As family physicians, our satisfaction at the end of a day comes from knowing that our patients received quality care. Often, a few encounters will remind us that we truly make a difference in their lives: the adult with severe dyspnea who is now stable after successful diuresis, the child with asthma who has remained symptom-free for a year, the smoker who finally succeeded in quitting after years of your urging. Our primary motivation for choosing this profession was our desire to be effective healers for all people who call us their personal physicians.

Yet how do we know we're extracting and applying the best medical knowledge and skills from the ever-expanding universe of medical literature? As important, how do we know we're providing quality care to every one of our patients when we examine or talk with only a small fraction of them each day? Family medicine educators might reframe these questions this way: How do we use evidence-based guidelines to practice population-based health care? Others might simply ask, “How do we provide the best care for all our patients?”

These questions may seem easy to answer. Some family physicians who have had stable patient panels for more than a decade tell me they “know” everyone with diabetes or asthma, everyone who smokes and others at risk and that they just “know” these patients are receiving the best care possible. Yet many family physicians in today's rapidly changing health care environment find they need to develop a more methodical and proactive approach to caring for their patient populations — a “systems approach” to primary care. A systems approach involves collecting data to help you identify and monitor those who need care; finding and implementing the latest care guidelines; and involving your office staff, nurses, other physicians and other caregivers in evaluating and improving your performance.

In helping us serve all our patients more effectively, a systems approach to primary care also helps us show the quality of our care to those who demand evidence. We family physicians increasingly are being asked by patients, fellow physicians, health plans and regulatory bodies to demonstrate objectively that our care really does make a difference and that our effectiveness extends to the entire population we care for, regardless of whether we've actually seen those patients recently.

This article describes a four-step systems approach that optimizes primary care for our patients with certain conditions:

1. Choose common conditions that lend themselves to a systems approach to care,

2. Identify the patients in your practice with those conditions,

3. Choose measurable outcomes that reflect the best evidence-based medical practice,

4. Regularly measure and try to improve these outcomes.

These four steps employ the principles of continuous quality improvement — that is, the notion that we can improve our care if we carefully choose and continuously monitor the outcomes we want to achieve and continuously explore ways to improve our system of care.

The extent to which a particular practice can implement this approach depends on many factors, such as your location (rural or urban); setting (solo, small group, multispecialty group or integrated health system); available resources (such as your computer system); and the availability of support services (such as home-health nurses, case managers and nurse telephone services) from your group, your health system or a particular health plan. The group practice with a large, relatively stable panel of capitated patients is ideal for the systems approach, although many other practices can approximate it to good effect.

How the work required to implement this model is paid for will also depend on your situation. If you're in a heavily capitated practice and you share risk for downstream utilization, proactive population-health programs are in your economic interest because they help keep patients from needing the services for which you're reimbursed only on a per-member-per-month (PMPM) basis. Even better, some capitation plans reimburse separately (beyond the PMPM payment) for services such as immunizations and visits by home-health nurses — services that can be important parts of the systems approach.

Even under discounted fee for service, you can recoup some of the costs of population health care by using codes that accurately reflect the services you and your staff provide. For example, you can bill for preventive medicine counseling using codes 99401-99404, immunizations using codes 90700-90749 and home visits using codes 99341-99350 (although not all insurers will cover these visits when they're provided by nurses rather than physicians). Also, be sure your coding reflects the true complexity of your visits with these patients when they do come to the office (of course, you'll need thorough documentation in this era of heightened scrutiny for health care fraud). Other components of population- health programs, such as enhanced computer systems or staff time involved in identifying eligible patients and reviewing their charts, may not be reimbursable.

In the end, your population-health program may depend largely on the resources you can devote to it, your own initiative and your commitment to doing what's best for your patients. Remember that you can implement this approach in stages. The keys are to be proactive, creative and realistic about what you can accomplish.

1. Choose the right condition

Choosing the right clinical condition is critical to the success of population-based care. Before applying this approach to any group of patients, ask yourself whether the condition meets the following criteria:

• The condition is commonly encountered in primary care;

• The cost and human burden of the condition is significant;

• Evidence exists of wide variations in care or outcomes (or such variations are likely);

• Evidence exists that best practices lead to predictable, improved outcomes;

• Appropriate evidence-based practice guidelines are available;

• Family physicians can provide most of the care the condition demands;

• Patient education and support can help improve outcomes;

• Best practices and outcomes are measurable, reliable and relevant.

