Evaluating the cost-effectiveness of cancer patient navigation programs: Conceptual and practical issues

Evaluating the Cost Effectiveness of Cancer Patient NavigationPrograms: Conceptual and Practical Issues

Scott Ramsey, MD, PhD1, Elizabeth Whitley, RN, PhD2, Victoria Warren Mears, RD, PhD3,June M. McKoy, MD, MPH, JD4, Rachel M. Everhart, MS2, Robert J. Caswell, PhD5, KevinFiscella, MD, MPH6, Thelma C. Hurd, MD7, Tracy Battalgia, MD, MPH8, and JeanneMandelblatt, MD, MPH9 For the Patient Navigation Research Program Group1 Fred Hutchinson Cancer Research Center, Seattle, Washington2 Denver Health & Hospital Authority, Community Voices, Denver, Colorado3 Northwest Portland Area Indian Health Board, Northwest Tribal Epidemiology Center, Portland,Oregon4 Northwestern University Feinberg School of Medicine, Departments of Medicine and PreventiveMedicine, Chicago, Illinois5 Ohio State University, College of Public Health, Columbus, Ohio6 University of Rochester School of Medicine, Department of Family Medicine, Rochester, NewYork7 University of Texas Health Science Center, San Antonio, Texas8 Women’s Health Research Unit, Boston, Massachusetts9 Georgetown University Medical Center, Cancer Control Program, Lombardi ComprehensiveCancer Center, Washington, DC

AbstractBackground—Patient navigators--individuals who assist patients through the healthcare systemto improve access to and understanding of their health and health care—are increasingly utilizedfor underserved individuals at risk for or with cancer. Navigation programs can improve access,but it is unclear whether they improve the efficiency and efficacy of cancer diagnostic andtherapeutic services at a reasonable cost, such that they would be considered cost effective.

Methods—We outline a conceptual model for evaluating the cost effectiveness of cancernavigation programs. We describe how this model is being applied to the Patient NavigationResearch Program (PNRP), a multi-center study supported by the National Cancer Institute’sCenter to Reduce Cancer Health Disparities.

Results—The PNRP is testing navigation interventions which aim to reduce time to delivery ofquality cancer care (non-cancer resolution or cancer diagnosis and treatment) after identification ofa screening abnormality. Examples of challenges to evaluating cost effectiveness of navigationprograms include the heterogeneity of navigation programs, the sometimes distant relationshipbetween navigation programs and outcome of interest (e.g., improving access to prompt diagnosticresolution and life years gained) and accounting for factors in underserved populations that mayinfluence both access to services and outcomes. In this article, we discuss several strategies foraddressing these barriers.

Address for correspondence: Scott D. Ramsey, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North M3-B232, POBox 19024, Seattle, WA 98109. Phone (206) 667-7846, Fax (206) 667-5977, [email protected].

NIH Public AccessAuthor ManuscriptCancer. Author manuscript; available in PMC 2010 December 1.

Published in final edited form as:Cancer. 2009 December 1; 115(23): 5394–5403. doi:10.1002/cncr.24603.

NIH

-PA Author Manuscript

NIH


NIH


Conclusions—Evaluating the costs and impact of navigation will require some novel methods,but will be critical in recommendations about dissemination of navigation programs.

Keywordscancer; navigator; cost effectiveness; modeling

IntroductionPopulations with limited access to or knowledge of the health care system often havedifficulty utilizing the system effectively for cancer services and this may result in delays incancer diagnosis,1, 2 added costs,3 and less efficient and effective use of recommendedtherapies. Patient navigation programs provide support and guidance to persons with thegoal of improving access to the cancer care system and overcoming barriers to timely,quality care.4–14 In this paper, we present a conceptual model for evaluating the costeffectiveness of cancer patient navigation programs, discuss methodological challenges, andsuggest approaches for addressing these challenges.

Rationale for and History of Patient Navigation ProgramsThe origins of patient navigator programs are widely attributed to Harold Freeman, who, aspresident of the American Cancer Society (ACS), commissioned a study of barriers tocancer care among the poor in the United States. The report documented substantialdisparities both in cancer care and outcomes between poor and non-poor Americans,identifying, among other issues, significant barriers to care and a sense of fatalism aboutcancer that prevented many from seeking care in the first place.15 As a result of this report,the ACS supported the first “Patient Navigation” program in 1990 at the Harlem HospitalCenter. A pre-post comparison of women diagnosed with breast cancer at this facilityshowed that 41% of breast cancer patients diagnosed between 1995 and 2000 werediagnosed with early disease compared to 6% of patients between 1964 and 1986.16, 17Five-year survival rates increased from 39% to 70% over the same period.

Due to the success of this pioneer program and in recognition that significant barriers toeffective cancer screening, diagnosis, and care continue to exist among minority andunderserved populations, patient navigation programs are becoming more common,particularly among health systems that serve these populations. The Centers for Medicareand Medicaid Services is funding demonstration projects to reduce barriers to care at alllevels.18 Despite their growing popularity and the publication of promising observationalstudies19–22 very few prospective, controlled trials have evaluated the efficacy of navigatorprograms. Controlled trials, most of which are small, have shown significant improvementsin time to diagnosis, reductions in anxiety, and greater levels of satisfaction with the careprocess.23–25 The impact of navigation programs on cancer-related morbidity and survival,and the cost-effectiveness of these programs is not yet known.

