Health professional’s conclusions and working assumptions that will guide treatment of the patient.

Data Element

Assessment and Plan of Treatment


NCPDP Comments

NCPDP recommends Assessment and Plan of Treatment belong in the Health Concerns section of the health plan. The appropriate coding should utilize LOINC and SNOMED. Pharmacists providing assessments document findings in the Health Concerns section of the NCPDP/HL7 Pharmacist eCare Plan using SNOMED CT and LOINC.  These conclusions and working assumptions will guide treatment of the patient.

Lantana Consulting Group Comment

Lantana recommends removing the word “assessment” and re-naming this class “Care and Treatment Plans.” Assessments are a collection of observations, which could be classified under Health Status, rather than combined with Care Plans. A Care and Treatment Plan category would improve focus on treatment plans and include types of orders related to medical services and treatments, therapy orders, end of life care orders, Advance Directive Information (ADI) orders, nutrition orders, and others.

Assessment and Plan is magical, needed, an separate. Here's why.

Hi. I'm Henry Wei, and I'm an internal medicine physician, a former health insurance executive, a clinical research executive, and a former technologist and clinical specialist from Google.  Here are some comments as a private citizen.

In my opinion, Assessment and Plan is perhaps the most valuable part of any medical data, from a clinical standpoint.  It conveys what the clinician, in aggregate, has summarized and assessed.  But moreover, that assessment justifies and motivates the plan -- an A&P (or Assessment & Plan in SOAP format) cannot be divorced from one another.  They are two parts of a narrative: "Who? What? Why?" and therefore "How? What next? When? With Whom?"

More broadly: the Plan -- when eventually structured better perhaps in future USCDI v(X+1) -- is an ideal way to accomplish the aims of ONC from the standpoint of multiple HHS jurisdictions:

A. CMS - Quality - rather than assess quality measures (which rely on past data), *PLANS* allow us to assess what the clinician set forward as their plan -- and whether or not that was actually right, with enough time to then modify the plan to, for example, follow evidence-based care, or otherwise address risks for that patient.

B. FDA - Real-World Data.  The thrust of the 21st Century Cures Act, amidst the Cancer Moonshot, was in discovery of new Cures and advancement of such.  This included congressionally-established instructions to the FDA to make use of real-world data that are primarily retrospective, but a major challenge is separating causation from correlation, even with propensity score-based methods (the mainstay of many analyses today).  The key is to understand the intent of why someone was assigned a treatment or not -- this is often revealed in the assessment and plan, especially biases (or errors of omission) that are evident only in clinician-documented assessment and plan, rather than, say, retrospective insurance claims or retrospective EHR data that document what happened, but not what was *supposed* to happen and why.  

C. FDA + NIH - Clinical Research.  *Protocols* are another form of plan, and in some specialties, e.g. oncology, they are inseparable from the actual treatment plan.  Over in primary care we see broad fragmentation of "gaps in care" and pathways with such wide institutional variation.  But supporting Protocols -- as very discrete *Plans* of care -- is something that can be conducive not just to new therapeutics, diagnostics and device evidence generation -- but moreover, a learning health system.  A protocol for, say, addressing the social determinants of health -- how are we to establish an evidence base?  Well, one way is executing a clinical research protocol on a SDOH intervention. Randomized or not, having a structured, well-classified way of expressing the *Plan* -- in this case a *Protocol* -- is a huge advantage, and can further be re-tasked over time from the clinical research arena and moved to care delivery.  This is the basis of not just the CTE (Clinical Trials Enterprise) that McClellan, Califf, and others have spoken of -- but also the basis of a learning health system, and clinical decision support.

