Environmental Scan - Health Risk Appraisals in Primary Care
Findings by Research Topic (continued)
Topic 3: How information derived from the use of has in other settings, such as worksite health promotion programs, is, or is not, shared with primary care providers
While health risk assessments (HAs) are used extensively among employer and health plan/health promotion groups (e.g., 241 of 453 large employers surveyed (53 percent) use financial incentives to encourage HA participation, according to Watson Wyatt Worldwide, 2008), HIPAA (Health Insurance Portability and Accountability Act of 1996)-compliant and privacy-conscious employers will not have access to information from their employees' HAs or readily be able to share HA reports with employees' primary care providers. For example, the Milwaukee Public School System collective bargaining agreement mandates HA response but forbids the HA vendor from sharing or discussing individual responses with medical providers or anyone else other than the respondents (Milwaukee Board of School Directors & Milwaukee Teachers' Education Association, 2006). Similarly, HA forms often state that all information provided will be kept confidential.
One approach used to deal with information sharing, for example in a major Medicaid demonstration of HA use by the elderly (Centers for Medicare and Medicaid Services, 2007), is to seek informed consent to share HA findings with a physician designated by the participant. An unknown number of employers also encourage employees to share the HA findings with their physician. For example, Galveston, Texas, requires employees to complete an HA and suggests they take the HA report to their "own physician for review and discussion" or "discuss it with a physician at the Galveston County Government Employees Healthcare Center" (Galveston County Government Texas, n.d.).
Linkage to the primary care setting are sometimes built-in when a research project collects HA data, especially if the project involves follow-up counseling with selected HA completers. As part of participant protection (possibly to meet Institutional Review Board requirements) or intervention planning, the research protocol often requires the counselor to refer completers to primary care, if selected response patterns indicate failure to receive medical intervention could endanger the participant (e.g., Boudreau et al. 1995, Ozminkowski et al. 2006, Goetzel et al. 2008, Baker et al. 2008, Pronk et al. 2002). The referral may be made by computer or by a counselor who interacts with the patients as part of the research protocols. But little is known about this process or its consequences. Articles resulting from these research projects describe this protocol in one short sentence or phrase; the volume and outcomes of these referrals are not reported.
The interviews confirmed these findings. Few HAs link with health promotion agencies or primary providers (P. Renner). Indeed, employers and physicians rarely communicate with each other (B. Bagley). Although aggregate-level data often are shared with employers and customers/clients for tactical and strategic planning, they typically are not shared with the employer's health care provider network (N. Pronk).
HAs present myriad confidentiality issues. One expert stated HA data at the individual level are not shared, in part due to HIPAA (N. Pronk). Another was more optimistic, suggesting HIPAA and other confidentiality requirements slow things down, but do not make information sharing impossible (R. Harmon). HAs that ask for a person's Social Security number impose a major, avoidable deterrent to response (R. Goetzel) as some respondents worry that the information provided could be used against them.
Another impediment to sharing HA data with physicians is many individuals may not have a regular doctor, may have multiple doctors, or may be unable to contact their primary health care professional. Information is rarely shared unless the patient brings the HA results to the physician (R. Goetzel).
Interview respondents described two efforts by health plans to administer HAs and use the results to improve physician management of preventive services:
Group Health Cooperative developed an evidence-based, clinically credible HA that is evolving over time (D. Grossman). Guideline changes at the health plan trigger changes in the tool. The HA is used as preparation for a preventive-services visit is considered an "omnibus screening tool" and is integrated into clinical care.
The HA score and interpretation go into reports that are given to the patient and the doctor, who receives a condensed summary. For example, if the patient's risk level for alcohol use merits outreach and follow-up, the recommendation goes to the physician, who is expected to act on it. Very few of the health plan's clinicians do alcohol screening, so this information is important. Some answers to the screener result in feedback to the patient, for the clinical team report, for the medical record, or for a "red flag" E-mail to the medical team. Other follow-up is available, such as appointments with a Lifestyle Coach.
It was a long process to obtain "buy-in" and get clinicians involved. Physician feedback was incorporated into the design of the tool, consumer feedback was gathered via focus groups, and changes were made based on their advice. The tool is designed to minimize user burden and the average time spent completing the HA is 12 to 15 minutes. The large majority of users have computer access, but the medical center also offers an HA kiosk. The insurer is working on a "Proxy Access" with nurses and allied health professionals.
