There is increasing coverage and discussion in the press about the empowered consumer and all the tools that are now available for people to use to help be empowered. The symptom checker is one of the key tools that consumers should use as it helps make sense of all the data such as symptoms, abnormal test results or abnormal readings from the various monitors that are now available.
There is a great regular Podcast series called “Stuff you should know” http://www.stuffyoushouldknow.com/ hosted by Chuck Bryant and Josh Clark and, earlier this month, they released a Podcast called “Will computers replace doctors?” The full podcast is 36 minutes long and can be accessed from this link http://www.stuffyoushouldknow.com/podcasts/computers-replace-doctors/
We have edited this to produce an abridged version if you want to listen to just the segments related to diagnosis and using diagnosis decision support tools or symptom checkers. Click here to listen to the edited podcast. A full transcript of the edited version is also provided in italics below.
As Chuck and Josh observe, there is an increasing desire by people to monitor their health and take more control in how they interact with their doctor. Most consumers have done some research, internet searches, etc prior to their visit to the doctor. This engagement by the healthcare consumer can lead to a more productive visit and assist with the outcome of the visit. No longer are they going just to be told what’s wrong with them but want to have a meaningful exchange with their physician based on what they have learned; asking questions like “what else could this be or what are the treatment options" etc.
A proven symptom checker, like diagnosis decision support mentioned in the blog, enables healthcare consumers to enter their symptoms and receive back a list of diagnosis possibilities, including both common and rare conditions. Having a list of likely diagnoses to discuss with the doctor rather than just sitting passively waiting to be told what the doctor thinks engages the healthcare consumer to care about their care and, in turn, are more likely to engage in their treatment as well.
Diagnosis decision support for physicians and symptom checkers for consumers are formidable tools for both in helping to get to the right diagnosis and treatment as soon as possible but also in facilitating an informed conversation between the patient and physician.
Narrator: Welcome to Stuff You Should Know from howstuffworks.com.
Josh: Hey, and welcome to the podcast. I'm Josh Clark. There's Charles W. 'Chuck' Bryant. Our guest producer Noel is here. To me, and I don't think it's over confirmation bias, it seems like there really is a growing desire among just average ordinary people to be able to track their health...
Josh: ...their well being, their activity...
Josh: ...and to do it easily.
Josh: Yeah. It seems to me this desire to kind of say hey, this is my health. This is my body.
Josh: I want to know more about it.
Josh: Like, I don't want to necessarily cut out doctors, but I want to decide if I should go to the doctor if it's time or not.
Josh: And, I want to use data to do that.
Chuck: Yeah. I imagine I frustrate a lot of doctors, because I'm one of those obnoxious people that goes in and is like well, here's what I think I have based on my research.
Josh: There's nothing wrong with that.
Josh: That is what... You're an informed patient.
Josh: That's exactly what you're supposed to do, and if you're getting on your doctor's nerves then go see another doctor.
Chuck: Yeah. I mean there's two sides to this. There's diagnoses and treatment. Some programs... A little bit of the history. This goes back to the 1970's at the University of Pittsburgh. They developed software to diagnose problems. Mass General since the 1980's has been working on their DX plan which provides ranked lists of diagnoses. Whereas the... What's the computer, the Watson?
Josh: Watson, who won at 'Jeopardy.'
Chuck: Yeah. That's more based, it looks like, on treatment options than diagnosis at this point.
Josh: It's both.
Chuck: They're using these... Well, yeah, but they said it's not... They haven't... I don't think they want to leave it alone with diagnosis yet...
Chuck: ...and to do its thing.
Josh: There's already something out there for diagnosis that's meant to support physicians.
Chuck: I know we looked into this one, sort of a savant diagnoser. Is that a word?
Josh: I don't know.
Josh: Diagnostician, yeah.
Chuck: Dr. Dhaliwal in San Francisco is sort of legendary for diagnosing things to the point where he does it on stage as almost like a parlor trick.
Josh: I would love to see it.
Chuck: I would, too. They give him 45 minutes and a bunch of symptoms basically, like really confusing, because they're trying to stump him.
Chuck: Generally, he comes out on top. But, he even uses a diagnostic program called Isabel.
Josh: Right. That's the one I said earlier that's already here.
Chuck: Yeah. Doctors are using these to help themselves out. But, he says that he's never had Isabel offer a diagnosis that he has missed.
Chuck: He's like the dude, though.
Josh: Yeah, and he also admits that he's like a freak of nature.
Chuck: Right - go ahead, quiz me.
Josh: He also reads case histories for fun, that kind of stuff. He's not...
Chuck: He really puts the time into it.
Josh: ...a normal physician. He's a complete and total outlier.
Josh: If every physician were like this guy then there probably wouldn't be this conversation going on right now.
Chuck: Yeah, you're right.
Josh: But, most physicians aren't, and it's not just with current medical research that they're just not aware of because they haven't had time to pick up 'The Lancet' the last few months.
