Prevalence of Lab-Ordering for Patients Discharged to SNF
Overuse of daily labs (DL), especially CBC and BMPs in hospitalized patients leads to increased costs and utilization of resources, iatrogenic anemia, and patient discomfort. Those awaiting discharge to post-acute care such as skilled nursing facilities (SNFs) may represent an important group of patients at risk of receiving unnecessary DL and labs on day of discharge (LOD) due to frequent non-medical delays in discharge. This study aimed to assess the prevalence of DL and LOD for patients discharged to SNFs. We examined associations of DL and LOD with provider and hospitalization characteristics. We also aimed to reduce over-ordering DLs and LOD in this patient population through several interventions delivered on a hospital medicine teaching service.
Project design:
This study was conducted at an urban academic tertiary medical center on 2 services each consisting of an attending hospitalist and third year resident (PGY3). Each provider is the primary physician for 7 patients; the attending hospitalist supervises the care of all 14 patients. DL and LOD ordering data were extracted for all patients discharged from these services to SNFs from 1/1/2021-12/31/2021. We determined the prevalence of DL and LOD, defined as BMP and/or CBC present every day of hospitalization including discharge. We performed bivariate and stepwise multivariate logistic regression to determine if DL and LOD ordering practices varied between provider characteristics (i.e., attending vs resident providers, hospitalist gender, years of hospitalist experience); or hospitalization characteristics (i.e., patient specialty type, academic quarter, patient LOS, and presence or absence of ICU stay during hospitalization). Subsequently, a retrospective review of the same patient charts from 02/2021 to 06/2021 was performed to establish a baseline of DL and LOD present prior to interventions. Physician reviewers then assessed the appropriateness of DL by reviewing lab values. Lab appropriateness guidelines were agreed upon by reviewers to reduce subjectivity. Several interventions were then enacted. First, emails were sent to hospitalists prior to their rotation asking them to focus on teaching their PGY3s lab stewardship. Next, posters were placed in workrooms reminding providers to avoid ordering DL for patients awaiting SNF placement. Lastly, a lecture was delivered to the PGY3s about lab stewardship. Post-intervention data were assessed by chart review of patients admitted between July 2021 to October 2022 in the same manner as the pre-intervention data.
Preliminary results:
From 1/1/2021-12/31/2021, 165 encounters met inclusion criteria. 91% (150/165) of patients had DL and 73% (121/165) had LOD. 71% (117/165) had attendings as the primary provider and 29% (48/165) had residents. In bivariate analysis, there were no significant differences in LOD by provider characteristics. However, among patients with ICU stay, 94% (n=16) had LOD compared to those without ICU stay (71%, n=105, chi2=4.19, p = 0.041). Likewise, patients on subspecialty services had proportionally more LOD (100%, n=15) than general medicine patients (70%, n=106, chi2=6.0, p = 0.014). In stepwise regression, no one factor was statistically significant, although patients with ICU stay had about seven times higher odds of having LOD (OR=6.98, 95% CI[0.84-57.8, p=0.07).
The baseline data prior to implementing interventions included patients hospitalized from 2/2021-7/2021. A total of 111 encounters met inclusion criteria. Of attending patients with DL, 35% (18/52) were deemed inappropriate, compared to 26% (8/31) of resident patients (p = 0.47). Among those with LOD, 42% (22/53) of attending patients were deemed inappropriate, compared to 33% (10/30) of resident patients (p = 0.49). When comparing these baseline data to post-intervention data, there was no significant difference in lab-ordering practices (Figure 1).
Significance of results:
For patients discharged to SNF, use of LOD and DL was high. There was an association between higher acuity measures (i.e., subspecialty patients and ICU stay) and increased prevalence of LOD and DL, though not statistically significant. Given these findings, variability in lab ordering may be explained by unmeasurable factors not specific to patient- or provider-related characteristics. Our findings suggest efforts should be broad-based, targeting all ordering providers and patients. The educational interventions we implemented did not reduce the amount of inappropriate lab utilization, suggesting additional measures may need to be implemented to reduce LOD.
Plans for future work:
Our team hopes to reduce inappropriate lab-ordering via a “Best Practice Alert (BPA),” an electronic health record (EHR)-based prompt. This prompt was initially designed as a “pop-up” when the EHR detects that a patient with labs is medically ready via a “medically ready” button that is clicked by a member of the healthcare team. The BPA suggests decreasing the frequency of labs when they have been stable for several days and the patient has been designated “medically ready.” We plan to change the design and wording of the BPA to encourage discontinuing DL for these patients. We plan to compare lab-ordering prevalence between the original BPA and the newly designed BPA.
Bloggers
