Abstract
Background:
The first decision confronting clinicians in the management of patients with community acquired pneumonia (CAP) is whether the patient is to be hospitalized or not. We sought to validate the pneumonia scoring system and assess the power of procalcitonin (PCT) level to predict in-hospital mortality (IHM) and intensive vasopressor and respiratory support (IVRS) requirements in patients with CAP.
Materials and Methods:
A total of 120 patients with CAP were evaluated for severity of illness based on the defined scoring systems including pneumonia severity index (PSI), confusion, urea, respiratory rate, blood pressure, age>65 (CURB-65), confusion, respiratory rate, blood pressure, age>65 (CRB-65), infectious diseases society of America/American thoracic society 2007 criteria (IDSA/ATS 2007) and systolic blood pressure, multilobar infiltrate, albumin, respiratory rate, tachycardia, confusion, low oxygen, low pH (SMART-COP).
Demographic, clinical, laboratory and radiographic data were collected prospectively. The accuracy of each scoring system in predicting IVRS requirement and IHM was assessed from the area under the receiver operating characteristic (ROC) curve (AUC). Level of PCT was determined by semi-quantitative PCT-Q method (BRAHMS). The accuracy of the defined scoring systems, PCT levels and each scoring system plus PCT levels in prediction of IHM and IVRS requirement was analyzed.
Results:
The accuracy of PCT levels in predicting IHM and IVRS requirement based on AUC was 0.542 and 0.658, respectively and the best threshold was ≥ 2ng/mL for both of them. Adding the level of procalcitonin to different scoring systems (based on the defined scoring systems) improved the accuracy of all systems.
Conclusion:
We do not suggest using the PCT level alone as a predictor for mortality and IVRS requirement. Instead, we suggest PSI plus PCT and IDSA/ATS 2007 plus PCT as accurate predictors for IHM and SMART-COP plus PCT for IVRS requirement in patients who presented with CAP.
Keywords: Hospital Mortality, Pneumonia, Procalcitonin, Severity of Illness Index
INTRODUCTION
The initial management decision after diagnosis of CAP is to determine the site of care. There is significant variation in admission rates among hospitals and among individual physicians. Overestimation of the severity of CAP is common among physicians and leads to hospitalization of a significant number of patients at low risk for death. Because of this, different specialty groups have tried to develop objective site-of-care criteria or severity scoring systems (1). Several scoring systems have been proposed. These scoring systems include PSI (20 variables including age, coexisting illness and abnormal physical and laboratory findings) (2), CURB-65 (3), CRB-65 (4), American Thoracic Society 2001 Criteria (ATS) (5), IDSA/ATS (major criteria: septic shock requiring vasopressor support and mechanical ventilation; minor criteria: respiratory rate ≥30 breaths/minute, PaO2/FiO2 ratio ≤250, multilobar infiltrates, confusion, blood urea nitrogen ≥20 mg/dL (blood urea 7 mmol/L), leukopenia, thrombocytopenia, hypothermia or hypotension requiring fluid support) and SMART-COP (6). The predictive power of each score is calculated and validated in different studies. According to the best cut off value, severe illness is defined as PSI class ≥ 4, CURB-65 ≥3, IDSA/ATS criteria (1 or 2 major criteria or ≥ 3 minor criteria), or SMART-COP score ≥ 5.
In addition, several inflammatory markers were identified with predictive capacity of the severity of pneumonia. Among the most widely studied biomarkers are C-reactive protein (CRP) and PCT (7–11).
Herein, we conducted a study to validate each scoring system and assessed the level of serum procalcitonin for prediction of IHM and IVRS requirement alone and in combination with other severity scoring systems in adult patients with CAP.
