home email us! sindicaci;ón

Archive for Variability in PHM Series

The Name of the Game in PHM Is Variability: Part 5 - Recruitment

by Scott MacStravic

Once members of the population have been targeted for participation in particular PHM interventions, the next step is to recruit as many as possible, or at least as many as will contributable value per participant, to each intervention.  Such recruitment may actually be part of the assessment process, when members of the population have to be persuaded to engage in a health risk assessment or biometric screening process, for example.  In any case, this element is yet another in which there is wide variability across the methods used by different DIY payers or the suppliers they hire.

One of the simplest is the “opt-out” strategy, in which every member of the population targeted is automatically enrolled without their involvement, once the targeting process is complete.  They are usually individually notified of their enrollment, and invited to take whatever the first step in actual participation is.  In some cases, they are automatically assigned a coach who will call them, or are automatically sent online or mailed information that is part of the intervention process.

This method often results in 90-95% enrollment, since only those who explicitly opt out are deemed non-participants.  But the significance of their being enrolled lies in their active participation, cooperation, and changes in behavior, and the rates of each may be quite a bit lower than when other enrollment strategies are followed.  The major two options with “opt-in” interventions are: 1) targeting each for a particular intervention based on the identified most promising PHM intervention for each; or less common 2) asking each target to choose among a number of options based on recommendations customized to each.

The enrollment process has to balance two considerations: 1) which intervention promises the greatest return to the investor; and 2) which has the greatest likelihood of delivering on that promise.   The opt-out approach probably is strongest on the first criterion, and weakest on the second.  Inviting targets to enroll in an intervention pre-selected for each is probably medium on the first and on the second criteria.  The “self-determination” approach would probably be weakest on the first and strongest on the second.  So it is a balancing act to select the best approach, and one or two may be tried before the best is identified in each case.

The “style” of the intervention and enrollment effort may also make a major difference to the numbers who enroll, participate effectively, and succeed.  Early PHM interventions were mostly “one-size-fits-all designed and implemented by the sponsor or supplier.  Increasingly, however, interventions are either partly or totally individualized, by either the person who contacts the target, or the automated online/mail communications based on analysis of assessment information.  And the customized approaches tend to work far better.

Enrollment is also influenced by whether or not financial or other incentives are offered – for enrollment, participation, or success.  Incentives paid for enrollment should achieve the highest enrollment, but not necessarily the highest levels of enthusiastic participation.  Incentives paid for participation in or completion of an intervention may achieve high levels of participation, but not necessarily of success.  Incentives paid for success may promote both enrollment and participation, particularly among those with the greatest confidence in their probability of succeeding.

Of course, people often delude themselves.  Repeated studies have found, for example, that those with the highest levels of capabilities tend to underrate themselves, while those with the lowest levels overrate themselves.  If the assessment process identifies targets based on predicted probability of success, it should be safer to pay incentives for participation, since those with low probability, despite confidence therein, would not be targeted.  Those with high probability, but perhaps less confidence, would be encouraged to participate by the incentive, where they might not be as encouraged by a reward only for success.

Since incentives add to the costs incurred by insurers or employers, they have to be used carefully.  PHM suppliers or peers who have used incentives, if they share their experience, may provide a basis for choosing which type and amount of incentive may work best, though each population is likely to be unique enough to make actual experience with it the best guide in the long run.  The combination of customization and the right incentives is likely to work best, judging by experience in PHM so far.

The Name of the Game in PHM Is Variability: Part 4 - Targeting

by Scott MacStravic

The first purpose of the assessment process is to determine the size of the PHM challenge and of potential gains from implementing a PHM strategy.  The assessment process is then also used in the evaluation process, to identify changes in measured “success” dimensions against the baseline data.  But another key use of the data is to identify and select which members of the population make the most promising targets for participation in which PHM intervention, together with how many available interventions will likely justify their investment.

This step begins with identifying the baseline costs, both direct and indirect when applied to employee populations, linked to individuals and the health-related factors each has been found to have.  Usually, without extensive multivariate analysis and predictive modeling, the “potential” is estimated based solely on the baseline costs.  But the real predictor of potential economic gains lies in the particular PHM interventions that will be applied on a DIY or outsourced basis.  And this information is peculiar to each PHM strategy, intervention, and supplier, rather than any generalizable average for PHM as a whole.

