Thomas Falasca, June 22, 2019, physician and author of Physician’s Guide to Better Medical Decision Making: Critical Thinking in Medicine It is encouraging to see Big Data used in other-than-commercial applications. A recent article described “Deep Learning–Assisted Diagnosis of Cerebral…
The new $2 billion, 368-bed, hi-tech Stanford Hospital opened November 17, 2019, prompting thought about technology and health care costs.
Stanford aside, I have deep concerns about technology and accelerating health care expense. The concerns are not with the responsible and specific use of technology but with its irresponsible and excessive use. The influences motivating this irresponsible and excessive use are powerful and diverse.
One set of these influences is economic, wherein new technologies are implemented primarily as an asset in the competition among health care systems. Once acquired
Another set of these influences is cognitive, wherein there is a systematic inclination to excessively demand and to overuse the new technologies.
Finally, excessive use of a technology, once embraced, increases costs methodologically.
In conclusion, the adoption of a new technology should be guided by legitimate need as demonstrated by the existence of a sufficiently large, specific population of pathology. Its continued use should be guided by specific, evidenced, indications. Existence of a technology should not justify its acquisition. Acquisition of a technology should not justify its use.
Dr. Thomas Falasca, Author of Physician’s Guide to Better Medical Decision Making
Thomas Falasca, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
The fearsome nature of pancreatic cancer makes screening and early treatment of this aggressive malignancy an attractive concept. Such screening and early treatment is the subject of a recent editorial in JAMA Surgery (published online August 6, 2019). While the article is optimistic for the future of screening and early treatment of pancreatic cancer, it is less sanguine about the present situation. Importantly, its reservations regarding the present provide special insight into screening’s limitations.
The article relates that the United States Preventive Services Task Force (USPSTF) “found no evidence that screening for pancreatic cancer improves disease-specific survival.”
One screening issue is that even a highly specific test will generate a number of false positives dependent on the number of people screened. If the number screened is large and the disease incidence is not, the number of false positives can be significant. The problem of false positives can be ameliorated if screening is confined to populations with a high baseline incidence of the disease. False positives are harmful when subsequent treatment involves chemotherapy and aggressive surgery.
These considerations could change if the disease incidence increases or if morbidity and mortality of the interventions decreases.
Another screening problem is that the rapid progression of pancreatic cancer makes the window of opportunity for screening very narrow. Patients might need to be screened repeatedly at very short intervals. While slow progression velocity makes colon cancer screening attractive, rapid progression velocity may make pancreatic cancer screening unattractive.
These are important issues regarding the screening of otherwise normal people and it is fortunate that the issue of pancreatic screening has illuminated them.
Thomas Falasca, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
The appeal of “real world” evidence is, compelling. First, “real world” evidence (RWE) avoids the high, and rising, cost of randomized controlled trials (RCTs). Second, RWE avoids the concern that evidence from RCTs might be incompletely applicable to the real world, since conditions in the real world inadequately imitate the idealized conditions under which RCTs are conducted. Third, some of the data for RWE may already be available. Fourth, previously unnoticed correlations revealed in big data, might suggest further scientific hypotheses.
However, RWE entails significant concerns. Data set size and data processing speed, may not correct for data inadequacies. In fact, increased processing speed and larger data size may produce errors faster and with greater apparent authority.
Certainly, incomplete data, edited data, suppressed data, multiply counted data, and agenda-directed data are still a problem. Existing data may not be a gift, as it may have been edited or agenda-directed and thus depart from the desideratum of random collection.
A couple of options for detecting unreliability in data suggest themselves, but these also have problems. The first option involves examining the data’s sources. However, the sources might not always be evident, either innocently or by deliberate obfuscation. The second option involves examining the data itself to detect departures from random selection. However, there are significant problems identifying lack of randomness (Falasca, 2019. Physician’s Guide to Better Medical Decision Making: Critical Thinking in Medicine.)
Certainly, “real world” evidence has undeniable appeal. However, there are associated problems that require intense examination.
Thomas Falasca, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
An instructive case of overlooked breast malignancy is described by Gordon Ownby. (Gordon T. Ownby. Malpractice Case: How Careful Physicians Can Miss Vital Information - Medscape - Dec 05, 2018.)
In this case, mammogram and ultrasound revealed two upper outer quadrant right breast masses in a middle-aged female. The biopsy report noted, “changes suggestive of ruptured cyst.”
