infocus€¦ · Head of Underwriting & Claims Strategy Blair Sievering Head of Product Development Source: The Syndicate/Defaqto IFA survey, June 2010 and ABI Going forward, aligning

Creating certainty by designing and pricing critical illness products that deliver on their promise

infocusIssue No. 60, September 2013

Critical illness insurance celebrates thirty years this year, having first been developed in South Africa by Dr Marius Barnard. In light of this it seemed fitting at our Annual Seminar in June to present recent developments from international Critical Illness (CI) markets. In this article we focus on a specific product feature that has recently been introduced by at least one insurer in South Africa.

The first critical illness insurance policy was launched on the 6 October 1983, making it 30 years old this year.

We look at how this could improve consumer perceptions of CI and other enhancements that could address concerns highlighted in recent consumer surveys. All of which are of interest to offices writing CI business in the UK.

Medical AdvancementsMedical Advancements Protection provides for claims to be assessed against new medical procedures or diagnostic protocols in clinical practice, where these have replaced the severity measures defined in the original policy terms and conditions.

A simple example of how this works might be:

After having the policy for ten years, the life assured is diagnosed with cancer. The original claim definition uses the TNM medical classification system to determine whether the claims criteria have been met. However, at that time the claim is submitted, the TNM classification system is no longer used in clinical practice to classify this disease. The severity required under the obsolete TNM system is mapped to the new classification and the appropriate amount will be paid.

In many respects this sounds a lot like what we would expect to do anyway in practice. Whilst this might be true for most conditions, it is not something we would automatically do without considering the future claims cost.

Consider heart attacks for example. We only need to look at the changing clinical diagnosis and the move from conventional to highly sensitive troponin tests, together with the growing numbers diagnosed each year, to understand the value this promise would have had if introduced alongside the current definition.

Heart attack incidence rates in Scotland had been showing a strong decreasing trend throughout the early and mid-2000s, but from 2009 this trend reversed for younger lives and levelled off for older lives. Across all ages, the overall incident rate has been rising. The Information Services Division (ISD) website expresses the view that: “The recent increase is likely to be due to the change in the clinical definition of AMI which is now diagnosed using more sensitive troponin tests”.

Critical Certainty

2 | Hannover Re UK Life Branch

Highly sensitive tests were introduced later in England than Scotland, creating a lag in the increased incidence in the data. But Hospital Episode Statistics (HES) data already suggests a similar upturn. Given the scope of conditions covered by current comprehensive critical illness plans, including such a promise would not be a free add-on, but a feature that offers genuine value and therefore carries a cost. When considering the introduction of such a promise, it is vital to understand the implications across all CI definitions, including those where the trigger is diagnosis only without supporting symptoms or other clinical features. For example, the potential identification of underlying conditions as a result of screening needs careful consideration if we are to avoid future claims cost increases. Introducing new screening programmes or expanding current ones could lead to diagnosis before symptoms arise.

As an industry, we are looking to build long term customer trust and loyalty. A promise like Medical Advancements Protection could be one way of achieving this.

UK MarketTo develop the right products for consumers it is vital to understand what they believe a policy will cover and what will motivate them to buy it. We also need to consider the distribution journey and what improvements advisers would most welcome.

Contents Critical Certainty ......................................................... 1 Insights into Critical Illness Experience .................... 5

We hope you enjoy infocus and we welcome your feedback, please forward comments to Kirsteen Grant at [email protected].

© Hannover Re UK Life Branch. All rights reserved.

The opinions expressed in this publication are those of the authors. This publication is subject to copyright. All rights reserved. Apart from any fair dealings for the purposes of research or private study, no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means without the prior permission in writing of Hannover Re. Single copies may be made for the purposes of research or private study. Multiple copying of the content of this publication without permission is always illegal.

Recent research carried out by HR UK in partnership with The Syndicate asked consumers to choose the description that best described what CI insurance provides. The survey covered a broad cross-section of consumers, some who currently held CI, some who had held it previously, and some who never had.

It is disappointing to see that not everyone who bought CI can recall what cover it provides – and that those who have never held a CI policy (who represent the potential market for new sales) have little understanding of what the product does. (See Graph 1)

Graph 1: ‘What cover does your Critical Illness policy provide?’

