• Open Access

Novel Infrastructure for Sepsis Biomarker Research in Critically Ill Neonates and Children


Correspondence: W. Charles Huskins (huskins.charles@mayo.edu)



Sepsis biomarker research requires an infrastructure to identify septic patients efficiently and to collect and store specimens properly. We developed a novel infrastructure to study biomarkers of sepsis in children.


Patients in pediatric and neonatal intensive care units were enrolled prospectively; enrollment information was stored in a secure, remotely accessible database. Researchers were notified of electronic medical record (EMR) orders for blood cultures (a surrogate for a diagnostic evaluation of suspected sepsis) by a page triggered by the order. Staff confirmed patient enrollment and remotely submitted an EMR order for collection of study specimens simultaneous with the blood culture. Specimens were processed and stored by a mobile clinical research unit.


Over 2 years, 2029 patients were admitted; 138 were enrolled. Staff received pages for 95% of blood cultures collected from enrolled patients. The median time between the blood culture order and collection was 34 minutes (range 9–241). Study specimens were collected simultaneously with 41 blood cultures. The median times between specimen collection and storage for flow cytometry and cytokine analysis were 33 minutes (range 0–82) and 52 minutes (range 28–98), respectively.


This novel infrastructure facilitated prompt, proper collection and storage of specimens for sepsis biomarker analysis. Clin Trans Sci 2012; Volume #: 1–5


Sepsis is defined as the systemic inflammatory response syndrome (SIRS) in association with a suspected or confirmed infection.[1-3] Research to improve prompt recognition and treatment of sepsis in infants and children has focused on identification of novel biomarkers, primarily immune cell surface and plasma markers, capable of reliably distinguishing sepsis from SIRS due to other causes.[4]

Researchers seeking to identify immunological biomarkers of sepsis face several methodologic challenges. First, episodes of sepsis are acute unpredictable events; consequently, researchers need reliable means of enrolling patients at risk for sepsis and promptly identifying those with suspected sepsis early in the disease process, any time of the day or night. Second, the clinical evaluation and treatment of sepsis is an emergency; consequently, researchers must have the capability to collect relevant specimens quickly before empiric treatment is initiated. Third, biomarkers are labile; consequently, researchers need efficient methods for prompt, proper processing and storage of specimens.[5-7]

In the context of a research study of immunological biomarkers of sepsis in critically ill infants and children, we developed research procedures to address each of these methodologic challenges. These research procedures were made possible through new, innovative use of the electronic medical record (EMR) combined with use of the comprehensive resources for clinical and translational research available through the National Center for Advancing Translational Sciences (NCATS)-funded Mayo Clinic Center for Translational Science Activities (CTSA).


Design, study setting, and population

The study was a cross-sectional study nested within a prospective cohort study of SIRS and sepsis among infants and children hospitalized in the 30-bed neonatal intensive care unit (NICU) and 16-bed pediatric intensive care unit (PICU) at the Mayo Clinic Children's Center in Rochester, Minnesota.

Patient screening and enrollment

Study coordinators prospectively screened newly admitted patients for eligibility each weekday using the patient's EMR and by discussion with the medical team, as needed, to clarify the expected duration of ICU stay. The inclusion criterion was an expected NICU/PICU stay of 3 days or longer. The only exclusion criterion was an immunocompromised state, defined as an underlying condition with decreased immune function (e.g., primary immunodeficiency, malignancy) or receipt of immunosuppressive therapy (e.g. chemotherapy, immunosuppressive medications) during the previous 3 months. Study coordinators met face-to-face with the parents/guardians of eligible patients and described the study. The informed consent and assent processes were completed and patients were enrolled as soon as feasible after admission. Screening and enrollment information and scanned copies of the signed consent forms were stored in the Research Electronic Data Capture tool (REDCap)[8] to enable secure, remote access by all members of the research team. The study was approved by the Mayo Clinic Institutional Review Board.

