Max Rady College of Medicine

Concept: Patient Allocation Algorithm: Assigning Patients to Physicians, Physician Groups or Clinics

 Printer friendly

Concept Description

Last Updated: 2020-06-04


    This concept contains information on the methods used in MCHP research over time to assign patients to physicians, physician groups or clinics in order to define a physician's practice population. A general approach to patient allocation is described, along with important considerations that should be decided before this process begins. Specific methods used at MCHP are described and two important "measures of care" that result from patient allocation are identified.

    In Manitoba, the population is not formally assigned to physicians and therefore they tend to visit more than one physician or a group of physicians over time. This creates complexity for research in which the physician is the primary unit of analysis. For analysis of continuity of care or majority of care, it is necessary to first derive a practice population for individual physicians or groups of physicians. This process of assigning patients to physicians or physician groups is called patient allocation.

General Approach and Considerations for Patient Allocation

    A general approach to patient allocation involves making decisions on the types of inclusions / exclusions and scope of physician involvement prior to assigning patients to physicians or physician groups. Once you have made a decision on these considerations, the conditions can be programmed and the SAS code can be run. This will create a working data file for your research that contains individual patient records for each visit with a physician. Once created, this file can then be used to allocate patients to a physician or physician group and to calculate specific indicators / measures related to physician services, such as total cost of services provided, continuity of care and majority of care.

    Some of the major considerations for patient allocation include:

1. What Types of Physicians Will Be Included?

    Decide on what types of physicians will be included in your research. Family physicians are usually included, but it may also include all, or a select group (e.g.: pediatricians) of physician specialists, depending on the focus of the research.

2. What Types of Physician Services Will Be Included?

    Decide on the types of ambulatory physician services that will be selected from the Medical Services (Physician Claims) data. For example, do you include emergency room and Personal Care Home (PCH) visits? This decision may involve excluding or including these types of visits based on the focus and scope of the research. Please see the Ambulatory Visits - Physician definition for more information on the types of visits should be considered.

3. Identifying Unique Physicians

    The same physician may have more than one unique MD number, therefore, it is necessary to create a physician identifier - call this variable PHYNO - that assigns those physicians with more than one MD number to a unique PHYNO. The PHYNO can be created based on a unique_md_number macro that is available in the MCHP SAS macro library.

    NOTE: The unique_md_number macro replaces the method that used phyn<yy>f formats (e.g. phyn98f) to assign a unique number to each physician.

    • In MCHP data, some of the MD values may be character values and these should all be converted to numeric values.
    • When converting MD numbers into unique physician identifiers, use the unique_md_number macro.

    For more information on unique physician MD numbers, please see:

4. Identifying Physician Groups / Clinics

    If required, you may need to identify the group practice or clinic that a physician works from. Currently, there is no simple way to identify physician group practices / clinics, as there is no group identifier available that indicates whether a physician is affiliated with a group / clinic or working in solo practice.

    The Medical Services (Physician Claims) data contains a variable called EUSLCD - (Electronic User Site Locator Code - previously known as TEXTCODE ) - that is used at MCHP to identify physicians who practice out of the same location. This is a site code that physicians use to file medical claims electronically. Physicians working within the same clinic or practice have the same EUSLCD value, so physicians with the same value are considered to belong to the same group. The process is not exact but it is the best we can do without a specific clinic identifier.


    1. Not all physicians can be grouped in this manner. For example, at one point in time, physicians who filed paper claims were all assigned an EUSLCD value = "000", which means that we cannot determine whether they have a practice associated with other physicians. Therefore, they are excluded from the analysis.
    2. The EUSLCD values merely tell us whether physicians work in a group setting. It does not indicate whether a physician is indeed affiliated with a group practice or whether they are essentially working as solo physicians in a shared space. This problem is unavoidable until all physicians file electronically or there is a provider group identifier.

    Identifying Solo Physicians
    The variable EUSLCD can also be used to identify physicians with a solo practice. Solo physicians were defined as those with an EUSLCD value associated with only one physician number. In Menec et al. (2000), they identified relatively few solo physicians (general practitioners and pediatricians) in Winnipeg. Less than 10% of the physicians having their usual practice in Winnipeg were identified as solo physicians satisfying the following conditions: they occurred in the data (Medical Services); they did not share their EUSLCD value with another PHYNO; and they did not have an EUSLCD value = "000".


