Point-of-Care Verification of Blood Culture Volume in Neonates: A Feasibility Trial

The management of neonates with suspected or proven bacterial infection is one of the most common clinical problems that neonatal providers deal with in their everyday practice [1]. A blood culture remains the gold standard for the diagnosis of sepsis [1]. The blood volume cultured is directly correlated with its positivity [2]. At present, the minimum recommended volume of blood submitted for culture in neonates is 1 ml [1]. This volume results in almost 100% sensitivity, even in cases of low level bacteremia, while inoculating 0.5 ml of blood decreases the sensitivity of the test by as much as 40% [3,4].

It has been proven by numerous studies that the volume of blood obtained for culture in neonates is often inadequate and below the recommended volume [5,6]. Connell et al. found that nearly one half of blood cultures drawn from infants and children were inadequate submissions, and therefore were unable to exclude bacteremia reliably [6].

In recent decades, many quality improvement projects have been undertaken to define interventions effective in improving the process of blood culture submission [2,6,7,8,9]. Continuous monitoring of blood volumes, followed by feedback to blood collectors, has been considered to play a significant role in projects that achieved the greatest improvement [2,7,10].

Determining the pre- and post-inoculation weights of blood culture bottles is the only method applicable to the unique newborn population [6,11]. To the best of our knowledge, this method, when used in existing trials, has consisted of measurements performed by the laboratory rather than at the bedside, with periodic rather than instant feedback on the results to blood collectors. We speculate that this method, if used as a bedside tool, could be highly motivating in collecting an adequate amount of blood for culture and could provide excellent information on the diagnostic accuracy of the test in every clinical situation. Additionally, when the calculated blood volume does not match the recommended value, an attempt can be made to repeat sampling. On the other hand, the increased workload of the front-line staff that this method creates may prevent its adoption into everyday practice. We performed this study to address the question of whether point-of-care verification of blood culture volume could be incorporated into routine clinical practice in a sustainable manner.

When reporting the trial results, we followed the guidelines included in the Consolidated Standards of Reporting Trials 2010 statement: extension to randomised pilot and feasibility trials [12].

During the study period, 244 blood samples were collected for culture, out of which 205 were weighed (84.0%, CI₉₅ [78.8% to 88.4%]). This high proportion remained stable throughout the study period, except for one month (Fig. 1). Of those weighed samples, 196 (95.6%, CI₉₅ [91.8% to 97.9%]) were subjectively assessed by blood culture collectors in terms of volume of blood that was drawn. There was a correlation between the time of blood culture sampling and the act of measuring the blood volume collected for culture. The proportion of blood cultures that were not subjected to weighing was significantly greater during the night shifts than the day shifts (24.0% vs. 12.5%; p = 0.039). Similarly, the proportion of blood culture samples that were not weighed was significantly greater during night shifts and/or weekends versus day shifts and/or workdays (22.6% vs. 10.9%; p = 0.022).

Figure 1.

Patients’ birthweight and blood sampling site distribution within the study group is presented in Table 1. Of the 205 blood cultures with calculated volume, 24 (11.7%) yielded positive results. There were 15 (7.3% of all weighed samples) true-positive blood cultures and 9 (4.4%) false-positive blood cultures (Table 1).

The percentage of adequate volume samples was 87.3%, CI₉₅ [82.0% to 91.5%]. The mean blood volume submitted for culture calculated for the study period was 1.52±0.58 ml. Out of 196 samples visually assessed for collected volume, 5 (2.6%) had an estimation of <1 ml, which was true in 4 cases. The remaining 191 samples (97.4%) were estimated as containing ≥1 ml of blood, which was true in 169 cases (88.5%). Of the 26 inadequate volume samples, 22 were deemed as adequate based on subjective assessment (84.6%, CI₉₅ [65.1% to 95.6]). On the other hand, of the 170 adequate samples, there was 1 sample classified subjectively as inadequate (0.6%, CI₉₅ [0.01% to 3.2%]). The study flowchart is presented in Supplemental Figure 1. The comparison of blood volume measurement versus blood volume estimation is presented in Table 2.

The blood volumes submitted for culture varied between birthweight categories. Post-hoc analysis confirmed that the volume of samples obtained from patients with a birthweight >2500 g (1.68±0.70 ml) was significantly higher than that from patients weighing ≤1500 g (1.38±0.42 ml) (Supplemental Table 1).

The source of blood samples also had a significant impact on the volume of blood that was drawn (p = 0.006). Post-hoc analysis confirmed that the mean volume obtained through umbilical catheters (1.20±0.28 ml) was lower than the volume of cord blood samples (1.72±0.78 ml) or the volume of samples collected from peripheral vessels (1.45±0.40 ml) (Fig. 2, Supplemental Table 2).

Figure 2.

