The clinical information system (CIS)

These are computerised databases which store and retrieve the ICU medical records. They may integrate diagnostic test review, test ordering, prescription/administration of drugs, and storage/retrieval of imaging studies. LITFL has an excellent summary. Whatever the public opinion of these things, at the time of the article's original date of writing (mid 2014) the inevitable invasion of ICCIS  was viewed as inevitable in NSW in spite of its many crippling flaws (though it was renamed eRIC, presumably to avoid association with the Islamic State of Iraq and the Levant). This tide has now (in 2020) washed over the author's own institution.

Question 15 from the second paper of 2007 asked the candidates to critically evaluate the role of a CIS in intensive care. In order to render answering this question easier, the topic discussion has been framed into familiar rationale-advantages-disadvantages-evidence framework.

As far as reading material goes, one cannot look past Oh's manual, Chapter 9 (pp. 69)  Clinical  information  systems by David  Fraenkel. Fraenkel used to be the president of ANZICS, in case you are wondering. He published a paper in 2003 about the quality benefits of a CIS, and may well have inspired the author of Question 15 from the second paper of 2007. Wherever possible, this paper along with Oh's Chapter 9 are used as the main resources for the ensuing discussion.

Critical evaluation of the CIS

Rationale

• The amount of data recorded for ICU patients increases every year (some say, by ~ 6% per year)
• Effective use of this information relies on the clinician's ability to easily retrieve and search it, so as to make informed decisions
• The increasing amount of information takes a toll by diverting time from patient-oriented workload towards non-patient-oriented documentation and administrative tasks
• The rationale behind the use of CIS is to render easier the process of documentation by automating some routine tasks, and by improving access to the growing patient record
• Much of the data in the ICU (eg. pressures, ECG, ventilator data) is already in digital format, and so it makes no sense that it should then be channeled though a human intermediate to finally come to rest in a paper record.

Advantages of CIS

• Increased legibility and improved searchability of records
• Automation and increased accuracy of repetitive tasks eg. recording of observations and ventilator settings
• May integrate with pathology results database
• Electronic record of prescriptions offers improved and security with electronic signatures and passoword-protected ordering
• Searchable database for audit and research
• Records simultaenously available from multiple points of access
• Built-in error checking, knowledge database and alert systems ensure prescribing is legal and accurate
• Built in spell checker and arithmetic function prevents medication dose calculation errors and spelling mistakes which lead to misinterpretation of orders.
• Theoretically, the digital record should not only be impossible to alter, but should also log all attempts at doctoring (pun  is unintentional)

Disadvantages of CIS

• Expensive to install and maintain
• Difficult to integrate with other existing electronic systems
• Steep learning curve
• Computer literacy among staff influences the degree to which the CIS changes workload patterns(many things previously easy may suddenly become more difficult)
• May degrade the quality of handover and ICU ward rounds (consultants focus their attention on surfing the CIS rather than on what the staff are telling them)
• There are concerns that electronically signed prescriptions are legally invalid
• There are concerns regarding the security of passwords, and the potential of untraceable "doctoring" of the records
• Built in spell checker is hardly protective from errors, and may malfunction disastrously (one is reminded of the autocomplete text function which leads to hilariously inappropriate mobile text messages).
• The system will dumbly record everything. The patient coughs or strains on the bed pan, and the system records a peak blood pressure of 300 mmHg.
• Susceptibility to attack - the CIS can be compromised by accident or through malice, as any other database can. In fact, this has already been demonstrated several times in Victoria, for incstance in March 2021 when Windows -based ransomware encrypted and locked Cerner ICU servers in Box Hill, Maroondah, Healesville and Angliss hospitals. 

Evidence in support of CIS implementation

• Even very early CIS versions led to a significant improvement in documentation.
• Fraenkel's paper 92003) reports on a small scale study, which detected "significant reductions in the rates of medication, intravenous therapy, and ventilator incidents"  as well as generally positive reports from nurisng staff
• In one study, nursing staff spend ~ 30% less time on documentation, and this time appears to redirected to patient-oriented tasks
• It seems to shorten ICU stay without influencing any other important outcomes
• It may improve ICU funding by decreasing coding errors (illness severity and procedures coding for the purposes of billing), and presenting a more accurate representation of the activity level.
• More recent studies of the impact of CIS have found in pooled analysis that CIS improve a large number of important clinical outcomes, perhaps reflecting that the systems of today are a vast improvement in comparison to the systems of yesteryear.
• In particular, there may be an improvement in mortality.

Evidence against the use of CIS

• No improvement in mortality in early studies of clinical decision support features of CIS
• Many nurses are strongly opposed to this technology, largely due to a perceived lack of training and support
• Meta-analysis of studies did not confirm an improvement in time spent on direct patient care.
• Computer keyboards act as a reservoir of nosocomial pathogens. Though all things considered, you can't exactly ban them).

Worldwide experience

• The British National Programme for Information Technology (NPfIT) was a complete mess, according to Fraenkel.
• US has implemented to the greatest extent (economic and efficiency improvement drove the change) but the sharing of data is limited by a lack of common standards. Everybody bought a different CIS and the records are non-portable.
• Australian use of CIS in ICU has been patchy and also plagued by inconsistent standards
• The New Zealand and Canadian systems are lauded as being the most efficient, as the ICU powers and health care administrators banded together to demand a basic minimal standard from the vendors. The EMR entries in these countries are more portable. Rather than rolling out a huge pork-barrel disaster like the NPfIT, focus has been on "incremental and iterative change on a more regional basis".

Briefly, the issues involved in designing and implementing a CIS

• The database must be easily searchable and adapted to rapidly processing queries, while at the same time being sufficiently vast to contain the massive amounts of patient data collected every second. These two priorities are in conflict with each other.
• The database and its content management system must be flexible enough to adapt to a variety of situations, while maintaining a basic standard to improve data portability. Again, these two needs are usually in conflict (standardisation of formats trends towards decreased flexibility and a one-size-fits-all approach)
• The CIS is mindblowingly expensive. Fraenkel quotes costs of $25,000-$50,000 for a basic system. Per bed.  For a 20-bed ICU that ends up being in the order of $500,000-$1,000,000 - enough to hire four or five intensivists with change left over. And then, annual recurrent costs are mentioned,which can be up to 20% of the initial establishment price (this essentially involves hiring IT staff to maintain the database and terminals).
• Cheaper open-source alternatives exist. A vast array of Windows and  Linux-based open source software exists to support electronic medical records in developing countries. If you've got the mad haxx0r skillz, you can write your own code and then publish about it (Massaut and Reper in 2008 wrote a CIS for their ICU in resource-rich Belgium using PostgreSQL, running it on Linux). Obviously, support for open-source and homemade systems is going to be patchy.