Finding 1: practice change disturbs the status quo equilibrium—establishing a new equilibrium that incorporates the change may be a struggle
When a change is introduced, be it through the introduction of new guidelines or self-initiated change, physicians face various struggles in order to successfully change practice and restore the “equilibrium.” One such set of struggles includes those related to the change itself, such as the ease of implementation or inherent discomfort with the change. Respondent 2, referring to new cholesterol guidelines, describes it as follows:
“Their whole background of thinking – this whole idea of not treating to a number. I think we like to have finite goals and endpoints, which some will argue we still have that, now we’re just treating to a goal dose instead of a goal LDL. I think it’s good to treat a patient and not a number, so that’s fine. It’s still a shift in paradigm. Many changes not only challenge old habits but also require a shift in paradigm regarding a physician’s approach to the practice, resulting in even more of a struggle in adopting the change.”
The nature of the change itself also affects the ease of unlearning. For instance, a change that requires an addition is perceived to be much easier by our respondents to implement than a change that requires the cessation of a current practice. Respondent 3 stated:
“It’s easier to introduce something new than to take out something old. Bringing in a new drug, a new pill to people and telling them that this has showed good evidence, people will introduce it right away, it will be on formulary in 2 weeks and physicians will be prescribing it left and right. Telling people to stop doing something is almost close to impossible. It takes so many years, it takes so many changes, it takes so many events, and it still doesn’t happen in 100% of the people.”
Moreover, changes in guidelines often introduce uncertainty into a field of practice that was very “certain” and well defined in the minds of many physicians. Respondent 2 adds:
“I guess just again thinking you knew something. So I guess that’s the hardest thing is feeling like ‘oh, I thought I mastered my knowledge about something but now I didn’t or now it’s changed’. It just adds a barrier to the patient outcome.”
Another source of struggle that physicians face is a result of disjointed decision making involving third parties, such as hospital management, physician colleagues, and support staff. For instance, a physician might have incorporated the change into his/her practice, but the other staff that the physician works with have not unlearned the current practice. Respondent 4, regarding the new cholesterol guidelines that call for the cessation of regular liver function tests (LFTs) for patients on statins, states:
“What is annoying to me as a clinician is that somebody who runs the clinic thought it would be a good idea for efficient patient care for patients that haven’t been seen in a while to have a nurse go through their record and see if they need labs done prior to their visit with their clinician. I don’t know what the nurse’s guidelines are or why she does it but she goes ahead and blanket orders a CMP (comprehensive metabolic panel), an A1c, I don’t know what prompts her to do whatever she does, LFTs, she’ll get everything. So then the patient gets all this stuff.”
Patients may also provide resistance to unlearning depending on both their understanding of and comfort level with the change. When Respondent 6 decided to change her recommendations regarding NSAID use, she explained, “This was challenging because telling patients that they can’t take ibuprofen doesn’t go over well, but it’s a product of learning new information and assimilating it into practice. Sometimes I get resistance from patients.”
While these opposing forces complicate the unlearning process, physicians that successfully unlearn continually balance these forces, implying that the new equilibrium they arrive at is dynamic in nature. For instance, part of establishing the new “equilibrium” requires physicians to build experience with the new practice to overcome the initial discomfort. As Respondent 4 explains,
“Sometimes it’s not so clear what to do, but I have a pretty high risk population and I’m dealing with a pretty safe drug and I’ve got a lot of years experience using it and reading about it, so I’m relatively comfortable with when and how to use the drug and when not to.”
While guidelines do help to guide practice, they do not instruct on how to implement a practice change in every patient scenario. As a result, successful unlearning requires physicians to not only use their prior experience and clinical judgment but also build new experience with the practice change in a continual fashion to engage in the new equilibrium.