MATERIALS AND METHODS
1. Selection and Description of Participants
Out of 166 patients with community acquired lower respiratory tract problems with chest infiltrate, 140 cases were eligible for this study. Sixteen patients with non-infectious etiologic diagnosis and four patients with extra-pulmonary infections were excluded from the study. Statistical analyses were only performed on the remaining 120 adult patients with the diagnosis of CAP admitted to Imam Reza Teaching Hospital in Mashhad, Iran. Severity of illness based on the defined scoring systems (PSI, CURB-65, CRB-65, IDSA/ATS 2007 and SMART-COP) was assessed. The diagnostic criteria for CAP included a new infiltrate on chest radiograph in a patient with either fever or clinical signs and symptoms of lower respiratory tract infection (cough, sputum production, dyspnea, pleuritic chest pain, crackles on auscultation), or both.
This study was approved by the vice chancellery, the institutional review board (IRB) and ethics committee of Mashhad University of Medical Sciences (MUMS). Written informed consents were taken from all subjects.
2. Technical Information
All patients underwent a diagnostic evaluation, including chest radiography with/without chest computed tomography (CT) scan, blood chemistry and ABG assessment, sputum/endotracheal aspirate (and possibly pleural fluid) staining and culture, blood for culture in standard aerobic/F BACTEC bottles, Binax NOW Legionella urinary antigen test (Binax, Scarborough, Maine, USA), Binax NOW Streptococcus pneumonia urinary antigen test (Binax, Maine, USA), and RT-PCR of nasopharyngeal swab specimen for influenza virus. The PCT levels were determined by a semi-quantitative solid-phase immunoassay (B.R.A.H.M.S. PCT-Q, B.R.A.H.M.S.-Diagnostica GmbH, Hennigsdorf, Germany) on 200 μl plasma. The PCT levels were categorized into four groups (< 0.5 μg/l; 0.5−< 2 μg/l; 2−< 10 μg/l; ≥ 10 μg/l) according to the provided reference scale. The test was performed within the first 12 hours of patient admission.
Demographic, clinical, laboratory and radiographic data were collected. The accuracy of defined scoring systems and PCT levels in prediction of IVRS requirement and IHM was analyzed (mechanical ventilation and/or vasopressor support during an unsuccessful CPR were not included). The PCT level factor of each scoring system as a risk factor for IHM and IVRS requirement was considered and calculated by the AUC of each new model. Patients included in our study were those who had new infiltrates on chest radiography with fever or lower respiratory signs/symptoms, (or both) and required hospital admission.
Chronic obstructive pulmonary disease (COPD) was defined clinically as the presence of a chronic productive cough for ≥ three months during two consecutive years (other causes of cough being excluded).
Bedridden status was defined as confined to bed by sickness or old age.
Opium addiction was defined as behaviors that include one or more of the following: Impaired control over drug (opioid) use, compulsive drug use or continued use.
3. Statistical Analyses
Statistical analyses were performed using SPSS software and R programming language. Discrete variables were expressed as percentage and continuous variables as mean ± standard deviation (SD) unless stated otherwise. Frequency comparison was done by the chi-square test or Fisher’s exact test for categorical variables. Student’s t-test or the Mann-Whitney U test was used for continuous variables. Standard definitions of sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were used. The accuracy of each scoring system in predicting outcome was evaluated using the ROC analysis. Discrimination was assessed by plotting ROC curve and calculating the AUC; AUC values were ranked as excellent (AUC≥0.90), good (AUC: 0.80–0.90), fair (AUC: 0.70–0.80), poor (0.60–0.70), and failed (0.50–0.60). The sensitivity and specificity of each score at the best threshold and after adding PCT levels to each score were calculated using R programming language.
RESULTS
One hundred twenty patients fulfilled the criteria for inclusion. The mean age was 50.4 ± 22.6 years (range 17–94). The male to female ratio was 1/7. The most common microbial diagnosis was S. pneumonia, followed by M. tuberculosis, S. aureus, and polymicrobial including anaerobes. The frequency of patients who presented with severe illness as judged by the PSI class ≥ 4, CURB-65 ≥3, IDSA/ATS criteria (1 or 2 major criteria or ≥ 3 minor criteria), and SMART-COP score ≥ 5 was 60.3%, 36.6%, 60.4% and 60.6%, respectively.
