USPSTF forms a recommendation from two axes only — certainty of net benefit, and magnitude of net benefit — read off a fixed grid into a letter grade A / B / C / D / I. There is no strength axis and no separate values/preferences determinant; the whole GRADE certainty+strength apparatus is replaced by a certainty x magnitude lookup. This is the wiki’s first fully specified non-GRADE recommendation-formation instrument. (US Preventive Services Task Force, 2022)
Net benefit — the unit being graded
USPSTF grades neither an effect estimate nor a body of evidence per outcome, but a single composite:
«The net benefit is defined as the benefits minus the harms of the preventive service as» implemented in a general primary care population. (US Preventive Services Task Force, 2022)
Two consequences the definition forces:
- The benefit-harm balance is inside the graded object, not applied afterward. GRADE rates certainty in an effect estimate and then weighs benefits against harms at the strength step. USPSTF folds the weighing in first: certainty is certainty about the netted quantity. -> GRADE vs USPSTF - Two Appraisal Systems
- It is absolute and population-scoped — «as implemented in a general primary care population». Net benefit is a per-1000-persons quantity, which is why USPSTF «generally prioritizes ARR over RRR» -> Baseline Risk and the Relative-Absolute Split. (inferred from US Preventive Services Task Force, 2022)
The two axes
Certainty of net benefit — three levels (high / moderate / low): «likelihood that the USPSTF assessment of the net benefit of a preventive service is correct». Judged across the entire analytic framework, not per outcome, and «not a quantitative calculation, but rather a judgment based on the six critical appraisal questions». -> Rating Certainty of Evidence (US Preventive Services Task Force, 2022)
Magnitude of net benefit — four levels (substantial / moderate / small / zero-negative): «The Task Force rates net benefit as substantial, moderate, small, or zero/negative. “Substantial” net benefit indicates that the benefits substantially outweigh the harms, whereas “zero/negative” net benefit indicates that the harms equal or outweigh the benefits.» (US Preventive Services Task Force, 2022)
The two are computed separately and deliberately so: «The Task Force is careful to assess the certainty of the evidence and the magnitude of benefit, harm, and net benefit separately. For example, the Task Force may have high certainty of the overall evidence and still determine that there is small (or even zero) magnitude of net benefit.» (US Preventive Services Task Force, 2022)
The grid (Table 5)
| Certainty \ Magnitude | Substantial | Moderate | Small | Zero / Negative |
|---|---|---|---|---|
| High | A | B | C | D |
| Moderate | B | B | C | D |
| Low | Insufficient (I) | Insufficient (I) | Insufficient (I) | Insufficient (I) |
(US Preventive Services Task Force, 2022)
Read literally, the grid encodes three rules:
- Low certainty erases the magnitude column entirely — it is always I. «If the certainty of the evidence is low, the Task Force is unable to assess the magnitude of net benefit». So I is not a low grade on a continuum with A-D; it is the cell where magnitude cannot be read at all. -> The Insufficient-Evidence Statement
- Grade tracks magnitude across the top rows — high certainty runs A (substantial) -> B (moderate) -> C (small) -> D (zero/negative). The letter is essentially how big is the net benefit, given we are confident it is real.
- A negative net benefit at moderate-or-high certainty is D — «recommends against the service… moderate or high certainty that the service has no net benefit or that the harms outweigh the benefits». D requires the same certainty as a positive recommendation; you must be confident of no-benefit to say so. (US Preventive Services Task Force, 2022)
What each grade licenses (Table 4)
- A / B — offer or provide the service. A = high certainty substantial net benefit; B = high certainty moderate, or moderate certainty moderate-to-substantial.
- C — selective, preference-sensitive offering. «recommends selectively offering or providing this service to individual patients based on professional judgment and patient preferences. There is at least moderate certainty that the net benefit is small.» This is where USPSTF puts values — not as a determinant that weakens a recommendation (GRADE’s move) but as the action content of the small-net-benefit grade itself. -> Certainty of Evidence vs Strength of Recommendation
- D — discourage the service. Confident no-net-benefit or net harm.
