Read Preparing the Master Patient Index text version

Preparing the Master Patient Index For an Integrated Delivery System

by Celia M. Lenson, RHIA and Linda M. Herr, RHIA Celia M. Lenson is executive director of Integrated Clinical Data, Concord, CA. Linda M. Herr is senior consultant with Integrated Clinical Data. Originally published in Journal of AHIMA, November-December 1995, Vol. 66, No. 10, pp. 56-60

In its Position Statement, "Managing Health Information in Facility Mergers and Acquisitions," the American Health Information Management Association (AHIMA) states, "To ensure availability of health information to all legitimate users, patient records should be consolidated or linked in the master patient index [MPI]."1 This statement was reiterated in a Healthcare Informatics article on the role of the MPI in an integrated delivery system (IDS). Golob and Quinn suggest that building an accurate enterprise MPI system is the first step toward addressing IDS needs. They state, "It is fundamental...that administration have the ability to track a patient across multiple facilities, through multiple episodes of care."2 Building the enterprise MPI is complex for many reasons. Variations in information systems, data capture, and institutional goals and objectives present multiple challenges to integrating patient data. Even on an individual institution basis, after technical and organizational problems are overcome, the purely operational task of linking patients across multiple entities is difficult due to duplication and error within the individual entity's MPI. Streamlining the individual entity's MPI is key to a successful enterprise MPI. This article discusses an operational approach to preparing the MPI for an integrated delivery system. Perception The MPI plays a major role in hospital information systems. It provides highway access to patient care, medical records, and billing information. Unfortunately, maintaining the integrity of the MPI has not been a high priority for hospitals. Typically, the MPI patient search is an unsophisticated Soundex or an alpha search using a simple algorithm to identify duplicate patient files. To compound this problem, registration departments where 500,000 to 1,000,000 registrations are created annually perceive 5 to 25 duplicates per day as insignificant. We have found this perception particularly in emergency room, outpatient, and clinic registrations at most of our client sites. Therefore, as HIM professionals, we find the first step in preparing the MPI is to have a reality check with all parties involved. Preparing the MPI: The Reality Check The first task is to educate the administration, medical information systems, (MIS), registration, and clinic staff members regarding the MPI and its direct ties: the patient's physical medical record, demographic and clinical data, and the investment the institution is about to make. When educating these parties, it is important to understand what is in the MPI file(s), to identify the satellite MPIs within the hospital, and to identify the number of duplicate/multiple files. These three steps begin the MPI education process. The first two steps consist of simple information gathering. The last step, however, requires assistance from MIS or a vendor who provides MPI data analysis. The MPI data analysis for identifying duplicate/multiple patient files typically is conducted via reports that identify duplicate/multiple patient files within the MPI. These reports, provided by MIS, usually consist of Social Security number error reports, misspelled last names, and date of birth errors. Unfortunately, we find that only 80 percent of the duplicate/multiple patient files are identified with this method. The question ensues: What happens to the unidentified 20 percent?

Copyright American Health Information Management Association. Reprinted with permission.

Page 1

We use software that matches patient records based on an algorithm using classical statistical theory. The software edits and standardizes the MPI files first: invalid values are deleted and hyphenated names and aliases are accounted for. Enhanced phonetic algorithms assure that thorough searches are completed on both first and last names. The software then applies several alternative search procedures to identify pairs of duplicate records for evaluation by employing a probability linkage algorithm. Patient records are grouped into "buckets" to determine which records are to be compared. The buckets are grouped three ways:

· · ·

Social Security number First name phonetic, birth date, sex Last name phonetic, first name phonetic

Because the software uses the above formulas, we find that the "fuzzy matches" are identified. "Fuzzy matches" are duplicate/multiple patient files that cannot be identified using ordinary phonetic procedures employed in vendor systems. The beauty of the software, so to speak, is that it identifies almost all duplicate/multiple files in the MPI. Once the duplicate/multiple files have been identified, the software assigns a confidence weight to the linked pair. See Table 1. The higher the confidence weight the greater the probability and vice versa.

