1St: C-Reactive Protein As a Prognosis Factor for Septic Patients - a Systematic Review

1st: C-Reactive Protein as a Prognosis Factor for Septic Patients - A Systematic Review and Meta-analysis

2nd: [Authors] Almeida, Marta; Araújo, Alexandra; Branco, J. Pedro; Couto, M. Eduarda; Martins, Camile; Meira, Jorge; Moura, Inês; Norinho, David; Pereira, Pedro; Reis, João; Rodrigues, Sara; Silva, Marília; Torre, A. Paula; Yakubovich, Margarita.

; ; ; ; ; ; ; ; ; ; ; ; ; (from the same order above)

3rd: Prof. Doutor AltamiroCosta-Pereiraand Prof. Doutor Armando Teixeira-Pinto Class 4

4th: ABSTRACT

5th: KEY-WORDS

6th: INTRODUCTION

Background

Justification

7th: RESEARCH QUESTION AND AIMS

8th: REFERENCES

ABSTRACT

Aim: To assess the value of CRP as a marker of prognosis outcome in septic patients admitted to an Intensive Care Unit (ICU).

Methods: A meta-analysis was performed. English-language articles published before April 2009 were used as the data source. Studies were identified by a computerized literature search through PubMed using several keywords (C - Reactive Protein (CRP), Sepsis, Prognosis and Intensive Care). From 221 articles a total of 15 studies met prestated inclusion criteria (N =15). Inclusion criteria included measurement of CRP among different samples of hospitalized patients in Internsive Care Units, evaluation of CRP as a possible prognostic factor and references to outcomes in terms of morbidity and mortality. Studies’ information was abstracted independently by 14 investigators using a standardized protocol. Data was collected and analyzed using SPSS software.

Results: Nine studies showed that there is no association between CRP and sepsis’ prognosis and four concluded the existence of such association. In almost all studies, there are records about inflammatory markers’ measurements and the association between CRP and sepsis’ prognosis, if existent, was independent of other baseline variables known to influence morbidity and mortality. Very different cutoffs have been proposed in each context across studies.

Conclusions: The prognosis power of CRP among septic patients couldn’t be clarified although relevant aspects of this issue were found. Further investigation concerning CRP and sepsis is suggested.

KEY-WORDS

[MeSH terms]: C-reactive protein, Biological Markers, Sepsis, Prognosis, Critical Care, Intensive Care, Meta-analysis, Systematic review.

INTRODUCTION

Sepsis is the body’s exacerbated response to infectionthat threatens patients of all ages and is also the leading cause of death in noncoronary intensive care units (ICU).1,2, 3, 4 It is defined as a complex syndrome characterized by simultaneous activation of inflammation and coagulation in response to infectious stimuli.1, 2When inflammation and the body’s normal anti-inflammatory response get out of balance, the result is systemic inflammatory response syndrome (SIRS).4 SIRS manifests itself through the release of, among other molecular agents, proinflammatory cytokines.1 It can progress to sepsis, severe sepsis, and septic shock. 4 According to the 2001 International Sepsis Definitions Conference, the septic spectrum consists of differentiated degrees of severity of the syndrome.4 A patient with (SIRS) has two or more of the following symptoms: core temperature below 96.8° F (36° C) or above 100.4° F (38° C), elevated heart rate (greater than 90 beats/minute), respiratory rate greater than 20 breaths/minute or Paco2 less than 32 mm Hg, white blood cell count less than 4,000 cells/mm or greater than 12,000 cells/mm3.4 A septic patient has a known or suspected infection plus two or more SIRS criteria.4 Severe sepsisis defined as sepsis plus organ dysfunction.4Finally, septic shock is an acute circulatory failure and persistent hypotension regardless of fluid resuscitation, not explained by other causes.4

Biomarkers are useful adjuncts to clinicians in monitoring septic patients.6 C-Reactive Protein, a plasma protein released by the liver in response to proinflammatory cytokines, is an indicator of inflammation.4 CRP is normally present in trace levels in serum but increases rapidly and dramatically during inflammation and after tissue damage.2, 4 CRP is thus a long-established marker of sepsis.6 However, hypothetically, measurements of circulating concentrations of CRP could also be proved to be useful in pinpointing the prognosis of sepsis. To the best of our knowledge there are currently no meta-analyses that address this problem. We aimed to perform one to investigate whether CRP could be used as a prognostic marker in septic patients admitted to an Intensive Care Unit (ICU).