Study specimens

Study specimens consisted of whole blood for flow cytometry analysis (0.5–1.0 mL) and plasma for cytokine analysis (0.5–1.0 mL). Specimens were collected only if it was possible to collect them simultaneously with clinically indicated blood samples to reassure parents/guardians and patients that collection of study specimens would be integrated into existing blood collection procedures and would not involve any additional discomfort or risk. Baseline specimens were collected from all patients as soon as possible after enrollment simultaneously with the next scheduled clinically indicated blood collection, except for infants <1 kg (to limit blood volume collection from these infants). Additional specimens were collected from patients who subsequently had a blood culture collected as part of a clinical evaluation for suspected sepsis, except for patients who had a blood culture drawn within the previous 7 days. These specimens were collected simultaneously with collection of the blood culture. Follow-up study specimens were collected 12–36 hours after the blood culture simultaneously with the next scheduled blood collection, except for infants <1 kg.

Collection and processing of study specimens

We initially asked phlebotomists to notify the research team of patients for whom a blood culture had been ordered, either by telephone or page. However, this system resulted in an unacceptable increase in the phlebotomist's workload and delay while the research team assessed whether the patient was enrolled in the study. An analysis of the first 21 blood culture orders revealed that the research team had been notified of only 13 (62%). Working with Mayo Clinic's Institutional Review Board and Information Technology Systems and Procedures staff, we created a function such that an electronic order for an urgent blood culture (i.e., not a follow-up blood culture) submitted through the EMR for a patient in the NICU or PICU automatically generated a page to the research team's dedicated pager that included the time of the order and the patient's bed space (but not the medical record number for privacy reasons). This was accomplished through custom software developed by Mayo Clinic that is integrated with the General Electric Centricity Enterprise, Enterprise Orders application and Mayo's paging system.

A research team member who received a page regarding a blood culture order remotely accessed the EMR to identify the patient by unit census records and then accessed REDCap to confirm enrollment of the patient. If enrollment was confirmed, the research team member remotely entered an order into the patient's EMR for study specimen collection. The order posted automatically to the phlebotomist's workflow sheet for simultaneous collection with the blood culture.

We utilized the Mayo CTSA Mobile Clinical Research Unit (CRU) to facilitate around-the-clock study specimen processing. After collection of specimens, the phlebotomist notified the Mobile CRU technician on-call. After electronically confirming study enrollment of the patient using REDCap, the technician retrieved the specimens from the bedside, processed them and stored them temporarily under the required conditions in the CRU laboratory. Research team members subsequently retrieved the specimens and stored them in the research laboratories.

Process measures to assess the effectiveness of study procedures

We used process measures to assess the effectiveness of our study procedures in addressing the three methodologic challenges described previously. We assessed the procedures for enrollment of patients at risk for sepsis and promptly identifying those with suspected sepsis early in the disease process by calculating the percentage of patients admitted to the NICU/PICU who were screened and enrolled, the percentage of enrolled subjects subsequently undergoing a clinical evaluation for a new episode of suspected sepsis (defined as an urgent order for collection of a blood culture in the absence of blood culture in the previous 7 days), and the percentage of these patients that research team members were able to identify in real-time in order to collect study specimens simultaneously with the blood culture. We collected these data by recording screening and enrollment information on individual case report forms entered into REDCap and by recording all automated pages for urgent blood orders and comparing them to a complete list of blood culture orders for NICU/PICU patients provided by Mayo Clinic's Department of Laboratory Medicine and Pathology.

We assessed the procedures for collecting study specimens from enrolled patients undergoing a clinical evaluation for suspected sepsis by calculating the percentage of blood culture orders for which study specimens were collected simultaneously.

We assessed the procedures for prompt, proper processing and storage of study specimens by calculating the time between collection and storage. We collected these data by recording the specimen's collection and storage times. We classified the blood culture orders and specimen collections by those occurring during weekday working hours (Monday–Friday 7:00AM-5:00PM), weekday after-hours (Monday–Friday before 7:00AM or after 5:00PM), and weekends (Saturday–Sunday).


