Critical care is an enormously complex system of care to render. Its complexity is from how sick and complicated patients are but this complexity is also spread across all facets of care. One of the roots of this complexity is variability. This variability is harbored in great differences in how individual patients respond to severe illness both in how sick they get and in how they may or may not respond to treatment. There is also variability in how critical illness is treated and this in turn occurs at multiple levels starting with how such care is initiated, maintained, and ended. There are big differences in how individual critical care physicians practice. There are differences in how patients, once they are identified as critically ill are started on appropriate treatment.

One factor has emerged in the care of the critically ill is that time is of the utmost importance: it is manifestly essential that critical care be started as soon as possible. Delays in care are associated with more severe illness and increased mortality. It is imperative that we catch such patients early and indeed, delays in the identification of severe illness and the institution of definitive care is an area that needs particular focus.

It can be difficult to identify such patients early in their illness because key signs of how sick they are or may become are often subtle and can easily be overlooked. The failure to identify and act in a timely manner can be catastrophic. Critical care illness is an absolute emergency. This also fits into this idea of variability that results in differences in how such sick patients are treated. One physician or institution may be especially good at getting the critically care patients the quick and definitive care they need whereas another practitioner may miss some of the subtle findings that herald severe illness. An institution may not have fostered a good communication system between the emergency room staff and the critical care physicians so that there delays result in both the identification and treatment of a critically ill patient. Such delays in care can also happen in the hospital when a patient is on a general hospital floor and gets sicker. We all have heard horror stories of patients languishing on the hospital floor or in the emergency room.

The variability in the care provided between institutions can also be great. Most of the care in this country is provided by small to medium size community hospitals. Nearly all the research occurs at major university teaching hospitals which only account for about 10-15% of the care in this country. There is also a separation between practitioners at teaching institutions and those at community hospitals. Those at teaching institutions tend to forge their entire careers in academics whereas those in the community likewise forge theirs in smaller hospitals and private practices. Needless to say, there are also variations in the culture, philosophy of care, and availability of resources between these large academic institutions and smaller community health systems.

One review article published within the last few years in the New England Journal of Medicine by two pulmonary and critical care physicians outlined an aggressive program for removing patients off of ventilators (McConville JF, Kress, JP. NEJM. 2012) . For patients in the intensive care unit on ventilators, half or more of their intensive care stay is spent trying to get them off ventilators. So this is an important issue in our field. These academic authors advocated a program in which they were very aggressive in removing the ventilator. For one thing, they took patients off of the ventilators during the night as long as certain parameters were met. This practice likely results in higher than expected failure rates  with more patients having to be put back on ventilators, which the authors openly point out. They justified the suspected higher failure rate surmising that this offset the risk of further and worsening illness associated with continued ventilator support.

There are several points that merit consideration. One is that there is medical literature supporting that those patients who fail being removed off ventilator successfully not only do worse but this worsening is associated with a failure to be removed from the ventilator and having to be put back on. Getting this right makes them better, getting this wrong can make them worse. Another point is that these practitioners are at an institution in where there is ample support to deal with a patient who had failed to be removed from the ventilator and has to be put back on even if this is in the middle of the night. This typically involves an anesthesiologist or a healthcare professional with anesthesiology skills re-inserting a breathing tube. Many community hospitals do not have that same degree of support as larger teaching institutions especially at night.

There is medical literature supporting that there higher success rates of patients being removed off of ventilators when this process is started so that the tube can be removed early in the day. At smaller community hospitals, there are more available staff around during the day to closely follow the patient and act to maybe avert having to place a patient back on the ventilator and if when the patient does need to be placed back on the ventilator,  then someone is  around to do it. There is much less staff around in the middle of the night immediately available.

The feasibility of such a program in a community hospital is suspect from a logistical perspective where there is not the resources or support that a big university system has. The application of such a program whole scale would probably be disastrous with many patients failing and having to be put back on ventilators. It’s not that such a program is untenable. At large academic teaching institutions where there are residents, fellows, and even attending physicians available even in the middle of the night, it may be feasible. Whether this actually would result in better care presumably offsetting the increased risk of getting sicker the longer otherwise patients  would need to be on the ventilator would to be studied. The point is that there are large differences in how medical care is delivered across the healthcare spectrum and this variability is a barrier developing and initiating performance improvement initiatives that can successfully undergo wide scale implementation.

