Design and staffing of intensive care units

The ICU is a specialised area where the sickest patients and the expertise to manage them are both concentrated. The patients are high-maintenance, as they have the need of special beds, numerous large machines which keep them alive, as well as towering stacks of drug pumps and complex monitoring equipment. All this stuff needs lots of electrical outlets and an uninterrupted power supply. Moreover, the expert staff who look after these patients are also high maintenance and require offices, break rooms, family conference lounges, overnight on-call beds,  education facilities, and an uninterrupted coffee supply. Clearly,  to meet the demands for both of these groups of organisms an artificial environment (functionally equivalent to a terrarium) must be constructed, which is complicated and expensive. 

Question 7 from the first paper of 2012 asked the candidates to "assist in the planning and development of a new level 2 intensive care unit". The college answer was long and reached to great depths, referring to some bureaucratic policy document with which the candidates were expected to be at least "broadly familiar". That document was IC-1, or "Minimum Standards for Intensive Care Units". It can be found along with numerous other such publications in the CICM "Professional Documents" page, where the college hide their various policies and standards. There, one can find guidance on such diverse topics as bullying, harassment, health of asylum-seekers (one paragraph), and how to gracefully transition into frail old age while still clinging on to one's FTE 1.0 staff specialist position. On a related topic, Question 6 from the second paper of 2021 asked the trainees to outline their approach to implementing a savage on-site weekend and evening intensivist roster. 

To answer Question 7 from the first paper of 2012, one might make notes based on the detailed college answer. Or, one might read the "Minimum Standards for Intensive Care Units" policy document mentioned above. Or, one might read the very first chapter of Oh's Manual ("Design and organisation of intensive care units" by Vineet V Sarode  and Felicity H Hawker). These options would be equal in their usefulness. For example, all three resources use exactly the same wording throughout their descriptions of Level 1 2 and 3 ICUs. With so much duplication, the time-poor candidate only needs to read one (ideally, the Oh's Manual chapter - because surely in the design of an ICU one would want to apply more than just the "minimum standard").

For the vast majority of people to go beyond this minimum level of preparation would be a waste of such resources as could be put to better use in feeding pigeons or observing the behaviour of ducks in a pond. However, if one is in need of greater detail, one may wish to explore the following:

• "Guidelines for intensive care unit design" by Dan Thompson et al (2012)
• Excellent article by Mahbub Rashid (2006), which discusses the various design aspects and the rationale for why some designs are better than others.
• The ESICM "Recommendations on basic requirements for intensive care units (2010), from which much of the CICM document seems to have been sourced (often, lifted en bloc in vast swaths)

Classification of Intensive Care Units

In summary:

• Level 1 ICUs provide basic critical care (ventilation, vasopressors, haemodynamic monitoring), and may be staffed by non-FCICMs
• Level 2 ICUs provide "general critical care" and are staffed with mainly FCICMs, but have no subspecialist support (eg. no ENT, no interventional radiology, ECMO, cardiothoracics, etc)
• Level 3 ICUs have everything under the sun, supposedly.

Intensive Care Unit

An ICU is "a specially staffed and equipped, separate and self-contained area of a hospital dedicated to the management of patients with life-threatening illnesses, injuries and complications, and monitoring of potentially life-threatening conditions". This line from the college document is also reproduced in Oh's Manual.

High dependency unit (HDU)

As separate from their statement on what an ICU is, CICM have Guideline Document IC-13 to explain to us what they mean by "high dependency". A High Dependency Unit is "a specially staffed and equipped section of an intensive care complex that provides a level of care intermediate between intensive care and general ward care." The following recommendations are made regarding their design:

• They should be attached to the ICU geographically
• They should be attached to the ICU operationally (i.e. the ICU should run them)>
• An intensivist should look after the patients admitted there
• A FCICM should be the director and at least one FCICM should be on the roster of staff specialists
• A HDU should have a minimum of 4 beds

These places are dangerous to the CICM trainee. Allow your roster to dip too deep into a HDU, and one might find the college extending the duration of your core training. Conceivably, one might have to do several extra years purely because the hospital where you work has a high volume of elective surgical turnover, requiring large numbers of ICU staff to be diverted to looking after high dependency patients.

