Early Recognition and Response to Increases in Surgical Site Infections Using Optimized Statistical Process Control Charts: The Early 2RIS Study

Duke University

Status: Completed

Conditions: Surgical Site Infection; Infection Control

Treatments: Other: Intervention Cluster

Study type: Interventional

Funder types: Other; Other U.S. Federal agency

Identifiers

NCT03075813

Pro00081013

Details and patient eligibility

About

The purpose of this quality improvement study is to measure the effectiveness of surveillance using optimized statistical process control (SPC) methods and feedback on rates of surgical site infection (SSI) compared to traditional surveillance and feedback.

The primary objective is to determine if hospital clusters randomized to receive feedback from optimized SPC surveillance methods collectively have lower rates of SSI compared to hospital clusters randomized to receiving feedback from traditional surveillance methods. Secondary objectives are 1) to estimate and compare the number of signals identified using optimized SPC methods and traditional surveillance methods; 2) to estimate and compare the time and effort required to investigate signals generated using optimized SPC methods and traditional surveillance methods; and 3) to estimate the number and proportion of false-positive signals identified using optimized SPC methods and traditional surveillance methods.

The Early 2RIS study will be a prospective, multicenter cluster randomized controlled trial using stepped wedge design. The active component of the quality improvement study will be performed in 29 DICON hospitals over three years, from March 2017 through February 2020. Clusters randomized to intervention will receive feedback on increasing rates of SSI identified through optimized SPC methods. This intervention is expected to decrease the subsequent rate of SSIs by closing the feedback loop on SSI outcomes.

Participating study hospitals will all be members of DICON, a network of 43 community hospitals in North Carolina, South Carolina, Georgia, Florida, and Virginia that provides community hospitals access to consultative services from infection prevention experts, data analyses and benchmarking, and educational materials designed by faculty from Duke. This study is considered part of routine quality improvement measures and a part of previously established agreements between DICON and the community hospitals. Data flow and communication are outlined in detail in approved protocols determined to be exempt research by the DUHS IRB. Briefly, existing clinical data are extracted from participating hospitals' electronic medical record into discrete files according to DICON specifications. Then a de-identification process removes direct patient identifiers into a limited dataset.

The majority of data collection will occur through methods already developed and utilized by study hospitals. In brief, each hospital routinely submits limited datasets to the DICON Surgical Surveillance Database, including the following variables: hospital, type of procedure, patient identifier, date of procedure, age, sex, surgeon identifier, start/stop times, ASA score, wound class, risk index, SSI (Yes/No), date of infection, type of SSI, location at diagnosis and organism. No identifiable patient or surgeon data are transmitted to the DICON Surgical Database. Data definitions and data collection methods are standardized across DICON hospitals. Following signal adjudication, additional data will be collected in a REDCap database to document actions and rationale.

Enrollment: 29 patients

Sex: All

Volunteers: No Healthy Volunteers

Inclusion criteria

• All patients who undergo one of 13 targeted procedures in 29 DICON study hospitals will be eligible for inclusion. The 13 targeted procedures include coronary artery bypass graft, cardiac valve replacement, colon surgery, herniorrhapy, knee arthroplasty, hip arthroplasty, Cesarean section, abdominal hysterectomy, vaginal hysterectomy, spinal fusion, laminectomy, carotid endarterectomy, and peripheral venous bypass. These procedures were selected because they are frequently performed in community hospitals and/or are associated with particularly adverse outcomes if complicated by SSI. Eligible procedures will be categorized by procedure type at each hospital using ICD9 codes published by the NHSN, per routine DICON activities. Six clusters were constructed from these procedures to ensure that surgeons who perform similar types of procedures were grouped together to limit potential bias. These clusters are labeled as Spine, GI, OB/GYN, Ortho, Vascular, and Cardiac. These clusters are the units for randomization and analysis.

Exclusion criteria

• DICON hospitals that did not submit a letter of support for participating in the study will be excluded. Patients not undergoing one of these 13 procedure types at the 29 study hospitals will be excluded from the analysis.

Trial design

Primary purpose: Prevention

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: Single Blind

---

29 participants in 2 patient groups

Intervention Cluster

Experimental group

Description:

Surgical surveillance data submitted to the DICON Surgical Database will undergo immediate analysis by optimized SPC methods. If a signal is generated, study personnel in DICON will be notified to adjudicate the signal and determine if further action is required. Optimized SPC methods include the application of two SPC charts. The investigator determined that when either chart identifies a signal, the study will have approximately 90% sensitivity and 65% specificity to identify important increases in rates of SSI.

Treatment:

Other: Intervention Cluster

Control Cluster

No Intervention group

Description:

Local personnel in clusters randomized to traditional surveillance and feedback will receive bar graph reports and data interpretation per routine DICON surveillance. These reports will be provided every 6 months.