A public record of what the calculator measures, what remains historical, which software defects were corrected, and which future changes require scientific review rather than routine code maintenance.
Evidence review: 2026-08-08.
The production build preserves the legacy GlobalRPh opioid calculation profile and the historical non-opioid denominators while correcting identifiable software defects and modernizing the interpretation language.
The application is structured for formal Technical Expert Panel review. A final public roster should be posted here when participation is confirmed. Until then, the calculator should not state or imply endorsement by named investigators or institutions.
| Version | Date | Status | Material changes |
|---|---|---|---|
| Legacy Perl / HTML | Historical | Original GlobalRPh implementation | Historical SDD denominators, medication list, opioid conversion logic, and original explanatory wording. |
| 2.0.0 | August 8, 2026 | PHP modernization / bug-fix release | Server-side PHP rebuild, isolated data/config layer, responsive interface, legacy-comparable calculation profile, eszopiclone summation correction, modern interpretation language, and separate research sensitivity profile. |
| 2.1.0 | August 8, 2026 | Prior TEP-ready build | Complete 62-drug SDD reference tables, expanded Beers/STEADI/current evidence background, public methodology/TEP page, explicit legacy limitations, and stronger scientific-governance language. |
| 2.2.0 | August 8, 2026 | Current public build; TEP-ready | Prominent standalone AGS Beers section navigation, expanded drug-specific Beers context, detailed legacy-versus-current evidence tables, explicit opioid coefficient comparison, and an expanded TEP scientific decision matrix. |
| Future TEP-reviewed release | Pending | Scientific review required | Any approved denominator, drug-universe, opioid-coefficient, threshold, interaction, or outcome-model revision should receive a new method version and validation record. |
Recommended TEP record: when formal review occurs, publish the review date, panel roster and affiliations (with permission), conflict-of-interest disclosures, evidence-search cutoff, approved method changes, validation status, and the exact calculator release reviewed.
The SDD framework standardizes each indexed CNS medication to a historical geriatric dose denominator and sums the individual standardized exposures. Opioids are first expressed as morphine-equivalent exposure and then standardized against 10 mg/day of oral morphine-equivalent exposure.
The key methodological principle is that these denominators function as index constants. They should not be silently replaced by a modern package-insert dose, a new guideline dose, or a clinician's preferred starting dose because doing so would change the numerical index and potentially invalidate the published 3-SDD threshold.
| Issue | Classification | Recommended handling |
|---|---|---|
| Eszopiclone omitted from legacy subtotal/final total | Software defect | Corrected in v2.x. The form already accepted and calculated the drug; omission from summation was not a scientific choice. |
| Fentanyl display/calculation inconsistency | Legacy ambiguity | Expose the coefficient actually used. Preserve historical computation until the TEP approves a method revision. |
| Replace legacy opioid factors with CDC 2022 factors | Method change | Do not silently substitute. Evaluate as a sensitivity analysis, revalidate thresholds, and publish a new method version if adopted. |
| Add a modern CNS-active medication with no historical denominator | Method change | Count it for Beers-style CNS polypharmacy when appropriate, but do not invent an SDD. TEP should define a reproducible derivation/validation process. |
| Update explanatory fall-risk wording | Evidence/communication update | Appropriate without changing the numeric algorithm as long as published associations are represented accurately. |
This matrix is intentionally more detailed than the public calculator summary. It separates changes that improve software correctness or communication from changes that redefine the scientific index and therefore require expert review, versioning, and potentially new validation.