Josh: It's also their training, too. Like, if a doctor's in practice for 20 years... The human brain tends to create habits, because it likes to expend as little energy as possible. It's trying to be as efficient as possible. I think the same thing happens with medical practice. You're trained. You understand. You come out of medical school with a lot of book learning. Then, you put it to practice and you kind of find your niche. Along the way you forget a lot of the stuff...
Chuck: Oh yeah.
Josh: ...that you haven't done in 20 years or haven't learned about in 20 years. It's not just current stuff. It's old stuff, too.
Chuck: Well, here's a scary stat. One in five diagnoses in the United States are incorrect or incomplete. One in five. A lot of times it's not that the doctor's a jerk or not any good. But, like you said, they just maybe haven't seen these cases that were written about in some obscure medical journal that the computer has scanned and indexed.
Josh: Dr. Dhaliwal himself says even with me a lot of it is intuition...
Josh: ...and intuition can be wrong.
Chuck: Yeah. I have this one stat, too. It says according to an expert - I'm not sure what that means, it sounds hinky - only 20 percent of the knowledge of physicians used to diagnose is evidence based. So, that means 80 percent is intuition?
Chuck: I like the idea of intuition to a certain degree, for sure.
Chuck: But, there's also got to be, like, data backing it up.
Josh: Sure, right.
Chuck: You know.
Josh: So, in your perfect world then it sounds like we still have physicians, but they go back and double check themselves using a program. Okay. Tell me your alarming stat.
Chuck: All right. Johns Hopkins did a study that found as many as 40,000 patients die in intensive care each year in the US due to missed diagnosis. 40,000.
Chuck: And, another study found that system related factors like lack of teamwork and communication or just poor processes...
Chuck: ...were involved in 65 percent of diagnostic error, and cognitive factors in 75 percent with premature closure as the most common - which is basically just sticking to that initial diagnosis and not being open minded to other second opinions.
Josh: Yeah. There's this thing called 'anchoring bias' that was in that 'New York Times' article...
Chuck: Yeah, yeah.
Josh: ...with Dr. Dhaliwal. The guy who created this program that's now around to support diagnostics where a physician will say I think it's this, but let me put in the symptoms and ask Isabel...
Josh: ...which is the name of the program. It's named after the guy who created the program's daughter.
Chuck: Oh, man, that story's rough.
Josh: Yeah. When she was three, he took her to the hospital and the doctors said well she has chicken pox. She did indeed have chicken pox, but that's all they looked at.
Josh: They completely missed a pretty nasty case of necrotizing fasciitis...
Josh: ...which we've talked about before.
Chuck: Oh yeah.
Josh: Flesh eating bacteria. She almost died from it and was disfigured from it as a result.
Josh: Her father who was a money manager said I'm going to take whatever computer programming skills I have and put it toward this program Isabel which is meant to say yes you're right with this diagnosis I agree with you, or have you considered these other diagnoses.
Josh: He said had Isabel been around and his daughters' doctors consulted it they would not have missed the necrotizing fasciitis.
Chuck: Well, it makes sense - as an assist, you know. There's this company called Lifecom that said in clinical trials if you use a medical diagnostic program as an assist those engines were 91 percent accurate without using exams or imaging or labs even.
Josh: Really just symptoms.
Josh: That's crazy. That's really, really, really good.
Chuck: But, as an assist then I think it's kind of a no brainer...
The Wall Street Journal ran a special report on healthcare this week and included two great articles that both referred to the Isabel diagnosis decision support tool:
Diagnostic error received scant attention only 2-3 years ago but now consumer press articles in highly respected papers such as the Wall Street Journal and peer reviewed studies in high impact journals are appearing frequently.
This lack of attention has always seemed odd when internal analysis of serious incidents invariably finds missed or late diagnosis to be the cause wherever it happens.
A recent study looking at appropriate prescribing of antibiotics using decision support even found the real problem was that the initial diagnosis made was incorrect over 30% of the time. This demonstrated the complete futility of trying to improve healthcare without improving diagnosis, the first and most important decision made about the patient
Bob Wachter is one of the leading patient safety figures in the USA and writes a very popular blog. He recently wrote one called “Diagnostic Errors: Central to Patient Safety, Yet Still In the Periphery of Safety’s Radar Screen”. The first comment made to the article was profound and is included in full below:
“You can either have excuses or get results–but you can’t have both.
Several years ago, our integrated system recognized the limitations of the National Quality Forum’s Serious Reportable Events and The Joint Commission’s Sentinel Events. Some of our leaders added a few additional events, one of which is: “Death or Serious Disability After a Missed Diagnosis”. This is a Board level report that triggers a full event response–immediate actions including a full Root Cause Analysis, identification of causative factors, action plans, and sharing across the system.
Next week, I will be presenting our monthly events to the Board. It’s a solemn and sombre experience. This month, half of our new events are diagnostic errors.