The ordering gynecologist did not notice that this comment described only specimen B and that on the preceding page, specimen A was described as showing “morphologic findings of invasive ductal carcinoma."
Three months later, responding to complaints by the patient of symptoms, the gynecologist examined the area, noted a mass, and then discovered the previous oversight.
This article prompted extensive reader reaction, much of it suggesting the ease of losing significant findings in a plethora of data.
Certainly, the incident seems an example of the search satisfaction bias, whereupon a search is terminated once some result is obtained. However, an effective countermeasure to search satisfaction bias is to conduct the search according to a plan and not to terminate prematurely.
Nevertheless, this is difficult to implement when the quantity of data increases explosively. Consequently, the comments of the readership seem justified.
It would be yet another victory for search satisfaction bias to stop looking for other causative factors in the case of this unfortunate patient. However, while examination of the case continues for other causative factors, it important to understand that the benefits of Big Data are realized only with machine processing of the data and may be counterproductive when human processing is involved.
Thomas Falasca, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
The Marik Protocol, an IV cocktail of vitamin C, thiamine, and hydrocortisone has generated much interest in the treatment of sepsis, albeit with little high-quality clinical evidence. In some places, it has almost reached the status of standard of care. However, science requires the suspension of judgment on effectiveness until sufficient high-quality evidence, including independent replications, has accumulated. The lesson of “cold fusion” remains.
Thomas Falasca, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
Praise to Kenneth Langa and James Burke for their insightful article “Preclinical Alzheimer Disease—Early Diagnosis or Overdiagnosis?” in JAMA Internal Medicine, July 8, 2019.
The authors draw attention to current clinical trials testing monoclonal antibody treatments (eg, solanezumab) aimed at decreasing the build-up of amyloid protein plaques in the brains of adults with preclinical Alzheimer disease as assessed with neuroimaging and biomarker technologies.
However, they emphasize, these technologies could identify as preclinical Alzheimer 30% of the US population over 50-years-old, whereas only 31% of women and 23% of men thus identified are estimated to eventually exhibit dementia, a clear example of overdiagnosis.
Further, clinical trials may not use dementia as the primary endpoint, but rather the surrogate endpoint such as a change in cognitive scores over a given time period, certainly not a guarantee of delaying or preventing dementia.
Finally, the consequent overtreatmentof pre-Alzheimer, potentially including PET scans, IV infusions, biologicals, and cholinesterase inhibitors would be associated with its own harms and misallocation of resources.
It is time to tread cautiously and give this potential problem of overdiagnosis the respect it warrants.
Thomas Falasca, June 28, 2019, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
“Appears to be a reliable (unreliable) informant.” These are the words I have found distressingly absent from patient histories presented by medical students.
Medical students need to assess reliability because relative confidence in the available information is critical in medical decision making.
The same is true of physicians and scientists evaluating information in medical journals, although the readers of medical journals can assess the reliability of studies by consulting the conflict-of-interest disclosures by article authors.
Granted, it may be difficult to avoid some conflicts, but the litany of conflicts after some authors’ names is highly disturbing. When there are multiple methods in selecting subjects, defining interventions, choosing analyses, writing the report, and submitting for publication, there are manifold variables subject to manipulation, and readers may have neither the resources to evaluate them nor even all the relevant data. In situations of incomplete information or compromised evaluative resources, the element of trust looms large.
Regardless of an author’s other credentials, I think it unwise to attribute substantial credibility to authors with excessive conflicts of interest.
Thomas Falasca, June 22, 2019, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
It is encouraging to see Big Data used in other-than-commercial applications. A recent article described “Deep Learning–Assisted Diagnosis of Cerebral Aneurysms.” The artificial intelligence (AI) application provided a red overlay that could be toggled on and off on the computed tomography angiogram (CTA) image at any area where the application interpreted a 0.5 probability of an aneurysm.
The AI augmentation of the interpretation of CT angiograms improved radiologists’ performance from 0.831 to 0.890 in sensitivity, 0.960 to 0.975 in specificity, and 0.893 to 0.932 in accuracy.
Such systems are in early development and have been applied so far in radiography, pathology, and ophthalmology. Further advances should prove a great boon to medical decision making.
Park A, Chute C, Rajpurkar P, et al. Deep Learning–Assisted Diagnosis of Cerebral Aneurysms Using the HeadXNet Model. JAMA Netw Open. Published online June 07, 20192(6):e195600. doi:10.1001/jamanetworkopen.2019.5600
Thomas Falasca, June 19, 2019, physician and author of Physician's Guide to Better Medical Decision Making: Critical Thinking in Medicine
Another drug manufacturer is in trouble over connections to the current US opioid epidemic.