Health Partners Health Behavior Group in Minnesota connected their HA to the health care delivery system, but later discontinued the effort (N. Pronk). Issues that surfaced included liability concerns—patients had risks, but were not being followed up; and confidentiality—in sensitive cases, the legal department stopped the information from going anywhere. As part of the employer-sponsored program, data were being collected through the workplace, but physicians had no chance to follow up. Patients were not being seen in primary care settings. The employer did not see these data.
In this example, the health insurer ended the arrangement and is debating how to integrate the HA with primary care. Some approaches include: asking insured persons if they want their HA data sent to their physicians (obtaining permission), or telling them to bring the HA with them to their next physician visit. HA data at the individual level are not normally shared due to HIPAA regulations. However, data at the aggregate level often are shared with employers and insurers to be used for tactical and strategic planning.
Rubinstein et al. (2003) report that in the 1980s, five Consolidated Omnibus Budget Reconciliation Act (COBRA) Medicare Prevention Demonstrations and a sixth related demonstration program all included an HA component, but the HA's role "as a health promotion tool was overlooked inadvertently" in evaluating the demonstrations. The HAs were health plan initiatives. "Some projects provided the HRA information to participants' physicians or trained nurses without assessing whether the providers actually discussed the HRA with the participants."
The COBRA demonstration at Johns Hopkins University (German et al., 1995; Burton et al., 1995) used a model oriented to primary care linkage. Rubinstein et al. (2003) describe that program and its outcomes. Participants "received an explanatory letter and voucher for a visit without charge to their primary caregiver in Year 1 and Year 2. Physicians were asked to review health risks; provide counseling where appropriate; take a complete medical history; and include breast, pelvic and rectal exams in the physical exam. Lab tests and immunizations were also provided. After two years, the intervention group did not significantly differ from the control group on smoking, problem drinking, or sedentary lifestyle. However, the health of participants in the intervention group declined less compared to the control group, as measured by the Quality of Well-Being Scale. This difference was mostly due to a differential death rate between groups." Unfortunately, the impact of HA use in this demonstration was not evaluated.
The Panel discussed roles, linkages, and methods of integration. Discussion focused on measurable risks, standardization, and "closing the loop." One member pointed out that "we know much more than we did five years ago." Some said to "try different theories out and see which work." Another suggested not getting into the micro level, that "change happens at the macro level," and recommended looking at what works best. Working with current models and comparing them using Thomson Reuters and other large data sets could be a good beginning. Several participants urged "mining existing data" and looking for practice based research.
Transfer of HA information from other settings to primary care providers presents a number of challenges, including lack of standards, practicality, ownership, and confidentiality.
Standards—No standards have been developed, and each HA is its own entity, and is administered and interpreted differently. Current HAs vary widely, from tools identifying basic medical markers to much more sophisticated versions. They can be patient-centered, clinician-centered, or some combination of the two. Some have no face validity, while others have a strong scientific base. The Panel agreed that it is imperative to create standards and have defined fields. Additional study should focus on identifying core HA questions, assessing whether a standard HA instrument could work across primary care settings, and developing consistent, age-appropriate scoring of HA responses.
Practicality—Some panelists said that practicality was a bigger issue than confidentiality in sharing information from HAs. In order to effectively share this data, individuals would have to fill out detailed information about their providers, include all current contact information, and would need to specify which of many physicians should receive it. Many patients cannot identify their primary care provider, or do not have one. Employer health plans may know who gets care where, but that does not guarantee that the information can or will be shared due to privacy issues and the logistics of contacting individual practices. For example, providers' billing addresses may differ from their office addresses or a health plan may only have access to a P.O. boxes.
Ownership of the information—What information should be sent? Who should send the information to the provider? How can confidential information be protected? Consistent with the scan results, patients do not want everyone seeing their medical information, especially since HAs may identify high risk lifestyles or become a liability for employment. One potential solution is to have the patient be the conduit of the information, which also could solve many of the HIPPA concerns. The challenge is getting the patient to accept this role.
The medical record itself is undergoing changes, and becoming more publicly viewable than a written record. Some are linking information across all units of care, so all medical staff can read it. Hopefully, this will increase transparency and facilitate sharing.
Both key informants and Panel meeting attendees discussed the role of the Medical Home, which has evolved since its introduction by the American Academy of Pediatrics in 1967. The basic premise is continual care that is managed and coordinated by a personal physician with the right tools will lead to better health outcomes. The Medical Home could serve as the focal point through which all patients receive their preventive, acute, and chronic medical services. This approach assumes that patient-centered care will result in a basic shift in the relationship between patients and their primary care physicians. It requires a greater degree of personalized care coordination, access beyond acute care, and identification of key medical and community resources to meet the patients' needs. Additionally, the widespread use of information technology for care management and improved quality of care is critical. This model shifts from acute care to continuous comprehensive care and services.
