Consumers' Use of the Internet for Health Insurance

Objectives: We examined consumers' search for informationabout health insurance choices and their use of the Internet for thatsearch and to manage health benefits.

Study Design: We surveyed a random sample of more than4500 individuals aged 21 years and older who were members of asurvey research panel during December 2001 and January 2002.

Methods: The survey included questions about searching forhealth insurance information in 3 health insurance markets:Medicare, individual or nongroup, and employer-sponsored group.We also asked questions about use of the Internet to manage healthbenefits. We tabulated means of responses to each question bymarket and tested for independence across demographic groupsusing the Pearson chi-square test.

Results: We identified important differences across and withinmarkets in the extent to which people look for information abouthealth insurance alternatives and the role of the Internet in theirsearch. Although many individuals were unaware of whether theiremployer or health plan provided a website to manage health benefits,those who used the sites generally evaluated them favorably.

Conclusions: Our results suggest that the Internet is an importantsource of health insurance information, particularly for individualspurchasing coverage individually in the nongroup andMedicare markets relative to those obtaining coverage from anemployer. In the case of Medicare coverage, studies focusing onbeneficiaries' use of Internet resources may underestimate theInternet's importance by neglecting caregivers who use theInternet. Many individuals may be unaware of the valuableresources available through employers or health plans.

(Am J Manag Care. 2004;10:609-616)

Although estimates of the extent to which consumersturn to the Internet to obtain informationabout health and healthcare varyconsiderably, even low-end estimates suggest that theInternet is an important source of consumer healthinformation.^1-5 To date, however, analyses of Internetapplications for consumer health has focused on theextent to which people use the Internet for informationabout health and healthcare. The role of the Internet inthe relationship between consumers and health planshas received considerably less attention. Nevertheless,the potential impact of Internet applications in this areais large. Health insurance is a complex and information-intensiveproduct, requiring extensive coordinationamong consumers, patients, insurers, and providers. Intheory, consumers, purchasers, and health plans couldbenefit substantially from technological innovation thateither improved the ability of consumers to obtaininformation about the types of health insurance availableto them or reduced the cost or increased the qualityof health insurance products.⁶

Many examples of these types of Internet-based innovationsalready exist. For Medicare beneficiaries, privatebrokerage sites focus primarily on individual supplemental"Medigap" coverage. Public and private not-for-profitorganizations have developed sites intended to presentcomprehensive information about the full range of choicesavailable to Medicare beneficiaries (eg, http://www.medicare.gov and http://www.calmedicare.gov) in responseto the increasing complexity of the choices availableto Medicare beneficiaries combined with evidence ofa lack of understanding among these individuals of theirchoices.⁷ Internet-based brokerage sites serve the individualand small group markets, potentially reducing thehigh loading costs and consumer search costs traditionallyconsidered major drawbacks of this market.⁸ Thesesites frequently provide information about the healthinsurance options and premiums available to an individual,often customized to his or her personal characteristicssuch as age, sex, health status, and geographiclocation. In the employment-based market, both employersand private not-for-profit organizations offer healthplan comparison information over the Internet. Privatenot-for-profit organizations also offer information aboutthe quality and coverage of health plans either in specificmarkets or nationwide (eg, Healthscope by the PacificBusiness Group on Health and the National Commissionfor Quality Assurance).

The Internet may also reduce the costs of offering andmanaging health benefits. Employers can provide theirown Web site, with a range of functionality, or can outsourcebenefits administration to online companies thathandle many of the administrative functions.⁹ Moresophisticated employer sites offer dynamic functionssuch as online enrollment and claims processing. A 1999survey of employers found that one third of employerswith 500 or more employees and two thirds of employerswith 20 000 or more employees used an Internet orintranet to manage health benefits, although interactiveapplications, such as online enrollment, were less common.¹⁰ Similar products are available directly fromhealth insurers and health plans, offering features suchas provider network and drug formulary searches, informationabout coverage, health information, and prescriptiondrug refills.

