SPO in Healthcare

Technically Speaking, a typology is just a hierarchical structure. Since computers can efficiently automate the hierarchy, it is now common to see them used on the Internet for searches or when gathering, ordering or presenting information. Furthermore, when the databases (DBs) they represent are highly normalized (made regular and consistent), these hierarchies can be implemented within tables and benefit from table relationships. That is, hierarchies can become the orientation or table of contents that allow one to get the most from a relational DB environment.

We are proposing a typology called SPO for health care. SPO is an acronym standing for Structure, Process and Outcome. From the informatics point of view, to literally get utility form this typology, one must sort or point to these three aspects of health care on the information platform.

The Clinical
It is proposed that SPO be adopted for persons to get answers or direction about health care clinical or administrative issues.

• "Structure" is that aspect of health care that is perhaps best represented by things like guidelines, standards, benchmark statistics, and other ways that define what is or what should be.
  
  
• "Process" is what doctors do. This can be imputed by examining the administrative or claims data base and translating such data as procedure codes (CPT's) and Diagnoses codes (ICD, MDC's, DRG's) into more encompassing ad comprehendible patterns of care or episodes of care. Although not a clinical application, Milliman and Robertson's MHC software, used in utilization review and for financial reporting can be somewhat revealing of process, however, it is the relationship of these various codes and constructs that give the reviewer information. In some advanced AI systems, one can actually determine the appropriateness of care or medical necessity from this translation.
  
  
• "Outcome" is the most difficult category to document or support with literature or opinion.

Some of the constraints on measuring outcomes is not having continuous enrollment data, a longitudinal record, or the ability to measure the results of the aforementioned "Processes." Often, research design is required to tell if X caused Y (causality).

Confounds also arise from population and because of acuity or case-mix differences (risk-adjusting software only partially correct for these disparities).

Health care management also suffers from the problem of small numbers

One can never be certain of the holy grail of outcome reporting--the "best practices."

To literally get to SPO, one needs to first orient the question that is being asked.

Used by medical diagnosticians or historians (doctors), the medical review of systems (ROS) helps document the present and past illness history. It also establishes the area of interest or focus of the patient's presenting symptoms or concerns--the "Chief Complaint(s) (CC)."

One peels the onion from the ROS to the inner core, the S, P and O of health care, which is, in total, is the clinical context.

For example, I have a backache = Musculoskeletal part of the body. I'm concerned with what can be done for this CC.

Now it should be easy to see that the next layer --use the "Process" context and available data to figure out the "Who" "What" "Where" or "When."

However, one must go to the "Outcome" layer to learn about "How Well."

Say you need to identify why a patient might have a CC of, for instance, "I wake up in the middle of the night and hear my heart beating in my ear; it makes it hard to get back to sleep." You look up "tinnitus" (ringing in the ear) and have to look to ear disease, heart disease, high blood pressure (hypertension), carotid artery disease, or psychological, neurological / neurosurgical causes, etc. This is the art of medicine. (It's and art to get an answer. But, go to the SPO typology and you may find the answers more easily. If not, you will be better able to articulate the question.

The full typology in this example could be as follows:
ROS: Ear, Nose and Throat (ENT)
Process issue - This began When? What? Have you had an ear or hearing symptom / problem? Who? Have you had an ear or hearing exam? ROS: Cardiovascular (CV)
Process issue- What? Have you had a heart symptom / problem? Who? Have you had a heart exam? When? How long have you had this symptom? ROS: Neurological / Neurosurgical (NS)
Process issue- What? Have you had a sleep[ing] problem? Who? Have you had an exam about this? When? How long have you had this symptom? ROS: Psychosocial (PsySoc)
Process issue- What? Have you had difficulties with sleep? Who? Have you seen or spoken to any professional about this? When? How long have you had this difficulty or concern?

When an individual has difficulty articulating a question, but needs a professional opinion, they will be able to use an algorithm or decision tree to take advantage of this typology.

---

Review of Systems:
Dent. - Dental/Oral Health
Derm. - Dermatological and Plastic Surgery
HENT - Head, Ears, Nose and Throat including infections thereof)
CV - Cardiovascular (anything primarily related to diseases of the Heart, e.g., Congestive Heart Failure)
Endo-Met - Endocrinology, Metabolic (e.g., Diabetes, Obesity)
GI - Gastro-intestinal
GU - Genito-Urinary including Renal (Kidney) Diseases
GYN - Gynecological
Hem-Onc - Hematology/Oncology (e.g., Cancer, HIV/AIDS, Organ Transplants) Infx - Infectious Diseases (e.g., complications of HIV/AIDS, TB, Otitis media or ear infection)
MSKL - Musculoskeletal and Rheumatological (e.g., Arthritis, Chronic Pain Syndromes, Fibromyalgia, Lupus)
Neuro - Neurological and Neurosurgical (e.g., MS, Seizures Stroke [Cerebral Vascular Accident]) Obstetrical
Opth - Ophthalmologic
Ortho - Orthopedics and Podiatric Medicine (e.g., Fractures, Osteoporosis)
Psy-Soc - Psychosocial (e.g., Depression, Insomnia, Poverty, Transportation, etc., mental disturbance as a primary or secondary condition)
Rs - Respiratory System (e.g., Asthma, Chronic Obstructive Pulmonary Diseases [COPD, Emphysema], and including related conditions such as Sleep Disorders as in Sleep Apnea])