Additional research from The Syndicate asked consumers what they look for when purchasing CI. The top five responses were:

When consumers were asked what top three features would help them buy an insurance product like critical illness the responses were:1. A clearer understanding of what the policy covers and

does not cover (47%)

0% 10% 20% 30% 40% 50%

A lump sum paid to youif you’re diagnosed with

one of the named seriousillnesses by your policy

A regular income is paidto you if you’re diagnosed

with one of the seriousillnesses covered by

your policy

A lump sum is paid to youif you are unable to work

due to illness or injury

Don’t know

Never held

Used to hold

Holders

Hannover Re UK Life Branch | 3

2. Easy to understand product and company information without any jargon (35%)

3. Ability to compare quotes from different product providers (23%)

4. An endorsement or recommendation from an organisation such as Which? or Money Advice Service (MAS) to reassure them about the quality of the product they were buying (18%)

5. The ability to apply easily online (18%)

Previous research by Defaqto complements these findings, revealing that when asked “Would you rather have CI products with improved definitions, or with additional conditions?” IFAs responded as shown in Graph 2.

The findings of these and other research initiatives consistently suggest two clear themes that we, as an industry, need to address when developing our products. Consumers are looking for products that are easier to understand and that do what they say.

The Medical Advancements Protection product feature discussed on Page 1, could be one way of helping to ensure that policies ‘do what they say on the tin’. But in addition to improving certainty in the future, we also need to ensure that current definitions also do what they say.

Table 1

Graph 2: IFA responses

Source: Defaqto IFA survey, June 2010

Improving DefinitionsConsumers often assume that being diagnosed with a named policy condition or undergoing a stated procedure will be sufficient cause for claim. The qualifying statements within definitions used to establish severity are often poorly understood. Policy definitions that are heavy with technical or medical detail can create uncertainty.

Heart attack and stroke are two of the three conditions that must be included within a policy for it to be called critical illness in the UK. Table one summarises how the Association of British Insurers (ABI) definitions for these conditions developed post 2006, along with some suggestions on ways of providing greater certainty for consumers.

72% Improve existingdefinitions

28% Add additionalconditions

Condition Pre-2006 ABI definition Current ABI definition Improvement from current

Heart Attack The death of a portion of heart muscle, due to inadequate blood supply, that has resulted in all of the following evidence of acute myocardial infarction:

• Typical chest pain; new characteristic electrocardiographic changes; the characteristic rise of cardiac enzymes; troponins or other biochemical markers;

where all of the above shows a definite acute myocardial infarction. Other acute coronary syndromes, including but not limited to angina, are not covered under this definition.

Death of heart muscle, due to inadequate blood supply, that has resulted in all of the following evidence of acute myocardial infarction:

• New characteristic electrocardiographic changes.• The characteristic rise of cardiac enzymes or

troponins recorded at the following levels or higher;– Troponin T > 1.0 ng/ml– AccuTnI > 0.5 ng/ml or equivalent threshold

with other troponin I methods.

The evidence must show a definite acute myocardial infarction.

For the above definition, the following are not covered:

• Other acute coronary syndromes including but not limited to angina.

Develop a definition aligned to the universal standard for myocardial infarction.

Stroke A cerebrovascular incident resulting in permanent neurological damage. Transient Ischaemic Attacks are specifically excluded.

Death of brain tissue due to inadequate blood supply or haemorrhage within the skull resulting in permanent neurological deficit with persisting clinical symptoms.

For the above definition, the following is not covered:

• Transient ischaemic attack.• Traumatic injury to brain tissue or blood vessels.

Develop a definition that would extend cover to include all strokes.

4 | Hannover Re UK Life Branch

Applying the diagnostic criteria within the universal definition used in current clinical practice to the pre-2006 ABI definition of heart attack would, in effect, lead to everyone who is diagnosed as having had a heart attack being paid. Likewise, all customers diagnosed with a stroke, irrespective of severity, would have been paid under pre-2006 ABI stroke definitions.

Going forward, aligning the definitions with clinical practice will provide consumers with a much greater degree of consistency between what they are told by medical professionals and the claim outcome on their CI policy. Such an approach is likely to lead to an increase in risk cost due to the increase in expected claims paid. Offering this wider cover at full benefit level is one option; but a cheaper alternative would be to offer wider cover only as a partial payment. Whilst this in itself could be seen as adding complexity, it would provide the consumer with the certainty of receiving a claim payment when they expect it.