In this study, we examined the extent to which consumerssearch for information about health insuranceand the role of the Internet in their search. We alsoexamined the extent to which individuals use theInternet to manage health benefits. Despite the emergenceof new technologies in this area, virtually noempirical evidence exists on the extent to which thesetechnologies are used by consumers. Studies of theMedicare market have focused on the extent to whichMedicare beneficiaries use the Internet resources availablefrom the Centers for Medicaid and MedicareServices (CMS). Approximately 3% of beneficiaries usedthe CMS site in February 2001, although a survey of siteusers indicated that families and friends of beneficiariesas well as healthcare professionals represent a large proportionof the users of the site.¹¹

METHODS

Participants

We surveyed a random sample of adults in the UnitedStates on their search for information about healthinsurance and their use of the Internet for healthinsurance applications. We drew our sample from anInternet-enabled research panel of more than 60 000households developed and maintained by KnowledgeNetworks, a survey research firm. Panel members wererandomly selected to receive free Internet accessthrough MSN WebTV in exchange for periodically participatingin short surveys. At the time of our survey,the panel acceptance rate was 41%, and the panel attritionrate was 14%.

From the Knowledge Networks panel, we randomlyselected 6601 individuals 21 years of age and older toreceive our survey, including 3967 individuals from thegeneral population and an oversample of 2634 individuals50 years and older. The survey was fielded betweenDecember 1, 2001, and January 21, 2002. The totalresponse rate was 68.3%, with a higher rate among theelderly (73.2%) than the general population (65.1%).Item nonresponse rates for individual questions used inthis analysis varied by question, but were very low,ranging from 0% to 8%.

We performed extensive analyses of the extent towhich our sample was nationally representative andfound that estimates of population demographics, theprevalence of health conditions, and provider utilizationfrom our data were consistent with those from otherhigh-quality population-based surveys.¹² Nonetheless, thepossibility remains that persons who chose to participatein the panel differed from those who declined based ontheir interest in or ability to use the Internet. In previousstudies, we restricted our analysis to the subset of surveyrespondents indicating they had Internet access prior tojoining the panel to ensure that our estimates were representativeof Internet users in the United States.¹ In thisstudy we did not make this restriction for 2 reasons. First,our study provides estimates of the extent to which individualslook for health information from any source.Restricting the sample to Internet users would havebiased this estimate. Second, this strategy allowed us toinclude subpopulations that traditionally are less likely tohave Internet access on their own, such as individualswith low income or levels of formal education, becauseeveryone on the panel had at least a minimum level ofInternet access provided by Knowledge Networksthrough WebTV. (Although some survey respondentsmay have been limited to MSN WebTV as their onlysource of Internet access, others had access through acomputer.) Fifty-three percent of the study sample hadInternet access prior to joining the panel.

Survey Questions

Our survey questions addressed 2 study domains:searching for health insurance information and managinghealth benefits; and 3 health insurance markets:Medicare, nongroup, and employer-sponsored. We alsoasked respondents detailed information about theirhealth insurance status, which we used to identifyappropriate subsamples for particular analyses.

For the Medicare market, we asked all survey respondents,regardless of age or health insurance status,about their search for information. Our hypothesis wasthat many beneficiaries looking for information turn toyounger relatives or friends, who are likely to use theInternet in their search. For the nongroup market, weasked questions about searching for information toeveryone younger than 65 years (n = 3303), to capturesearches for either oneself or someone else. We hypothesized, however, that those without either employer-sponsoredor public coverage would be more likely to belooking for information on coverage through the nongroupmarket. The study sample for the employer-sponsoredmarket was defined as individuals indicatingthat they had a choice among employer-sponsoredplans, either through their own employer or someoneelse's (n = 1410).