Enrollment of patients at risk for sepsis and prompt identification of those undergoing clinical evaluation for a new episode of suspected sepsis

From July 1, 2009 to April 30, 2011, 2,029 patients were admitted to the NICU and PICU (Figure 1). Thirty-eight percent of these patients had an ICU stay of less than 24 hours and were not screened for eligibility. Thirty-seven percent of these patients were excluded from the study, most commonly because their expected ICU stay was 3 days or less, which indicated they were at low risk of developing a new episode of sepsis. Of the 423 (23%) patients eligible for the study, the majority was not enrolled because the study coordinator was unable to meet face-to-face with the parents/guardians, either because the parents/guardians were unavailable or could not be approached. Of the parents/guardians who met with the study coordinator, 81% provided consent. A total of 138 patients (7% of admitted patients) were enrolled in the study yielding a 31% and 25% consent rate among all eligible NICU and PICU patients, respectively. Of the enrolled NICU patients, 13% were less than 1 kg at birth.

Figure 1.

Patient study flow diagram of all critically ill neonates and children admitted to the Mayo Clinic Children's Center between July 1, 2009 and April 30, 2011. *Categories are not mutually exclusive.

A total of 1,186 blood cultures were collected from 464 patients across the entire cohort of NICU and PICU patients: 33% were collected during weekday work hours, 37% were collected during weekday after-work hours, and 30% were collected during the weekend. Among the 138 enrolled subjects, 158 blood cultures were collected after study entry and followed a similar temporal distribution (Figure 2). The automated paging system notified study staff of 94% of blood culture orders across all patients and 95% of blood culture orders for all enrolled patients. The page was typically received less than one minute after the order was issued in the patient's EMR.

Figure 2.

Blood culture orders and study specimen collection for 138 enrolled patients. *Based on random sample of 50 blood culture orders.

Collection of study specimens from patients in whom blood cultures were ordered

The median time between issuance of a blood culture order and its collection was 34 minutes (Figure 2). Nonetheless, among the 158 blood cultures drawn from enrolled patients, specimens were collected successfully in simultaneous fashion in 41 (26%) instances (Figure 2). Among the 117 instances in which a specimen were not collected, the most common reason was that the patient had a blood culture drawn in the previous 7 days. A specimen was not collected due to a failure of study procedures in 33% of situations, most frequently because the research team member did not response to the automated page quickly enough or because the page was not received.

Processing of study specimens

Sixty-two study specimens were collected at baseline, 41 specimens were collected simultaneously with a blood culture, and 22 specimens were collected in follow-up after the blood culture. These specimens yielded a total of 121 whole blood samples for flow cytometry and 116 plasma samples for cytokine analysis. The median time from collection to storage for was 25 minutes (interquartile range [IQR] 16–38, range 0–109, n = 107, 14 missing data) for flow cytometry and 47 minutes (IQR 37–57, range 15–156, n = 112, 4 missing data). Similar turnaround times were achieved for specimens collected with a blood culture (Figure 2).


We successfully addressed three key methodologic challenges in the study of biomarkers of sepsis in critically ill neonates and children. First, because episodes of sepsis are acute unpredictable events, we enrolled critically ill neonates and children prospectively prior to any sepsis episodes. We used a screening procedure involving the EMR and the patient's medical team to efficiently focus our enrollment efforts on subjects with a higher risk of sepsis. Our overall enrollment rates of 31% and 25% in NICU and PICU patients, respectively, were affected primarily by the availability of the parents/guardians to meet face-to-face with our study coordinator. For those parents/guardians available to meet with study staff, our enrollment rate was 81%, which is comparable to rates in large neonatal clinical trials.[9-11] Because patients were enrolled prospectively, we were able to efficiently identify patients who subsequently developed an episode of sepsis and to collect specimens from them, regardless of the timing of the episode.