Level 1 ICU

Judging by the college document, this is supposed to be essentially a HDU. The Level 1 ICU is usually found in a small rural or regional centre, which (in Australia) may be about 350km from the next highest level of care. The population serviced by such a unit may consist of 25,000 people, and may be dispersed over a vast terrain which shares many features with the surface of Venus. Alternatively,  Level 1 ICU may be found attached to a small metropolitan private hospital, where it is used as an extension of the theatres and recovery room. Actual "intensive care" (such as non-invasive or invasive mechanical ventilation or invasive haemodynamic monitoring by arterial and central lines) can only be provided for several hours - presumably, after that you need to get the patient retrieved to a real ICU.

"Immediate resuscitation and short term cardio-respiratory support" is what this sort of unit can offer, with the expectation that the person managing it has some experience with such techniques. Locally (and according to rumors) such places are usually run by anaesthetists respiratory physicians or cardiologists, i.e. medical specialities who have at least some idea about the haemodynamic effects of vasoactive drugs and NIV. The after-hours cover typically consists of disinterested locums or SRMOs trying to get in to other specialities. Little education or research activities take place,

The college document calls for the following minimum standards:

• If nobody else, then at least the director should have a CICM fellowship. The operative word is should: the college do not make this mandatory. The only thing they must be is "a registered medical specialist". With such solid standards, a psychiatrist could easily take control of a Level 1 unit.
• A specialist roster with 1 in 3 on-calls.
• Somebody medical (eg. intern or resident) in the ICU at all times.
• At least half of the nurses in any given shift should have some sort of postgrad ICU training

Realistically, these place have:

• Little medical cover after hours (possibly, nobody on site and consultant on call)
• 1:2 nursing ratios
• NIV (CPAP or BiPAP) machines
• Perhaps a transport ventilator (eg. an elderly Oxylog) for short term invasive ventilation
• Arterial and central line pressure transducers, which offer at least the promise of vasoactive substance use (whether the nursing staff will be comfortable with this is another matter)
• Usually, no dialysis machine of any sort

What sort of patient can you have here? HDU patients. The college policy document describes them as:

• Patients with uncomplicated myocardial ischaemia
• Post-surgical patients requiring special observations and care
• Unstable medical patients requiring special observations and care beyond the scope of a conventional ward;
• Patients requiring short term mechanical ventilation.

So, it's a HDU. The college allows for the possibility that a FCICM-qualified intensivist might be managing the patient, and - if that were the case - "provision of mechanical ventilation and simple invasive cardiovascular  monitoring for more than 24 hours is acceptable". However, it must be pointed out at this stage that such units are classically managed by people who are not fellows of the college. And yet, these people go around calling themselves intensivists. So little protection is afforded by CICM to its workforce that essentially anybody can walk in off the street and start fiddling with the ventilators. One wonders as to how much of the current Australian ICU specialist job shortage is caused by people who can't intubate, nesting in peripheral centres and claiming a salary as an "intensive care specialist".

Level 2 ICU

A "high standard of general intensive care" is supposed to be on offer in a Level 2 ICU. These places can be found in large regional centres of Australia ("base" hospitals) as well as medium-sized urban hospitals with bed numbers in in the 250-400 range.

According to the college document, the ICU in such a hospital should consist of at least six beds. It is unclear how the college arrived at this number, or how six beds is superior to five or seven as a threshold for being able to "adequately discharge clinical and teaching functions".

The presence of several basic medical and surgical specialties makes for a more diverse practice, and therefore for a greater diversity of complications. In order to deal with these, the ICU must have facilities permitting more extensive and sophisticated organ support, and for a longer term. The college document recommends that such an ICU be exposed to a least 200 mechanically ventilated patients per annum in order to maintain some sort of minimum standard of expertise, of which (presumably) some are intubated. Mechanical ventilation, renal replacement therapy or cardiovascular monitoring should be available for an "indefinite period".

What's the bare minimum for medical staffing?

• The director must be a doctor whose practice is predominantly in ICU, and needs to have a "substantial commitment" to it. What precisely is that, you ask? Well. The college policy document also weighs in with the relevent definitions of "substantial" and "commitment".  They are surprisingly insubstantial.  The director needs to be working on the floor for at least seven weeks every year (i.e. at least 70% of a 1.0 FTE staff specialist) and must dedicate a minimum of 10 weeks to non-clinical tasks (i.e. paperwork and meetings). For the remaining 35 weeks, the college does not have a plan. Presumably, the director can be relaxing in the sun during these weeks.
• There should be at least four specialists on the roster, and the majority should have their FCICM letters.
• Of these specialists, there should be somebody in charge of the ICU at any given time (i.e. either in it or available to come in at short notice). This is different to the Level 1 ICU description, where there is no mention of how available the senior medical staff should be. There is also no mention of ward rounds in the Level 1 description, whereas in the Level 2 ICU the patients are serious enough to merit the attention of a rounding intensivist at least twice every day.
• In Australia, CICM have recommended that an intensivist is available on site during daytime hours and available on short notice at other hours. This is the "Leapfrog group standard" which produces both cost savings and an improvement in mortality (Pronovost et al, 2004).

