Professional Documents
Culture Documents
However, compared to the benefits that the hospitals thus saving a couple of days
are evident, the negative factors in the payment cycle.
overweigh the positives. Most of the hospitals
The issue of negotiated rates for TPAs still
So, insurers can either dispense with this
do not have a remains open though the PPN concept
requirement or look at published rate chart was targeted at achieving package rates.
For the negotiated/package rate scheme
• Issuing the policy itself in the card and in its absence, it is to become a success, insurers/TPAs should
form
hard to make a look at
• Or incorporate a tear away portion in
comparison. We have • Restricting the number of network
the policy itself (like the frequent flyer
hospitals so that such hospitals get
membership cards of airlines) to accept the fact that
increased footfalls
• Or like any other policy, let the the hospitals are free
• Agree to a volume based discount
customer produce the policy copy to provide services structure with the hospitals so that
along with any other photo ID (in
according to their hospitals also see a business sense
these days of ubiquitous ID
requirements at all places, all the perception of the • Get the powers to direct traffic to
customers would have one such ID select hospitals based on published
proof handy). paying capacity of the success rates of treatment by such
20
However, the TPAs face a huge employee Fraud management: • Software development with rule
turnover which adversely impacts their engines for automated claims
Fraud management is altogether a
efforts on points 2, 3 and 4. There is one processing, outlier detection and
different stream and to expect the TPAs to
more challenge of training their online access for customers for
discharge this function too is stretching it
employees because of the variety of viewing claims status.
too far. Moreover, there is inherent conflict
policies being marketed by the insurers.
of interest in clubbing together claims • Increased availability of network
As of today, there are at least 30 insurers
management and frauds management. hospitals at all nooks and corners.
(both Life and Non-life) who sell Health
In markets abroad, fraud management is
insurance policies. This means there are at One could argue that much more could
a specialist function entrusted to
least 30 types of policies, not to speak of have been done in the past 10 years of
specialists. Till recently, at least one private
the multiple policies marketed by each TPAs existence. But it is always better late
insurer had a special investigation unit,
insurer. Though all such policies are built than never. The present trend towards
headed by an ex CBI officer to fight frauds.
on the standard Mediclaim platform, there having the claims management in-house
are slight differences from policy to policy. Conclusion: might result in reduced business for the
What is excluded for 2 years under one existing TPAs but the targets to be
Overall, the introduction of the TPAs has
insurer’s policy might be excluded for 4 achieved remain the same for TPAs as well
resulted in
years under another policy. In the as in-house claims management teams.
insurers’ quest for innovation, there could • Improved Turnaround times in claims Let us hope that the ultimate beneficiary
be many other such small variations. processing. is the customer and we would see the
Many insurers just send the policy • Making cashless treatment a reality, reduction in loss ratios being passed on to
wordings to the TPAs and it is left to the though only 50-60 per cent of the the customers in the form of increased
TPAs to understand and interpret the claims are cashless claims. benefits at lesser premium.
policy. Won’t be surprised, if the number
• Using medical professionals for claims
of interpretations is more because there
processing
are 29 TPAs operating today and each is
entitled to at least one interpretation –
one only wishes that at least within the
TPA, there is one uniform interpretation!
Though all such
Thanks to IRDA, there are some
standardization initiatives like uniform policies are built on
pre-existing condition definition, critical the standard
illness definition and non-medical
Mediclaim platform,
expense heads definition. The recent
exposure draft talks of even standard two there are slight
year exclusions. These are steps in the differences from
right direction and my suggestion would
be to look at having standard Health
policy to policy. What
policy clauses on the lines of Marine is excluded for 2 years
insurance. The additional
under one insurer’s
irda journal January 2013
21
issue focus
Harvey Lee exhorts the importance of prioritizing the precious resources so that a proper balance of
various functions is achieved, which would eventually result in management efficiency.