The Patient Navigation Research ProgramThe National Cancer Institute (NCI) and the ACS are sponsoring a 9-site Patient NavigationResearch Program (PNRP) [Table 1]10 The primary aim of the PNRP is to evaluatenavigation programs’ impact on the time between an abnormal finding (from a screening testor clinical examination for case finding) to definitive diagnosis and treatment initiation.Secondary aims include evaluating the impact of navigation on patient satisfaction and thecost-effectiveness of navigation.

Ramsey et al. Page 2

Cancer. Author manuscript; available in PMC 2010 December 1.

NIH


NIH


NIH


PNRP sites serve diverse patient populations. Navigation programs focus on follow-up ofabnormal breast, cervical, prostate, and colorectal cancer screening tests, among minoritypopulations including African Americans, American Indians, Asians, Hispanics, and therural underserved. Navigation models vary across sites, utilizing different professionals andhealth care systems [Table 1] to follow patients through the completion of initial treatment.

Rationale for Evaluating the Cost-effectiveness of Patient Navigation ProgramsPatient navigator programs can be time and resource intensive. Like other interventions thatmay improve the health of poor and underserved populations, navigation programs must beviewed in the context of allocating resources such that health outcomes are maximized underlimited budgets. It is particularly important to evaluate the cost-effectiveness of publiclyfunded navigator programs, since funding for these programs typically come from globalhealth budgets that are fixed in the short run with many competing needs. Cost-effectivenessanalysis can assist decision makers by showing the health benefit for expenditure ofnavigator programs relative to other interventions, particularly those that are targeted to thesame disease or condition of interest. The desirability of navigator programs can also beassessed in terms of commonly accepted thresholds (e.g., $100,000 per quality adjusted lifeyear gained) in the health system or country. 26

Conceptual Model for Cost Effectiveness Analysis of Patient Navigation InterventionsFor the PNRP, we are using cost-effectiveness analysis (CEA) to compare the added(incremental) costs of navigation interventions versus those of the status quo for the giventarget populations.27 Cost-effectiveness analysis is a comparison of alternatives, typically anew intervention such as navigation vs. usual care, which in patients and their familymembers seeking care without formal assistance. Costs and consequences flow from eachalternative (navigated vs. usual care) are summarized over the time period that is relevant tothe episode of care (Figure 1). The incremental cost effectiveness of navigation is derivedusing the following formula:

(1)

where CNav and CUC refer to the incremental difference in total costs of the navigationprogram compared to usual care, and ENav and EStd refer to the difference in totaleffectiveness between navigation and usual care [Figure 1]. While the comparator istypically “usual care;” that is, care as it occurs in usual practice in the absence of navigators,one could also compare two or more navigation programs vs. usual care, or one programwith another. Both the navigator program and usual care have costs that flow from the pointof entry (e.g., abnormal finding on mammogram) to short- and long-term downstream costsand consequences. Generally the time horizon is the individuals’ remaining years of life.Since the PNRP program will only observe individuals over a maximum of the 5 years ofthe program, examining impact on survival and costs per quality-adjusted life years savedwill require estimation using mathematical models.

Navigation Cost-effectiveness Analysis and Approaches for Addressing ChallengesEvaluating the cost-effectiveness of patient navigation programs poses several uniquechallenges [Table 3]. In this section, we describe particular challenges for evaluating thecost effectiveness of the PNRP and how we plan to address those issues.

Defining the Navigation Intervention—The first issue in conducting the cost-effectiveness analysis of navigation is that the navigator intervention itself is not uniform forall patients, since part of the principle of navigation is to identify patient-specific issues and



NIH


NIH


NIH


tailor the program to those needs. Moreover, navigation interventions (including PNRP) arequite heterogeneous, and are typically tailored to the needs and available resources of aparticular region and the cancers of interest. Even within a single program site, the navigatorwill tailor the intervention to the needs of the particular patient-client, with wide variation inservices provided between individuals. A related issue is that programs differ inexpectations, qualifications, training, and supervision of navigators. In many settingsnavigators are trained to assist patients with abnormal screening tests for several cancers(e.g., cervical and colorectal, or breast and prostate). Although there are economies of scalein these situations, it is more difficult to segregate the time costs for each cancer and modeleach separately. One could capture the economies of scale by modeling all screening, butthis requires extension of the time horizon in a model capturing the natural history ofmultiple cancers at once.

At present, we are not aware of models that are designed to incorporate the natural history ofmultiple cancers simultaneously. However, this is an important research priority since mostproviders recommend screening for multiple cancers to their patients and navigators assistindividuals in navigating through to diagnostic resolution for more than one cancer type.

Therefore, we address the issue of the heterogeneity of interventions by defining thenavigation programs broadly, as specified by the study protocols. 28 This approachemphasizes the type of navigator (e.g., nurse, layperson) and the general scope of servicesthat that individual is able to provide. We will then have to model the cost-effectiveness ofnavigation for each individual cancer separately; allocating navigator time and other effortsin proportion for each cancer site.