Other staffing requirements:

• The requirements for critical care specialisation of the nurses are greater (50-75% of the staff) as well as restrictions on how one may use enrolled nurses (ENs), whereas in the Level 1 ICU there are no such recommendations.
• The Level  1 ICU also needs to have nurse educators for every 50 RNs
• The colege document also calls for a secretary, equipment officer, allied health staff such as pharmacists and dietitians, a research coordinator, etc.

Unlike the Level 1 ICU description, to maintain the elevated position of Level 2 the ICU needs to demonstrate some interest in data collection, so that an SMR may be calculated and benchmarking may be carried out.

What's not available? Neurosurgery, ENT, cardiothoracics, plastics, ophthalmic surgery, interventional radiology, rheumatology, immunology, haematology, medical and/or radiation oncology, basically any specialised area of medicine where a successful first-time job applicant needs to be waving around a PhD to even get an interview. The Level 2 ICU does not have access to such refined services and must make external referrals for transfer. 

Level 3 ICU

A Level 3 ICU is a "centre of excellence", swarming with professors and full of expensive equipment.  "Comprehensive critical care" is provided here, for an indefinite period.

The description of the medical director in the Level 3 ICU is no different to the description of a Level 2 director, nor are any additional qualities listed. "Superintelligent sociopath" is not mentioned anywhere. As far as medical staffing goes, each 12-14 bed pod of the ICU requires at least 4.0 FTE staff specialists. All of them should be FCICMs. Beyond these directives, the description of a Level 3 ICU is no different to the description of a Level 2 ICU.

In the document, further mention is made that a single "pod" of such a large unit should ideally be no larger than 15 beds, as this is a sensible number of patients to look after.

Design of the Intensive Care Unit

Historical aspects

Broadly speaking, ICUs arose out of the need to keep post-operative cardiothoracic patients intubated for some time following their sternotomy. An excellent article by Robinson (1966) details some of the early progress in ICU design. Robinson described the establishment of a small 8-bed ICU at the entrance to the general medical ward at the 516-bed Whiston Hospital of Prescot in the UK.  The following issues were discovered during this early experience:

• Convenience of near position of the unit in relation to places like theatres and radiology was found to be "immaterial", as the patients were usually too sick to be moved anyway. Being able to reach all parts of the hospital was more important.
• Unit size should be small to concentrate the expertise and equipment of "real" ICU patients; Robinson scoffs dismissively that "one can only assume that units of forty-eight beds ...admit patients who are not in need of or do not receive intensive care.".These large units had average lenths of stay of only a few days, whereas  Robinson's unit had average stays of 3 weeks. The total patient workload of one year was 168 patients, 38 of whom had died.
• The bed spaces should all be single rooms. The author derisively refers to shared-area ICU design as a "barn full of expensive machinery", and "a most retrograde step". 
• Cross-infection was the single biggest problem in the experience of these early pioneers; ample handwashing facilities and adequate space between beds was cited as a priority of design. Each room had its own supply of drugs and equipment to minimise traffic between rooms.
• Initial bed spaces were 216 square feet (20m²) which was felt to be inadequate; 300 (28m²) would have been better.
• Specific recommendations were made for light intensity (25-35 lumens per square foot) and special attention was for some reason paid to paralysed patients who cannot look away from painfully bright lights. 
• Senior medical staffing in the pre-CICM era consisted of a physician and an anaesthetist whose skills were complimentary. Egotistic wankery was apparently minimal ("deliberations as to who has the ultimate clinical responsibility are Victorian and unrealistic" they said).
• Working conditions were harsh and weird. "It is believed that there is no place for a medical officer's bedroom; if one of the  medical staff is in the unit he is working and has no time for sleep".
• Robinson was in agreement with a closed-ICU policy. "The suggestion that the referring physicians or surgeons could be trained to use the facilities of the unit is ridiculous", he opined.

