The ‘Doorknob’ Test & How EHRs Could Change Wound Care’s Future

After years of practicing wound care, most clinicians can develop fairly good “doorknob” instincts — the ability to walk into an exam room, take one look at the patient’s wound, and immediately predict the likely outcome, even with their hand still on the doorknob. The best clinicians hone that instinct, even though they may not know exactly how they do it. This phenomenon is the subject of Blink, one of the fascinating books by author Malcolm Gladwell, which opens with the story of a Roman statue and the antiquities experts who recognized it was a fake with a single glance, even though they were not able to explain exactly how they knew. Of course, as clinicians we’ve all been wrong about our assessments on occasion, but competent clinicians are correct more often than wrong when that “little voice inside” sets off a warning. Most experienced wound care providers have an instinct that tells them which ulcers will heal with routine care and which will need additional help. Unfortunately, they are not sure how reliable their instincts are or how to better harness them to improve patient care.

This article will discuss how electronic health records (EHRs) can play an important role in helping clinicians to strengthen their doorknob instincts as the emphasis on risk stratification and outcomes reporting within the wound care industry increases. 

Predicting Outcomes for Better Healing Rates

Why would predicting outcomes or response to treatment be useful for wounds or wound care patients, assuming we could reliably do it? Outcome prediction is important in part because we need to report healing rates in relation to how likely (or unlikely) the patient was to heal in the first place. The Centers for Medicare & Medicaid Services (CMS) requires risk stratification for outcomes reporting. Wound care is basically the only “specialty” that does not do this. For years, the reporting method utilized by the wound care industry has been to simply not report treatment failures at all while claiming healing rates above 90%. This deceptive approach to outcomes reporting simply cannot continue. Outcomes reporting is a major component of the new Quality Payment Program (QPP), and risk stratification is required by CMS to report outcomes so that practitioners who are caring for the sickest patients will not be penalized by appearing to have worse outcomes than their peers, who may be caring for patients who are not as unhealthy. The future of reporting wound healing rates will be to do so in relation to the predicted likelihood of healing. Risk stratification (or outcomes predicting) can also help us target advanced therapeutics more effectively. We know that some (perhaps all) advanced therapeutics (ATs) work better if used sooner rather than later. However, we don’t want to provide ATs to all patients because if we did, we would waste some ATs on wounds that could heal without them. Unfortunately, there are wounds so severe that a given AT might not be of benefit because they are unable to respond to the treatment. So, if possible, in addition to targeting ATs for the wounds that need them, we would like to identify futile situations so that we do not waste time and resources on treatments that we know are doomed to fail. Individual clinicians may actually already be pretty good at this process. The problem is that some clinicians never develop that doorknob instinct, and the ones who do likely require years of clinical experience to develop it. Being able to teach this skill could be possible, if we could break down exactly what it is about a patient (or a wound) that contributes to predictive ability. There must be specific elements that clinicians are interpreting, even if that analysis happens in the subconscious mind (eg, patient coloring, facial expression, posture, peculiar characteristics of a wound that can’t be measured with a ruler). 

An exciting dividend of EHR adoption is the possibility that we can “teach” the EHR to conduct the doorknob test and perform healing predictions. If we could identify the factors that determine whether or not a wound will respond to a specific treatment, then we could automate that process within the EHR. If the EHR can first identify which wounds are not likely to heal with routine care and which wounds are likely to respond to a given AT, suddenly any wound care practitioner can be a “genius” at making treatment decisions, regardless of experience. An EHR that could do this would reduce the total cost of care, improve the effectiveness of the treatments provided, and improve outcomes, because the patients who require an AT would be identified immediately and prioritized for those interventions. The good news is that this capability currently exists. The bad news is that practitioners have little motivation to employ it until the reimbursement structure demands it. Since we are on the cusp of a new physician payment structure, it’s time to discuss what can be done with EHR technology and to contemplate how we might make clinical decisions more accurately. 