I will go further and say that much of the research in our field is severely hampered by all of this variability. In medical statistics, this variability is more commonly referred to as heterogeneity. It is difficult, if not impossible, to find out what works and doesn’t when the care is all over the map. There needs to be more consistency in our system care for us to optimally find ways to improve it. How to go about reducing this variability is also very complicated. One obvious place is to start with physicians and from the most rudimentary perspective: the processes of how we see and treat patients and this is a whole other topic.

There has been much focus recently on the ICU and with good reasons. It is the most expensive care we provide to patients and represents care rendered to the sickest of sick. These patients are the most complicated medically and caring for them is likewise also complicated. Taking care of ICU patients requires multiple health care providers coordinating care within a complex setting. It mixes advance technology, complex medical procedures, advance life support, state-of-the-art monitoring systems with good nursing and doctoring skills. All this needs to be coordinated well for an ICU to operate efficiently and effectively and make the most out of saving lives. However, despite all the technology, in the end, it is people who provide the care.  Critical care medicine is a complex system that requires a substantial human element. For such systems to work well, efficiency is of key importance along with systems that promote good communication.  Unfortunately, a lot of ICU care is neither.

There are basically two models current medical intensive care units operate under and these are based upon which physicians have direct care responsibilities. The open model ICU is set up so that any physician can admit patients to the ICU and care for them. A critical care physician may also be involved but the idea is that all the physicians involved co-manage the patient. This is how most of ICUs operate in the United States. The primary care physician, specialists (sometimes many of them) are all following the patient and ordering tests and procedures. If a critical care physician is involved, he or she will also be caring for the patient and ordering tests and procedures. In addition, ideally, the ICU physician will be coordinating all the care provided by all these other doctors. This may seem like a good system; all physicians co-managing the patient in a collaborative environment with each one bringing their own expertise and experience. Unfortunately, the reality of how this system operates is much less than ideal and, in fact, it is quite dysfunctional.

The closed model of how an ICU operates is based upon the critical care physician who may also be called an intensivist as the main care giver. Typically, this means that this physician is the one ordering tests and procedures. Other physicians can still see the patient but they are restricted from ordering tests and procedures and in some instances even kept from directly consulting other physicians. Primary care physicians and other consulting physicians are still involved but only are permitted render recommendations. It is ultimately up to the critical care physicians to follow through on the advice and recommendations of these other physicians. All other health care providers must work through the critical care specialist. Another important aspect of the closed system is that the critical care physician is in the ICU for a full work day (and longer if necessary). Typically, this physician has no other duties like seeing patients in the office, or going to other hospitals. This is protected time. It’s easy to see how this system may improve efficiency and it is this model that is advocated over the closed model in care for critically ill patients.  Unfortunately, only a minority of ICUs operate under this model.

There are several problems with the open system. One issue allows any physician to render treatment to a critically ill patient regardless of whether this physician is qualified or experienced in the care of such patients. Critical care medicine is a highly specialized field. Physicians who are specialist in critical care undergo years of training and acquire extensive experience. Relegating the care of such sick patients to health care professionals who neither have training nor experience in the care of the critically ill is questionable. If there happens to be a critical care specialist involved, he or she may also have other obligations beyond caring for patients in the ICU. Many critical care physicians are also lung and sleep specialists. They may have to see other patients in the hospital or in the office in addition to patients in the ICU. Like all the other physicians on board, they come to the ICU see their patients, write their notes and orders and then leave. One physician may write an order or procedure not known by the other physicians or many physicians may be writing orders that are redundant or in conflict with other orders. In the setting of high patient complexity, things get mixed up pretty quickly, with the right hands not knowing what the left hands are doing so to speak. Efficiency drops, patients end up getting tests and procedures that may not be necessary, much energy and time is squandered in clarifying and correcting orders, and patients can be exposed to otherwise avoidable risks in tests that may not be needed, and all in the end, jeopardizing care. In short, there is redundancy, inefficiency, and little coordination of care.

Another big problem with the open system is communication and continuity of care. Ideally, all these physicians involved should communicate with each other each day. This rarely happens to the level it should with each health care provider tending to their busy schedule and only seeing ICU patients on the fly. All these providers are also coming by at different times throughout the day. Critical care medicine can be fast paced with patients capable of deteriorating quickly. What one physician may see in the morning is not what another physician may see in the afternoon. Again, communication would help ameliorate this issue but physicians don’t talk to each other enough. There are too many cooks in the kitchen all at different times doing their own things and not talking to one another.