- I — insufficient (see The Insufficient-Evidence Statement). (US Preventive Services Task Force, 2022)
Where values and cost live — and where they do not
- Values enter through grade C + shared decision-making, not through a strength axis: «For interventions where the relationship between benefits and harms is influenced heavily by personal preferences, the Task Force advocates that providers and patients engage in shared decision-making.»
- Financial cost is excluded; opportunity cost is not. «The USPSTF does not consider the financial costs of providing a service in its assessment of the balance of benefits and harms, but may provide contextual information regarding costs for use by providers, including cost-effectiveness studies.» Yet «Harms may also be considered in the form of opportunity costs for both patients and providers» — so cost enters as context, opportunity cost as harm, and neither as a priced axis in the balance. -> Which Objective Moved This Recommendation
- Harms are held to a lower evidence bar than benefits — «The quality of evidence for harms need not be as strong as that for benefits because of the ethical imperative to do no harm, especially when caring for asymptomatic persons.» (US Preventive Services Task Force, 2022)
Decision relevance
- The grid maps onto the wiki’s Layer-1 ranking more directly than GRADE’s strength does. Layer 1 ranks interventions by effect size x certainty; USPSTF’s grid is literally magnitude x certainty. GRADE deliberately severs strength from magnitude («strong recommendations are not necessarily high priority»), so its output cannot be read as a ranking; USPSTF’s can, because magnitude is a named axis. -> Layer 1 - Ranking Interventions for a Stratum. Caveat: USPSTF’s magnitude is of net benefit (harms already subtracted), Layer 1’s is of gross effect — close, not identical. (inferred from US Preventive Services Task Force, 2022)
- Reading a USPSTF grade: recover the two axes. A B can be high-certainty-moderate or moderate-certainty-moderate-to-substantial — different evidence states behind one letter. A C is not a weak B; it is a small net benefit known with at least moderate certainty, an affirmatively small effect, not thin evidence.
- I vs D is the load-bearing distinction — we cannot tell (low certainty) versus we are confident it does not help (moderate/high certainty, zero/negative magnitude). Collapsing them is the commonest misreading. -> The Insufficient-Evidence Statement
Worked cell — aspirin 2022 (the grid applied to a real topic)
The 2022 aspirin primary-prevention statement is the first held USPSTF recommendation whose two grid axes can be read off directly, and it lands two adjacent cells on the moderate-certainty row (US Preventive Services Task Force, 2022b):
| Recommendation | Certainty | Magnitude | Grid cell -> grade | Action content |
|---|---|---|---|---|
| Initiate aspirin, adults 40-59 at >=10% 10-yr CVD risk | moderate | small | moderate x small = C | individual/shared decision |
| Initiate aspirin, adults >=60 | moderate | zero/negative | moderate x zero-neg = D | recommend against |
This instantiates two of the grid’s structural rules on one topic. The C confirms the small-net-benefit-is-its-own-action-content reading: USPSTF concludes «with moderate certainty» a «small net benefit» and the licence is shared decision-making, not a weakened offer (US Preventive Services Task Force, 2022b). The D confirms the load-bearing D-vs-I distinction — it is a confident no-net-benefit («moderate certainty that initiating aspirin use… has no net benefit»), not an I: the evidence is sufficient, the magnitude is what is zero/negative (US Preventive Services Task Force, 2022b). The full effect/what-to-do layer lives on Aspirin for Primary Prevention of Cardiovascular Disease; here it serves only as the grid’s worked audit. Caveat — one topic is not a consistency test: that USPSTF read this topic onto the grid coherently does not show it applies the grid uniformly; that still needs more worked cells (inferred from US Preventive Services Task Force, 2022b).
Limits
- The grid is described by its own manual; application is under-sampled. Whether USPSTF applies the grid consistently across topics needs many worked cells, not one. The aspirin 2022 cells above are the first worked audit (moderate x small = C, moderate x zero/negative = D); the wiki also holds USPSTF statements (statins, multivitamins, supplements) that could supply more.
- The grid is coarse — two axes at 3x4 resolution, with the entire low-certainty row collapsed to one cell. It buys auditability and communicability at the cost of resolution, the same trade GRADE makes.
- Coherence, not validity (method-risks R1): the grid is an instrument for forming a recommendation transparently, not evidence any recommendation it produces is correct.
(inferred from US Preventive Services Task Force, 2022a)