Evidence-based ICU design

Rashid (2006) presents a study of 19 units each of which had become the recipient of an ICU design award from SCCM and AACCN. Analysis of these units was carried out, and characteristics (positive and negative) were identified. 

Positive characteristics included the following:

• Custom-built facilities
• Private single rooms
• Free-standing patient beds for easy access from all sides
• Handwashing facilities in every room

Negative characteristics included the following:

• Renovated facilities (i.e. "Didn't this place used to be the morgue ten years ago?")
• Mixed-service units (i.e. inefficient general ICUs, as opposed to specialist neuroICUs or cardiothoracic ICUs)
• Poor space use, thoughtless design which does not consider patient or staff movement
• Family space outside the unit, restricted family access

After reviewing the available literature, Thompson et al (2012) published the "Guidelines for intensive care unit design" featuring recommendations which have been absorbed into the CICM guidelines.

How big should the ICU be?

That depends on how loose your admission policy is and how incompetent your ward staff are. The models vary from place to place. In the American civilization, up to 20% of all beds in a hospital might be classified as "ICU", but be staffed by 1:3 or 1:4 nursing ratio and contain reasonably healthy patients. Contrast that with the British model where only 2% of the beds in a hospital are ICU beds and where only the sickest of the sick end up admitted.  O's Manual (p. 4) seems to support the elitist British model. Some points are made in the Oh's chapter:

• For every 100 hospital beds, there should be 1-4 ICU beds
• More beds are required if the ICU also looks after HDU patients (as indeed they should)
• Units smaller than 4 beds are not cost-effective
• Units larger than 12 beds suffer from decreased efficiency
• Units larger than 26 beds are though to suffer from "management problems"

There does not seem any evidence behind the decision as to how large an ICU should be. The larger the ICU, the more permissive the admission practice. In ICU which is too large for its hospital, the concentration of truly sick (ventilated, dialysed) patients will be smaller and the staff will not have exposure to them frequently enough to maintain the skill level required for high quality care provision. 

Open-area vs. single room models

A barn full of expensive machinery has several advantages:

• All the patients are visible, making it possible to observe them all with fewer nurses
• All procedures are in the public view, making education and supervision easier
• It is less expensive to build

However, single rooms are favoured in modern ICUs;

• Expense of construction is though to be outweighed by the savings in cleaning and infection control
• Single rooms offer more privacy for dying and grief.
• There is the advantage of "isolating" staff, preventing them from being used for other tasks and preserving the continuity of observation and care. The implication of this argument is that without single rooms nurisng and medical staff will wander aimlessly and irresponsibly like cattle without a paddock.

Subspecialised ICUs

For whataver reason, the Australian environment has favoured the development of "multidisciplinary" ICUs in most hospitals. Some places with specialised interests may separate the cardiothoracic ICU from the "general" area, and in some other places a "neuro and trauma" pod may develop - but the Oh's Manual chapter speaks against such single-discipline organisation.  Arguments against having superspecialised ICUs are offered:

• With multidisciplinary units there is less unnecessary duplication of equipment and infrastructure.
• Staff "trapped" by a high level of expertise in one area will de-skill in other areas
• Exposure to multiple disciplines builds a more well-rounded specialist, and trainees benefit from such exposure.

However, specialised ICUs have the advantage of being cheaper, because general ICUs have to be able to solve a greater range of problems and would therefore require a larger support infrastructure and a greater diversity of equipment.  The concentration of specialised skilled staff also gives the impression of expertise. Are outcomes are improved? This is uncertain.

Positioning of the ICU within the hospital

Contrary to the gruff dismissal of the importance of positioning seen in the paper by Robinson et al (1966), the position of an ICU is an important conasideration for logistics of patient movement. Recent developments locally had yielded an initial hospital redesign where the acute care services building (Emergency, operating theatres etc) were well over 1000 metres away from the main ICU area, and another 500m away from the blood bank.  This is an extreme example of how logistics could be important. 

Specific features of the bed spaces and ICU facilities

Design of the ICU layout follows a series of stereotypic patterns. The picture demonstrating the main patterns here is stolen shamelessly from an 1986 textbook of hospital architecture by James and Tatton-Brown, which was quoted in the abovementioned Rashid paper from 2006. Of the American ICUs, the majority were of the "racetrack" type with circumferential corridors to facilitate several directions of simultaneous staff and patient bed movement. This also places patient rooms out on the periphery, where they can be exposed to more natural light.

