Larry is the sort of person who comes 30 minutes early to his appointment and brings a novel to read in the reception room. We talk blood pressure, the publishing industry, his prostate cancer history, his grandchildren. At the end of our visit, I recap our plan. We’ll try the higher dose of his beta blocker, keep a sleep diary to address his insomnia and start physical therapy for his arthritic knee. Then the clincher: Let’s see each other again in three months.

We rarely think about all the calculations that go into those final words. There’s the clinical question of when a diagnostic test might yield results or an intervention might take effect. Practically speaking, putting an appointment on the calendar means the patient and his problems are not (as we say in medicine) lost to follow-up. And it constitutes a relationship-building expression of concern: I care about your health. We are in this together.

As doctors, we usually base the timing of follow-up visits on some mix of habit and a gestalt of patient need, all within the arbitrary structure of the lunar calendar. Not surprisingly, then, Dartmouth researchers have shown that visit rates vary tremendously. In a study of Medicare data, they found that seniors in Grand Forks, N.D., average less than three visits each year, while those in East Long Island, N.Y., go to the doctor as many as 12 times a year on average. Patients tend to have more visits per year if they are sicker, the study found, but also if they live in an area with more doctors or with doctors who tend to ask patients to come in more often, even when adjusting for factors such as health status. What the patient prefers seems to have no significant association with visit rates.

The timing of follow-up visits, in other words, has tended to fall under the art, rather than the science, of medicine. While studies suggest that connecting with a doctor is generally a good way to build a trusting relationship and to promote health, we don’t really know the right frequency of visits. The few studies that exist on the subject pertain to specific conditions, such as kidney failure: In one observational studyof patients on hemodialysis, patients who had four visits per month had the same risk of death and only a slightly lower risk of hospitalization in the following year compared with those who had less-frequent visits. Several other studies, of prenatal care, for instance, have found that fewer visits may be just as good.

So the broader trend toward evidence-based practice in medicine hasn’t quite caught up with the basic question of how often to see your doctor. With appointments harder and harder to come by, medical costs still rising and a long-standing problem of both overuse and underuse of health care in the United States, doctors ought to pay closer attention to how and how often we ask patients to see us. Cutting out unnecessary visits would free up doctors to see patients who truly need care, shrinking lengthy wait times (like the 66 days it takes to book a physical in Boston, where I practice primary care) and potentially reducing visits to specialists or emergency departments for problems that primary-care doctors could treat better. This is particularly important in light of the projected increase in visits by a growing insured population under the Affordable Care Act.

Then there is cost. Outpatient care is the largest and fastest-growing sector of the U.S. health-care system, accounting for $436 billion of “above expected” spending, by one estimate. The price tag for doctors’ visits also adds up for patients who must take time off from work and shell out cash for transportation and co-payments. As my colleagues and I recently argued in the Journal of the American Medical Association, health system administrators will need to think more about decreasing the number of unnecessary visits (and the associated risk of unneeded, possibly harmful additional tests and treatments) as they adopt new care models in which insurers set targets for how much healthcare for a given population of patients will cost.

To figure out the “right” number of visits, we can start with better information about what we already do. Clinicians need to ask one another how often they request that patients come back, and why. We should examine the topic more closely and intentionally, both through well-designed studies and by scrutinizing the messy, real-time data we can get from electronic health records. Another idea that could immediately cut out some unnecessary appointments: Coordinating visits between primary-care doctors and specialists, and getting each to agree on who does what for shared patients.

Most important, we must be more nimble in how we connect. Many goals of Larry’s three-month follow-up visit may be accomplished just as well, or even better, without getting him into my office again so soon. If Larry is game, I can sign him up for a telemonitoring application that allows me to see his blood pressure readings remotely. I can schedule a virtual visit with him by telephone or video to check on his insomnia. I can order an automated e-mail survey that will arrive in his inbox in four weeks to assess his knee pain. I can use an educational video or handout to explain when to call me about a suspicious mole on his arm, and I can virtually consult a dermatologist to review a photo of it if necessary. I might ask him to schedule a follow-up in a year, to examine his knee and maybe check some bloodwork, knowing that if I don’t hear from him, he can still stay on my radar with the help of a master list of patients and their needs that I co-manage with a population health coordinator on my team.

These approaches complement and enrich the traditional face-to-face visits that remain the anchors of the doctor-patient relationship. The barrier to more widespread use is not technology (which already exists and keeps getting better), but the mundane forces of inertia and antiquated fee-for-service payment models. Many insurers are still reluctant to reimburse doctors for nontraditional visits and care coordination, for instance, which have made them challenging to adopt.

As we learn more about the impact of visit frequencies and get better at nontraditional visits, we can start to tailor care for individual patients with their unique medical problems, personalities, goals, and attitudes about health and health care. Rather than lamenting the uselessness of the annual physical (a popular target these days), we can examine the traditional visit and strip it for parts: What aspects are useful for a patient like Larry? Today, it might be managing his knee pain from afar. Two years from now, if his prostate cancer recurs and metastasizes, it might be a frank discussion — in person — of how he wants to spend his final months.

Traditionally, a doctor’s work has centered on caring for the patient sitting in the exam room. It follows, then, that the best way to make sure I’m taking care of the patient in front of me is to get him back into my office. But this sensibility is shifting as we realize that we must expand our reach beyond clinic walls and recalibrate how we connect with patients. Let Larry read his novel in the comfort of his living room. He and I will be in touch when, and how, it makes sense for his health.