Retained surgical items (RSIs) remain among the most preventable yet persistently recurring surgical “never events.” Despite decades of implementing process checklists, counting protocols, and technology solutions, surgical items continue to be left inside patients across the country.
The question “How do RSIs happen?” is no longer sufficient. Instead, we must ask: “How, where, and why are our safety nets failing us?”
In this exclusive three-part article, the University of Michigan Health’s Periprocedural Safety and Quality Improvement (PSQIP) team shares how it conducted an analysis to find answers to some of these questions and what their findings reveal about counting discrepancies, imaging decisions, and opportunities to strengthen patient safety in the operating room.
To evaluate the effectiveness of existing safety nets, the PSQIP team began by examining the final safeguard in the retained surgical item (RSI) prevention process: the use of intraoperative X-ray imaging to locate potentially retained items. From the electronic medical record (EMR), the team extracted a list of all X-rays ordered with the indication “Retained Surgical Item.” Between August 2021 and June 2025, a total of 1,537 intraoperative X-rays were ordered for this purpose.
Our dataset revealed that the system averages approximately 32 RSI X-rays per month, corresponding to one to two events per day. While AORN1 reports that one in every 10,000 procedures results in an actual RSI, our data indicates that 64 out of every 10,000 procedures are at risk for RSI.
The analysis began with a closer examination of the reasons for placing imaging orders. Of the 1,537 intraoperative X-rays, 53% were initiated in response to a surgical count discrepancy. Notably, an additional 13% were ordered because baseline counts were not performed in accordance with protocol, rendering the initial counts unreliable. Together, these two factors accounted for approximately two-thirds of all intraoperative X-ray orders. The remaining third were obtained as a precautionary measure to confirm the absence of RSIs.
Next, we explored whether recurring patterns were evident based on the items involved, the clinical service, or the procedures performed.
None of this is surprising, as each of these services and procedures typically involves a large number of sponges, small needles, and screws, thus increasing the complexity of an already complex count process.
We next turned our attention to how often intraoperative X-ray imaging successfully identified a missing item. Of the 1,537 X-rays performed, only 63 detected a missing item, yielding an overall positivity rate of 4.1%. However, closer examination revealed that 23 of these 63 positive cases involved known device fragments – items the surgeon was already aware had been retained within the operative field and for which imaging was obtained to determine whether retrieval was feasible or whether the fragment could safely remain in place.
After excluding these cases, only 40 previously unknown missing items were identified on X-ray, resulting in an adjusted positivity rate of 2.6%. Importantly, most of these items were located and removed prior to skin closure and therefore did not meet the criteria for RSIs. Additionally, among patients with negative intraoperative X-rays, postoperative imaging also failed to identify retained items, reinforcing the accuracy of intraoperative radiologic interpretation.
When we examined how often intraoperative X-rays successfully identified specific missing items, clear differences emerged. Surgical sponges were the most frequently detected item, with imaging locating 25% of missing sponges. By contrast, suture needles were identified in just 2% of X-ray examinations. We attribute the higher detection rate for sponges, in part, to the use of a non-RFID electronic sponge counting system. Since its implementation in 2010, the institution has seen a marked reduction in both retained sponges and X-rays ordered to locate missing sponges. We hypothesize that the backup count verification provided by this adjunct technology improves count accuracy, making discrepancies more likely to reflect a true missing item rather than a counting error.
When we evaluated the effectiveness of confirmatory intraoperative X-rays in higher-risk situations, such as emergent procedures or spine cases with high instrument counts, the diagnostic yield dropped markedly. Of the 580 X-rays obtained in these contexts, only two missing items were identified, corresponding to a positivity rate of just 0.3%. Both items found were surgical sponges discovered during emergent cases in which baseline counts could not be performed.
*Excludes X-rays for imaging in lieu of and for device fragments, leaving a denominator of 1,168 X-rays
In conclusion, our analysis demonstrated that intraoperative X‑rays were effective in identifying surgical sponges and other large, radiopaque items. In contrast, they were considerably less effective as a safety net for locating smaller items, such as suture needles and screws. While our analysis showed that an initial negative intraoperative X‑ray was consistently followed by negative subsequent X-rays, this result should not be interpreted as definitive evidence that an item was never retained. Rather, it indicates that no item was detected within the limited images obtained or identified by the interpreting radiologist at the time.
References:
In Part I, we focused on unpacking the quantifiable insights gleaned from our analysis of intraoperative RSI X‑ray data. In Part II, we step further back along the safety‑net continuum to examine the surgical count process itself, and the people behind those counts.
When a surgical count discrepancy occurs, staff are required to submit a Patient Safety Report Form (PSRF). Between August 2021 and June 2025, 1,195 PSRFs were filed. These reports allow staff to document contributing factors associated with each event. Review of the data identified two dominant themes: communication breakdowns during breaks or shift changes (cited in 44% of reports), and failure to adhere to standard practice, (cited in 24% of reports).
To deepen our understanding of these findings, the PSQIP team conducted a targeted survey of operating room nurses and scrub technologists and interviewed radiologists and surgeons. This qualitative inquiry aimed to better understand perceptions of how retained surgical items occur and to identify common themes across disciplines.
Nursing and scrub technologist perspectives
An electronic survey was distributed to all nursing and scrub technologist staff across our six operating room locations. Among the 144 survey respondents, 98% believed that incorrect counts occur between zero to two times per week. This perception stands in stark contrast to our X-ray data, which shows that such events occur one to two times per day.
