A Framework for De-implementation in Surgery

Despite spending nearly twice as much on health care as other high-income countries, the United States falls behind in multiple metrics of population-health outcomes.1 As a result, there is an urgent need to prioritize value in medicine. A key component to this equation is the elimination of low-value services. Currently, the United States spends >$100 billion annually on overtreatment through unnecessary medical tests, treatments, and procedures that expose patients to harms without improving survival.2 Recognizing this, a number of surgical societies have partnered with the national Choosing Wisely campaign to generate >110 recommendations to avoid low-value care in surgery, including 24 surgical procedures.3 Although these recommendations are useful in identifying low-value surgical and perioperative care, they do not provide guidance on how providers or organizations can achieve de-implementation. De-implementation is the science of identifying factors that facilitate persistent use of low-value services to develop evidence-based strategies to eliminate these practices. To date, most efforts to reduce overtreatment have focused on unnecessary diagnostic testing or medications which are more easily addressed with interventions such as ordering restrictions, market withdrawals, or changes in insurance coverage. De-implementation of low-value surgical procedures may require unique considerations due to direct financial incentives to do more irrespective of value, a need to maintain surgeon discretion in the operating room, and more entrenched practices before efficacy data are available.4 Although some low-value procedures are naturally de-implemented (eg, laparoscopic adjustable gastric band surgery after introduction of sleeve gastrectomy), others continue to be performed at high rates (eg, screening colonoscopy for patients with limited life expectancies). This underscores the importance of developing effective strategies to achieve de-implementation to reduce cost and harms from low-value services. Most successful attempts at de-implementation require multifaceted and multilevel strategies.5,6 Multifaceted strategies target de-implementation through a combination of interventions and are likely more effective than single component strategies. For example, a potential provider-facing multifaceted strategy to reduce unindicated antibiotic prescribing might be to educate clinicians on national recommendations and audit and provide feedback on individual provider performance. Multilevel strategies recognize that complex health behaviors are driven by factors at guideline, patient, provider, and organizational levels and that targeting 1 level alone is usually insufficient. For example, a multilevel intervention to decrease overtreatment of low-risk prostate cancer might target patients by addressing individual preferences for more aggressive treatment, target clinicians by providing tools for accurate risk communication, and target organizations through alternative payment models. One useful model for understanding differences in natural de-implementation is early-stage breast cancer. Early-stage breast cancer is common, has an evidence base that relies strongly on evidence from meta-analyses and randomized controlled trials, involves multidisciplinary providers, and has decentralized treatment largely away from academic centers to community practices. Currently, major breast surgical oncology organizations through the Choosing Wisely campaign have recommended against the routine use of 4 breast cancer procedures. Of these, the most successful example of de-implementation has been the rapid decline in national rates of completion axillary lymph node dissection (ALND) after the American College of Surgeons Oncology Group (ACOSOG) Z0011 trial demonstrated that ALND does not improve survival or recurrence rates in patients with limited nodal disease receiving post-lumpectomy radiotherapy. In this case, de-implementation occurred naturally after trial publication without the need for specific strategies to target unnecessary procedures. In contrast, the Society of Surgical Oncology recommends against the routine use of sentinel lymph node biopsy (SLNB) in women ≥70 years’ old with early stage, clinically node negative hormone receptor positive (HR+) breast cancer. This guideline is based on retrospective and prospective data demonstrating that early-stage, estrogen receptor positive tumors in older women carry an excellent long-term prognosis and that omission of axillary staging does not impact overall survival.7 Despite this, national data suggest that >80% of women ≥70 years’ old with HR+ invasive breast cancer continue to receive SLNB.8 A key component to success when attempting to eliminate entrenched practice patterns is pragmatic application of implementation-science frameworks.9,10 One novel framework for reducing low-value care is the Choosing Wisely De-implementation Framework.11 This 5-phase strategy largely focuses on provider behavioral change. In phase 0, potential areas of low-value health care are identified, and in phase 1 local priorities for de-implementation of Choosing Wisely recommendations are determined, in phase 2 barriers to de-implementation and potential interventions to overcome them are identified, in phase 3 de-implementation programs are evaluated, and in phase 4 de-implementation programs are promoted and effectively spread. The authors describe their strategy in a “how-to” framework that can be broadly applied to a variety of health care settings. Here, we present a real-world example of a provider-facing de-implementation strategy to