Trends in the Use of Procalcitonin at US Children’s Hospital Emergency Departments Free

We performed a cross-sectional study using data from the Pediatric Health Information System (PHIS), an administrative database that contains ED, inpatient, observation, and ambulatory surgery encounter-level data from 52 not-for-profit tertiary care pediatric hospitals in the United States. The Children’s Hospital Association and participating hospitals are responsible for ensuring data quality and reliability. Portions of the data submission and data quality processes for the PHIS database are managed by Truven Health Analytics. At the time of data submission, data are deidentified and subjected to validity and reliability checks before inclusion in the database. Data from 33 of these hospitals were included in this study; 19 hospitals were excluded from analysis because of data quality issues or for absent or incomplete ED data for the entire study period. This study was approved by the Boston Children’s Hospital Institutional Review Board.

We included ED encounters for children <18 years of age that occurred in a participating institution between July 1, 2016 and June 30, 2020. Because the Federal Drug Administration first approved the PCT assay in 2016, we limited our sample to that period. Encounters for patients who were transferred from other institutions were excluded because diagnostic testing may have been obtained before arrival.

PCT testing was identified as clinical transaction code 316210. For children discharged from the ED, PCT testing was considered to have been performed in the ED. For children who were hospitalized after the index ED encounter, PCT testing performed on the initial or the next day was assumed to have occurred in the ED. Our analyses were conducted at the level of the visit. We did not account for repeated visits by the same patient nor for clustering because we were mostly interested in testing rates across EDs and over time.

Patient-level variables included age (<1 year, 1–4 years, 5–8 years, 9–12 years, 13–17 years), sex, race/ethnicity, type of payor, presence of a complex chronic condition at the visit using a validated classification scheme,²¹  hospitalization status, season of presentation, and the All Patients Refined Diagnosis Related Groups (APR DRG), which is a common classification system for research studies that categorizes encounters into diagnostic groupings according to assigned diagnosis codes, severity of illness, and risk of mortality.²²  Given the large sample sizes and the fact that these analyses are not central to the study, we used the totals and calculated an χ². Then, among all ED encounters with each of these APR DRG designations, we calculated the proportion with PCT testing. To examine the population most likely to have PCT testing considered, we calculated the proportion of ED encounters with PCT testing among those with complete blood count (CBC) testing and each of these 10 APR DRGs.

We calculated the annual rate of PCT testing over the study period for each hospital as well as the all-hospital yearly median rate. We used a test for linear trend (as implemented by the Stata command ptrend) to evaluate whether the slope of PCT testing among ED encounters (measured annually) increased over the study period. With descriptive statistics, we characterized the patient-level demographic features of the sample using frequencies with proportions for categorical variables. We used an χ² test for trend to evaluate whether the proportion of ED encounters with PCT testing increased annually over the study period. To describe PCT testing across hospital-level characteristics, we calculated the proportion of ED encounters with PCT testing within hospital groupings defined by quartiles of hospital-level annual ED volume, admission rate, and payor mix.

We identified the 10 most common APR DRGs among ED encounters with PCT testing based on absolute number. Then, among all ED encounters within each of these 10 APR DRGs, we calculated the proportion with PCT testing. We also calculated the PCT testing proportion among all ED encounters with each of these 10 APR DRG categories that also underwent CBC testing as a proxy for patients who were ill enough to require a blood draw.

To examine hospital variability in PCT testing, we calculated hospital-level PCT testing proportions among all ED encounters, as well as among the subset with any of the 10 most common APR DRG categories. To further examine hospital-level variability in PCT testing, we ranked hospitals according to their PCT testing proportion within each of the 10 most common APR-DRG categories and divided the resulting rankings into deciles. We depicted the hospital-level decile rankings for each ARP-DRG category as a heat map, with darker colors indicating higher levels of PCT testing.²³  All tests were 2-tailed, and α was set at 0.05. We are unable to use corresponding letters from Fig 2 for Fig 3 because each graph is ranked by a different metric and the order of hospitals differs from graph to graph.

Data analyses were conducted by using Stata SE version 16.0 (StataCorp, College Station, TX).

We included 9 666 177 ED encounters for children presenting to one of the EDs between July 2016 and June 2020. A total of 85 958 ED encounters had PCT testing (0.89%). Patients with PCT testing were younger and more likely to be white, non-Hispanic, have a complex chronic condition, and have private insurance (Table 1) when compared with patients with visits in which PCT testing was not performed. Children with PCT testing were also more likely to be hospitalized, admitted to an ICU, and receive antibiotics than children who did not have PCT testing performed.

PCT was more commonly obtained in hospitals located in the Northeast and among hospitals with higher admission rates, although rates of testing were not associated with ED volume (Supplemental Table 3).

We observed an increase in the rate of PCT testing from 0.2% in 2016 to 1.6% in 2020 (test for linear trend; P <.001). In 2016, PCT testing was used in only 17 of 33 hospitals (4 of which had PCT testing rates of 0.01%) and in 30 of 33 hospitals in 2020, with marked variability in overall use across hospitals (Fig 1).

The 10 most common conditions for which PCT testing was obtained were identified (Table 2). The most common discharge diagnosis for children who had a PCT test obtained was fever (10.7% of all PCT tests), followed by upper respiratory tract infection (9.3%) and bronchiolitis and pneumonia (5.9%). Among encounters with CBC testing performed, the rate of PCT testing was highest among children with sepsis (28.7%), fever (21.4%), pulmonary edema and respiratory failure (17.3%), and bronchiolitis and respiratory syncytial virus (RSV) pneumonia (15.6%). Trends in PCT use for fever, pneumonia and sepsis are shown in Supplemental Figure 4. Among 61 096 infants <90 days old with fever, PCT testing was performed in 6.8%, overall, and in 21.4% of infants this young with CBC testing performed.

