Improving Patient Safety Through An Effective Harm Index
ROSEMARY BACZEWSKI, details how her facility is working toward its goal of zero harms.
Most adult Americans are familiar with the healthcare dictum, “First, do no harm.”
Yet as we all know, harm in some form occurs all the time in even the best hospitals and with the best clinicians. It’s always been assumed to be an unavoidable byproduct of care, enabling administrators and clinicians to think in terms of “acceptable levels.”
But if you put a face on that harm – your spouse, your child, your neighbors, etc. – reducing harm to nil takes on greater urgency.
That was the starting point for Greater Hudson Valley Health System’s (GHVHS) implementation of a Harm Index. With three main facilities and dozens of provider offices serving the residents of southeastern New York State, we aren’t just faceless providers; our patients are often people we know personally.
We believed developing a Harm Index – a concept sponsored by the American Hospital Association (Center for Healthcare Governance), Health Research & Educational Trust (HRET), Trustee Magazine and the Partnership for Patient Safety – and the processes around it, would enable us to document and evaluate every patient harm and near-miss. We could then use big data analytics to track trends to determine if our efforts to improve care quality and patient safety were making a difference. We committed to building it in 2014 and launching it Jan. 1, 2015.
The first step was to define exactly what we meant by harm. For that we created a hierarchy that placed standards based on or reported from the Agency for Healthcare Research and Quality (AHRQ), National Quality Forum (NQF) or Centers for Disease Control and Prevention (CDC) at the top level, then the Centers for Medicare & Medicaid Services (CMS), and finally New York State.
Next came populating the Harm Index. We used three sources – our internal occurrence reporting programme (both declared and anonymous reports), our infection control surveillance programme and data that’s fed from our electronic medical records (EMR) system to our clinical analytics software. The analytics are used in real time to answer queries about performance in key areas, such as hospital-acquired pressure ulcers (Stage III, IV and unstageable), injuries from falls (e.g., fractures, concussions), obstetrical adverse events (i.e., serious injuries to mothers and/or neonates during delivery), post-operative venous thromboembolisms (VTE), surgical complications and unexpected deaths.
Once the data has been gathered it can follow one of two workstreams depending on the nature of the incident. The two workstreams ultimately converge, and events are evaluated once a week by a team of experts to determine if there was actual harm based on the previously mentioned standards. If so, the event is added to the Harm Index. If not, it still goes back for peer or quality review, where it may lead to disciplinary action or other remediation.
As we were developing this process, we also revised our internal systems to transition coding from being a finance function to a quality function. Prior to the Harm Index, coders were instructed to put as few codes as possible on the bill to get it out the door as quickly as possible. By putting coding under quality, we were able to focus on ensuring it accurately represented the condition of the patient while he/she was in our facility.
We also worked with our physicians to help them think of care in terms of “care is rendered, care is documented, care is coded and then care is billed.” This change in thinking helped them get on-board with our clinical documentation improvement (CDI) efforts by demonstrating how better documentation leads to better outcomes with a reduction in harm. Additionally, we worked with our clinicians and non-clinical staff to help them learn to embrace big data and the critical information it can provide.
Huge gains in harm reduction
Using our clinical analytics platform, with 2014 as our baseline year, the results were nothing short of remarkable. In the first year of the Harm Index (2015), there was a 35 per cent decrease in total harms. From 2014 to 2018, the total reduction was 50 per cent.
Since our goal is zero harms, however, we dug deeper to see how the trends looked per 1,000 discharges. The numbers here were similar, with a 34 per cent decrease from 2014 to 2015 and a 53 per cent decrease through 2018.
Finally, we used the analytics to measure risk-adjusted mortalities using the 3M All Patient Refined Severity of Illness and Risk of Mortality (APR SOI+ROM) measure. It’s our proxy for externally calculated risk-adjusted mortality. Additionally, any mortality that comes across as less than a seven is reviewed in-depth for optimization of clinical documentation and gaps to clinical care.
The data was once again impressive. We saw a 52 per cent decrease from 2014 to 2015, and a staggering 80 per cent decrease from 2014 to 2018.
We still have a long way to go to get close to zero, and we know the closer we get, the more difficult it will be. But through continuous commitment to education, CDI and the use of big data analytics to identify and understand trends, we strongly believe the Harm Index can help us get to the “First, do no harm” standard with every patient in every site of care.
* Rosemary Baczewski is chief clinical officer at Greater Hudson Valley Health System (now known as Garnet Health).
*Editor’s Note: Charts that originally were included in this article are eliminated for space constraints.