Flu season forecasting with year over year data for November staffing

Why year over year works for flu season planning

Flu season planning fails when it relies on gut feel or a single bad year. Most outpatient clinics and many hospital based ambulatory departments see a repeatable seasonal pattern that is visible in appointment volume, call volume, walk ins, and nursing tasks. Year over year comparisons are not perfect, but they are usually good enough to set staffing targets for November when you treat them as a planning baseline, not a promise.

The goal is not a perfect prediction. The goal is to reduce preventable overtime, reduce last minute schedule changes, protect patient access, and keep the clinical team from burning out in the first three weeks of the surge.

Year over year planning is especially useful when your clinic has stable hours, stable service lines, and stable panel size. It is less reliable when you have major provider turnover, a new location, large payer changes, or a new scheduling policy that affects access. In those cases you still use year over year data, but you adjust with a short list of known changes and you build wider safety buffers.

Choose the right demand signals

Start with the simplest demand signals that you can extract consistently. You want signals that are hard to game and easy to measure.

Good signals for flu related surges include

• Completed visits by day
• Same day add ons by day
• Nurse visits and vaccine visits by day
• Telephone encounters by day
• Portal messages by day
• After hours nurse line volume by day
• No show rate by day
• Average visit length by provider template by day

Pick two to four signals. If you pick too many, you will argue about the data instead of staffing. If your EHR reporting is slow, prioritize completed visits and same day add ons first, then layer in message volume if it is reliable.

Also track constraints that create workload without creating visits

• Prior authorization requests
• Lab result callbacks
• Medication refill requests
• Care gap outreach that still runs during surges

If you ignore these, your forecast will understate real workload for medical assistants, nurses, and front office staff.

Build a clean year over year baseline

Use at least two prior years if possible. Three years is better. If you only have one year of trustworthy data, treat the output as a starting point and keep your buffers larger.

Do not compare raw totals only. Compare patterns by week and day because scheduling happens on specific days. November staffing is often stressed by Monday and Tuesday peaks plus holiday week disruptions.

A practical baseline process looks like this

• Pull daily totals for your chosen demand signals from September through December for each prior year
• Mark holidays and known closures for each year
• Mark policy changes and one time events that affected access, such as a mass vaccine clinic, provider leave, or construction limiting exam rooms
• For each week, compute a week index that compares that week to a stable reference week in September or early October

The index approach helps when your clinic is larger this year than last year. You can apply the index to your current year baseline volume rather than trying to copy last year volumes exactly.

Example of a simple index logic using words, not formulas

• Identify a stable week in September for each year
• Compute the ratio of each November week volume to that September week volume for that same year
• Average those ratios across years
• Apply the averaged ratio to your current year September baseline volume

This yields a November forecast that adapts to growth and still captures the seasonal lift.

Adjust the baseline for known changes

Year over year data assumes the system is similar. Make it similar by applying a small number of explicit adjustments. Keep this list short, documented, and agreed by clinical leadership.

Common adjustment categories include

• Panel growth or loss, such as new contracts, new providers, or a new employer group
• Template changes, such as more same day slots or fewer nurse visits
• Service line changes, such as new urgent care blocks, new vaccine offerings, or added chronic care management visits
• Location changes, such as a new site that will absorb demand
• Documentation and inbox policy changes, such as message routing that shifts work between roles

A practical way to adjust without overengineering is to convert each change into an expected percent effect on the demand signal you use. If you cannot estimate the percent with confidence, do not pretend you can. Instead, place that uncertainty into your buffer staffing plan.

Convert forecast demand into staffing workload

Forecasted demand is not staffing. Staffing depends on how long work takes and who can do it.

Start by translating demand into workload blocks by role. You can do this with time standards, even if they are approximate. The standard does not need to be perfect. It needs to be consistent and reviewed.

For many clinics, a workable starting standard set is

• Rooming and turnover time per visit for medical assistants
• Vaccine visit time including screening, administration, documentation, and supply restock
• Phone call handling time for front office and clinical triage
• Portal message handling time for nurses and providers, separated by message type if possible
• Lab draw and result follow up time per lab heavy visit types

Then compute capacity per FTE per day. Capacity depends on shift length and non patient time. If you schedule an eight hour shift and assume eight hours of productive time, your plan will fail. Build in time for huddles, breaks, supply stocking, training, and the unplanned interruptions that are normal in November.

A realistic capacity approach is

• Define productive minutes per staff member per day, such as six hours and thirty minutes
• Assign non patient time categories explicitly so they do not disappear
• Apply the workload standards to forecast volume to estimate required staff hours by role per day

Now you have a staffing need curve that can be mapped to shifts.

Plan November staffing around peak days and peak tasks

November stress often comes from a few peak patterns, not the entire month equally.

