|
ROI Calculator Methodology
This page describes the ROI Calculator methodology, data sources, and
limitations using non-technical language. This page also provides brief
descriptions for specific model inputs from the data entry screen. Users are
encouraged to adjust the model inputs using information for their population.
Users interested in further details about the methods can contact the ROI Team.
I. Methodology Overview
The ROI Calculator uses a longitudinal cohort approach to estimate the
return on investment for alternative system-level smoking cessation
interventions. This approach is used to track the experiences for a specific
group of individuals over time. For adult smokers enrolled in a health
insurance plan in year one, the model calculates the number of smokers
receiving cessation services, the number of self-quitters and program quitters,
and the costs to the health insurance plan for each intervention strategy. Each
intervention is assessed “as if” it is the only strategy available to smokers.
The ROI Calculator then follows the cohort of smokers and new quitters over a
five-year period.
For simplicity, the model assumes the insurance plan offers the interventions
for a single year, and that all smokers in year one will have access to the
service prior to any change in health, smoking, or eligibility. The model uses
annual probability estimates for a smoking-related disease (SRD) diagnosis,
spontaneous quitting and relapse, and disenrollment to move a large population
of smokers through the model over a five-year period.
The ROI Calculator models annual smoking-related disease (SRD) status, smoking
patterns, and health plan enrollment for a cohort of health plan members who
were smokers at the beginning of the study period. At the end of each year,
mean annual medical expenditures and smoking-attributable productivity losses
are estimated for individuals in each distinct SRD, quitting and relapse, and
plan disenrollment group, controlling for age, sex, and smoking intensity.
The cohort approach clarifies the future impacts of implementing a system-level
intervention for a known population. Assessing programs in real world
situations requires controlling for the smoking, disease, and disenrollment
patterns of new plan members, and for continuing the program over several
years. This added complexity would require additional computational resources
that, we believe, would not change the conclusions of the model.
The model compares to usual care (2 A's) the impact of medical expenditures and
the productivity of 5 A's (ask, advise, assess, assist, and arrange)
interventions, including medications and proactive telephone counseling. Annual
estimates over a five-year period are generated from both health insurance plan
and employer perspectives.
For each system-level intervention, the ROI Calculator estimates the number of
participants, new quitters, and costs for a one-year cessation program. The
model is rerun using the new distribution of smokers and quitters. Separate
probability and cost data are assigned to intervention and non-intervention
quitters. Incremental return on investment is calculated by estimating the
differences in program costs, annual medical care expenditures, and
productivity costs for each intervention compared to no intervention (usual
care). ROI per intervention participant is estimated from the health insurance
plan and employer perspectives. Annual and cumulative ROI, per health plan
member per month (PMPM), are estimated for health insurance plans and employers
combined. All costs are in 2002 dollars and have been discounted to the present
value using a cost-of-capital approach.
II. Components of the ROI Calculator
The ROI Calculator has three components. A population component uses
probability estimates of SRD incidence, quitting and relapse, and disenrollment
to move groups of individuals through each year of the model. A cost component
assigns mean annual medical expenditures and smoking-attributable productivity
losses to each group. An intervention component estimates each intervention’s
reach, efficacy, and one-year program costs. Each component is described below.
II.1. Population Component
The population component moves groups of individuals through each year of the
model based on their experiences with new smoking-related disease (SRD)
diagnoses, quitting or relapse, and plan disenrollment. The initial population
of current smokers is divided into subgroups based on sex, age group (18-34,
35-64, and 65 plus), heavy or light smoking, and SRD diagnosis. The subgroups
were used to predict the probability of a new SRD, quitting and future relapse
given a SRD diagnosis, and disenrollment given SRD and smoking status.
We used the following formula to simulate annual SRD incidence, quitting (and
relapse) by SRD status (prior year and current year), and disenrollment (by SRD
and quit/relapse status):
Popi x (pSRDi)
x (pQuit(Relapse)|SRDi) x (pLeavei|SRD
and Quit(Relapse))
where:
-
Popi
= the number of current smokers, new quitters, and future relapsers in each
population subgroup,
- pSRDi
= the probability of a smoking-related disease diagnosis during the year for
each subgroup,
- pQuit(Relapse) = the probability of quitting
(relapse) during the year, and pLeave is the probability of disenrollment given
SRD and quitting (relapse) during the year.
