Carboplatin shortage policy met uneven execution across a seven-site cancer network
A formal allocation policy can look rigorous on paper and still fracture at the bedside. During a severe carboplatin shortage across a seven-site health system in Massachusetts and New Hampshire, clinicians had to decide which patients would receive the drug and which would be switched to an alternative regimen. The policy used predetermined priority categories built around treatment intent, the availability of inferior or contraindicated substitutes, clinical trial participation, and whether therapy was ongoing or newly starting. In practice, those rules were applied by clinical groups and disease-specific pharmacists after a training session, then compared with an independent post hoc review of medical records.
The study captures a familiar pressure point in oncology: when supply disappears faster than consensus can form, allocation becomes a real-time test of institutional discipline. In this cohort, decisions were made for 483 patients, with a median age of 66 years and 73% women. Of those patients, 222 received carboplatin and 261 were switched away from it. The numbers alone show how abruptly a shortage can redraw treatment plans for a large cancer population.
One sentence stands out for its clinical bluntness: 26% of allocation decisions were not aligned with policy intent.
Documentation gaps drove much of the mismatch between policy and practice
The independent review agreed with 56% of priority category assignments, or 189 of 335. Agreement was better for the final allocation decision itself, where review concordance reached 74% of 483 decisions, but even that left a substantial minority of cases vulnerable to inconsistency. The two most common reasons for disagreement were incomplete documentation, which accounted for 51% of the 146 discordant cases, and discrepancies in treatment contraindication assessment, which accounted for 21%.
Those failures are not administrative trivia. Allocation policies depend on explicit clinical facts: why carboplatin was preferred, why a substitute drug was inferior, whether a patient was on an active regimen or beginning treatment, and whether trial participation created a different standard of urgency. If the medical record does not preserve those details clearly enough for another reviewer to reconstruct the decision, the policy becomes difficult to audit and easier to bend unintentionally. Shortage management is supposed to distribute harm fairly; missing documentation makes fairness harder to prove and, in some cases, harder to achieve.
The study also shows how a shortage policy can drift when local context changes the interpretation of a rule. A patient’s priority category may appear straightforward in theory, but the assignment becomes more fragile when contraindications are not recorded consistently, or when a clinician at one campus applies the policy differently from a colleague at another site. In a network with main and regional campuses, consistency is a governance issue as much as a clinical one.
Site and sex differences emerged in the allocation model
Multivariable analysis identified two factors associated with actual allocation. Male gender, compared with female gender, was associated with higher odds of receiving carboplatin, with an odds ratio of 2.72 and a 95% confidence interval of 1.26 to 5.99. Treatment at the main campus, compared with regional campuses, was also associated with higher odds of allocation, with an odds ratio of 2.13 and a 95% confidence interval of 1.30 to 3.45. The results point to an allocation process that was not purely determined by the policy categories themselves.
Those findings invite a practical clinical question about how shortage rules are operationalized. Gender imbalance in allocation during a rationing event may reflect differences in disease mix, treatment pathways, or documentation quality, but the analysis still signals that patient characteristics and site of care were associated with access. In oncology, supply constraints rarely distribute themselves evenly; they can amplify existing differences in referral patterns, staffing, and review processes. A shortage policy that depends on local interpretation will tend to reproduce those differences unless the institution actively monitors them.
The main campus effect is equally consequential. When one site receives more consistent access than another, the policy may be functioning less as a single network-wide standard and more as a set of locally adapted practices. For a drug such as carboplatin, which plays a central role in multiple malignancies and treatment combinations, uneven application across sites can alter whether patients remain on intended therapy or are forced into substitutions that may be less effective, less tolerated, or simply less familiar to the treating team.
What the shortage revealed about institutional readiness
The study was conducted after the shortage, between September and November 2025, by independently reassessing records against the policy criteria. That retrospective design makes the findings especially useful for quality improvement because it exposes where the policy broke down rather than where the team believed it was working. The central lesson is not that allocation policies are futile. It is that policy performance depends on the quality of the underlying record, the clarity of contraindication assessment, and the extent to which a multicampus health system can enforce the same rules in the same way.
For oncology programs, the implications are immediate. A formal priority system can only be as reliable as the documentation that feeds it. If treatment intent, trial participation, substitute drug limitations, and ongoing versus new therapy are not captured in a standardized way, review after the fact will uncover drift that may never have been visible in real time. During a severe shortage, that drift can determine who receives the scarce drug and who does not.
The study does not describe patient outcomes after carboplatin was withheld, and it does not quantify whether the alternative regimens preserved efficacy. What it does show is narrower and more operationally important: even with predetermined criteria and trained reviewers, nearly one in four allocation decisions departed from policy intent, and the discordance concentrated in documentation failures and contraindication disagreements. For institutions building shortage protocols, the next step is not simply writing stricter rules. It is building a record system and review process that can actually carry those rules through a crisis.