Drug maker Insys Therapeutics, on June 11, filed for Chapter 11 in U.S. Bankruptcy Court in the District of Delaware. The company manufactured
SubsysTM, a fentanyl under-the-tongue spray, approved by the FDA in 2012 only for cancer pain.
A month previously, in Boston, a federal jury determined Insys founder and majority stockholder, John Kapoor, along with four other former executives and managers, guilty of engaging in a racketeering conspiracy.
The company paid medical practitioners bribes for prescribing and promoting Subsys, bribes disguised as speakers’ fees ostensibly to educate clinicians about Subsys.
The problem is threefold: the offering of kickbacks by the company, the acceptance of kickbacks by the physicians, and the off-label promotion of the drug for non-cancer pain.
Of course, the racketeering conspiracy charges are meant to deter future companies and physicians from engaging in such activities.
However, off-label promotion needs to be more strictly defined. It does little good to prohibit explicit off-label advertising and then allow drug companies to pay representatives to tell physicians that although a drug has not been approved for an application, it has been used for that application. If a drug is not FDA approved for an application, then that application should not be mentioned in the company’s paid promotions.
Thomas Falasca, June 17, 2019, physician and author of Physician’s Guide to Better Medical Decision Making: Critical Thinking in Medicine Critical thinking is often the neglected stepchild of medical disciplines. The newest test, the newest procedure, and the newest model for medical practice all intrude into medical consciousness, often promoted by shameless special economic interests, and often to the exclusion of limitations and true scope of application. The progress of medicine depends not just on the instruments in the doctor’s bag but on the judgment and ingenuity of the physician.
Thomas Falasca, June 13,
A compelling article in Anesthesiology
These errors exemplify the change blindness described in Physician’s Guide to Better Medical Decision Making. The errors could have been protected against by
Since we are hard-wired to be influenced by such blindnesses and biases, it is reasonable and important to adopt the structural alterations of configuration and sequential attention to help neutralize these influences.
https://www.anesthesiologynews.com/Clinical-Anesthesiology/Article/06-19/Spinal-Administration-of-Tranexamic-Acid-Catastrophic-and-Avoidable/55134?sub=76EAFDF3939A3CCD22857BB2E0F3BB91A96FB5C8A45F937444AC0F92112B225&enl=true&dgid=&utm_source=enl&utm_content=1&utm_campaign=20190612&utm_medium=title
The American Pain Society board, on May 20, suggested to its members that it cease operations in the face of legal liability connected with the opioid crisis and its funneling of funds from pharmaceutical companies, including Purdue, the manufacturer of OxyContin.
Patient advocacy groups and professional organizations have both a need for money and the appearance of impartiality. This makes them attractive targets for organizations with money and private agendas. Patient advocacy and professional organizations have an obligation to the public and to their memberships to be judicious in their associations. Failing this, they should be accountable.
Recently, I read on Medscape the commentary of Dr. Pies about his experience as a patient having an elective procedure by a surgeon who did not present himself beforehand, did not introduce himself, and only gave the patient brief post-op instructions.
Dr. Pies story brings to mind and contrasts with my personal experience as a medical student several decades ago.
It was my first surgical rotation. My small student group was assigned to a hospital, where each of us was allocated to a different surgeon. I was paired with surgeon M.
In the OR, I was scrubbed and standing beside the resident, across the table from Dr. M. Then, Dr. M looked up in my
Thank you, Dr. M. Your comment has been with me all these years.
https://www.medscape.com/viewarticle/913232?nlid=129991_1381&src=WNL_mdplsnews_190528_mscpedit_surg&uac=118506HK&spon=14&impID=1976910&faf=1&fbclid=IwAR2mtXy3e7U-QWpSSC2v8E0UnU1hY0l3CrfqmbQscV_5ymOqmjxDPmuiztYDr. Thomas Falasca, May 28, 1019
Kristen Lee Currier died at the age of 45 from a pulmonary embolism. The “care” provided to her was abominable. It reveals a textbook of failures.
In outline, a first nurse practitioner saw the patient for routine examination and prescription of oral contraceptives.
A first and then second family practitioner subsequently saw the patient for her complaint of persistent cough but failed to include pulmonary embolus, the “diagnosis you can’t afford to miss,” in the differential.