Measuring Effectiveness of Navigation Programs—The recommended measure ofeffectiveness of navigation programs for cost-effectiveness analyses is the Quality AdjustedLife Year (QALY),29 which requires data on survival with and without the program andevaluation of health state preferences (utilities). However, outcome measures being directlytracked by the PNRP research sites are intermediate outcomes: time to definitive diagnosis/resolution and time to initiation/completion of recommended cancer therapy for those with acancer diagnosis.28 Moreover, the period of observation under the five-year PNRP programwill be too short to observe any mortality endpoints.

Estimating QALYs will require simulation modeling. To address the need to extrapolatefrom the observation period to estimate the impact of navigation over a lifetime, we will usesimulation models to extend the time frame of observation and look at stage distribution ofpatients diagnosed under navigation and usual care, using local cancer registries, hospitals,and patient charts. Age-, race-, and stage-specific survival from cancer registries (local ornational) can then be used to project the life expectancy, or mortality experience of eachgroup of patients.

Even using this approach, modeling the effects of mortality based on delays in diagnosis ortreatment is challenging and requires modeling assumptions. For instance, most modelsportray screening benefits in terms of decreases in tumor size (and number of nodesinvolved) or stage shifts. In this situation, for navigation to show a benefit, the interventionwould have to lead to an early stage diagnosis in a patient who would have otherwise havebeen lost to follow up and only presented clinically at more advanced stages. Less dramaticwithin-stage shifts (e.g., early in the course of local disease vs. later in local disease, butbefore transition to regional spread) are also likely to improve survival, but there are limitedprimary data upon which to model these effects. It is also possible that small within-stageshifts do not affect cancer-specific mortality. We will use sensitivity analysis to evaluatehow different assumptions about stage shift or cure affect results. If navigation is not cost



NIH


NIH


NIH


effective under the most favorable assumptions about small effects, then one could concludethat the investment does not yield a return on investments in QALYs. However, if programswould be considered cost effective under assumptions that are clinically reasonable, thenprograms with small effects could be considered to have the potential to be cost effective.

The relationship between the intervention (navigation) and the endpoints (survival, QALYs)may not be straightforward, since the intermediate outcome of navigation—adherence totimely diagnostic services (where the majority does not have cancer) and to recommendedtherapy—will not necessarily be uniform and linear in its relation to endpoints. We addressthis issue with simulation modeling and sensitivity analysis, the latter evaluating howchanges in the association between specific input parameters (e.g., expenditures onnavigation services and adherence to screening recommendations over time) influences longterm outcomes.

Even if navigation interventions do not improve survival, they still may improveindividuals’ quality of life. In cost-effectiveness analyses, these effects are recorded ashealth state utilities to be used in computing QALYs. Utilities are measures of health statepreference, measured on a scale from 0 (death) to 1 (ideal health). QALYs are a summarymeasure of survival weighted by utilities over the period following the intervention.29Utility weights for navigator program participants and a comparator group can be measuredusing a generic multi-attribute utility instrument such as the EQ-5D.30 Multi-attribute utilityinstruments are questionnaires filled out by respondents assessing their quality of life acrossseveral domains. The individual responses are weighted using data derived from largepopulation surveys on the utility of the different quality of life states. Scores are summedand converted to a 0 to 1 scale with zero representing the worst health imaginable (or death)and one representing perfect health. This approach provides societal, rather than individualpatient ratings of the potential quality of life improvements that might occur withnavigation, so that results are generalizable.

Due to budget constraints, not all PNRP sites will collect multi-attribute utility data for theirparticipants. Utility weights for the comparison (no navigator) group will be based on theliterature, and when available, surveys of low income populations with cancer but nonavigation services.31 We will compare patient populations where utilities are beingcollected and those where they are not. In cases where health and socioeconomic status aresimilar, we use data from the populations were utilities are collected as proxies for thesewhere utilities were not collected. We will also explore the use of regression models basedon navigator study populations with utility data to impute utilities for those without utilitydata.

It should be noted that problems that are highly prevalent in underserved populations thatare being targeted by navigation—such as low literacy rates and frequently changingresidences—pose challenges to measuring outcomes following navigation using existingutility surveys. For example, populations with very low literacy or special groups such as thehomeless or persons with mental illnesses may have great difficulty completing writtenquestionnaires. The PNRP address this issue by 28 allowing telephone and face-to-faceinterviews with patients and, if necessary, patient representatives.

Another issue that is embedded in the navigation program that poses a challenge to cost-effectiveness analysts is that patients with significant barriers to access of health systemsoften have complex social and health issues — such as poor educational attainment or non-cancer-related comorbidity — that themselves may influence long term outcomes, such aslife expectancy and/or cancer – specific survival rates following treatment.32, 33 Education,health status and comorbidity are measured in parent PNRP study. In our projections of



NIH


NIH


NIH


effects from the trial horizon to a lifetime horizon, we will construct multivariate modelswith covariates to account for these characteristics. This will allow us to vary projectedoutcomes based on the characteristics of the cohort of interest; we can also use national dataon the distribution of these factors to conduct sensitivity analyses to estimate the impact ofnavigation in broader settings and populations.