The overall IHM rate was 23.6%. The third day mortality rate was 3.3%; 19.1% of hospitalized patients admitted to the ICU; however, it was far less than actual number of patients who needed intensive care. If the actual number of patients who need ICU admission was equal to the number of patients who need IVRS, then the former would be 1.6 times more than patients admitted to the ICU (37 vs. 23). The IHM rate was higher among patients who admitted to the ICU compared to those who did not (60% vs. 14.8%; P<0.001). As expected, it was significantly higher among patients who needed IVRS compared to other patients (67.6% vs. 4.1%, P <0.001). Evaluation of different demographic, clinical, laboratory and radiographic characteristics of patients demonstrated several risk factors for IHM and IVRS requirement (Table 1). The mean length of hospital stay (LOS) was 11.43±13.2 days. Factors associated with longer LOS were: leukopenia (P=0.045), leukocytosis (P=0.021), PCT level > 0.5 ng/mL (P=0.002), presence of hypoxia on admission (P=0.001), IVRS requirement (P=0.028), ICU admission (P=0.004) and severe illness as judged by the IDSA/ATS 2007 and SMART-COP criteria (P=0.031 and P=0.009, respectively).
Table 1.
The relation between patient characteristics, IVRS and IHM
| IVRS | IHM | |||
|---|---|---|---|---|
| Risk factor | Odds Ratio | P value | Odds Ratio | P value |
| Demographic characteristics | ||||
| Age>=65 yrs. | 1.723 | 0.211 | 1.964 | 0.063 |
| Female sex | 1.645 | 0.297 | 1.25 | 0.958 |
| Addiction | 2 | 0.101 | 1.6 | 0.556 |
| History of incarceration | 0.968 | 1.000 | 1.75 | 0.387 |
| Comorbidity | 1.639 | 1.62 | ||
| Diabetes | 0.364 | 0.596 | 2.821 | 0.381 |
| IHD/CHF | 0.066 | 0.068 | 1.795 | 0.619 |
| COPD | 1.8 | 0.301 | 2.029 | 0.095 |
| Bedridden | 4.31 | 0.054 | 8.1 | 0.02 |
| Initial Clinical Characteristics | ||||
| Confusion | 7.736 | <0.001 | 5.416 | 0.001 |
| Hypotension SBP<90 | 3.6 | 0.011 | 3.611 | 0.003 |
| Tachycardia HR >=90 | 1.571 | 0.575 | 0.415 | 0.369 |
| Tachypnea RR>=30 | 7.032 | 0.005 | 1.86 | 0.494 |
| Fever T>=37.9 | 0.659 | 0.393 | 0.774 | 0.686 |
| High grade fever T>=38.3 | 0.684 | 0.485 | 1.178 | 0.714 |
| Initial clinical finding | ||||
| Hypoxia SaO2=<93% | 31.158 | <0.001 | 7.723 | 0.001 |
| Severe hypoxia SaO2=<85% | 11.143 | <0.001 | 8.821 | 0.001 |
| Hct<36% | 1.374 | 0.628 | 1.111 | 0.946 |
| Thrombocytopenia PLT:100000–150000/μl | 1.453 | 0.464 | 0.542 | 0.251 |
| Severe Thrombocytopenia PLT =<100000/μl | 1.327 | 0.894 | 0.763 | 0.975 |
| Leukopenia WBC<4000/μL | 1.184 | 1.000 | 1.466 | 0.866 |
| Leukocytosis WBC>12000/μL | 0.875 | 0.8 | 0.403 | 0.142 |
| BUN>20 mg/dl | 4.16 | 0.012 | 2.204 | 0.138 |
| Sodium<135 mg/dl | 2.045 | 0.432 | 3.536 | 0.056 |
| PCT level>0.5 ng/ml | 1.091 | 0.843 | 0.644 | 0.331 |
| PCT level>2 ng/ml | 3.481 | 0.003 | 1.73 | 0.211 |
| PCT level>10 ng/ml | 6.735 | 0.001 | 3.516 | 0.004 |
| Bilateral CXR involvement | 4.127 | 0.002 | 1.62 | 0.302 |
| Pleural effusion | 1.815 | 0.164 | 1.333 | 0.335 |
IVRS: Intensive Vasopressor and Respiratory Support;
IHM: In-Hospital Mortality; Yrs.: Years Old; IHD: Ischemic Heart Disease; CHF: Congestive Heart Failure; COPD: Chronic Obstructive Pulmonary Disease; SBP: Systolic Blood Pressure; HR: Heart Rate; RR: Respiratory Rate; T: Temperature; SaO2: Arterial oxygen saturation; Hct: Hematocrit; PLT: Platelets; WBC: White Blood Cells; BUN: Blood Urea Nitrogen; PCT: Procalcitonin; CXR: Chest X-ray.