In SummaryIf we align our definition wordings to clinical diagnostic practices, the technical wordings within the policy terms and conditions should not be so contentious at the point of claim. If we get this right, then adding a promise to update our definitions to match clinical practice at point of claim would further reinforce trust in the product, confidence in its benefit, and more importantly, meet customer expectations.

If you would like to discuss this article in greater depth, please contact us at [email protected].

Julie HopkinsHead of Underwriting & Claims Strategy

Blair SieveringHead of Product Development

Source: The Syndicate/Defaqto IFA survey, June 2010 and ABI

Going forward, aligning the definitions with clinical practice will provide consumers with a much greater degree of consistency between what they are told by medical professionals and the claim outcome on their CI policy.

All heart attacks

Heart attacks meetingcurrent criteria

All strokes

Strokes with persistingclinical symptoms

Hannover Re UK Life Branch | 5

The Continuous Mortality Investigation (CMI) Critical Illness Committee has been collecting experience data since 2000 and publishing results since March 2003. The Committee’s last published results appeared back in December 2009 – covering the 2003 to 2006 quadrennia.

There has been a delay in producing further results due to complications with the move to per-policy data, although the CMI CI Committee has continued to keep busy, and in 2011 produced a suite of interesting working papers detailed below:

• January 2011: Working Paper 50 presented some base incidence rates for accelerated CI business (referred to as the AC04 rates)

• June 2011: Working Paper 52 provided a breakdown of those rates into cause-specific rates

• December 2011: Working Paper 58 looked at various features of the experience compared to the derived rates and offered a brief consideration of the standalone experience

The CMI are hoping to publish new results later this year. In the meantime, this article focuses on taking a fresh look at some of the results from the data available in Working Paper 58, but from a slightly different angle in order to see if we can extract any additional conclusions.

Graph 1: Assumed Select Patterns within AC04 Rates

The Continuous Mortality Investigation (CMI) has been analysing data on mortality and morbidity risks arising under life assurance, annuity and pension business for nearly 90 years

Select PatternsThe AC04 rates were derived in a fairly pragmatic way by simply adjusting the existing CIBT02 base table in line with the ratio of actual-to-expected experience based on the existing table. This adjustment varied by age for each sex and smoker status. The select adjustments used were not age specific, but were instead derived for each sex and smoker combination. These are illustrated in Graph 1.

Dur 0 / 5+ Dur 1 / 5+ Dur 2 / 5+ Dur 3 / 5+ Dur 4 / 5+ Dur 5+ / 5+

MNS MS FNS FS0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Insights into Critical Illness Experience

6 | Hannover Re UK Life Branch

Overall the fitted rates aligned well to the observed experience from 2003 to 2006. For non-smokers this suggests that the select effect is quite marked in the first year after underwriting, before reducing quite considerably in the second year. It then remains fairly constant before taking a small jump to the ultimate level. For smokers, a less pronounced initial select effect comes through, together with a more rapid blending to the ultimate rates.

Working Paper 58 compared the 2003 to 2006 experience with these fitted rates, split by various features, with the results generally shown by sex, smoker status, age band and duration.

The results by duration effectively showed how the actual select experience compared to the expected select pattern in the base rates. This was useful in assessing the suitability of the expected rates, but did not make it easy to get a feel for the actual select experience. We have therefore adjusted the results by duration, allowing for the expected select pattern, in order to show the actual select pattern in the experience (i.e. as if comparing all experience to the ultimate rates).

Product TypeGraph 2 shows the 2003 to 2006 experience, split by duration and product type, compared to the AC04 ultimate rates. This highlights the large select effect in Duration 0 for the term business – together with the wearing off of the more subtle select effects in Durations 1 to 4.

What really stands out, however, is the hugely anti-selective effects seen in the endowment and whole of life experience (the actual figures at duration 0 are way off the scale shown). Although the volumes of this business are much lower, and the numbers of early-duration claims are pretty small during 2003 to 2006, the results are still highly statistically significant.

Sum AssuredThe CMI paper split the results into three basic sum assured bands (based on current sum assured rather than original): up to £40k, £40K to £80k and over £80k.

