Allocation Policy Implementation During a Severe Chemotherapy Shortage

Cancer drug shortages require health systems to rapidly develop and enact allocation policies, yet little is known about how such policies perform in practice.

To evaluate clinical implementation of a formal allocation policy during the 2023 carboplatin shortage.

This cohort study comprised a carboplatin shortage at a 7-site health system in Massachusetts and New Hampshire that lasted from June 1 to July 17, 2023. Patients with cancer who were eligible for carboplatin during the shortage were assigned priority categories using predetermined criteria based on treatment intent, inferior or contraindicated alternative treatments, clinical trial participation, and ongoing vs new treatment; these categories were assigned collectively by clinical groups and disease-specific pharmacists after a training session. A structured process conducted between September and November 2025 was used to review medical records and independently reassign categories post hoc.

Allocation of carboplatin during the shortage vs mandate to switch to an alternative treatment.

Main outcomes were concordance between actual and review-based priority category assignments, actual allocation, and independent review-based allocation, as well as associations of demographic and clinical variables with allocation. Descriptive statistics and multivariable logistic regression assessed actual and review-based outcomes.

Allocation decisions were made for 483 patients (median age, 66 years [IQR, 58-73 years]; 354 women [73%]). A total of 222 patients (46%) were allocated carboplatin and 261 (54%) were not. An independent review agreed with 56% of priority category assignments (189 of 335) and 74% of allocation decisions (357 of 483); 26% of allocation decisions (126 of 483) were not aligned with policy intent. Incomplete documentation (51% [74 of 146]) and treatment contraindication discrepancies (21% [31 of 146]) were the most common causes of disagreement. In multivariable regression, male gender (vs female gender; odds ratio [OR], 2.72 [95% CI, 1.26-5.99]; P = .01) and treatment at main campus (vs regional campuses; OR, 2.13 [95% CI, 1.30-3.45]; P < .001) were associated with increased likelihood of receiving carboplatin. In models of policy-concordant allocation based on independent review, these associations were not significant (male gender: OR, 1.39 [95% CI, 0.16-13.08]; P = .77; treatment at main campus: OR, 1.47 [95% CI, 0.33-7.14]; P = .61).

In this cohort study of allocation policy implementation during a chemotherapy shortage, most allocation decisions aligned with policy intent, but key implementation gaps were identified and may have been associated with biases in allocation. Explicit documentation, contemporaneous monitoring, and enforcement mechanisms may improve policy implementation during oncology drug shortages.