The associations of PCT levels with IHM and IVRS requirement were also analyzed and showed different levels of PCT; the odds ratio values for IHM were 2.95, 1.467 and 0.595 for PCT level>10ng/mL, >2ng/mL and >0.5ng/mL, respectively. The associations for IVRS requirement were: 6.846, 3.555, and 1.136 for PCT level>10ng/mL, >2ng/mL, and >0.5ng/mL, respectively (Table 2).
Table 2.
Frequency of IHM and IVRS and LOS in patients with CAP
| PSI Class | IHM | P value | IVRS | P value | LOS (days) | P value | ICU | P value |
|---|---|---|---|---|---|---|---|---|
| I | 0 | <0.001 | 0 | <0.001 | 7 | 0.98 | 0 | <0.001 |
| II | 4.3% | 2 (8.7%) | 6.9±5.5 | 0 | ||||
| III | 4.5% | 3 (14.3%) | 10.6±12.1 | 2(9.1%) | ||||
| IV | 20% | 7 (23.3%) | 10.6±8 | 4(13.3%) | ||||
| V | 48.7% | 25 (65.8%) | 15.7±18.8 | 17(45.9%) | ||||
| CURB-65 | ||||||||
| 0 | 0 | <0.001 | 0 | <0.001 | 7.5±8.3 | 0.096 | 0 | <0.001 |
| 1 | 5 (14.7%) | 4 (12.1%) | 8.3±6 | 2 (5.9%) | ||||
| 2 | 5 (21.7%) | 8 (34.8%) | 13.3±12.8 | 6 (26.1%) | ||||
| 3 | 7 (31.8%) | 12 (54.5%) | 14.1±15.8 | 7 (31.8%) | ||||
| 4 | 7 (50%). | 11 (78.6%) | 18.7±24.6 | 6 (42.9%) | ||||
| 5 | 2 (66.7%) | 1 (50%) | 5±1.4 | 0 | ||||
| SMART-COP | ||||||||
| I | 1 (5.3%) | <0.001 | 0 | <0.001 | 5.4±2.5 | 0.003 | 0 | <0.001 |
| II | 1 (4%) | 2 (8%) | 9.7±1.9 | 2 (8.7%) | ||||
| III | 6 (18.2%) | 13 (40.6%) | 14.5±11.9 | 7 (30.4%) | ||||
| IV | 19 (59.4%) | 22 (68.8%) | 14.5±19.7 | 14 (60.9%) | ||||
| IDSA/ATS 2007 | ||||||||
| Severe | 24 (36.9%) | <0.001 | 35 (54.7%) | <0.001 | 13.8±16.2 | 0.013 | 21 (32.8%) | <0.001 |
| Non-severe | 3 (6.8%) | 2 (4.7%) | 7.9±6.6 | 1 (2.3%) |
IVRS: Intensive Vasopressor and Respiratory Support; IHM: In-Hospital Mortality; LOS: Length of Hospital Stay; CAP: Community Acquired Pneumonia; PSI: Pneumonia Severity Index; ICU: Intensive Care Unit; IDSA: Infectious Diseases Society of America; ATS: American Thoracic Society.