The Medical Home can help standardize and promote best practices. Many participants believe that developing the Medical Home is a prerequisite for transfer of medical information, including health appraisal data from other settings to the primary care setting. Some said the health care system itself needs to be re-designed, that "we can't just shove HAs into the system we have now."
Priorities for Moving Ahead
First, the roles of all players must be defined. This includes the patients and their families, interpreters as needed, primary care physicians, nurses, community providers and organizations, employers, schools, vendors, government, insurers, and the research community. Second, prototypes that have been used should be identified and assessed for organization, accessibility, efficiency, and clinical and financial outcomes. Third, new models need to be developed and tested in real world settings.
The need to keep personal information confidential hampers information transfer from employer programs to primary care providers. Ongoing programs and randomized trials that incorporate strategies like informed consent or counseling to facilitate transfer, at a minimum should document the yield from these strategies and the use that primary care physicians make of the HA information. Other barriers include the lack of HA standards and common elements and access to provider identification.
What does it take to achieve behavior change? Evidence supporting HAs as interventions from the scan, the key informant interviews, and the Panel participants emphasizes that HAs alone do not change risk factors and that HA linkage/follow-up is required to achieve desired health outcomes. The HA is "an opportunity, but not the end point." Trust is the essential element in a one-to-two minute health promotion intervention prompted by HA findings. The HA information needs to go to someone the patient trusts, but many lack that kind of provider relationship.
Is there value in pursuing more knowledge in this area? Yes, participants definitely believed that it would be worth studying these issues further, even with significant barriers. The environmental scan and the meeting discussion emphasized how little integration is occurring, and the need to explore how to make integration happen. One participant remarked that linkage of HAs from the workplace or community setting to the clinical setting is very promising. Administering them in the work place with supportive behavior change is also quite promising. Another related that the Centers for Medicare and Medicaid Services (CMS) is testing a model that connects vendor-administered HAs to clinical care. Few people, however, have looked into creating an effective system.
Undisclosed proprietary scoring algorithms impose a further barrier to primary care use of HA risk scores. Before acting on scores, many providers will expect to know the underlying science. It also is unclear how well the scores measure disease or mortality risk, which is a question AHRQ is uniquely well-positioned to pursue.
Topic 4: How has are being used (in any setting) to provide care to individuals that integrates preventive services and care management for chronic conditions
Minimal information was available on HAs and chronic care. In a 2003 review, the RAND Corporation concluded "With the exception of studies involving older adults," the literature "provided limited evidence on which to evaluate the effectiveness of HRAs in vulnerable populations" (Rubinstein et al., 2003).
Yet the potential is recognized. An Institute of Medicine report entitled Promoting Health: Intervention Strategies from Social and Behavioral Research (Smedley et al., 1999) suggests that substantially improving prevention and management of chronic conditions requires integrating social, public health, and medical approaches. Rubinstein et al. (2003) conclude that "HRAs coupled with health promotion programs have the opportunity to be an important part of such integration, by identifying risks to health, providing recommendations tailored to an individual's specific medical and cultural context, and linking this with information on resources available within the community, such as senior centers." The interview respondents suggested focusing on care management services in HAs for people at high risk of chronic illnesses such as diabetes and heart disease.
A CMS (2007) demonstration program, based in part on the RAND report, may point the way. The 3-year demonstration started in 2008 at five Medicare providers: Health Dialog Services Corporation, Focused Health Solutions, Health Partners Health Behavior Group, Pfizer Health Solutions Inc., and StayWell Health Management. The demonstration excludes institutionalized and managed care patients. Each provider is linking with two Department of Health and Human Services' Aging and Disability Resource Centers and will deliver HAs and tailored feedback reports to 17,000 elderly plan members. Meeting participants reported that three demonstration providers had trouble enrolling subjects and only two providers are going forward. Each will serve 20,000 beneficiaries plus a comparison group. Participants will "receive health education and behavior change materials, and health coaching, provided on an ongoing basis using their preferred communication method, either through the mail, telephone, or internet. Participants will also have the option of having information shared directly with their physicians. In addition, participants will receive referrals to national and local programs, such as physical activity, falls prevention, smoking cessation, and other types of health promotion programs, or if needed, referrals to their physician for recommended clinical preventive services." Program goals include improving health and reducing avoidable health care utilization. Somewhat overlapping, the goals of chronic-illness care management are to prevent premature death and disability and improve quality of life (J. Mold, Z. Nagykaldi).