Navigator programs also aim to improve patient satisfaction, self efficacy, and reduce theshort-term distress associated with evaluation of an abnormal screening result. However, selfefficacy and satisfaction with care are generally not incorporated in surveys that measureutilities. In such situations, one could calculate a cost per unit decrease in distress.34However, there are no established benchmarks for comparison to determine whetherparticular reductions in stress are cost-efficient compared to other ways to accomplish thesame goal.

Navigation programs aimed at cancer patients may also have goals such as informed use ofprocedures based on patient preference (e.g. lumpectomy vs. mastectomy) or completionrates of planned therapy. These measures of outcome as well as distress and other outcomes(e.g., stage at diagnosis, time to diagnostic resolution, satisfaction) can be summarized usingcost-consequence analysis.35 Cost-consequence analyses summarize program costs andeffects in tabular fashion [Table 2]. For instance, one can evaluate the costs per patient oftimely diagnostic resolution for the navigator program vs. usual care. Cost-consequenceanalysis can be useful to a decision makers who use components of cost-effectivenessanalysis rather than the cost per QALY ratio. 36

Interpersonal styles and commitments of navigators may influence the outcomes ofparticular programs. Although this factor is very difficult to measure and account for acrosssites, we will evaluate variations in sensitivity analysis, using proxy measures such asvolume-outcome relationships (e.g., volume of patients seen and adherence to follow-up ofabnormal mammograms) and socio-demographics of the navigators themselves (age, sex,education).

Cost Impact of Navigation ProgramsNavigation program costs include allocated “fixed” (office space, proportional allocation ofsupervisory personnel, new equipment or contracts initiated for the program, etc) and“variable” (navigator time and transportation costs, direct medical care, etc) components[Table 2]. We denote the sum of allocated fixed and variable costs as Cnavigator(program).There are also costs associated with training navigators, including replacements oradditional navigators as needed (Ctraining(program)). We denote the total direct medical carecost of diagnostic services and treatments received for persons utilizing navigation programsas Cmedical(program). Patients who receive care without utilizing navigator services have acost, denoted Cmedical(usual care).

Patients and their caregivers incur nonmedical costs when seeking care, such astransportation costs, time costs related to testing and treatment, and time lost from work. Wedenote related nonmedical patient costs for those receiving and not receiving navigatorservices as Cnonmed(program) and Cnonmed(usual care). Note that in the short run, medical andrelated nonmedical costs are likely to be higher for the navigation program because ofimprovements in patient access to care and adherence to protocols for care. Longer termcosts for the navigation program may be lower if a program results in diagnostic resolutionat an earlier stage based on an abnormal screening test, since patients lost to follow-up arelikely to present again with more advanced, more time consuming (and costly) stages ofdisease. Navigation may also lower costs if patients use care more appropriately and



NIH


NIH


NIH


efficiently or better adhere to planned therapy such that cancer relapse rates fall. Thus, in thelong run the net cost of navigation programs can be more or less than those under usual care.

One of the potential cost-offsets of a navigator program is decreasing the time required bythe medical staff and office support staff in trying to support patients who need help throughthe complex medical system. Due to the heterogeneity of care settings involved, it is notpossible to track these offsets directly. We will explore the impact of offsets, based on timenavigators spend with patients, in sensitivity analyses.

Direct medical care related to navigation (e.g., screening tests and care related to follow-upof abnormal tests) will be assessed based on the routine core data elements collected by thePNRP and valued use representative reimbursement rates, such as regionally adjustedMedicare payments. Longer term costs, such as lifetime costs related to cancer treatment,will be estimated based on the stage at diagnosis, using published sources. 37 Navigators’time costs are likely to be the most significant program cost. Time costs will varysubstantially depending on training (e.g. professionals vs. laypersons), the complexity of thecare system, and the needs of the target population. Time spent by volunteer navigators isnot “free” and should be valued as the opportunity cost of those persons, given other optionsfor spending their time. Time costs for professionals can be valued based on their wages.Valuing time costs for volunteers can be more difficult. For persons who are employed, timeis typically valued based on their wages or the prevailing national wage rates for those of theindividuals’ age and gender. For those who do not work for pay (e.g., homemakers or retiredpersons), there is no generally agreed upon method, but most base costs on national wagesurveys.27 Using navigator logs, the PNRP will collect self-reported information on the timespent by navigators in direct contact with patients and in activities required for coordinationof care.

In the process of seeking care, patients incur costs which may be significant barriers toaccessing care in the first place.38 Patient costs can be evaluated using patient logs, or if thisis infeasible, by estimating time and associated expenses when traveling to specific services.Although the PNRP will not collect patient log data, navigator logs will include informationon the provision of these patient services, including transportation and child care costs.Patient time costs will be valued using census region specific wage rates for individuals thatmatch the age and sex of the patient population.

It is important to separate research-related costs from intervention costs. For the PNRPevaluations, research costs will be identified from audits of research budgets during sitevisits with investigators (e.g., navigator time filling out study-related paperwork andcomplying with Institutional Review Board documentation). In practice, it can be difficult toseparate research from intervention costs, thus necessitating the documentation andreporting of assumptions made when there is uncertainty.

In cases where navigation influences the use of multiple cancer screening programs, we willdisaggregate costs to particular services (e.g., mammography) based on the patient andnavigator diaries. If feasible, we will also estimate the cost-effectiveness of a bundle ofservices (e.g., mammography + pap smear + colorectal cancer screening) compared to usualcare.

