Averages Lie: Rebuilding Clinic Par Levels Around Procedure Mix and Delivery Variance
Photo: clinic supply room shelves labeled inventory bins, via tophealthdoctors.com.au
Most clinics inherit their min/max levels rather than design them. A practice manager opens the inventory module, finds par values entered years ago, adjusts one or two after a painful stockout, and moves on. The figures look authoritative because they occupy a software field, but they often trace back to a distributor representative's suggestion, a predecessor's habit, or a single month of usage divided by four.
For clinics that perform procedures, that inheritance is especially fragile. Procedure-driven demand does not arrive evenly across the calendar. It clusters around block scheduling, surgeon and clinician availability, seasonal case volume, and the periodic push to close out year-end referrals. A par level built from a monthly average assumes a smoothness the schedule never delivers.
Why the Average Misleads
Consider a clinic that consumes 120 units of a sterile drape pack in a typical month. Thirty per week sounds like a defensible planning figure. But if two of that month's four weeks carry 45 units each and the remaining two carry 15, a minimum set at one week of average demand will trigger a reorder after the first heavy week and still run dry before replenishment lands. The average was accurate. The plan was still wrong.
Averages conceal lead-time variance in the same way. A distributor may deliver in three days for most of the year and nine days when a manufacturer backorder ripples through the network. Planning against the three-day figure builds in no cushion for the tail, and the tail is precisely the scenario in which a stockout reaches the procedure room.
Three Inputs That Belong in the Calculation
Procedure mix, not total volume
Total case counts tell you how busy a clinic is. Procedure mix tells you what will actually be consumed. Map each stocked item to the procedures that require it, then translate the procedure calendar into demand bands: a high week and a low week rather than a monthly mean. Most clinics discover that a modest number of items serve a large share of scheduled cases, and that those items deserve a different planning treatment than the long tail of convenience stock.
Lead-time spread, not lead-time average
Request the 90th-percentile delivery interval from each primary distributor, not the median. If the answer is not available, reconstruct it from receiving logs over the past twelve months. Planning against the tail rather than the midpoint is the single most effective way to prevent a stockout without inflating every par level in the catalog.
Clinical criticality tiers
Sort items into three tiers. Tier A items cannot run out without canceling or delaying a case. Tier B items are substitutable with effort, a phone call, or a short drive. Tier C items are convenience stock whose absence irritates but does not interrupt. Safety stock belongs disproportionately in Tier A, and most clinics find that their current levels apply roughly the same cushion to all three tiers.
Where the Two-Bin Model Breaks
Two-bin kanban is an elegant system resting on two assumptions: reasonably steady demand and dependable replenishment. When both hold, the physical bin size becomes the reorder point and the system runs itself. Procedure-heavy clinics violate the first assumption routinely and the second often.
A subtler failure is drift. Someone buys a deeper bin to reduce trips to the stockroom, and the reorder point changes silently because no one updated the system. A year later the two-bin system is still nominally in place while the actual reorder point has moved twice. Two-bin is not the problem; it is a demand-driven method that requires demand inputs reflecting reality.
A Worksheet for Resetting Levels
Work item by item for Tier A first, then Tier B. A spreadsheet with the following columns is sufficient:
| Column | What to enter |
|---|---|
| Item and manufacturer number | Exact identification |
| Procedures served | Which scheduled cases consume it |
| Criticality tier | A, B, or C |
| High-week demand | Units used in a heavy procedural week |
| Low-week demand | Units used in a light week |
| Median lead time | Days, typical delivery |
| Lead time, 90th percentile | Days, worst realistic case |
| Current min and max | As entered in the system |
| Proposed min and max | Calculated below |
| Notes | Substitution options, storage limits |
A workable formula is daily high-week demand multiplied by the 90th-percentile lead time, plus a criticality buffer for Tier A items. Set the maximum at the minimum plus one high week of demand, or one full order cycle if that is longer. The result is a level tied to the schedule the clinic actually runs rather than a seasonal average.
Guardrails Against Inventory Inflation
Resetting par levels can quietly become an exercise in raising them. Three constraints prevent that.
- Cap total inventory value at the current level for the first quarter of the new system. Reallocate dollars from Tier C rather than adding new spending.
- Review only Tier A and Tier B in the first pass. Tier C can wait, and often shrinks on its own once attention moves elsewhere.
- Keep the calculation visible in the inventory system or an adjacent spreadsheet. Par levels that cannot be explained cannot be defended when a new manager or a distributor representative proposes changes.
What to Change on the Shelf
A recalculation that lives only in software will decay. Mark the minimum on the shelf edge with a printed label, and use a colored tag for Tier A bins so that a depleted bin is visible from the doorway rather than discovered at the case cart. Separate Tier A stock physically from the general supply shelves; the extra ten seconds of walking is repaid the first time a case is saved.
Set a Review Trigger, Not a Calendar
The final step is deciding when to revisit the numbers. Calendar-based quarterly reviews tend to become paperwork. A better trigger is a change in circumstances: a new procedure added to the schedule, a surgeon joining or departing, a distributor change, or two consecutive weeks outside the high or low band. Those events carry more signal than the turn of a quarter.
Par levels are a forecast expressed as a shelf rule. When the forecast is built from averages, it will fail on the weeks that matter most. When it is built from procedure mix, delivery spread, and clinical stakes, it can be modest and still hold.