PARTICIPANTS AND METHODS

The study design used was a systematic review, therefore we used other articles as unit of analysis in its development.

We used a computerized literature search in Pubmed database until April 2009 using the following query: "C-reactive protein” AND (“prognosis” OR “prognostic factor” OR “prognostic studies”) AND ("sepsis” OR “infection” OR “cross infection” OR “hospital acquired infection” OR “multiple organ dysfunction syndrome” OR “MODS” OR “septicemia” OR “septic shock” OR “systematic inflammatory response syndrome” OR “SIRS”) AND (“hospital” OR “critical care” OR “intensive care”). In this search we found 221 articles.

Articles were included if they fulfilled the following criteria: written in Portuguese, Spanish or English; studies realized among hospitalized patients in ICU (Intensive Care Unit) with a septic diagnosis; studies that mention CRP as a prognosis factor of sepsis (alone or conjugated with other biomarkers); those thatincluded consistent and organized data about CRP’s values in different samples of patients and references to outcomes in terms of morbidity and mortality. As exclusion criteria we defined: articles without free full-text or those with insufficient or inconclusive data.

In the first selective phase, titles and abstracts of the retrieved studies were screened for relevance each by two reviewers. If they were concordant in excluding the article it would be removed, if they were concordant in including it, it would pass to next phase. If they had not the same opinion it would be necessary a third reviewer, who decides if the article would have the right conditions to the next phase. In second phase, the full-texts of the articles were analyzed and identified as potentially relevant through the same review method as the first phase.

In order to develop the quantitative analysis we used different software such as: SPSS® 16.0 to edit the data, create common variables and work with statistical measures; Microsoft Visio to build the fluxogram and other diagrams; Microsoft Project to organize the group work.

Through the SPSS® 16.0 software, we created common variables based in the articles selected in the second phase to organize them and find similarities and differences between them. The variables were: name of the article, author, country, year of publication, number and age of patients, percentage of male patients, follow-up, percentage of non-survivors, serum levels of CRP in survivors and non-survivorson different times, summary measures, p value of statistical significance and other relevant data.

The articles were classified using a qualitative test chose due to their heterogeneity. The assessment method selected was the Newcastle-Ottawa Scale (NOS), which was developed to assess the quality of nonrandomized studies including case-control and cohort studies. NOS has a “star system” in which a study is judged on three areas: the selection of the study groups, the comparability of the groups and the outcome of interest, in our case, these are the perspectives for cohort studies.A high quality item is identified with a star, besides a maximum of one star for each numbered item within the Selection and Outcome categories, and a maximum of two stars can be given for comparability. Thus, the parameters: Selection, Comparability and Outcome with four, one and three items respectively make the scale from zero to nine stars.

RESULTS, TABLES AND GRAPHICS

A total of 221 articles were identified using the search strategy planned(Figure 1). After screening titles and abstracts, 149 articles were excluded for the following reasons: no available full text (60), no reference to sepsis (63) or serum levels of CRP (22) and other languages besides those referred in criteria (4). The remaining 72 articles were retrieved for the full paper evaluation; 57 were further excluded due to the lack of conclusions about the association between CRP levels and prognosis of sepsis (8), references to CRP only as a diagnosis factor (16) and insufficient data about CRP values (33). The reports included in the final selection, 14 articles, were divided into two groups according to the association between prognostic value of CRP and sepsis: 4 of them had a positive and 10 had a negative association.