The second FP ordered a chest x-ray, with sensitivity for pneumonia and hiatal hernia, but no sensitivity for PE. There was apparently no contingency plan for CTA or other PE-sensitive testing.
Husband James Currier’s concerns about his wife’s worsening coughing and breathlessness were inappropriately acted upon when the patient was assigned a second nurse practitioner, who maintained the failed diagnosis, did not order PE-sensitive testing, treated the patient only symptomatically, and did not seek consultation.
The patient was then seen again by the first nurse practitioner, who also maintained the failed diagnosis, did not order PE-sensitive testing, treated the patient only symptomatically, and failed to seek consultation. Further, PE should have been in the foreground of diagnosis since this practitioner had ordered the oral contraceptives.
In short, continuity of care was sorely deficient. Additionally, the combination of availability bias and confirmation bias referred to by Groopman as “diagnosis momentum” pervaded the entire process.
Despite the catalog of mistakes leading to Kristen Lee Currier’s death, the involved corporation, LRGHealthcare, strangely contends, “Reasonable judgment was exercised.” The decision errors that took her life were both corporate and human. All were avoidable. The tort law system can address the corporate contribution. Greater attention to critical thinking in medicine will be needed to address the human contribution.
A wrong-patient prostate surgery with serious complications is reported in the Des Moines Register of January 3 & 7, 2019.
Based on a pathology report of prostatic carcinoma, 67-year-old Rickie Huitt had an unnecessary prostatectomy that left him incontinent. The pathology report was correct about the carcinoma, but wrong about the patient. The incident was alternately attributed to a barcode glitch or to human error, which would be more probable and a case of change blindness.
We have described many cases of change blindness in Physician's Guide; and, the ease with which it occurs suggests that many cases go unnoticed. However, countermeasures to change blindness exist in the forms of alteration of configuration and sequential attention. Perhaps one of these countermeasures could have avoided the medical error that injured Mr. Huitt.
Interesting letter supporting high-quality RCTs as the minimum standard for all procedural interventions (Sharma, May 21, 2019, https://jamanetwork.com).
Although the theoretical basis for an intervention may seem compelling, RCTs are still the gold standard because of the existence of unconsidered factors. The high-quality RCT is a safeguard against the insidious unknown unknowns.
This is reminiscent of David Salsburg's citation of the lady tasting tea, who contended that she could tell if milk was added before or after water. The answer: experiment. When theory fails, RCT prevails!
This interesting study by Kristin Ray, MD, in Pediatrics, April 08, 2019, indicates that 52% of children diagnosed with acute respiratory infections (ARIs) during direct-to-consumer (DTC) telemedicine are prescribed antibiotics compared with 31% diagnosed during in-person primary care visits. The antibiotic management was appropriate in 59% of the DTC telemedicine kids vs. 78% of the kids seen in the primary care setting.
This seems no surprise in light of what we know of cognitive biases. These biases affect medical decision making under uncertainty. Reducing critical information seems to Increase uncertainty and potentiate these biases.
The critical information reduced here is from reduced continuity of care and physical examination. The operative biases include availability bias, action bias, and novelty bias.
The findings published by Dr. Ray in Pediatrics should be a wake-up call.
While I have written much previously about the publication bias, finally I have some optimistic news possibly signaling a change.
Publication bias exists when studies with positive findings, e.g., that a new drug is effective, are published in preference to no-difference studies, e.g., that the new drug is no different from placebo control. This has been documented in several past studies as outlined in Physician’s Guide to Better Medical Decision Making.
Interestingly, Murray et al published a research letter in the May 14, 2019 issue of JAMA that examined each issue of 10 JAMA Network Journals published between January 1, 2013, and December 31, 2015, They found in these journals no significant difference between publication in these journals and the direction of the findings.
If these findings are true, the JAMA Network Journals are to be congratulated. Hopefully, this signals a change in publication tendencies.
I am a physician and I care about my patients' welfare. I don't retail a commercial service concerned only with the number of such services I can sell in a limited time. The language and mindset of industry and business are deleterious to medicine.
This has been captured in this excellent presentation by Arthur Caplan.
https://www.medscape.com/viewarticle/911562
An interesting research letter in JAMA Network describes how falsified data makes its way into meta-analyses thus contaminating them and changing conclusions. The article cites as an example, apixaban studies. Thank you, investigators.