I
n an insurance industry where use existing resources more effectively to structure resembling a project plan.
pressure on both the top line and the reduce claims losses. By achieving higher Like any system that prioritises some
bottom line are driven simultaneously, efficiencies, insurers should be able to things over others, there are winners and
it is essential to understand the reduce both their claims losses and losers – some casualties will be assigned
consequences of only focusing on expenses simultaneously. treatment by the best burns specialist, put
pushing the top line. Unlike a physical on the ventilation machine, sent to
Origins of Triage
goods business which presumably makes hospital in the first ambulance, etc. while
a margin on each sale, indiscriminately The word triage (pronounced tree-AHJ) others are left untreated until those same
pushing the top line in insurance – e.g. stems from the French verb trier, meaning resources are freed up. However morbidly
selling more policies at cut prices, with lax to sort. It was first used by French doctors unfair that may seem, the fact is that
underwriting discipline – will lead to in World War I and refers to a system of when resources are constrained, it can be
disaster for the bottom line, as the prioritising medical patients in order to impossible to treat all of the individual
resulting claims losses and expenses achieve optimal overall outcomes given casualties being faced simultaneously. In
outweigh the insufficient premiums. On resource constraints. For example, in a these sorts of situations, any approach
the other hand, as long as customer situation involving a high number of would result in an implicit prioritisation of
service levels are maintained, reducing casualties such as a train wreck, the some casualties over others.
costs to boost the bottom line generally availability of vital resources such as time,
leaves the top line unscathed. However, doctors, life-support devices and Evolution of Medical Triage
tinkering with the two variable claims transportation to hospital may be A triage team – usually a team of
costs (expenses and claims losses) can be overwhelmed by the sheer number of paramedics – makes sorting assessments
like running back-and-forth on a seesaw; casualties that require treatment. A based on both the current status of the
one month you are squeezing down system of triage can be used to sort the casualty (e.g. non-life-threatening
claims leakages by increasing the level of casualties and assign resources to treating injuries, life-threatening injuries or
scrutiny, and the next you are trying to them in order to maximise the overall deceased) and a prediction of the
reduce the headcount and time required outcomes, with the side benefit of incremental benefit of treatment (i.e. how
to process a single claim. It can seem organising the treatment effort into a
despairingly like a zero-sum game: either
your claims ratio is down but your
expense ratio is high, or vice versa. Now
that the Indian general insurance industry
has endured several years of “land
irda journal January 2013
22
much is prompt treatment likely to help in • If a claim is made within 2 weeks of
improving the outlook). Each component policy initiation and there is a break in
Even if the term
presents very clear limitations. coverage, send it to the special
‘triage’ is new to the investigation unit (SIU).
Firstly, the triage needs to be performed
quickly and systematically to direct vocabulary, most However, like simple forms of medical
immediate aid to casualties – implicitly, triage, business rules like these have
claims departments
there is little time to gather data and limitations based on the coarseness of the
perform a complete diagnosis on each already have simple rule. It is easy to see the inefficiencies that
casualty. Over time and repeated triage forms of triage remain, or are even created by these sorts
efforts, medical triage systems have of business rules: perhaps there are a
evolved standardised diagnostic instituted in terms of huge number of `5,000 claims on perils
processes that paramedics can perform coarse business rules. not covered; or perhaps the referral
on each casualty in less than 60 seconds to threshold of `30,000 is too low and leads
collect an informative set of data on which to excessive workloads for in-house
to base triage decisions. The key is to surveyors; and if only 1% of break-in-
collect the relevant data – typically, bottom line (both losses and handling), an coverage claims are repudiated then 99%
medical signs such as blood pressure, insurer needs to reduce the cost of the of the time the SIU is not doing productive
pulse rate, respiratory signs and sources of trade-off required between handling work. Whilst claims triage cannot
bleeding – no more and no less. expenses and claims losses. That is, for the necessarily help a special investigation
same amount of spending on handling unit (SIU) do a better job of identifying and
Secondly, triage involves a prediction of
expenses, how can an insurer reduce uncovering evidence of fraud, it can
future outcomes – these predictions can
claims losses; or, how can an insurer certainly reduce the number of specious
never be 100 per cent accurate.
reduce the amount of spending on claims that are referred to it, which also
Historically, these predictions were very
handling expenses required to achieve relieves the workload and enables
subjective in nature. Some guesswork was
the same level of claims leakages (e.g. investigators to spend more time on each
required from paramedics who may or
fraudulent claims, inflated claims, claims case or look into other suspicious cases.
may not have had substantive relevant
on perils not covered)?
experience. For example, even an Additionally, it is probable that the rules
experienced paramedic may find it Like medical practitioners in medical were originally created based on high-
difficult to differentiate between a chest triage situations, a claims department level, general observations and almost
wound that is repairable and one that is sees many problems simultaneously and never updated. Again, like medical triage,
beyond help. An inexperienced paramedic has limited resources to solve them. Large insurance claims triage of today can be
will have even more trouble claims and small claims, own damage and improved by basing the sorting criteria on
differentiating. This guesswork often led third party claims, own administered and a scientific, objective model backed by
to sub-optimal results: some casualties third party administered claims, empirical data.
that would really benefit from immediate reimbursement and cashless claims.