Comparison of PCT levels in patients with CAP of different levels of severity showed that the PCT levels increased with increasing severity of CAP according to PSI, CURB-65, SMART-COP, and IDSA/ATS scores (Table 3). It was significantly higher in patients with PSI class≥ 4 (P< 0.001), CURB-65≥ 3 (P< 0.001), SMART-COP≥ 3 (P< 0.001), and patients with severe pneumonia based on IDSA/ATS 2007 criteria (P< 0.001).
Table 3.
The PCT levels in patients with CAP of different levels of severity
| PCT level | >10 ng/mL | 2–10 ng/mL | 0.5–2 ng/mL | <0.5 ng/mL |
|---|---|---|---|---|
| PSI Classes | ||||
| I | 0 | 0 | 0 | 1 (2.9%) |
| II | 0 | 1 (5.9%) | 8 (34.8%) | 13 (38.2%) |
| III | 4 (10.8%) | 6 (35.3%) | 4 (17.45) | 7 (20.6%) |
| IV | 11 (29.7%) | 6 (35.3%) | 6 (26.1%) | 5 (14.7%) |
| V | 22 (59.5%) | 4 (23.5%) | 5 (21.7%) | 8 (23.5%) |
| Total | 37 (100%) | 17 (100%) | 23 (100%) | 34 (100%) |
| SMART-COP | ||||
| I | 1 (2.8%) | 1 (6.7%) | 5 (22.7%) | 11 (33.3%) |
| II | 3 (8.3%) | 4 (26.7%) | 9 (40.9%) | 7 (21.2%) |
| III | 9 (25%) | 7 (46.7%) | 7 (31.8%)) | 9 (27.3%) |
| IV | 23 (63.9%) | 3 (20%) | 1 (4.5%) | 6 (18.2%) |
| Total | 36 (100%) | 15 (100%) | 22 (100%) | 33 (100%) |
| CURB-65 | ||||
| 0 | 1 (2.9%) | 0 | 1 (4.5%) | 1 (2.9%) |
| 1 | 4 (11.8%) | 8 (53.3%) | 10 (45.5%) | |
| 2 | 7 (20.6%) | 3 (20%) | 6 (27.3%) | 13 (38.2%) |
| 3 | 11 (32.4%) | 3 (20%) | 3 (13.6%) | 7 (20.6%) |
| 4 | 8 (23.5%) | 1 (6.7%) | 2 (9.1%) | 5 (14.7%) |
| 5 | 3 (8.8%) | 0 | 0 | 8 (23.5%) |
| Total | 34 (100%) | 15 (100%) | 22 (100%) | 34 (100%) |
| IDSA/ATS 2007 | ||||
| Severe | 31 (88.6%) | 9 (56.3%) | 10 (43.4%) | 15 (45.5%) |
| Non-severe | 4 (11.4%) | 7 (43.8%) | 13 (56.5%) | 18 (54.5%) |
| Total | 34 (100%) | 16 (100%) | 23 (100%) | 33 (100%) |
PCT: Procalcitonin; CAP: Community Acquired Pneumonia; PSI: Pneumonia Severity Index; ICU: Intensive Care Unit; IDSA: Infectious Diseases Society of America; ATS: American Thoracic Society.
The accuracy of different scoring systems in prediction of IHM and IVRS requirement is shown in Tables 4 and 5. The accuracy of PCT levels in predicting IHM and IVRS requirement based on the AUC was 0.542 and 0.658, respectively.
Table 4.