II.1.a. Population Characteristics
The ROI Calculator assesses the impact of cessation services provided to the
general population of insured smokers during primary care visits. The model
considers smokers’ age, sex, smoking intensity, and smoking-related disease
status (current and previous diagnoses). The model does not enable analyses of
interventions targeted to specific population subgroups, e.g., race, ethnicity,
income class, or smokers after a heart attack. We did not include income in the
ROI Calculator because other plans were not likely to have these data, and
initial analyses using geo-coded household incomes indicated that income data
did not change the model conclusions. Income did affect the probability and
expenditure estimates consistent with other studies. However, within each of
three income groups (low, middle, high), the relative affects of continued
smoking and quitting did not change substantially. Contact the ROI Team if you
are interested in conducting analyses for population subgroups.
We encourage you to enter data relevant for your own population. The age and
sex data are straightforward. We provide state and regional prevalence
estimates for users lacking accurate smoking rate data for their population.
The default smoking prevalence data were obtained from the CDC’s Behavioral
Risk Factor Surveillance System (BRFSS) (CDC, 2001-2003). BRFSS defines current
smokers as having smoked at least 100 cigarettes in their lifetime and who
currently smoke everyday or some days. The state estimates represent the
weighted average for 2001–03, and were calculated using the combined surveys
for these years. We combined estimates over three years to minimize the
year-to-year variations due to sampling error. You may be able to find data for
your population in medical records, ongoing member surveys, or from recent
research projects.
Regional prevalence estimates reflect the population-weighted average of the
relevant states. Regional estimates are provided as an aid to insurers and
employers operating in multiple states.
The model uses separate data for heavy (= pack per day) and light (< pack
per day) smokers. The ROI Calculator models heavy (= 1 pack per day) and light
smokers separately. Currently, users cannot change the distribution of heavy
and light smokers. The model splits smokers into heavy and light smokers based
on medical record analyses. For males, 36% of smokers ages 18-34, 58% of
smokers ages 35-64, 49% of smokers ages 65 and over are heavy smokers. For
women smokers, heavy smokers are 32% for ages 18-34, 51% for ages 35-64, and
48% for ages 65 and over.
II.1.b. Probability Data
The transition probabilities (i.e., pSRD, pQuit(Relapse), pLeave) were
calculated by estimating multivariate logistic regression models using EMR data
for about 200,000 adult Kaiser Permanente Northwest (KPNW) members with known
smoking status, who were enrolled from January 1, 1997 to December 31, 1998.
KPNW is a comprehensive HMO serving 440,000 members in the Portland,
OR—Vancouver, WA area. The ROI Calculator includes information for 37,450
members ages 18 and over who were documented in the EMR as current smokers on
January 1, 1998. Data were obtained and analyzed for each year, through 2002.
We used multivariate logistic regression to estimate the annual predicted
probabilities and 95% confidence intervals of a new SRD diagnosis, quitting,
relapse among new quitters, and disenrollment. Separate data were calculated
for each group based on sex, age group (18-34, 35-64, 65+), heavy or light
smoker, year since quit, year since relapse, and SRD diagnosis in previous
years. Chi-square tests indicated differences in the parameter estimates
between population subgroups and were significant at the p<.01 level. While
there were substantial variations in the predicted probability estimates
between population subgroups, particularly by age, across-year differences
within each population subgroup were largely trivial. The exception was for new
SRD diagnoses among quitters. In the two years following quitting, the
probability of a new SRD fell below that of a continuing, current smoker.
II.1.c Disenrollment
Disenrollment is a critical concern for health insurance plans considering
tobacco cessation services. High disenrollment rates can reduce the future
expected savings from investments in cessation services for health insurance
plans. The ROI Calculator accounts for disenrollment. Disenrollment rates
differ substantially depending on age and health status. Users also should note
that the annual disenrollment rates for the cohort of adults in the model are
likely less than the annual disenrollment rates for the entire health insurance
plan population. The model does not include the disenrollment of subscriber’s
dependents. Also, disenrollment rates tend to stabilize at lower rates after a
year or two.
In the study cohort, about 64% remained in the HMO at the end of five years.