Perspective and Time HorizonIn cost-effectiveness analysis, perspective refers to the point of view taken for evaluating theimpacts and costs of the study. The societal perspective is favored for cost-effectivenessanalysis where public health issues are under evaluation, 27 and is particularly important fornavigation programs, since the resources for navigators may come from one source (e.g.,



NIH


NIH


NIH


foundations, government programs, hospitals), while payment for medical care may comefrom another (e.g., Medicaid). As discussed above, navigation programs have short and longterm impacts. Thus, the cost effectiveness of navigation programs is best estimated over theentire period that the program is expected to influence costs and outcomes. The relevanttime horizon for navigation programs that assist patients with evaluation of abnormalfindings is the time from the initial point of detection of abnormal findings to theirresolution. For navigation programs that change care such that longer-term endpoints areaffected (e.g., survival), this implies using a lifetime time horizon. Because the PNRPprogram will only observe participants over a 4 to 5 year horizon, evaluating costeffectiveness will require simulation modeling to estimate the lifetime impact of navigationon populations.

Uncertainty AnalysisOne-way sensitivity and multi-way uncertainty analyses can identify factors that mostsubstantially influence the cost effectiveness of the programs.39 One way sensitivityanalysis is a process of varying individual parameters across a range, then re-calculating thecost-effectiveness ratio. This gives a sense of the relative influence of individual factors(e.g., the hourly wage of navigators) on overall cost-effectiveness of the program. Multi-wayanalysis is a process of varying all parameters simultaneously such that a distribution orconfidence interval can be derived around the point estimate of cost effectiveness.

Particular attention should be paid to the impact of varying assumptions regarding costs,quality of life, and survival for the usual care (non-navigator) group. The comparison or“usual care” group in some PNRP studies utilizes historical data from the period prior tonavigation or convenience samples from comparable communities that are not involved inthe PNRP; few use randomized controlled trials [Table 1]. Navigator program-specificfactors that should be considered for sensitivity analyses include patient time, type ofnavigator used, ranges of time to navigate different sub-groups of patients, and the basis fortime costs (local vs. national, average or race-specific wages, etc).

ConclusionsIt is rare for an economic evaluation to be free of conceptual and/or practical challenges, andcost-effectiveness analysis of cancer patient navigation is no exception. In this report, weoutline several special conceptual challenges to evaluating navigation interventions, as wellas many practical issues of data collection, instrument choice, and cost measurement. Wehave outlined several issues related to assessing costs and effectiveness in navigationprograms, as well as methods PNRP investigators will take to identify them. Although it ispossible to derive nationally representative estimates of cost effectiveness for particularprograms, many navigation programs are tailored to specific local situations, and thus alsomerit evaluation of economic value in a local context. However, we do not know ifnavigation will translate into improved cancer survival, and if it will improve theeffectiveness of cancer care at a reasonable cost (i.e., be cost-effective).40–43 Thus, theprocess of defining processes, costs, and outcomes that is part and parcel of cost-effectiveness analysis can also provide valuable information for local decision makersallocating limited health resources to navigation programs.

AcknowledgmentsFunding provided by the National Cancer Institute, via the Center to Reduce Cancer Health Disparities, throughContract 263-FQ-612391. Also supported by NIH grants: U01 CA116892; U01 CA117281; U01 CA116903; U01CA116937; U01 CA116924; U01 CA116885; U01 CA116875; U01 CA116925; and American Cancer Societygrant SIRSG-05-253-01. Dr. Mandelblatt’s work is also supported in part by NCI grants U01 CA88283 and KO5CA96940.



NIH


NIH


NIH


PNRP PIs and Affiliation

Charles L. Bennett, M.D., Ph.D., M.P.P., The Robert H. Lurie Comprehensive Cancer Center, Feinberg School ofMedicine, Northwestern University, Chicago, IL

Elizabeth Calhoun, Ph.D., Health Policy and Administration, University of Illinois at Chicago, Chicago, IL

Donald J. Dudley, M.D., University of Texas Health Science Center, San Antonio, San Antonio, TX

Kevin Fiscella, M.D., M.P.H., University of Rochester Medical Center, Rochester, NY

Karen M. Freund, M.D., M.P.H., Tracy Battalgia, M.D., M.P.H. (Co-PI), Boston University Medical Center,Women’s Health Research Unit, Boston, MA

Victoria Warren Mears, Ph.D., Northwest Portland Area Indian Health Board, Portland, OR

Electra D. Paskett, Ph.D., Marion N. Rowley Professor of Cancer Research, Division of Epidemiology, College ofPublic Health, Associate Director for Population Sciences, Comprehensive Cancer Center, Ohio State University,Columbus, OH

Steven R. Patierno, Ph.D., The George Washington University Cancer Institute, Washington, DC

Peter C. Raich, M.D., F.A.C.P., Denver Health & Hospital Authority, Denver, CO

Richard G. Roetzheim, M.D., M.S.P.H. H., Lee Moffitt Cancer Center & Research Institute, Tampa, FL

CRCHD PNRP Directors, Roland Garcia, Ph.D., Mary Ann Van Duyn, Ph.D., Emmanuel Taylor, Dr.PH

NOVA, Amanda Greene, Ph.D., M.P.H., R.N. (Project Manager), Paul Young, M.P.H., M.B.A. (Project Manager)

References1. Strzelczyk JJ, Dignan MB. Disparities in adherence to recommended followup on screening

mammography: interaction of sociodemographic factors. Ethn Dis. 2002; 12(1):77–86. [PubMed:11913611]

2. Gwyn K, Bondy ML, Cohen DS, et al. Racial differences in diagnosis, treatment, and clinical delaysin a population-based study of patients with newly diagnosed breast carcinoma. Cancer. 2004;100(8):1595–604. [PubMed: 15073845]

3. Center to Reduce Cancer Health Disparities. National Cancer Institute, National Institute of HealthPublication No. 07–6131. 2007. Economic costs of cancer health disparities.