New Paradigm for HBOT 

By way of example, let’s use patient selection for hyperbaric oxygen therapy (HBOT) to illustrate a new paradigm for the use of ATs. (This concept was published in detail in 2016.¹) For all the controversy surrounding HBOT in the treatment of diabetic foot ulcers (DFUs), there’s wide agreement that tissue hypoxia, usually due to ischemia, will delay or prevent wound healing and that ATs (such as cellular products) are unlikely to be helpful without adequate tissue oxygen levels because tissue oxygen levels control angiogenesis and the neutrophil function that mitigates infection. Since the 1990s, many ATs have come to market, but their proven efficacy in randomized, controlled trials (RCTs) is limited to superficial ulcers that do not involve tendon or joint capsule among patients with adequate arterial supply. In other words, the majority of the advanced products currently on the market and in clinical use have been demonstrated to work only in easiest-to-heal ulcers. Their effectiveness among real-world patients is unknown. Despite the critical relationship between tissue oxygen and healing, when 1,172 RCTs in wound care were reviewed, none were directed at improving angiogenesis in ischemic ulcers except those involving HBOT and vascular endothelial growth factor.² When the RCTs for cellular products are compared with the RCTs involving HBOT, only the HBOT trials include patients living with significant comorbid diseases, Wagner Grade III or greater DFUs, and ischemic vascular disease. The superficial ulcers that have been enrolled in cellular product RCTs use epithelialization as the primary outcome. However, the endpoints for the more serious DFUs enrolled in HBOT prospective trials measure outcomes such as amputation rate, infection incidence, and/or improvement in transcutaneous oximetry values (TcPO₂). The mechanism of action of HBOT reviewed by Thom³ in 2011 includes an increase in growth factor levels, the activation of growth factor receptors, and reduced neutrophil adhesion. Ultimately, HBOT reduces pathological inflammation and stimulates angiogenesis. The systematic review of HBOT for chronic wounds by the Cochrane Collaboration’s wounds group showed an increase in the rate of ulcer healing with HBOT at six weeks compared to non-HBOT-treated ulcers.⁴ Huang’s systematic review,⁵ which formed the basis of the Undersea & Hyperbaric Medical Society’s clinical practice guidelines for the diabetic foot, analyzed nine RCTs and more than 20 observational studies. This evidence supported the benefit of HBOT in preventing amputation and promoting complete healing in patients living with Wagner Grade III DFUs or greater who have undergone a surgical debridement or who have shown no significant improvement after at least 30 days of conservative care (level of evidence assessed at moderate). There’s inadequate evidence to justify HBOT in Wagner Grade II or lower DFUs. These conclusions align with the CMS national coverage determination for DFUs, which specifies that HBOT coverage is limited to Wagner Grade III ulcers that have failed to respond to 30 days of standard wound care.

Given the positive prospective trial data supporting the efficacy of HBOT for severe DFUs, why were Thom and Margolis et al recently unable to demonstrate the effectiveness of HBOT in a large retrospective analysis?⁶ Using data obtained from a wound center management company database, they studied 6,259 patients living with DFUs and determined those undergoing HBOT were less likely to heal a DFU and more likely to have an amputation. Although the authors stated the cohorts were defined by CMS criteria for using HBOT in DFUs, the majority of the DFUs treated with HBOT in this dataset had Wagner Grade II ulcers, lesions for which the data do not suggest HBOT is useful and that are excluded from Medicare coverage. The database contained no information that could ascertain whether the other requirements for HBOT patient selection had been met (eg, vascular screening, control of infection), nor was there sufficient information to distinguish major amputations (eg, below or above the knee) from minor amputations (eg, toe or partial foot) that preserved function. Without this vital information, patients whose ambulation was preserved with a partial foot or toe amputation were still considered hyperbaric failures. In other words, the majority of patients treated at 83 wound centers in 31 states received unnecessary (and thus inappropriate) HBOT, since they were treated for Wagner II ulcers that might have been managed via less-costly methods. Additionally, HBOT may have been found to be ineffective in the Wagner III ulcers because vital conservative care and treatment may not have been provided prior to HBOT. What can be done to improve this situation?