Disenrollment was inversely related to age, with disenrollment averaging 15–20%
for 18-34 year olds, 5–10% for 35-64 year olds, and 2–3% for persons 65 years
and over. While never smokers, new quitters in the quit year, and existing
former smokers were least likely to leave the plan, current smokers and new
quitters after the quit year were most likely to disenroll. Heavy smokers were
more likely to leave the plan than light smokers. Individuals with a new SRD
diagnoses were less likely to leave the plan than individuals without a new
SRD. Differences were significant at the p<.001 levels. In the default data,
annual disenrollment was somewhat higher in the first two years compared to
later years. For model simplicity, we used the average disenrollment rate over
the five-year period. The impacts on the model results were minute.
II.2. Cost Component
The ROI Calculator’s cost component includes annual medical expenditures and
estimated productivity losses for years 1998–2002. At the end of each year, we
estimate mean annual medical expenditures and productivity costs for
individuals in each population subgroup as defined in the population component.
The ROI calculator includes data for individuals who were current smokers at
the beginning of the study.
It is important to note that these costs are based on retrospective analyses,
not prospective predictions of future costs. Retrospectively, individuals who
remained smokers during the year have low mean expenditures primarily because
they did not have an adverse health event that typically prompts smokers to
quit. Some degree of service under-utilization is also likely to have
constrained costs for continuing smokers.
Analyses were conducted using a stratified sub-sample of 62,717. We assigned
population weights to each population strata to match the 200,000 population
used in the probability analyses. Medical costs were calculated for persons
enrolled in the health plan at the beginning of the year. Costs were annualized
for individuals who left the plan during the year.
We used a one-part multivariate Generalized Linear Model (GLM) with a gamma
distribution and log link function (proc GENMOD in SAS Version 8.2 ©) to
generate predicted mean annual expenditures for each population strata
described above. This approach allowed us to account for individuals with zero
costs and the non-normal expenditure distributions for individuals with
utilization. This approach has been shown to be an efficient predictor of mean
costs compared to other econometric approaches (Diehr et al., 1999).
Data analysis highlights were consistent with previous studies, showing
continuing smokers have lower mean annual expenditures compared to existing
former smokers and new quitters. For new quitters, mean costs were highest in
the quit year and fell each year after quitting, but did not reach the average
for continuing smokers.
II.2.a. Medical Costs for Cessation Program Quitters
A critical issue for modeling quitting’s cost implications is determining the
future expenditures for smokers who quit because of an intervention program.
Studies of smoking-attributable medical costs show mean expenditures are higher
for new quitters than continuing smokers (Wagner et al., 1995; Miller et al.,
1998; Miller et al., 1999; Fishman et al., 2003; Warner et al., 2004). However,
these studies did not properly control for the temporal relationship between
disease diagnoses and quitting.
Most intervention participants, and subsequent quitters, are healthy. About
half of all quitters site future health as a reason for quitting, but only
about 25% cited current health concerns prompted quitting (USDHHS, 2001). Thus,
it is likely that most would have remained healthy had they not quit.
We wanted to avoid falsely assigning the high costs of self-quitters to healthy
smokers who quit because of an intervention. Therefore, we estimated predicted
mean annual expenditures using smoking status at the beginning of the period
(instead of using smoking status at the end of the year). The predicted costs
for smokers at the beginning of the year were slightly higher than for those
who were still smokers at the end of the year.
We further assumed that a program-induced quit would not immediately change the
health status of a quitter (except for pregnancy, which was not assessed). To
correctly assign predicted costs to program quitters, while maintaining cost
comparability in the first year with the base population, we assigned all
current smokers and quitters the same mean costs in the first year in both the
usual care and intervention cohorts.
In the year following a program-induced quit, quitters without a new SRD were
assigned the average for all smokers (at the beginning of the year) and
continuing healthy smokers (assessed at the end of the year) for that year.
Program quitters who had an SRD were assigned the baseline mean costs for
self-quitters. In subsequent years, we assigned continuing quitters, without an
SRD, the mean costs for healthy continuing smokers. We believed this reasonably
represented the likely true future costs of healthy quitters without an SRD.
II.2.b. Productivity Costs
The productivity loss estimates are limited to lost work time associated with
excess smoking breaks and absentee days. The model does not include long-term
disability due to smoking-related disease episodes, replacement costs for sick
employees, the costs of accommodating workers’ needs when they return, and the
impact of sick workers on the employer’s health insurance risk pool rating.
Therefore, the estimated productivity savings are conservative.