The closed model system has many advantages over the open model. It places the care of such sick patients directly under a physician with training and experience in dealing with the critically ill. It relegates most or all of order writing and procedures to this individual. The ICU physician’s duty is to be in the unit with no other outside obligations making it easier to coordinate care by other specialists. Restricting order writing, forces physicians to communicate with each other and especially with the critical care specialist. A physician may want to obtain a CT scan of the brain but the patient may not be stable enough for transfer to the scanner. The critical care physician can weigh in on the necessity of such testing and veto it until the patient is stable enough to be safely transported. If a patient deteriorates or is especially sick, there is someone there or readily accessible who has training and experience in treating critically ill patients.

There are many reasons why more ICUs are not closed. This is actually a very complicated issue. There is a shortage of critical care physicians for one. There are not enough to go around and staff ICUs under the closed model. Not all ICUs have enough sick patients to support an on-site dedicated critical care physician. A small community hospital with a four bed ICU is not enough to justify a closed system and hiring a dedicated specialist. The patients in that four bed unit are likely going to be not that sick compared to a larger health care system. Many physicians who are not trained nor have the expertise in critical care medicine may be against a closed model and perceive a loss of autonomy in giving up their ability to write orders when they have always been permitted to do so. These are some of the issues related to closing an ICU. Clearly, in those settings where there is a need along with the resources available to permit such staffing, the closed system is the way to go.

In the process of making a correct diagnosis we want to make sure patients do not have other disease(s) by excluding specific diagnostic possibilities. This is in part done by ordering diagnostic tests. The results of such testing are said to rule in or rule out diagnoses. A simple example of this is a patient who shows up to a doctor with chest pain. Diagnostic testing in the form of an electrocardiogram (EKG-and electrical map of how the heart is working) and laboratory tests that are indicative of a heart attack are obtained and depending on the results of this testing, the patient is said to either rule in for a heart attack if the tests are positive or rule out if the tests are negative. Obviously, in the case of a patient with chest pain, it is very important to diagnose a heart attack and it is just as important to make sure a patient with chest pain is not having a heart attack. This paradigm of either ruling in or ruling out diagnostic possibilities has become a strong tenet in medicine.

There is obvious utility in this process. It is reassuring to patients, families and health care providers to make sure someone does not have some god awful diagnosis and, god forbid, if they do, it is important to know this. It is also imperative to obtain as much of a definitive diagnosis as possible in order to treat a patient. There is a lot of emphasis on diagnostic testing in medicine both on the part of practitioners and health care consumers.

This all would be straight forward if ruling in or out a particular diagnosis was a sure thing; if this testing as part of the diagnostic process was perfect. If the elevation of cardiac enzymes and a positive EKG meant 100% that the patient was having a myocardial infarction.  But nearly all diagnostic testing is imperfect and has pitfalls. What is more, a positive test or finding in one patient may be unimportant or mean something else in another.  And most times, it is just not one or a few diagnostic tests that are run, but many with the idea ruling out many possible diagnoses. This shot gun approach is the norm rather than the exception.

It is typical for a patient presenting to an emergency department with symptoms of chest pain or shortness of breath to have a myriad of diagnostic tests ranging from ones for the evaluation for a heart attack to a tests for blood clots in the legs or lungs. These tests could include an electrocardiogram, a chest x-ray, possible Doppler Ultrasound studies of the veins of the legs, and a spiral CT scan of the chest-the last two looking for the aforementioned blood clots. Additionally, blood work would entail a complete blood count which may show anemia (low blood) or an increase in the white blood cell count possibly indicating an infection like pneumonia. Electrolytes looking at sodium, potassium, chloride, bicarbonate, and glucose in the blood, along with blood work indicative of kidney function will be routinely ordered. An arterial blood gas may be obtained looking blood pH, and the concentrations of carbon dioxide and oxygen in the blood along with another blood test, if elevated, that would be supportive of heart failure (BNP-brain natriuretic peptide). Tests of liver blood tests and of the how the blood clotting system working may also be ordered.

What may seem at the surface and initially to be a logical, straight forward and streamlined process becomes very complicated. As in the above example, there can be many diagnostic tests generating numerous data points each of which have to be accounted for and interpreted in the context of a single patient encounter. And some of these tests are not simply just positive or negative but involve interpreting multiple findings over a broad range of possibilities as is the case with interpreting chest x-rays or EKGs.  Given all of this, some of this data will be normal and some maybe abnormal. Some of what is abnormal may be important and some may be of little or no importance. And of course, these same principles apply to the normal data as well: some of it important and some of it no so important. What the clinician must do is to figure out what is important and what is not. In borrowing from engineering terminology, a clinician must figure out what is signal (important) and what is noise (not important). It gets even trickier because the noise can interfere with the signal; that is, what is unimportant or extraneous can get in the way of what is important. The noise can muck up the signal.