We asked each eligible respondent whether he or shesearched for information in the past year, customizingthe questions by market. For the Medicare market,respondents were asked about their search for Medicaremanaged care plans or Medigap insurance. For the nongroupmarket, respondents were asked about "healthinsurance they could purchase themselves, not throughan employer or union." For the employer-sponsoredmarket, respondents were asked about health insuranceplans offered to them through an employer or union. Weasked individuals responding positively if they used theInternet in their search. To respondents indicating theyused the Internet, we asked whether they stronglyagreed, agreed, disagreed, or strongly disagreed with 5statements evaluating their search, ranging fromwhether they were able to find the information theywere seeking to whether the information influenced adecision about health insurance. Again, the questionswere customized by market and the exact wording ispresented in each table.

When asking about managing health benefits, we surveyedonly respondents younger than 65 years with privatecoverage, either employer-sponsored or nongroup.We asked those with employer-sponsored coverage (n =2338) questions about their employer's health benefitsWeb site and those with either employer-sponsored ornongroup coverage (n = 2472) questions about theirplan's Web site. The question format was similar acrosssamples, but the wording was customized to employeror health plan. We first asked respondents if theiremployer or health plan offered a Web site. For respondentsindicating yes, we asked them if they had visitedthat Web site. For those indicating they had visited theWeb site, we asked 3 questions that were intended todifferentiate among broad categories of functionality.Once again, the exact wording of the questions is presentedin the relevant tables.

Statistical Analysis

We calculated means of responses to each questionoverall and, in some cases, by insurance coverage usingpoststratification weights to adjust for oversampling andsurvey nonresponse. The weighting corrected the distributionof respondents to match the known distributionof the US population based on age, sex, race, education,region, metropolitan residence, and veteran status,using the Current Population Survey as the standard.(Standard errors in the regressions were not adjustedfor the complex survey design. Many of our analyseswere based on relatively small subsamples of the surveypopulation, resulting in primary sampling units withvery small numbers of respondents, in many cases 1.When possible, we adjusted the standard errors fordesign effects and found that it had virtually no effect onthe statistical significance of our results.) We tested forindependence across groups using the Pearson chi-squaretest.

Table 1 presents the distribution of our sample bysex, self-reported health status, education, income, andage.

RESULTS

Consumer Search for Information AboutHealth Insurance

Medicare Market.

Overall, 6.7% of the study populationsaid they had looked for information about Medicaremanaged care or Medigap coverage during the past year(Figure). Among those looking for information, 18% usedthe Internet. The rate of information seeking aboutMedicare coverage was relatively low among youngerindividuals, just 4.4% and 3.6% for individuals youngerthan 40 years and 40 to 59 years, respectively. Not surprisingly,the rate of information seeking peaked at 64 to66 years (28%), the time period during which most individualsbecome eligible for Medicare coverage. Informationseeking declined among older adults (15.4%).

Although persons 64 years and older were morelikely than younger individuals to look for informationabout Medicare alternatives, younger individuals weremore likely to use the Internet for their search(Figure). Five percent of individuals 67 years and olderused the Internet compared with 33% of those youngerthan 40 years. Individuals younger than 64 years representedmore than half of those seeking informationabout Medicare and more than 80% of those who usedthe Internet (results not shown), providing support forour hypothesis that younger adults are an importantaudience for information about Medicare coverage.

Source of insurance coverage was associated withinformation seeking among the elderly population, withMedicare managed care enrollees (34%) and Medicarebeneficiaries with Medigap coverage (26%) most likely toreport looking for information. Seventeen percent ofMedicare beneficiaries with supplemental coverage froman employer reported looking for information on supplementalcoverage or Medicare managed care plans.Beneficiaries with employer-sponsored supplementalcoverage were the most likely to usethe Internet in their search. This findingmay be because of either greaterfamiliarity among these individualswith the Internet or an important roleof employers in providing or directingretirees to Internet-based resources.Twenty-seven percent of elderly Medicarebeneficiaries indicated "othersource of supplemental coverage" orthat they did not know what type ofsupplemental coverage they had, indicatingsome difficulty among this populationin responding to questions aboutthe source of their insurance coverage.Thus, our results by insurance statusmay reflect a subset of beneficiarieswith greater knowledge about issues ofinsurance coverage in general.