We estimated smoking-attributable productivity losses using published estimates
from Warner et al. (1996). Male current smokers were assumed to have 3.9 more
absentee days each year than male non-smokers, and women smokers had 2.1 excess
absentee days. In addition, smokers were assumed to spend an additional five
minutes each day for break time compared to nonsmokers. We used Warner’s
assumption that quitting resulted in a 25% reduction in excess absentee days in
the quit year and 25% proportional reductions in subsequent years, and that
excess break time dropped to zero upon quitting. We valued lost productivity
using age-specific, per-capita earnings from work from Haddix et al., adjusted
to 2002 dollars using the compensation component of the Employment Cost Index
(BLS, 2004).
II.3. Intervention Component
The interventions tested include the “5 As” (including brief physician
counseling), 5As plus pharmacotherapy (Rx), 5As plus proactive telephone
counseling (quitline), and 5As plus pharmacotherapy plus quitline. We compared
each intervention to a baseline (usual care) practice of asking about smoking
status and brief advice to quit (<3 minutes) such that 60% of smokers
receive quit advice. We assumed the interventions were delivered as part of a
routine care visit with a General/Family Practitioner or General Internal
Medicine practitioner, with referral to an existing proactive telephone
quitline service when appropriate. Clinician time for each intervention was
assumed to be 10 minutes for the 5As plus a 3 minute telephone follow-up, 15
minutes for the 5As plus Rx (2 minutes to discuss dosage and other related
issues), 5 minutes for 5As when referral to a quitline is accepted, and 7
minutes for 5As plus Rx plus quitline. Clinician time is reduced when telephone
counseling is included because the phone counselors take responsibility for the
last two As (assist and arrange).
II.3.a. Intervention Costs
Intervention costs include physician time and overhead costs to deliver the
intervention, printed quit materials, prescription medication costs, contracted
telephone counseling services, provider training and incentives, and other
program support costs.
We valued clinical time using mean net income ($146,987) for FP/GP
practitioners (Wassenaar and Thran, 2003), adjusted to 2002 dollars. We also
included facility overhead cost estimates of $16.88 per in-office hour from
Peden and Baker (2002), adjusted to 2002 dollars using the medical CPI.
Individuals receiving the 5 As were also assumed to have received printed quit
materials valued at $7 each.
The model assumes the cost per participant for prescription medication is $204
(includes $20 shipping) and $195 for proactive multi-session telephone
counseling. Nicotine replacement therapy (NRT) is the default medication in the
ROI Calculator. The model assumes an eight-week course of NicoDerm®
CQ® (an over-the-counter patch), a three-step
program of 21 mg daily dose for four weeks, followed by two weeks each of 14 mg
and 7 mg daily doses. Total costs of $204 in 2004 were obtained from
www.Drugstore.com Wellbutrin SR®, a brand of
the generic anti-depressant buproprion, has similar effectiveness (Fiore et
al.) but is somewhat more expensive than NRT. We found a retail price of $230
for Wellbutrin SR® at
www.net-rxmeds.com for a 10-week treatment program of 150 mg per day,
beginning two weeks before the quit date.
The ROI Calculator assumes no co-payments, based on USPHS recommendations
(Fiore et al.). If co-payments are required, the model reduces intervention
acceptance by 50%. We valued the additional clinical time patients spent with
the provider using the earnings data used to assess productivity losses.
Physician training and incentives, and other program support costs are an
important part of assessing a system-level change in clinical practice. These
costs are difficult to quantify, given the wide-range of organizational
structures and practice characteristics. The model assumes an exclusive
provider relationship to guide estimation of the default costs of physician
training. The costs were estimated by multiplying physicians’ net revenue and
overhead costs per hour by the amount of time spent in training and reviewing
performance feedback (using an achievable benchmark approach). We assume
physicians attend an initial 1-hour training session (including lunch),
followed by 15-minute review sessions after 2 weeks, 1 month, and at 3, 6, and
9 months. The review sessions include discussion of feedback reports,
intervention strategies, and problems. The estimated total cost per physician
was $248. The default data assumes there are 295 primary care physicians in
general FP/IM specialties. Thus, total costs for physician training were
approximately $73,000.
For practice arrangements, outside group/staff and integrated health system
models, health insurance plans may face unique challenges implementing
cessation programs in their network or contracted physicians’ practices. Plans
will likely need to depend more on provider incentives, clinical reminders,
marketing services to members, and more diffuse methods of provider training.