Predicting Likely Wound Failure

Although we know hypoxia prevents healing, basically the only RCTs directed at improving the critical issue of tissue hypoxia are those involving HBOT. However, the real-world use of cellular products is increasing, even among severe ulcers where their value is unclear, while access to HBOT is being restricted among patients with few alternative treatments and in whom the value of HBOT has actually been proven. Although we know HBOT is ineffective when it’s provided to patients who could heal without it (unnecessary care) or when it’s provided to patients who cannot be helped (futile or inappropriate care), CMS has only blunt tools to limit the inappropriate or unnecessary use of ATs. Medicare has sought to limit unnecessary treatments by requiring that HBOT be reserved for patients who exhibit “no measurable signs of healing for 30 days,” per most local coverage determinations (LCDs). The goal of this policy is to use time to separate wounds that are going to heal on their own from wounds that cannot heal without an AT. Unfortunately, there are no data to support the requirement for “no measurable signs of healing over 30 days” as a patient-selection tool. However, there are data confirming that wound healing trajectory based on surface area (SA) measurement over four weeks can be used to identify ulcers unlikely to heal. Wounds that fail to reduce in SA by at least 40% over four weeks are not likely to heal within 12 weeks. In other words, some decrease in SA may occur over four weeks in wounds that are still destined to fail. That means limiting HBOT to wounds exhibiting absolutely no evidence of healing is overly restrictive and prevents some patients who need HBOT from receiving it. On the other hand, per LCD policy, after 30 days any wounds that meet this restrictive criterion can be treated with HBOT. The result of this policy is that some wounds that probably would not benefit from HBOT due to their severity may still receive it. As a result, the policy CMS is currently using for patient selection denies HBOT to some patients we know need it and provides HBOT to some patients we know can’t benefit from it. A more accurate method to predict the likelihood of healing is via mathematical modeling, which can be performed using the data collected at the first visit. The Wound Healing Index (WHI) is a suite of mathematical models for seven wound types that combines patient and wound variables. A predictive model specific to DFUs has been validated, which, by using the data available on the first visit, can predict the likelihood of the DFU healing with conservative care alone. It’s not known whether payers will support the use of models to select the patients most in need of ATs. However, doing so would allow interventions like HBOT to be employed earlier, at a time when they are more likely to be of benefit, and without the added expense of four weeks of care that could have been predicted to fail. 

Reducing Wasted HBOT 

The best predictor of benefit from HBOT in DFUs is the TcPO₂ obtained while the patient is undergoing HBOT. In-chamber TcPO₂ alone is 74% accurate at predicting benefit from HBOT in DFUs. In a large retrospective study, when the in-chamber TcPO₂ was ­­­­­­
> 200 mmHg, 84% of DFUs benefited; when it was <100 mmHg, only 14% benefitted.⁷ Predicting the response to HBOT can also be improved with mathematical modeling. A relatively simple model available since 2007 uses baseline TcPO₂, pack-per-year smoking history, Wagner grading, patient age, and the number of years of diabetes to predict the approximate number of HBOT treatments needed to achieve benefit.⁸ This model, which does not require an in-chamber TcPO₂, has the advantage that the patient does not have to undergo a hyperbaric treatment in order to determine the likely benefit of HBOT. This same retrospective analysis of more than 1,000 DFUs suggested there’s no incremental benefit after 40 hyperbaric treatments. Thus, if the model predicts more than 40 treatments will be necessary to achieve a positive outcome, HBOT may not be a realistic treatment option. However, if the in-chamber TcPO₂ is ≥ 200 mmHg and the predictive model suggests between 15-40 treatments will be required, there is a reasonable likelihood of benefit from HBOT for a DFU.

Operationalizing Predictive Models Within EHRs

The DFU WHI is now embedded within an EHR and a prediction regarding the likelihood of wound healing is available at the conclusion of the initial visit. Clinicians using EHRs that are not equipped with DFU WHI can simply ask patients the 10 questions listed in the Table (page 22), and the U.S. Wound Registry (USWR) can provide the healing prediction. 

The Figure above compares the use of predictive models to guide patient selection compared to current Medicare coverage policy. Currently, clinicians implement conservative care and wait 30 days, after which only patients who have no evidence of healing are HBOT candidates. This approach excludes some patients who need HBOT because they had small reductions in SA (but who still are unlikely to heal) and delays HBOT for all patients who need it. An alternative approach is to perform the DFU WHI on the first visit to identify ulcers unlikely to heal with conservative care. The next step would be to determine which of the ulcers “predicted to fail” can actually benefit from HBOT. One method is to perform an in-chamber TcPO₂. Another is to use the HBOT predictive model to determine whether or not benefit can be expected within a reasonable number of treatments. This stepwise approach would enable the clinician to decrease the inappropriate use of HBOT by accurately predicting which DFUs will not heal spontaneously, and then predicting whether HBOT is likely to be of benefit in them. As you see in the Figure, this new paradigm of patient selection would be more accurate than current Medicare policy at identifying both patients who need an AT and the wounds that are likely to benefit from it. In other words, we have taught the EHR to do the “blink” test, but to do it with more accuracy and consistency than even the most experienced human clinician.