Nongroup Market.

Sixteen percentof people younger than 65 yearsreported looking for information aboutcoverage in the nongroup market, and 27% of these peoplesearched the Internet (Table 2). As hypothesized,people with nongroup coverage and the uninsured weremore likely to look for information about coverage inthe nongroup market than those with employer-sponsoredor public insurance, although they were less likelythan those with employer-sponsored coverage tosearch the Internet. Thirty-seven percent and 32% ofindividuals who reported no insurance or coverage purchasedin the nongroup market, respectively, reportedsearching for information about coverage in the nongroupmarket.

Employer-sponsored Market.

Although many peoplewith a choice among employer-sponsored plans lookedfor information before enrolling (67%), relatively fewsearched the Internet (16%) (Table 2).

Evaluation of Information Foundon the Internet

Overall, respondents favorably evaluated the informationthey found on the Internet by indicating thatthey either agreed or strongly agreed with 5 statementspositively evaluating their search. However, we notedimportant differences in the evaluations across the markets(Table 3). In the employer and nongroup markets,fewer respondents reported agreeing or strongly agreeingwith each statement as the items identified increasinglysubstantive effects. Respondents who searched theInternet for information about nongroup coverage wereless likely than those in the employer-sponsored marketto agree or strongly agree with each of the statements.Among people searching the Internet in the nongroupmarket, 75% indicated they were able to find the informationthey were looking for, whereas 58% reportedthat the information they found influenced a choiceabout health insurance. In the employer-sponsoredmarket, in contrast, 92% were able to find the informationthey were looking for, whereas 79% reported thatthe information they found influenced a decision abouthealth insurance. In the Medicare market, in contrast,people more frequently reported difficulty locating theinformation they were looking for, but favorably evaluatedthe information they found. Twenty-two percentdisagreed or strongly disagreed with the statementthat they were able to find the information they werelooking for. More than 90%, however, indicated thatthey were able to understand the information theyfound, that they learned something new from theinformation they found, and that the informationimproved their understanding of the insurance choicesavailable to them. Sixty-seven percent reportedthat the information they found influenced a choiceabout health insurance. Because this sample includedboth elderly and nonelderly individuals who usedthe Internet for the Medicare market, these resultscannot be generalized to Medicare beneficiaries.

Managing Health Benefits

Many people with private health insurance wereunaware of whether their employer or health planoffered a Web site related to their coverage. Almost half(49%) of individuals with employer-sponsored coveragedid not know if their employer offered a related Website, and nearly two thirds (64%) of those with any privatecoverage did not know if their plan had a Web site(Table 4). Statistical significance in this table refersto tests of independence across groups. Awareness ofemployer Web sites was similar between people with achoice of plans and those with no choice. However,individuals with a choice of plans were more likely toreport that their employer did offer a Web site. In thecase of health plan Web sites, people with a choice ofplans were more likely than those with no choice to beaware of whether the health plan offered a site. Awarenessof health plan sites was extremely low among thosewith nongroup coverage (71% did not know whethertheir health plan offered a site).

Relatively few people reported visiting their employer'sor health plan's Web site. Among people indicatingtheir employer or plan had a Web site, 29% and 33%reported visiting the Web site, respectively (Table 4).Individuals with a choice among employer-sponsoredplans were more likely to visit their employer's Web site.

Individuals who visited an employer or plan Web sitegenerally evaluated their functionality favorably,although the sites more often contained coverage andhealth information than tools for managing health benefits.Eighty-eight percent of individuals who visited theiremployer's site indicated that it provided informationthat helped them understand their coverage, whereasonly 70% indicated that they used their employer's Website to manage their health benefits (Table 5). Theresults were similar for plan sites. When assessingemployer Web sites, individuals with a choice amongplans were more likely than those with no choice to indicatethat the site provided information that helped themunderstand their coverage (91% compared with 74%) butequally likely to indicate that the site allowed them tomanage their health benefits (71% compared with 67%).