We recognize that the costs of implementing a cessation program will be more
costly for less-centralized practice arrangements. Fewer adherences to
protocols also will be likely, which will reduce the overall reach of the
intervention program. While the web-based model does not adjust for reach
associated with protocol adherence, analyses using the spreadsheet model
indicate lower rates of intervention delivery do not change the conclusions of
the ROI modeling.
We suggest users adjust the physician training and incentives estimate to
account for the potential costs of achieving the model’s intervention reach of
5% and 10%, with and without member co-pays, respectively. Based on our review
of the literature and discussions with colleagues, we suggest the following
costs as a proxy for achieving the default reach values.
If your plan has:
-
An exclusive provider relationship, multiply $248 by the number of primary care
physicians for training costs.
-
A network of physician groups or individual physicians, multiply the estimated
number of adult smokers in your plan by $20 to use as an incentive to
physicians for achieving and documenting tobacco status and cessation delivery
targets, including
-
95% of patients asked about smoking status
-
80% of smokers advised to quit and readiness to quit assessed
Training may include written training materials, chart reminder stickers, and
conversations with the tobacco control coordinator (see other program support
costs). Training materials and chart reminders may cost about $10 per
physician.
Note: We conducted analyses of the impact of lower reach on the ROI results.
Holding start-up costs constant, we found that declining reach improved the ROI
estimates slightly, compared to usual care, as use of medication and counseling
declined. This shift occurs because the percent of smokers receiving at least
advice to quit is higher for each intervention than for usual care.
For other program support costs, the ROI Calculator assumes the health
insurance plan hires an experienced Masters-level interventionist to serve as a
tobacco control coordinator and trainer at .5 FTE, provides a research analyst
at .1 FTE, conducts provider training, and uses performance feedback for
physicians based on achievable benchmarks. We assumed $30 per hour for the
interventionist, $25 per hour for the analyst, a 40% fringe benefit rate, and a
55% indirect cost rate. The indirect rate is used to capture overhead costs and
other support. The estimated total other costs are approximately $80,000.
II.3.b. Reach and Efficacy
An intervention’s reach was estimated as the product of the percent of
individuals, by age and sex, with at least one annual routine care visit
(Cherry et al., 2003), 75% of smokers asked about smoking status, advised to
quit, and assessed for readiness to (75% based on Healthy People 2010
objectives (USDHHS, 2000)), 46% ready to quit in the next six months (Hollis,
2001), and acceptance rates for each intervention. We conservatively assumed
50% of those smokers who were ready to quit would accept a 5 A's regimen (with
or without medications) if no co-payment was required. We assumed 40% would
accept the 5 A's plus telephone counseling (with and without medications) if no
co-payment was required. These assumptions lead to an overall intervention
reach of about 10% for a 5 A's program with telephone counseling and 11% if
telephone counseling is not included. The percentages vary slightly by age and
sex, and are consistent with Curry et al. (1998). We assumed reach falls by
half if co-payments are required.
Intervention efficacy data were obtained from Fiore et al. We used estimates
for brief physician advice to quit, 3-10 minute of physician counseling,
nicotine replacement therapy, buproprion (Wellbutrin SR), and proactive
telephone counseling. We calculated quit rates by multiplying odds ratio
estimates by the probability of quitting for each sex and age strata without a
new SRD diagnosis. The model assumes a 10.2% quit rate for smokers receiving
advice to quit, 16% for individuals accepting the 5 A’s regimen, 24.3% for the
5 A’s with prescription medication, 19.2% for the 5 A’s with proactive
multi-session telephone (quitline) counseling, and 29.2% with 5 A’s and both
medication and quitline counseling. We assumed 5 A’s counseling was 10 minutes
in duration with a 3 minute follow-up call attempt. If medication was included,
an extra 2 minutes was spent discussing dosage and potential side-affects. If
smokers received quitline counseling, the 10-minute provider counseling was
reduced to 5 minutes. The model assumes that 3%-6% of the population quits on
their own each year, depending on age and sex.
II.3.c. Modeling the financial impact of smoking cessation interventions
The Model evaluates each intervention’s effect by following the population
through each year of the model assuming the usual care condition and one of the
four cessation interventions are implemented for one year, and comparing
differences in the population (SRD incidence, quitting, disenrollment), annual
medical care expenditures, and smoking-attributable productivity costs. All
costs are stated in 2002 dollars, and future costs are discounted to the
present value using a standard 3% rate that represents the real social discount
rate (Haddix et al., 2003).