We certainly don’t have all the answers when it comes to diagnostic error, nor do we have all of the solutions.
We have changed the culture by identifying Diagnostic Error as a system failure until proven otherwise. In doing so, we’ve provided visibility to a closeted issue, begun the long journey of learning, and most important, started to heal our patients and staff for many events that seem frankly incomprehensible.
You can wait around for Superman or Godot or you can control your own destiny. I strongly advocate the creation of a structure within your own healthcare systems to learn, share, prevent harm, and heal.”
I firmly believe that the great tool to help improve diagnosis is the differential diagnosis. It's been taught for over 100 years yet is not carried out in routine practice in spite of research showing a differential diagnosis that includes the correct diagnosis is the most accurate predictor of diagnostic accuracy. Lack of time is often the reason given for not doing this but with modern diagnosis decision support tools available for clinicians and symptom checkers available for patients there should be a renewed focus on carrying out the valuable discipline of working up a comprehensive differential diagnosis.
A study entitled ‘The Ecology of Medical Care Revisited’ from the NEJM of June 2001 graphically shows how important it is to try and guide and influence the patient from a very early stage in order to ensure appropriate flows into primary and secondary care. Encouraging patients to use a symptom checker could be one way to help them at that crucial stage when they are considering seeking care.
The study looks at a population of 1,000 men, women and children and shows how many of those 1,000 in an average month report symptoms, consider seeking medical care, visit their doctor and finally end up in hospital.
The results showed that, on average, for 1,000 people each month:
800 report symptoms
327 consider seeking medical care
217 visit a doctor, of which 113 go to primary care
65 visit a complementary or alternative provider
21 visit a hospital outpatient clinic
14 receive home health care
13 visit an emergency department
8 are hospitalised
1 or less end up in an academic medical centre
Clearly the behaviour of the 80% that report symptoms and 33% that consider seeking medical care has an enormous bearing on subsequent flows to primary and emergency care.
You can start guiding your patients by providing them with the right tools. The Isabel Symptom Checker enables patients to enter multiple symptoms in everyday language and read up on possible diagnoses. Each diagnosis is linked to knowledge from various commonly used resources such as Medline Plus, Wikipedia, NHS Choices, patient.co.uk etc.
Perhaps a controversial headline but a study recently appeared in JAMA (Journal of the American Medical Association) ‘Physicians' diagnostic accuracy, confidence and resource requests which revealed some rather shocking figures about the levels of physicians’ diagnostic accuracy, confidence and the contribution to diagnostic accuracy made by lab tests and imaging.
- The study is brilliant and the authors are to be congratulated. However, the results, which show a diagnostic accuracy rate of 55% for the easy cases and just 6% for the hard cases, are truly shocking and the authors’ statement that "overall diagnostic accuracy was rather low- 31% across the 4 cases” must be the understatement of the year!
- The cases used for this study are known to be quite difficult so one could excuse a low level of accuracy if the physicians weren’t so confident that they were right. The study looked at how confident the physicians were in their diagnoses and what was also shocking was how little their level of confidence changed from the easy to hard ones. For the easy cases it was 7.2 out of 10 and for the hard cases it was 6.4 out of 10. So, even with an accuracy rate of only 6% for the hard cases, the physicians were still 64% confident that they were right! The issue here from the patient’s perspective is they can put up with 6% accuracy if they know that the person is only say 10% confident of being right as they know where they stand but if it’s a respected professional who appears confident this can be very dangerous. With the high confidence and low accuracy the patient has the illusion that he’s being reasonably well looked after.
- The study also looked at the accuracy at 4 different chronological stages in the diagnostic process:
- Chief complaint and medical history
- Physical examination
- General laboratory and imaging
- Definitive or specialized laboratory and imaging.
What is alarming is that the accuracy of diagnosis appears to have been barely improved after the history and physical stages with the labs and imaging, begging the question what is the point of all that expensive testing? Perhaps this shouldn’t be a surprise when one of the old adages in medicine is that 75-80% of the diagnosis is revealed by the patient’s story. The lab tests should just be used to confirm a suspected diagnosis but not as a scattergun approach that hopefully reveals the answer.
- Lack of time did NOT seem to be an issue in the low levels of accuracy, although physicians often cite this as a problem. It seems that the high levels of confidence meant that the physicians did not request additional resources. Rather than the term 'over confidence' a more apt explanation may be the 'illusion of knowledge'; the over confidence resulting from the illusion of knowledge. My personal view is that often a lack of time is simply an excuse not to do something which is either not perceived as sufficiently important or where the person believes they know what they need to already.
- The 31% overall rate of physician accuracy could be compared to results from a Pew Survey written up in a previous blog which showed that 41% of patients ‘own diagnoses were confirmed by their physicians! This is clearly taking both studies out of context but the point to emphasise is that physicians' diagnostic accuracy is far less than they would like to believe and that patients’ diagnostic accuracy is far better than physicians tend to assume. This is a perfect argument for the two groups to work together as a team towards a jointly created differential diagnosis.