Second, because the clinical evaluation and treatment of sepsis is an emergency, we devised procedures to collect study specimens simultaneously with clinically indicated blood cultures before empiric therapy was initiated. The paging system reliably notified the research team that blood cultures were being collected from a patient and we were able to access the EMR and REDCap remotely to confirm patient enrollment and to order simultaneous collection of the study specimens. Consequently, despite the fact that the median time between placement of an urgent order for a blood culture and its collection was only 34 minutes, we were able to successfully collect study specimens within this narrow time window. Key features of these novel procedures were that they were independent of any effort by clinicians at the bedside and were integrated with the workflow of clinically indicated management.

Third, because the biomarkers we were studying are labile,[5-7] we developed procedures to process and storage of study specimens promptly after collection at all times of the day and night. The mobile CRU was invaluable in this effort as demonstrated by the excellent collection to storage times for flow cytometry and cytokine analysis specimens.

The transition to EMR promises integration of existing medical information and better workflow but also affords opportunities to seamlessly link translational research with clinical practice.[12, 13] These opportunities include automated notification of study staff of orders, events or laboratory results, secure remote access to patient medical records, and placement of research orders alongside clinical orders for simultaneous processing and collection of study specimens. Additionally, the research infrastructure created through the NCATS CTSA Consortium has generated new ways to recruit and enroll research subjects, collect, store, and access research information and specimens.[14, 15] Indeed, without the innovations to address the methodologic challenges described above, research to identify biomarkers of sepsis in critically ill infants and children in our institution would have been much more difficult, if not impossible. These innovations may have relevance to other translational research conducted in the context of clinical practice.

The study has several limitations. While our screening methods quickly identified eligible subjects without requiring direct contact with parents/guardians, our overall consent rates for this study were only 25–30%. This lower consent rate reflected both limited study coordinator resources (weekday screenings only) and the difficulty in approaching parents/guardians of critically ill neonates or children. Once approached, enrollment rates were high (81%). We used an order for a blood culture (without a history of a previous blood culture in the prior 7 days) as a surrogate marker of a clinical evaluation of suspected sepsis. While this is likely to be a sensitive marker, it is not likely to be very specific. Finally, while we were able to collect many study specimens simultaneously with a blood culture, the very short time between placement of the blood culture order and collection of the blood sample and the difficulty of responding to pages at all times of the day and night, made it difficult for even our relatively efficient study procedures to ensure simultaneous collection of study specimens in all cases.


In summary, we addressed three key challenges involved in conducting sepsis biomarker research in critically ill neonates and children using EMR and CTSA research resources. In this process, we developed several novel research procedures, specifically use of a dedicated study pager linked directly into EMR laboratory ordering system and use of the EMR to remotely order simultaneous collection of study specimens with a clinically indicated blood culture. Combined with the use of REDCap for remote access to enrollment information and signed study consent forms and the Mobile CRU, these procedures facilitated efficient and proper collection and storage of specimens for sepsis biomarker analysis.


We thank Mayo Clinic's Institutional Review Board for their guidance on how to integrate these new methodologies into our study design. We also gratefully thank the Mayo Clinic REDCap support staff (Michael Lin, Joseph Wick), Systems and Procedures staff (Munawwar Khan, Mark Foley, Rachelle House), Mobile CRU staff (Nancy Hawley, Diane Swanson RN), our NICU (William Carey, MD, Christopher Colby, MD) and PICU (Sheri Crow, MD) collaborators, and all other NICU/PICU staff for their assistance in the day-to-day execution of this study.

Sources of Financial Support

This work was supported in part by an intramural grant from the Mayo Clinic Department of Pediatric and Adolescent Medicine, F30 DK084671 from the National Institute of Diabetes and Digestive and Kidney Diseases and UL1 RR024150 from the National Center of for Advancing Translational Sciences. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases, the National Center for Advancing Translational Sciences, or the National Institutes of Health.