D. CDC - Public Health.  The Plan can, in particular, support broader plans that don't speak to the care of the patient specifically, but of plans that guard against risk to broader populations -- particularly for infectious disease, and most recently, pandemics such as COVID-19.  Having a well-structured, specifically classified plan that can be reliably found and encoded into electronic health records will be critical for public health agencies including HHS/CDC to understand what clinicians, institutions, and other organizations were *planning* on doing with patients... and ideally, also developing ways to provide input into the Plan as well.

and finally

E. (All agencies within HHS) -- the Patient.  Patients are to receive an after-visit summary for certified EHR use.  This has collapsed into a retrospective summary, which is informational -- but for patients, what's extraordinarily helpful is a well laid-out and specific *Plan* -- and a clear rationale as to why the Patient will be following that Plan.  Although the raw Assessment and Plan may not be understandable to many patients, having it structured and classified -- separate from, say, "Health Concerns" -- introduces potential for 3rd party developers and EHR suppliers and others to translate this for *Patients* and their Caregivers to follow -- particularly when assigned for people to complete tasks associated with that plan.  In this way, the Plan transforms EHRs from pure information to actual real-world action and implementation.  Very few other parts of the health record (perhaps medications and orders otherwise) have this property of affecting future care, rather than just documenting past care.

Overall, the Plan is a critical element -- perhaps one of the most critical ones -- of the electronic health record, because it represents the Future State -- what *should* happen in the future.  The more specific, time-bound, and actionable it is, the better.

I would agree with the comments that Dan Vreeman made back last year in 2021 -- there needs to be clarity.  But I would disagree with its removal.  While it *seems* redundant, the future tense (what we used to call "mood codes"), as well the conveyance of the *intent* of the clinician are distinct.  These are not mere representation of facts and events, but rather instructions for the future state of the patient.  

Overall, FHIR CarePlan resource is a good start for this Class -- but the end goal would be something more akin to supporting a full-bore set of tasks or ideally a complete clinical pathway or protocol, as described for a given patient.  While controlled vocabularies are relevant, the broader issue is representing prospective, future state instructions -- the most structured of which (and regulated by the FDA!) are clinical research protocols, of which there are 100,000s of thousands of examples and even more real world experience monitoring for protocol deviations (as opposed to gaps in care in clinical care).  CDISC -- which has only now begun to emerge in the standards community predominantly governed by HL7 -- did have a Protocol Representation Model (PRM) which incorporated requirements from the WHO also.  HL7 Vulcan also has some good early efforts at basic structuring of data useful in clinical research, such as Adverse Events.  But neither CDISC nor HL7 Vulcan's domains/elements are yet sufficient to describe a plan of care when it is a research protocol, or a care delivery protocol (eg as in oncology).

Above all, please note that the 21st Century Cures Act *Public Law No: 114-255) starts with "To accelerate the discovery, development, and delivery of 21st century cures, and for other purposes."  Let's ensure that we haven't just focused on delivery, but also on *discovery* and *development* in USCDI, which appears otherwise to exclusively be purely about delivery.

In other words, USCDI classes and elements to support Discovery and Development are a logical outgrowth of 21st Century Cures.  If neglected, we may not be fulfilling the intent of the directive from Congress.  And certainly as we witnessed with COVID-19, discovering and developing cures -- and then rapidly moving a protocol into a care plan (delivery) -- the first two, Discovery and Development of 21st century cures (i.e. therapeutics) has been critical to the survival of millions of Americans for COVID-19 alone.

Thank you for your consideration, and I would be happy to elaborate further on the impact from driving more specificity -- while being realistic -- along this USCDI Data Class (Assessment and Plan).   While care delivery is an important "final mile", when it comes to Assessment & Plan -- not to mention, say, randomization in healthcare services research -- please do not neglect or under-value the opportunity to advance actual cures, and not just care delivery, be it for therapeutics research, SDOH research, or other interventions to improve the standard of care rather than just deliver it.


Henry Wei, M.D.

Commenting as a private citizen and former US Federal Employee working closely with HHS/ONC

NCPDP Comments on USCDI draft V3

Assessments belong in the Health Concerns section of the health plan. The appropriate coding should utilize LOINC and SNOMED.

NCPDP Comment

NCPDP requests ONC add the FHIR® questionnaire resource for sharing assessments which is used in the Pharmacist eCare plan.

Log in or register to post comments