Finding 2: part of the struggle to establish a new equilibrium incorporating a practice change involves both the “evidence” itself and tensions between evidence and context
While evidence-based medicine is often cited as a driver for change and unlearning, what is considered to constitute evidence varies greatly among providers. When we asked respondents to describe what drives change in their practice, all 15 made references to evidence-based medicine. However, they differed in their interpretations of evidence and the extent to which their practice is truly evidence based. Initially during the interviews, 9 of 15 physicians cited guideline-driven changes as examples of changes in their practice. As physicians had a chance to reflect on their practice, many also came up with examples of non-guideline driven changes—reflecting on their practices and changes in it led to new kinds of changes being recognized. While guidelines are a common example of evidence-based drivers of change, our interviews revealed that there are other sources that provide sufficient evidence to drive change.
First, not all guidelines are created equal and some individuals review the quality of guidelines for themselves [46–48]. Respondent 12 described the process she uses to analyze the validity of guidelines:
“So I had the opportunity to read through all the data that prompted this guideline change. I critically reviewed the evidence myself and convinced myself that this is a guideline I want to follow, given that there are multiple other guidelines out there that other providers are following. I had to convince myself that I believed the evidence and that was the first thing.”
Thus, the quality of the evidence backing guidelines and large studies is an important factor in determining whether physicians will use such evidence to support their unlearning.
When guidelines and randomized controlled trials (RCT) are not available, physicians use local data that they gather from their patient population to support unlearning. Respondent 1 illustrates this using the hesitancy of administering Lovenox without a RCT as an example, “Heparin drips take 12 hours to be therapeutic and Lovenox is therapeutic immediately. So the amount of time that a person spends out of target range on heparin drip is pretty significant yet because there was no evidence, no RCT, people were very reluctant to use Lovenox even though from the common sense standpoint you could easily measure time within target range which was always 100% for Lovenox and something under 100% for heparin drip.” These personal deductions that Respondent 1 made regarding the efficacy of Lovenox were grounded in critical evaluations of patients on heparin versus Lovenox. Respondent 1 used her judgment rooted in clinical expertise to change her practice and provide what she thought to be the best care for her patients. In addition, Respondent 1 had access to local patient data to support her decision. “The pharmacists had done this excellent study looking at our patients at this VA, all of them that had been on heparin drips, and what percentage of time they fell outside the therapeutic range and it was highly disturbing. So that kind of put the nail on the coffin for me.”
Besides clinical observations, there are many other unstructured sources of evidence that guide practice change, such as casual conversations and discussions with colleagues. Respondent 13 describes:
“I think being in an academic environment makes adapting change easier. I think you have local experts that you trust, you have an opportunity to discuss the pros and cons, it’s hard to talk to a paper, and I think that’s an advantage to being in an academic center is having colleagues that are similarly interested in providing evidence-based medicine, you’ve got sub-specialists that similarly do that.”
Without input from colleagues, physicians find it difficult to change their practice solely based off the literature—the opinions of different physicians and discussions on how to incorporate data into practice are critical forms of evidence that physicians use to support their unlearning.
In addition, the respondents differed regarding their opinions of the extent to which it is possible to practice purely evidence-based medicine. Respondent 5, who is at one end of the evidence-based medicine spectrum, argues,
“How you take up a revised recommendation depends heavily on your personal experience and your personal biases, how you interpret literature. And I think this is how it works for everybody, we all have a mouthful with evidence-based medicine, but very few of us are truly evidence-based. Most of us are like a snippit-based, based off of what you hear, what you read, what other people explain to you, what a mentor or somebody you really respect says. It’s a mixture.”
Respondent 5 revealed that a physician’s clinical practice in reality is affected by many sources of information, including classical evidence consisting of published literature, but also unstructured sources of data, such as conversations with colleagues and personal clinical experience. All of these inputs assist physicians in making clinical decisions that are both grounded in data and practical experience.
While every physician we spoke with attested to practicing evidence-based medicine, evidence was very rarely the only determining factor of a physician’s clinical decision-making process. Many physicians stated they used guidelines to “guide” but not necessarily dictate their clinical decisions. Respondent 4 exclaims,
“So certainly anyone with a history of LFT abnormalities, chronic liver disease of any kind, obviously the guideline doesn’t apply to them – if they’re on a statin, I keep monitoring, as being on an additional potential hepatotoxin is a concern. The guidelines are just for patients that have healthy livers and who have no other issues going on, they’re otherwise doing well, no other concern that they’re developing any illness that might involve the liver. I individualize the guidelines. The guidelines don’t say don’t order LFTs, they say only don’t order them for folks who are healthy and who don’t have baseline liver disease. You individualize, obviously we are physicians, we are not robots that just blindly follow guidelines.”