Based on these selection criteria, you can identify a special subset of acute and chronic conditions, as well as important preventive services, that lend themselves to a population-based approach. Here are some examples:

• Acute conditions such as headache, low-back pain, otitis media, urinary tract infection and myocardial infarction;

• Chronic diseases such as congestive heart failure (CHF), asthma, diabetes and hypertension;

• Preventive services such as immunizations, Pap smears and mammograms;

• Preventive counseling in areas such as diet, exercise, smoking cessation and cessation of the use of alcohol and other drugs.

2. Identify patients with the condition

Once you choose a condition or a preventive service, you must develop a system to identify the patients in your practice who have the condition or need the preventive service.

First, clearly define a target population according to variables such as age, gender and specific medical data. For example, if your focus is immunizations, the target populations may be your two-year-old patients, your patients over age 65 or your pregnant patients. If your focus is a chronic disease, the target population might be all patients with the signs and symptoms of the disease or just those who have been hospitalized with that diagnosis. How you define your target population will clearly affect the size of the population and the scope of your project.

Then define the condition clinically in a way that will identify the patient population accurately and make the identification process feasible. The capabilities of your computer system may largely determine what you can accomplish. For example, if you use a computerized patient record (CPR) system, it may be as simple as pushing a few buttons to identify all patients with asthma, all patients with CHF who have a recent ejection fraction of less than 40 percent, or all adults over age 65 and all other patients with medical conditions that require annual influenza vaccinations.

Without a CPR system, you will be more limited in the patient populations you can identify accurately and feasibly. For example, if you use your computer system only for patient demographics and billing, you could identify patients over 65 but perhaps not those who should be vaccinated because of their medical conditions. This illustrates why many see CPRs as the phone and fax of the 21st century — tools family physicians won't be able to practice without.

Even without a CPR system, you can create a system to identify most of your patients with the condition you select. For example, you could ask your patients to fill out a simple survey added to your patient intake form or have your staff review your charts and apply colored dots to indicate certain diagnoses or conditions (e.g., red dots for all current smokers, green dots for all children who need to be immunized or heart stickers for all patients with CHF). To limit the number of chart reviews, you may be able to use your billing system to identify patients with certain conditions by searching for ICD-9 or CPT codes.

Ideally, you would review your entire patient population, but your computer system, staff time and expertise may limit what's feasible. If you can identify 50 or more patients who have a target condition and for whom your care may vary from the “best practice,” you probably have a large enough group to make population-based care worth your effort. The key point here, and the essence of continuous quality improvement, is that you need to start somewhere and can always improve.

3. Choose outcomes linked to guidelines

Once you identify a condition or preventive service and a target population, choose measurable outcomes that result from following high-quality, evidence-based care guidelines. The outcomes you choose will be the bases for evaluating the success of your program.

What outcomes are reasonable to measure depends on many variables, including the condition you target, the capabilities of your computer system and the amount of effort you can expend. For example, although you can measure your effectiveness in managing hypertension by tracking admission rates for hypertensive emergencies, stroke and myocardial infarction, this may require a complex methodology and a sophisticated information system. But other outcomes related to hypertension — such as the percentage of hypertensive patients who have normal blood pressure readings or the percentage of hypertensive patients who don't use tobacco — are easier to measure. Your nurse could maintain a simple computer spreadsheet or a handwritten log to track such variables as name, date of visit, blood pressure reading, whether the patient smokes and date of next visit.

Whatever outcomes you select must be linked to evidence-based care guidelines, since the guidelines justify the effort. For example, the percentage of patients who have controlled hypertension and who don't smoke is an appropriate outcome to measure because consensus guidelines exist for managing hypertension, and clinical studies demonstrate that controlling hypertension and eliminating smoking reduces morbidity and mortality from hypertensive emergencies, stroke and myocardial infarction.

Clinical guidelines for a variety of conditions have been published (see “Sources of clinical guidelines”). They can help you avoid reinventing the wheel for your own practice.

4. Measure and improve performance

The final step in implementing population- based care is to set up a system for regularly measuring and improving your outcomes. Again, being proactive is the key.

When caring for a population, you should measure outcomes for all your patients with the targeted condition, not just those who come to your office. This is largely what differentiates population-based care from traditional, individual-centered care. For example, if your target outcome is that all hypertensive patients will have blood pressure within normal limits, documented by twice-a-year readings, then you must identify those who haven't been in the office for a recent blood pressure check and let them know they need to come in. Computer programs are available (to augment a CPR system) that can identify patients who meet given criteria, create letters and even place telephone calls to them, reminding them to have their blood pressure checked.