Moreover, when measurement problems such as side-by-side comparison “self-selection bias” and before/after “regression to the mean” effects have been overcome, there remains the inherent difficulty of predicting realistic potential cost impacts or other benefits based on past data.  While past data may accurately describe the past of the very population to be addressed by PHM, it does not translate easily into predictions of the future.  And while a given PHM supplier’s own past performance may be “statistically valid and reliable”, it may not be a sound basis for accurate prediction of results in a new population.

Within these limits, PHM investors and suppliers will want to determine which and how many population members represent the best risk/reward potential for which particular PHM interventions.  Some suppliers and employers have employed a completely customized approach to PHM, where each individual either chooses each’s own goals or is assigned a coach who will customize the intervention to the individual’s mix of challenges, and address as many of each’s individual problems as possible.  Experience suggests that individuals can handle more than one, but rarely more than two or three, so the number of problems to be addressed is normally small.

In many cases, there is one specific problem that is the basis for the PHM intervention to which members of the population will be invited.  This does not mean that the intervention will have a very narrow focus, however, though it may.  A smoking cessation program normally focuses exclusively on enabling participants to quit, but may have to address a wide range of perceived barriers and related issues, including stress and weight management along with the smoking behavior, per se.  And diabetes disease management interventions routinely include attention to blood pressure and cholesterol levels, in addition to blood glucose.

While the baseline costs linked to individuals may guide targeting to some extent, the psychographic data indicating the probability that each will make a needed change and achieve success thereby, together with PHM suppliers’ demonstrated performance will add significantly to how “informed” the choice of targets can become.  In many cases, the supplier may guarantee results for particular population segments, interventions, or success dimensions, making the targeting that much better informed.

Targeting necessarily includes consideration of how PHM suppliers charge, and what additional costs may be incurred by targeting more and particular segments or individuals for interventions.  When the supplier charges on a “per population” of flat fee basis, it will not add fee costs to include as many members of the population in any given intervention, though most charge a fee for each separate intervention, which will multiply costs by the number of interventions selected, rather than the number of members participating.

When the supplier charges per participant, perhaps with graduated fees based on the risk/reward potential it determines for each and the different intensity/expense for the graduated interventions, it is the number of members targeted for each intervention that will represent the upper limit on fees.  Moreover, payers will likely incur additional costs, per member, per participant, or even per successful participant (e.g. when incentives are offered for success rather than participation alone), which have to be considered as well.  And while payers may recommend steps their clients should take that incur costs, the client will have the last word on which kinds of support it provides and what it will pay for what.

There will always be the unknown costs of participation or success incentives that will have to be paid, but this will typically be substantially less than the economic gains associated therewith.  The past performance of the supplier applied to the baseline costs and economic value of members of the actual population to be addressed should help in ensuring that the client’s total costs are kept lower than the client’s benefits. Controlling the number of members targeted and number of interventions invested in based on risk/reward vs. predicted or controllable costs, is the best overall approach to take.

The Name of the Game in PHM is Variability: Part 3 - Assessment

by Scott MacStravic

The sheer number of dimensions of the PHM challenge that are to be measured partly determines the relative difficulty, uncertainty, and costs involved, but the types of dimensions makes an even greater difference. One fairly common practice, however, is to count only the numbers of health risks, chronic conditions, or impairment causes, and link the number of such “factors” to the amounts of sickness care, disability and workers compensation (‘direct”) costs, and productivity or performance impairment (“indirect”) costs, for each individual in the population. This greatly simplifies the measurement challenge, but it may also mislead it.

The difficulty with assessing the specific factors of concern in PHM is that it is normally only possible to identify the overall direct and indirect costs that apply to each individual, together with the presence/absence or degree of the separate factors for each. This means that there is no basis for determining how much each individual factor contributes to the total costs for each, nor predicting how much might be saved through a particular PHM intervention addressing each. In practice, PHM interventions are customized to individuals, at least by what is usually one particular factor affecting each, though potentially for a number of such factors simultaneously, depending on the types of interventions available.