Figure 1_ Flowchart with results of the revision phase.

The two groups selected were organized in detail in Table 2 according to the various variables previous described in the participants and methods section. The articles with references 8 – 16 have a negative association between prognosis value of CRP and sepsis, whereas the articles with references 17 – 20 have a positive association.

Reference / Place / n / Age (years)
Sex (% M) / APACHE/SAPS/Others / Death
Rate / Follow-up
Herrmann W et al.8, 2000 / Department of Trauma Surgery, University Hospital of the Saarland, Homburg/Saar, Germany / 35 / 37,1% / 28 days
Muller B et al. 9, 2001 / Farmacologic Research Institute “Mario Negri," Milan, Italy / 101 / SAPS II
APACHE II
Claeys R et al. 10, 2002 / Department of Clinical Chemistry, Vrije University of Brussels, Brussels, Belgium / 53 / 66 (18-93) years 2
66% / APACHE II
S: 22 (18-26)2
NS: 27(21-32)2 / 47,2% / 6 days
Neslihan C et al. 11, 2002 / Departments
of Surgery, Medical Biology and
Anesthesiology, Faculty of Medicine, Istanbul and the Division of
Immunology, Institute of
Experimental and Medical, University of Istanbul,
Istanbul, Turkey. / 16 / 58,5 (33-70) years2
62% / APACHE II
21 (11-31)2 / 56,3% / 28 days
Marti L et al. 12, 2003 / Infections, Biochemistry, Microbiology and Emergency Service of Hospital Clinic Barcelona, Spain. / 50 / >= 60 years
75,6± 8,98 years1
58% / APACHE II
17,3±4,541 / 20%
Moller HJ et al. 13, 2006 / Five University Hospitals, Denmark / 133 / 66 (23-99) years2
44,36% / 24 hours
Heper Y et al. 14, 2007 / Department of Microbiology and Infectious Diseases, Medicine School, Turkey / 39 / 54,85 ±15,75 years1
61,54% / ACCP / SCCM
sepsis: 53,8%;
severe sepsis: 33,3%;
septicchock: 12,8%. / 12,8% / 72 hours
Marti L et al. 15, 2007 / Hospital Clinic, Barcelona, Spain / 100 / 75 ± 8,4 years 1
60–93 years (IR)
59% / APACHE II
16,60 ± 4,68 1
IR:4–32 / 14% / 9,7± 9,6 days
Silvestre J, et al. 16, 2008 / IntensiveCareUnit (ICU) of
Garcia de Orta Hospital, Lisbon, Portugal / 158 / 59 ± 17 years 1
62,02% / APACHE II
S: 18,8 ± 6,31
NS: 26,0 ± 7,31
SAPS II
S: 41,5 ± 11,71
NS: 58,2 ± 14,71
(p<0,001) / 50% / 1 day

Table 1_ Table containing data about the general description of included articles that stat a negative association between CRP levels and prognosis of sepsis. Abbreviations: M, male; S, survivors; NS, non survivors.

1mean±SD (standard deviation); 2 median (percentiles 25 – 75); 3 median (minimum – maximum); 4 median (IR – interquartiles range).