The net return on investment is calculated per intervention participant, per
quitter, and per member per month using the following formula:
Net ROI per participant: |
(MCb–MCi)
+ (PCb–PCi) – ( ICj–ICb) |
| Pi
– Pb |
Where MCb and MCi,j are
discounted annual medical care expenditures for baseline (b) and jth
intervention (i), PC is discounted smoking-attributable productivity losses for
baseline and each intervention, IC is the intervention cost for each
intervention, and P is the number of participants in each intervention. The
intervention costs are not discounted since the intervention is assumed to
occur only in year one. Net ROI per quitter is calculated similarly with the
appropriate population groups in the denominator. The net ROI per member per
month (PMPM) formula is estimated with the adult health insurance plan member
population in the denominator.
The ROI Calculator estimates incremental ROI for health insurance plans and
employers using disaggregated intervention cost and cost savings data. Health
plan results included only medical care utilization and intervention costs,
while employer ROI included only productivity data.
II.3.d. Discounting Future Costs
All costs are stated in 2002 dollars, and future costs were discounted to the
present value using real discount rate of 7.63%. This rate is the
cost-of-capital for the Medical Services sector rate as of July 2004. The cost
of capital can be thought of as the real (inflation-adjusted) discount rate
that equates the price of obtaining investment funds with the expected real
rate of return from the investment. The estimate was obtained from the website
of Aswath Damodaran, PhD, Professor of Finance, Leonard Stern School of
Business, New York University, NYC. Available at:
http://pages.stern.nyu.edu/~adamodar/New_Home_Page/datafile/wacc.htm.
II.3.e. Other Adjustments
The model allows users to make additional adjustments in the input data in
order to better fit the health plan population being assessed. We adjusted
medical care costs for health plan type (HMO, FFS, PPO, and POS) and Census
region using health care premium data from the Medical Expenditure Panel Survey
online data retrieval tool available from (MEPSnet/IC, 2003). The model
multiplies expenditures by 1.08 for Preferred Provider Organizations (PPOs) and
1.06 for Fee-for-Service (FFS) plans. The multiplication factors are based on
the ratio of average total family premiums for families covered by insurance
obtained from 2001 Medical Expenditure Panel Survey (MEPS) (AHRQ, 2004).
Regional expenditure differences are addressed by multiplying the default
medical expenditure data by 1.20 for the Northeast, 1.15 for the South, and
1.28 for the Midwest (the default region is West).
III. ROI Calculator Data Inputs
Smoking intensity
The ROI Calculator models heavy (= 1 pack per day) and light smokers
separately. Currently, users cannot change the distribution of heavy and light
smokers. The model splits smokers into heavy and light smokers based on medical
record analyses. For males, 36% of smokers ages 18-34, 58% of smokers ages
35-64, 49% of smokers ages 65 and over are heavy smokers. For women smokers,
heavy smokers are 32% for ages 18-34, 51% for ages 35-64, and 48% for ages 65
and over.
Disease (smoking-related diseases (SRD))
The model includes disease diagnosis data for smoking-related cancers,
cardiovascular diseases, and chronic lung disease. We used conditions that the
Surgeon General concluded are causally linked to cigarette smoking (USDHHS,
2004). A complete list is available from the Business Case project Principle
Investigator.
The ROI Calculator estimates the number of individuals who have one or more new
SRD diagnoses each year. In future model years, individuals SRD diagnosis
history is used in estimating the probability of a new diagnosis, quitting and
relapse, and disenrollment.
Smoking Relapse
The model accounts for relapse, but does not yet allow you to change the
relapse rate. The default data were estimated from medical records analyses of
HMO members. Separate estimates are used for each sex, age group, smoking
intensity, and SRD history. Relapse rates in the model are 3%-6% per year
depending on age.
Smokers
Enter your estimated percentage of current smokers for each age and sex.
Estimates can be obtained from medical records, member surveys (e.g. CAHPS),
population surveys, or research studies of your members. Read the
population section to learn how we calculated smoking prevalence.
If you do not know the smoking rate for your population, use the default data
provided for your state or region.
Cost of Capital
Enter the cost-of-capital for your organization.
This rate will be used to discount future medical care expenditures and
productivity estimates to the present value. The default cost of capital is
7.63% , which reflects the rate for the medical services sector as of July
2004.
