Clinician documentation and practices after declined medications for patients with sickle cell disease

Abstract Introduction Documentation in the electronic medical record (EMR) frequently includes the negative descriptor “refused”. Our prior work demonstrated that patients with sickle cell disease (SCD) had higher odds of having “refused” in their EMR compared to Black patients and patients with chronic pain. It is not known whether differences in use of “refused” in the EMR is due to clinician documentation practices that vary based on patient characteristics or due to actual differences in patient behaviors. We examined patient behaviors in declining medications and clinician practices after declination. Then, we compared patient and clinician behaviors based on whether clinicians had documented “refused” in the EMR of hospitalized patients with SCD and other stigmatized patient groups. Methods This single-center study included adult hospitalizations from March 19, 2019, to October 5, 2020, in which patients received scheduled constipation medications, nonopioid analgesics, opioids, or antibiotics. We dichotomized hospitalizations by whether an administration of a medication from each of these 4 classes had been declined by the patient during their hospitalization. Clinician practices were assessed by whether a medication was discontinued after it was declined. We also dichotomized hospitalizations by whether the EMR included the negative descriptor “refused” in clinician notes. We compared patients with SCD to 4 patient groups without SCD: Black patients, patients with chronic pain, patients with opioid use disorder (OUD), and counterfactuals (non-Black patients without chronic pain or OUD). Multilevel modeling accounted for clustering of hospitalizations within patients, and all models were adjusted for sex, marital status, insurance type, and comorbidities. Results Of 9,740 patients across 14,533 hospitalizations, 213 (2%) had SCD. Patients with SCD declined constipation medications in 83% of hospitalizations, nonopioid analgesics in 59%, opioids in 20%, and antibiotics in 12%. Patients with SCD had higher odds of declining any medication than all patient groups: Black (aOR 8.11, 95% CI 6.13-10.7), chronic pain (aOR 3.73, 95% CI 2.81-4.94), OUD (aOR 4.79, 95% CI 3.26-7.05), and counterfactuals (aOR 8.31, 95% CI 5.94-11.6). Clinicians discontinued the declined medication in 36% of the 6856 hospitalizations with a declined medication. This was highest for antibiotics, as clinicians discontinued a declined antibiotic in 38% of hospitalizations, then opioids (37%), nonopioid analgesics (31%), and constipation medications (29%). Patients with SCD had lower odds of clinicians discontinuing declined constipation medications than all patient groups except for OUD; they had higher odds of clinicians discontinuing declined nonopioid analgesics than patients with OUD. Only 6% of hospitalizations with declined medications had the negative descriptor “refused” documented in the EMR in clinician notes. Patients with SCD had higher odds of “refused” in the EMR than patients without SCD when declining a medication (aOR 1.74, 95% CI 1.06-2.87). When “refused” was documented, only antibiotics had significantly higher odds of clinician discontinuation after declination (aOR 2.26, 95% CI 1.22-4.18). Conclusion While patients with SCD declined medications more frequently than other stigmatized patient populations, they also had higher odds of the word “refused” being documented in their EMR when declining a medication, and clinicians had lower odds of discontinuing a declined medication for a patient with SCD. This suggests that clinician documentation of “refused” in the EMR may be driven by patient behaviors and clinician biases. Opioids, nonopioid analgesics, and constipation medications were selected given their relevance to patients with SCD, chronic pain, and OUD to manage pain through complementary mechanisms and reduce side effects from opioids. While antibiotics were least commonly declined, clinicians had higher odds of discontinuing antibiotics when documenting “refused” in the EMR. This may reflect medicolegal concerns and defensive charting. In summary, clinicians may not consider the medication preferences of patients with SCD as much as other patient groups. Clinicians should explore why patients decline medications, adjust the treatment plan based on patient needs and medication alternatives, and document in the EMR the patient-centered care and shared decision-making that occurred, rather than merely noting “patient refused”.