There are four common approaches to gauging individuals’ and populations’ risk/reward levels in terms of costs and economic gains: 1) past claims analysis, 2) health risk assessments); 3) biometric screenings; and 4) “psychographic” surveys. Champions for each tend to think theirs is the best approach, though there is likely to be added value in each of them that is not available in the others. The trouble is, however, that adding more methods adds to costs for insurers and employers, and to participating employees as well, which can directly threaten payers’ return on investment (ROI) by increasing the cost denominator. It can also indirectly threaten ROI by making participation either lower in numbers of people involved, reducing the economic gains achieved, or more expensive to achieve by requiring incentives to be paid, which will affect both numerator and denominator.

Claims analysis is often easiest, because the data already exist, in insurance plans or the employers’ own operational data. This will only apply to direct costs, unless the employer has a P4P system that involves already measuring employee productivity or performance in a way that can be applied to gauging their health-related impairment. Otherwise, only direct costs will be gauged by using existing data, and this will greatly understate the size of the problem and extent of economic benefit for employers, regardless of how much they pay their employees. The degree of understatement will be directly related to the amount of average compensation, with or without team/peer effect or value-based multipliers.

Claims analysis also carries with it the greatest risk of overestimating potential and actual economic gains. This is particularly true when individuals with high/outlier levels of sickness care expense in the baseline year will greatly influence the estimate of potential. Since such individuals are more likely to have lower costs in the following year, thanks to what is called “regression to the mean” than to have the same or higher costs in the next year, a major portion of the potential savings in the assessment, as well as the actual savings in the evaluation, will not be linked to the PHM intervention, but to unrelated “natural” causes.

There are half a dozen validated measures for employee productivity, which should also be fairly good estimates for performance, though I know of none validated for this added dimension of employee value. The book — R. Kessler & P. Stang (Eds), “Health & Work Productivity “, U. Chicago Press 2006 — offers the best compilation of these methods that I know of, and any one might be chosen for PHM use with some confidence. The trouble is that different methods tend to produce different results. Moreover, where comparisons have been made between employees’ self-reported impairment and objectively measured output, it has been found that employees often overstate their degree of impairment. So whichever method is used, there should be sound data available for converting employee reports into most probable reality.

In practice, many PHM suppliers use psychographic measures, along with, or even instead of the others. A few rely totally on individual’s self-reported perceptions and attitudes to predict their future costs, and the same may be used to predict their impairment. This can be inexpensive, where the survey is administered online, for example, and responses are automatically analyzed by computers. But the proof of all such assessments will be in the degree to which they lead to positive ROI based upon them. No merely “scientific” superiority should overcome pragmatic advantages.

Most suppliers who use employee surveys, whether health risk assessments or productivity/impairment questionnaires, can and usually do combine these with psychographic questions to gauge things such as employee motivation, commitment to change, confidence and self-efficacy in self-governance, for example. Answers to such questions, combined with the costs linked to each, can be used to identify and target the most promising individuals for inclusion in PHM interventions. Or they may be used to determine risk/reward segments, whose members will be assigned different levels of intensity and cost for the interventions they get.

There are equally wide variations on how identified risks are defined. “Depression”, for example, may be defined as a diagnosis, with only a small portion of the population identified as having a diagnosed “disease”, or as a “disorder” where a far larger portion is affected, or as an “emotional state” where an even larger portion would be identified as having it. It may even be defined as a general catchall covering all emotional problems and “down” feelings, including anxiety and stress, making it likely to affect the majority of the population.

The variety across internal PHM programs, as well as outsource suppliers, in the methods they employ to assess the baseline situation, which are typically repeated in the evaluation process, add greatly to the overall variability across the large numbers of options available to employers and insurers. We can hope that in the future, there will be improved identification of which are truly the “best practices” in PHM, though there is likely to be constant experimentation with new methods that will compete with even those proven best in the past.

The Name of the Game in PHM is Variability: Part 2 - Economic Impact

by Scott MacStravic

The first basic split among PHM clients in terms of which dimensions of economic impact will be used in assessing problems and evaluating gains is that between commercial and government insurers on the one hand, and employers on the other.  Insurers focus mainly, though not exclusively, on reductions in sickness care costs as the economic gains they are after, with many true “health” care costs part of the PHM intervention.  They may also look at member/beneficiary health status and satisfaction with their PHM experience, as well as the duration of their “membership” in specific insurance plans.

Employers tend to look at a far greater number of dimensions, though these include the same ones that insurers look at, since they pay for such insurance, or if self-insured, pay the costs directly.  But in addition to sickness care costs, employers worry about workers compensation and short- plus long-term disability (STD and LTD) insurance or direct costs as well.  These may be as great as are insured sickness care costs, with some employees, though they are normally far less overall.