4. Lurie JD, Welch HG. Diagnostic testing following fecal occult blood screening in the elderly. J NatlCancer Inst. 1999; 91(19):1641–6. [PubMed: 10511591]

5. Baig N, Myers RE, Turner BJ, et al. Physician-reported reasons for limited follow-up of patientswith a positive fecal occult blood test screening result. Am J Gastroenterol. 2003; 98(9):2078–81.[PubMed: 14499791]

6. Bickell NA, Wang JJ, Oluwole S, et al. Missed opportunities: racial disparities in adjuvant breastcancer treatment. J Clin Oncol. 2006; 24(9):1357–62. [PubMed: 16549830]

7. Griggs JJ, Sorbero ME, Stark AT, Heininger SE, Dick AW. Racial disparity in the dose and doseintensity of breast cancer adjuvant chemotherapy. Breast Cancer Res Treat. 2003; 81(1):21–31.[PubMed: 14531494]

8. Shavers VL, Harlan LC, Stevens JL. Racial/ethnic variation in clinical presentation, treatment, andsurvival among breast cancer patients under age 35. Cancer. 2003; 97(1):134–47. [PubMed:12491515]

9. Freeman HP, Muth BJ, Kerner JF. Expanding access to cancer screening and clinical follow-upamong the medically underserved. Cancer Pract. 1995; 3(1):19–30. [PubMed: 7704057]

10. Center to Reduce Cancer Health Disparities. Patient Navigation Research Program (PNRP)Overview. 2007 [[Accessed April 11, 2007]]. Available fromhttp://crchd.cancer.gov/pnp/pnrp-overview.html



NIH


NIH


NIH


http://crchd.cancer.gov/pnp/pnrp-overview.html

11. Psooy BJ, Schreuer D, Borgaonkar J, Caines JS. Patient navigation: improving timeliness in thediagnosis of breast abnormalities. Can Assoc Radiol J. 2004; 55(3):145–50. [PubMed: 15237774]

12. Farber, JM.; Deschamps, M.; Cameron, R. Canadian Breast Cancer Initiative, Health Canada.2002. Investigation and Assessment of the “Navigator Role” in Meeting the Information,Decisional and Educational Needs of Women with Breast Cancer in Canada.

13. Battaglia TA, Roloff K, Posner MA, Freund KM. Improving follow-up to abnormal breast cancerscreening in an urban population. A patient navigation intervention Cancer. 2007; 109(2 Suppl):359–67.

14. Wells KJ, Battaglia TA, Dudley DJ, et al. Patient navigation: State of the art or is it science?Cancer. 2008

15. Freeman HP. Patient navigation: A community centered approach to reducing cancer mortality. JCancer Educ. 2006; 21(1 Suppl):S11–4. [PubMed: 17020496]

16. Freeman HP, Wasfie TJ. Cancer of the breast in poor black women. Cancer. 1989; 63(12):2562–9.[PubMed: 2720605]

17. Oluwole SF, Ali AO, Adu A, et al. Impact of a cancer screening program on breast cancer stage atdiagnosis in a medically underserved urban community. J Am Coll Surg. 2003; 196(2):180–8.[PubMed: 12595043]

18. Evidence report and evidence based recommendations: cancer prevention and treatmentdemonstration for ethnic and racial minorities. Centers for Medicare and Medicaid Services, andThe Schneider Institute for Health Policy and the Heller School for Social Policy andManagement, Brandeis University, 2003.

19. Smedley, BD.; Stith, AY.; Nelson, AR. Unequal treatment: confronting racial and ethnic disparitiesin health care. Washington, DC: National Academy Press; 2002.

20. Weinrich SP, Boyd MD, Weinrich M, Greene F, Reynolds WA Jr, Metlin C. Increasing prostatecancer screening in African American men with peer-educator and client-navigator interventions. JCancer Educ. 1998; 13(4):213–9. [PubMed: 9883780]

21. Freeman HP, Chu KC. Determinants of cancer disparities: barriers to cancer screening, diagnosis,and treatment. Surg Oncol Clin N Am. 2005; 14(4):655–69. v. [PubMed: 16226685]

22. Dohan D, Schrag D. Using navigators to improve care of underserved patients: current practicesand approaches. Cancer. 2005; 104(4):848–55. [PubMed: 16010658]

23. Christie J, Itzkowitz S, Lihau-Nkanza I, Castillo A, Redd W, Jandorf L. A randomized controlledtrial using patient navigation to increase colonoscopy screening among low-income minorities. JNatl Med Assoc. 2008; 100(3):278–84. [PubMed: 18390020]