- One of the solutions suggested in the study is "engaging patients in creative ways". One could be actively encouraging patients to use a sophisticated symptom checker, like Isabel, before the consultation so that they could contribute more productively. With the Isabel symptom checker integrated into a patient portal, the patient could enter their details before the consultation which could then be available through the institution’s EMR system so that when the patient presented the physician could bring it up on his computer and discuss it with the patient.Technology is the driving force behind patient engagement which allows patients to become more involved and more proactive in managing their healthcare outcomes. Patient engagement tools like symptom checkers, drug databases and medical calculators provide instant access to health information and ensure patients become active participants in their healthcare.
It is with great pleasure that we introduce our guest post from Robert Hitchcock, M.D., FACEP, Chief Medical Informatics Officer, Vice President, T-System, Inc. describing 'How to avoid Diagnostic Errors in the Emergency Department'.
Dr. Robert Hitchcock is a frequent contributor to highly regarded industry print and online publications, and was the reader's choice No. 3 pick for ONC National Coordinator in a poll conducted by Modern Healthcare magazine. He has more than 20 years of experience in healthcare with more than 10 years as a practicing emergency physician. Along with his role as T-System’s CMIO, Hitchcock is a practicing ED physician. He currently serves on the Emergency Department Practice Management Association (EDPMA) board of directors and the ACEP EM Informatics section. He received his medical degree from SUNY Stony Brook and his Bachelor of Arts in Computer Sciences from SUNY Oswego.
The unique environment of the emergency department (ED) makes it more prone to diagnostic errors than other settings of care. It’s imperative for EDs to optimize the diagnostic process in order to avoid serious errors in patient care.
What makes the ED unique?
- ED physicians only have one shot at getting it right. They’re likely not seeing a particular patient again – and, even if the patient comes back, he or she will likely see another physician.
- The time frame to reach the correct diagnosis is measured in minutes to hours and follow-ups the next day are rarely an option. This creates an imperative to get the diagnosis right the first time.
- Without a previous relationship with the patient and often limited access to previous records, ED physicians lack much of the information necessary to help inform their decisions. Even with the data, the physicians and staff generally don’t have the time to parse through it.
- The ED is a chaotic environment with numerous opportunities for interruptions. Physicians and nurses can easily get distracted and, without the right tools, distractions can lead to errors.
- ED physicians are accustomed to treating common chief complaints with common diagnoses. Diagnostic decision support tools can help avoid cognitive errors and provide assistance with atypical presentations of diseases.
Top criteria to look for when selecting an ED diagnosis tool:
- Limits data entry burden
- Requires only data expected to be available to the practitioner at any time
- Presents support based on the most current medical literature
- Includes “artificial intelligence” learning – the output is based on a system’s ongoing fluid and dynamic learning as opposed to human programmed rules
- Produces plausible diagnoses and next steps
- Provides available references as to why the system concluded with a decision
- Output can be filtered and ranked based on need (high-risk diagnoses, likely medication side effects, etc.)
Diagnosing and treating most patients is straight forward. In atypical cases, physicians need the right tool to help them quickly and easily reevaluate the diagnostic process. The decisions the ED physician makes drives the next stages of patients’ care, and diagnosis errors can have a broad negative impact on more than just the patients’ ED visit. For example, if a patient is diagnosed with pneumonia in the ED and admitted, the admitting physician will likely treat them for the next day or more for pneumonia. Before you know it, you’re two or more days into admission before it becomes evident something else is going on.
Increasingly, technology provides opportunities to improve patient care, but also has the
potential to complicate a physician’s work flow. The right technology solutions with the right attributes can naturally fit in and help ensure patient safety while protecting efficiency.
With the Duchess of Cambridge being admitted to the Lindo Wing at St Mary’s Hospital, Paddington, London there is worldwide interest as to when the royal baby will arrive, especially as her husband the Duke of Cambridge was also born there. Isabel Healthcare which produced the Isabel clinical decision support system for Health Professionals and the Isabel Symptom Checker has strong links with St Mary’s Hospital in Paddington as it is where Isabel Maude was treated, at age 3, with Necrotizing Fasciitis on the Pediatric Intensive Care Unit (PICU) after being misdiagnosed at her local hospital. It is where Jason Maude and Clinicians from St Mary’s conceived the structure of what became the Isabel clinical decision support system. Further reading on how Isabel was developed can be found here.
Background and Overview:
Labor is the sequence of physiological events that results in a fetus being transported from the uterus through the birth canal to delivery as an Infant. Labor is a clinical diagnosis where many things happen including:
- Changes in cervix ( lower part of uterus which connects with vagina) to allow passage of the fetus through the birth canal
- Synchronous, coordinated contractions of the cervix to produce the hormone Oxytocin which facilitates the thinning of the cervix membranes and movement of the fetus into the vagina for delivery. The Contractions progress in magnitude, duration and frequency as the labor progresses.
