The Hospital Internet of Things

A wireless communications chip embedded in each device or “thing” is touted as the step that will create the Internet of Things (IOT). How can that affect healthcare and medicine? Can we have a Hospital Internet of Things?

While it would be fantastic to have stents, coils, implants, pellets et cetera with the capacity to sense and report, there is the issue of how would such devices affect the human body in the long term (the effects of electromagnetic energy). But there are other aspects of the IOT that healthcare and in particular the hospital, could use.

Imagine the little bracelet that each patient gets on registration to the ED, and if that bracelet could identify hospital sectors and communicate with an in-hospital patient tracking system. Such a system then wouldn’t have to rely on human updates (which are often delayed or omitted), and transport personnel would be able to easily locate the patient. Also, the swarm of agents (residents, attending physicians, consulting physicians, physical therapists, portable X-ray techs, etc) involved in the care of the patient could modify their rounds* so that they do not go to a ward only to find a patient missing because they’re in radiology, dialysis etc. And after some time, enough data might be generated to allow optimization of transport and subsequently other functions.

We could also have clever medication pumps that notify that the bag of infused medication is almost empty, at a pre-specified interval before the end, rather than having an alarm after the medication is actually over. This would help ensure uninterrupted administration, esp. in cases that continuous infusion is critical to achieve the desired effect (e.g. vasopressors) or pharmacokinetics (chemotherapy).

And the use cases extend to pretty much all other aspects of inpatient care, like vital signs monitoring, medication administration, blood sampling etc. Integration with an EHR that could serve also as a control system might help to cut down on waste and improve response times. The inpatient environment provides a good opportunity for such developments, which could then be transferred to the outpatient environment. There are challenges regarding security of information and function (data breaches and hacking) but the benefits would be very significant if implemented.


* Since most hospitals are labyrinths of interconnected buildings, elevators and stairwells, the problem of rounding route optimization has to be a 3D version of the travelling salesman problem.

Internal Medicine: discordance of training and goals

Much is being said about the necessity for primary care physicians and the importance of primary care, esp. as a point-of-entry to healthcare as compared to the ED. Much is also being discussed about the increasing numbers of people skipping primary care and using the hospital ED for the whole spectrum of non-preventive care, including chronic complaints, urgent care and emergencies.

A fairly common explanation for this seems to be that while a visit to the ED is time-consuming, it is still less time consuming than the cycle of primary care visit – referral – testing – follow up visit. Many physicians in training (students and residents) would agree.

Isn’t it strange though? The average resident in internal medicine gets trained in inpatient medicine and the subspecialties. Obviously, the level of care is only increasing, from the general medical floor to the critical care. Practice in these environments is backed up by the laboratory, the imaging and the subspecialties, all at an arms reach, so these patterns become embedded. Furthermore, emergency medicine training is limited, as low as 1 month in some programs. Primary care exposure occurs either as the “1 week every fourth” model or as a half-day weekly session interspersed in the rest of the rotations; albeit the former is preferable, both systems result in a fractured experience.

Some of the people that will go through this training will choose primary care. The question then becomes, how adequately can they manage urgent care when they have had minimal exposure to the actual urgent care? How parsimoniously (regarding testing, imaging and referring) can they deal with chronic problems? Because it should be obvious that their training is not preparing them for that, but for the role of hospitalist. Is it unreasonable to expect that if you get trained in flying fighter jets you won’t be able to do a good job flying C-130s? Most probably not. It is time to re-evaluate the scope of training in medicine if we want primary care to have a role other than a testing and referral distributor.

The multimedia EHR

For the most part of modern medicine, charts have been paper-based, and communication of physical exam findings has also been either verbal or written. So what do you do when you have to describe images (e.g. skin lesions) and sounds (lung, heart, bowel)? You come up with descriptive terms, which you can define with variable precision (skin lesion v. heart murmur).

This effort to describe and correlate the findings with the diseases is the cornerstone of the medical textbooks, teaching and diagnosis. Gradually however the use of images found its way into textbooks and teaching, and even more recently the same happened for sound clips: medical education made its first steps to the world of multimedia. Rich content is enhancing learning: it is way better to read the description of a murmur while listening to it, rather than while trying to imagine it, or having somebody attempting to reproduce it by whistling, taping, et cetera. Hopefully such multimedia will soon be accompanying any didactic presentation/discussion, esp. given their availability on the internet.