Better Triage for Insurance Claims
treatment are left waiting, others that are Given the limited resources available, a
beyond help are fast-tracked to hospital, claims triage process can optimise the A true improvement to the status quo of
and others still with only minor injuries effort in resolving all of the claims in the existing claims triage systems would:
are assigned scarce resources same way triage can maximise outcomes
1. Increase the detail of the sorting to
unnecessarily. To improve the accuracy of in medical treatment scenarios.
improve accuracy using more data
the sorting process, medical triage
Even if the term ‘triage’ is new to the and more sophisticated analytics
systems have recently started using
vocabulary, most claims departments
quantitative, empirical results to inform 2. Base the sorting on an evolving blend
already have simple forms of triage
sorting decisions and reduce the amount of empirical data and general
instituted in terms of coarse business
of guesswork required. Modern medical experience
rules such as:
irda journal January 2013
23
issue focus
for a hypothetical healthcare claims predictive modelling techniques – should be referred to the SIU. Claims that
department might look like. i n c l u d i n g re g re s s i o n m o d e l l i n g , score highly on the “pay immediately”
classification trees, neural nets and model should be fast-tracked for
Increase Detail
nearest neighbour clustering – eventually payment. This sorting process for claims
The more detail that is considered by a boil down to a scoring model, but the ease based on scores is directly analogous to
triage process, the finer and more of use, sophistication, accuracy and the sorting done in medical triage that
accurate the triage can be, leading to amount of subjective human input determines whether a patient should be
more appropriate treatment being required to perform the analysis will vary. treated now or later, and what resources
applied and improving efficiency. In an Scoring models can be created to rate any to assign to treating him or her.
advanced claims triage process, the aspect of interest of a claim, but some
Base Decisions on Both Empirical Data and
sorting would be based on many more likely ones are:
General Experience
factors than just the size of the claim and
• Likelihood that a claim is fraudulent
whether there was a break in coverage. Traditional claims triaging efforts are
• Likelihood that the best thing the unstructured and generally based solely
Some other factors to consider might
insurer can do is to pay this claim on the experiences of a handful of
include:
immediately operations managers making decisions
• Was the treatment conducted in an that can be improved by increasing the
• L i ke l i h o o d t h a t a c l a i m w i l l
“in-network” hospital? Were the level of detail – if a manager sees a trend
significantly increase in severity (or
“preferred” rates charged or are they of inflated claims from out-of-network
lead to more claims)
inflated? hospitals, a too-common response would
• Deviation from an established be to apply additional scrutiny to all out-
• How many claims per month are
benchmark for a claim of this type of-network hospital claims. For the most
received involving treatment by this
part, this is a grouping error that is solved
doctor? Is this number suspiciously • Estimated benefit of handling
by sorting with increased detail, but the
high? internally versus assigning to a third
whole effort can still be left unstructured
party administrator
• What is the probability that the and likely to either stagnate into
diagnosis will lead to more claims in For example, a high score on the fraud irrelevance (the managers who set the
the future? Does the reserve need to model could indicate that there is a high rules move on to other roles) or fluctuate
be adjusted? likelihood that a claim is fraudulent. Some wildly (the rule setters react to every new
claims may score highly on multiple event). Systematically incorporating
Even if these factors seem obvious, they
models, such as an excessive claim that is empirical data into decision-making can
must be combined in an appropriate
also likely to be fraudulent. help regulate the effort to maintain an
manner to be helpful – otherwise, too
effective claims triage process.