The college document describes (briefly) the following list of mod cons:

• Drug preparation area
• Dirty utility
• Storage spaces
• Staff relaxation facility, i.e. tea room
• Seminar room
• Nursing and medical offices
• Relatives area
• Secretarial area
• Computer facilites
• Cleaners area
• Blood gas laboratory
• Library facilities

Evidence suggests that ICU design can improve the quality of infection control. According to a review by O'Connell et al (2000), putting about 20m² between each bed space is one such design feature (i.e. one does not brush past patient beds and curtains as one make their way around the ICU). Ensuring that there is ample storage space also encourages people to store equipment there, and prevents the corridors of the unit from being packed full of pathogen-collecting boxes and apparatus.

The college has prescriptive rules regarding what the ICU bed spaces should look like:

• 25m² (single room)
• 20m² (bed space)
• Wash basin
• Four oxygen outlets
• Three air outlets
• Three suction outlets
• Four data ports (RJ45?)
• Sixteen power points
• Cardiac-protected power supply

Access to fresh air and natural light is "desirable", and patient privacy needs should be "taken into account", but no firm recommendation is made for either. Allowance is made for older style units, presumably dark windowless dungeons where naked patients are perpetually on display. While making sure their rude bits are discretely hidden, there is an imperative to make sure all the patients are highly visible: keeping them in low-visibility rooms is a recipe for disaster and has been associated with increased mortality (Leaf et al, 2010).

The specific size of the bed space is not exactly a highly scientific determination. Locally, the designers took one ICU bed, positioned all possible equipment around it, and then walked around that equipment. The perimeter described by their trajectories also described the size of the ICU single rooms they asked for, which ended up being about 25m². 

After-hours senior ICU cover

Question 6 from the second paper of 2021 presented the trainees with a situation where the medical director of the ICU has decided to take their 9-to-5 intensivists and make them work evening shifts. "Outline important aspects that will need to be considered in the implementation of such changes", the examiners guffawed. This was an excellent SAQ, as it represents a real and serious challenge to anybody in charge of an ICU, trying to govern a frequently ungovernable group of strongly opinionated people. At the time of writing n November 2021, the college has not yet provided anyone with their model answers, which we should all anticipate with some excitement, as it would be interesting to see what the "official" stance on this might be.

What follows, therefore, is a rambling expansion of the author's own uneducated take on this matter.

Rostering aspects

• Decisions need to be made regarding how the evening roster works, including:
  • How many hours are worked - the hours worked during a week of evening cover should be equivalent to the hours worked during any other week, in order for the salary to remain stable; the exact duration of this evening shift needs to be determined
  • After-hours service provision has implications for time in lieu, leave accumulation, and hourly rate of pay in some services, and this needs to be factored into the department’s staff budget
  • Provisions should be made for staff who work this roster, such that protect them from excessive workload (eg. they should not be expected to start a week in the ICU at 07:00 on the next day, if they finished at midnight)
• A fair system would distribute the evening workload evenly between all clinical staff according to the fraction of their clinical appointment
• The number of existing staff may not be sufficient to cover the increased roster and maintain safe working hours, which means more intensivist staff would need to be hired

Logistic aspects

• Space for the after-hours staff needs to be allocated
• Structures allowing for rest (i.e. rooms for tired intensivists to sleep) and break facilities may need to be expanded.
• If more intensivists are needed to staff the roster, they will need offices, an increase in the administrative capacity (eg. hiring more secretaries), etc. 
• External services (eg. the medical emergency response team, the emergency department, operating theatres, etc) need to be informed regarding the new arrangement so that the existing patterns of patient referral are preserved

Clinical governance aspects

• The after-hours specialist covering the ICU may have to take decisions regarding the care of patients which are admitted under the care of another ICU specialist, eg. in an arrangement where one intensivist provides evening cover for several pods in the ICU. This has medicolegal and governance implications. The craft group of intensivists will need to take a consensus decision regarding this matter, for example:
  • Everybody agrees to manage their own patients in the after-hours period, which means the on-site intensivist is mainly employed as a pair of skilled hands
  • Everybody agrees that they trust each other to manage each other’s patients, which means the evening intensivist takes responsibility for the unit in the evening
• The hospital administration needs to be informed of the decision, so that appropriate channels of communication can be established (eg. the switchboard operators will forward emergency calls to the correct intensivist)

