Outcomes Framework - Background

Mirah has, since its inception, been helping behavioral healthcare providers measure the outcomes of the care they provide. This help has primarily taken three forms:

  • Helping structure data collection so that providers are measuring the right thing at the right time.
  • Allowing customers to easily access their data in a variety of formats, both inside the Mirah platform and through data exports designed to be used in Microsoft Excel or a data warehouse system.
  • Providing consulting for organizations who would like more aid in understanding their data, including helping create and track appropriate outcomes metrics.

However, it has become apparent over time that although these three activities provide a strong basis for outcomes, they aren’t always sufficient to give providers the tools they need to operationalize their outcomes metrics.


In particular, through our work with many providers, we identified the following shortcomings:

  • Constructing outcomes metrics is complicated. There are many different ways to make metrics and many different decisions to make.
  • Once you have decided what your metric is, it can often take significant time to transform the raw data into the final result. Sometimes it involves analyst or other specialist time and tends to only be done on a quarterly basis.

The Goals of the Mirah Outcomes Framework are to allow Mirah customers to:

  • Collect the appropriate measure data at the appropriate interval to enable high quality outcomes metrics.
  • Help our customers understand their data so they can create a set of metrics that allow them both to:
    • Report effectively on outcomes to external parties like insurance companies in a way that best highlights the value their care provides.
    • Be a data-driven organization that uses the outcomes data to iterate and improve on their standard of care.
  • Allow our customers to easily keep track of these metrics over time, and allow all people in their organization who need to see these metrics to have quick and easy access to them.

Design Decisions


The Mirah Outcomes Framework is an opinionated framework. This means it makes a set of simplifying design decisions which allow customers to get going quickly, both in terms of the outcomes the platform supports and how these outcomes are represented.


Customers always have the option of downloading the full dataset and constructing outcomes metrics from scratch as they have done historically.


The Mirah Outcomes Framework makes the following assumptions:

  • Each measure/scale supports a set of simple analyses, depending on how the scale is constructed. These include:
    • Raw scale, where appropriate
    • Trend and clinical cutoffs, where these are supported by the psychometrics of the scale.
  • Each metric is designed to give a single numerical answer.
  • Best practice is to combine several simple metrics to cover all the necessary parts of your practice.
  • Metrics are designed to be simple to understand in aggregate. This may lead to unusual results in some particular cases.
    • As an example, consider a metric which evaluates treatment episode response at 12 weeks. This excludes data in week 13 and beyond, which will mean both patients who finally see response after week 12 will not be counted as a success, but also that patients who have a sudden deterioration in week 13 will still be counted as a success even though they have since regressed.


Explicit Limitations


There are several limitations that may impact any conclusions drawn from the outcomes analysis in the Mirah platform: 

  • Incomplete data - in most cases, the data in Mirah will not be 100% complete and may contain minor inaccuracies (which stem not from the platform itself, but may be related to human error in data entry, imperfect workflows, etc.).
  • Data Silos - because Mirah only contains a small subset of the possible data on any given patient population (eg: there are likely many data points in the EHR that are not represented in Mirah), the ability to understand the multitude of potential contributing factors to any one outcome is limited.
  • Measurement limitations - not all measurements have been validated for use in the diverse settings in which they are being employed (eg: many measurement tools are initially validated on college students which represents a specific subset of the larger community), which may lead to biased interpretations of measurement results
  • Self-selection bias - patients who complete MBC measures may fundamentally differ from patients who opt not to complete MBC measures, or specific groups of patients (eg: racial groups who have historically been harmed by the misuse of healthcare data) may not respond to MBC differently than other patient groups
  • Data collection bias - by virtue of the specific measurement tools that are implemented and analyzed, organizations are prioritizing specific types of potential outcomes over others; there is always the possibility that some patients would have shown different results if a different set of measurement tools were employed
  • Interpretation bias -  whenever interpretation of data is completed on a “post hoc basis” (that is, the hypotheses about the data are generated after the data has already been collected), there is a chance for confirmation bias, which is the phenomenon of noticing results that support pre-existing beliefs or hypotheses.  This can lead to multiple types of errors including the possibility of incorrectly interpreting a non-significant result as significant or the possibility of overlooking significant results that are inconsistent with the existing beliefs/hypotheses.

Users of the Mirah Outcomes tool are therefore encouraged to consider conclusions drawn from MBC outcomes with some degree of caution.


See FAQ.


Get started with the Quick Start Guide!