OK, multimedia is changing the face of medical education. What about the realities of clinical practice? In case presentations, how often does doubt exist about the accuracy of the description of that rash or that murmur? And did the patient have wheezing? I think that we should start to augment the EHR by adding the actual findings on top of their descriptions, at least in the initial encounter of an admission. Describe the rash, but take a photo of it; describe the murmur, but record it; the same for the lung sounds. Would it inflate the size of the chart? Yes it would, but not so much, and the potential value would be great: reduced ambiguity and subjectivity, increased comparability and therefore ability to monitor, increased value for later presentations, including consultations and case reviews. Would it create HIPAA concerns? Not if the actual capture media belong to the hospital and upload directly to the EHR, eg. using the medical record number (by manual input or scanning the ID tag) and connecting via a secure network. One would argue that while any smartphone or tablet could be used to obtain images, the sound clips would pose a problem, as not many people use a digital stethoscope. True, though it probably won’t be long since somebody builds a cheap one.

Furthermore, the EHR (in the broad sense) contains the imaging reports and the actual images, which the clinician can review, and they can be exported and submitted for a second opinion. It also contains the endoscopy reports which have images of the areas of interest. Why not then the pathology slides? A scanned image of at least a representative slide should be included with the report. Some might argue that all slides prepared and stained should be scanned to the EHR, and that they should be exportable just like the CT/MRI scan images, facilitating second opinion and telepathology.

So, while I appreciate an elegant description of a rash or a murmur, I know that most often the descriptions are basic, and I feel that the ability to see/hear the finding would improve the ability to assess a clinical presentation.

The medications bag

Electronic systems are nice and convenient, depending hugely on one factor: their usability. If not usable, then the user will in the end misuse them. That’s pretty much what’s happening with medications modules in the systems I’ve had personal experience with: lists are cumbersome to fill, have mistakes, don’t carry over easily, medications auto-expire and then vanish from the list, and of course there’s no interoperability.

Therefore for the needs of medication reconciliation the most reliable source remains the patient’s own hand-written list and/or the medication bag. So then why not create a list, print it and give it to the patient at each visit? Why not create a QR code containing the medications list that could fit in a wallet? It might be an easy solution to have a portable medication list.

Classes and formulations of antiretroviral medications

The task of learning about highly active anti-retroviral treatment is not an easy one, given the combinations of classes of medications, the number of members of each class and the synthetic difficulty of their names, and the various available combinations. It’s of course worse if you have an episodic contact with them and not a continuous one. Therefore I’ve tried to compose a short simplistic presentation about them, following the building block hierarchical logic. I am sharing this presentation here, hoping it might prove useful to others as well.


The EHR is not your smartphone

The message that arrived in my inbox 30 days prior to starting a new clinical rotation was clear: they’re using another EHR down there, and I need new EHR training. Guess what however: only one date is available at the training center and coincidentally it’s right in the middle of one of the busy days of my current rotation.

Fast forward to the training day: predicted duration 3 hours. There are four of us, and we ‘re all anxious to return to our posts. We all have seen this system while walking by the workstations of the ED, but we have no previous experience with it. Apparently, so does the trainer: a non-clinician, that has tried to model the workflow and the needed task and teach it. It is to their advantage that the ED workflow is somewhat more linear that the inpatient one, but still, the lack of hands-on experience shows. The session finished after 2 hours and we all left the place commenting on it being yet another lost opportunity.

Opportunity for what? Well, for efficiency. Regardless of how intuitive is or isn’t an EHR, there’s a learning curve there. Not investing in properly designed and delivered training is a very poor decision. A set of densely spaced brief and task-oriented sessions would pay back significantly on active involvement, by removing yet another thing that the physician has to figure out in the busy rotation. This is the analogy of the post’s title: most people when they get a new smartphone with an unfamiliar OS just start using it and over the course of time become more comfortable. But the basic task in this analogy is very simple and surprisingly standardized among OS’s. See the button/icon with the handset? Task virtually completed. Not so easy with the EHR though, and despite the unparalleled benefits that EHR provides, usability hurdles and half-boiled upgrades can seriously impact the value from its use.

So, the message is clear: well designed and delivered training in the EHR is not a luxury, but an investment with an immediate and lasting return. It is current users at the same site that should present the tasks one needs to perform and how to best achieve that, much like with all other aspects of medical training.

As a side-note: I got to use HMED (HealthMatics ED) by Allscripts. In comparison to SCM (Sunrise Clinical Manager) for acute care by the same company, it appears to be quite better. Perhaps the icon-centered design is making all the difference in usability? Maybe.

Triage via telephone in the UK

The Guardian published a harsh critique of the new helpline of the UK’s NHS, based on “Pathways“, a prioritisation and assessment system.

