MMHA 6520 Week 3 Data Source Reconciliation Review Example

Three Sources, One Member: Reconciling Social Needs Data Across Screenings, Z Codes and Care Manager Assessments

Student Name

Walden University

MMHA 6520: Health Informatics and Population Health Analytics

Instructor Name

Month Day, Year

Three Sources, One Member: Reconciling Social Needs Data Across Screenings, Z Codes and Care Manager Assessments

Purpose

Riverline Health Plan, a composite Medicaid managed care plan, keeps what it knows about members' social circumstances, whether they have enough to eat, a stable place to live or a way to get to appointments, in three places: screening answers sent by network practices, ICD-10 Z codes on claims, and assessments recorded by the plan's care managers. Leaders want to know how many members have social needs and which members to reach. This review asks whether the three sources agree, what each can and cannot show, and what role each should play. It uses a composite sample of members who appear in all three sources, so the comparison is fair.

The Crosswalk

Table 1 shows twelve members from the composite sample, selected to illustrate the patterns found. Each row records the member's status for food insecurity in each source.

Table 1

Food Insecurity Status by Source, Composite Sample

Members and results are composite and illustrative. Z59.4 is the ICD-10-CM code for lack of adequate food. Not assessed means the member was not in care management at the time.

The Pattern

Three patterns appear. First, screenings that flag a need are often not followed by a Z code on any claim. In the sample, most members who screened positive for food insecurity had no corresponding code. This matches national findings that Z codes are rarely used, even among hospitalized patients for whom social needs are documented elsewhere (Truong et al., 2020). Coding a social need takes time, rarely affects payment and depends on whether the clinician or coder thinks to do it.

Second, care manager assessments hold the richest detail. They record not only whether a need exists but how severe it is, how long it has lasted and what the member wants to do about it. But assessments exist only for members already in care management, which at Riverline is a small and selected group, typically those with high costs or recent hospital stays. Population health questions concern the whole membership and how outcomes are distributed within it (Kindig & Stoddart, 2003), so a count drawn from this group cannot answer them. A count based on assessments would describe that group, not the plan's membership.

Third, the sources sometimes disagree outright. Member C screened negative at the clinic but described food insecurity to a care manager; member I screened positive but told the care manager the problem had resolved. Disagreement can reflect a change over time, the setting in which the question was asked, or who asked it. A member may be more willing to describe a need to a care manager who has called several times than to answer a form in a waiting room.

Instruments Differ

Network practices do not all use the same screening questions. Some ask the ten core questions CMS published for its Accountable Health Communities model, which cover five domains from housing to personal safety (Billioux et al., 2017). Others use the two-question Hunger Vital Sign for food, a locally written form or questions built into their electronic record. A positive food screen at one practice may reflect a question about the past twelve months; at another, the past month. When results are pooled, these differences are invisible. Any count drawn from screening data should therefore be reported by instrument, or limited to practices using the same one.

A Role for Each Source

Screenings: prevalence among screened members. Screening data are the best available source for estimating how common a need is, but only among members who were screened, at practices using a known instrument. Reports should state the denominator, members screened, and should not be read as the plan-wide rate, since practices that screen may serve different populations from those that do not.

Care manager assessments: individual outreach. Assessments are the best source for deciding what to do for a particular member, because they include severity, history and the member's own preferences. They should drive referrals and care plans for members in care management.

Z codes: neither, for now. Until coding practice changes, Z codes undercount needs too severely to support either prevalence estimates or outreach. A member without a Z code is not a member without a need. The plan should continue to collect them, since a rise in coding would be useful in itself, and should consider working with high-volume practices to code needs identified in screening.

Recommendations

Three steps follow. First, the informatics team should label every screening-based report with the instrument used and the number of members screened, and should never present screening counts as plan-wide prevalence. Second, care managers should be able to see clinic screening results in the care management system, so that a positive screen prompts a conversation and disagreements like member C's are caught. Third, the plan should ask its five highest-volume practices whether they would adopt a common screening instrument and code positive results, offering technical help with the electronic record build. If that work succeeds, Z codes could become a useful source within a few years; for now, the plan should not rely on them.

Limits of This Review

The composite sample includes only members present in all three sources, which excludes most of the membership: members never screened, never in care management or never billed with a Z code. The patterns found here describe how the sources relate for those members, not how common needs are. The review also covers one need, food insecurity; agreement may differ for housing, where questions are more standardized, or for interpersonal safety, where members may be reluctant to answer at all.

References

Billioux, A., Verlander, K., Anthony, S., & Alley, D. (2017). Standardized screening for health-related social needs in clinical settings: The Accountable Health Communities screening tool. NAM Perspectives. https://doi.org/10.31478/201705b

Kindig, D., & Stoddart, G. (2003). What is population health? American Journal of Public Health, 93(3), 380-383. https://doi.org/10.2105/AJPH.93.3.380

Truong, H. P., Luke, A. A., Hammond, G., Wadhera, R. K., Reidhead, M., & Joynt Maddox, K. E. (2020). Utilization of social determinants of health ICD-10 Z-codes among hospitalized patients in the United States, 2016-2017. Medical Care, 58(12), 1037-1043. https://doi.org/10.1097/MLR.0000000000001418