Population-level statistical attribution — not individual causation.
Outputs estimate excess deaths attributable to nurse-to-patient ratio deviation during vacancy periods, using the peer-reviewed Aiken dose-response relationship.
The MAM does not identify specific patients, determine individual causation, or constitute a finding of negligence. Column D applies a [THEORETICAL] suppression multiplier.
Select an evidence tier — each adds exactly one analytical layer
Col B · OIG 2025 — Annual attributable deaths
—
vs. Col A baseline · —
—
attributable deaths per year
Deaths / Vacancy
—
National Estimate
—
Annual Departures (V)
—
Option 3 Dual-Output · National vs Facility Baseline
Annual deaths · selected tier
Output 1 · National baseline VERIFY
—
B_mort_nat 2.9% · CMS Hospital Compare
Output 2 · Facility CMS rate BENCHMARK
—
B_mort_fac — · enter above to activate
Output 3 · Facility deviation (Δ)
—
B_mort_fac − B_mort_nat · positive = above national avg
Enter a facility-specific CMS 30-day mortality rate above to activate Outputs 2 and 3. A positive delta means your facility's 30-day mortality exceeds the national average — compounding exposure when paired with a positive M_suppress tier.
Captured vs Hidden Mortality
Col A (reported floor) vs selected tier — deaths not captured in institutional reporting
—
Four-Tier Comparison · Annual Attributable Deaths
OIG correction expands the observation window; suppression tier scales by ownership governance. Active column highlighted.
Col A · Reported
Col B · OIG 2025
Col C · OIG 2012
Col D · Suppression
Attributable Deaths by Tier · Annual Facility Estimate
Metric
Col A
Col B
Col C
Col D
Deaths / Vacancy
—
—
—
—
Annual Facility Deaths
—
—
—
—
National (534,047 dep.)
—
—
—
—
Calculation Detail
Active tier: —
Annual Departures (V)—
Vacancy Patient-Days (V×D×PPD)—
Baseline Mortality (B_mort)—
OIG Correction Applied—
OR_excess (= 0.07)0.07
M_suppress (Col D only)—
Deaths / Vacancy (output)—
National Scale
534,047 annual RN departures (HRSA 2028 × 17.6%). Col B is primary policy figure.
Col A
—
Reported floor
Col B
—
OIG 2025 primary
Col C
—
OIG 2012 high
Col D
—
OIG 2025–2012 × M_suppress
Master Formula
MAM = V × D × PPD × B_mort × OR_excess [÷ captureRate] [× M_suppress]
Col A · Reported: V × D × PPD × B_mort × 0.07
Col B · OIG 2025: V × D × PPD × (B_mort ÷ 0.35) × 0.07
Col C · OIG 2012: V × D × PPD × (B_mort ÷ 0.14) × 0.07
Col D · Suppression: low = Col B × M_suppress · high = Col C × M_suppress [THEORETICAL]
OR_excess = OR_mort − 1 = 1.07 − 1 = 0.07. OIG fix is denominator correction; OR is causal mechanism — separate variables, separate columns.
Col A · Reported: V × D × PPD × B_mort × 0.07
Col B · OIG 2025: V × D × PPD × (B_mort ÷ 0.35) × 0.07
Col C · OIG 2012: V × D × PPD × (B_mort ÷ 0.14) × 0.07
Col D · Suppression: low = Col B × M_suppress · high = Col C × M_suppress [THEORETICAL]
OR_excess = OR_mort − 1 = 1.07 − 1 = 0.07. OIG fix is denominator correction; OR is causal mechanism — separate variables, separate columns.
- Annual Departures (V) = RN FTE × (turnover ÷ 100). NSI 2026 national average: 17.6%.
- Vacancy Patient-Days = V × D × PPD. 78-day vacancy, 0.5 patients absorbed per vacancy per day.
- Baseline 30-day mortality (B_mort). National default 2.9% from CMS Hospital Compare [BENCHMARK — VERIFY current year]. Option 3 dual-output: Column A–D each computed twice, once with B_mort_nat and once with B_mort_fac. Delta = B_mort_fac − B_mort_nat.
- OR_excess = 0.07. Aiken 2002 (JAMA; OR 1.07, 95% CI 1.03–1.12; N=10,184 nurses; 232 hospitals). Replicated Aiken 2014 (Lancet; OR 1.07; N=422,730 patients; 9 countries). OR_excess = OR_mort − 1 = the 7% excess mortality above baseline per additional patient per nurse. [VERIFIED]
- OIG capture-rate correction — Cols B and C only. OIG (2025) confirmed 35% capture rate [VERIFIED]. OIG (2012) confirmed 14% historical floor [VERIFIED]. Applied to B_mort: B_mort ÷ captureRate. Not to OR_excess — these are independent constructs.