The Future is Now

Why aren’t we using this method in daily practice if it’s available? It’s not clear whether patients, payers, or practitioners will support the allocation of healthcare resources like HBOT using tools like mathematical models. Although the WHI is calculating the predicted likelihood of healing within one currently used EHR, the prediction has not yet been revealed to the clinicians because there hasn’t been a community-wide conversation about the possible unintended consequences of revealing it. The WHI was intended to help identify patients who need ATs at an early stage (eg, the initial visit), so that these treatments could be provided earlier. However, the predictions could also be used by clinicians or payers as an excuse to withhold care. As we move toward payments based on “episodes of care,” both clinicians and payers may be motivated to avoid treating patients who are predicted to require expensive ATs. Home nursing agencies might decline to provide services to such patients because they will likely lose money in a capitated environment. On the other hand, the increasingly draconian methods by which Medicare Administrative Contractors (MACs) are limiting access to HBOT are the result of not having better methods to predict the need for and benefit of such ATs. Without better ways to allocate resources, the MACs have no alternative other than to craft coverage policies with increasingly restrictive language, or to use expensive individual chart reviews to decide whether to pay for treatments based on the highly variable interpretations of that language by human reviewers.

Are We Ready For This?

Are we ready to allocate treatments using models? How reliable are they? These first-generation models are more reliable than Medicare’s current methodology for patient selection. However, better models could be developed if there were even nominal funding for the work. The WHI was partially funded by a grant from KCI (now Acelity,^™ San Antonio, TX), with the balance of funding from the USWR and the Institute for Clinical Outcomes Research. There was no funding for the development of the HBOT predictive model in 2007, and no payers were interested in using it a decade ago. However, they may be interested now. The Meaningful Use mandate that providers submit data to a specialty registry has resulted in a significant expansion of the data available in registries like the USWR for clinical research. Those data could be used to improve the current models and create new models if funding for the work existed. If the wound care industry does not support this type of work, the only method payers will have to ration the use of ATs is to craft increasingly restrictive coverage policies. The result may be that, instead of making better decisions using predictive models, payers will simply deny coverage for an AT. That appears to be what is happening to HBOT. More restrictions on the use of other ATs like cellular and/or tissue-based products and negative pressure wound therapy will likely be implemented by payers. Predictive models can be developed for all of these ATs at relatively low costs compared to prospective clinical trials. The other good news is that when predictive models are based on real-world patient data, unlike RCTs, the models can be used to inform the care of real-world patients. We are at a crossroads in the QPP, and the question is whether we are willing to adopt a new paradigm for patient selection or accept the loss of treatment options for patients we know are going to need them, even while we still have our hand on the doorknob. n

Caroline E. Fife, MD, FAAFP, CWS, FUHM, is chief medical officer at Intellicure Inc.; executive director of the U.S. Wound Registry; medical director of St. Luke’s Wound Clinic, The Woodlands, TX; and co-chair of the Alliance of Wound Care Stakeholders. 

References

1. Fife CE, Eckert KA, Carter MJ. An update on the appropriate role for hyperbaric oxygen: indications and evidence. Plast Reconstr Surg. 2016;138(3 Suppl):107S-16S.

2. Carter MJ, Fife CE, Walker D, Thomson B. Estimating the applicability of wound care randomized controlled trials to general wound-care populations by estimating the percentage of individuals excluded from a typical wound-care population in such trials. Adv Skin Wound Care. 2009;22(7):316-24.

3. Thom SR. Hyperbaric oxygen: its mechanisms and efficacy. Plast Reconstr Surg. 2011;127(Suppl 1):131S–41S.

4. Kranke P, Bennett MH, Martyn-St James M, Schnabel A, Debus SE, Weibel S. Hyperbaric oxygen therapy for chronic wounds. Cochrane Database Syst Rev. 2015;(6): doi: 10.1002/14651858.CD004123.pub4.

5. Huang ET, Mansouri J, Murad MH, et al. A clinical practice guideline for the use of hyperbaric oxygen therapy in the treatment of diabetic foot ulcers. Undersea Hyperb Med. 2015;42(3):205–47.

6. Margolis DJ, Gupta J, Hoffstad O, et al. Lack of effectiveness of hyperbaric oxygen therapy for the treatment of diabetic foot ulcer and the prevention of amputation: a cohort study. Diabetes Care. 2013;36(7):1961–6.

7. Fife CE, Buyukcakir C, Otto GH, et al. The predictive value of transcutaneous oxygen tension measurement in diabetic lower extremity ulcers treated with hyperbaric oxygen therapy: a retrospective analysis of 1,144 patients. Wound Repair Regen. 2002;10(4):198–207.

8. Fife CE, Buyukcakir C, Otto G, Sheffield P, Love T, Warriner R 3rd. Factors influencing the outcome of lower-extremity diabetic ulcers treated with hyperbaric oxygen therapy. Wound Repair Regen. 2007;15(3):322–31.