I have left out until now the whole part of getting a history from the patient and performing a physical exam. This is the starting point for diagnosing and treating patients and so should guide the ordering all of those diagnostic tests but this too, not surprisingly, can get lost in all the noise. And we are not done yet. Multiply this by factors ranging from 8 to 20 that make up the number of patients that can be in a busy emergency department or intensive care unit. Mix in multiple health care providers from nurses, to medical students, to residents and fellows, to attending physicians from multiple specialties all running around trying to figure what is important and what is not. Spice it up with health care providers having differing opinions on the best way to diagnose and treat a given patient assuming that they have some semblance of knowledge as to what is actually going on. Throw in pressures to get patients in and out; the tight constraints placed upon health care providers to be time-efficient, cost-saving conscious while at the same time trying to provide the best and most thorough care. Top all this off with the emotional stress of being a patient or health care provider in such an environment; the din of what is important can easily be lost in a sea of cacophony.

There are numerous other factors I have not mentioned that make all this a lot more complicated and add even more noise. Like life, all this is… well…very, very complicated. So, where do we go from here as patients and care providers?  Maybe start at the beginning with the patient showing up and the clinician taking the time to get a detailed history and perform a careful physical examination. This is going back to what all those sage and masterful clinicians had been telling us doctor types in medical school and throughout all the years of our subsequent training: that everything in this process starts with the history and physical.

This may seem simple and obvious but in reality it can be quite challenging. Again, getting those patients in and out and those tight time pressures clinicians face in caring for patients. It may seem easier for a busy and overwhelmed clinician to briefly speak to a patient, do a cursory physical exam, and just check off a bunch of tests on a lab order sheet and then go on to the next patient. We’ll just sort everything out when all those tests come back. This is the shot gun approach and it does not work well.

By proceeding based on a careful history and physical and upon knowledge and experience, a clinician will form a working diagnosis and order testing based on what is carefully thought out to be going on with the patient. This can do more than cut down on the on all the noise but greatly improve the diagnostic precision of testing. The patient can also be started on treatment. As the data comes rolling in, the clinician will have a clear idea about what to do with the results. If the results of testing does not fit well with the working diagnosis, than a re-evaluation can be made and additional testing and treatment can be initiated or changed. Further support of a diagnosis is made if the patient is responding favorably to treatment targeting the working diagnosis.

An important point that is sometimes lost is that tests do not diagnosis disease, people (clinicians) do. All this testing is part of, and an extension of the history and physical. This is about choosing thoughtfully and carefully what to ask for and about fitting data from tests and procedures with the history and physical to form a complete a picture of what is going on with the patient. This is really what being a good clinician is about and aside from doing right by the patients, it can also be immensely gratifying when everything comes together.

World War II fighter pilot aces were not only damn good shots but they also had what is called situational awareness. They could concentrate on the enemy aircraft in their sites in front of them but also have a sixth sense about what was going on around them and if there was an enemy looking to get behind them. They had to watch their 6 along with everything else. When clinicians care for sick patients it is also important for them to have situational awareness. I will again mention the pulmonary embolism issue which is a blood clot in the lungs.

If a health care provider seeing a patient feels strongly that the patient may have a pulmonary embolism, they may want to start treatment immediately. The earlier the treatment is started for this potentially life threatening disorder the better. However, sometimes patients may be too sick to be transported safely and undergo the definitive test for this called a spiral CT scan. If their kidneys are not working up to par, a common finding in critically ill patients, this also is a reason to not be able to get this test. Starting treatment makes sense there if is a high index of suspicion and when a definitive diagnosis cannot obtained in a timely manner. The test can be done at a later time when it is safer.

The first option in treating a pulmonary embolus is usually thinning the blood (anticoagulation) with one of many available drugs. Coumadin is one such commonly known drug that thins the blood but it is not used first because it may take days to get the necessary effect and there are risks that it may worsen the potential for further clot formation when it is the only drug started.  Another well-known drug is heparin. This is given by intravenous infusion and historically this has the drug that has been used when needing to start treatment for clotting disorders that includes pulmonary embolism.