Labor is divided into three stages:
Stage 1 (Cervical stage): This occurs from onset of uterine contractions until full dilation of the cervix at 10 cms. Stage 1 contains the latent phase where contractions are mild, short and irregular (less than 45 seconds). The uterus contracts but there is little change in cervical dilation or effacement (thinning of the cervix). The next phase of stage 1 labor is called the active phase which begins around the time of cervical dilation of 3-4 cm and contractions are strong, regular (every 2-3 minutes) and last longer than 45 seconds.
Stage 2: This is from onset of complete cervical dilation (at 10 cms) to the time the Infant is delivered. This stage is influenced by the 3 P’s: Passenger (Infant size and presentation), Passageway (size of pelvis and soft tissues), Power (uterine contraction strength).
Stage 3: This stage occurs from when the Infant is delivered until delivery of the placenta. This stage can be as quick as 10 minutes but can sometimes take up to 30 minutes.
The length of time it takes for these three stages to be completed varies for each woman. Typically the lengths of the first 2 stages for women who have never given birth to a live infant before are significantly longer.
Symptoms of Labor:
- Intermittent low abdominal pain with or without low back pain, occurring regularly at least every 5 minutes.
- Each episode lasts 30-60 seconds.
- Sudden release of clear fluid from vagina or constant perineal wetness can represent rupture of membranes (rupture of amniotic sac the baby is in inside the body). This can also be called “breaking of waters”. This may not happen until the fetus is in the second stage of labor. Sometimes the waters can rupture prematurely before labor contractions start. If you experience this release of clear fluid then you should contact your Obstetrician or Midwife.
It should be noted that patients in the third trimester (final three months of pregnancy) who have abdominal pain or vaginal bleeding should contact their Obstetrician or Midwife as vaginal bleeding is not associated with labor and could be an indication of abnormal labor complications such as placental abruption or placenta previa.
Physical Examination and Workup:
Once you have been seen by a Midwife or Obstetrician they will perform some examinations and tests to confirm you are in labor and what stage you are at:
- Assess fundal height – height of top of uterus to the top of the pelvic bone to determine fetal growth and development compared to number of weeks of pregnancy.
- Sterile pelvic exam to assess cervical dilation (how many centimetres dilated) and effacement how thin the cervix is) unless vaginal bleeding is present then the pelvic exam shouldn’t be performed.
- A complete blood count, type and screen should be sent.
The patient is assessed to correctly determine they are in labor and rule out other possible diagnoses which could explain their symptoms including:
- Braxton Hicks contractions (false labor) which are irregular uterine contractions without associated cervical changes and contractions can be every 10-20 minutes
- Muscoskeletal back pain
- Uterine ligament pain
- Other causes of abdominal pain including appendicitis, ovarian cyst or a urinary tract infection
Pain Control during Labor:
Various methods can be employed during labor to relieve pain and your Obstetrician or Midwife would have discussed these with you in your birth plan prior to being admitted in labor.
Some common methods are:
- Self-help: learn about labor, how to cope with the pain, visit the antenatal unit where you may give birth. Learn how to relax and stay calm. Bring a partner, relative or friend with you.
- Gas and Air (Entonox) – This is a mixture of oxygen and nitrous oxide gas. It won’t remove all the pain but can help to reduce it and make it more bearable. It’s easy to use as delivered through a mouthpiece and you control how often you want to use it and when. There are no harmful side effects to the baby but it can make you feel light-headed, sick, sleepy or unable to concentrate.
- Pain Killing Injections – Intramuscular injection into the thigh or buttock of a drug such as Pethidine can help you relax and lessen the pain. After the injection is administered it can take 20 minutes to work and the effects last 2-4 hours. The side effects can make you feel sick or woozy and if the effect hasn’t worn off by the time you have to start pushing it can make it difficult to push in the second stage of labor.
- TENS – Transcutaneous Electrical Nerve Stimulation. These can be hired or some hospitals have one you can borrow. It’s most effective during the early stages of labor during the latent phase of stage 1 rather than the active phase. Electrodes are taped onto your back and are connected to the machine or simulator. You press a button on a handheld attachment which delivers small amounts of current through the electrodes. It is believed to work by simulating the body to produce endorphins which are its own natural painkillers. It also reduces the number of pain signals sent to the brain via the spinal cord.
- Epidural anaesthesia – An anesthetist in the hospital sets up an epidural (injection into lower back containing an anesthetic) which numbs the nerves which carry the pain impulses from the birth canal to the brain. For most women, an epidural gives complete pain relief and can be helpful for women who have a long or painful labor. Side effects can occur including drowsiness, sickness, make your legs heavy. It can also prolong the second stage of labor as you can’t always feel the urge to push.
Labor and delivery is an amazing and exciting process which, after nine months of nurturing and protecting your growing baby – you finally get to meet them.