24. Ferrante JM, Chen PH, Kim S. The effect of patient navigation on time to diagnosis, anxiety, andsatisfaction in urban minority women with abnormal mammograms: a randomized controlled trial.J Urban Health. 2008; 85(1):114–24. [PubMed: 17906931]

25. Ell K, Vourlekis B, Lee PJ, Xie B. Patient navigation and case management following an abnormalmammogram: a randomized clinical trial. Prev Med. 2007; 44(1):26–33. [PubMed: 16962652]

26. Braithwaite RS, Meltzer DO, King JT Jr, Leslie D, Roberts MS. What does the value of modernmedicine say about the $50,000 per quality-adjusted life-year decision rule? Med Care. 2008;46(4):349–56. [PubMed: 18362813]

27. Gold, MR.; Siegal, JE.; Russel, LB.; Weinstein, MC. Cost-Effectiveness in Health and Medicine.New York: Oxford University Press; 1996.

28. Freund KM, Battaglia TA, Calhoun E, et al. National Cancer Institute Patient Navigation ResearchProgram: methods, protocol, and measures. Cancer. 2008; 113(12):3391–99. [PubMed: 18951521]

29. Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB. Recommendations of the Panel onCost-effectiveness in Health and Medicine. Journal of the American Medical Association. 1996;276(15):1253–8. [PubMed: 8849754]

30. The EuroQol Group. EuroQol--a new facility for the measurement of health-related quality of life.Health Policy. 1990; 16(3):199–208. [PubMed: 10109801]

31. Earle CC, Chapman RH, Baker CS, et al. Systematic overview of cost-utility assessments inoncology. J Clin Oncol. 2000; 18(18):3302–17. [PubMed: 10986064]



NIH


NIH


NIH


32. Du XL, Fang S, Coker AL, et al. Racial disparity and socioeconomic status in association withsurvival in older men with local/regional stage prostate carcinoma: findings from a largecommunity-based cohort. Cancer. 2006; 106(6):1276–85. [PubMed: 16475208]

33. Ward E, Jemal A, Cokkinides V, et al. Cancer disparities by race/ethnicity and socioeconomicstatus. CA Cancer J Clin. 2004; 54(2):78–93. [PubMed: 15061598]

34. Mandelblatt JS, Cullen J, Lawrence WF, et al. Economic evaluation alongside a clinical trial ofpsycho-educational interventions to improve adjustment to survivorship among patients withbreast cancer. J Clin Oncol. 2008; 26(10):1684–90. [PubMed: 18375897]

35. Polsky D, Mandelblatt JS, Weeks JC, et al. Economic evaluation of breast cancer treatment:considering the value of patient choice. J Clin Oncol. 2003; 21(6):1139–46. [PubMed: 12637482]

36. Mauskopf JA, Paul JE, Grant DM, Stergachis A. The role of cost-consequence analysis inhealthcare decision-making. Pharmacoeconomics. 1998; 13(3):277–88. [PubMed: 10178653]

37. Yabroff KR, Lamont EB, Mariotto A, et al. Cost of care for elderly cancer patients in the UnitedStates. J Natl Cancer Inst. 2008; 100(9):630–41. [PubMed: 18445825]

38. Yabroff KR, Davis WW, Lamont EB, et al. Patient time costs associated with cancer care. J NatlCancer Inst. 2007; 99(1):14–23. [PubMed: 17202109]

39. Briggs AH. Handling uncertainty in cost-effectiveness models. Pharmacoeconomics. 2000; 17(5):479–500. [PubMed: 10977389]

40. Mandelblatt J, Freeman H, Winczewski D, et al. The costs and effects of cervical and breast cancerscreening in a public hospital emergency room. The Cancer Control Center of Harlem Am J PublicHealth. 1997; 87(7):1182–9.

41. Mandelblatt JS, Fryback DG, Weinstein MC, Russell LB, Gold MR. Assessing the effectiveness ofhealth interventions for cost-effectiveness analysis. Panel on Cost-Effectiveness in Health andMedicine J Gen Intern Med. 1997; 12(9):551–8.

42. Russell LB, Gold MR, Siegel JE, Daniels N, Weinstein MC. The role of cost-effectiveness analysisin health and medicine. Panel on Cost-Effectiveness in Health and Medicine Jama. 1996; 276(14):1172–7.

43. Zhou X-HA, Ramsey SD. Assessing the equality of means of healthcare costs. Expert Review ofPharmacoeconomics & Outcomes Research. 2003; 3(5):561–67. [PubMed: 19807390]



NIH


NIH


NIH


Figure 1. Conceptual model of patient navigator intervention vs. usual care*Examples may include persons eligible for cancer screening procedures or those withcancer who are eligible for treatment.



NIH


NIH


NIH


NIH


NIH


NIH



Tabl

e 1

Patie

nt N

avig

atio

n R

esea

rch

Prog

ram

stud

y po

pula

tions

, set

ting,

and

pro

gram

s

PN S

ites

Can

cers

Popu

latio

nsN

avig

ator

Stud

y D

esig

nSe

tting

PN In

terv

entio

nC

ontr

ol

Bos

ton

Uni

vers

ityB

rC

vB

H U

6 O

Gro

up R

ando

miz

ed, c

ontro

lled

Com

mun

ity h

ealth

cen

ter

1,20

01,

200

Den

ver H

ealth

& H

ospi

tal A

utho

rity

Br

Cr

Pr

B H

U A

/PI A

I/A

N4.

