Patient-focused care over time: Issues related to measurement, prevalence, and strategies for improvement among patient populations

by Robert Reid, Morris Barer | Jul 01, 2003

Key Implications for Decision Makers

Measuring continuity of primary care with administrative data

  • Administrative data can be used to measure continuity in primary care by examining the degree to which a patient's care is concentrated with certain providers.
  • We found that easily constructed and understood measures of continuity performed as well as or better than more complex measures.
  • Selection of measures should be based on the ease of calculation and interpretability. Our work here suggests that some indices are best suited for routine reporting while others are most appropriate for research uses.

Continuity provided by primary care physicians and large practice groups in B.C.

  • Existing administrative databases may not routinely provide accurate information on which physicians are involved in primary care group practices.
  • In 1996/97, fewer than one-quarter of B.C.'s primary care physicians practiced in large groups (four or more full time physicians). Large-group practices are more common in rural and suburban areas than in the large metropolitan areas.
  • Adult patients had relatively high continuity of primary care in B.C. over the period 1994-96. However, children had much lower continuity, compared with adults or seniors.
  • Most large clinics did not appear to "share" a practice list; instead, patients generally saw the same physician. Even patients of large group practices saw a single physician for an average 70 percent of their primary care.
  • In the average large clinic, about 50 percent of all patients receive most of their care from that clinic.
  • Continuity of a personal physician had a significant but small effect on the likelihood of future hospitalization (higher continuity was associated with a lower probability of hospitalization in the subsequent year). Using site of care rather than an individual primary care provider did not increase the ability to predict hospitalization in the subsequent period.

Continuity of primary care in patients with severe and persistent mental illness

  • Administrative databases (at least in B.C.) are not well-suited to investigations of the extent and effects of continuity of primary care in these populations. Unless those gaps can be addressed, or better ways found to identify the relevant patient population and construct continuity measures for them using available data, attempts to study continuity of care in this population using such databases is not recommended.

Continuity of primary care in patients with workplace injuries

  • Individuals who report workplace injuries experience lower levels of continuity both before and after their injury, suggesting there may be room for changing patterns of health service use, which could, potentially, affect both propensity for and recovery from such injuries.

Observations on working with decision making partners

  • Maintenance of ongoing interest by decisions makers in a longer term program of research can be very challenging, especially when there is a change in staffing and/or organization within the decision maker partner (which is common).
  • As a general rule, decision makers are more interested in applicable results than they are in the research process, particularly when the research is concentrated on the development of methods, as much of this program of research was.

Executive Summary

Continuity of care has been identified as critical to effective primary healthcare. Sustained, continuous relationships between patients and their providers are arguably key to improving the outcomes of primary healthcare. However, there has been little systematic quantitative research focused on the impact of different levels of continuity of care, in a Canadian context, on important process and outcome variables. Key to such work is robust, valid measures of continuity of care, which are both easily computed and easily applied and interpreted.

This research program investigated ways of measuring continuity in the context of primary care. The program was divided into four separate research projects that were designed to address the following: identify the best measures of relational continuity in primary care; measure continuity for the population of British Columbia; and measure the impact of continuity of care on health outcomes. Once the best measures for continuity of primary care were identified, they were applied to three patient populations: (1) general primary care practice populations; (2) persons with severe and persistent mental health disorders; and (3) persons with workplace related injuries.

Data for this program came from the B.C. Linked Health Database, a population-based research database that contains a range of healthcare data for all people registered in the provincial healthcare system. The key advantage to using these data for this program is that they provide a population based perspective.

Project A - Measuring continuity of primary care with administrative data

In this project, we compared self report of having a regular doctor (from the National Population Health Survey) with six different measures of relational continuity constructed with administrative records of visits to physicians. Measures were constructed for persons who had three or more visits over one or two years. The different measures were compared and those measures that performed best were constructed for a five percent random sample of B.C. residents. The measures were constructed with three different configurations of visit data: primary care visits only; primary care visits and specialist visits; and primary care visits with specialist visits attributed back to the referring physician. The measures were also evaluated for their ability to predict hospitalization.

The way the measures were operationalized influenced their performance. In general, the measures constructed over a two year data window outperformed those constructed over only one year, both in their concurrence with self report of a regular source of care and in their prediction of future hospitalizations.

Similarly, the measures constructed with primary care visits with specialist visits referrals attributed back to the originating physician significantly outperformed those that included primary care visits only. This finding is consistent with the notion that specialty and primary care visits are "connected" (and thus continuous) through the process of referral.

Project B - Comparing continuity for primary care physicians and practices

This project focused on the question of whether, for patients who receive most of their care from physicians in group practices, calculating continuity on the basis of only one physician might be unduly restrictive and not representative of the continuous integrated care that they actually receive. The project focused on patients of practices with four or more full time-equivalent physicians. The patient populations for each practice were divided into a core group who made three or more visits in the study period and who made most of their visits to that practice, and others. Continuity measures were calculated for the core group of patients, based on physician level and practice level definitions. As with Project A, hospitalization was the main outcome measure.

About 30 percent of B.C. primary care physicians were found to practice within a large group. Large groups were the predominant practice organizations outside metropolitan Vancouver and Victoria. Thus, most B.C. physicians who practice in these settings were not affiliated with a large group. The distributions of physician continuity measures were similar to those found in Project A. As we expected, measures based on practice definitions were higher - patients got seemingly more continuous care if it was measured based on the practice rather than the individual physician. Physician level continuity measures (constructed with specialist referrals re attributed to the general practitioner) were statistically significant predictors of subsequent period hospitalization, while practice-level measures added no additional explanatory power.

This latter finding reflects the fact that physician based continuity was relatively high in these large groups; on average patients making at least three visits during the year saw a single physician for about three-quarters of them. Thus, most large practices do not appear to share common patient "lists."

Project C - Continuity of primary care in mental health

In this project, we attempted to measure continuity of primary care for persons with severe and persistent mental health conditions. Because of large gaps in the available administrative data for this population, several data sources - hospital records, pharmaceutical payments, a mental health service database, and physician fee for service records - had to be combined to define the cohort. Continuity measures were constructed using physician payment records as well as records of pharmaceutical prescriptions, treating each prescription as a physician "visit." The outcome measures were hospitalization, number of hospitalizations, and death in the subsequent period.

Because of gaps in the physician payments data, we analyzed pharmaceutical prescription data as a proxy source. In a sub study using a population for which we had relatively complete data from both sources, however, we found poor agreement - the pharmaceutical based measures tended to produce continuity measures considerably higher than those based on the physician payment data. Not surprisingly, then, using pharmaceutical based measures of continuity for this population produced results suggesting greater continuity of primary care than that found in Projects A and B.

In general the continuity measures were poor predictors of all of the outcome measures. Only continuity measures for those on income assistance with mental health conditions were significant.

The primary conclusion from this project was that existing administrative data, at least in B.C. for this period, provide an inadequate source of information for assessing continuity of primary care for patients with severe and persistent mental health conditions. Some combination of more comprehensive data and more nuanced continuity measures based on what data exist will be required if further progress is to be made. This should be a high priority for the research community, given the considerable potential impact (at least in theory and suggested by our limited results) of improved primary care continuity in this very difficult and costly population.

Project D - Continuity of primary care in persons with workplace injuries

This project compared continuity of care for people with workplace injuries to those without injuries, before and after an injury. Measures were constructed according to the methods developed in Project A. The population was identified using injury claims data from the Workers' Compensation Board. Continuity of care was slightly lower for those who had a workplace injury than for those who did not report an injury, both before and after the injury.