DISCUSSION

Our study identified important differences both acrossand within markets in the extent to which people look forinformation about health insurance and the role of theInternet in their search. We found that the nonelderlycomprised a large proportion of the population seekinginformation about Medicare and its alternatives, suggestingthat family and friends, many of whom assume caregivingduties, play an important role in findinginformation and assisting older individuals with their coverage choices. This finding is consistent with qualitativeresearch indicating that caregivers play an importantrole in assisting the elderly in health insurance decisionmaking.¹³ Currently, these younger information seekersare more likely to search the Internet, implying that themost frequent users of Web sites targeted to Medicarebeneficiaries are nonelderly. Although we found littleevidence of difficulty in understanding or using the information,we were unable to determine whether these generalfindings for Medicare coverage were true for bothyounger and older populations due to small sample sizes.Nevertheless, our results indicate that the nonelderlypopulation is an important target for information aboutMedicare coverage and its alternatives, both on and offthe Internet. Although further research to identify thebarriers the elderly face in using the Internet to obtainthis type of information would be helpful, these barriersare likely to diminish over time as Internet usage for alltypes of applications continues to increase among thispopulation and the aging of younger cohorts creates anelderly population with greater experience with thesetechnologies.

Our results also suggested that employers play animportant role in identifying valuable information andmaking it accessible to workers and retirees. Amongindividuals aged 65 years and older, the most frequentusers of the Internet for health insurance choices werethose with employer-sponsored retiree benefits. Theemployer's role in the provision of health benefits thereforeappears to facilitate the use of the Internet as aninformation source for older Americans. More generally,individuals searching the Internet for information in theemployer-sponsored market evaluated the informationthey found more favorably than those searching forinformation in the nongroup and Medicare markets.

We found that many people with a choice amongemployer-sponsored plans looked for information beforeenrolling, but most used "traditional" sources and relativelyfew used the Internet. This finding may be becauseemployers offered both online and offline resources forworkers. Greater efforts by employers to direct individualsto information resources available on the Internet aswell as greater promotion of publicly available resourcesmay help individuals make these choices.

We found evidence that many people look for informationabout coverage in the nongroup market, particularlyindividuals without employer-sponsored orpublic coverage. Although many of these people turn tothe Internet, the information available there appears tobe less useful than that available for other types of coverage.Although it is difficult to determine whether thisis due to the quality of the information available or thequality of the choices available, this finding suggeststhat further study of the experiences of individualsshopping for coverage in the nongroup market, particularlytheir use of the Internet, may generate a betterunderstanding of the barriers to obtaining health insurancein the United States.

Finally, our results suggested that many people maynot be aware of potentially valuable resources availablethrough their employer or health plan to help themunderstand their coverage, manage their health benefits,or find health and healthcare information. Althoughawareness of these Web sites is low, site visitors tend toevaluate them favorably, perhaps because people aremore likely to be aware of good Web sites. Although wefound that plan and employer Web sites were equallyvaluable, this result could have been because of the difficultysome individuals may face in identifying theoperator of a particular site. For example, manyemployers may offer sites that enable covered workersto click through to the plan Web site.

Our survey was administered to individuals randomlychosen to receive Internet access in exchange for participatingin a survey research panel. Whereas somesurvey respondents had Internet access prior to joiningthe panel, others did not. In other work, we found thatthe use of the Internet for health applications by surveyrespondents who had Internet access prior to joiningthe panel was similar to estimates from nationally representativesurveys¹ and those who gained Internetaccess by joining the panel generally had lower rates ofuse of the Internet for health applications.¹⁴ This observationsuggests caution in interpreting our estimates asrepresenting the prevalence of national Internet use forthese types of applications. If our sample introducedany bias relative to a nationally representative sample,we likely would have overestimated the extent to whichindividuals used the Internet for health insurance orunderestimated the prevalence of use among the populationwho obtained Internet access on their own.