5 LA

YR

ando

miz

edC

omm

unity

hea

lth c

ente

r,ho

spita

l87

087

0

Geo

rge

Was

hing

ton

Uni

vers

ity D

CB

rB

H U

1 N

P 1

SW 7

ON

on-r

ando

miz

ed, c

ontro

lled

Clin

ic80

080

0

H. L

ee M

offit

t Can

cer C

ente

rB

rC

rB

H U

3 LA

YG

roup

rand

omiz

ed, c

ontro

lled

Clin

ic a

nd h

ospi

tal

600

600

NW

Por

tland

Are

a In

dian

Hea

lthB

oard

Br

Cv

Cr

Pr

AI/A

N3

RN

1 L

AY

Non

-ran

dom

ized

, con

trolle

dC

linic

650

650

Uni

vers

ity o

f Illi

nois

at C

hica

go/

Nor

thw

este

rn U

nive

rsity

, Chi

cago

Br

Cv

Cr

Pr

B H

U2

SW, 5

LA

YR

ando

miz

ed, c

ontro

lled

Com

mun

ity h

ealth

cen

ters

,cl

inic

s, ho

spita

l2,

500

2,50

0

Uni

vers

ity o

f Roc

hest

er, N

YB

rC

rB

U3

LAY

Ran

dom

ized

, con

trolle

d (p

atie

nt)

Hos

pita

l40

040

0

Uni

vers

ity o

f Tex

as H

ealth

Sci

ence

Cen

ter S

an A

nton

ioB

rC

vB

H U

4 PR

O, 2

RN

, 2 S

WN

on-r

ando

miz

ed, c

ontro

lled

Clin

ic70

070

0

Ohi

o St

ate

Uni

vers

ity, C

olum

bus

Br

Cv

Cr

B H

U3

LAY

Gro

up ra

ndom

ized

, con

trolle

dC

linic

4,25

84,

258

Tota

ls11

,978

11,9

78

KEY

Can

cers

: Br=

Bre

ast,

Cv=

Cer

vix,

Cr=

Col

orec

tal,

Pr=P

rost

ate

Popu

latio

ns: B

=Bla

ck, H

=His

pani

c, A

/PI=

Asi

an &

Pac

ific

Isla

nder

, AI/A

N=A

mer

ican

Indi

an/A

lask

a N

ativ

e, U

=Und

erse

rved

Nav

igat

or: A

PN=A

dv P

ract

ical

Nur

se (N

P, n

urse

clin

icia

ns, p

hysi

cian

ass

ista

nt),

RN

=Reg

iste

red

Nur

se, L

PN=L

icen

sed

Prac

tical

Nur

se, S

W=S

ocia

l Wor

ker,

Lay=

Lay

or C

omm

unity

Wor

ker,

PRO

=Pro

mot

oras

, O=O

ther


NIH


NIH


NIH



Table 2

Cost consequence analysis sample table, with specific elements of interest in navigator interventions

Costs

Training Costs [Ctraining(program]

• Initial training

• Training replacements and additional navigators

Costs-Navigation Program [Cnavigator(program)]

Fixed Costs-Navigator Program

• Costs associated with developing navigator-related materials (e.g. pamphlets, telephone scripts)

• Allocated fixed operation costs (Office space leasing, telephone, furniture, etc.)

Variable costs-Navigator Program

• Time spent in navigation (travel, meeting with patients, documentation)

• Travel associated costs

Variable direct nonmedical costs-all patients [Cnonmed(program) and Cnonmed(usual care)]

• Patient time costs seeking treatment

• Travel-associated costs

Variable direct medical costs-patients [Cmedical(program) and Cmedical(usual care)]

Outcomes

• Time from abnormal screening test or suspicious finding to diagnosis

• Time from diagnosis to initial therapy

• Time from initial therapy to resolution (end of initial therapy including therapeutic combinations such as surgery pluschemotherapy)

• Percentage of patients receiving initial therapy (surgery, chemotherapy, radiation therapy)

• Percentage completing therapy

• Satisfaction with care

• Quality of life during care

• Quality of life following care

• Survival (years of life)*

• Quality adjusted survival (QALY)*

*Modeled


NIH


NIH


NIH



Table 3

Unique challenges to evaluating cost effectiveness of navigation programs

• Relationship between navigation and endpoints (costs, survival, QALY) is non- linear

• Content (and costs) of navigation interventions are variable due to site-specific program needs

• Confounding between need for navigation and stage, mortality end points

• Difficulty in allocating costs and effects over multiple cancers

• Short-term intervention outcomes (e.g., distress) do not map easily to QALYS

• Difficulty collecting uniform data across sites and at relevant time points (e.g. time costs)

• Difficulty detecting the impact of modest reductions in diagnostic or treatment delays on mortality

• Personal characteristics of navigators (difficult to measure) may influence program effectiveness


Evaluating the cost-effectiveness of cancer patient navigation programs: Conceptual and practical issues

Documents