A retrospective cohort study of wrong-patient imaging order errors: how many reach the patient?

Abstract

Studying near-miss errors is essential to preventing errors from reaching patients. When an error is committed, it may be intercepted (near-miss) or it will reach the patient; estimates of the proportion that reach the patient vary widely. To better understand this relationship, we conducted a retrospective cohort study using two objective measures to identify wrong-patient imaging order errors involving radiation, estimating the proportion of errors that are intercepted and those that reach the patient. This study was conducted at a large integrated healthcare system using data from January 1 to December 31, 2019. The study used two outcome measures of wrong-patient orders: 1) wrong-patient orders that led to misadministration of radiation reported to the New York Patient Occurrence Reporting and Tracking System (NYPORTS) (misadministration events); and 2) wrong-patient orders identified by the Wrong-Patient Retract-and-Reorder (RAR) measure, a measure identifying orders placed for a patient, retracted, and rapidly reordered by the same clinician on a different patient (near-miss events). All imaging orders that involved radiation were extracted retrospectively from the healthcare system data warehouse. Among 293,039 total eligible orders, 151 were wrong-patient orders (3 misadministration events, 148 near-miss events), for an overall rate of 51.5 per 100,000 imaging orders involving radiation placed on the wrong patient. Of all wrong-patient imaging order errors, 2% reached the patient, translating to 50 near-miss events for every 1 error that reached the patient. This proportion provides a more accurate and reliable estimate and reinforces the utility of systematic measure of near-miss errors as an outcome for preventative interventions.

INTRODUCTION

Studying near-miss errors is vital to understanding system vulnerabilities and preventing errors that reach patients and cause harm.[1, 2] Near-miss events are errors caught before reaching the patient.[1, 3–5] Near-miss events and errors that reach the patient are posited to share the same causal pathway,[3, 6–8] estimates of the proportion that reach the patient vary. An accurate estimate of this relationship is needed to inform interpretation of studies using near-miss errors as the outcome.

Limited prior evidence shows relative frequencies of 100:1 to 7:1 of near-miss events to errors that reach the patient.[3, 5, 7, 9–12] The accuracy of these estimates has been limited by biases inherent to the detection methods used to identify cases. Chart review relies on accurate documentation, voluntary reporting vastly underrepresents the number of events, and errors that reach the patient are subject to overreporting bias.[10, 13–16] We sought an estimate using a type of error that could be detected by more reliable measures, including intercepted errors (near-miss events) and errors that reach the patient. Therefore, we focused on a “never event,”[17] wrong-patient misadministration of radiation ordering errors. Imaging errors are among the most common contributors to serious harm events,[18, 19] and 26% of imaging errors occur at the ordering phase.[5, 20]

In New York, a wrong-patient misadministration event must be reported to the New York Patient Occurrence Reporting and Tracking System (NYPORTS).[17] Mandatory reporting has been shown to more accurately capture errors that reach the patient than voluntary reporting.[21] The Wrong-Patient RAR measure is a reliable indicator of near-miss wrong-patient orders.[22] Therefore, these data can be combined to more accurately estimate the proportion of errors that reach the patient.

METHODS

This retrospective study was conducted January 1 to December 31, 2019 using healthcare system data comprising 4 inpatient facilities and 4 emergency departments. Sites used Allscripts Sunrise Clinical Manager (Chicago, IL) as their electronic health record (EHR) system. The affiliated Institutional Review Board approved the study protocol.

The study used two outcome measures: 1) wrong-patient orders leading to misadministration events reported to NYPORTS (misadministration events); and 2) wrong-patient orders identified by the Wrong-Patient RAR measure (near-miss events). Imaging orders, including computed tomography, nuclear medicine, and X-rays, were extracted retrospectively from the data warehouse from 2019. For misadministration events, wrong-patient NYPORTS events were reviewed to determine errors at the ordering phase. For near-miss events, we executed the Wrong-Patient RAR measure, which identifies wrong-patient orders, restricting results to imaging orders with radiation.[23]

We reported wrong-patient orders, and wrong-patient orders per 100,000 imaging orders by patient, clinician, and order characteristics. Age was specified as a dichotomous variable (< 10 vs. ≥ 10 years).[24–26] A multivariable logistic regression was conducted using patient, clinician, and order characteristics. Model calibration was assessed with the Hosmer-Lemeshow chi-square test, and discrimination with area under the receiver operating characteristics curve. We used cluster-robust standard errors and a two-level logistic model with a random intercept at the clinician level to adjust for the nesting of orders by clinicians. Proportion of errors reaching the patient was calculated as misadministration events divided by total wrong-patient orders, with the corresponding exact Clopper-Pearson test. Statistical analyses were conducted using Stata version 15.1 (StataCorp).

RESULTS

A total of 293,039 imaging orders involving radiation were ordered during 1-year. We identified 3 wrong-patient x-ray ordering errors that reached the patient, resulting in harmful additional treatment. Orders were placed by an attending, a resident, and a physician assistant; 2 in the emergency department and 1 in a general medical unit. The Wrong-Patient RAR measure identified 148 wrong-patient imaging orders.