The AUC*, best threshold, sensitivity, specificity, PPV, and NPV of different scoring systems of CAP in prediction of IVRS requirement
| Predicting the need for IVRS | AUC | Best Threshold | Sensitivity | 95% Confidence Interval | Specificity | 95% Confidence Interval | PPV | NPV | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Lower Limit | Upper Limit | Lower Limit | Upper Limit | |||||||
| PSI | 0.805 | 122 | 80.2% | 71% | 88.1% | 75.6% | 62% | 89.1% | 87.1% | 65.1% |
| CURB-65 | 0.806 | 2 | 57.9% | 46.3% | 69.5% | 88.8% | 77.7% | 97.2% | 90.9% | 52.4% |
| CRB-65 | 0.745 | 2 | 69% | 57.7% | 78.8% | 69.4% | 55.5% | 83.3% | 81.6% | 53.1% |
| IDSA/ATS Minor | ||||||||||
| SMART-COP | 0.853 | 6 | 75.7% | 65.7% | 85.7% | 64.8% | 78.3% | 91.8% | 86.8% | 63% |
| PCT level | 0.658 | >2ng/ml | 81.3% | 72% | 90.6% | 61.1% | 44.4% | 77.7% | 81.3% | 61.1% |
| PSI+PCT | 86% | 77.7% | 93% | 75% | 58.3% | 88.8% | 87.3% | 72.9% | ||
| 0.858 | >2ng/ml | |||||||||
| 83% | 75% | 91.6% | 77% | 61.1% | 91.6% | 88.2% | 70% | |||
| CURB-65+PCT | 0.835 | >2ng/ml | 96% | 88% | 100% | 61.8% | 51.3% | 72.3% | 45.2% | 97.9% |
| CRB-65+PCT | 0.865 | >2ng/ml | 64.7% | 52.9% | 76.4% | 94.2% | 85.7% | 100% | 95.6% | 57.8% |
| IDSA/ATS Minor.PCT | 0.828 | >2 ng/ml | 67.1% | 54.6% | 78.1% | 96.6% | 90% | 100% | 97.7% | 58% |
| SMART-COP+PCT | 0.881 | >2ng/ml | 79.1% | 68.6% | 88% | 88.8% | 77.7% | 97.2% | 92.9% | 69.5% |
AUC: 0.90–1 = Excellent; 0.80–0.90 = Good; 0.70–0.80 = Fair; 0.60–0.70 = Poor; 0.50–0.60 = Failed
AUC: Area Under the Curve; PPV: Positive Predictive Value; NPV: Negative Predictive Value; IVRS: Intensive Vasopressor and Respiratory Support; LOS: Length of Hospital Stay; CAP: Community Acquired Pneumonia; PSI: Pneumonia Severity Index; ICU: Intensive Care Unit; IDSA: Infectious Diseases Society of America; ATS: American Thoracic Society.
Table 5.
The AUC*, best threshold, sensitivity, specificity, PPV, and NPV of different scoring systems of CAP in prediction of IHM
| Predicting IHM | AUC | Best Threshold | Sensitivity | 95% Confidence Interval | Specificity | 95% Confidence Interval | PPV | NPV | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Lower Limit | Upper Limit | Lower Limit | Upper Limit | |||||||
| PSI | 0.796 | 122 | 80% | 64% | 96% | 74.7% | 64.3% | 83.9% | 47.6% | 92.8% |
| CURB-65 | 0.712 | 3 | 60% | 40% | 80% | 71.2% | 61.2% | 81.3% | 39.4% | 85% |
| CRB-65 | 0.699 | 2 | 65.3% | 46.1% | 84.6% | 62.9% | 53% | 72.9% | 36.1% | 85% |
| IDSA/ATS 2007 | 0.747 | --- | 52.3% | 33.3% | 76.1% | 93.3% | 86.6% | 98.6% | 68.7% | 87.5% |
| IDSA/ATS 2007 Major | 0.751 | 2 | 57.6% | 38.4% | 76.9% | 92% | 86.3% | 96.5% | 68.1% | 88% |
| IDSA/ATS 2007 Minor | 0.635 | 2 | 76.2% | 57.1% | 90.4% | 46.6% | 36% | 58.6% | 28.5% | 87.5% |
| SMART-COP | 0.817 | 6 | 84% | 68% | 96% | 70% | 60% | 80% | 46.6% | 93.3% |
| PCT level | 0.542 | >2ng/ml | 50% | 30.7% | 69.2% | 75% | 65.4% | 84.5% | 38.2% | 82.9% |
| PSI+PCT | 0.869 | >2ng/ml | 84% | 68% | 96% | 80.4% | 71.9% | 89% | 56.7% | 94.2% |