- M_suppress (Col D only). Governance suppression coefficient by ownership tier. NP/Public/Academic 1.0× [BENCHMARK]. For-profit chain 2.5× · PE-owned 4.3× [THEORETICAL — CMDS Stages 5–6; directional support Kannan 2023]. Col D displayed as a low (÷0.35 × M) to high (÷0.14 × M) band. Must be labeled [THEORETICAL] in all outputs.
- Output measure: attributable deaths (not dollars). Per-vacancy, annual facility, and national scale.
Methodological Framing · Mandatory Disclosure
The MAM produces population-level statistical attribution estimates. It does not identify specific patients, determine individual causation, or constitute a finding of negligence or liability. Population-level attributable risk estimation is the standard methodology used in environmental epidemiology, tobacco litigation, pharmaceutical adverse event reporting, and public health burden-of-disease calculations. Consistent with Bradford Hill criteria. Before use in litigation support or regulatory enforcement, counsel review is required.
Primary Evidence Anchors
| Study | Design | Finding | MAM Application |
|---|---|---|---|
| Aiken 2002JAMA 288:1987 · doi:10.1001/jama.288.16.1987 | Cross-sectional; N=10,184 nurses; 232 hospitals; Pennsylvania | OR 1.07 per pt/nurse (95% CI 1.03–1.12) | PRIMARY anchor. OR_mort = 1.07; OR_excess = 0.07. 30-day surgical mortality outcome. |
| Aiken 2014Lancet 383:1824 · doi:10.1016/S0140-6736(13)62631-8 | Retrospective; N=422,730 patients; 300 hospitals; 9 EU countries | OR 1.07 per pt/nurse (95% CI 1.03–1.12) | Independent replication at identical point estimate. High-confidence signal. |
| OIG 2025 · OEI-06-18-00401July 2025 | Random sample; Medicare records review | 35% capture rate | captureRate_B. Col B low-side denominator (÷0.35). |
| OIG 2012 · OEI-06-09-00091January 2012 | Medicare record review; independent methodology | 14% capture rate | captureRate_C. Col C high-side denominator (÷0.14). |
Confidence Interval Sensitivity Range
Aiken OR 1.07 carries 95% CI 1.03–1.12. Primary outputs use the point estimate. This tab shows the full CI range applied to Col B (OIG 2025) per vacancy, at national baseline B_mort. [VERIFIED — CI bounds]
Lower Bound · OR 1.03
1.03
OR_excess = 0.03
Col B / vacancy: —
Annual (— dep.): —
National (534,047): —
Primary · OR 1.07 VERIFIED
1.07
OR_excess = 0.07 · point estimate
Col B / vacancy: —
Annual (— dep.): —
National (534,047): —
Upper Bound · OR 1.12
1.12
OR_excess = 0.12
Col B / vacancy: —
Annual (— dep.): —
National (534,047): —
Research Gap · Vacancy-Period Hazard Accumulation
Griffiths 2019 (BMJ Qual Saf; PMC6716358) established aHR 1.03 per day below ward mean staffing [VERIFIED]. Compounded across 78 days: 1.03^78 = 10.0×. This compound application to a continuous multi-week vacancy is an author extrapolation, not a Griffiths finding, and is excluded from all MAM outputs. The MAM uses Aiken OR 1.07 as a conservative single-shift constant. No published study has measured how 30-day mortality risk accumulates across a continuous multi-week RN vacancy period. The upper bound of vacancy-period mortality is empirically unknown.
Full Sensitivity Table · Col B (OIG 2025) at National Baseline · Annual Facility
| OR Bound | OR_excess | Deaths / Vacancy | Annual (facility V) | National (534,047) | Status |
|---|---|---|---|---|---|
| Lower · OR 1.03 | 0.03 | — | — | — | VERIFIED · CI lower |
| Primary · OR 1.07Point estimate — all primary MAM outputs | 0.07 | — | — | — | VERIFIED · primary |
| Upper · OR 1.12 | 0.12 | — | — | — | VERIFIED · CI upper |
M_suppress by Ownership Type · Column D only
| Ownership Type | Est. Reporting Rate | M_suppress | Status |
|---|---|---|---|
| Private equity ownedContractual NDA; cost extraction governance | 10–14% | 4.3× | THEORETICAL |
| For-profit corporate chainMargin pressure; risk-management protocols | 14–18% | 2.5× | THEORETICAL |
| Nonprofit, non-academicReputational protection; TJC risk mgmt | 16–20% | 1.0× | BENCHMARK |
| Nonprofit academic / publicResearch mission; stronger institutional oversight | 20–28% | 1.0× | BENCHMARK |
| Safety net / public / CAHRegulatory scrutiny; union presence | 20–28% | 1.0× | BENCHMARK |
M_suppress values are model-constructed coefficients derived from the Clinical Moral Disengagement Scaffolding (CMDS) framework (Torrez 2026; doi:10.5281/zenodo.18985075), anchored to OIG OEI-06-09-00091.