Across all hospitals, the median rate of PCT testing was 0.6% (interquartile range 0.1%–1.1%). Twenty-three hospitals had PCT testing rates of <1% of all ED encounters; of these, 8 were also <1% for encounters with a CBC obtained. Among children with CBC testing, the range of hospital-level PCT testing ranged from 0% to 34.6% (Fig 2).

Although hospitals varied in utilization across APR DRGs, hospitals often remained consistent in their level of utilization, and hospitals that use PCT testing use it broadly across conditions (Fig 3). Across hospitals, the proportion of ED encounters with PCT obtained ranged from 0.0005% to 4.3% with marked variability in overall use.

In this large, cross-sectional study of PCT use across the EDs of 33 US children’s hospitals, we identified widespread variability in PCT utilization, both across conditions and across hospitals. High-utilizing hospitals tend to use PCT broadly. We identified patient and hospital characteristics associated with PCT utilization and the most common conditions for which PCT testing occurred, including its use along with CBC testing. To our knowledge, this present study is the first to examine trends and variability in the use of PCT testing across US children’s hospital EDs.

Overall, we found that PCT utilization occurred at a relatively low rate among these children’s hospitals, and rates began increasing in 2016. We found that PCT was more likely to be obtained for children with complex conditions, children hospitalized from the ED or admitted to an ICU, and younger children. For patients with concurrent CBCs sent, PCT testing was most common in patients with sepsis, fever, and respiratory diagnoses (pulmonary edema/respiratory failure and RSV pneumonia/bronchiolitis) compared with other conditions. Treating providers likely order PCT for conditions with available literature to guide clinical decisions (ie, febrile infants, suspected septic shock, and lower respiratory infection vs pneumonia).^18,24–26  Interestingly, some of these conditions with higher utilization have established guidelines supporting use (febrile neonates, for example),^19,20,27  whereas others have weak evidence only (suspected septic shock or helping to identify bacterial vs viral respiratory tract infection). Practice patterns in adults, in which there are mounting data to suggest the potential utility of PCT in discriminating bacterial from viral lower respiratory tract infections, the need for antibiotic therapy in community-acquired pneumonia and for prognostication in sepsis may also influence the disease states for which PCT was obtained for children.^2–9,28,29 

Although PCT utilization increased for all conditions, we found variation in the use of PCT among conditions as well as across institutions, with some institutions being high utilizers of PCT testing across conditions and others using it much less commonly, regardless of the specific condition. CBC testing was used as the reference standard because it is the most commonly performed test when infection is suspected. A recently published study examining rates of PCT use in febrile neonates also similarly revealed a trend toward increased use of PCT; however, the authors were unable to demonstrate changes in the rates of lumbar puncture, antibiotic administration, or hospitalization.³⁰  These findings are supported in a recent publication on trends in PCT use in admitted children with pneumonia and sepsis and febrile infants in which there was a trend of increased PCT use over time in the cohort and variability noted across institutions.³¹ 

We hypothesize that mounting evidence revealing that PCT is a useful biomarker in febrile infants, in the discrimination of viral from bacterial lower respiratory illnesses in children and adults, and, potentially, in sepsis may be influencing these practice patterns because these are the most common conditions for which it is used in the children’s hospitals included in this study. We suspect that, as hospitals began using PCT in these conditions, it became more widely used in decision-making in a variety of other conditions, as can be seen by its use in a wide array of conditions described above. Additional studies are required to determine if these are, in fact, the driving force behind the uptick in PCT utilization, as well as to determine if the use of PCT in these conditions is associated with changes in outcome.

This study has several limitations. Given the nature of administrative databases, the study is at risk for the misclassification of admission or discharge diagnoses, and it is possible that some PCT testing was included erroneously because it occurred on the floor and not in the ED. Although we included >9 million patient encounters, this study is limited to the 33 hospitals that consistently provided data to PHIS between July 2016 and June 2020 and may not be representative of national trends. We chose the study period between July 2016 and June 2020 to focus our data analysis because that is when PCT was approved by the Federal Drug Administration. Although patients with PCT testing were more frequently hospitalized, prescribed antibiotics, and admitted to an ICU, this may be due to the fact that PCT is more frequently obtained in those patients in whom an invasive bacterial infection is clinically suspected. These findings may also be due to the fact that these patients were hospitalized and treated with antibiotics because of the elevated PCT value; we are unable to differentiate them in this administrative dataset. We were unable to assess whether individual institutions used guidelines for the use of PCT testing (overall or condition-specific) given the nature of the database. Future studies should be conducted to identify PCT prescribing habits and the impact of guidelines on test utilization.

Subsequent prospective studies to identify PCT prescribing habits in the variety of conditions noted above, as well as, potentially, others, and the impact of guidelines on this utilization will be important in informing best practices moving forward.

PCT utilization in the ED has increased over the past 5 years with variation between hospitals. Differential rates of increasing use of PCT testing emphasize the need for standards for the use of this testing modality for specific pediatric conditions encountered in the ED setting. Additional investigation is required to determine how PCT will be used in the future as it becomes more widely available.