Structure examples:
Benchmark performance objectives
Standards § Guidelines
Protocols
Comparative statistics (for 'apples to apples' comparisons, these data should be demographically controlled and acuity-adjusted)

Process examples:
Trends, Patterns
Care management observations
Performance (utilization) reports
Quality indicators
Factor analysis
Quality sentinel reporting

Outcome examples:
Variance analysis
Patient satisfaction surveys
Treatment efficiency studies
Cost-effectiveness analysis
Cost-benefit analysis
Cost-utility analysis
Cost-minimization analysis
Treatment effectiveness studies

The "? Branch" contains the following questions:
WHY? The issue of appropriateness, medical necessity, precedents
WHAT? Function, procedure, PDx or diagnostic plan / PTx or treatment plan / PEd or educational plan)
WHERE? Venue choice, alternatives, safety, appropriateness
WHO? What expertise do I need? To whom do I speak?
WHEN? What can happen if I don't do this now?
HOW WELL? What is the possible outcome?

References:
Donabedian A. Basic approaches to assessment: structure, process, and outcome. In: The Definition of Quality and Approaches to Its Assessment. Ann Arbor, MI: Health Administration Press; 1980:77-128.

Hibbard JH, Jewett JJ. Will quality report cards help consumers? Health Affairs. May/June 1997;16(3):218-228.

Hibbard JH, Sofaer S, Jewett JJ. Condition-specific performance information: assessing salience, comprehension, and approaches for communicating quality. Health Care Financing Review. 1996;18(1):95-109

---

To demonstrate the functionality of this typology, we will need to answer a few prototypical questions such as:
On Formularies: How can a patient easily obtain information about why an HMO won't cover a particular Rx for example, for an uncommon disease like multiple sclerosis (MS)? Answer: ROS--MusculoSkeletal system (MUSKL) ; SPO--S; ?-- WHAT? Print the Formulary Rule or guideline about coverage of experimental drugs; then, if asked, show formulary listing, sorted by diseases of the MUSKL system.

On Lifestyle Issues : Where can the ordering physician get help managing obesity for his patients? Why can he not get insurance coverage for weight reduction pills or residential weight-loss treatment center care (fat farms)? Answer: ROS--ENDOMET; SPO--P; ?--WHERE? to link to sites that can help patients do calorie counts, attend support groups like Weight Watchers, learn about "bariatrics," the practice of obesity management, etc.; or SPO--O to learn about the effectiveness of programmatic weight-loss efforts; SPO--S to find a policy or links to relevant sites.

On Best Practices: An employer wants to identify the "Centers of Excellence" for managing congestive heart failure (CHF) in her community. Answer: ROS--CV; SPO--O; ?--HOW WELL? A comparative listing of outcomes called "best practices" data on treatment compliance, use of case managers (RN's) and disease state management (DSM) programs; SPO--S for incentive programs that help managed care organizations (MCO's) do their thing, and their thing, by the way, is SPO--P.

---

Technical Notes:
In XML, hierarchies are implemented with nested tagging, and constraints, etc. In object models, hierarchies are implemented in the structure of classes and class relationships. In automating a typology, the primary concern should be specifying the structure of the typology itself. Once you have a generalized typology superstructure (within which all regions of the typology will fit), implementing it in any number of hierarchical computer models is relatively straightforward.

Once modeled, one can begin to address some very interesting issues:
· Should the typology have an open platform, allowing it to be used by third parties as an embedded system?
· Should the systems that use the typology write out all human interaction to back-end log files so that the typology authors can analyze their responses for where there are bottle necks, or where users get lost, or where users enter less or more frequently than the authors expected.
· Should the typology have software that analyzes its own logs suggesting to the authors what links might be more appropriate based on frequent user misdirection.
· How will the authors load the thousands of data needed to add meat to the structure. Hopefully it will be parameter driven. It should have a UI so that the authors can easily enter / change / delete structured data without the aid of a programmer. It should also have an API so that programmers can help the authors "bulk load" data where hand entering would be tedious or unnecessary.

All of these features should be implemented in both Win32 and Internet applications.

---

Charges for reports are as follows:
1) LogicalLinks using a compendium of bookmarks, accessed through key word searches. There will be no charge but a cookie will be generated, with or without an actual subscription (online registration)
2) Boilerplate If the answer is generated using multiple and/or nested links and/or a few paragraphs of available text is downloaded, a $15 charge is applied.
3) HandyResearch If research must be done by a specialist, $75 is the fee* for up to 1/2 hour to do research and generate a report. (* $25 to engage the researcher and $50 upon receipt of the findings with no charge for up to 3 related questions based upon that submission).