Abstract

METHODS

RESULTS

All 16 (100%) invited institutions participated. After excluding 84 responses (two consultants and 82 insufficient completion), a total of 551 of 1,266 e-mailed HCWs completed the survey (44%, range of 13–107 responses per site). The median site response rate was 60% (range 17–83%). By role, there were 192 physicians (35%), 173 nurses (31%), 134 RTs (24%), and 52 APPs (9%). Respondents had worked in the PICU setting for 0–5 (n = 227, 41%), 6–10 (n = 146, 27%), 11–20 (n = 113, 21%), or greater than 20 years (n = 65, 12%).

PICU characteristics are presented in Table ​Table11. Across the 16 units, there was a median number of 34 beds (IQR 25–39) with 1,949 annual admissions (1,492–2,046), and 20% of admissions received invasive mechanical ventilation. Thirty-eight percent had cardiac surgery patients, 88% provided extracorporeal membrane oxygenation, and 75% had stem cell transplant and solid organ transplant patients. A few units had previously provided education (n = 3) or guidance (n = 2) for respiratory culture practices. There was variability in laboratory processing practices with 69% having some type of laboratory rejection criteria (specimens consistent with saliva not worked-up [n = 7], specimens repeated within 0–3 days not worked-up [n = 4], or not repeating antibiotic susceptibility testing within 3–7 days [n = 6]).

TABLE 1.

Participating Site Characteristics (n = 16)

Size of Unit	Median (IQR)
No. of ICU beds	34 (25–39)
Annual Admissions 2019	1,949 (1,452–2,046)
Annual Admissions with ventilation 2019	382 (296–614)
Antibiotic stewardship/infection control support	No. Sites (%)
Conduct Surveillance for VAE or VAP	9 (56)a
Antimicrobial stewardship team supports ICU	16 (100)
Treatment guidelines for ventilator-associated infections	4 (25)
Prior education for indications of respiratory cultures in unit	3 (19)
Prior algorithm/guidelines for respiratory cultures in unit	2 (13)
ICU participates in national quality improvement collaboratives	12 (75)
IPSO	8 (50)
SPS	10 (63)
Clinical staff and training programs
Pediatric residents	15 (94)
Emergency medicine residents	14 (88)
Pediatric critical care fellows	13 (81)
Critical care attendings have overnight in-house call	16 (100)
Advanced practice providers (NPs and PAs)	16 (100)
Respiratory therapists specifically assigned to unit	11 (69)
Open unit	6 (38)
Patient populations
Medical patients	16 (100)
Surgical patients	16 (100)
Neurosurgical patients	15 (94)
Cardiac surgery patients	6 (38)
Nonconventional ventilation modes	15 (94)
Extracorporeal membrane oxygenation	14 (88)
Hematopoietic stem cell transplants	12 (75)
Solid organ transplants	12 (75)
Congenital diaphragmatic hernia	6 (38)
Microbiology laboratory restrictions of respiratory cultures
No rejection criteria or restrictions	5 (31)
Laboratory has some type of stewardship criteria	11 (69)
If specimens are consistent with saliva, no further work-up	7 (44)
Rejects specimens repeated in certain number of days	4 (25)b
Do not repeat susceptibility testing within certain number of days	6 (38)c

Open in a separate window

^aBreakdown of surveillance included ventilator-associated pneumonia (VAP, n = 5), VAEs (n = 7) and other ventilator-associated infection definitions (VAI, n = 1).

^bResponses from sites included: 0, 0, 2, 3 days.

^cResponses from sites included: 3, 3, 3, 3, 7, 7 days.

IPSO = improving pediatric sepsis outcomes, NP = nurse practitioner, PA = physician assistant, SPS = solutions for patient safety, VAE = ventilator-associated event, VAP = ventilator-associated pneumonia.

