Many family physicians may see disease management (DM) as one of those trends born of managed care that haven't yet matured enough to be dangerous. But as we discussed in the first part of this series (“Disease Management: Who's Caring for Your Patients?” October 1998), DM as an industry is flourishing, and its success could present a real challenge to family medicine. Family physicians not only need to become more involved in DM programs developed by health plans and managed care organizations (MCOs), they also need to begin developing DM programs for their own patients. By taking a page from the playbook of organizations already involved in DM, a practice can position itself to show MCOs that it's improving patient outcomes while limiting unnecessary episodes of care. Because it helps physicians maximize quality and minimize cost, DM also sets the stage for a practice to share risk with insurers for the cost of patient care. But essentially, DM is simply a cost-effective way of doing the right thing: improving the quality of the care you deliver. Now we'll show you how.
Yes, you can
The heart of DM is preventing disease and intervening early with patients who have chronic conditions such as asthma, diabetes or congestive heart failure — an approach that family physicians have long espoused and practiced with individuals. To an extent, DM is little more than a way of systematizing and organizing familiar concepts and focusing them on a given condition. Implementing the DM approach successfully requires committed physicians, a willing and well-prepared staff, a structured process of change management, an organized approach to analysis and a well-developed quality improvement system with the patient at the center of the process. Because of their training and practice philosophy, family physicians should be able to implement DM as well as, or better than, anyone else.
However, as it's practiced by the large MCOs and drug company spin-offs, DM has become dependent on information technologies and data-management tools that the average family practice doesn't possess. For example, large DM programs use databases of enrollee and claim information to track patient populations; electronic links with pharmacy and laboratory databases; intelligent software that assesses patients' risk and stratifies patients according to risk profiles; and communication networks that link nurse triage staff and case managers with patients, providers and payers to monitor compliance and sometimes even collect physiological data directly from patients at home.
So how can family physicians in small practices develop and use DM programs for their patients? We're not going to suggest that the average family practice can put the same resources into DM that drug companies, investor-owned firms and MCOs can afford. But information technology today offers family physicians a whole new set of desktop tools for managing information in ways that can promote disease prevention, early intervention and better patient education — the core of any DM program. Most family practices can begin to use off-the-shelf software to design and implement practice-based DM programs that will improve patient care and quite possibly improve the bottom line.
The informatics of DM
Our framework for developing and implementing practice-based DM programs is a set of information-management tasks:
Evaluate your patient population for high-volume, high-risk conditions that lend themselves to DM.
Identify practice guidelines or treatment protocols for patients who have those conditions.
Identify your patients in the atrisk population, and use risk-assessment tools to stratify them according to their need for DM services.
Collect and store information about the patients, their conditions and your interventions.
Communicate with patients as part of the DM program, and educate them about their conditions.
Analyze the outcomes of your interventions.
Notice that these steps are not a comprehensive recipe for doing DM or for setting up a DM program. Rather, this is the information-management road map of any DM effort, the common denominators for processing the data and information necessary for DM. You have many choices about the information technology — computer systems, databases and analytical software programs — that you will use to perform and automate these steps. For example, if your practice has a computerized patient record (CPR) system, some of the information- management tasks in DM may be easier. But in this article, we are assuming only that you have a practice management information system capable of doing billing and claim processing and that you have one or more PCs running Windows 95 or Windows 98.
Evaluate high-volume, high-risk conditions
A key to success in practice-based DM is keeping things simple. This means focusing on one or two chronic diseases that are highly represented in your practice and, furthermore, that interest you. Nothing will kill a practice-based DM program faster than trying to tackle too many diseases. If you try to do too much at once, you and your staff will quickly tire of the extra work required to capture and manage patient data for DM.
A prerequisite for analyzing high-volume, high-risk conditions is a claims data set from your practice showing the volume of visits and patients by ICD-9 diagnosis codes. Your practice management system should be able to produce a file like this by capturing data from the HCFA-1500 universal claim form that each visit generates. If it can't, work with your vendor until you can get this critical data set, by month and year.
You can pay software vendors to produce analyses for you, but it's surprisingly easy (and less expensive!) to work with the data yourself. Doing so also gives you much greater flexibility in how you use the data. If your practice management system can produce a paper report, it can also export the same information as a data file. Once you put this file on a floppy disk, you can perform the next step: importing it into an analytical software application so that you can manipulate the raw data and turn them into information.