Rubinstein et al. (2003) found that the Healthtrac Program (Fries and McShane, 1998) was the only tailoring of an HA intervention to the needs of participants at high risk for health problems. "Developed by James Fries, Stanford University, this program, based on self-efficacy theory, consists of completing health risk questionnaires at six-month intervals" for most individuals and at three-month intervals for high-risk individuals. "Computer-based serial personal health risk reports are provided after each administration, along with 86 individualized recommendation letters, newsletters, self-management and health promotion books, and other program materials. The instrument contains 14 modules on various health risks and major chronic diseases…. Although results indicated an 11% decline in health risk scores (from baseline) at a 6-month follow-up, this study did not include a control group of high-risk individuals who did not receive the full Healthtrac Program" (Rubinstein et al., 2003).
Consistent with the scan results, the Panel said little research exists on HAs and chronic care. It is important to distinguish between disease management and care management (e.g., managing hypertension that leads to heart disease vs. managing the chronic illness). Many chronic diseases that clinicians spend the most time managing have the same risk factors prioritized in HAs. The HA should be integrated into chronic care management so the case manager can link people with the proper care and services.
Attendees discussed complex case management for people with very high risks. The HA is integrated into a system with a case manager and has follow-up efforts to ensure the person is connected to the proper care. However, as one participant pointed out, "we don't know how to manage the seriously multi-disease patient" and that clinicians are unsure what preventive services to use and how to set priorities for these patients. External resources are needed to help manage the needs of chronically ill individuals and participants thought that while it would be helpful for HAs to prioritize care "the science is not there yet."
Nico Pronk described his conceptual project that is looking at chronic conditions as preventable outcomes that fit into the HA process. It is looking at primary, secondary, and tertiary prevention to increase risk management.
Minimal research has been conducted on HA strategies for patients with chronic illnesses. It is clearly an opportunity for further exploration, with the initial focus probably on identifying preventive service needs and priorities for multi-disease patients. Both the patients and their providers tend to be focused on immediate needs, so research also needs to focus on effective ways to promote a more preventive orientation.
The Medicare demonstrations warrant monitoring. It might be desirable to work with CMS to review the data elements being collected to ensure that information needed by AHRQ and the U.S. Preventive Services Task Force (USPSTF) will be included.
Topic 5: Strategies that might improve the delivery and coordination of clinical preventive services when has are used, regardless of setting
The National Committee for Quality Assurance (NCQA) has worked on designing evaluation protocols for health-promotion vendors, and developing process and outcome measures regarding use of HA tools. NCQA is reviewing common industry practices and validated tools for workplace programs. The group is focusing on how to administer HAs, what to do about confidentiality, risk factors (e.g., BMI, smoking), motivational levels, HA completion rates, frequency of use and updates, and risk reduction (P. Renner).
In the clinical setting, both physicians and patients are being overwhelmed by possible interventions and recommendations. As Dr. Mold indicated, "We've gotten too good at it," with many more detailed guidelines and improved identification and testing. In theory, HAs could help pare down the information overload.
Collectively, interview respondents suggested the following strategies:
- Study the cultural and work environment that drives HA response rate, participation, and completion.
- Scale back the number and depth of questions if the response burden is too great.
- Incorporate questions into HAs about use of an age-appropriate evidence-based list of the highest ranking preventive services and about preventive care (e.g., self-report on immunization).
- Traditional HAs looked mostly at functional objectives such as staying in one's own home. A patient-centered focus on quality-of-life goals (e.g., who you are, what you want, life expectancy) seems desirable to add (J. Mold).
The group discussed strategies to improve the delivery and coordination of preventive services, including the return of the worksite-based clinic for primary care. Participants viewed this as an opportunity to develop a model of coordinated care, with some "sorting out" of roles (e.g., the provider, the nurse). The health plan would be a key partner to gain access to the patient base.
Participants promoted the team approach to reduce the burden on clinicians. Also, developing and using CPT codes to reimburse HA administration and follow-up would greatly increase usage. Panel members suggested this might occur through a government initiative. Redesigning systems, a major challenge, is likely needed to accommodate new ways of communicating and providing services.
Participants are interested in assessing if patients receive appropriate care for their conditions. One of the keys to this is developing a personalized wellness plan and adhering to it. Self-care management would be an important component. And, a measure of quality should be built in. This is not yet a mature area, and needs more exploration.
Panel participants disagreed whether to limit HA questions to areas where there are evidence based interventions. Some believed that all risk factors are modifiable. And even if interventions were not totally effective, new ones might be developed. Others said that HAs limited to risks with proven interventions would result in more concise tools, an advantage with patients.
