MLTC & Healthcare: Star-Rating for MLTC plans
Updated: Jun 30, 2020

What is The Managed Long-Term Care (MLTC) performance data?
It is a semi-annual evaluation of New York state- certified MLTC plans. Rates are calculated for each performance measure by the plan, and describe their population or assess their quality of care.
Plans are evaluated on the quality of care they provide and on enrollees’ satisfaction [1].
Type of performance data.
Different type of measures, survey-based measures or population-based measures across all plans in MLTC.
The results are reported per semester (in this case 2018), and there is a process where, depending on the results of these measures, a final grade is assigned to each plan (stars).
Areas, Domains and Measures.
There are a total of 18 measures, grouped in 12 domains, distributed in 4 areas. Here we present a sample of measures grouped by Domains and Areas. We can notice that different domains can contain a single measure or more.

The process for Star rating per plan
The star-rating process consists of 7 steps. The first 3 steps are related to the measures, this means we only do some calculation on the result of the single measures.
The last 4 steps are carried out at the domain level. This means that if a domain only has one measure we keep the result of the first 3 steps. On the other hand, if the domain contains 2 or more measures after step number 3, we will average over the total number of measures in the domain.

Example
We will work with the area Stability or Improvement, this area contains 4 domains: ADL Stable or Improved, Pain Intensity Stable or Improved, Shortness of Breath Stable or Improved and Urinary Continence Stable or Improved.
In this specific case, all the domains have only one measure. For simplicity, we will use the domain Urinary Continence Stable or Improved to exemplify the calculation process.

Standardization of measures
Here it is important the distinction between the two types of data. When the measure we are working with is population-based measure, we need to use Nelson's H-statistics to make the calculations. When the measure is a survey-based measure we use Student's t-statistic.
In the calculation, the difference resides, among other characteristics, on the variables we use to calculate steps 1 and 2. In both cases, we will use plan rate and statewide rates as variables. Nevertheless, when the measure is survey-based measure, the standardization is carried out by subtracting from the plan rate the statewide rate and that result goes over the standard error.
When the measure is population-based measure, the formula is a little more complex, and it involves the denominator of the measure. The denominator is the total of members that qualify for a specific measure. Here the size of the plan can contribute to improving the rating or lowering it.


Here we can see the transformation, from the original results to the standard measure. The size of the dots are related to the denominator of each plan, i.e if the size is relatively small, this plan has a small denominator.
Measure limits and trimming
Once again, in this part of the calculation, since we are working with a population-based measure to find the limits of the trimming process is carried out using Nelson's H-statistics.
In this case, the limits are the quantile from the t distribution based on a probability, multiplied by 3 and its positive and negative values are the limits. The probability is obtained using the formula on the left-hand side, where n is the total of plans participating in this specific measure.