Directionally supported by Kannan 2023 (PMC10751598: +25.4% HAC, +27.3% falls, +37.7% CLABSI post-PE acquisition vs matched controls).
Kannan 2023 is VERIFIED; M_suppress values are THEORETICAL — not independently validated as vacancy-period mortality multipliers.
Column D outputs must be labeled [THEORETICAL] in all formal presentations and must not be cited in legal or regulatory submissions without expert review and appropriate disclosure.
Primary Sources · Evidence Disclosure
| Aiken LH et al. JAMA. 2002;288(16):1987-1993.doi:10.1001/jama.288.16.1987 · OR_mort 1.07 (95% CI 1.03–1.12) · PRIMARY anchor | VERIFIED |
| Aiken LH et al. Lancet. 2014;383(9931):1824-1830.doi:10.1016/S0140-6736(13)62631-8 · OR 1.07 replicated; 422,730 patients; 9 EU countries | VERIFIED |
| OIG OEI-06-18-00401. Hospitals Did Not Capture Half of Patient Harm Events. July 2025.captureRate_B = 0.35 · Col B denominator | VERIFIED |
| OIG OEI-06-09-00091. Hospital Incident Reporting Systems Do Not Capture Most Patient Harm. January 2012.captureRate_C = 0.14 · Col C denominator | VERIFIED |
| NSI Nursing Solutions. 2026 National Health Care Retention and RN Staffing Report.vacancyDuration 78 days · turnoverRate 17.6% · 527 hospitals; 40 states | BENCHMARKCOI: for-profit staffing vendor; not peer-reviewed |
| HRSA. Nurse Workforce Projections, 2023–2038. December 2025.2028 projected 3,034,360 RN FTE → 534,047 annual departures (× 17.6%) | BENCHMARK |
| CMS Hospital Compare. 30-Day Mortality Rates. CMS.gov.B_mort_nat source · all-condition 30-day average · current year figure requires verification before production deployment | BENCHMARK · VERIFY |
| Kannan S et al. JAMA Health Forum. 2023;4(1):e221578. PMC10751598.+25.4% HAC; +27.3% falls; +37.7% CLABSI post-PE acquisition — directional support for M_suppress PE tier | VERIFIEDNot a direct measurement of M_suppress |
| Griffiths P et al. BMJ Qual Saf. 2019;28(8):609-617. PMC6716358.aHR 1.03/day below ward mean staffing · referenced for research gap only · NOT used in primary MAM output | VERIFIEDCompound application (1.03^78) is author extrapolation — excluded from MAM |
| BLS Occupational Employment Statistics. May 2024.3,391,000 employed RNs nationally | VERIFIED |
| Torrez J. Clinical Moral Disengagement Scaffolding (CMDS). Zenodo. 2026. doi:10.5281/zenodo.18985075.M_suppress value architecture · Column D only | THEORETICALAuthor self-published preprint; not independently validated |
Evidence Disclosure · MAM v1.0
Four-tier architecture per MAM Source of Truth Specification v1.0, June 2026. Mortality OR: Aiken 2002 JAMA / Aiken 2014 Lancet [VERIFIED] OR 1.07 (95% CI 1.03–1.12). OR_excess = 0.07 (= OR_mort − 1). Capture rates: OIG 2025 (35%) / OIG 2012 (14%) [VERIFIED]. B_mort_nat ~2.9% requires verification against current CMS Hospital Compare before production deployment [BENCHMARK — VERIFY]. Col D applies M_suppress as a [THEORETICAL] governance band — low = ÷0.35 × M, high = ÷0.14 × M. Must be labeled [THEORETICAL] in all outputs.
The MAM does not identify specific patients, determine individual causation, or constitute a finding of negligence. For litigation, regulatory, or government recoupment use, counsel review required. PHC Calculator companion at prismqd.github.io/PHC-Calculator. Not legal or medical advice.
© 2026 Jennifer Torrez, BSN, RN / PRISMqd. CC-BY 4.0. Cite as: Torrez J. (2026). Mortality Attribution Module v1.0. PRISMqd Research Division.