We are often asked by clinicians whether they would be better or worse off in a malpractice case if there was a record in the medical notes of their full differential diagnosis and the diagnosis missed was in that list. Our view is that it’s always better to have documented what you have done and thought. If you were wrong and had a reasonable explanation for why you did not think the diagnosis missed and on the list was the most likely and the one that you treated for, then it is better to be able to show that you thought about other possibilities. At the very least you appear to be a concerned and caring clinician rather than one that couldn’t be bothered.
A case described in “The Clinical Advisor” this week demonstrates this very clearly.
The case concerned a urology physician assistant (Ms P) who saw a teenage boy who presented 6 weeks earlier in the ED with blood in his urine. The teenager had been seen in the ED by a urologist (Dr B) who had ordered lab results which had revealed both hematuria and proteinuria. The urologist diagnosed a urinary tract infection and prescribed antibiotics.
The teenager had returned to the ED as he had not responded to the antibiotics and was complaining of continued hematuria, sore throat, fever and right flank pain.
The urology PA called the urologist who made the original diagnosis. He maintained his view it was a UTI and said to continue the antibiotics. The PA challenged the diagnosis and questioned whether it could be something else but was overruled.
2 years later the teenager returned to the ED spitting up blood and with pain in his side below his ribs. Tests showed that his kidneys had failed and that he had immunoglobulin A nephropathy, a severe kidney disease that would require hemodialysis three times a week. A malpractice case followed.
“In his certification for the lawsuit, the nephrologist alleged that Dr. B and Ms. P had departed from the standards of practice among members of the same health professions with similar training and experience by failing to include nephritis as a differential diagnosis for the patient when he presented to the ED.
Ms. P explained to her defense attorney that she had had reservations about the diagnosis at the time, but had not noted anything on the chart and had deferred to Dr. B. The attorney explained that without notes proving that she had challenged the diagnosis, they had no choice but to proceed on the basis that the presentation appeared to be a UTI.”
It is interesting to note that Isabel’s list of possibilities for just the clinical features from the initial visit to the ED shows the actual diagnosis at the top of its list but does NOT include UTI within its top 10 suggestions. Had the PA used Isabel it may have given her much more confidence in challenging the urologist’s diagnosis and also she would have had a documented differential diagnosis proving her challenge.
“Old GP” is an actual GP based in North America who is close to retirement. He has a keen interest in strange presentations of diagnoses and a wealth of experience. He has been using Isabel over the last few months and has gone back over some memorable old cases to see whether and how Isabel could have helped. We thought that his experiences would be of great use and interest to other clinicians, not only to hear about these cases but also how Isabel could have helped in building the differential diagnosis. “Old GP” wishes to remain anonymous out of respect for the privacy of patients and colleagues.
Down memory lane…
Some pre-Isabel diagnostic failures of the past . . .We missed them – Isabel would have got them first time. It may be tedious and unproductive to put every case through a computer program, but at the end of a busy day there are always a couple of uncertain cases worth computing.
- There's the uncommon presentation of a common disorder . . .
A 3-year old boy, presented increasingly weak and listless over the past two days, has rapid pulse and breathing, and warm dry skin. He had no neck stiffness,rash, his chest was clear, belly soft and a dry diaper.
Our immediate thoughts was a viral infection and obviously in need of hospital support. The Pediatrician did a spinal tap in the emergency department. The laboratory technician had the thoughtfulness to check the spinal fluid for glucose as well as everything else. Strongly positive! The child had Type 1 Diabetes Mellitus, was given Insulin and all was fine.
The parents thanked us from the bottom of their hearts for our speed of action and investigative thoroughness. We didn't feel so clever ourselves. What a complicated way to reach a simple diagnosis. Turned out the mother had put on a new diaper just before coming to our clinic – she hadn't wanted to offend us with a wet one. (Since then we've always asked parents to bring in a used diaper. Even then it's not always easy to squeeze a few drops from the new super-absorbent brands. And then we had to fire a college-trained secretary who said that dipping urine wasn't in her job description. After that we hired pizza-waitresses – they're perfect for medical clinics, good memories, don't mind dirty work, good at handling drunks and delighted to get clinic rates of pay.)
How did Isabel do in the retrospective trial? A checklist of tachypnea, tachycardia, dry skin and lassitude put Diabetic Ketoacidosis top of the search list.
Diabetic Ketoacidosis top of Isabel’s differential diagnosis
- And the common presentation of an uncommon disorder . . .
An Irishwoman in her 30s staggered into an emergency department, confused, irritable and smelling strongly of beer. The Emergency Medical Officer (not me, I'm glad to say), took offence at a visit by an apparent drunk, and ordered her to leave. In the small hours of the next day she had a seizure and her friends took her to a different hospital. Again, a diagnosis by spinal tap – Meningeal Tuberculosis.
Could Isabel have expedited matters? Yes. A checklist of two features only, confusion and irritable, gave a search-list of possibilities including Meningeal Tuberculosis .