Copyright American Health Information Management Association. Reprinted with permission.

Page 2

Table 1: Duplicate/Multiple Files

Medical Record # Original Name 2032200 2465447 Faith Parrish Faith Pierce

BMBDBY

SSN

LAdmit Sex Race Date F F 1

Partial Address

Confidence Weight 4.2

01-24- 04174471 966 01-24966

02-08-95 135 Broad

2356900 2716792

Gregoria 11-17Encarnacion 929 Gregoria Goitia 11-17929 Lelo, Rob Lelo, Robert 10-07988 10-07988

1252453

F F

5 1

01-02-95 360 Main St.

5.3

000000579906 000000579907

M M

W W

3-24-95

13.0

000000535354 000000535355

Gram, John Gram, James P

12-17992 12-17992

M M

W W

02-19-95 02-19-95

11.9

000000467158 000000467159

Janson, Amanda Marie Janson, Andrea M.

07-27990 07-27990 09-24990 09-24990

F F

W W

02-24-95 05-03-95

12.4

000000472472

Avila, Roger A Avila, Richard 0000004720473 R

M M

W W

09-10-94 03-18-95

10.6

Note: The names in this table are fictitious.

Copyright American Health Information Management Association. Reprinted with permission.

Page 3

Minimizing fuzzy matches is extremely important when building an accurate enterprise MPI. Table 2 shows some of the findings in reviewing MPI data with this software.

Table 2: MPI Errors % of MPI Duplicates 4. 42 7.82 7.18 4.34 1.46 2.01 11.05 11.65 2.74 2,875 # of Records to be Corrected

MPI Files

128,000 867,968 917,888 250,388 1,123,894 111,004 111,634 117,021 104,943

5,657 67,875 65,904 10,866 16,408 2,231 12,335 13,632

The average error rate is 5 to 10 percent for a single MPI, and 10 to 25 percent in overlap (multiple) MPIs. There are two major differences between single and overlap MPI files. The linkage files in overlaps have a higher confidence weight due to exact matches on name and birth date. This follows logically since internal duplication occurs primarily when patient demographics (last name) change or when registrars enter data incorrectly. Additionally, overlap pairs are characterized by a large number of linkages. In several overlap evaluations, there were more than 100,000 linkages. This makes human evaluation impossible. This data is crucial when beginning the educational process. It is not unusual to see a hospital without this data conduct two to five meetings to discuss MPI issues. What is surprising is that they discuss MPI issues without getting to the root of the problem. Diagnosing the MPI Once the MPI duplicate/multiple files have been identified, questions remain: How do we manage this project? Do we clean all the duplicate/multiple patient files identified? What is this going to cost? We have developed an eight-step work plan to tackle operational and system issues and the future of the MPI: (1) review MPI duplicate/multiple file reports; (2) review policies and procedures; (3) review storage; (4)

Copyright American Health Information Management Association. Reprinted with permission.

Page 4

review record retention policies; (5) review impact on other departments and interfaces between MPI and other MPI databases; (6) review clerical staffing; (7) review project management; and (8) review budget. Step 1: MPI Duplicate/Multiple Reports When medical records management staff members review MPI output, their first inclination is to clean the entire report. The MPI output can consist of as few as 2000 duplicate/multiple patient files, or as many as 100,000. We recommend reviewing the output to determine the types of errors created by registration and to determine what duplicate/multiple files absolutely need merging. Reviewing and editing MPI reports is a tedious process, but will have payback later by reducing MPI cost and providing quality improvement information to registration. Table 3 displays the data field and the average percentage errors by data field.

