In 2013 we were winding down the analysis needed to create the Wound Healing Index (WHI), a mathematical model to stratify wounds by their likelihood of healing. The goal was to enable the reporting of honest healing rates. Susan Horn and I have published a lot about the WHI. The variables are for the most part, just what any wound care doctor might guess (e.g. wound size, wound age, total number of wounds, the type of tissue exposed, whether there is evidence of infection, whether hospitalization was required, and others that are more patient specific).
There’s a lot of data that wounds with evidence of infection are less likely to heal, so we had developed a query to capture the concept of high bioburden and/or infection from the US Wound Registry’s structured data. Wounds with purulent drainage, peri-wound erythema, peri-wound warmth, a foul odor, or that had undergone a culture were all less likely to heal based on our analysis. And then I got the bright idea to use “being put on antibiotics” as a surrogate for “had infection” as we began to evaluate variables for the predictive models. And rather than predicting that the wounds would fail, giving a patient with a non-healing wound antibiotics made the wound more likely to heal. That is one of our dirty little secrets.
But, let me back up. Although most of the evidence suggests that infection and bio-burden have a negative effect on wound healing, most randomized, controlled trials (RCTs) in wound care exclude patients with active infection or current antibiotic use, so it isn’t possible to use RCT data to settle the question of whether antibiotics are valuable in the real world.
The U.S. Wound Registry (USWR) is a repository of linked, de-identified electronic health records (EHRs). Currently 130 facilities in 34 states contribute data to the USWR, a 501(c)(3) non-profit organization. Computers are present in every room; point of care documentation is performed by the nurse and the physician; medications prescribed by electronic prescribing (eRx), the EHR internally audits the chart to calculate physician and facility reimbursement; quality data is transmitted to the USWR for use in the Merit Based Incentive Payment System (MIPS) reporting; and the USWR archives the data for research.
To create the WHI, we analyzed data from 256,671 wounds of which 106,272 wounds met our criteria (41.4% of total) for 9 wound types: Amputations, Arterial ulcers (chronic wounds related to atherosclerosis), Burns, Diabetic Foot Ulcers (chronic ulcers related to diabetes), Pressure Ulcers, Surgical wounds, Traumatic wounds (wounds related to trauma), Failing Flaps, and Venous ulcers. Factors known to be associated with decreased healing were evaluated and logistic regression models were created based on variables that were significant (p<0.05). A different model was predictive of healing for each wound type. Seven models were well-validated when applied to a hold-out sample of data.
It turns out that about half of all patients treated for chronic wounds and ulcers at hospital-based outpatient wound centers are treated with systemic antibiotics (43% to 54% of all patients), many times with more than one course of antibiotics. Using bivariate analysis, when we evaluated 106,272 wounds, evidence of infection/bioburden was associated with a decreased likelihood of healing in all wound types (p < 0.001). However, the administration of systemic antibiotics was consistently associated with an increased likelihood of healing in all wound types (p<0.01). I doubt I could get that dirty secret published.