Logistic regression (Table 1) showed wrong-patient orders were significantly more likely among patients aged <10 vs ≥ 10 years (OR 2.3, 95% CI 1.4–3.9) and orders placed by residents (OR 2.1, 95% CI 1.2–3.6) or physician assistants (OR 2.4, 95% CI 1.3–4.3) vs attendings. The two-level logistic regression showed similar results (not shown).

Table 1.

Wrong patient order characteristics

	Imaging Orders*	Total Wrong-Patient Imaging Orders† (2019 Census)	Wrong-Patient Imaging Order† Events per 100,000 Imaging Orders (95% CI)	Odds Ratio (95% CI)
Overall	293,039	151	51.5 (43.6 – 60.4)
Patient age, n (%)
Age greater than 10	267,187 (91.1)	128 (84.8)	47.9 (40.0 – 57.0)	Reference
Age less than 10	26,045 (8.9)	23 (15.2)	88.3 (56.0 – 132.5)	2.3 (1.4 – 3.9)
Patient sex, n (%)
Female	147,713 (50.4)	72 (47.7)	48.7 (38.1 – 61.4)	Reference
Male	145,469 (49.6)	79 (52.3)	54.3 (43 – 67.7)	1.1 (0.8 – 1.5)
Not reported	5 (0.0)	0 (0.0)	-	-
Clinician role, n (%)
Resident	126,971 (43.3)	75 (49.7)	59.1 (46.5 – 74.0)	2.1 (1.2 – 3.6)
Attending physician	65,278 (22.3)	20 (13.2)	30.6 (18.7 – 47.3)	Reference
Physician assistant	65,000 (22.2)	42 (27.8)	64.6 (46.6 – 87.3)	2.4 (1.3 – 4.3)
Nurse practitioner	34,206 (11.7)	13 (8.6)	38.0 (20.2 – 65.0)	1.1 (0.5 – 2.4)
Nurse	297 (0.1)	-	-	-
Other	1,435 (0.5)	1 (0.7)	69.7 (1.8 – 387.6)	2.6 (0.6 – 12.3)
Visit type, n (%)
Inpatient	184,886 (63.1)	97 (64.2)	52.5 (42.5 – 64.0)	1.1 (0.5 – 2.7)
Emergency	86,205 (29.4)	46 (30.5)	53.4 (39.1 – 71.2)	14 (0.6 – 3.4)
Outpatient	22,096 (7.5)	8 (5.3)	36.2 (15.6 – 71.3)	Reference
Modality, n (%)
X-ray	222,115 (75.8)	113 (74.8)	50.9 (41.9 – 61.2)	0.9 (0.6 – 1.3)
Computed tomography	69,970 (23.9)	38 (25.7)	54.3 (38.4 – 74.5)	Reference
Nuclear medicine	1,102 (0.4)	-	-	-

Area under the Receiver Operator Characteristic (ROC) Curve = 0.61 (95% CI 0.57 – 0.65); Hosmer-Lemeshow X² test = 9.53

^*Imaging orders = computed tomography, nuclear medicine, and x-ray

^†near-miss events and misadministration of radiation events.

The overall rate of wrong-patient imaging orders was 51.5 (95% CI 43.6 – 60.4) per 100,000 imaging orders (Table 1). The proportion of errors reaching the patient was 2.0% (95% CI: 0.4–5.7).

DISCUSSION

This study yielded an estimate of approximately 50 near-miss wrong-patient misadministration of radiation orders for every 1 that reached the patient. This ratio provides a more accurate estimate, using two reliable methods of measuring wrong-patient order errors involving misadministration of radiation, which have not been previously used. The estimate reported adds to the limited prior evidence of this ratio.[3, 5, 7, 9, 10] Prior studies have focused primarily on medication errors, and used chart review, voluntary reporting, and direct observation, which have inherent limitations.[10, 13–16] An accurate estimate is crucial to interpret studies using near-miss errors as an outcome to improve patient safety, ultimately reducing patient harm.

Misadministration of radiation is a never event.[17] Preventing these errors due to inadvertent orders on the wrong patient is critical. Intervention studies to reduce wrong-patient orders have shown reductions of 16% to 41%.[23, 27–29] Findings from this study can estimate the number of errors prevented errors in intervention studies using near-miss errors as outcomes.

We found that patients under 10 years have over twice the risk of wrong-patient errors than those over 10, similar to prior research.[26, 30] We also found a more than two-fold risk of wrong-patient errors placed by residents and physician assistants than attending physicians. These findings provide potential targets for future intervention studies.

This study has multiple limitations. First, events identified by the RAR measure are not all wrong-patient orders, making our estimate conservative.[23] Second, the EHR employed interventions (e.g., patient verification alerts) to reduce wrong-patient errors; however, we expect the proportion of near-miss events to errors would remain the same. Third, the frequency of order errors was low; however, these are never events. Lastly, this study was conducted at one institution, limiting generalizability.

CONCLUSIONS

We estimate 50 near-miss wrong-patient orders for every 1 that reaches a patient. These results provide a more precise estimate and can inform the interpretation of intervention studies employing near-miss errors as the primary outcome.