We highly recommend using Microsoft Excel as an analytical and statistical tool for working with claims data sets because it's almost universally available (it may already be on your PC), relatively inexpensive, user-friendly and year-2000 compliant. Also, numerous add-ins are available to automate common data-gathering and analytical processes using Excel. One of these is QI-Tools, a free Excel add-in from Future HealthCare Inc., which you can download from our web site (http://www.futurehealthcare.com) and which will run on Windows 95 or Windows 98 systems. It automates many of the most common descriptive statistical chores used by health care data analysts, such as those described below. QI-Tools can take some of the labor out of producing QI and financial charts and graphs.
In the figure below, we have used QI-Tools to perform a Pareto analysis and create a Pareto chart that sorts diagnostic codes by frequency of visit. This allows you to tell at a glance which diagnoses are the “critical few” in this hypothetical practice. The figure shows the most common diagnoses for 429 patient visits during one month, arranged in decreasing order of frequency. Each bar displays the frequency of the diagnosis (seen on the primary Y axis on the left) as well as the percentage of the total visits that diagnosis represents (seen on the secondary Y axis on the right). The dotted line over the bars represents the cumulative percentage. Because the “sliding bins” feature of QI-Tools has been used to limit the number of displayed diagnoses, the primary Y axis accounts for slightly more than 200 of the 429 visits and the secondary Y axis extends only to slightly more than 50 percent.
The conclusion: For the period under study in this practice, asthma and diabetes are the most common diagnoses likely to lend themselves well to DM; together, they account for more than 20 percent of all visits related to the practice's top 15 diagnoses. In contrast, congestive heart failure (another frequent target of DM) doesn't appear among the top 15 diagnoses in this practice. If this pattern of visits were to be seen over a longer time period, these family physicians would want to focus their DM efforts on asthma or diabetes; congestive heart failure would be a less-productive choice.
A Pareto analysis of diagnostic codes
This Pareto analysis shows the most common diagnoses in one hypothetical family practice. Of these conditions, diabetes and asthma are the best candidates for a disease management program.
Identify guidelines and DM protocols
Once you've spotted your best opportunity for DM, you need to adopt a specific protocol for how your practice will manage the care process for patients in your DM program — when and by whom various interventions will be made. Without a doubt, the Internet and World Wide Web are your best desktop technologies for locating practice guidelines and DM protocols. You may not think of it this way, but the Web can be your expert-knowledge database for DM, including reviews of medical literature, published DM protocols and networks of DM professionals willing to share their experiences about what works well and what doesn't. And, of course, you can use the Web to stay abreast of the latest scientific advances.
All you need to access this expert knowledge is a PC running web browser software such as Netscape Navigator or Microsoft's Internet Explorer, a telephone connection and an account with an Internet service provider.
Although it does take some time to learn where to locate information on the Web, here are some examples of resources that could help you design a practice-based DM protocol for patients with asthma. Similar resources are available for other conditions.
Internet Grateful Med is the National Library of Medicine's database of more than 9 million abstracts from the medical literature, available at http://igm.nlm.nih.gov. Internet Grateful Med offers free access to a number of databases, including MEDLINE, HealthSTAR and several others.
The “Expert Panel Report 2: Guidelines for the Diagnosis and Management of Asthma,” published by the National Heart, Lung and Blood Institute, is available free (in PDF format) at http://www.nhlbi.nih.gov/nhlbi/lung/asthma/prof/asthgdln.htm.
The Cochrane Database of Systematic Reviews has several reviews on asthma treatment and patient education. The Cochrane database, which focuses on systematic reviews of randomized controlled trials, is available online for a modest subscription through Synapse Publishing Inc. at http://www.medlib.com.
JAMA's Asthma Information Center, at http://www.ama-assn.org/special/asthma/asthma.htm, includes a wealth of information on asthma prevention, treatment and education, including links to evidence-based guidelines, updated full-text articles from the medical literature and resources for patients and family members.
“Asthma Management: Guidelines for the Primary Care Physician” by Miles Weinberger, MD, of the Department of Pediatrics, University of Iowa College of Medicine, is available on the Virtual Hospital web site, http://www.vh.org/Providers/ClinGuide/Asthma/Asthma.html.
Following a guideline isn't always as easy as it would seem. You can't assume your staff will know what to do and when to do it unless you carefully spell out a program for managing your target disease. We strongly suggest that you convert your DM guideline or protocol into one or more flow diagrams that clearly indicate each step in the process and can be understood by everyone in your practice. Diagrams that illustrate steps and decision points make guidelines easier to follow and more likely to succeed. (For an example, see “A protocol for vascular disease check-up.”)