illustrate how a multifaceted strategy based on behavioral determinants can work to reduce low-value surgery Table 1. TABLE 1 - Application of the Choosing Wisely De-Implementation Framework to an Institutional Pilot to Reduce Rates of Sentinel Lymph Node Biopsy in Women ≥70 Years’ Old With Hormone Receptor Positive Cancer Framework Phase De-Implementation of SLNB Phase 0Identification of potential areas of low-value healthcare Choosing Wisely identified 4 low-value breast surgery procedures:1. Axillary lymph node dissection for limited nodal disease2. Reexcision for close but negative margins3. Contralateral prophylactic mastectomy for unilateral cancer4. SLNB in women ≥70 years’ old with hormone receptor positive (HR+) cancer Phase 1Identification of local priorities for implementation of recommendations Systematic review:Demonstrated that national rates of SLNB for women ≥70 years’ old with HR+ cancer is >80% Phase 2Identifying barriers and potential interventions to implement CW recommendations Identified barriers to SLNB de-implementation:1. Lack of familiarity with the guideline2. Belief that the results of the biopsy would influence downstream treatments3. Influence of other oncology providers Phase 3Evaluation of CW implementation programs Piloted a multi-faceted de-implementation program:1. Provider education. Presented at multi-disciplinary educational sessions regarding the guideline and potential consequences of low-value SLNB2. Feedback. Provided feedback on facility and provider-specific SLNB rates3. Modeling. A surgeon champion led educational sessions and provided reminders related to the guideline for eligible patientsSLNB rates in women eligible for omission decreased from 77% to 46% Phase 4Spread of effective CW implementation programs Next steps:1. Develop multi-level interventions2. Expand to other institutions Adapted From Grimshaw et al.11 Our previous qualitative work of a national sample of breast surgeons found that specific barriers to de-implementation of SLNB in women ≥70 years’ old with clinically node-negative HR+ breast cancer in were a lack of familiarity with the guideline, the belief that the results of the biopsy would influence downstream treatments, and the influence of other oncology providers.12 We therefore performed a multifaceted strategy at our institution targeting provider-level barriers to de-implementation which consisted of education through presentations at multidisciplinary breast cancer conferences that focused on downstream harms and costs for older patients receiving SLNB, feedback on facility and surgeon-specific rates, and modeling through a surgeon champion for de-implementation. We found rates of SLNB in women ≥70 years’ old with HR+ breast cancer at our institution decreased from 78% in 2014 to 46% in 2018 with a decrease after implementation of provider education in Spring 2017 (Fig. 1). In contrast to the belief that nodal status greatly influences adjuvant chemotherapy, we found that <5% of all patients and only one-fifth of node-positive patients received chemotherapy; the vast majority of SLN-positive patients either declined, were not candidates for chemotherapy, or omitted chemotherapy after Oncotype Dx testing indicated a low to intermediate recurrence risk.FIGURE 1: Results of a multifaceted strategy targeting low-value SLNB for Women ≥70 years’ Old With Clinically Node Negative HR+, HER2− invasive breast cancer at Michigan Medicine.Overall, despite the success of our provider-facing intervention, approximately half of the candidates for omission continue to receive SLNB. These findings suggest that patient or organizational-level factors likely contribute to continued use and will need to be targeted with complimentary strategies. For example, our preliminary work suggests that 1 patient-level barrier to SLNB de-implementation is difficulty interpreting the rationale for reducing low-value procedures (in this case a favorable rather than poor prognosis). Therefore, as a next step, an intervention might address gaps in education through improved patient-facing materials or reframing of national guidelines to emphasize the good prognosis. Ultimately, however, any effort to reduce rates of low-value surgical procedures will need to contend with the prevailing fee-for-service model that encourages doing more regardless of value. Potential strategies that may work to incentivize provision of high-value services and disincentivize low-value breast cancer surgery are bundled payments, eliminating reimbursements, or instituting financial incentives for providers with low rates of unnecessary procedures. These strategies (eg, bundled payments for joint replacement surgery and eliminating reimbursements for low-value vitamin D testing) have previously been applied successfully with evidence of cost savings and can be adapted to reduce rates of other low-value surgical procedures. Taken as a whole, the de-implementation framework presented here can be broadly applied in any practice setting to reduce low-value surgery. Although the participation of surgical societies in promoting specialty-specific Choosing Wisely recommendations has been a critical first step in reducing low-value care, we know that national guidelines alone are not enough to decrease use of an unnecessary procedure. However, our experience demonstrates that targeted efforts to assess specific barriers to de-implementation can be effectively translated to a reduction in overtreatment. Through wide-spread dissemination of these strategies, we have the opportunity to eliminate the costs and harms associated with unnecessary services and sustainably improve value in the US health care system.