Table 3 Data Field and Average Percentage of Errors Data Field

% Discrepant 39 22 16 3 7 10 36

Last name First name Middle initials Birth month Birth day Birth year Social Security number

The data fields displayed are of particular importance when building, maintaining, and standardizing the MPI. They invite questions such as: what fields will be shared by multiple hospitals on an enterprise MPI? who will be the gatekeeper for updating the MPI data fields? and what procedures will be in place for overriding data? One client had to deal with a patient assigned two Social Security numbers. The obvious problem was how to decide which Social Security number to use. The client's first response was to use the most current data. However, the client never realized that it did not matter which Social Security number was used since neither was verified. The question the client should have posed was, how do we plan to verify Social Security numbers in the future--social security card, paycheck stub, or other evidence? Step 2: Policies and Procedures After reviewing the MPI output, we review and streamline the health information department merge procedures for maximum medical record number merge output. To date, we have modified all client medical record number procedures. What works with a small volume of corrections will not work with a large volume of corrections. Envision a health information department that has 20,000 duplications to

Copyright American Health Information Management Association. Reprinted with permission. Page 5

correct in six months. The department will need to correct 3,333 duplications per month to meet its deadline. Now imagine the procedure taking 45 minutes to merge each pair. The cost may become prohibitive. Step 3: Storage The once simple task of pulling and filing records in off-site record storage becomes problematic when merging files. It is important to communicate in writing the purpose of off-site record storage and the number of records to be pulled (300, 500, 800) per week. Read an existing off-site storage contract to determine if there are hidden costs to such a project. If you have microfilm records, develop new procedures to handle this population. Step 4: Record Retention Policies Record retention policies can be problematic when either converting a single MPI or merging multiple MPIs. The health information department must be able to track and retrieve every record within the institution. It is common to see two to four MPIs in the HIM department. We had one medical center merge with two hospitals in the last six years, with the following mix of MPIs: MPI on computer-350,000 files; MPI 3 x 5 card system--277,120 cards; MPI on microfilm--317,000 files; and MPI on second computer system--200,000 files. The medical center was planning to consolidate all MPIs, to clean duplicates within each MPI, and to identify overlap patients within all MPIs. The health information directors had different points of view about record retention and the consolidated MPI. The director of MIS was against converting all data. We reviewed all the MPIs and recommended a 19year record retention policy with the exception of minors and psychiatric patients. To convert existing manual MPIs to tape for a 25-year record retention period would cost $154,000; a 19-year record retention period would cost $64,000; a 10-year record retention period (19 years for minors and psychiatric cases) would cost $15,000. The directors of health information agreed on a 25-year record retention policy. No one used the manual MPI microfiche or the MPI 3 x 5 cards for retrieving patient records. This created further problems. For example, new numbers had been created for these patients, so when the manual MPIs were converted, the hospital created additional duplicate numbers. Of course, this created more work in the individual MPI clean-up. Step 5: Impact on Other Departments and Interfacing When beginning a large project, it is important to take inventory of the departments that may be affected by medical record number corrections. We find many departments with independent MPIs. We maintained these MPIs for retrieval purposes. Table 4 illustrates sample inventory information.

Copyright American Health Information Management Association. Reprinted with permission.

Page 6

Table 4: Sample Inventory

Department Radiology

Do they have an existing MPI? Yes

How many years on MPI? 2

Is the file Is the MPI system an alpha interfaced or medical with hospital- record number based MPI? system? Yes alpha medical record # alpha

Do they correct folders? No

What is the work standard for correcting numbers?

Laboratory Clinic

Yes Yes

1 3

No Yes

Yes Yes

5 mins. 2 mins.

