Measuring Quality of Service Based on Social Media Engagement in Healthcare
This blog post is just an initial draft of me brainstorming about a model to evaluate Quality of Service based on input received from the community over Social Media.
Key Points to consider:
- The model has to be feasible for implementation and wide adoption as much as possible:
This means that Semantic Queries are out of the question at the moment. No point in coming up with a complex model, that while it will provide high value, can only be implemented by a few health care establishments. Feasibility encompasses the availability of financial, technological, regulatory, and human resource mix to successfully implement the solution.
- The model should address the structure by which data is captured:
This part is tricky as community social media engagement could be in multiple forms including 1) personally generated micro-blogs (tweets and FB Status updates), 2) Re-posts of information shared by other individuals/groups such as re-tweets. (best to distinguish between re-posts from individuals with direct contact [e.g. friends, family, colleagues, etc] and re-posts of info provided by organizations or websites [e.g. Hospitals]) 3) Personal Blogs, 4) Posts on public patient websites (e.g. imedix and patientslikeme), 5) Posts on Healthcare organization-hosted custom SM tools (e.g. blogs inside the cancer society website), etc.
Since data in each form of SM is recorded in a different format, there needs to be a clear outline of how data will be captured (e.g. based on keywords or hash tags) in each of the 5 forms, and how it will be aggregated and linked together (e.g. grouping tweets about surgical complications with Facebook Status updates or blog posts about the same topic). This aggregation, if it is to provide real value, it should be beyond word counts.
Finally, and while this may be somewhat subjective, but it is good to classify SM posts (if possible) to differentiate between fact and perception. For example, a delay in seeing an ER doctor maybe viewed by a patient as an inefficiency in the Hospital (perception), while the fact maybe that the hospital is receiving a patient load beyond its staffing capabilities for the day.
- The model should address the layout of the final report, and how the information in the report will support decision making:
Once the previous point is finalized (how and what data elements will be captured), we need to identify what key data summaries (e.g. percentages, totals, averages, etc) we need to generate from the captured data. Those data summaries should be directly linked to factors that affect decision making (e.g. fund/resource allocation, alignment with corporate goals, etc).
- The model should address what quality measures we track/don’t track internally and what we can track externally (through SM) only:
Since the goal of the model is to utilize data/information input into SM to evaluate Quality of Service, we need to first establish our internal measures of QoS. From there, we can move one to establish 3 categories of Quality Measures: 1) Quality Measures that we can only track internally. Those measures affect internal processes directly, and are not apparent to the outside (patients and the rest of the community), and yet affect the outside indirectly (a demotivated employee because of duplication of work is not an efficient employee), 2) Quality Measures that we can only “truly” track externally. Those measures are mostly related to the patient experience and for the sake of credibility, should only come from the outside (e.g. patient and community satisfaction surveys), and 3) Quality Measures that are a combination of both internally and externally tracked data/information. Those measures include processes where feedback from both internal staff and external entities is required to evaluate the quality of service (e.g. time sitting in an ER waiting area before seeing a doctor).
- The model should address the platform “mix” from which data will be collected, and how data will be centrally aggregated.
Customized to each organization’s needs, objectives, and budget, the model should include a list of the various social media tools, the objective of each tool, the data to be collected from each tool (overlapping may occur here), and finally, where (storage platform) and how the data will be aggregated from the various tools (e.g. count of re-posts on twitter, FB, or custom SM tools, made by patients will be stored in a central SQL database, from which dashboards will be created).
- The model must draw a clear distinction between on-going monitoring and on-time occurrences:
Gathering data about on-going activities (e.g. daily work in the Outpatient Clinic) differs from gathering data about one-time events (e.g. Pilot project or PR campaign for a one-time event). This distinction must be outlined in the model as for each category, data to be collected, and the structure in which it is to be collected would differ.
- The model should include a feedback cycle for the findings:
Customer-based Total Quality Management (TQM) is an iterative process of gathering information from customers, aggregating the information, releasing the findings, and receiving feedback on the findings from customers in order to validate or disprove our conclusions. After all, we are using social media to analyze our customers’ perception of our services; we don’t want to aggregate the information and reach conclusions influenced by our biases while missing the main points from our customers’ perspective.
Like this post? Subscribe to my RSS feed and get loads more!