Procalcitonin for predicting catheter-associated bloodstream infection

Abstract

Objective:

The predictive accuracies of procalcitonin (PCT) in the diagnosis of catheter-associated bloodstream infection (CABSI) vary widely. This meta-analysis aimed to explore the predictive value of PCT for CABSI.

Methods:

We searched PubMed, EMBase, Web of Science, ScienceDirect, Cochrane Library, and studies published up to 10 March 2019. Odds ratios (ORs) with 95% confidence intervals (95%CIs) were calculated to evaluate PCT predictive value using Stata 14.0 software.

Results:

The meta-analysis was composed of 7 studies, consisting of 347 subjects. Pooled analysis demonstrated that a high PCT was significantly correlated with CABSI (pooled OR = 23.36, 95%CI 12.43–43.91, P < .001) and medium heterogenicity (I ² = 36.9%, P = .147). The pooled sensitivity and specificity were 85% (95%CI 0.76–0.91) and 89% (95%CI 0.68–0.97), respectively. Although Begg funnel plot (P = .007) indicated the presence of publication bias among the included studies, the stability of the pooled outcomes was verified by the trim-and-fill method. Furthermore, sensitivity analyses did not show important differences in effect estimation.

Conclusion:

PCT is an effective predictor of CABSI. However, high-quality randomized controlled trials are needed to determine whether PCT could predict CABSI.

1. Introduction

Central venous Catheters (CVCs) play an important role in facilitating infusion, drug administration, blood sampling, and close monitoring in seriously ill patients. The reported prevalence of CVC utilization in intensive care patients is 48%.^[1] CVC use has been associated with many complications.[²,³,⁴] Catheter-associated bloodstream infection (CABSI) is one of the most common intractable healthcare-associated infections because catheters can be easily contaminated by bacteria and are associated with high mortality.[³,⁵,⁶,⁷] An estimated 250,000 cases of CABSI occur in the United States annually, with 10% mortality rate.[⁸,⁹] The alarmingly high rates of CABSI have forced changes in clinical practices, including increased adoption of catheter lock solutions and improved anti-infective and anti-microbial surveillance protocols to reduce catheter-related infection rates.[¹⁰,¹¹] Controlling and managing these infections have imposed a heavy financial burden on the government.

Procalcitonin, a 116-amino-acid peptide, which undergoes posttranslational proteolysis into calcitonin, is synthesized in C cells of the thyroid gland and secreted from leukocytes of the peripheral blood. The rise of PCT in bacterial infection is remarkable, but its incidence in viral infections only showed a negligible increment.[¹²,¹³] Increased PCT level is considered a key laboratory indication of acute infection, and PCT is confirmed as a marker of CABSI among patients with CVCs.[¹⁴,¹⁵,¹⁶,¹⁷,¹⁸,¹⁹,²⁰] Although PCT is implied to have superb specificity and sensitivity for CABSI, controversies remain due to inconsistent opinions. Therefore, we performed a meta-analysis from clinical trials to investigate the predictive value of PCT for CABSI.

2. Materials and methods

The manuscript has been prepared according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines for protocols.^[21] Ethical approval was unnecessary in this study, because it was a meta-analysis of existing articles, and no individual patient data were handled. This protocol has been registered in the PROSPERO network (registration number: CRD42019129219).

2.1. Search strategy

We searched PubMed, EMBase, Web of Science, ScienceDirect, Cochrane Library, and studies published up to March 20, 2019. The search terms were as follows: “catheter associated bloodstream infection” AND “procalcitonin.” The reference lists of relevant literature were manually searched for additional eligible articles. Searching language was restricted to English and Chinese.

2.2. Inclusion and exclusion criteria

The inclusion criteria were as follows: all the subjects included in the study were given definitive diagnoses of CABSI; all prospective and retrospective full-text studies detecting the predictive value of PCT for CABSI; sufficient survival data were provided for the odds ratios (ORs) with 95% confidence intervals (CIs); The exclusion criteria were as follows: duplicated studies, letters, case report abstracts, comments, reviews, and conference papers. Only the study with the largest number of subjects was included, when multiple studies were based on the same case series.

2.3. Data extraction

Data were extracted independently by two reviewers from the included studies. When conflicts or different opinions arise during data extraction, a resolution was achieved by consensus with a third reviewer. Data were abstracted from included studies, including first author, year of publication, study design, country of origin, gender, sample size, cutoff values, PCT level, and study period.

2.4. Risk of bias

Risk of bias assessment of the included studies was independently fulfilled by 2 reviewers using the Quality Assessment of Diagnostic Accuracy Studies 2 tool (QUADAS-2),[²²,²³,²⁴] which consisted of four domains, as follows: patient selection, index test, reference standard, and flow and timing. Discrepancies were resolved by discussion.