Population Name
The population name you provide is used in the title of the results report. The
population name should be short and descriptive to help you manage the results
of your analyses.
Health Insurance Plan Type
Enter the percentages of your population in HMO, PPO, and FFS plans. The total
must equal 100%. We use the data to adjust the medical expenditure data for
differences based on health plan type. The default is 100% HMO
State/Region of Residence
Users must select the primary state or region of residence for the study
population. The ROI Calculator automatically adjusts the medical expenditure
estimates to account for regional cost differences (based on annual family
premiums), and calculates the number of current smokers using BRFSS prevalence
data for the selected state or region.
Prevalence estimates for each state can be used if you do not know the smoking
rates for your population. Regional prevalence data are provided as a guide for
insurers and employers whose populations include multiple states within a
single Census region. The state and regional estimates are averages for
2001–03. We used multiple years to reduce variations in year-to-year estimates
associated with sampling.
The states included in each region are defined below:
New England (CT, MA, ME, NH, RI, VT)
Mid Atlantic (NJ, NY, PA)
South Atlantic (DC, DE, FL, GA, MD, NC, SC, VA, WV)
South Central (AL, AR, KY, LA, MS, OK, TN, TX)
East North Central (IN, IL, MI, OH, WI)
West North Central (IA, KS, MN, MO, ND, NE, SD)
Mountain (AZ, CO, ID, MT, NM, NV, UT, WY)
Pacific (AK, CA, HI, OR, WA)
Age Group
The age groups were selected for statistical estimation purposes and to
minimize computational time. CDC uses the 35-64 and 65+ age groups to estimate
smoking-attributable deaths.
Leave the Plan/Employee Turnover
Review the default annual member disenrollment rates for each age group for
your plan, and change your plan rates if needed. We will use this rate to
adjust the default values in the model. The default rates appear lower than
overall plan averages due to the cohort approach of the model and exclusion of
children under age 18.
Members are assumed to leave the employer when they disenroll from the health
insurance plan, which reduces productivity savings. Employers and health
benefits managers can substitute employee turnover for plan disenrollment.
Cost per Participant
Review the default costs per participant for medication and telephone
counseling, and adjust if necessary.
For medication, enter the formulary costs for your plan for an 8-week course of
nicotine replacement therapy (e.g., Nicoderm CQ) or 10-week 300 mg/day course
of buproprion SR (e.g., Wellbutrin SR).
For telephone counseling, enter the costs paid by your plan for a proactive
multi-session telephone cessation-counseling program.
Enter the combined cost of medication and telephone counseling if the total
cost is different than the sum of each component.
Member Paid Cost
Enter the dollar amounts you expect each participant to pay out-of-pocket for
prescription medication and telephone counseling, separately and combined. The
default is no member paid costs.
The costs paid by the intervention participant affect intervention utilization.
If member paid costs are positive, the model reduces acceptance of the
interventions by 50%.
Provider Training and Incentives
Provider training for primary care practitioners is crucial, but very difficult
to estimate. The level of training and system-level efforts to promote the
provision of services will affect intervention reach. The model adjusts
intervention reach if member co-pays are required, but is not based on the
organizational structure of the plan or physician adherence to protocols.
However, based on our review of the literature and discussions with colleagues,
we suggest the following costs as a proxy for achieving the default reach
values.
If your plan has:
-
An exclusive provider relationship, multiply $248 by the number of primary care
physicians for training costs.
-
A network of physician groups or individual physicians, multiply the estimated
number of adult smokers in your plan by $20 to use as an incentive to
physicians for achieving and documenting tobacco status and cessation delivery
targets, including
-
95% of patients asked about smoking status
-
80% of smokers advised to quit and readiness to quit assessed
Training may include written training materials, chart reminder stickers, and
conversations with the tobacco control coordinator (see other program support
costs). Training materials and chart reminders may cost about $10 per
physician.
Note: We conducted analyses of the impact of lower reach on the ROI results.
Holding start-up costs constant, we found that declining reach improved the ROI
estimates slightly, compared to usual care, as utilization of medication and
counseling declined. This occurs because the percent of smokers receiving at
least advice to quit is higher for each intervention than for usual care.
Other Program Support Costs
Enter your costs for a tobacco control coordinator/trainer to oversee the
program. Include the costs of a research analyst.
The default assumes a .5FTE Masters-level tobacco control coordinator and
trainer and a research analyst at .1 FTE. The costs include 40% fringe benefit
and 55% indirect cost rates.