    • One should check for both multiple EUSLCD values associated with a PHYNO and overlapping physician numbers in group practices (e.g.: for physicians who move). How to deal with these issues is up to the individual researcher.

      Example: Let's say you are trying to place physicians in specific clinics (e.g.: associate physician numbers ( PHYNO ) with distinct EUSLCD values, but when running frequencies, you notice that the same PHYNO appears in two different EUSLCD values (e.g.: PHYNO 0001 has claims under both 111 and 222 EUSLCD values). What do you do? If one EUSLCD has a very large percentage of the claims, then consider assigning the physician to that clinic. Then, you can either ignore the claims made to the other clinic, or group those claims all into the clinic where most of the claims were made. However, if the percentages are similar, the answer is less clear, and you should consult with the researcher.

    For more information on identifying physician groups / clinics, please see:

5 - Consideration of Low Users

    Patients with a low number of total visits during the study time period are considered low users. These individuals can have a marked effect on the measures of continuity and majority of care in a population. Should these low users be included or excluded in the analyses of patient allocation? This is a question for the researcher.

    The impact of low users was systematically explored by Menec et al. (2000). Individuals were called low users if they had less than four visits over the three-year study period. Low users may require special consideration, since the majority of care definitions do not conveniently apply to them. If a person makes only one visit he or she can be classified as a regular patient of that physician/group, regardless of which patient allocation method is used. As a result, these individuals are more likely to be classified as regular patients than low users. Low users were more likely to be male, younger, and healthier than high users. Income group was not associated with number of visits incurred. In addition, whether low users are included or excluded had a major impact on the size of the determined practice population for both rural and urban groups. When patients who made only one visit during the three year study period were excluded from the analyses, the size of the assigned practice population dropped - 9.5% for the rural groups and 12.9% for the urban groups, relative to the practice size when all patients were included.

    In MCHP research, "low users" are typically excluded from some of the analyses. Usually, a minimum number of visits over a specific time frame identify "low users". Examples of this in MCHP research include:

    • in Frohlich et al. (2006) they required a minimum of 2 visits in a one-year time period;
    • in Katz et al. (2010) they required a minimum of 3 visits over a 2-year period;
    • in Martens et al. (2010) individuals with less than three visits over a two-year period were excluded. Tables describing the proportion of different age groups of the population excluded from the analyses are available from the Continuity of Care glossary term in this report; and
    • in Chateau et al. (2015) they required a minimum of 4 visits over a 3-year period.

Patient Allocation Methods Used in MCHP Research

    The following section describes three different patient allocation methods / algorithms that have been used in MCHP research for investigating and reporting on physician practice populations.

1. Patients Allocated Based on Number of Ambulatory Visits

    In our earliest research with patient allocation, patients were assigned to physicians / physician groups based solely on the number of ambulatory visits provided. The physician providing the most number of visits was identified as the primary provider. This method was used in the following research:

    • Mustard et al. (1996) developed a measure where the physician providing the greatest proportion of care was defined as the regular source of care (RSOC). Please read the Regular Source of Care (RSOC) concept for more information on the methods used in this research.
    • Roos et al. (1998) developed a measure based on the number of visits and on the sequential nature of provider continuity. Please see the Measuring Continuity of Care concept for more information on the methods used in this research.
    • Menec et al. (2000) developed a measure, called Majority of Care that investigated different percentages of visits to a physician group/clinic. Please see the Measuring Majority of Care concept for more information on the methods used in this research.

2. Plurality Approach to Patient Allocation - (Katz et al., 2004 and 2016)

    This method assigns patients exclusively to the practice where they receive the highest intensity of care, determined by the total number of ambulatory visits - physician and/or the greatest proportion of expenditures (cost). In this method, all visits are allocated to an individual physician or group of physicians, regardless of who provided the service. Over time, different methods have been used. These are described below.

    PLEASE NOTE: The Plurality approach is the current MCHP standard method used to assign patients to a physician, physician group or clinic.

    The method is currently based on point 3 above, but has been modified slightly over time, depending on the needs of the research. For example, in Katz et al. (2014), this method only included direct costs of the ambulatory visits, and did not include any indirect costs.

