1 Patient Origin Study Guidelines PRIMARY CARE NEED
Transcription
1 Patient Origin Study Guidelines PRIMARY CARE NEED
Patient Origin Study Guidelines PRIMARY CARE NEED/DEMAND ASSESSMENT PART 1: SERVICE AREA DEFINITION A health center’s service area is defined as the geographic area in which 80% of its current patients live. It should be: • outlined on a map, and • defined in U.S. Census Bureau geopolitical units (e.g., village, town, township) listing either counties, census tract numbers or minor civil divisions. The percent of patients living in each of these census areas must be identified. Use census tracts in Standard Metropolitan Statistical Areas (SMSAs) and minor civil divisions elsewhere. If an area is not in an (SMSA) but is tracted urban area, census tracts may be used. Note: Throughout this guidance it is assumed that centers will use the most current census boundaries and data. EXAMPLE: The center’s service area consists of the following census tracts and patient loads: Census Tract Number Health Professional Shortage Area Estimated # of Patients % of Patients by Census Tract *107 109 110 56 57 58 No residents of defined service area yes no yes yes yes no N/A 2,000 500 1,000 1,500 2,700 300 2,000 20% 5 10 15 27 3 20 10,000 100% *Census tract where center located. Census tract or minor civil division breakdowns for the center’s area can be obtained from local city or county planning offices. 1 In order to locate current patients the center will need to identify their local addresses. This can be done either by prospectively counting all patients for a month or retrospectively by random sampling of clinic records. The center should choose the easier method. The center may wish to sample some records to see how complete their addresses are and how easily they may be matched to United States Census areas. If, for example, more than 20% of records show only RFD numbers or third party addresses (e.g., Medicaid), the center should turn to a prospective universal count. METHOD A: Prospective Universal Count: Over a month period record the current address of each patient as he/she requests service. Payment source and position on sliding fee scale (e.g. 50% of 100, 150, or 200% of poverty) could also be recorded if this information could be used later. METHOD B: Retrospective Universal Count (For those centers with Computerized Record Systems): The center may choose to do a universal count retrospectively as long as only the records of those users with an encounter within the last 2 years are used. METHOD C: Retrospective Random Sampling: The appropriate set of center records (billing, registration, or medical) is randomly sampled for information on patient residence and payment source. Since we are interested only in current users, only records of patients seen within the past 2 years should be sampled. • Sample Size For centers having fewer than 2,000 patient records, the sample must consist of at least 100 records. • All other centers will sample 200 records. • Muti-site projects • Grantees which have multiple sites that maintain separate records systems may apportion the total sample size among the sites based on the percentage of records kept at each site. (For example, if billing records were being pulled for this information, 20% of them would be sampled where the site generates 20% of the total project billing records.) If the apportioning yields on or more sites with less than 50 records to be sampled, then a minimum of 50 records should be included in the sample data for each of the sites. 2 • Procedure 1) Determine which set of center records will be sampled for the information (billing, registration, medical, etc.). 2) Put the records in order, alphabetically or numerically. 3) Start by pulling the 52nd record (The number 52 was chosen from a table of random numbers.) 4) If this set of records is filed in one place then pull the records as in this example: EXAMPLE: The center has decided that medical records will be sampled because each record contains the needed information on patient residence and payment source. It is also known that center files these records in one place and that there are a total of 1,200 presently filed. Since the center has fewer than 2,000 records, a total of 100 records will be pulled, or every 12th record in the file. The first record pulled is the 52nd one from the beginning of the file then every 12th record after this one is pulled until 100 records are obtained. Each family counts as one record. Remember to pull only records of current users (those patients or families with an encounter within the last two years.) If a pulled record cannot be used, then select the next record in the file and resume counting from that one. If the sampler reaches the end of the file without completing the sample, begin again, at the front while keeping the same sequence. Record how many records were not used because they were too old, had incomplete addresses, etc. 5) If the center chooses to sample a type of record which is not filed in one place, then sampling can be done as in this example: Example: The center has decided that billing records will be sampled because this is the only record system which contains information on both patient address and payment source. However, these records are divided into three separate files: File #1 (Awaiting billing)……… 200 records File #2 (Awaiting payment)…… 600 records File #3 (Payment received)……1200 records 2,000 records Since the center has 2,000 records, 200 records will be sampled. To make sure that the sampling is done proportionately to the number of records in each file, this number of records was sampled from each: 3 File # % of Total Records 1 2 3 10% 30% 60% # Sampled 20 60 120 200 records in sample Beginning with the 52nd record in each file and every 10th record thereafter, 20 records are pulled from File #1, 60 records from File #2, and 120 records from File #3. The total number of records pulled is thus 200. And, as in the previous example, only records of current users are to be sampled. If a pulled record is not usable, the next record should be pulled and the count started over with that record. Finally: (1) match the addresses to the appropriate Census division; (2) outline the service area on a map. Be sure to draw boundaries only around U.S. Census areas (e.g., do not show ½ a census tract or minor civil division); (3) list the census areas, whether they are MUAs or HPSAs, or state designated underserved areas, the estimated number of patients, the percent in each census area, and the service delivery sites’ locations. Note that: Use of zip codes rather than addresses is not accurate enough for this purpose. 