Time Period for the Analysis
The model will calculate cumulative results for the time period you select. All
figures will be discounted to the present value using a cost-of-capital rate of
7.63%.
References
Agency for Healthcare Research and Quality. Health Insurance Component
Analytical Tool (MEPSnet/IC). September, 2003. Agency for Healthcare Research
and Quality, Rockville, MD. http://meps.ahrq.gov
(Accessed November 30, 2003).
Bureau of Labor Statistics. Employment Cost Index U.S. Department of
Labor, Bureau of Labor Statistics, Office of Compensation Levels and Trends,
Washington, DC, 2004.
Available at: http://www.bls.gov/ncs/ect/
Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance
System Survey Data. Atlanta, GA: U.S. Department of Health and Human
Services, Centers for Disease Control and Prevention, 2001-2003.
Cherry DK, Burt CW, Woodwell DA. National ambulatory medical care survey: 2001
Summary. Advance data from vital and health statistics; no 337.
Hyattsville, MD: National Center for Health Statistics, 2003.
Curry SJ, Grothaus LC, McAfee T, Pabiniak C. Use and cost-effectiveness of
smoking-cessation services under four insurance plans in a health maintenance
organization. New Eng J Med 1998:673-679.
Damodaran A. Cost of Capital by Sector Leonard Stern School of Business,
New York University, NYC, 2004.
Available at:
http://pages.stern.nyu.edu/~adamodar/New_Home_Page/datafile/wacc.htm (Accessed
on September 15, 2004).
Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care
utilization and costs. Annu Rev Public Health 1999;20:125-144.
Fiore MC, Bailey WC, Cohen SJ, et al. Clinical practice guideline: Treating
tobacco use and dependence. Rockville, MD: Public Health Service;2000.
Fishman PA, Khan ZM, Thompson EE, Curry SJ. Health care costs among smokers,
former smokers, and never smokers in an HMO. HSR: Health Services Research
2003;38:733-749.
Haddix A, Teutsch SM, Corso PS. Prevention effectiveness: A guide to decision
analysis and economic evaluation. Oxford University Press;2003.
Hollis JF. Population impact of clinician efforts to reduce tobacco use. In Population
based smoking cessation: Proceedings of a conference on what works to influence
cessation in the general population. Smoking and tobacco control
monograph no. 12. Bethesda, MD: U.S. department of health and human services,
national institutes of health, national cancer institute, nih pub. no. 00-4892,
November 2000.
Miller LS, Zhang X, Novotny TE, Rice DP, Max W. State estimates of Medicaid
expenditures attributable to cigarette smoking, fiscal year 1993. Pub Health Rep
1998;113:140-51.
Miller VP, Ernst C, Collin F. Smoking-attributable medical care costs in the
USA. Soc Sci Med 1999;48:375–91.
Peden A, Baker J. Allocating physicians’ overhead costs to services: An
econometric/accounting –activity based- approach. J Healthcare Fin 2002;29:57-75.
SAS Institute Inc., Cary, NC: 2000.
U.S. Dept of Health and Human Services. Healthy People 2010. 2nd ed. With
Understanding and Improving Health and Objectives for Improving Health.
2 vols. Washington, DC: U.S. Government Printing Office, 2000.
U.S. Dept of Health and Human Services. Women and smoking: a report of the
Surgeon General. Rockville, MD: U.S. Dept of Health and Human Services,
Public Health Service, Office of the Surgeon General; Washington, DC: U.S.
Government Printing Office, 2001.
U.S. Department of Health and Human Services. The health consequences of
smoking: A report of the surgeon general. U.S. Department of Health and
Human Services, Centers for Disease Control and Prevention, National Center for
Chronic Disease Prevention and Health Promotion, Office on Smoking and Health,
2004.
Wagner EH, Curry SJ, Grothaus L, et al. The impact of smoking and quitting on
health care use. Arch Intern Med 1995;155:1789-1795.
Warner KE, Smith RJ, Smith DG, et al. Health and economic implications of a
work-site smoking cessation program: A simulation analysis. J Occup Environ Med
1996;38:981-92.
Wassenaar JD, Thran SL (eds). Physician Socioeconomic Statistics 2003 Edition:
Profiles for detailed specialties, selected states, and practice arrangements.
American Medical Association, Center for Health Policy Research Chicago, IL:
AMA Press, 2003.
|