In this exercise, we will perform a detailed analysis of the actual 2017 HEDIS c
In this exercise, we will perform a detailed analysis of the actual 2017 HEDIS clinical measure supplied by National Committee for Quality Assurance (NCQA). Our measure analysis will focus on diabetes care. In preparation for this exercise, please download and review the Comprehensive Diabetes Care (CDC) measure provided in the Unit 4 readings tab. Be sure to read through and understand the measure in its entirety. The major parts of the measure are as follows:
Summary of changes: updates for the current year (measures are typically updated annually).
Description: basic summary of the measure.
Eligible population: defines inclusion criteria.
Denominator: the number representing the total population.
Numerator: definition criteria for identification of patients for the measure.
Hybrid Specification: other criteria combinations for the measure (FYI – will not be utilized for this exercise).
We will utilize ICD, CPT, LOINC, and RxNORM codes as, and where applicable for the entire measure building exercise. For all parts, use an example output table in the following general format, but you can deviate as you see fit and to whatever matches your style for presenting data. Formatting is not important; your understanding of the measure building process and the meaning of clinical data codes are important.
Step 1. Define eligible population in a table format and label it Eligible Population Table. 10 points.
Locate Eligible Population section of the HEDIS measure on pp. 1 – 2. Select appropriate medical data vocabularies for each of the categories, find necessary codes, and provide a table in a format of or similar to the example table above. Do not worry about specific dates (just use “last year” or any time definition of your choice when needed) and financial/insurance measures such as Product Lines, Allowable Gap, and Benefit. In the real world, you will have access to this data from various systems and will be required to include every step and every data element, but we will disregard financial and time data for now, for academic purposes.
Step 2. Define the denominator in table format and label it Denominator Table. 10 points.
Apply the same strategy you employed to build eligible population under Step 1 to define the entire denominator. Denominator description is on page 4 of the measure specification. You already have eligible population defined from Step 1, so no need to repeat, and you can skip to the next section that describes HbA1c controls.
Step 3. Define the numerator in a table format and label it Numerator Table. 10 points.
We continue applying the same strategy to define the numerator. In this section:
a. LOINC would be your main vocabulary that also includes some of the value sets listed in the measure specification.
b. RxNorm may be necessary for drug definitions.
c. If you are unable to locate some of the value sets, such as CKD (chronic kidney disease), replace with codes pertaining to this disease or skip the value set. In real data analytics environment, you will not be allowed to skip, but you will also have access to a greater variety of tools to locate necessary codes.
d. Note that medical documentation codes will not delineate percentages, i.e. <7%, >9%; this is done by analysts on paper and implemented by software engineers programmatically. For our purposes, you can just specify percentages in the description – applicable codes will be the same no matter percentages. Or you can list all percentage descriptions on the same line within the same specification.
e. You can skip the Eye Exam section on Page 6: there is no easy way for you to locate relevant codes.
f. On page 7, use RxNORM search to locate codes for ACE Inhibitors.
g. We stop on page 8 and do not take on Hybrid Specification that concludes measure definition.
Follow the steps described in the NCQA measure for the numerator, accounting for the notes above.
Your output is tables representing Steps 1, 2, and 3. Again, feel free to use a different output format and/or more tables as necessary. The format of these assignment instructions is purposely ambiguous and does not provide you with line-by-line directions. Finer detail is not omitted. Instead, it represents academic and industry realities of the modern data science development times; there are no directions, out-of-the-box solutions, or specific one-ways of doing “things”. It’s on you to demonstrate your creative thinking to get through the sea of options or no options, to find a workable solution.
The ultimate output of your measure is a rule-based program that would be built upon foundations you have just laid out in this exercise. Were you a medical professional with an ability to create new ideas for measures and actually define these measures, you could find yourself working for either organizations like NCQA that define new measure standards or numerous technology or consultancy firms that either program standard measures and/or define custom ones.
APA requirements for this exercise are relaxed to reflect its technical nature, but please cite all references using APA style, as usual. You do not need to reference ICD-10, LOINC, CPT, or RxNorm databases that we know you will be using for this exercise. But please do reference any literature that you did utilize, as/if applicable.
See attached document for reading materials, guideline, and complete assignment.