    The advantage of this approach is that the resulting profiles align with the understanding that a physician's responsibility is to "integrate and organize care across visits, illnesses, and care sites for a defined population." The disadvantage of the plurality approach is that it doesn't take into account patients whose plurality of care happens elsewhere, resulting in skewed practice-based profile for clinics/physicians who provide the majority of service to patient who are not their own (e.g.: an off-hours 'walk-in' urgent care clinic) (Reid et al., 2001).

    NOTE: An example of the SAS code developed for investigating the plurality approach to patient allocation is available in the SAS Code and Formats section below (internal access only).

Katz et al. (2004)

    In Katz et al. (2004), four approaches were investigated:

    1. A patient is allocated to the physician with the most number of visits. In case of a tie, the patient is arbitrarily allocated to the physician with the lowest physician billing number.

    2. A patient is allocated to the physician with the most visits as defined by cost. In case of a tie, the patient is arbitrarily allocated to the physician with the lowest physician billing number.

    3. A patient is allocated to the physician with the most visits. In case of a tie, the patient is allocated to the physician with the greatest total cost. Total cost calculations may include direct care (i.e., visits) and indirect care (e.g.: referrals to other physicians or for services such as lab tests and x-rays).

    4. A patient is allocated to the physician who generates the most cost providing both direct and indirect care.

    In Katz et al. (2004), approach #3 was chosen due to the highest correlation values of .96 for visits and .79 for patients. Patients were assigned to the most responsible physician (based on number of visits and costs), and all services rendered to a patient were credited to that physician regardless of who provided the services.

Katz et al. (2016)

    In Katz et al. (2016), the patient allocation algorithm was based on the following rules:

    1. The patient is allocated to the primary care provider with whom the patient has had the most visits. In the event of a tie for the most visits, the providers with fewer visits with that patient are eliminated from this process and step two is performed with the remaining providers.
    2. The patient is allocated to the primary care provider with the highest total billings. In the event of a tie for the highest total billings, providers with lower total billings with that patient are eliminated from this process and step three is performed with the remaining providers.
    3. The patient is allocated to one of the remaining primary care providers at random.

3. Equivalent Approach to Patient Allocation - (Reid et al., 2001)

    Unlike the plurality approach, in the equivalent approach, patients can be assigned to more than one physician. The physician is allocated the percentage of care that it provided per patient. The percentages are then summed. For example: "If a practice delivered 50% of the care to one patient, 80% to a second, and 20% to a third, the clinic would be assigned a total of 1.5 'equivalent' patients." (Reid et al., 2001).

    The advantage of this approach is that it demonstrates all care provided by a clinic/physician regardless of where patients receive their care. The disadvantage is that it assumes that patient characteristics (including morbidity) have the same distribution as expenditures, where as in truth, expenditures are related to many factors beyond patient factors.

Measures of Care Related to Patient Allocation

    There are two common measures of care related to patient allocation that are used in MCHP research. Both these measures use the number of visits to physicians in their calculations. The measures are:

1. Continuity of Care

    Continuity of Care measures the extent to which an individual patient sees a given provider over a specified period of time. This involves the creation of a Continuity of Care Index that identifies the percentage of total care provided by each physician to an individual patient. For more information on this measure, please read the Measuring Continuity of Care concept.

2. Majority of Care

    Majority of Care is a measure of whether individuals receive most of their ambulatory care from a single provider (versus two or more other providers). This measure usually is set at greater than 50%, but can be a higher percentage (e.g.: 75%) if required. For more information on this measure, please read the Measuring Majority of Care concept.

Notes and Limitations

  • The Plurality approach and the Majority of Care approach are the same if a patient attended a physician group or clinic more than 50% of their time, however, below this percentage the Plurality approach would assign them to the physician group or clinic with the highest percentage, whereas the Majority of Care approach would leave them unassigned.

  • Grouping physicians is a difficult task given the data available to develop a comprehensive system for identifying physician groups or individual clinics. The definition of "physician groups" should be well defined in each study. For additional information on this limitation, please read section 4 - Identifying Physician Groups / Clinics above.

  • Additional information on the methodology for identifying physician groups / clinics can be found in the MCHP document: Medical Group Characteristics (General Practitioners) (internal access only) available in the Links section below.