2.5. Statistical analysis

Data analysis for this meta-analysis was performed using STATA version 14. Statistical heterogeneity was assessed through the Chi-square test and I-square test, which were checked through the Q test, and a P > 0.10 indicated a lack of heterogeneity. The percentages of I ² of approximately 25%, 50%, and 75% indicate low, medium, and high heterogeneity, respectively.^[25] Without statistical heterogeneity (I ² < 50%, P > .1), a fixed-effects model was employed to calculate the pooled HRs with 95% CIs. Otherwise, a random-effects model was used. Publication bias was estimated using Begg test and a funnel plot with the trim-and-fill method to adjust for publication bias from potential unpublished studies. Sensitivity analysis was used to evaluate the robustness of the pooled results. Sensitivity analysis was performed by removing individual study at each turn to investigate the influence of any single study on the pooling summary. Fagan plot analysis was also performed to assess the relationship among an estimated pretest probability of the disease, the likelihood ratio of the diagnostic test, and the post-test probability of the disease. A pre-test probability was calculated depending on the included studies, and the corresponding positive and negative post-test probabilities were calculated. P < 0.05 was considered statistically significant.

3. Result

3.1. Search strategy

The detailed selection procedure is listed in Figure 1 According to the search strategy, a total of 133 potentially eligible studies were identified. After excluding duplication and reading the texts for further examination, a total of seven studies[¹⁴,¹⁵,¹⁶,¹⁷,¹⁸,¹⁹,²⁰] were included for analysis.

Figure 1.

Flowchart of study selection.

3.2. Study characteristics

Table 1 revealed the main characteristics of the included studies. All 7 prospective studies consisting of 347 subjects were published between 2011 and 2018. Among the included studies, three studies[¹⁴,¹⁸,²⁰] were from China. Turkey,^[17] Egypt,^[15] USA,^[16] and Greece^[19] had 1 study each. Six studies[¹⁴,¹⁵,¹⁶,¹⁷,¹⁹,²⁰] were published in English, and 1 study^[18] was published in Chinese.

Table 1.

Characteristics of included studies.

3.3. Assessment of risk of bias for included studies

QUADAS-2 assessment is summarized in Figure 2. The patient selection risk of bias domain in 6 studies[¹⁴,¹⁶,¹⁷,¹⁸,¹⁹,²⁰] was labeled as unknown, because the authors did not report whether the subjects were consecutively enrolled. One study^[15] was labeled as high risk, because the authors did not report whether the subjects were consecutively enrolled in the study period. The index test domain of all studies[¹⁴,¹⁵,¹⁶,¹⁷,¹⁸,¹⁹,²⁰] was labeled as high risk, because a pre-specified threshold was not used.

Figure 2.

Quality assessment of the included studies by using QUADAS-2 graph.

3.4. Association of PCT and CABSI

Pooled analysis showed that a high PCT was significantly correlated with CABSI (pooled OR = 23.36, 95%CI 12.43–43.91, P < .001; Fig. 3) and with medium heterogenicity (I ² = 36.9%, P = .147). The pooled sensitivity and specificity were 85% (95%CI 0.76–0.91) and 89% (95%CI 0.68–0.97), respectively. The positive likelihood ratio was 7.4 (95%CI 2.3–23.6), the negative likelihood ratio was 0.17 (95%CI 0.10–0.29), and diagnostic OR was 43 (95%CI 9–189). AUC of PCT for CABSI was 0.90 (95%CI 0.87–0.93; Fig. 4)

Figure 3.

Forest plot of the association between procalcitonin and catheter-associated bloodstream infection.

Figure 4.

SROC of procalcitonin for the prediction of catheter-associated bloodstream infection. The area under the SROC curve (dashed line) is 0.90, the pooled sensitivity is 85%, and the specificity is 89%. The 95% confidence region displays the 95% confidence interval of the pooled sensitivity and specificity. SROC = summary receiver operating characteristic curve.

3.5. Meta-regression analysis for heterogeneity

The following specific variables were separately evaluated for their effects on heterogeneity: sample size (≥50 or <50), publication year (before 2016 or after 2016), prevalence (43%), and cutoff (≥3.96 or <3.96). Meta-regression showed that these variables did not explain heterogeneity observed for specificity (Table 2).

Table 2.

Meta-regression analysis of potential sources of heterogeneity.

3.6. Sensitivity analysis and publication bias

Sensitivity analysis was performed by removing all included studies sequentially to detect the influence of each study on the pooled OR. The corresponding pooled ORs were not significantly changed after excluding the studies one by one, thereby suggesting the significant stability of our results (Fig. 5). Although Begg funnel plot (P = .007; Fig. 6A) indicated the presence of publication bias among the included studies, the adjusted fixed-effects pooled OR of 20.59 (95%CI 10.58–40.08, P < .001) calculated using trim-and-fill method was consistent with the original analysis (OR = 15.64, 95%CI 8.26–29.61, P < .001) (Fig. 6B).

Figure 5.

Sensitivity analysis of procalcitonin for catheter-associated bloodstream infection prediction. Leave-one-out method was used to confirm the stability of the results. No single study was detected to incur undue weight in the analysis.

Figure 6.

Funnel plot of publication bias. (A) Begg test; (B) Trim-and-fill method.