Patient safety aspects

• Clinical handover times will need to be coordinated; the on-site specialist will need to take handover from somebody, and will need to give handover to somebody, in order for the continuity of care to be preserved
• The junior ICU medical staff and ICU nurses need to know whom to contact, and there needs to be clear guidance provided to them
• Rostering evening cover should aim to maintain a specialist-level quality of care in the after-hours period, which has implications for rostering transition year fellows or postgraduate fellows (or senior staff with limited recency of practice). 
• The breadth of cover should comply with CICM safety regulations (section 6.2.5 of IC-1), which recommend that "after hours it is acceptable for one intensive care specialist to manage two such pods [16-30 patients in total]  provided there is an additional intensive care specialist on second call", 

Staff safety aspects

• The evening roster needs to protect the ICU staff from unsafe working hours. Most health services have specific guidelines related to the maximum number of safe working hours for senior medical staff, and these guidelines need to be followed
• A maximum number of evening shifts per annum or per month needs to be established in order to prevent burnout
• A sufficient recovery time should be afforded to the staff involved in the evening roster
• Nonclinical work (for example, committee attendance, education, research or audit activities) should be suspended for the duration of the rostered evening shifts, as it would be unreasonably onerous to expect the specialist to perform administrative work all day, and clinical work all night. 

Change management aspects

• A group of staff may be resistant to this model for a variety of reasons, ranging from totally legitimate (eg. child care arrangements) to entirely preposterous (seniority, for example, mass inertia, or some sort of ideological opposition to shiftwork).
• The introduction of the roster needs to be handled carefully and collaboratively, by:
  • Introducing it in terms of the needs it addresses,
  • Establishing trust for the process by transparently discussing each aspect of it with the entire ICU specialist group,
  • Building consensus for various elements of the process (involving the more resistant participants to take decisions on various minor aspects of the overall process, so that they feel involved and in control but don’t have to opportunity to sabotage anything important)
  • Planing the implementation together as a group, and setting up a clear timeframe for the implementation
  • Communicating the steps in this process widely, including senior hospital administration, junior medical officers, and nursing staff
• Resistance is not unexpected, and is a predictable aspect of any change process. One should be patient and firm. It would be wise to prepare to at least rebut some of the more asinine arguments, such as the complaint that shiftwork is somehow unbecoming of senior specialist medical staff. Emergency physicians routinely work shifts, including evening and in some cases nights, and they are no less professional than the ICU senior staff. Moreover, it is not clear exactly what this objection is aimed at, other than the perception that being on call from home is somehow more prestigious. 

Audit and monitoring

• This change, it clearly did not arise from a vacuum. It is never a particularly popular decision, and is usually driven by necessity. Specifically, when a department starts extending the rostering of senior medical staff, it is usually a reaction to a serious adverse event (or example a patient’s death) or the finding of some sort of critical failure of after-hours trainee supervision or service provision (eg. the hospital noticed that none of the on-call intensivists are answering their phones). Either way, it is a reaction to a some kind of thing. That thing, then, becomes the main outcome measure of the audit and monitoring process, i.e. the department observes the variable of interest following the implementation of the roster, and demonstrates to society that they are comletely reformed and everything is back in normal working order.
• In the unlikely event that everything is already working well and the department for some reason decides to change the rostering for some unrelated reason, possible outcome measures to track could include the following:
  • Patient satisfaction (or famly satisfaction)
  • Staff satisfaction (one might like to canvas the other departments as well)
  • Improvement in SMR
  • Impovement in the outcomes of the ICU outreach service
  • Improvement in a specific metric looking at the rate of some specific incident or complication which could be associated with senior supervision, for example procedure-related complications 
• Like with all things audit-y, some kind of regularly repeated pattern of audit and reaudit should be established, with a regular forum to present the findings (eg. some kind of ICU senior staff meeting). 

So. Evening and night rostering for ICU senior staff. Is this a a good thing? Emergency physicians and intensive care physicians are routinely rostered to night shifts in America, Europe, the Middle East, and generally throughout the world. British and Australian intensivists tend to stay on call from home. But is this really better for them? A shift roster can guarantee periods of sleep, whereas a being on call 24/7 for several days in a row guarantees nothing.  On the other hand, the same person being in charge for a sustained block of hours and days results in increased continuity of care, and less information loss through handover.