Related concepts 

Related terms 


  • Chateau D, Enns M, Ekuma O, Koseva I, McDougall C, Kulbaba C, Allegro E. Evaluation of the Manitoba IMPRxOVE Program. Winnipeg, MB: Manitoba Centre for Health Policy, 2015. [Report] [Summary] (View)
  • Katz A, Valdivia J, Chateau D, Taylor C, Walld R, McCulloch S, Becker C, Ginter J. A Comparison of Models of Primary Care Delivery in Winnipeg. Winnipeg, MB: Manitoba Centre for Health Policy, 2016. [Report] [Summary] [Additional Materials] (View)
  • Katz A, Soodeen RA, Bogdanovic B, De Coster C, Chateau D. Can the quality of care in family practice be measured using administrative data? Health Serv Res 2006;41(6):2238-2254. [Abstract] (View)
  • Katz A, Bogdanovic B, Soodeen R. Physician Integrated Network Baseline Evaluation: Linking Electronic Medical Records and Administrative Data. Winnipeg, MB: Manitoba Centre for Health Policy, 2010. [Report] [Summary] (View)
  • Katz A, Chateau D, Bogdanovic B, Taylor C, McGowan K-L, Rajotte L, Dziadek J. Physician Integrated Network: A Second Look. Winnipeg, MB: Manitoba Centre for Health Policy, 2014. [Report] [Summary] [Updates and Errata] (View)
  • Katz A, De Coster C, Bogdanovic B, Soodeen R, Chateau D. Using Administrative Data to Develop Indicators of Quality in Family Practice. Winnipeg, MB: Manitoba Centre for Health Policy, 2004. [Report] [Summary] (View)
  • Martens P, Brownell M, Au W, MacWiliam L, Prior H, Schultz J, Guenette W, Elliott L, Buchan S, Anderson M, Caetano P, Metge C, Santos R, Serwonka K. Health Inequities in Manitoba: Is the Socioeconomic Gap in Health Widening or Narrowing Over Time? Winnipeg, MB: Manitoba Centre for Health Policy, 2010. [Report] [Summary] [Updates and Errata] [Additional Materials] (View)
  • Menec V, Black C, Roos NP, Bogdanovic B. Defining Practice Populations for Primary Care: Methods and Issues. Winnipeg, MB: Manitoba Centre for Health Policy and Evaluation, 2000. [Report] [Summary] (View)
  • Menec V, Sirski M, Attawar D. Does continuity of care matter in a universally insured population? Health Serv Res 2005;40:389-400. [Abstract] (View)
  • Menec V, Black C, Roos N, Bogdanovic B. What is the potential for formal patient registration in Canadian primary care? The scale of 'informal registration' in Manitoba. J Health Serv Res Policy 2001;6(4):202-206. [Abstract] (View)
  • Menec VH, Roos NP, Black C, Bogdanovic B. Characteristics of patients with a regular source of care. Can J Public Health 2001;92(4):299-303. [Abstract] (View)
  • Menec VH, Sirski M, Attawar D, Katz A. Does continuity of care with a family physician reduce hospitalizations among older adults? Health Services Research Policy 2006;11(4):196-201. [Abstract] (View)
  • Mustard CA, Mayer T, Black C, Postl B. Continuity of pediatric ambulatory care in a universally insured population. Pediatrics 1996;98(6 Pt 1):1028-1034. [Abstract] (View)
  • Reid RJ, Bogdanovic B, Roos NP, Black C, MacWilliam L, Menec V. Do Some Physician Groups See Sicker Patients Than Others? Implications for Primary Care Policy in Manitoba. Winnipeg, MB: Manitoba Centre for Health Policy and Evaluation, 2001. [Report] [Summary] (View)
  • Roos NP, Carriere KC, Friesen D. Factors influencing the frequency of visits by hypertensive patients to primary care physicians in Winnipeg. CMAJ 1998;159(7):777-783. [Abstract] (View)

Contact us

Manitoba Centre for Health Policy
Community Health Sciences, Max Rady College of Medicine,
Rady Faculty of Health Sciences,
Room 408-727 McDermot Ave.
University of Manitoba
Winnipeg, MB R3E 3P5 Canada