Reference / CRP initial Values / Significant Differences / CRP Last Values / Significant Differences / Other Data
CRP values
(S) / CRP values
(NS) / CRP values
(S) / CRP values
(NS)
Herrmann W et al.8, 2000 / 22,3(12,6-38,7)mg/dL4 / 17,6(15,1-31,6) mg/dL4 / p>0,05 / 20,9(11,2-35,7)mg/dL 4 / 22,0(9,7-30,9)mg/dL 4 / p>0,05
Muller B et al. 9, 2001 / Odds Ratio of 1,13 in association with mortality, p=0,466 (no statistic significance).
Area under ROC curve equals to 0,55, regarding the same parameter.
Claeys R et al. 10, 2002 / 19,7(13,6-29,0) mg/dL2 / 21,2(11,0-28,2) mg/dL2 / p=0,777 / 8,9(3,5-14,0)mg/dL2 / 14,2(5,5-18,2) mg/dL2 / p=0,260
Neslihan C et al. 11, 2002 / Non survivors were found to have higher CRP levels than survivors / p=0,048
Marti L et al. 12, 2003 / 1,99 mg/dL1 / 1,61 mg/dL1 / p<= 0,05 / 1,53 mg/dL1 / 1,32mg/dL1 / p<= 0,05
Moller HJ et al. 13, 2006 / 1st group: 1090 nmol/L 2
2nd group: 681 nmol/L 2 / Relative Risck
RR = 7,0
(2,4–21,6 CI 95%)
1st group: <75 years
2nd group: >=75 years
Heper Y et al. 14, 2007 / 1st group: 18,02±12,15mg/dL1
2nd group: 16,83±10,1 mg/dL1 / 1st group: 11,21±7,95 mg/dL1
2nd group: 12,71±9,2 mg/dL1 / 1st group: sepsis
2nd group: severe sepsis
Marti L et al. 15, 2007 / 7,61 mg/dL2 / 6,55 mg/dL2 / p>0,05 / p>0,05
Silvestre J, et al. 16, 2008 / 25,3±13,7 mg/dL1 / 28,2±13,1 mg/dL1 / p=0,15 / Area under the ROC curve:
sepsis: 0,55 (0,45–0,65)
documented sepsis: 0,66 (0,53–0,79)

Table 2_ Table containing results presented in articles that stated a positive association between CRP levels and prognosis of sepsis. Abbreviations: S, survivors; NS, non survivors.

1mean±SD (standard deviation); 2 median (percentiles 25 – 75); 3 median (minimum – maximum); 4 median (IR – interquartiles range).

Articles with a negative association between CRPlevels and prognosis of sepsis

In 9 studies which concluded none association between the levels of C-reactive protein and prognosis of sepsis, described in detail in Tables 1 and 2, different values of CRP, measured in different conditions and according to various procedures among distinct patients, were observed and analyzed as a potential factor of sepsis’ prognosis. All the studies stand that no association was found.

In Herrmann W et al.8 published in 2000, CRP values were measured at admission, 24th, 48th and 72nd hours, showing that CRP values decreased during the follow-up period among patients in the survivors group and, on the contrary, increased in the non survivors group. Values presented (median [minimum – maximum]) were higher in survivors compared to non survivors (22,3 [12,6-38,7] mg/dL vs. 17,6 [15,1-31,6] mg/dL) at admission and after 24 hours (21,6 [11,6-43,7] mg/dL vs. 19,9 [11,3-28,2] mg/dL). Non survivors registered a lower value in the 48th and 72nd hours in relation to survivors (21,3 [3,1-39,4] mg/dL vs. 22,1 [9,8-35,2] mg/dL and 20,9 [11,2-35,7] mg/dL vs. 22,0 [9,7-30,9] mg/dL). All values were not significant, p>0,05. These results show that CRP isn’t associated with prognosis of sepsis regarding the perspective of mortality.

Muller B et al.9, 2001, presented an odds ratio of 1,13 in association with mortality, which does not have statistic significance, and also an area under ROC curve equals to 0,55 regarding the same parameter related to mortality.

Only two of the studies analyzed presented CRP values concerning results obtain in more than two measurement times. Claeys R et al.10, 2003, showed that, at 24th and 48th hour, CRP levels in non survivors were lower than in survivors (26,9 [19,7-32,8] mg/dL vs. 25,4 [15,3-34,4] mg/dL , p=0,806, and (25,2 [19,2-32,3] mg/dL vs. 22,8 [16,4-25,0] mg/dL, p=0,215) corresponding to statistically significant values. However, levels of CRP were higher in non survivors comparing those with the survivors group both at admission (19,7 [13,6-29,0] mg/dL vs. 21,2 [11,0-28,2] mg/dL, p=0,777) and at the 120th hour (8,9 [3,5-14,0] mg/dL vs. 14,2 [5,5-18,2] mg/dL, p=0,260) representing values with no statistic significance. Study also reported a general decreased at 48th hour in both groups of patients, in 11 of 24 survivor patients and 9 of 19 non survivor patients. Another decreased was registered in all patients of the survivors group and in 7 of 11 non survivors at the end of 120 hours; but in opposition to the first lowering, this second one was statistically significant, p=0, 037.