As Respondent 4 demonstrated, physicians individualize guidelines and ascertain whether the guidelines are applicable to a particular patient.
Physicians also approach guidelines with caution because the study population in papers that support guidelines do not always match the much smaller sample of patients that physicians see in their clinics. Respondent 6 expanded upon this discrepancy regarding prostate-specific antigen (PSA) screening guidelines,
“I had a lovely patient, but he does everything I say and he went under my influence and we didn’t get a PSA. When he was 70 he started having back pain and dizziness and he walked into an outside ER, they did a PSA and it was over 100 and he had metastatic prostate cancer and he lived for 6 months or a year, not a long time. I went to visit him in his nursing home and he’d gone down hill. It just killed me. I kept thinking if I had done the PSA maybe I could’ve saved him. So after that, evidence-based medicine was out the window and I’ll sneak in PSA into people’s blood work and not even tell them. I’m exaggerating a little but I’d say let’s get a prostate test today, so they’d say yes so I’d throw it in there. I would rather live with causing biopsy problems or a surgical complication than to have someone die because I didn’t check. For this patient, he was 70 and he wasn’t even supposed to get a PSA but he was complaining about back pain so he got one. But these guidelines are guidelines and they don’t always fit every person and you’re going to order a lot more labs than you really should because of things like this. Something in me changed where there’s a responsibility in taking care of people and I was asking myself what can I live with.”
Respondent 6’s poignant story sheds light upon the disconnection between PSA screening guidelines and the actual patients that Respondent 6 sees in her practice who might actually benefit from PSA screening. On a similar note, Respondent 5 added,
“In the end, on a population level, the heavy use of HRT probably did cause some problems, but for me as an individual I probably didn’t have a single patient in my practice who got breast cancer from HRT, right. So this is sort of the disconnect you have between population management and the few 100 patients I saw.”
Since Respondent 5 had a hard time relating to the guidelines, she left the ultimate decision-making to her patients. Respondent 5 described,
“And I think the key really is that you have to decide with your patient, not for your patient. And I had women that said I’m not coming off HRT, so I said ok you are not coming off HRT.”
Respondent 5’s lack of personal experience with the adverse effects of HRT and respect for patient autonomy drove her to continue prescribing hormone replacement therapy (HRT) for patients that desired it.
The care that the respondents have for their patients was best elicited during discussions regarding the implementation of practice change. All 15 physicians were willing to make exceptions to the guidelines for the sake of respecting patient autonomy and providing patients with care that catered to their needs. Respondent 13 describes,
“Where I have some issues is the moderate versus high intensity statin especially when it comes to side effects for the patients. In someone that is naïve to statins, putting them on a high intensity makes me a little nervous. I think patient factors are important, what’s their adherence like, what is their tolerance in general to medications and side effect profiles. I sometimes err on the side of moderate intensity because for a lot of my patients, if they develop side effects, they’re afraid, and they want to do nothing. I’m looking at the long term, I don’t want him to take the statin for 3 days, I want him to take it for potentially years.”
While the guidelines might have indicated a high-intensity statin for a particular patient, Respondent 13’s care for her patients and concern for possible side effects and long-term medication compliance led to her incomplete adherence to the guidelines.
Other respondents developed similar contingency theories where they presented various patient-specific scenarios that would cause them to be non-compliant with guidelines and use clinical judgment to guide decision-making. Respondent 14 elaborates,
“So I had one person who was like I don’t care I just want the mammogram, because people get worried about cancer. It’s hard to explain harm from screening to patients. I felt like most of them still wanted it. I felt like I tried to explain it but cancer scares a lot of people and a lot of people think ‘I rather go through all this and not have cancer than to have cancer and not know’. I am still trying to respect my patients’ autonomy.”
In these situations, both Respondent 14 and the other respondents’ inherent care for their patients and desire to address their needs outweighed consistent practice change supported by literature-based evidence.