AHIP

Return Users:
Type in your email address
OK
ROI Calculator
 
- The Business Case
- Executive Summary
(pdf version)
- Methodology
- FAQ's
- Resources and Bibliography
- Related Links
- ROI Team
- Feedback
- Privacy/Security Statement

 

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
II. Components of the ROI Calculator
  II.1. Population Component
    II.1.a. Population characteristics
    II.1.b. Probability data
    II.1.c. Disenrollment
  II.2. Cost Component
    II.2.a. Medical costs for cessation program quitters
    II.2.b. Productivity costs
  II.3. Intervention Component
    II.3.a. Intervention costs
    II.3.b. Reach and efficacy
    II.3.c. Modeling the financial impact of smoking cessation
    II.3.d. Discounting future costs
    II.3.e. Other adjustments
III. ROI Calculator Data Inputs
  Smoking intensity
  Disease
  Smoking Relapse
  Smokers
  Cost of Capital
  Population Name
  Health Insurance Plan Type
  State of Residence
  Age Group
  Leave the Plan (disenrollment)
  Cost per Participant
  Member Paid Cost
  Provider Training and Incentives
  Other Program Support Costs
  Time Period
  References


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:
  1. An exclusive provider relationship, multiply $248 by the number of primary care physicians for training costs.
  2. 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:
  1. An exclusive provider relationship, multiply $248 by the number of primary care physicians for training costs.
  2. 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.