| Topic | Legacy state | Current evidence / standard | Current v2.2 handling | TEP question |
|---|---|---|---|---|
| Purpose of SDD | Dose-standardized summation used to identify higher CNS medication burden; page language linked ≥3 directly to high fall risk. | The threshold is supported by observational nursing-home studies, not by an individual absolute-risk prediction model. | Use “published research threshold” and “medication-review signal.” | Should the next-generation tool remain a burden index, or be developed/validated as an outcome-prediction model? |
| Historical denominators | Drug dose divided by a “minimum effective geriatric daily dose.” | Contemporary prescribing doses, FDA labels, renal adjustments, and Beers recommendations do not provide a one-to-one replacement denominator for this research index. | Preserved and relabeled as historical index constants. | Define a reproducible denominator source hierarchy before any denominator is changed or a new drug is added. |
| Opioid coefficients | Codeine 0.15; fentanyl TD actual Perl 4.1667 but display 7.2; hydrocodone 1.3; hydromorphone 5; methadone 4; morphine 1; oxycodone 1.5; tramadol 0.2. | CDC 2022 lists 0.15, 2.4, 1.0, 5.0, 4.7, 1.0, 1.5, and 0.2 respectively. | Legacy actual computational profile remains default; CDC profile is sensitivity-only. | If modern factors are adopted, what dataset and endpoint will be used to recalibrate or revalidate the ≥3 threshold? |
| Fentanyl inconsistency | Code and user-facing factor disagree. | CDC 2022 provides a third value, 2.4. | Disclose all three facts; reproduce actual code behavior for legacy comparability. | Which factor belongs in a future method and how should historical results be mapped? |
| Eszopiclone subtotal | Individual SDD calculated but omitted from class and final totals. | No scientific rationale supports the omission. | Bug corrected. | No method decision required unless strict reproduction of old erroneous outputs is needed for a research archive. |
| Drug coverage | Fixed 62-drug index. | Beers CNS-active polypharmacy spans broader contemporary classes and drugs, including skeletal muscle relaxants and agents without historical SDD denominators. | Additional CNS drugs may contribute to the medication-count screen only. | Which new drugs/classes merit a validated SDD denominator, and should historical and expanded modes coexist? |
| Beers drug-specific PIMs | No separate PIM logic. | Drug-specific recommendations can apply regardless of SDD, including strong anticholinergic antidepressants, benzodiazepines/Z-drugs, and antipsychotic use in dementia/delirium. | Educational/current-guidance layer remains parallel to SDD. | Should future versions generate structured drug-specific alerts, and if so, which alerts are in scope? |
| Beers ≥3 CNS rule | No medication-count criterion. | Beers recommends avoiding concurrent use of ≥3 CNS-active drugs from specified classes when possible because of falls/fracture risk. | Separate count flag; does not change SDD. | Keep as parallel flag or build a multidimensional display combining dose burden and medication count? |
| Drug-drug interactions | Not modeled. | Important Beers interactions include opioid + benzodiazepine and opioid + gabapentin/pregabalin. | Educational warning only. | Should interaction flags become automated structured outputs, and how should exceptions be represented? |
| Renal function | No renal input. | Beers includes renal dosing/avoidance guidance for multiple CNS-active medications. | Do not change SDD automatically; prompt independent renal review. | Would a future version benefit from CrCl/eGFR input and drug-specific renal alerts without altering the SDD numerator? |
| PRN exposure | Daily dose entry did not fully standardize how PRN use should be represented. | Reproducible research use requires a prespecified exposure window and source. | User guidance favors actual administered exposure when known, but method remains explicitly reviewable. | Choose ordered maximum, prior-24-hour administered dose, rolling mean, or another protocol-defined exposure. |
| Deprescribing claims | Suggested reducing/stopping medications when SDD >3. | Deprescribing can be appropriate, but randomized/systematic evidence does not support promising that lowering SDD will itself prevent falls; abrupt withdrawal can be hazardous. | Recommend individualized review/taper planning, not automatic discontinuation. | Define allowable action language and whether the tool should link to drug-class-specific deprescribing resources. |
Governance principle: changing wording, fixing a coding defect, or adding an educational safety notice does not necessarily change the index. Changing a denominator, included drug, opioid coefficient, exposure definition, weighting rule, or threshold does.
Geriatric medicine and long-term care should anchor the panel. Additional expertise should include clinical pharmacy/geriatric pharmacotherapy, pharmacoepidemiology and biostatistics, falls research, pain/opioid pharmacology, psychiatry or geriatric psychiatry, implementation science, nursing/long-term-care workflow, and software/data validation.
The University of Pittsburgh currently lists David A. Nace, MD, MPH as Professor of Medicine, Chief of Geriatric Medicine, Thomas P. Detre Chair, Director of Long Term Care, and Chief Medical Officer, UPMC Senior Services. His public academic profile describes work in long-term care, adverse drug events, patient safety, dissemination/implementation, and quality improvement. Those areas are directly relevant to expert oversight of a nursing-home medication-burden tool.
Suggested format: “Scientific methodology and future index revisions are reviewed through a multidisciplinary Technical Expert Panel. Panel membership, affiliations, conflicts of interest, evidence-review dates, and version history are available on this methodology page.”
Do not use language such as “validated by” or “endorsed by” an institution unless the validation or endorsement has actually occurred and the institution has authorized that wording.