You can create flow diagrams with any of a number of user-friendly software programs. We made ours with Visio Standard, but we could just as easily have used Microsoft Powerpoint, a standard component of the Microsoft Office 97 suite.
A protocol for vascular disease check-up
With widely available software, you can create flow charts like this one to communicate your treatment protocols clearly to others on your DM team.
Identifying and stratifying those at risk
Now comes one of the most challenging information-management tasks in practice- based DM: identifying patients with the target condition who are at risk for overutilization of services or exacerbations of their illness and then stratifying them based on their level of risk. This step is essential because it lets you target interventions to those patients most likely to benefit from them. It's not cost-effective, for example, to have every patient with asthma receive case management or home monitoring of pulmonary functions, especially if your practice would bear the cost! You want to target your resources first to the patients whose conditions are most difficult to manage.
DM companies and health plans are using software that extracts claim, pharmacy and lab data on tens of thousands of patients and then applies proprietary clinical rules that stratify patients with a given disease and flag those who are appropriate for different levels of DM — everything from patient education to intensive case management. One advantage you have over large health plans in assessing patient risk is that you know your patients better and, in theory at least, should be able to assess their risk more accurately than anyone else, even without the DM companies' sophisticated software. But remember that DM relies on objective data. There's no getting around the need to identify patients using risk-assessment tools and to stratify them based on standard classifications of risk so that you can document why you used particular interventions.
Health risk assessments (HRAs) are available for a number of the most common DM conditions. So once you've chosen a condition to target, turn to the Web to find a tool for assessing risk related to it. Here are some examples:
A diabetes risk assessment from the American Diabetes Association at http://www.diabetes.org/ada/risktest.html;
A cardiac disease risk assessment from the American Heart Association at http://www.amhrt.org/risk/quiz.html;
A breast cancer risk assessment from Wayne State University at http://www.cs.wayne.edu/~faf/care/w_risk.html.
A logical next step is to administer the HRA to all your patients who might be at risk for the condition (for example, if your focus were cardiac disease, you might have your patients older than 35 complete the assessment).
Once you've identified your patients who have the condition you want to target, we suggest you stratify them using a risk classification methodology that identifies certain clinical criteria that define progressive levels of illness. It's then up to you to determine which classification of illness patients must fall into (i.e., how sick they are) before they qualify for your DM program. Here are two examples of these classifications:
The New York Heart Association classification of patients with heart failure, available from the American Heart Association at http://188.8.131.52/Heart_and_Stroke_A_Z_Guide/classf.html;
The National Heart, Lung and Blood Institute's classification of asthma severity, given on page 20 of “Expert Panel Report 2: Guidelines for the Diagnosis and Management of Asthma” (http://www.nhlbi.nih.gov/nhlbi/lung/asthma/prof/asthgdln.htm).
You can also use a measurement of patients' perceived health and functional status to bring additional patient data into your efforts to evaluate risk. The SF- 12, a 12-item survey of general physical and mental health status, is an easy-to-administer survey that's gaining widespread use for this purpose. A somewhat more involved version, the SF-36, also is available. The software for collecting and assessing survey results runs on a PC and is available from several vendors, among them Velocity Healthcare Informatics Inc. in Minneapolis (800-844-5648; http://www.velocity.com).
However you choose to enroll patients into your DM program, keep the numbers small initially and let everyone on your staff know the criteria for enrollment.
Collecting and storing the information
Among the many information-technology breakthroughs in the past few years, none is more important than the availability of user-friendly relational databases that can collect, store and manage large amounts of information on a desk-top PC. Relational database technology is the heart of all current CPR systems, such as those from MedicaLogic and Epic, but you don't need a full-blown CPR system to create a registry of patients in your practice-based DM program. You do need an electronic database that will function as a miniature medical record for the patients in your program so you can track their visits, diagnoses and secondary conditions, and key test results, and so you can document DM interventions. And you will need to invest the staff time to enter the relevant data consistently and in a timely manner.
Microsoft Access has become the PC database system of choice for customized outcomes, case-management and DM efforts. It is part of the Microsoft Office 97 suite, is easy to use and maintain, comes with a variety of database templates and can handle a large number of records. But Paradox (originally from Borland but now available from Corel), Lotus' Approach, and FileMaker's File-Maker Pro also run on desktop PCs and are used by physicians for keeping patient records.
Most family physicians won't want to spend the time to design their own DM database or modify one of the templates that come with these programs. A reasonable alternative is to hire a database consultant from a local community college, university or private firm to build your application, which will consist of data-entry forms, tables in which the data are actually stored and reports that you can configure and view as a summary for each patient.