Use of the umbilical cord as a source of blood for culture compared to the peripheral vessel did not result in an increased rate of false-positive cultures [4/72 (5.55%) vs. 5/112 (4.46%); p = 0.739] or inadequate submissions [11/72 (15.27%) vs. 11/112 (9.82); p = 0.379]. There were also no significant differences between the proportion of pathogens and contaminants grown with adequate compared to inadequate blood culture volumes (Supplemental Table 3). Blood cultures that yielded true-positive results with inadequate volumes contained > 0.9 ml of blood.

The results of a qualitative questionnaire conducted among medical personnel (14 doctors and 22 nurses) involved in the study are presented in Table 3. Blood culture collection was perceived as an easy procedure by 31% of personnel, while 58% described it as a moderately difficult, and 11% as difficult. A total of 83% of responders answered that verification of blood culture volume by weighing is necessary (for 75%, it was not a substantial impediment; for the remaining 8%, it was a substantial impediment). Two out of 36 responders (5.6%) answered that weighing blood culture samples is unnecessary. There were no significant differences in the assessment of the difficulty of blood culture collection and the necessity of blood sample measurement between doctors and nurses.

Bacterial sepsis is still one of the most common clinical problems faced by neonatal care providers. At present, a blood culture remains the gold standard for diagnosing neonatal sepsis, provided that an adequate volume of blood is collected. The phenomenon of “culture-negative” sepsis is likely attributable to falsely negative blood culture results due to insufficient volumes sampled [3]. Harewood et al. demonstrated that despite educational intervention, 39% of 145 cultures from patients less than 1 month of age had a blood volume calculated as 0.0 ml [15]. In such circumstances, negative culture results are obviously worthless, either from a clinical or economical perspective.

Our study showed that point-of-care verification of blood culture volume using a scale was successfully incorporated into routine neonatal practice. The high percentage of blood cultures with calculated volume was observed during the entire study period except for during one month. In November 2020 our unit experienced a shortage of medical staff, including research staff, due to COVID-19, which we believe was responsible for the poor compliance with the intervention. Moreover, the vast majority of medical staff involved in the study reported the intervention as necessary and not as an impediment. All the above findings support the feasibility of conducting a definitive trial regarding the optimization of blood volume submitted for culture in neonates.

We wish to point out that nearly 90% of blood culture submissions in our study contained an adequate volume of blood. We speculate that the specific and immediate feedback to blood culture collectors may support performance in this practical task [16]. The additional benefit offered by our approach was that clinicians taking care of the patients were informed about the potential decreased sensitivity of the results due to suboptimal volume. Conversely, they could rely on negative blood culture results when adequate blood volume was collected. Such information is not available with quality improvement projects based on periodic feedback.

The overestimation of the collected blood volume (understood as calculated volume smaller than estimated) took place in 34% of cases (67/196), whereas the remaining 66% of samples were underestimated (129/196). This was not the case in the group of blood cultures with suboptimal volumes. In this group, more than 80% of blood samples were overly optimistically estimated as adequate submissions. The observed discrepancy could potentially lead to improper clinical decisions.

It is worth noting that drawing blood for culture from isolated cord segments proved to be not inferior to the other two methods in terms of volume adequacy and the rate of contamination. There is a growing body of evidence that cord blood sampling is a painless and blood-saving alternative to initial blood sampling from neonates [17,18,19]. In their randomised study, Balasubramanian et al. demonstrated that this strategy, combined with other anemia prevention efforts, reduced the need for blood transfusions in the neonatal period [18].

We assume that bedside measuring of blood culture volume is a simple and low-budget intervention that could easily be applied in many neonatal units. To achieve better compliance with the intervention, it is important that pre-weighed bottles be immediately available to staff members. Connell et al. observed a significant loss of weight in capped blood culture bottles with time under normal storage conditions [6]. For this reason, we have tested the blood culture bottles we normally used for a possible loss of weight before commencement of this trial. We have noted only negligible loss of weight in the capped glass culture bottles over a period of six months, and a considerable loss of weight in the plastic bottles. We believe that such a test should be performed in any unit willing to incorporate the studied intervention. Moreover, we have decided to use only glass culture bottles in our definitive trial. The surveyed medical staff has tabled no amendments to the intervention. However, our data showed that adherence to the intervention was lower during the night shifts and/or weekends. These findings harmonize with the evidence of the negative impact of the night shifts on overall work performance among healthcare providers [20]. We will conduct monthly feedback and coaching sessions with the staff from each participating unit to overcome this challenge in the definitive trial.

In conclusion, our study showed that point-of-care verification of blood culture volume using a scale is feasible to deliver. The high uptake of the intervention was sustained throughout the study period. Since we have met the pre-established criterion for success, we plan to proceed with the definitive trial.