Typical peak patterns include

• Early month surge when flu and other respiratory symptoms increase
• Early week surge when patients delay care over weekends
• Holiday week distortions when fewer appointment slots create more messages and more calls
• Vaccine demand spikes when employers or schools promote campaigns

To handle this, you should schedule for peak days with buffers, and then use flexible time on lower days to recover.

Practical scheduling tactics that work in real clinics include

• Add coverage on Mondays and Tuesdays for front office and medical assistants
• Add a short midday surge shift for phones during the busiest call window
• Add a dedicated vaccine block staffed by a nurse or trained assistant so it does not consume rooming staff for regular visits
• Reserve a small number of same day slots early in the day to reduce late day overload
• Use a defined float role for an experienced medical assistant who can shift between rooming, vaccines, calls, and supply work

These tactics only work if you protect them. If your float role is always pulled into rooming, then you did not create float capacity. You created an illusion.

Build a buffer plan that you can actually execute

A buffer plan needs to be concrete. It should not be a vague statement about calling in help if it gets busy. Your buffer should be scheduled, trained, and reachable.

Useful buffer sources include

• Per diem staff prebooked for peak weeks
• Cross trained staff from related departments who can support phones or vaccines
• A small internal pool of staff who prefer extra hours and can be preapproved for overtime within limits
• Temporarily expanded hours for part time staff who want more shifts

Define clear activation triggers. Triggers should be measurable and visible daily.

Examples of triggers that work

• Same day add ons exceed your planned level for two consecutive days
• Call wait time exceeds a set threshold in two windows in the same day
• Portal message backlog exceeds a set count by noon
• Staff report missed breaks for two consecutive days

When a trigger is hit, you execute a predefined action. The action should specify the role and the shift time, such as bringing in a per diem medical assistant from 11 to 7 or adding a phone support shift from 8 to 12.

Protect quality and safety while you increase throughput

Flu season staffing is not only about volume. It is about safe throughput. When the team moves too fast, documentation slips, vaccination screening errors rise, and infection control steps get skipped.

Build non negotiables into your staffing plan

• Ensure enough staff to allow breaks without abandoning rooms
• Ensure a clear triage path for high risk symptoms
• Ensure supplies are stocked daily so staff do not spend time hunting materials
• Ensure cleaning and turnover standards remain intact

If you are short, it is safer to reduce low value appointments than to overload staff until errors happen. This should be an explicit leadership decision, not a quiet drift.

Operate the plan with a daily cadence

Forecasting is only valuable if you operate to it.

Set a daily operational cadence for November

• A short morning huddle that reviews yesterday demand vs forecast and today staffing vs plan
• A midday check that reviews triggers and backlog items
• An end of day recap that logs what shifted demand and what staffing actions were used

Keep it short and consistent. The goal is decisions, not reporting theater.

Assign ownership. One person should own the daily demand review. One person should own staffing actions. They can be the same in a small clinic, but the responsibilities must be clear.

Common forecasting mistakes and how to avoid them

Treating last year as the only truth

One year can be an outlier. Use multiple years, and when you cannot, enlarge buffers and shorten the feedback loop.

Ignoring non visit work

Messages, calls, and prior auth can exceed visits as the main driver of stress. Include at least one non visit signal and translate it into workload.

Using averages that hide peaks

A monthly average hides the Monday and Tuesday crunch. Forecast by day and plan for peak days.

Scheduling buffers that are not real

If buffer staff are not trained, not credentialed, or not actually available, they are not buffers. Validate availability in advance.

Failing to communicate the plan

If staff do not understand why schedules change, they assume leadership is disorganized. Share the logic, share the triggers, and share what will happen if the triggers are hit.

A practical November staffing playbook you can run this year

Use this as a simple implementation sequence.

Two to three months before November

• Pull and clean daily demand data for at least two prior years
• Build week indexes and create a first pass daily forecast for November
• List known changes and apply small explicit adjustments
• Define time standards by role for major tasks
• Convert forecast to required staff hours by role per day

Four to six weeks before November

• Draft schedules around peak days with a visible buffer plan
• Confirm per diem availability and cross training coverage
• Prepare vaccine workflows, supply plans, and room allocation
• Agree on triggers and actions with clinical leadership

During November

• Run daily huddles and midday checks
• Track demand vs forecast and backlog metrics
• Activate buffers quickly when triggers hit
• Protect breaks and safety steps
• Log lessons weekly so you can improve next year

Year over year forecasting is not glamorous. It is a discipline. When you do it with clean data, honest assumptions, and an executable buffer plan, November becomes manageable. Staff feel the difference immediately because the schedule starts to match reality instead of reacting to it.