Health IT and the Move Toward Mobile Computing

— By Ishir Bhan, MD, MPH

doctor with handheld mobile tablet deviceHeathcare notoriously lags behind other industries when it comes to adoption of information technology. With many healthcare providers still resorting to paper records, the expectations on the medical community are comparatively low, even ignoring the obvious safety and quality improvements that would result from basic interoperability between systems.

While the health IT community struggles to make inroads, there is a larger change afoot in computing which, if left ignored, will cause the field to fall further behind: the change from traditional computing, driven largely by desktop computers, to mobile devices. Led by the smartphone and followed closely behind by tablets, these devices are increasingly capable of being fully-fledged siblings to the relative behemoths that sit at our desks, with one key advantage: they are always with us.

These devices are increasingly capable of being fully-fledged siblings to the relative behemoths that sit at our desks, with one key advantage: they are always with us.

The past decades have seen healthcare providers increasingly driven from patient rooms to workstations to complete their work, squirreled away behind monitors instead of at the bedside. Both patients and their physicians have lamented the increasing brevity of interactions, driven in no small part by the need to retreat to a keyboard to document, review data, enter orders, and complete other various tasks of the day.

Mobile computing has the ability to offer a course correction. Using Mojo, the Laboratory of Computer Science’s clinical smartphone app, I have been able to review results on my way to a patient’s room, share results and images with them in person, and records thoughts that would have faded by the time I returned to my desk. Communication becomes instantaneous instead of delayed, and the flow of healthcare becomes enabled, rather than encumbered, by the growing adoption of information technology.

We are at a new stage in the evolution of computing: a shift from traditional, fixed machines to truly personal, always connected, versatile tools. How quickly we fully embrace this technology will shape how effectively we can move forward as an industry.

Ishir Bhan, MD, MPH, of MGH LCSAbout the Author
Ishir Bhan, MD, MPH, is the Director of Collaborative Informatics at the Massachusetts General Hospital Laboratory of Computer Science, an Associate Physician and Director of Nephrology Informatics at MGH, and the Associate Program Director for the Nephrology fellowship program at MGH. His clinical software innovations are widely used to improve care delivery across Harvard-affiliated hospitals.

The difference between ICD-10 and ICD-10-CM

— By Jaime Chang, MD, MS

On October 1, providers and electronic health record systems across the United States faced the transition from using the ICD-9-CM set of 14,567 diagnosis codes to ICD-10-CM set of 69,823 diagnosis codes for purpose of medical billing. The prospect of this transition was stressful enough that it was postponed a number of times, and while the switch went forward at the start of the month, the Centers for Medicare & Medicare Services (CMS) announced flexibility in acceptance of imperfect coding for the first 12 months to ease the transition.

It has been said that the transition from ICD-9-CM was long overdue, with the codes used for diagnoses and billing not updated in the US for more than 35 years. According to the World Health Organization (WHO) website on the International Classification of Diseases (ICD), ICD-10 came into use in 1994, and ICD-11 is planned to be released in 2018.

If the US had decided to transition from ICD-9-CM to the unmodified ICD-10, we could have actually decreased the number of codes from 14,567 to 14,199!

However, ICD-9-CM and ICD-10-CM are not the same as ICD-9 and ICD-10.  The Clinical Modifications (-CM) are American modifications of the WHO diagnosis codes that expand on them greatly. ICD-9 has 6,969 codes, and ICD-10 has 14,199 codes (including fourth-character place of occurrence codes), so the clinical modifications roughly doubled and quintupled the number of codes, respectively. If the US had decided to transition from ICD-9-CM to the unmodified ICD-10, we could have actually decreased the number of codes from 14,567 to 14,199!

I picked three ICD-10-CM codes at random to demonstrate how ICD-10-CM differs from ICD-10.

  • A84.8 Other tick-borne viral encephalitis
  • O64.5XX4 Obstructed labor due to compound presentation, fetus 4
  • T45.522A Poisoning by antithrombotic drugs, intentional self-harm, initial encounter

The closest matches in unmodified ICD-10 are these, respectively:

  • A84.8 Other tick-borne viral encephalitis
  • O64.5 Obstructed labor due to compound presentation
  • T45.5 Poisoning by anticoagulants

A84.8 remained unchanged between ICD-10 and ICD-10-CM.  O64.5 was clinically modified into 7 codes (fetuses 1 to 5, unspecified, and other).

T45.5 was clinically modified into 36 codes by making several layers of distinctions:

  • anticoagulant vs. antithrombotic
  • poisoning vs. adverse effect vs. underdosing
  • accidental vs. intentional self-harm vs. assault vs. undetermined intentionality
  • initial encounter vs. subsequent encounter vs. sequela

If coded reliably, all the extra information available from using the clinical modification instead of the unmodified ICD-10 could be nice to have, but the price for having this extra information is steep. Offhand, I do not see why it is important to specify that fetus 4 was the one with the compound presentation, and how that specification should affect reimbursement. It also seems unlikely that fetus number would be coded reliably enough for research purposes.

The closest SNOMED-CT concepts are these, respectively:

  • 415821008 Viral encephalitis transmitted by tick (navigational concept)
  • 199755001 Obstructed labor due to compound presentation (finding)
  • 67438002 Poisoning by anticoagulant (disorder)

Mapping to and from SNOMED-CT is important because, while ICD-10-CM will be used for billing, SNOMED-CT is the preferred terminology for interoperability between systems. The granularity in ICD-10 better matches the granularity in SNOMED-CT, allowing for relatively painless forward and backward mapping.  The finer granularity in the clinically modified ICD-10-CM makes mapping between ICD-10-CM and SNOMED-CT more awkward, either by losing specificity in going from specific to general or by requiring more information to go from general to specific.

The complexity of ICD-10-CM probably contributed to its late adoption and continues to contribute to difficulty in achieving interoperability between systems that are expected to use a combination of ICD-10-CM and SNOMED-CT. My life as a practicing physician and informatician would be simpler if the US had decided not to clinically modify ICD-10 so much. Instead of dreading the transition to ICD-10-CM, we could perhaps already be well on our way to ICD-11, which the WHO ICD website suggests is being designed with electronic health record systems in mind.

If you would like to explore ICD-10 (unmodified), ICD-10-CM, and SNOMED-CT for yourself, you can use these websites:

ICD-10: http://apps.who.int/classifications/icd10/browse/2016/en#/
ICD-10-CM: http://www.icd10data.com/
SNOMED-CT International Edition: http://browser.ihtsdotools.org/

What do you think?

Jaime Change, MD, MS, of MGH LCSAbout the Author
Jaime Chang, MD, MS, is a practicing physician with over 10 years of experience in supporting problem and medication terminologies in a locally developed electronic health record (EHR) system used by multiple outpatient practices at Massachusetts General Hospital.