4 PRIMARY CARE NEED/DEMAND ASSESSMENT APPENDIX A TECHNIQUES FOR CONDUCTING PATIENT ORIGIN STUDIES A patient origin study is a method for determining those geographic areas from which a major portion of a primary care center’s current patients are derived. (Current patients are those patients who have been seen within the last year.) Patient origin studies can also be used to compile information on the economic status of center patients, their insurance coverage patterns, and their range of health problems. To conduct a patient origin study, a center must have the following information: 1. A complete listing of the center’s current patients and their addresses, and 2. A census tract map by which patient addresses can be linked to specific geopolitical (e.g., census tract, county) boundaries. The decision as to how the patient origin study will be conducted will depend on the center’s age, the manner in which patient records are maintained, and the total number of center patients. There are three options: Method A – Prospective Universal Count Information is recorded on the residences of all patients who are seen in a given month. Since patient utilization will show seasonal variations, prospective counts should be done for a month when patient utilization patterns are considered typical. Method B – Retrospective Universal Count Information is recorded on the residences of all current patients (i.e., patients seen during the last year). Method C – Retrospective Random Sampling Information is recorded on the residences of sample of current patients. The reliability of the patient origin results will depend on the reliability of the information base that is used. The information base should be as complete as possible. If the entire patient file is not used, sampling procedures should be chosen to ensure that the patient information recorded is representative of the center’s patients. Centers should select the method that is suitable to the center’s resources and easiest to perform. 5 1. RECOMMENDED PROCEDURES FOR COMPLETING PATIENT ORIGIN STUDIES 1.1. METHOD A – PROSPECTIVE UNIVERSAL COUNT This method is particularly appealing for a small center or a new center. A center may also choose this method if its patient files cannot easily be used to compile information on patient residences because the addresses of registrants are either incomplete, out of date, or not suitable for assigning to census tracts. The following cautions should be observed if this method is chosen: 1. In recording patient information, the center should be aware of seasonal variation and should select a month considered to be typical of patient utilization. 2. In recording patient information, attention should be given to users rather than encounters. Information on individual patients should not be compiled more than once. The objective of this method is to obtain information on a sufficient number of patients to be able to generalize to the center’s entire patient population. Centers with fewer than 2,000 records, should collect information on at least 100 records. Centers with more than 2,000 records should collect information on at least 200 records. If a center has an average of 400 encounters per month, an unduplicated count of users would be estimated at between 200 and 300 per month, an appropriate number for determining service area boundaries. Some center may need to record patient information for slightly more than one month to obtain information on at least 200 users. Recording information will be facilitated by the use of a form such as the one provided below. This will allow the recording of all the information necessary for the need/demand assessment including patient address, payment source, and income. As each patient arrives, the appropriate information should be recorded during the sample month. Patient index cards should be filed alphabetically so that persons with multiple visits can be identified and duplication in the recording of information can be avoided. 6 SAMPLE INDEX CARD Sex: Age: (use date of birth & age) Ethnicity: (if available) Address: Census Tract #: Minor Civil Division: Payment Source: level) Family Income: (note sliding fee eligibility as well as annual income Family Size: Name: (use to check sex of patient) Health problem prompting the visit: 7 1.2 METHOD B – RETROSPECTIVE UNIVERSAL COUNT This method is generally used when an automated data base is available. In order to use a retrospective universal county to conduct a patient origin study, a center must be able to: 1. Identify current users (those patients who have been seen during the last year). 2. Link patient residence information to geopolitical boundaries. The Census Bureau has a Street Index Finder for matching residences to census tracts. While compiling information on patient residences, centers should also obtain other information that will be helpful in completing the need/demand assessments such as: 1.3 1. Information regarding the age/sex distribution of patients. 2. Information regarding patient income and family size (or sliding fee scale information). 3. Information regarding patient payment source. 4. Information regarding those other patient characteristics that many influence service utilization and will be useful information for strategic planning, such as age, sex, race or ethnicity. 5. Information regarding health status problems. METHOD C – RETROSPECTIVE RANDOM SAMPLING This method is a means by which summary information regarding patient residences can be obtained through sampling of a center’s patient records. The objective of this method is to obtain information on a sufficient number of patients to be able to generalize to the center’s entire patient population. Centers with less than 2,000 records should sample at least 100 records. Center with more than 2,000 records should sample at least 200 records. Any set of patient files can be used if they contain the required information. (To avoid bias in the sampling of the records, the patient files should not be organized by patient characteristics. An alphabetic or numeric ordering is preferred.) Once the patient record system to be sampled has been chosen, procedures for conducting the records sampling should be as follows: 1. Determine the manner in which sampling will be conducted. This total should be divided by the number of records to be sampled to determine which records should be pulled. 8 EXAMPLE: 1,200 divided by Patient records on file 2. 100 = Number of records to be sampled 12 Record selection count (e.g. every 12th record to be pulled) Start by randomly selecting a number for the first record selection. For the example above you might start with any number from 1 to 12, such as 3. (This was chosen from a random number table.) Example: Pull the third record and compile the necessary information. Continue pulling every 12th record. 3. Every sampled record must meet the selection criteria: 1) patients must have been seen in the last year and 2) sufficient information must be available for the need/demand assessment. If a record does not meet these criteria, the next record in the file should be pulled until an eligible record is found. Center record systems are often organized by family. Information on patient origins can be obtained by using any household member. If the patient origin study is also being used to profile the age/sex distribution of center patients, attention should be given to the selection of patient records. To avoid over selection of the heads of household, centers should choose the second, third, or fourth household) and then continue counting records from this eligible record. When the end of the record system is reached, the count of records should begin again at the start of the record file if a sufficient number of records have not yet been sampled. 4. A count should be maintained of those records found to be ineligible for the sample. This information should be presented in the need/demand assessment. 5. The information required for the need/demand assessment should be recorded on an index card (as displayed for Method A). 9