The goal of the study is to determine whether clinical decision support (CDS) for radiology affects the number, type, or appropriateness of targeted high-cost radiology imaging orders (i.e. magnetic resonance (MR), computed tomography (CT), nuclear medicine (NM) and Positron Emission Tomography (PET) scans). The CDS will be delivered to physicians in the Aurora Health Care system. It will be delivered in Epic, an industry-standard electronic medical record software, through ACR Select, which is a leading decision support tool based on the American College of Radiology (ACR) Appropriateness Criteria (see http://www.acr.org/Quality-Safety/Appropriateness-Criteria). The ACR Select tool rates imaging orders on a scale of 1-9 with 1-3 labelled as 'usually not appropriate', 4-6 'May be appropriate', and 7-9 'usually appropriate'.

With healthcare spending accounting for almost one-fifth of the U.S. economy and an even larger share of public sector budgets, there is substantial interest in innovations in healthcare delivery that can reduce the "over use" of resources that have no or low value to patients. As a result, there is a key need for rigorous evidence on scalable interventions aimed at improving the efficiency of the U.S. healthcare sector in general, and in the public sector in particular, which accounts for $1.25 trillion in annual healthcare spending.

In particular, there is widespread concern in both the medical profession and the public sector of the cost and health risks of "over-scanning". Estimates suggest that as many as 30% of imaging in the U.S. are unnecessary. Medicare direct spending on "high-cost" scans (e.g. MRs and CTs) was about $10 billion in 2012, or about 2% of total Medicare costs ; the indirect costs are likely considerably greater, since imaging often triggers additional follow up care. It is also estimated that about 2 percent of cancers in the U.S. are due to CT use.

Reflecting this concern, starting in 2018 Medicare will no longer reimburse for high-cost scans unless ordered using an "acceptable" Clinical Decision Support (CDS) system. Despite this upcoming policy change, the investigators know of no large-scale randomized trials on the impact of CDS for imaging.

The intervention in this study provides Clinical Decision Support (CDS) for targeted high-cost radiology orders, (MR, CT, NM, and PET scans), to healthcare providers treating patients in settings affiliated with the Aurora Health Care system headquartered in Wisconsin. CDS is a tool embedded in an order entry system that provides information and guidance to providers on whether their intended order is "appropriate" and whether there are more highly recommended alternatives. The randomization is at the provider level: half will receive the CDS, while the remaining half of providers in the study will serve as the control group.

The CDS will be delivered through the order-entry software, Epic, through ACR Select software, which is a leading decision support tool based on the American College of Radiology (ACR) Appropriateness Criteria.

Recommendations that appear in the CDS tool are a digitized version of guidelines created by the American College of Radiology (ACR). The guidelines score the appropriateness of a scan order for a given health indication, where indications include common symptoms and diagnosis keywords, such as "acute headache." In particular, indication-scan pairs are assigned an "appropriateness rating" from 1-9. Scores 1-3 are 'usually not appropriate,' 4-6 are 'may be appropriate,' and 7-9 are 'usually appropriate.'

To learn more about how CDS impacts ordering behavior, the investigators will investigate whether those exposed to CDS orders in ways that avoids CDS, such as placing them via clerical workers or choosing different indications. The investigators plan to analyze outcomes across different settings (Inpatient vs Outpatient and in the ED). Last, if the investigators find effects of CDS on ordering they will examine effects on utilization, including length of inpatient stays.

The primary outcome is the number of imaging orders that would produce a best-practice alert suggesting a change. The investigators conducted power calculations using a 7-month intervention window and 6 months of pre-period data to calculate the lag of the dependent variable as a control. With this control, the minimum detectible effect is a 13% reduction compared to a mean of 11 scans over that time period.

In addition, the investigators will investigate the number of scans (all high-cost, high-cost scoring 1-3, high-cost scoring 4-6, and substitution to low-cost scans. The investigators will also investigate whether treatment group orders in ways that avoids the CDS, such as placing them via clerical workers or choosing different indications. The investigators plan to analyze outcomes across different settings (Inpatient vs Outpatient and in the ED). Last, if the investigators find effects of CDS on ordering they will examine effects on utilization, including length of inpatient stays.

Inclusion Criteria:

• Medical provider actively employed at Aurora with a valid Epic log-in.
• Is a Medical Doctor (MD), Doctors of Osteopathic Medicine (DO), podiatrist (DPM), nurse practitioner (NP), physician assistant (PA), or certified nurse midwife (CNM)
• Has imaging order permissions at Aurora Health Care.
• Has at least one high- or low-cost imaging order in the year from November 1, 2015 to November 1, 2016 or is medical resident who joined Aurora since that time.

Exclusion Criteria:

• Opted out of the study prior to November 24, 2016