Ordering Indications

Attendings (86%), fellow physicians (82%), and APPs (69%) often made the decision to order a respiratory culture, whereas resident physicians (43%), nurses (5%), and RTs (5%) less often made the decision. The top most common clinical indications for obtaining respiratory cultures were changes in the description of secretions (33%), new fever (30%), persistent fever (14%), and increased ventilator/oxygen requirement (13%). Figure ​Figure11 presents the perceived likelihood to obtain cultures for different clinical indications. Respondents indicated cultures were most likely to be obtained for patients with new fever and increased ventilator/oxygen support (90%), new fever and increased secretions without increased ventilatory support (85%), concurrent with blood and urine cultures as a “pan culture” (76%), request from consulting service (76%), and persistent fever without other symptoms (72%). Across hospitals, the scenarios with the most variability in likelihood to obtain cultures were fever with inflammatory laboratory changes (e.g., elevated WBC or C-reactive protein) (median 67%, IQR 54–79%), isolated change in color or thickness of secretions (67%, 54–78%), isolated increase in secretions (53%, 40–62%), laboratory changes without respiratory changes (49%, 38–57%), and isolated new fever (48%, 38–59%).

Open in a separate window

Figure 1.

Survey responses: likelihood that respiratory cultures would be obtained for different clinical indications. Box and whiskers plot showing the proportion of respondents across sites who strongly agree or agree that cultures are likely to be obtained in the clinical indications shown on the y-axis. The box indicates the median and IQR, the whiskers indicates the next value within 1.5× IQR, and the points show the outliers.

Perceptions About Respiratory Culture Practices

Figure ​Figure22 presents the agreement with statements about respiratory cultures. The majority of respondents felt confident about indications to obtain respiratory cultures and that cultures would be helpful to guide antibiotic therapy (80%). Most were aware that medication changes may influence secretion characteristics (84%). Nearly half of respondents did not anticipate that cultures from endotracheal tubes are likely to grow bacteria (42%). Less than one third believed cultures were always indicated for ventilated patients with a new fever (27%) or increased secretions (28%).

Open in a separate window

Figure 2.

Survey responses: healthcare worker attitudes and perceptions about respiratory culture practices in general and practices in their PICU. Box and whiskers plot showing the proportion of respondents across sites who strongly agree or agree with statements about respiratory cultures in general and respiratory culture practices in their PICU. The box indicates the median and IQR, the whiskers indicates the next value within 1.5× IQR, and the points show the outliers.

Regarding perceptions of their local PICU practices, a median of 24% of respondents perceived clinicians were likely to conduct a physical examination, 49% likely to review the patient’s microbiology history, and 58% likely to review other clinical data prior to ordering a respiratory culture. Only 30% perceived their unit obtained cultures too frequently. Most respondents agreed they would find clinical decision support (CDS) helpful (79%), anticipated they would follow CDS (82%), and that CDS would align the ICU team and consulting teams (82%).

Barriers to Changing Practices

Respondents anticipated the most likely barriers to changing practices would be reluctance to change practices (70%), differing opinions of consulting services (64%), and clinician concern for missing a diagnosis of VAI (62%) (Fig. ​Fig.33). Less likely barriers included insufficient evidence that respiratory cultures could be reduced safely, and workflow challenges, such as patient workload, conducting physical examinations, or reviewing clinical data.

Open in a separate window

Figure 3.

Survey results: anticipated barriers to changing respiratory culture practices. Box and whiskers plot showing the proportion of respondents across sites who strongly agree or agree with potential barriers to changing respiratory culture practices in their PICU. The box indicates the median and IQR, the whiskers indicates the next value within 1.5× IQR, and the points show the outliers.

DISCUSSION

In this multicenter survey of PICU HCW perceptions about respiratory culture practices, we identified variations in reported specimen collection practices and indications to obtain cultures across hospitals. Importantly, participating hospitals, similar to national patterns, had disparate respiratory culture rates (4). Respiratory cultures were consistently perceived to be obtained for fever and increased ventilatory needs, or for fever and secretion changes. Obtaining cultures for isolated fever and as part of a pan culture concurrent with urine and blood cultures was also common at some institutions. Perceived indications to order cultures that varied the most across hospitals correlated with higher respiratory culture rates, in particular, persistent fever without respiratory changes, fever, and laboratory changes without respiratory symptoms and pan cultures. These correlations strengthen in a sensitivity analysis limited to clinicians’ responses (data not shown). These associations support the concept that culture rates may decline if respiratory cultures were deferred in patients without respiratory tract symptoms.

The method of collecting respiratory secretion specimens impacts culture results and thus interpretation. Saline lavage was a common practice, but no centers reported a specific protocol regarding instilled saline volume. Notably, respiratory therapy guidance does not recommend saline lavage prior to suctioning as it may be harmful to patients without benefit (21–23). The addition of saline also dilutes secretions impacting cell counts, Gram-stain interpretation, and bacterial concentrations in cultures. Inline endotracheal aspirate or open suction from tracheostomies was the most common sampling approach, but this survey did not clarify if HCWs routinely use new vs. existing inline suction catheters. Similar to prior reports, we identified inconsistency in microbiology laboratory sample quality rejection criteria and variation in evaluation of repeat samples submitted within a certain number of days (9). These findings support a need for optimized secretion sampling and processing techniques.