Interestingly, the overall results suggest that the smallest band appears to have the lightest experience. There then appears to

0 1 2 3 4 5+

Decreasing Term Level Term Unclassified Term Endowment WoL0%

20%

40%

60%

80%

100%

120%

140%

Graph 2: Experience by Product Type & Duration (compared to AC04 Ultimate Rates)

Hannover Re UK Life Branch | 7

they produced some implied standalone rates by deducting the death-only element of the accelerated rates from the total (Working Paper 52 had already broken the AC04 rates down into the main causes of claim, including deaths).

Comparing the actual standalone experience with these implied rates showed that the experience was much higher than might have been expected: around 13% higher on average. Although this approach does not constitute a direct comparison between standalone and accelerated experience, it does provide a good proxy. These results suggest significantly higher experience for standalone compared with accelerated critical illness business.

Multivariate ModellingWorking Paper 58 also described some multivariate analyses carried out using generalised linear modelling (GLM) techniques. Having reviewed these, we looked to see whether we could take this any further to consider some of the other features ourselves.

Using the published All-Office results, we were able to split the exposure and actual claims for 2003 to 2006 by two sexes, two smoker statuses, six durations, five age bands and two product types (i.e. accelerated and standalone). We excluded

be a ‘hump’ in the middle band, before the experience comes back down again for the highest band. This is contrary to the decreasing pattern by sum assured band which we might traditionally expect for mortality-only business.

It is interesting to look at these results by duration compared with the ultimate rates, to reflect the actual select patterns (See Graph 3).

Graph 3 clearly suggests that any difference in experience between the bands comes in after the initial select effect has worn off. This makes perfect sense. We would hope that the initial underwriting would be a reasonable leveller in terms of risk, and that any impact from socio-economic factors (for which the sum assured is essentially a proxy) would begin to have more of an effect as we move further away from that initial selection.

StandaloneThe AC04 base rates used in the above comparisons were derived using accelerated critical illness experience, including deaths. There was much less data available for standalone critical illness and so the CI Committee did not attempt to fit any rates to this business. Instead, in Working Paper 58,

£0 - £40,000 £40,001 - £80,000 £80,001+0%

20%

40%

60%

80%

100%

120%

0 1 2 3 4 5+

Graph 3: Experience by Sum Assured Band & Duration (compared to AC04 Ultimate Rates)

www.hannoverlifere.co.uk

death claims from the accelerated results to compare purely the critical illness element of the accelerated experience with the standalone.

We then modelled this quite simply using a Poisson model with a log link function, with the log of the exposure included as an offset, i.e.:

In(Actual Claims) = In(Exposure) + Σβ·Χ + ε

Or in other words:

Actual Claims = eΣβ·Χ + ε

Exposure

Using a forward stepwise process with minimisation of the Akaike Information Criteria (AIC) as our fitting criteria, the initial results provided no surprises. Age was the most significant of our five factors, followed by smoker status. Then came age and smoker status combined, suggesting that having different smoker differentials by age was the next most important consideration.

The surprising result was that product type (i.e. standalone or accelerated) was the next most significant factor. This suggests that the product type is more significant in explaining differences in experience than sex and duration.

Adding in sex, and then differentials for sex varying by age and then by smoker status produced results that were broadly as expected. Finally, duration came into the model in various combinations with the other factors until we ended up with an optimal model as follows:

[Age * Sex * Smoker] + [Product] + [Age * Sex * Duration] + [Smoker * Duration]

This suggested that, as well as being a more significant factor than both sex and duration, a single constant differential may be all that is needed to adequately reflect the effect of product type on the experience. The differential calculated was 119%, suggesting that overall – even allowing for different mixes in the other factors – standalone experience is almost 20% worse than accelerated.

Future AnalysesThe CMI has already revealed the revised format that the future All-Office results will take once these are released. The underlying data will be provided split by sales channel and detailed product type as well as by sex and smoker status. With individual ages, and individual durations up to 25+, we should have the data to enable much more detailed analyses following the format described, which in turn will help ensure we reflect the most appropriate features of the experience in our pricing bases.

Brian SewellManager, Actuarial Pricing

Source: Continuous Mortality Investigation (CMI)

Contact [email protected]