How Much Time Do Physicians Spend Providing Care Outside of Office Visits?

Context Much of primary care takes place outside of office visits and is not reimbursed. Contribution By using the information recorded by physicians in an academic geriatrics practice, the authors estimate that providers spent an additional 6.7 minutes providing care outside of every 30-minute office visit, which amounted to an estimated extra 7.8 hours of clinical work per week for a full-time provider. Caution The findings may not be generalizable to full-time clinicians in community practices. Implication Providers spend a lot of unreimbursed time outside of office visits providing care. The time can be quantified, and similar data should be collected in broader settings to inform debates over reimbursement reform. The Editors Primary care is becoming increasingly focused on the management of chronic disease in a health care system that is more suited for episodic care of an acute illness. Seventy-eight percent of Medicare beneficiaries have 1 or more chronic health conditions and 46% have 3 or more chronic conditions (1). The 23% of Medicare beneficiaries with 5 or more chronic conditions account for 68% of all Medicare spending (2). Patients with chronic disease take more medications, see more specialists, and receive more formal and informal care. It is widely maintained that the care of patients with chronic illness involves nonreimbursed interactions with patients, caregivers, and others outside of the standard office visit (1). However, no data are available to support this claim. Because chronic disease management and coordinated care are particularly necessary in geriatric medicine, we evaluated the amount of time geriatricians spend and the nature of care they provide outside of office visits (35). Methods Two of the authors documented all clinical interactions that occurred outside of office visits during a 5-day workweek and used the information to develop a structured interaction form that collected the time spent (in minutes) in the interaction, type of participant (patient or family), method of interaction (telephone or electronic), content of the interaction (new symptom or chronic disease management), and outcome of the interaction (office visit or medication use). The form was then used by the authors in a 5-day workweek and was further modified. The final version of the interaction form was used to document all clinical interactions outside of office visits during 3 randomly selected, 1-week periods (excluding weeks with legal holidays) in late 2005 and early 2006. Forms were completed by physicians in a large urban, academic geriatric medicine ambulatory practice. Eighteen attending physicians and 13 geriatric medicine fellows, who spent varying amounts of time in the practice, formed the staff. Four office assistants, 3 social workers, 2 nurse practitioners, and 2 registered nurses also worked at the practice when the study was conducted. The practice had no electronic health record (staff used paper charts), and it provided care for about 2300 community-dwelling older adults. Because geriatricians typically work in many settings outside of the office (for example, hospitals, nursing homes, or home health care), we asked participants to record only interactions related to their ambulatory care practice. We included only physicians with at least 2 half-days of office visits scheduled in the week as eligible to participate, because we did not want to include physicians who had only minimal involvement in the practice. Physician eligibility was measured separately for each study week. A given physician, therefore, might be eligible to participate in 1, 2, or all 3 weeks of data collection, depending on the number of patient sessions scheduled that week. Participating physicians were briefed and were provided a written protocol with the structured interaction forms. Physicians were encouraged to complete the interaction forms at the time the work was done to minimize uncaptured events and document the time spent to the nearest minute. Time measurement reflects the time involved in performing the interaction itself and not in documenting the interaction on either the chart or the interaction form. Physician participants typically did not have the chart available to document these interactions at the time they occurred. We used the office's electronic scheduling system, which captures all visits to the practice with any provider, to determine whether interactions were temporally related to an office visit. We defined an interaction as related to an office visit if it occurred within 1 week of an office visit (that is, up to and including 7 days before and after the interaction). We used the practice's electronic database of clinical information to determine whether a patient had a diagnosis of dementia for those interactions that lacked such documentation. A diagnosis of dementia was assigned if the patient's record documented dementia, use of a cholinesterase inhibitor or memantine (a typical medication used to treat dementia), or a Mini-Mental State Examination score less than 23. Information on the content of the interaction was classified into 10 categories: physician ordering, follow-up laboratory tests, scheduling, referrals, correspondence, new (acute) medical symptoms, medication management, chronic disease management, coordinating care with family members (family counseling), and coordinating care with other health care professionals (professional collaboration). When an interaction had more than 1 component, it was assigned to a category on the basis of the most time-consuming component. Interactions that involved changes in any single medication in the absence of other care-plan adjustments were categorized as medication management, and interactions that involved anything more than titrating a single medication (for example, instructions about monitoring glycemic control) were categorized as chronic disease management. Multistep or linked interactions involving the same patient were grouped into episodes of care to more accurately represent the time spent on and type of work done in an episode of care involving a specific patient. By consensus, we developed 4 categories for these episodes. We then prioritized the categories into new medical symptoms, management of chronic disease and medications, family counseling, and miscellaneous. For example, speaking on the telephone with a patient's home health aide to discuss a new symptom, scheduling an urgent visit with the practice's front desk (scheduling), then speaking with the physician on duty who will see the patient later that day (professional collaboration) are 3 interactions grouped into 1 episode of care involving 1 patient, categorized as involving a new medical symptom. This project was exempted by the institutional review board at Mount Sinai School of Medicine, New York, New York. Statistical Analysis We present data on time spent in interactions and episodes of care involving physicians and patients. We created sample weights for each week, taking into account the number of eligible and participating physicians for each week of data collection. This was done to account for variability in the ambulatory clinical effort of individual physicians and physician nonparticipation, and because some physicians participated in more than 1 week of data collection. Our estimates accounted for clustering of interactions and episodes by patients and physicians. Some interaction forms had incomplete data. One author directly approached physicians to obtain the missing data, after which only the diagnosis of dementia sometimes remained uncertain. For these case-patients, the lead author used the patient's medical record to determine whether the patient had dementia, as previously described. Complete information on patient characteristics, obtained from medical record review, was not available for 12 patients (5%). Observations were reduced if missing data were involved in that specific analysis. All analyses were done by using SAS software, version 9.1 (SAS Institute, Cary, North Carolina). Role of the Funding Source The National Institute on Aging played no role in the design, conduct, and analysis of the study or in the decision to submit the manuscript for publication. Results Sixteen of the 22 physicians who were invited to participate in the study provided data. Nine were geriatric medicine fellows, and 7 were attending geriatricians. Physician participants were mostly women (88%), with a mean age of 37 years (range, 30 to 58 years). Mean attending physician experience was 10 years out of training (range, 2 to 27 years), whereas mean fellowship experience was 13.2 months (range, 6 to 18.5 months). The participating physicians accounted for 68% of the ambulatory office sessions scheduled for physicians who were eligible to participate in the study. Physician participants and nonparticipants did not differ in attending status (44% vs. 47%) or mean number of 3.5-hour half-day clinical sessions in the week (2.4 vs. 2.3 sessions). Nine physicians (56%) participated in 2 weeks of data collection, 6 (38%) contributed to 1 week, and 1 (6%) participated in all 3 weeks. This resulted in information on a total of 27 physician-weeks. Patients involved in the study were mostly women (83%) who were elderly (mean age, 83 years [range, 67 to 101 years]), had several medical problems, took many medications (mean, 11 medications [range, 1 to 26 medications]), and were community-dwelling (that is, not institutionalized or living in a nursing home) (Table 1). Physicians reported 472 discrete interactions within 296 episodes of care involving 226 patients. There were a mean of 2.0 (range, 1 to 10) interactions per patient, 1.6 (range, 1 to 9) interactions per episode, 1.3 (range, 1 to 10) episodes per patient, 18.0 (range, 1 to 53) interactions per physician-week, and 11.4 (range, 1 to 23) episodes per physician-week. The mean total time spent per physi