Panel members agreed that all parties are affected by financial concerns, and that this is perhaps the main reason for stakeholders to be at the table. One member said "It all comes back to money." The bottom line for health care organizations as well as workplace settings is: What health outcomes are achieved for the dollar investment?
The interviewees and the meeting participants were in substantial agreement on this topic conceptually, but the meeting attendees offered more input on strategies. They suspect major systems redesign is needed to achieve desired health outcomes and coordination of preventive care and follow-up services. Studying existing models would be very informative. Participants agreed that using a team approach could reduce the burden on the clinician. Self-care management also deserves exploration. All agreed that cultural and work environments need to be studied for their impact on HA response rates, use, and user accessibility.
The NCQA work on HA protocols and measures seems important to monitor and may warrant AHRQ participation. The panel participants and interviewees felt strongly that more evaluation of HA outcomes and cost-effectiveness would be informative.
Topic 6: What are the gaps in knowledge and evidence regarding the use of has and how do has affect the delivery of effective preventive and care management services?
Interview respondents identified five gaps and issues:
- Better evidence is needed about HA use in primary care. What works in different practices? What do HA completion rates look like in different settings? How can linkage be accomplished effectively and efficiently? What are the key roles and who are the actors in a coordinated system of HAs and primary care follow-up?
- More evidence-based research/randomized trials are needed (Grossman), especially ones that focus on how HAs affect primary care outcomes and cost of care? The trials should pay attention to the influence on outcomes of differences in the way a design is implemented (Goetzel).
- Standardizing HAs requires better evidence about design. What is the optimal length of an HA? What is the most appropriate reading level? What are the critical items to ask on an age-appropriate HA (R. Soler)? How good are primary care electronic health records (EHRs) compared to HAs (e.g., how do the questions on smoking compare) (Harmon)? For ethnic and other special populations, within the primary care setting, what considerations arise in designing, promoting, and using HAs? How can we best handle confidentiality concerns? And critically, if standardized HAs for primary care were developed, would physicians use them?
- How frequently should HAs be repeated? Are data on outcomes valid if they are not longitudinal?
Panel participants saw a need for research on ways to integrate HAs into systems of care and for evaluation of outcomes of HA integration. Like the interview respondents, they also focused on the mechanics of HA design and administration. The latter discussion overlaps with the discussion under Topic 2 on HA design and response rates in primary care.
System Integration and Evaluation
A "best practices" review would be very informative. The first step is to study existing models—how they operate, what works, what does not, and how they have overcome obstacles. What system components facilitate effective HA use? How are data used to provide care and improve health? The review should explore the vendor/employer perspective and its effect on relationships with health care systems.
Studies also are needed on the economic and cost implications. One outcome should be cost savings, but this needs to include patient productivity and functional status.
Using a conceptual model for an ongoing demonstration rather than a time-defined project is desirable. Pairing a large HA vendor with an EHR vendor and a health system could measure many of these concepts. Collaborative research with other Federal partners such as CMS, Department of Defense (DoD), and Department of Veterans Affairs (VA) seems desirable.
Research is needed on "patient activation" and "clinician activation" addressing patient response to different types of feedback, and "readiness to change" effects. Basic research on risk and risk reduction in specific demographic populations is needed, as is research on physician support from the community, especially for the highest risk patients.
Prioritization of HA follow-up is key, and should be based on the likely health impact of possible interventions. Integration of clinical and community prevention services should be a priority. Guidance is lacking on ways to expand services from the clinical office to the community. For example, the doctor's office will not tackle the obesity epidemic. Within clinical preventive services, reasonable outcomes could be preventive services uptake and effects on disease management.
Participants echoed the interview discussion that cultural and work environments affect response rates of HAs. Studying this is critical to understanding how and how much these factors affect HA participation.
Participants discussed health information technology (IT) needs and HAs, including readability format and skip patterns, and interactive tools. The group agreed that technology should ease the way and improve communications, resulting in more patient-centered services. The HA tools must be user-friendly, and responsive to health and information technology literacy levels. Some attendees also mentioned the need for HA tailoring to specific populations (e.g., by ethnicity, gender, occupational groups).
Generally, the meeting seemed to focus more on broader systems-based research, while the scan/interviews brought up more detailed operational functions. The major research recommended included a best practices study; evaluation of linkages between clinicians and communities, including physician support from the community; HA effects on costs of care and patient functioning and productivity; HA design (e.g., length, reading level, population tailoring); and basic research on risk and risk reduction research targeted toward gender/ethnic specific populations and people with chronic conditions. IT around HAs also needs to advance and become more standardized.