much labour will be allocated to perform The process workflow for a claim can be
claim edits, any doctor being too influenced or even completely Fast Track Model
hardworking will be treated as a fraudster, determined by the scores given to the
Recall the scoring models introduced
and solvency requirements will be sent claim, depending on the appetite and
previously. How should the relevant
sky-high with excessive reserves. Most requirement for human oversight. Claims
factors and the assigned scores for each
doctors involved frequently in claims are that score highly on the fraud model
model be determined? Take the example
not trying to defraud the insurer – the goal
of a fast-track model: it would be possible
is to sort the ones that are from the ones
for an experienced claims manager to list
that are not, using more information than
a handful of attributes that would indicate
just a single attribute (i.e. the number of
recent claims). The key is to take an overall
Can the claims claims which should be fast-tracked, such
as low claim amount or where complete
view of the attributes of the claim in manager be sure that
supporting documents were submitted at
combination to determine how to process
all the relevant factors intimation. However, can the claims
it.
irda journal January 2013
24
empirical data – observations of objective The pitfalls of incorporating empirical
facts attributed to relevant claims. data into claims triage can be mitigated
Fraud Model Experienced managers by understanding the limitations of what
statistical analysis can and cannot do.
For claims triage, one useful form of are still critical for
Experienced managers are still critical for
empirical data is a dataset of historical claims triage, but the claims triage, but the nature of their roles
claims describing the attributes of each nature of their roles changes from being the traffic controllers
claim (e.g. policy details, treating hospital to maintaining and optimising the traffic
changes from being
and doctor, diagnosis, treatment, claim lights.
amount, etc.) matched with the outcome the traffic controllers
of interest. For example, for a fraud model, to maintaining and Reduce Overall Cost
the attributes of each claim should be optimising the traffic The point of a triage process is to assign
matched with whether or not the claim resources efficiently and to organise the
lights.
was found to be fraudulent. Running a process. Assuming that financial cost is a
statistical modelling analysis on this data consideration of the claims handling
can then highlight which factors are process, then the cost of actually
relevant to detecting fraud, and to what performing triage must be factored into
extent. The output of the analysis on the people are not particularly adept at any decisions aimed at reducing the
historical data is a predictive statistical understanding and acting on probabilistic overall cost of claims. Thus, it is important
model – a scoring model where the future outcomes. An everyday example is to consider how to manage the cost of
factors and scores are “trained” to the when we curse the weather bureau for claims triage so that a net benefit can be
historical patterns. When this scoring making us pack our umbrellas and gained.
model is applied to new claims, the raincoats because they announced a 90%
resulting scores will represent the chance of rain, even though the day turns Like medical triage, the sorting process for
statistically best prediction of whether a out to be beautifully sunny – on average, claims triage needs to be fast and not
claim is fraudulent. In other words, the this will happen 1 in 10 times! In this case, resource intensive. Unlike medical triage,
model extrapolates the historical patterns packing an umbrella still represents a which generally deals with limited data
into the future, setting an expectation of logical decision, but the mind has (basically restricted to what a paramedic
events to come. “rounded up” the expectation of rain to can diagnose and measure within 60
100% without justification. seconds or so), the significantly larger
If a statistical analysis exercise is regularly timescales and reduced urgency of claims
conducted using “fresh” historical data – Another caveat that affects all forms of handling means that much more
for example, an analysis every six months statistical analysis is that models based on complete data can be collected and
on the latest two years of data – then historical data can only discover “more of considered, and more data can be
models can kept up-to-date without the same.” That is, they can only predict generated or collected over the course of
overreacting to fads. Building a claims future events if similar events were the handling process – patient reports,
triage process around objective scoring detected in the past. For example, if an claim histories, benchmark comparisons,
models that are updated regularly will insurer has never identified any claims etc. The greater amount of detail in the
institutionalise accurate triaging and keep t h a t p re s e n t e d o p p o r t u n i t y fo r data implies a higher processing cost –
the claims handling process running subrogation, then statistical analysis on how can this be managed?
efficiently. historical data will not be able to predict
future subrogation opportunities. Greater Detail, Greater Processing Cost?
Caveats of Statistical Analysis
Statistical predictive modelling can only
A trained triage paramedic arriving at the
extrapolate the known patterns of the
irda journal January 2013
Whilst statistical analysis and predictive scene of a building fire has a lot of
modelling represent a scientific gold past into the future. As a consequence,
information to digest at once – the
standard, there are some caveats that even if a claims triage process were to be
number and status of casualties, the
should be kept in mind. Statistical analysis informed by statistical analysis, there is
safety of the situation, the resources
is not a silver bullet that can describe the still a need for human inspection to
available for administering treatment etc.
future with 100% certainty and in general, identify novel trends that fall outside the
The training that a triage paramedic has
historical experience.
25