From the Department of Health Research and Policy, Stanford University School ofMedicine, Stanford, Calif (MKB, THW, LB); the Center for Health Policy, StanfordUniversity, Stanford, Calif (SJS); and the VA Palo Alto Health Care System, Palo Alto, Calif(THW).

Funding for this research was provided by the Department of Veterans Affairs, StanfordUniversity Office of Technology and Licensing, and the National Institute on Aging, Grant#AG17253. Dr Bundorf was also supported by grant KO2-HS11668-01 from the Agency forHealthcare Research and Quality.

A preliminary version of this report was presented at the 2002 Academy HealthConference.

Address correspondence to: M. Kate Bundorf, PhD, Department of Health Researchand Policy, Stanford University School of Medicine, HRP Redwood Building, Room 257,Stanford, CA 94205-5405. E-mail: bundorf@stanford.edu.

JAMA

JAMA

1. Baker L, Wagner TH, Singer S, Bundorf MK. Use of the Internet and e-mail forhealth care information: results from a national survey [published correctionappears in . 2003;290:334]. . 2003;289:2400-2406.

The Online Health Care Revolution: How theWeb Helps Americans Take Better Care of Themselves

2. Fox S, Rainie L, Horrigan J, et al. . Washington, DC: PewInternet and American Life Project; 2000.

Health Aff (Millwood)

3. Brodie M, Flournoy RE, Altman DE, Blendon RJ, Benson JM, Rosenbaum MD.Health information, the Internet, and the digital divide. .2000;19(6):255-265.

A Nation Online: How Americans Are Expanding TheirUse of the Internet

4. Victory NJ, Cooper KB. . Washington, DC: US Department of Commerce, NationalTelecommunications and Information Administrations, Economics and StatisticsAdministration; 2002:1-90.

Seeking Health Care Information: Most Consumers Still onthe Sidelines

5. Tu HT, Hargraves JL. . Washington, DC: Center for Studying Health Systems Change; 2003.

J Econ Perspect

6. Borenstein S, Saloner G. Economics and electronic commerce. .2001;15:3-12.

Health Aff (Millwood)

7. Hibbard JH, Slovic P, Peters E, Finucane ML, Tusler M. Is the informed-choicepolicy approach appropriate for Medicare beneficiaries? .2001;20(3):199-203.

Health InsurancePurchasing and Privacy Online for Individuals and Small Groups

8. Wilson KB, Goldman J, Hudson Z, Smith R, Mulkey M. . Oakland, Calif:California HealthCare Foundation; 2000:1-28. Available at:http://www.chcf.org/topics/view.cfm?itemID=12503. Accessed March 30, 2004.

E-Health Options for Business: Evaluating theChoices

9. Silow-Carroll S, Duchon L. . New York, NY: The Commonwealth Fund; 2002. Available at: http://www.cmwf.org/programs/insurance/silowcarroll_ehealth_508.pdf. Accessed March 30,2004.

Mercer National Survey of Employer-sponsored HealthPlans

10. William H, Mercer I. . Louisville, Ky: Mercer Human Resource Consulting LLC; 1999.

HealthCare Financ Rev

11. Goldstein E, Teichman L, Crawley B, Gaumer G, Joseph C, Reardon L.Lessons learned from the National Medicare & You Education Program. . 2001;23:5-20.

Validity of the Survey of Health andInternet and Knowledge Network's Panel and Sampling

12. Baker L, Bundorf M, Singer S, Wagner T. . Stanford, Calif: StanfordUniversity; 2003:1-13. Available at http://www.herc.research.med.va.gov/SHI%20appendix.pdf. Accessed August 2, 2004.

Health Care Financ Rev

13. Sofaer S, Kreling B, Kenney E, Swift EK, Dewart T. Family members andfriends who health beneficiaries make health decisions. .2001;23:105-121.

Med Care

14. Wagner TH, Bundorf MK, Singer SJ, Baker LC. Free Internet access, the digitaldivide, and health information. . In Press.