Tuberculosis Meningitis presents on Isabel’s differential with 2 key symptoms
The key word here is possibilities to help build a good differential diagnosis. Medical and Nursing training give us a good sense of the probabilities, but we humans are not so good with the possibilities, and that's where a computer search can help so well.
- Sometimes the viewpoint of a beneficent alien like Isabel is better than human assumptions . . .
An 80-year old man, a widower, living alone, was on our regular visiting-list. (My father used to say that he'd rather see ten old people in the daytime – even if only one of them showed any hint of decompensation – than be called out at 3 a.m. to deal with a single paroxysmal nocturnal dyspnea.)
This old gentleman, though never acutely ill, did seem to be going gradually downhill. His house became messier, he complained that his dentures hurt, and we saw petechiae on his forearms. His diet appeared good – there was always a bowl of fresh fruit on his living-room table. It took referral to an internist to reveal that he had scurvy.
How would Isabel have done, if we'd had her back then? A checklist of lassitude, gingivitis and petechiae gives Vitamin C Deficiency top of the list. But what about the fruit bowl? “Well, doctor, I never liked fruits. I just keep them as treats for my grandchildren when they visit.”
Vitamin C Deficiency (Scurvy) – a reminder never to assume!
- Last of all I want to boast of a diagnostic triumph of mine, made without any computer aid.
A LOL in NAD (Little Old Lady in No Apparent Distress) came to our general practice office on a rather busy day with the complaint of frequent falls. She needed a complete examination, but I was daunted by her many layers of clothing (it was winter-time). Undressing and dressing again, even with the help of our excellent assistant, would have delayed the day's appointment sequence to a crawl.
“I notice, Ma'am, that your house is on the road between my home and our clinic. Would it be possible for me to drop by tomorrow morning, say half-past-seven? Perhaps you could stay in your nightgown, so I could check out your heart more easily?”
She agreed, said she'd leave the door unlocked. Next day I entered the house and called her name.
She called back “I'm still in my bedroom, doctor , come right on up.”
So up I went, and damn near broke my neck.
Though Isabel has a checklist for falls frequently, her list of diagnostic possibilities does not, as yet, include Loose Stair Carpet!
Well, time to sign off,
The Isabel project was inspired after my own daughter’s near fatal misdiagnosis in 1999. Last week I experienced my own misdiagnosis. Although far less serious, the episode is probably as instructive for the lessons it teaches. It showed me how easy it is to happen, how it’s often nothing to do with lack of time but just down to sloppy thinking. The discipline of doing a differential diagnosis (even for what seems a blindingly obvious, straightforward diagnosis) is crucial to avoid the sloppy thinking traps.
It was the last day of my vacation, skiing with the family. In the white out conditions we had experienced most of the week, we started down a run that the day before had had wonderful snow but this time had become slush that had turned to ice. My left ski got knocked off as I hit a lump of ice and I tumbled over. As I fell, I put my hand down and my ring and little fingers must have taken the brunt of the fall. It was extremely painful and I felt slightly nauseous, reminding me of the feeling I got when I fractured my knee a few years earlier. By the afternoon, when we were starting to travel home, the hand had completely swollen and was about 30% larger than my good hand.
We arrived home very late so I went to see my family doctor the following day to have it checked. I wasn’t in continuous pain but with the amount of swelling that hadn’t reduced with ice and anti-inflammatory drugs, it seemed sensible to have it looked over. The doctor asked what had happened. His first question that was also asked by all the subsequent doctors I saw was “where had I been skiing!”
He seemed to examine it carefully and thought that I had fractured the bone in the hand at the 4th metacarpal (ring finger). I agreed that it hurt there but said that it also hurt on the soft area (dorsal tenderness) just around the 5th. He referred me for an x-ray that was carried out shortly afterwards at the attached small community hospital. The radiologist and nurse practitioner there both agreed that I did have a fracture of the 4th metacarpal. I asked to look at the x-ray and was impressed when they showed me the fracture as it seemed nothing more than a slight difference in shading and difficult to spot. I felt reassured by the diagnosis and went home.
Two days later the hospital called to say that a senior radiologist had reviewed my x-ray and that I had, in fact, dislocated the 5th carpometacarpal joint. The orthopaedic surgeon at the main city hospital said that I should come in as soon as possible as the joint had now been dislocated for 3 days; I wasn’t to eat anything and should expect to be operated on that night. This came as a bit of a jolt! When the orthopaedic surgeon at the main hospital examined my hand, (after establishing where I had been skiing!), he pointed out that my little finger was considerably lower relative to the ring finger than it was on my other hand. Wow! that was so obvious now it was pointed out, how was it missed?
I have been chewing over this episode over the last few days and a number of points have struck me:
- I was angry with myself for not asking the doctors the key question “what else could it be?” We had all fallen into the trap of finding one easy explanation (the fracture of the 4th metacarpal) and been satisfied and not looked any further.
