Health systems across the U.S. have had to set up sophisticated COVID-19 data reporting systems and dashboards to track the spread within their institutions as well as their community.
In recent weeks, hospital data has been scrutinized more deeply as HHS changed reporting protocols and the federal government further tied reporting to resource allocation. Many states have reported challenges in gathering COVID-19 data from local hospitals and labs as hospitals focus primarily on treating COVID-19 patients. As a result, many health systems plan to boost investment in data gathering and reporting in the future.
Here, 12 CIOs and IT leaders answer the question: What is the biggest issue with COVID-19 data that you think should be addressed first?
Jason Fischer. CIO, Information Systems of PIH Health (Whittier, Calif.): Data aggregation can create challenges when the guidelines for reporting are not clear and data received does not correlate accurately across the base. What has been problematic, particularly in the early days of COVID-19, were the daily adjustments and lack of clear guidance related to the reporting of data that was requested.
Richard Temple. Vice President and CIO of Deborah Heart and Lung Center (Browns Mills, N.J.): The big issue is ensuring that we properly define a ‘positive’ both for reporting, and for patient safety reasons. We have had so many different scenarios that we’ve had to work through in terms of properly defining true positive cases that the reporting truly has been daunting. For instance, picture a patient who may have been previously tested and arrived with a negative COVID result but is exhibiting symptoms, is subsequently tested, and shows up as positive.
We’ve also had other situations where patients have had tests both internally here at our hospital and from an external lab, and the results conflict. Tracking our persons under investigation also is a critical part of the dashboards we have created and adds a whole new layer of complexity for our quality and infection control coordinators. Having to reconcile the COVID status of patients with multiple stays here at our center, where one stay correlated to a negative result and another stay correlated to a positive result also was something we had to work through. Having a centralized FAQ with guidance on these real-life situations would have greatly assisted us as we worked through addressing these issues.
Tamara Havenhill-Jacobs. CIO of Bozeman (Mont.) Health: We know there are a number of issues; however, I believe that the issue to be addressed first is to consolidate the reporting to the most important data points that will drive the distribution of critical supplies and staff. The required fields change frequently and sometimes without warning. There lacks transparency in how the data provided is assessed and what is actually being executed as a result.
Greg Bryant. Director of IT at Baylor Scott & White Texas Spine and Joint Hospital (Tyler): Due to the nature of the pandemic and how the response was individual to every state health officials, we had a decentralized reporting structure. I believe having a united reporting structure to better see trends and be able to allocate resources accordingly.
Aaron Young. CIO of Summit Healthcare (Show Low, Ariz.): Due to the rapid progression of COVID-19, the mandates for data reporting were developed and in many cases left to interpretation of each facility without a true standard to allow for normalization. Data reporting requirements came from multiple agencies and each agency seemed to have different requirements. Early data appeared to have driven trend-lines higher at a faster rate than what was actually experienced, perhaps due to the increased testing volumes.
Data trends seemed to suggest that perhaps early data wasn’t adjusted for increased testing volumes. Delays in reporting resulted in inaccurate daily totals that skewed information provided to the public. The biggest issue with COVID-19 data that I feel should be addressed first is to agree on a data definition to ensure everyone is reporting apples-to-apples while ensuring duplicative data is not reported from multiple organizations.
Roger Neal. Vice President and COO of DRH Health (Duncan, Okla.): To me, in an ideal world it would have gone like this:
1) The federal government works with states to determine what data needs to be collected.
2) The federal government works with states and data representatives from several small and large health facilities to define each data point being collected.
3) One database is created where health facilities report to daily. This data is available to the federal and state governments at all levels.
It has been a very frustrating experience given trying to plan, execute, and care for patients and being told to report this here, that there; we want this but don’t give us that, etc. We had a lot of other concerns going on caring for the actual people and not just inputting data.
Cara Babachicos. Senior Vice President and CIO of South Shore Health System (Boston): It’s harder to retrospectively identify a COVID positive patient than it may seem. Many patients are tested elsewhere (skilled nursing facilities, mobile site, primary care physicians’ offices, etc.) prior to admission. If the patient comes as a confirmed positive, we would prefer not to run a test on them, especially when test kits and turn around on test results were delayed.