Step 6: Clerical Staffing Clerical staffing can be the most challenging facet of the MPI project. It is critical that clerks are well managed to ensure budgetary compliance, quality work, and high output. Hiring competent clerks for a short term project is difficult. Health information managers must be able to recruit in an expeditious manner. Unfortunately, health information managers do not have 5, 10, or 15 clerks at their disposal. Recruitment and coordination with the human resources department is difficult, especially in a union setting or when recruitment takes 60 to 90 days. The skills required of clerks are flexibility, and the ability to follow procedures and to work with minimal supervision. A basic understanding of computers is mandatory. Concentration and attention to detail is important. Training an individual generally takes 15 to 20 hours with a 15 to 30 day learning curve. Training includes becoming familiar with the HIM department, filing and retrieving records, merging files on computer, correcting systems if interfaces do not exist, and correcting folders. If the file clerk is not up to speed after 15 to 30 days, release the clerk and start again. These conditions typically influence health information managers to outsource. However, outsourcing may not be worthwhile if the makeup of the duplicate/multiple files identified is less than 2000 linkage pairs or the institution has allowed ample time (6 to 12 months) to complete a project. Step 7: Project Management The success of a project rests with the project manager. The project manager typically is a supervisor in the HIM department or an outside consultant. The project manager must: oversee a budget that can be as low as $36,000 or as high as $200,000; coordinate workspace, terminals, and passwords for staff members; recruit, train, monitor and enforce staff disciplinary actions; ensure that staff members meet work standards; monitor the MPI schedule to ensure target completion date; provide weekly statistical reports to administration or MIS departments addressing major obstacles and their projected resolutions; and communicate problems with the MPI and possible enhancements decreasing workload. The project manager must be a self-starter who solves problems creatively. The project manager also must be a good communicator with sharp analytical skills. Avoid the type of project manager who gets entangled in detail without knowing how to reengineer work flow.

Copyright American Health Information Management Association. Reprinted with permission.

Page 7

MPI project time varies according to procedures, complexity of MPI merges, and system issues. We find project management hours average 22 percent of the project's clerical hours. Step 8: Budget Typically, MIS will assign dollars to a budget before any background work begins. This type of budget, however, has no backbone. In developing the budget, it is important to do all the background work before submitting a dollar amount to the facility's administration. Background work should consist of performing steps 1 through 8, as discussed in this article. Also, assign hours to clerical staff and project management. Be sure to allow for materials (labels, folders, outguides, etc.) and outside vendors who might convert manual systems to tape, analyze MPI data, provide outside storage facilities, and so on. One hospital performed all the work plan steps except step 2: Policies and Procedures. Initially MIS reviewed the work flow, but did not suggest procedural improvements. The budget assigned was $210,000. When the process was reviewed again and simple changes to procedures made, the cost was reduced to $60,000. Sometimes hospitals do not want to go through this exercise and thus just assign a dollar amount. The rule we use is that it will cost the hospital between $7 and $10.50 a pair to correct. The caveat is that the hospital can save money if the process is done methodically. Specifically, we have found, by going through these steps, the hospital can reduce cost between 30 to 50 percent of the original cost (based on existing systems with no modifications to procedures). Merging of healthcare organizations and development of integrated delivery systems highlights the importance of the MPI. Streamlining the individual entity's MPI is not only key to a successful enterprise MPI, it is a first step to both organizational and budgetary success. Making MPI integrity a priority is essential to assure the quality and accessibility of clinical information.

Table 5: Management Hours

Hospital

Management hours 619 277 210 1,060

Clerical hours

Number of files

A B C D

1,966 1,660 931 6,640

40,478 24,762 3,148 108,496

Notes 1. Wanerus, Priscilla, and Mary D. Brandt. AHIMA Position Statement, "Issue: Managing Health Information in Facility Mergers and Acquisitions." Journal of the American Health Information Management Association 65, no. 4 (1994): insert. 2. Golob, Randy, and John Quinn. "America's Best Networked Organizations." Healthcare Informatics 11, no. 11 (1994): 84.

Copyright American Health Information Management Association. Reprinted with permission.

Page 8

Information

Preparing the Master Patient Index

8 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

161864


You might also be interested in

BETA
Microsoft Word - GSA_pricelist_product_thru Mod 7.doc
Preparing the Master Patient Index
http://www.medicaltabletpc.com/content/view/144/26/