- The doctors said this was an unusual injury and is hard to pick up on x-ray. However, the real clue was the physical examination that showed my little finger was at a different position relative to the ring finger. Examination of both my hands side by side would have highlighted this. The degree of swelling and the tenderness was also a clue as it was more than would have been explained by a minor fracture of the 4th metacarpal. The community hospital that did the the first x-ray was told that I needed an x-ray for a suspected fracture of the 4th metacarpal but not to consider dislocation of the 5th. This meant that when they found what had been suspected they were satisfied and didn’t look for anything else. Even though the x-ray clearly showed the dislocation right next to the fracture site and the misalignment of the finger joints! There is a very apt adage in radiology that says “what is the most commonly missed fracture on x-ray?” Answer: “the second one”.
- Although we often attribute diagnostic errors to the pressures of time, this mistake was not due to lack of time as neither my doctor nor the radiographer or nurse practitioners were particularly rushed during the consultations.
- I concluded that the only thing that could reduce the chances of this type of relatively simple mistake happening again was if my doctor was forced to record in the medical notes a full differential diagnosis of possible causes or maybe even 3 possible diagnoses. This would have acted as a thinking trigger and reduced the chances of the classic premature closure trap. By ruling in and out the diagnoses on the differential list, he would have been more likely to reach the two diagnoses I was eventually diagnosed with: (i) fracture of the 4th metacarpal (ii) dislocation of the 5th carpometacarpal joint. Alternatively, or in addition, I (the patient) could have acted as the thinking trigger by asking the doctor what else this could be.
In the end this was, I admit, a fairly minor medical error but if dislocated joints are left unrepaired for several days there can be long-term implications. However, because it was so minor and such an easy a trap to fall into it seems more important to examine it closely and try and understand how these minor mistakes happen and can be avoided. Their minor nature may make them seem not worth addressing when we should be focusing on life threatening diseases. However, the root cause is the same so, if we can fix these seemingly minor events, then the serious ones will be fixed as well.
Over 45 years ago, in 1968 Dr Lawrence Weed published an article on Medical Records that Guide and Teach which described the problem-orientated medical record (POMR) where organized problem lists and medical records are critical to clear decision making. The POMR was implemented worldwide and became a standard for medical documentation.
In 1971, Dr Weed spoke at the Grand Rounds at Emory University where he discusses the POMR approach:
- How Physicians cannot be expected to analyse the volumes of information that exist in a single patient’s paper medical record
- The practice of medicine is the way you handle data and determines what you think. What happens over time is the structure of the data determines the quality of the output.
- To analyse a patient record would take around 3 hours but even then its only as good as the information entered at the time and due to the variety of people involved in the care of a single patient the problem is huge. Information within the medical health record is separated into individual sections so it is impossible to see a timeline across all the data of what is happening at a particular point in time.
During the 1970’s Dr Weed lead an effort to develop an electronic version of the POMR. In Lee Jacobs, MD 2009 article Interview with Lawrence Weed Dr Weed describes how the development of the POMR then lead to the importance of integrating detailed patient data with comprehensive medical knowledge. Due to computer technology making available voluminous amounts of data which couldn’t be processed completely by the human mind he realized that medicine must transition to utilizing information technology to provide knowledge and processing capacity to the detailed patient data it receives. This information is then worked through by the Physician to rule in or rule out diseases, treatment and to evaluate the patient and their current health situation.
In 2011, Dr Weed and his son, Lincoln Weed published their book Medicine in Denial. In Terry Graedon’s recent editorial "Is Larry Weed right?" he states:
“According to the Weeds, misdiagnoses “are not failures of individual physicians. Rather they are failures of a non-system that imposes burdens too great for physicians to bear.” They argue that software tools should be employed first. Software linked to the medical evidence base could present a true list of probable diagnoses for physician and patient to consider together, rather than the ad hoc list of differential diagnoses that a doctor may construct based on his or her particular interests or specialization.
This scenario may sound like science fiction, but even science fiction aficionados have noticed that computers are gaining a diagnostic edge over many doctors. In research from Indiana University, the computer demonstrated greater accuracy in diagnosis than the human physicians. IBM’s Watson and the diagnostic checklist software Isabel are also being tested and refined to facilitate accurate differential diagnoses. Computers don’t suffer from sleep deprivation or distraction, and they shouldn’t display the kinds of unintentional biases (based on gender, ethnicity, or age) that beset human doctors.”
It may seem that over 45 years later Dr Larry Weed’s original vision “We need to better organize our records, better utilize paramedical personnel and appropriately use computers” is now available. Mark Graber, MD makes the comment at the end of Terry Graedon’s editorial that as systems such as Isabel are now available integrated into the clinical workflow that utilisation will increase but only with the realisation that Physicians need to be using tools such as these to enable them to cope with the increasing demands of their jobs and not to just rely on the limits of their own minds which have to process great volumes of information in ever decreasing time limits.