Even without a CPR system, you can implement simple office procedures to accomplish the same task. For example, your staff could review the spreadsheet or handwritten log of hypertensive patients to identify those who failed to keep appointments, those who simply have not been in to see you for some time and those with poorly controlled blood pressure. These patients would be targeted for telephone and office follow-up.

In addition to monitoring all your patients with a given condition, you must ensure that those patients actually receive optimal care. Once again, this may require proactive efforts to reach patients, educate them and monitor their results. For example, suppose your chosen outcome is that no children with asthma will be seen in the emergency department or admitted. Evidence-based guidelines suggest you can accomplish this if all patients know and avoid their asthma triggers, use their inhaled medications properly, learn to measure their peak flow and have a patient-initiated treatment plan for acute exacerbations. Meeting these guidelines may require that you send a home-health nurse to do environmental assessments, teach patients to use inhalers, review patient-initiated plans for exacerbations and tell patients whom to call with after-hours questions. Finally, your staff (or volunteers from among your patients or the community, if you have access to them) may need to contact all asthma patients who miss their regular appointments or periodically call parents of children who have moderate or severe persistent asthma.

Finally, to improve your performance, it's important to evaluate your care outcomes regularly and consider improvements. For example, you might schedule quarterly review meetings with your office manager and (depending on how ambitious your effort is) physician colleagues, home-health agency staff, patient or community volunteers, and others involved in your population-health initiative. You might survey your patients with the targeted condition and your staff to identify barriers to meeting your goals and determine how to overcome them. You can't fix what you don't know is broken.

The systems approach in action

Does a primary-care systems approach to managing patient populations really work? Let's use the four steps described earlier to review how you might implement it. Then we'll examine some results of a CHF program instituted to help the primary care physicians and members of AvMed Health Plan, a Florida-based, nonprofit HMO. The program is directed by primary care physicians.

Step 1: Choosing the condition. Almost 5 million Americans have CHF, and family physicians commonly treat them. CHF is the leading cause of hospitalization for people older than 65; it contributes to 250,000 deaths annually and costs $10 billion each year. The five-year mortality rate approaches 50 percent in some studies. Research has demonstrated that adopting practice guidelines for CHF leads predictably to improved outcomes, and a set of evidence-based practice guidelines (“Heart Failure: Evaluation and Care of Patients With Left- Ventricular Systolic Dysfunction”) is available from the Agency for Health Care Policy and Research (AHCPR). Family physicians can provide most CHF care, and both physician and patient education can help improve outcomes. Measurable, reliable and relevant outcomes can be identified.

Step 2: Identifying patients. Most patients with signs and symptoms of CHF have left-ventricular systolic dysfunction revealed by an ejection fraction on echocardiography of less than 40 percent. For this program, the target population would be all patients who meet the clinical diagnosis of CHF (e.g., dyspnea, poor exercise tolerance, rales, etc.) and have an ejection fraction of less than 40 percent. If you use a CPR system, you could identify these patients through a series of database queries. Alternately, your nurse or office manager could maintain logs of all patients with this diagnosis by reviewing claims records, administering a survey to all patients or reviewing charts.

Step 3: Choosing outcomes linked to guidelines. From the AHCPR guidelines for CHF, you could identify and measure a number of key outcomes, such as these:

• Patients receive a prescription for an angiotensin-converting enzyme (ACE) inhibitor and adhere to their regimens,

• Patients record their weight daily,

• Patients contact the physician's office if they note weight gain of three to five pounds in a week or since the previous visit.

The patients' weight logs also would include their weights at the beginning of the program and when you prescribe the ACE inhibitor. You would ask patients to bring their weight diaries to office visits, and staff would record their weights at those times as well.

Because nonadherence is a major cause of morbidity and avoidable hospital admissions, your staff (or volunteers) might need to contact patients with more severe heart failure (e.g., those with a recent hospitalization or with dyspnea on minimal exertion) every couple of weeks to make sure they have scales in their homes, are recording their weights daily, are taking their medications, haven't gained weight and don't have worsening symptoms. For patients whose adherence is in doubt, you might need to have a home-health nurse conduct a home assessment and provide additional education and support.