When asked what they perceive to be the top contributing factors to count discrepancies, respondents identified the following:
Overall, participants described persistent challenges with reliably performing the count process and tracking item location throughout a procedure. Many also cited reluctance among surgeons to pause the operation to verify that an item was not retained, along with frustration due to prolonged delays in obtaining radiology interpretation and clearance.
Radiology perspectives
Radiologists and radiology technologists play a critical role in detecting RSIs through diagnostic imaging. However, interviews with two radiologists and three technologists revealed several concerns:
These challenges, compounded by a fast-paced work environment, contribute to heightened anxiety among radiology staff regarding the possibility of misdiagnosis.
Surgeon perspective
From the surgeons’ perspective, many assert that in “99% of cases,” their clinical suspicion is that the missing item is not retained within the patient (our data indicates that it is closer to 97%). As a result, they may feel justified in continuing the procedure, even when nursing staff recommend pausing to verify that no items have been left behind.
During interviews, surgeons shared the following insights:
Frequent miscounts, coupled with low X-ray positivity rates and a prevailing belief that the item is unlikely to be in the patient, have fostered skepticism among surgeons about count discrepancies. As confidence in the count process erodes, so does the adherence to established policies and procedures.
This analysis highlighted the critical role of coordinated multidisciplinary teamwork in preventing RSIs. However, when trust erodes within any single discipline or step in the process, its effects quickly cascade across the broader system. Ultimately, failure in one element of the process compromises the integrity of the entire safety net.
Parts I and II focused on understanding the circumstances surrounding surgical miscount events. In Part III, we shift the lens to examine the broader impact of miscounts on the health system.
To do this, we compared miscount cases with a control population by compiling all surgeries performed during the same timeframe that did not involve an X-ray for a suspected RSI. These cases were then compared with surgeries in which an X-ray was ordered, enabling assessment of the operational and clinical consequences of missing‑item events.
Compared with the control population, cases involving a missing item were associated with an average increase of 49 minutes of OR time and 48 additional minutes of anesthesia exposure. These events also caused downstream delays, disrupting subsequent cases and overall OR flow and staffing schedules.
Staff further described workflow disruptions not fully captured in administrative data. Anecdotally, respondents reported increased OR traffic as additional personnel were called in to locate missing items. The announcement of a miscount often heightened frustration and anxiety among surgical team members, further compounding operational impact.
Timing amplified these challenges. Nearly 46% of incorrect counts occurred after hours (between 5:00 p.m. and 7:00 a.m. or on weekends), even though after-hours cases accounted for only 18% of overall surgical volume. This imbalance highlights heightened system vulnerability during periods of reduced staffing and limited resource availability.
From a cost perspective, we estimate that each miscount event requiring an intraoperative X-ray carries an average financial impact of approximately $2,720.76. This estimate is based on our institution’s average operating room cost of $35 per minute, plus the cost of obtaining the X-ray (not billed to the patient) and additional expenses for faculty and anesthesia time, derived from the literature. These costs were multiplied by the average additional time observed in our dataset.
This estimate assumes that a single X-ray image was obtained and that only one faculty member remained for the additional duration of the case. It does not include any staff overtime incurred due to prolonged case length, suggesting the true financial impact of miscount events may be underestimated.
| Cost per miscount event | |||
| Cost of X-ray | $200 per image | X 1 image | = $200.00 |
| Additional OR time | $35 per minute | X 49 minutes | = $1,715.00 |
| Additional Anesthesia Time1 | $8.62 per minute | X 48 minutes | = $413.76 |
| Additional Surgeon/ Resident Time2 | $8.00 per minute | X 49 minutes | = $392.00 |
| Total system cost per miscount event | = $2,720.76 | ||
In applying these figures to the 815 X-rays performed for incorrect counts and the 203 obtained due to unreliable baseline counts (totaling 1,018 potentially avoidable X-rays) the health system incurred an estimated loss of $2,769,733.68 over the 48-month study period, or an average cost of $57,702.79 per month.
Beyond the staff psychological and hospital financial consequences of miscount events, there are meaningful implications for patient outcomes. A 2017 study by Cheng et al.4 demonstrated that the risk of surgical site infection (SSI) increases by 17% for every additional 30 minutes of operative time. Similarly, a 2018 meta‑analysis5 reported a 14% increase in the likelihood of postoperative complications for each additional 30 minutes spent in the operating room.
The magnitude of risk becomes even more pronounced with longer delays. A study published in the Journal of the American College of Surgeons6 found that each additional hour of operative time was associated with a 14.1‑fold increase in the risk of superficial SSI, a 14.4‑fold increase in organ‑space SSI, a 3.6‑fold increase in renal failure, and a 16.6‑fold increase in sepsis. Together, these findings highlight how incremental extensions in operative duration, such as those caused by unresolved miscounts, can translate into substantial and avoidable patient harm.
Overall, this study underscores the fragility of current safety nets designed to prevent retained surgical items. When developing improvement interventions, the Hierarchy of Intervention Effectiveness6 emphasizes prioritizing system‑level solutions over those that rely on individual performance. Yet many current RSI prevention protocols remain highly dependent on human vigilance and behavioral compliance, rendering them inherently vulnerable. Sustainable prevention will require a deliberate shift toward more reliable, system‑based approaches that reduce reliance on individual memory, judgment, and workarounds.
Successful systems are intentionally designed to anticipate human limitations and provide built‑in protections that prevent error before harm can occur. So, where do we go from here?
Sree Ponnaluri-Wears, MPH, CIC (sponnalu@med.umich.edu) is manager, PSQIP, with University of Michigan Health.
References