The really large economic benefits sought in PHM by employers tend to be that derived from reductions in absences and both productivity and performance impairment while at work (“presenteeism”).  Economic losses from lost productivity/performance has been measured at levels two to five times as great as the sickness care costs associated with the many health-related problems that cause impairment.  Moreover, employers may achieve economic gains through “positive presenteeism”, the potential that healthy and happy employees will contribute significantly more economic value than is “normal”, not merely “impaired” levels.

While productivity, per se, is the more frequently included dimension, compared to performance, there have been a few cases where specific benefits of improved performance have been addressed by employers.  These include employee impact on customer satisfaction and loyalty, word-of-mouth impact on new business, and similar revenue-related benefits, in addition to labor cost reductions.  And many employers have examined improvements in employee retention as adding value in the balanced scorecard dimension of “organizational knowledge” as well as reducing costs of replacing employees who leave.

The determination of which dimensions to include affects the complexity and costs of the PHM effort, whenever measuring these dimensions requires going beyond information already being collected as part of normal operations.  This is less often the case for insurers, since claims data is frequently all they look at, though if they also consider member health status, satisfaction, and retention, for example, special efforts and added expense may arise in collecting such information.

With employers, complexity and costs of measuring dimensions of the PHM problem and progress in its solution are likely to be significantly greater than for insurers (though insurers may have to address these when they provide PHM services for their employer clients).  Measuring workers compensation and disability expenditures is normally routine, already being tracked overall, though there may be added effort used in analyzing collected data so that it serves better to guide and evaluate the PHM effort, itself.

The biggest challenges will be in gauging productivity and especially performance impairment levels and their reduction, i.e. the economic benefits of improving either or both.  While productivity is susceptible to both objective measurement in some cases, and validated estimates via surveys that collect employees’ and sometimes team supervisors’ perceptions of output, performance includes multiple dimensions of its own.

Of course, some dimensions of performance, such as customer satisfaction, retention, market share, new business revenue, etc. are likely to be routinely measured.  But they are normally measured on an overall or market/segment-specific basis, rather than linked to individual or segments of employees, except for sales results linked to specific sales staff members.  Customer satisfaction may be linked to the specific employees known to have served them, where this is likely to be the same individual or team over time, but market share and new business increases may have to be credited more widely, perhaps to the PHM effort overall.

There are probably a half dozen commonly used methods for estimating productivity based on employee self-reported data.  These may directly ask employees to recall how many hours of lost productivity they have had in a recent period, perhaps a week, two weeks or a month, where their memory may be reliable.  Or it may ask for an estimate of the percentage of their full potential they have been able to deliver, considering their health and any problems identified, that may affect their output.

Similar surveys may be used to gauge the levels of performance employees have been able to deliver, since both absence and presenteeism will likely affect performance as much as they do productivity. And the impairment found in individuals may be “multiplied” by their estimated “team/peer impact”, which has been measured at between 1.25 and 1.35 times the impact of the individual’s absence on average. [“Multiplier Effect: The Financial Consequences of Worker Absences” Knowledge@Wharton Dec 14, 2005 (knowledge.wharton.upenn.edu)]

The economic impact of absent or impaired employees may also be based on the economic value of employees in terms of their overall contribution to the employers’ performance.  When employees are paid on a “pay-for-performance” basis, their performance will naturally be measured, though it may prove difficult to determine the extent to which any individual is impaired by their health problems, as opposed to other non-health-related limitations.  Fortunately, once performance improvements have been linked to PHM interventions, the measured performance may be more easily translated into economic benefit.

By far the most common approach currently used in evaluating the economic impact of employees’ health-related impairment is to multiply their degree of impairment times their annual compensation levels.  This means the different employees will have different costs of lost productivity, whenever they have different levels of compensation.  It also fails to take into account any team/peer impact or the likelihood that employees are worth more than they are paid.  One study, for example, found employees worth an average of 3.28 times their annual compensation, though the median was only 1.92 times. [P. Strassman “How Much Is an Employee Worth?”,  Jan 14, 2006]

While insurers have relatively few and commonly used dimensions when measuring their PHM problem/challenge and the results of their PHM investments, employers have far more complex and potentially costly choices.  Fortunately, employers can use the same relatively simple and inexpensive-to-measure dimensions as do insurers, and when they go beyond those, they should also be gaining many times the cost of doing so in the added value they discover across the full range of benefits in having a healthier/happier workforce.