3.7. Fagan plot analysis

When a pre-test probability of 43% (149/347) was calculated depending on these included studies, the probability of a correct diagnosis following a positive measurement of PCT was 85% for CABSI. The misdiagnosis rate was 11% for patients under a negative measurement (Fig. 7).^[26]

Figure 7.

Fagan plot analysis of procalcitonin for the prediction of catheter-associated bloodstream infection.

4. Discussion

To the best of our knowledge, the present study is the first meta-analysis to investigate the predictive value of PCT for CABSI, which might provide useful information for clinicians. The results from the combined data of 7 studies should that a 22-fold risk increase of CABSI is present among clinically suspected CABSI with high PCT level, with a diagnostic sensitivity of 85%, a specificity of 89%, and 0.90 AUC.

Early diagnosis of CABSI is of primary importance, because when an appropriate antimicrobial regimen is administered after accurate diagnosis in infected patients, a better outcome is expected.^[27] Clinical findings are unreliable for establishing the diagnosis of CABSI because of their poor sensitivity and specificity. Establishing a true “gold standard” for the diagnosis of CABSI is a challenge. The practical use of blood cultures has a significant limitation due to the delay in reporting of culture results, the issues with colonization and contamination, and the inability to grow atypical pathogens in standard cultures.

Colonization of the lumen of an intravascular catheter precedes infection in many cases. Schuetz et al^[28] showed the increased values of PCT 1 day prior to clinical manifestation of BSI and collection of blood cultures. The early increase of PCT might reflect the colonization of the catheter with subclinical infection ultimately leading to bloodstream infection.^[29] Notably, nonspecific increases in the absence of bacterial infection can be observed in stress conditions, such as severe trauma and surgery. Under these conditions, the increases are moderate and a rapid decrease might ensue.^[30] Thus, dynamic observation on change of PCT is objective and stable.

PCT's role as a biomarker for bacterial infection in adults is well documented, but its role in infections affecting neonatal children remains controversial.[³¹,³²] The number of studies interpreting the worth of PCT as a diagnostic and screening test in pediatric infections is insufficient. The pooled OR with 95% CI was OR = 13.17 (95% CI 4.33–40.08, P < .001) on the basis of 2 studies in children[¹⁶,¹⁷] and OR = 30.97 (95% CI 14.19–67.59, P < .001) on the basis of 5 studies in adults. Our results demonstrated that a high PCT is significantly correlated with CABSI in children and adults.[¹⁴,¹⁵,¹⁸,¹⁹,²⁰] However, further research is needed due to the insufficient number of studies, especially in children.

A comprehensive evaluation of every patient with a bloodstream infection includes an attempt to identify the infectious source. Pathogens can originate from various sites, including the lung, biliary tract, urinary tract, surgical site, skin, or catheter. Biomarkers are used for identifying the presence of infection, but not the specific sites.

Missing comorbidity data result in uncertainties; under-/overestimation of PCT level; or failure to address confounding factors such as burn injury, trauma, and chronic heart failure. Notably, all seven studies included in our meta-analysis did not adjust for comorbidity. Thus, to improve research on comorbidity, studies should include information on the variety and severity of comorbidity.

The strength of our study is the use of Fagan plot analysis to explore the clinical utility of PCT. At 43% pre-test probability, the probability of a correct diagnosis following a positive measurement of PCT was 85% for CABSI. The misdiagnosis rate of patients under negative measurement was 11%. Fagan plot analysis further supports the importance of PCT for diagnosing CABSI.

The cutoff value of PCT level varied across different studies, and a consensus value was difficult to reach. The range of cutoff value among the seven included studies was from 0.3 ng/mL to 15.5 ng/mL. The discrepancy could be attributed to differences in the proportions of bacterial species. Infection by different pathogens can induce different levels of PCT production.^[33] GNB can produce endotoxins that can be released upon cell death, resulting in persistently high levels of PCT.[³⁴,³⁵]

Nonetheless, several limitations should be considered when interpreting our results. First, our meta-analysis included a total of published studies with 347 subjects, which might decrease statistical power. Second, some potent reports in other languages might have been missed due to language limitation in search strategy. Thus, Begg funnel plot might have not detected possible publication bias. Third, medium heterogeneity was found in this meta-analysis.

In conclusion, the interesting but preliminary data of this meta-analysis demonstrate that PCT is an effective diagnostic method for detecting CABSI in patients. Our study has several limitations and further investigation is needed; however, these findings might help clinicians identify CABSI early and formulate treatment strategies.

Author contributions

Conceptualization: Chun Mei Jia.

Data curation: Chun Mei Jia, Shun Yi Feng, Zong Xun Cao, Cheng Pu Wu, Yan Zhao Zhai, Meng Zhang.

Methodology: Yong Li.

Project administration: Yong Li.

Software: Shun Yi Feng.

Writing – original draft: Chun Mei Jia, Shun Yi Feng.

Writing – review & editing: Jie Cui, Jie Gao.