There are several newer agents that in some cases have been shown to be better than that old standby, heparin. These new agents include enoxaparin (Lovenox), daltaparin (Fragmin), and fondaparinux (Arixtra). There are several advantages to these agents. Heparin is usually given by continuous intravenous infusion (IV drip) and has to be adjusted to target a narrow range of blood thinning. These new agents are given by one shot at a time injection, not by continuous infusion, and do not requiring regular monitoring of how thin the blood is as with heparin. Typically with heparin, there is a large (loading) dose given followed by the infusion. There are no loading doses with these newer agents and they are given by scheduled injection ranging from every twelve to twenty four hours. So, one dose given, no continuous infusions, no frequent blood draws to monitor how thin the blood is, and no adjustments to dosing once the drug is started. This all sounds great except there is a big problem with these newer agents.

There is that famous phrase by Donald Rumsfeld, the Secretary of Defense under George W. Bush, about known knowns, known unknowns and unknown unknowns in describing the Iraq War. This all applies to medicine and to especially sick patients. The situational awareness comes in when stepping back and going down the path of starting anticoagulation (a blood thinner) in a sick patient. The known knowns are the known history of the patient. The big risk of all these agents is bleeding and if there is something in the history that places the patient at high risk for bleeding then starting blood thinners may not be such a good idea. This all needs to be thought through before starting one these drugs. And yes there are alternatives to blood thinners in such patients. The known unknowns are those patients that may have a bleeding risk but it may be worth it to start anticoagulation anyway-balancing risk and benefit. The last are the unknown unknowns and in the spirit of really sick patients harboring all kinds of potential problems, even when there is no history of potentials for bleeding, there still may be bleeding when one of these agents are started.

The other unknown unknown is that it is difficult to predict which sick patient may get sicker.  A patient may be admitted to a hospital in stable condition and become unstable and critically ill. When one reads or hears of such patients in the media, the term commonly used is, ‘critical’. Stabilizing these unstable patients in the ICU sometimes requires performing invasive procedures which carry risks of, you guessed it, bleeding. If a patient has a dangerously low blood pressure, they may require placement of a special catheter called a central line. These lines are best placed in a large vein typically either in the neck (internal jugular vein) or in the chest (subclavian vein). These catheters permit measurement of pressure and oxygen inside the veins in the chest that can aid in helping to stabilize the patient. They can also permit intravenous infusions of drugs to treat low blood pressure and which can only be given safely through these large central veins. A patient with a suspected central nervous system infection may need a spinal tap. A risk of these invasive procedures is bleeding and anticoagulation substantially increases this risk. In most instances, anticoagulation would preclude doing invasive procedures.

The situational awareness comes in with bearing all of this in mind when deciding whether to anticoagulate a patient and which agent to use. If a clinician is worried about bleeding but thinks it is reasonable to treat (the known unknown) then this may be okay. They would bite the bullet so to speak and not only watch very closely for bleeding but also have a plan if this complication arises. Again, this requires thinking ahead and anticipating likely problems. If a patient is critically ill or it is thought they may become so, than this also needs to be born in mind in deciding anticoagulation and what agent to use in terms of prior planning to cover for this. Then there are the unknown, unknowns which a provider will know nothing about other than again forcing the issue to plan too for this contingency.

All of these new blood thinning agents are essentially irreversible; there is no way to reverse their blood thinning properties short of letting the drug wear off (enoxaprin is partially but mostly not reversible). Remember these are given at twelve to twenty four hour intervals and it is at least these time intervals that must lapse before their effects are gone. Bleeding with one of these drugs on board can be a disaster. These agents also significantly increase the risk of bleeding with invasive procedures which otherwise could be life-saving. However, there is a solution to this quandary.

Heparin, that old standby, marred with the inconvenience of continuous intravenous infusion, frequent blood draws and dose re-adjustments of the infusion rates is by far a safer agent in instances where there is risk of bleeding. It can be shut off with its effect dissipating in four to six hours but more importantly, it has an antidote, called protamine that can reverse its effect in a matter of minutes.

If an otherwise healthy patient has a blood clot or pulmonary embolism, then these newer agents certainly have a role in the initial treatment of blood clots. Again, such a patient will likely have a very low risk of bleeding with the clinician not anticipating a need for invasive procedures over the short time it takes to initiate treatment. Unfortunately, blood clot disorders tend to occur in more complicated and sicker patients. Having crashing patients admitted or transferred to the ICU with one of these irreversible agents on board really complicates providing life-saving treatment. Ironically, sometimes with starting one of these agents empirically as described above because the patient is strongly suspected of having a blood clot, the patient ends up not having one but still is critically ill. If a clinician suspects a clot and wants to start treatment but there is a possibility for bleeding or if the patient is sick or has the possibility for getting really sick, then heparin is probably the safer choice.