Other study that followed the same structure is Heper Y et al.14, 2006. This study’s results showed that larger levels of CRP were found in sepsis group compared to severe sepsis group at admission (18,02±12,15mg/dL vs. 16,83±10,1 mg/dL) and at the end of the 24th hour (18,06±10,45mg/dL vs. 16,35±10,41 mg/dL). Although the initial stage-setting, CRP levels in the sepsis group became lower than in severe sepsis group at the 48th hour (13,15±8,01 mg/dL vs. 15,71±10,27 mg/dL) and at the end of the 72nd hour (11,21±7,95 mg/dL vs. 12,71±9,2 mg/dL). These results show that CRP values tended to decrease after the 24th hour among patients diagnosed with sepsis, having a significant decrease from the 24th to the 48th hour. Patients with a diagnostic of severe sepsis registered a great increase in levels of CRP during in the last two measurements compared to the initials. Despite the differences registered none of the results was statistically significant, p>0,05.

Another study, published in 2002, Neslihan C et al.11, only reported that non survivors were found to have higher CRP levels than survivors, indicating a p value equals to 0,048. Article also contained other data included in graphics whose contours did not allow the collection of exact and precise data.

Marti L et al.12, 2003, reported higher values of CRP in survivors group compared to the non survivors (1,99 mg/dL vs. 1,61 mg/dL) at admission and also at the end of the 4th day (1,53 mg/dL vs. 1,32mg/dL) (mean values). This values were statistically significant, p <= 0,05. This study seems to corroborate the conclusions presented by other study form the same authors published in 200715. This second study also shows that survivor patients had higher levels of CRP (7,61 mg/dL) in comparison to non survivors (6,55 mg/dL), p>0,05, however this results weren’t significant in terms of statistics when they were measured at admission to ICU.

A study performed in 2006 by Moller HJ et al.13 had a specific cutoff dividing patients through their age into two groups. CRP values were higher in the group of patients with <75 years compared to patients included in the >=75 years group (1090 nmol/L vs. 681 nmol/L) (mean). Article also presented a relative risk value of 7,0 (RR = 7,0 [2,4–21,6 CI 95%]) of mortality among patients considered in the study. The article presented CRP levels considering nmol/L as a measurement unit and because no other values of CRP were presented, namely a CRP standard value, no adjustments of units were done.

The most recent study found about the subject of the present systematic review was Silvestre J, et al.16, 2008. The results of this study showed that levels of CRP varied from 5,5 mg/dL – 43,9 mg/dL (median 19,0 mg/dL). They were higher in non survivors group (28,2±13,1 mg/dL) compared to the survivors group (25,3±13,7 mg/dL) at admission time. Although that record (mean±SD), differences were not significant, p=0,15. It was also reported that the area under ROC curve for patients diagnosed with sepsis was 0,55 (0,45–0,65) and 0,66 (0,53–0,79) for patients with a positive diagnosis for severe sepsis.

All studies included in this group also measured and presented results concerning values of an inflammatory marker, APACHE/SAPS, expressed in Table 2, except Heper Y et al.14 that presented values of ACCP/SCCM and Herrmann W et al.8 and Moller HJ13 et al. that did not present values of any marker of inflammation.