All this isn't just hypothetical. For the past four years, Paul Dunn, PA, in the Department of Family Medicine at the University of North Carolina School of Medicine in Chapel Hill has kept a Microsoft Access database of his practice's patients with diabetes. Similarly, Douglas Kelling, MD, an internist practicing in Concord, N.C., manages the care of almost 600 of his patients with diabetes using an Access database that he designed with a consultant's help. That database keeps track of patients' weights, test results and visits to referral specialists, as well as issuing reminders for eighteen tests or exams (such as HbA1c tests every 90 days, LDL cholesterol tests yearly, etc.).
The cost of Kelling's database was about $16,000 over three years for software, database design and development, and maintenance. But you may be able to create a basic registry of patients that would let you track their visits, conditions, key test results and DM interventions for less than this. Database consultants' charges do vary, and you may not need many bells and whistles. In any case, the cost of a database is less daunting when you think of it as being spread over three years or so and shared by the physicians in a small group. Consider also that the payoffs include a stronger bargaining position with your MCOs, the likelihood of obtaining additional managed care contracts and the ability to provide better patient care that you can document. It's certainly cheaper in the short term not to start up a DM program. But the long-term cost could well include losing patients to plans and providers that do.
Patient communication and education
It's well-known that patients with chronic diseases frequently don't adhere to treatment regimens. Explicit and thorough communication with your patients about DM does make a difference and can help produce positive clinical outcomes. Researchers at the Johns Hopkins University School of Public Health noted that patients with chronic diseases benefited when their clinicians encouraged them to adhere to treatment regimens.1 The multiple interventions commonly included in DM — checking on whether patients are taking medications, refilling prescriptions, keeping appointments, etc. — all translate into improved outcomes.
You, the physician, are the most important educational resource in your practice-based DM program. Getting your advice and direction is why patients come to you in the first place. But physicians who have built successful practice-based DM systems have told us repeatedly that doctors can't do it alone. You need a system that gives patients the right information at the right time and from the right person — perhaps a nurse, a nutritionist or a public-health educator.
Here again, the Web can help by making patient education materials easily and inexpensively available to anyone involved in your DM program. You can find patient education materials at a number of web sites, but we especially recommend two sources:
“Information From Your Family Doctor,” patient education materials from the AAFP at https://familydoctor.org;
Patient education materials from the University of Iowa's Virtual Hospital web site, http://www.vh.org/Patients/PatientsAnnotatedList.html.
Other desktop options for producing educational materials include commercial products that offer personalized patient instruction sheets, forms for tracking medication lists and charting tools. Here are three of the leading resources:
Medifor Inc. of Port Townsend, Wash. (http://www.medifor.com), which offers Patient Ed, a program developed by family physician Peter Geerlofs, MD;
Clinical Reference Systems Inc. of Broomfield, Colo., which offers a variety of patient education programs (http://www.patienteducation.com);
The Patient Education Institute at the University of Iowa's Technology Innovation Center, which offers X-Plain (http://www.patient-education.com/about/X-Plain/index.html).
Analyzing and studying outcomes
Of course, the main point of going to the trouble and expense of building a practice-based DM program is to improve patient care. If you have collected the patient data and stored it in a database, then proving to yourself and your payers that you've improved patient outcomes should be relatively easy. Kelling has no doubt about the value of the analytical phase. “We can prove conclusively that our diabetes patients' average hemoglobin A1c has gone from 8.4 to 7.1 over the last two years,” he says. “That information not only assures me that I've done my job well as a physician, but it gives me ammunition with the HMOs when it comes to contracting. For the first time, I have more information on the quality of the care I deliver than they do.”
What we were trained for
As health care shifts more and more toward managed care, family physicians need to become more proactive about marketing their DM skills. But at the same time, they need to learn how to practice DM on a larger scale, similar to the risk management of defined patient populations that MCOs now carry out. Family physicians should be asking themselves, “Where do I fit into the disease management picture?” After all, this work approximates what family physicians were trained to do: identify patients in need of interventions by assessing risk (the patient history is a risk-assessment tool), use evidence-based guidelines to prevent complications and unnecessary episodes of illness or care, treat patients using a biopsychosocial model that accentuates patient education and involvement, use a team approach to after-care, and keep good records in the interest of tracking trends and improving outcomes. If this is the picture of disease management, why don't we see the family physicians of America standing in the middle of it?