote

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

All times are GMT -4. The time now is 07:14 AM.

Powered by vBulletin®
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
*PLEASE NOTE: Posts are not checked for accuracy, and do not
represent the views of the Actuarial Outpost or its sponsors.
Page generated in 0.14702 seconds with 9 queries