Reference / Place / n / Age (years)
Sex (% M) / APACHE/SAPS/Others / Death
Rate / Follow-up
Aoki Y et al. 17, 2000 / Saga Medical School Hospital, Saga, Japan / 73 / Men:66 (33-90) 3
Women:65 (20-82)3
60,3% / APACHE III
S: 2,91
NS: 76,51
(p<0,001) / 34,2% / 28 days
Memis D et al. 18, 2007 / Trakya University
Hospital, Edirne, Turkey / 96 / 56 (18-75)2
42% / APACHE II
S: 14,8 (3,8)4
NS: 20,1 (4,3)4 / 57% / S: 11,1±7,0 days1
NS: 9,9± 4,9 days1
Schmit X et al. 19, 2008 / Erasme
University Hospital and CHIREC Hospital, Brussels, Belgium / 50 / 63 (51–73)2
76% / 27% - >50%
(depending on severity) / 4 days
Lee CC et al. 20, 2008 / Department of Emergency Medicine, National Taiwan University Hospital Yun-Lin Branch, Douliou, Taiwan / 525 / 64 (47-76)2
57,5% / MEDS score
Total: 3 (2-6)2 / 90% / 30 days

Table 3_ Table containing data about the general description of included articles that stat a positive association between CRP levels and prognosis of sepsis. Abbreviations: M, male; S, survivors; NS, non survivors.

1mean±SD (standard deviation); 2 median (percentiles 25 – 75); 3 median (minimum – maximum); 4 median (IR – interquartiles range).

Reference / CRP initial Values / Significant Differences / CRP Last Values / Significant Differences / Other Data
CRP values
(S) / CRP values
(NS) / CRP values
(S) / CRP values
(NS)
Aoki Y et al. 17, 2000 / 11,78,0 mg/dL1
(measured within 24 h) / 16,510,1 mg/dL1
(measured within 24 h) / p=0,031
Memis D et al. 18, 2007 / 10 (6-14) mg/dL2 / 32 (20,5-64,5) mg/dL2 / p<0,001 / 6 (3-9) mg/dL2 / 30 (22-5) mg/dL2 / p<0,001 / Area under ROC curve (95%CI):
Initial: 0,947 (0,904-0,991)2
Last: 0,997 (0,992-1,002) 2
p<0,001
Schmit X et al. 19, 2008 / Total: 16,7 ± 10,6 mg/dL1
Group 1 vs. Group 2:
Δ CRP day 0-day 1: –2,4 ± 6,8 mg/dL vs. -5,6 ± 5,0 mg/dL1 / p>0,05 / Group 1 vs. Group 2:
Δ CRP day 0-day4: 6,3 ± 11,1 mg/dL vs. -1,0 ± 8,7 mg/dL1 / p>0,05 / Area under ROC curve:
Major difference between days D0 and D2 (0,71 ± 0,08)1
Group 1 - patients with a favorable response to the antibiotic therapy; group 2 - patients who required a change in antibiotic therapy; group 3 - patients who needed a procedure to control infection.
Lee CC et al. 20, 2008 / CRP >6 mg/dL
in 41,9% / CRP >6 mg/dL
in 75,0%*in “early mortality” group (5 days); in 61,5%* in “late mortality” group (5-30 days) / Area under ROC curves (95% CI):
NS: early mortality - 0,68** and late mortality” – 0,76***;
Best cutoff value (sensitivity, specificity):
NS: early mortality – 70 (73%, 63%)** and late mortality - 60 (64%, 58%)***

Table 4_ Table containing results presented in articles that stated a positive association between CRP levels and prognosis of sepsis. Abbreviations: S, survivors; NS, non survivors.

1mean±SD (standard deviation); 2 median (percentiles 25 – 75); 3 median (minimum – maximum); 4 median (IR – interquartiles range); *significant higher frequency compared to survivors (p<0,05); **p=0,002; ***p=0,011.

Articles with a positive association between CRPlevels and prognosis of sepsis

Only 4 of the 13 articles included in this systematic review concluded a positive association between values of CRP and a prognosis produced based on those values regarding patients with a positive diagnostic for sepsis. These 4 articles are described detailed in Tables 3 and 4.