Other important outcomes could include reducing hospital admission rates and maintaining or improving patients' perceptions of quality of life. But measuring these outcomes requires computers, access to survey tools (such as the SF-36 Health Survey, which measures perceived quality of life) and careful attention to methodology.

Step 4: Measuring and improving performance. Based on data from patient visits and proactive phone calls, your staff could track these variables related to your targeted outcomes:

• The number of your patients with CHF,

• The date of their last office visits,

• The number and percentage of those who keep a daily weight diary,

• The number and percentage of those who take their ACE inhibitors,

• The number and percentage of those who gain more than three to five pounds and those who call the office if they do,

• The number of hospitalizations for CHF among these patients.

For those who are readmitted, you would try to find out why. Were they taking their medications? Were they keeping daily weight diaries? Did they remember to call the office when they gained weight? Had they visited the office recently? Did your staff call them to check on their clinical status? Answers to these questions could suggest strategies for managing CHF more effectively.

Results of a systems approach

Our health plan implemented a program, called Healthy Hearts, following the model outlined above. After receiving approval from 225 CHF patients and their primary care physicians, AvMed's nurse case managers and home-health nurses conducted home assessments, patient education and telephone follow-up. We found that when the patients entered the program, more than half didn't even own scales, much less check their weight daily. So AvMed purchased scales for these patients. (Yes, this is clearly an advantage of working in a fully capitated environment; the scales were necessary for effective care, and we were fairly sure they would pay for themselves in reduced health care costs.)

Here are the clinical results of the program after only six months (comparisons are with data collected before the program began): The percentage of patients taking ACE inhibitors increased from 60 percent to more than 70 percent; readmissions for CHF decreased by 40 percent; and admissions for non-CHF causes (such as pneumonia and myocardial infarction) decreased by almost 50 percent.

Based on pre- and postenrollment surveys of quality of life, the CHF population's perceived health didn't decline during this time; in fact, a number of patients rated their own health as improved. This is an accomplishment given the worsening natural history of untreated CHF. Up to half of the patients in some studies died within five years of diagnosis.

Of course, you don't have to take my word for it. The literature offers several examples of the benefits of a systems approach to primary care. (See “Suggested reading.”)

Implementing a systems approach

How to implement a primary care systems approach to population-based medicine depends largely on your type of practice. Some health plans, integrated delivery systems and large group practices already are providing population-based care by forming quality improvement committees to identify conditions, establish outcomes, review and approve care guidelines, and involve their nurses and other health professionals to assist physicians in outreach and education.

With determination and creativity, a solo family physician or small group can make this approach work, too. The major differences may be the number of conditions you can afford to target and the scope of the interventions you can afford to implement. Here are some concrete suggestions:

• Use the clinical guidelines and other materials available from national organizations such as AHCPR to help you develop population-health programs.

• Seek out local resources to be part of your practice's primary care system. Contact the local hospital, home-health agencies, pharmacies and pharmaceutical companies, the public health department and health-related associations such as your local American Heart Association chapter. You might ask them for help in setting up computerized tracking systems, developing surveys, conducting home visits or providing patient education materials and support.

• Consider soliciting volunteers from local organizations to make follow-up calls and the like.

• Contact your health plans for assistance. All health plans accredited by the National Committee for Quality Assurance provide ongoing quality improvement programs targeted at common conditions and preventive services. So your plans may already have resources to help you reduce hospital admissions for asthma patients or improve immunization rates. Many plans will help you care for their members by arranging or performing home assessments and education, providing after-hours telephone advice and triage services staffed by nurses, providing clinical pharmaceutical and specialty consultation services, sending patients educational materials and helping you monitor the progress of your population-based care.

Ideally, health plans should work together on these initiatives, but in practice they target only their own members. If you have contracts with multiple health plans, you'll get the biggest bang for your buck by working with those that cover the largest percentage of your patients. The plans' medical directors usually are your best points of contact. And if you'd like to gain skills and experience in quality improvement and population-based health care, you can ask to serve on these plans' local quality improvement committees.

You can do it

The most important part of implementing population-based care is your own desire to do it. Evidence-based care guidelines are readily available; and with a bit of effort and creativity, you can find additional resources to help you put a program into action. So pick a condition of interest to you and of significance to your patients, and implement your own primary care systems approach to population- based medicine. With regular monitoring and evaluation, you can continually find new ways to improve the quality of care for your various patient populations.