19.4 Using Data to Design CDI Workflow and Staffing

Using Data to Design CDI Workflow and Staffing

Staffing decisions translate the program's mission into review capacity. Begin with the population the program intends to serve, the tasks reviewers perform, and the documentation risks being addressed. Then compare demand with available time. Equal bed assignments, a purchased prioritization score, or another hospital's staffing ratio does not establish the right capacity for every department.

Choose an assignment model that fits the work

Geographic, specialty, and centralized assignments are practical design choices. They can also be combined without buying artificial intelligence or natural language processing software.

These are potential advantages and limitations, not universal outcomes. A service-line assignment need not require lengthy physical travel if the team communicates electronically or at scheduled rounds. A geographic model need not fail during a surge if there is a clear handoff and rebalancing process. Evaluate results rather than assuming one model is inherently modern or superior.

Estimate workload from activities and available time

Use local observations to estimate initial reviews, follow-ups, query work, reconciliation, education, audit duties, and other assigned work. Define which activities each observed time includes so that query creation is not counted twice. Separate paid hours from the hours actually available for the modeled tasks.

Worked staffing estimate

Consider an explicitly hypothetical daily review workload:

• 40 initial reviews at an observed average of 25 minutes: 1,000 minutes.
• 60 follow-ups at an observed average of 10 minutes: 600 minutes.
• Total modeled chart-review demand: 1,600 minutes, or about 26.67 hours.

If each scheduled specialist has 6 hours available for these review activities after other duties, the simple capacity estimate is $26.67/6=4.44$ specialist shifts. Five such shifts provide 30 hours of modeled capacity. This is an estimate for the stated activities, not a national five-person staffing standard or a guarantee that every day will fit.

Translate the estimate into an actual roster by considering variability, leave, training, coverage days, skill mix, and additional duties excluded from the model. Avoid double counting: if meetings have already reduced availability to 6 hours, do not add the same meeting time again to the modeled review demand. Follow-up demand also depends on the current census and review policy; admission counts alone do not capture it.

AHIMA recommends local pilots and review-time assessment when setting expectations. ACDIS publishes examples of different organizations' practices, including differing activities and staffing assumptions. Such examples do not establish a universal FTE-per-bed or FTE-per-discharge requirement. Exceeding a bed ratio does not mathematically prove that coverage must fall below a particular percentage.

Prioritize documentation need within the approved scope

A workqueue helps decide which eligible records to review first. It does not establish a diagnosis or guarantee that lower-ranked records have no documentation needs. Useful inputs can include an unresolved contradiction, a time-sensitive clarification, an approaching discharge, newly available test information, a clinical validation concern, or a documented review gap. Select priorities consistent with the program's mission and monitor what the process misses.

For example, antibiotics and a vasopressor may warrant closer evaluation of the record. They do not alone diagnose sepsis or septic shock. A postprocedure return to surgery or transfusion may warrant review, but neither independently establishes a reportable complication. Evaluate the documented circumstances and use a compliant query when clarification is needed.

A record with one CC or MCC may be selected for accuracy review. The purpose is to determine whether the documentation and reporting are supported, including whether an existing diagnosis needs clarification. Seeking another diagnosis merely to protect payment if the first is removed undermines that purpose. A second reportable condition must independently be supported and meet the applicable reporting rules.

Avoid declaring entire obstetric, orthopedic, observation, or short-stay populations “zero opportunity.” Eligibility, review depth, and priorities can differ, but clinical documentation remains important in each setting. MS-DRG 795 is Normal Newborn, not uncomplicated vaginal delivery. A short stay, low relative weight, or software score cannot certify that a record needs no attention. Audit an appropriate portion of lower-priority or unflagged records to test the queue's limitations.

Technology can help sort records. Human process owners remain responsible for compliant query practices. Under the 2026 query brief, oversight of technology-generated queries may occur before delivery or through structured ongoing auditing. Review directional bias, source information, answer options, and missed opportunities; do not assume automation guarantees equal workloads or eliminates the need for judgment.

Rebalance coverage using observed demand

If local data show a rise in medical admissions and a fall in elective surgical demand, consider temporarily moving appropriately trained staff to the overloaded queue. Check existing assignments and skill requirements, define handoffs, and continue to monitor the population losing coverage. Seasonal patterns are planning hypotheses; confirm them with the hospital's own data rather than assuming all hospitals experience the same winter decline in surgery.

Track backlog age, eligible encounters discharged before review, unresolved queries, available review hours, and audit findings. A floating assignment can be useful, but its size is a local capacity decision rather than a required percentage of a senior specialist's time. Reassess whether the change improves coverage without causing new omissions or inappropriate queries.

Test weekend coverage before promising a return

A business case should identify the actual problem. In a hypothetical four-week baseline, 90 eligible weekend admissions include 30 discharged before initial review: $30/90=33.3\%$. Record the cohort and definition. Do not compare this weekend-admission rate with a percentage of the whole weekly census and call the difference an improvement.

Pilot a defined weekend schedule, record the additional hours and costs, and measure the same outcomes over a comparable period. Include review quality, timeliness, handoff success, provider response opportunities, and the effect on weekday work. A different case mix or staffing change can explain part of a before-and-after difference. Compare like cohorts before attributing the result to weekend coverage.

Query-related DNFB balances may help identify a billing bottleneck. Distinguish gross charges held from reimbursement or cash actually collected. Releasing a hold is not automatically new revenue, and faster queries do not guarantee a particular agreement rate. Weekend staffing levels and coverage, response-time, and retrospective-query goals require a locally defined and evaluated plan.

Expand the program with setting-specific rules

Before adding pediatric, emergency, or ambulatory reviews, define the purpose, eligible population, payment context, coding rules, clinical expertise, and compliance oversight. Pediatric programs may use APR-DRGs where the applicable payer uses that system; not every children's hospital or payer uses the same classification. Emergency department review can support accurate documentation and appropriate collaboration with utilization review. CDI does not replace the practitioner admission decision, establish inpatient status by bed placement, or use inpatient uncertain-diagnosis rules for outpatient encounters.

For inpatient POA reporting, the inpatient admission order is the reference point. Documentation from earlier emergency or observation care can help establish timing; physical arrival on the inpatient floor is not the reference point.

For Medicare Advantage risk adjustment, identify the applicable CMS-HCC model and data-collection period. CMS describes a prospective model that uses base-year diagnoses and demographic information to predict payment-year expenditures. A risk score does not simply reset to zero on January 1. Do not equate Medicare Advantage, every commercial ACO, and Hospital Value-Based Purchasing with one HCC formula. Current conditions must be documented and reported under the applicable encounter rules; a historical diagnosis or a revenue target cannot justify unsupported recapture.

Present expansion proposals with explicit assumptions, costs, measurable objectives, and a review date. Preserve accurate reporting and independent provider judgment while testing whether the proposed staffing change actually serves the intended population.