What does the author think about it? Well, he thinks it’s not working and it’s leading the people to the ED (A&E). The author also summarizes some basic reasons that lead to over crowding of the A&E (boldface mine): “the combination of a risk-averse computer algorithm, call centre staff who don’t want blood on their hands and the scarcity of GP appointments on public holidays, weekends and when there’s good skiing to be had, is driving thousands of 111 callers into A&E to be put on a trolley and ignored like a rum baba in a 70s restaurant.”

The author puts forth his idea of a solution to this problem: “the algorithm needs some backbone… But we need to be honest with ourselves: the whole point of this phone line is to fob people off – to stop them going to hospital, not encourage them. Deep down, they want to be told they’re fine and to stop moaning – otherwise they would’ve dialled 999 – so a new version of Pathways needs to be designed with that in mind…”

How can we check the validity of the critique? An evaluation report is available online and is an interesting read.

Who built the NHS Pathways? It “was developed and is maintained by a group of NHS Clinicians with extensive experience of both urgent and emergency care provision, and also of clinical decision support tools. This includes GPs, nurses, paramedics and many more.”

What’s interesting about the system is its data analytics module, the “Intelligent Data Tool” that “presents data in an easily accessible form to show clinical demand and gaps in service provision through a wide range of reports”.

Imprecision medicine

The term precision medicine has been put forth recently as “a phrase that is often used to describe how genetic information about a person’s disease is being used to diagnose or treat their disease” [TCGA site]. In reality this is an alternative term for the -rather unfortunate- term “personalized medicine“, as described in an editorial in NEJM: “the fundamental idea behind personalized medicine: coupling established clinical–pathological indexes with state-of-the-art molecular profiling to create diagnostic, prognostic, and therapeutic strategies precisely tailored to each patient’s requirements — hence the term “precision medicine.””

But this is not a post on precision medicine and genomics. Instead, I am trying to focus on the imprecision that plagues the everyday reality of clinical medicine, in the form of vague descriptions. “Facial droop” states the HPI, yet it doesn’t report the side of the droop, as if it’s unimportant. “We gave some antibiotics” reports a colleague, “and the heart rate was really high”. “The patient has had a valve replacement in the past” (which one?). “The patient will have a procedure tomorrow, therefore will be NPO”. “Please start ceftriaxone, and give some furosemide”.

The above are examples of vagueness that can actually negatively affect the care of the patient, and that wouldn’t require a tremendous effort to be remedied. So the question is, why are young (and not only) doctors less precise in their communication than the chefs in restaurants and the reporters? If clinical medicine and healthcare are to such a great degree a continuous aggregation and reporting of data and decisions, why do we do such a bad job in communications? Out of all the possible reasons, I would just bring up the absence of focus on this matter: training occurs in real time, and little attention is drawn to the “formal” aspects of the message; people are interested in reaching their conclusions, and don’t have time to dwell on these “minor” omissions, they simply assume that you’re going to improve with practice and experience. Unless we decide that precision is a significant goal and lack of it can hurt patients, the situation won’t improve.

Networking and Interoperability anecdotes

Anecdote 1: So the patient was originally planning to have their elective operation in Upper West Manhattan Hospital. However they ended up in Lower West Brooklyn Hospital and changed their minds: the operation should be performed in the second facility. But wait! We need the already performed imaging in order to proceed, and in fact the actual images, not just the reports. The year is indeed 2014, but apparently the only solution to retrieve the images is to have somebody actually dedicating a whole morning (and perhaps more) to go to the other facility and retrieve the CDs.

Anecdote 2: the patient was very recently discharged from another hospital where they were admitted, and now is being admitted to your hospital. They tell you they had that blood test and that scan you are recommending, but they aren’t sure about the results. Can’t you just call the other hospital to have those results sent over?  Yes, but it might take some time.

It a sad waste of time and resources, and it shouldn’t be so in the age of networks and Health Exchanges and Health IT.

Actual tests and the “raw” data (i.e. in imaging the actual images, not only the reports) should be passed on to the patient on discharge or made available for download and transfer to personal or enterprise EHRs with a simple electronic transaction.

Paleogenomics: archeology, history and biology

The plague that hit Ancient Athens in 430 BC is a well known and significant cause of the Athenian defeat. The cause of the plague per se however is obscure, being subject to hypotheses. There are however human remains from that era:


At least 89 persons were buried in the Athens plague-era group burial shown in the above photograph (from an article in Greek). With the recent advances in paleogenomics and metagenomics one can expect that the identification of a pathogen would be very likely. What would the actual obstacles be? Well, resources, expertise and of course the interest to do it.

Since the Greek scholars would use the former two as excuses for the lack of the third, we would have to wish for some other School/Center of Archeology or Paleobiology/ Paleogenomics to try to do it. I think it’s a question that deserves an answer; hopefully sometime in the near future it will be answered.