What does the college say? The CICM document outlining minimum standards (IC-1) only states that the intensivist needs to be directly available during normal working hours. With respect to protecting the intensivist from the brutality of a junior-registrar-like roster, there is no policy statement per se, but CICM do have a welfare SIG document which describes the professional risks of sleep deprivation and offers suggestions as to how one might mitigate against them. The author finds some aspects of this document particularly relatable, including especially the statement that one should protect the hours of 3-6am for sleep (as 90% of Deranged Physiology is written during the hours of 3-6am).

What do other colleges say? ACEM, the college of emergency physicians, actually has a statement on night shift rostering, with some excellent points made to support the evening specialist shift (but not the night shift, which they don't support). Some of these can be extrapolated to the ICU with the addition of a little evidence. 

• 86% of ED presentations occurred between 8am and midnight, which means these hours are the most critical for trainee supervision. For the ICU, admission and referral patterns are also reasonably predictable, and old data from Bhongari et al (2011) suggests that the vast majority of ICU admissions also occur during these hours. 
• In the ED, the majority of patients seen during a night shift are patients which arrived during the evening, but who were not seen until the night because of emergency department overcrowding and understaffing. The ICU equivalent of this would be consult referrals and bed block. Just as in the ED, the increased number of after-hours workload does not reflect intrinsic ICU process deficiencies, but rather systemic problems (eg. the failure of after-hours supervision of hospital ward staff, who resort to making ICU referrals, or poor availability of ICU beds because of hospital bed block). In either case, extending ICU senior staff rostering to encompass this after-hours period would represent a band-ad solution which merely redistributes the responsibility for after-hours trainee supervision from physicians and surgeons, transferring it to the ICU team.
• Other specialities are not available after hours. As much of the most important definitive patient care comes from specialist input (gastroenterologists doing scopes, cardiologists doing caths, surgeons operating, etc), one cannot expect much improvement in the quality of patient management in the ICU (or ED) without these people being around. If the patient needs a relook laparotomy, and the surgeon is not available until the next morning, even a hundred evening intensivists will not be able to improve their outcome.
• Only large intensive care units would be able to justify providing evening or night cover. Smaller units, running with fewer full time specialists, would not be able to expand the roster in this way. To provide specialist care after hours in major metropolitan centres would therefore further exacerbate the rural-urban inequality in healthcare access. If you get sick in Prahan, you'd get Chris Nickson resuscitating you, whereas if you get sick in Bathurst, you would merely get an on-call yartsev.  
• Evening and night shifts, with their effect on sleep patterns and fatigue, would have an effect on physician performance and burnout. Already ICU is a speciality which places substantial stresses on its practitioners, and increasing the level of physical and emotional exhaustion experienced by these people will just lead to early retirement and poor senior staff retention. 
• Senior specialists are not cheap. It remains to be seen whether there is a benefit to patient care from after-hours rostering which could justify the cost of hiring additional senior staff. As an outlandishly extreme example, to go from a 9-to-5 model to a 24/7 three-shifts-a-day model may require a doubling or tripling of the ICU medical specialist workforce. This does not look good for the ICU budget. 

What is the evidence? Some data suggest that the presence of an intensive care specialist is beneficial. For example, Arabi et al (2006) observed patient mortality in a hospital from Saudi Arabia where intensivists are rostered 24/7. The authors observed that there was no major difference in mortality in that ICU, no matter what time of day or night you were admitted there. This is entirely unlike other studies (eg. Bhongari et al, 2011) which demonstrate a clear increase in mortality during the out-of-hours and weekend periods. The authors concluded that it must have been the consistent intensivist input that made the outcomes stable and positive.

Other data suggests that evening or night staffing of ICUs with senior staff does not improve mortality. For example, Wallace et al (2012) and Kerlin et al (2013) did not find any mortality benefit from night-time intensivist cover. Garland et al (2011) looked at something completely different - nonintensivist staff satisfaction - and found that basically everyone hated having an intensivist around at night. Nurses reported more "role conflict"and junior staff reported less autonomy, which basically means that the intensivists compulsively micromanaged everything and stuck their nose in everybody's business. The intensivists, on the other hand, loved the shiftwork lifestyle. The same feedback comes through from reading Wise & Frost (2006). The authors, speaking from two years of experience in a large British ICU, reported marked improvement in their quality of life, arising mainly from uninterrupted sleep.