The Name of the Game in PHM Is Variability: Part 1 - Introduction

by Scott MacStravic

Population Health Management (PHM) has become a major market for insurers, governments, and employers to invest in – and for a wide range of organizations, from traditional healthcare providers to specialized PHM suppliers to insurers to deliver to that market.  The numbers of suppliers keeps growing as the number of payer clients does also. The market is beginning to look like the early stages of many new markets, from automobiles to radios to TV sets, cell phones, and electronic gadgets in general – it offers a vast number of options to purchasers, including the “do-it-yourself” (DIY) alternative to purchasing PHM.

Many traditional healthcare organizations (HCOs), from physician practices to hospitals to integrated health systems (IHSs) have entered this market, as DIY providers to their own workforces, or as competitors in marketing PHM services to payers, mainly to employers.  They may do so in order to support their sickness care services marketing strategy by creating and sustaining stronger and more lasting relationships with employers as influencers of both insurance plan decisions on provider networks, and employees’ choices of providers.  Or they may see the PHM market as one they would rather join than merely suffer its sought-after effects in reducing sickness care use and expenditures and thereby their sickness care revenue.

Employers have been involved in a “toe-in-the-water” approach to PHM since the 1970s at least.  Many began with some kind of DIY “worksite wellness” program for their employees, and a growing number are offering onsite medical clinics that engage in PHM along with traditional sickness care.  But most appear to be outsourcing the PHM function to specialized suppliers or HCOs, since there are so many complications in their knowing about their individual employees’ health problems — and most have no great competence in PHM.

Insurers have often been influenced by their own interests or their employer clients’ demands to become PHM DIY suppliers, as well as insurers, to employers, at least.  Both Aetna and CIGNA, for example, offer PHM services to employers other than those to which they provide health insurance plans.  Since insurers can gain substantial experience, and with large populations, by delivering PHM services to their own insured populations, they can gain both the competence and track record needed to become viable suppliers in the overall market, as well as capable DIY providers for their own health plan members.

Specialized PHM suppliers have been in the PHM market, as operators of worksite medical care clinics for employers, or suppliers of PHM services for commercial and government insurers, for a few decades.  There has been a significant movement toward merger and acquisition among such suppliers, resulting in consolidation of their numbers, but many new ones keep entering the market, so the overall numbers of suppliers is still huge.  And while some have captured large market shares, there appears to be no single supplier who could be called dominant.

One of the reasons for this is the vast degree of variability among PHM suppliers, due to both their large numbers and the fact that there is nothing close to agreement on what are the “best practices”, the most cost-effective approaches to PHM in any one of its essential elements.  These elements include:

  1. Determination of what dimensions of economic impact will be included in the PHM planning, management, and evaluation
  2. Assessment of the PHM current state of these dimensions in whatever population (insured plan members, employees, dependents, retirees) is to be involved
  3. Targeting of which individuals shall be sought as participants in either standardized or differentiated PHM approaches to particular problems, risks, or potential gains are to be addressed
  4. Recruiting targeted individuals and segments to become active, engaged, and enthusiastic participants
  5. Sustaining participants throughout the PHM intervention program, at least long enough to achieve some desired effects, if not to the end of a limited intervention, or on a continuous basis if there is no intended end.
  6. Carrying out the intervention process on those participating, using a wide range of interaction and communications technologies and methods
  7. Evaluating the effects of interventions over whichever time frame(s) clients wish

Not only are there vastly different approaches to each of these elements being used by different PHM suppliers, but clients may engage different suppliers for two or more of such elements, including doing one or more themselves, or at least sharing the responsibility with outsource suppliers.  As a consequence of the numbers of and variations among suppliers, the overall PHM market contains far more variation than is true for almost any other example in the business-to-business (B2B) or business to consumer (B2C) markets.

In subsequent postings, I will describe and discuss the seven separable elements in PHM that are subject to such variations.  The combination of this number of basic elements, and the variability in how they are carried out by different suppliers, even within the same supplier for different clients, is what makes the PHM market so replete with variability.