However, we still need to confirm and document that they are positive even though we don’t have immediate record of the positive result. There are multiple approaches to this and likely each hospital is handling it a little differently. In Massachusetts we are also now required to test on all patients prior to admission and, due to testing turnaround times, it could sometimes take days to get the test results back (unless your in-house lab has unlimited capability), so you must presume positive until patient is confirmed.
Similar to above, the timing of the positive is also important. We may have record of the patient having tested positive before, during or after their hospitalization. We should only count the patients who were hospitalized while they were positive, but this is also tricky. Predictive models were all over the place. It felt like we had to shop around for one that seemed to match our actual experiences, and many were not even close to being accurate.
There is also the fallacy of averages. Some COVID patients stayed in the hospital/ICU 45 days or more, but that was not the norm. Using average length of stay was thrown off by these outliers and became a flawed statistic as a result.
Counting PPE is hard. Scarcity leads to sourcing that is atypical. Utilization is difficult to track and the standards continued to evolve, so utilization in March did not match utilization in April, even if the same number of staff and patients were in play.
Finally, trying to map deliverables to federal and state requirements is challenging. Typically, one would breakdown data needs into three categories: on demand, daily summary and weekly summary. It was very difficult to glean which artifacts were required at which interval. Sometimes something that really should have been on demand was published daily even though it was readily available on demand if anyone wanted to look at it. Sometimes things that were available on demand really weren’t needed more than once a week.
Gene Thomas. CIO of Memorial Hospital at Gulfport (Miss.): Turnaround times are a huge problem. Many results come back in paper (fax) format, which is sad and ridiculous in a digital healthcare world. That is not standard results data format.
Jim Feen. Senior Vice President and CIO of Southcoast Health System (New Bedford, Mass.): This has to start with better, more granular data definitions between HHS and the contributing hospitals and/or states. Acknowledging that when we don’t have complete understanding of new metrics or measures, introduce redundancy or conflicts in requested data, this adds significant time and overhead to an already very complex process for COVID-19 reporting for everyone involved.
Bob Foster. Senior Vice President and CIO/HIPAA security officer at South Georgia Medical Center (Valdosta, Ga.): I think the biggest issue is a lack of clarity, standardization, and definition in the data being asked for and collected. This causes discrepancies in the data collected resulting in distrust in the people and entities consuming the data and finger-pointing by the various agencies and entities reporting the data. We need to think like data scientists and be clear in our ask, our definition of data elements and the data we are asking for. There needs to be a single standard that agencies are adhering to. Variation lends itself to distrust and opportunity to discredit the data.
Ash Goel, MD. CMIO of Bronson (Kalamazoo, Mich.): Changing defined data points and streamlining the process with open access to geographic and regional data will be very helpful.
Jordan Tannenbaum. Vice President, CIO and CMIO of Saint Peters Healthcare System (New Brunswick, N.J.): Two places to start:
1. Decentralization of reporting. Reporting of results had to be done at the state level to the DOH, as well as reporting of hospitalized case counts, ICU patients, patients on vents and deaths and other information. Similar but not quite the same information was required to the CDC through normal channels, as well as the portal that was stood up. For New Jersey hospitals, much of this information was centralized in a portal stood up by the New Jersey Hospital Association in conjunction with the NJ DOH. The data from this portal was then distributed upstream to meet federal data requirements, thus hospitals had one central reporting portal.
2. Data definitions. We were dealing with a new disease, and defining the disease took some time. The primary goals of data collection were to understand the prevalence of disease in the community and the impact on hospital bed capacity, ICU capacity, and ventilator availability. Mortality data was also critical to the impact assessment. However, data definitions were inconsistent across various reporting portals, resulting in dissection of the hospital data into several different frameworks. For example, some data definitions defined a patient as COVID positive, while others included patients with suspected disease in spite of a negative test.
More articles on health IT:
10 concerning trends in health IT
Texas Medical Center’s COVID-19 dashboards prove challenging to read and design
Resist the urge to ‘go back to normal’: Mayo Clinic CEO sees digital health, virtual hospital as keys to improving access to healthcare
© Copyright ASC COMMUNICATIONS 2020. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.