Single-center studies have described within-hospital variability of respiratory culture utilization (5, 17, 24–27). Most survey respondents perceived variability in indications to obtain respiratory cultures between different clinicians and patient populations within their PICU illustrating across and within-site variability in respiratory cultures. Interestingly, clinicians were more likely to perceive variability and overuse compared with nurses and RTs. These survey results suggest there may be tension or an incongruency between clinical knowledge and practice habits as a contributing factor to practice variability. Although many HCWS reported confidence in indications for cultures, many indicated that cultures are likely to be obtained for isolated change in secretions, isolated fever, or pan cultures while reporting that not all ventilated patients with new change in secretions or new fever without associated symptoms should get a respiratory culture. Among PICUs with higher rates of use, there was not a consistent perception that cultures were obtained too frequently. These patterns suggest that PICU clinicians and/or individual units may have developed practice habits that are normalized but not optimized within their setting. Building on prior studies, our survey demonstrates a lack of consistency in respiratory culture practices across hospitals and an opportunity to standardize clinical practice and management of VAI through the development of practice recommendations.

Identification and engagement of key stakeholders are essential components of diagnostic stewardship programs. The survey identified RTs as the most likely to obtain specimens, followed by nurses; however, neither group felt that they were typically involved in the decision to obtain cultures. Nurses and RTs were less likely to think respiratory cultures would have bacterial growth, suggesting opportunities to fill knowledge gaps that respiratory secretions are likely to grow bacteria (6, 8, 28). Importantly, conducting physical examinations and reviewing the clinical data prior to obtaining a new culture was not perceived to be routine, but reassuringly HCWs did not perceive these steps as barriers.

Centers interested in implementing CDS to improve respiratory culture practices may consider the anticipated barriers highlighted in this survey. Encouragingly, clinical data obtainment and workload were not anticipated barriers. Instead, conceptual and psychologic barriers, such as reluctance to change practices and concern for missing VAI, were more likely. Interestingly, HCWs from hospitals with the highest culture rates also viewed CDS less favorably. Single-center studies have identified similar themes of excessive endotracheal aspirate culturing including that clinicians or individual units have developed normalized “default” practices (24), have low thresholds to obtain cultures (29, 30), have high vigilance for sepsis (26), and have fear of missing infection (24–26, 31). These patterns align with psychologic factors relevant for clinical de-implementation, including difficulty breaking habits, and asymmetry of outcomes or loss aversion (e.g., clinician concern for missing a diagnosis feels stronger than the perception of harm from the intervention) (32). The observed patterns also relate to either intrinsic (e.g., regret for errors of omission) and extrinsic (variation in medical practice) drivers of low-value practices in healthcare (33). Future research should consider if additional implementation strategies (e.g., education, provider feedback) are needed in addition to CDS to effectively address these complex clinical decision-making factors.

Limitations of this survey included that the survey asked about perceptions of practices and did not measure actual behaviors. We considered internal validity in several ways: we piloted the survey at another center prior to this study; reduced nonresponse error by aiming for large sample sizes and including HCWs with different roles and perspectives; each site’s responses were reviewed by the site leads. These findings align with single-center studies in which clinicians’ practice behaviors were independently recorded (24) that support that HCWs’ perceptions of practices are representative of actual behaviors. Supporting generalizability, the results demonstrate diversity in respiratory culture rates and perceived practices, the hospitals are geographically dispersed in the United States and most U.S. PICU beds are within academic centers (34). However, this study may not represent all U.S. PICUs, and future studies should consider prospective assessment of clinical practices and include community-based settings.

CONCLUSIONS

This multicenter survey demonstrated variability in perceived respiratory culture practices and attitudes across PICUs and identified clinical patterns that may be associated with higher respiratory culture use. These findings support a need to build consensus and highlight opportunities to optimize respiratory culture practices.

ACKNOWLEDGMENT

We appreciate the engagement and responses from all contributing respondents to make this study possible.

Supplementary Material

Footnotes

REFERENCES