| CURB-65+PCT | 0.835 | >2ng/ml | 96% | 88% | 100% | 61.8% | 51.3% | 72.3% | 45.2% | 97.9% |
| CRB-65+PCT | 0.822 | >2ng/ml | 100% | 100% | 100% | 53.2% | 41.5% | 64.9% | 41.9% | 100% |
| SMART-COP+PCT | 0.852 | >2ng/ml | 88% | 72% | 100% | 73.6% | 63.1% | 82.9% | 52.3% | 94.9% |
| IDSA/ATS 2007+PCT | 0.868 | >2ng/ml | 76.1% | 57.1% | 95.2% | 84.5% | 76% | 92.9% | 59.2% | 92.3% |
| IDSA/ATS Major.PCT | 0.838 | >2 ng/ml | 88.4% | 76.9% | 100% | 62.6% | 51.8% | 72.3% | 42.5% | 94.5% |
| IDSA/ATS Minor.PCT | 0.756 | >2ng/ml | 80.9% | 61.9% | 95.2% | 67.6% | 56.3% | 77.4% | 42.5% | 92.3% |
AUC: 0.90–1 = Excellent; 0.80–0.90 = Good; 0.70–0.80 = Fair; 0.60–0.70 = Poor; 0.50–0.60 = Failed
AUC: Area Under the Curve; PPV: Positive Predictive Value; NPV: Negative Predictive Value; IHM: In-hospital Mortality; LOS: Length of Hospital Stay; CAP: Community Acquired Pneumonia; PSI: Pneumonia Severity Index; ICU: Intensive Care Unit; IDSA: Infectious Diseases Society of America; ATS: American Thoracic Society.
Comparison between the AUC values of different scoring systems of CAP in predicting IHM and IVRS requirement showed that AUC value of each scoring system increased when we added PCT levels to them as risk factor for IHM and IVRS (Tables 4 and 5, Figure 1).
Figure 1.
ROC curve of PCT level, scoring systems, and scoring systems plus PCT level in prediction of hospital mortality and IVRS requirement
The best threshold for prediction of IVRS requirement and IHM was also re-calculated for each defined scoring system alone and after addition of PCT levels and the results are shown in Tables 4 and 5 and Figure 2.
Figure 2.
Comparative ROC curve of suggested models in prediction of IHM (A) and IVRS requirement (B)
DISCUSSION
Almost all of the major decisions regarding management of CAP, including diagnostic and treatment issues revolve around the initial assessment of severity (1). Several studies assessed and compared different scoring predictors of pneumonia severity. Some authors concluded that there is no significant difference in the capability of each scoring system to effectively predict CAP mortality (12), whereas others noted that different severity scores have different strengths and weaknesses as prediction tools (13). In our study, patients with CAP were evaluated for severity of illness according to different scoring systems and the association between several demographic, clinical, laboratory and radiographic characteristics and prognosis was analyzed. We found several risk factors that significantly increased patient’s need for IVRS and/or IHM. Risk factors for IVRS requirement (in decreasing order of significance) included hypoxia, altered mental status, tachypnea, bedridden status, blood urea nitrogen (BUN)>20 mg/dL, bilateral involvement, hypotension, PCT level>2ng/mL, hyponatremia and opium addiction. Factors related to increased IHM were bedridden status, hypoxia, confusion, hypotension, hyponatremia, diabetes mellitus, BUN>20 mg/dL and COPD.
Several biomarkers and cytokines have been proposed as potential predictors of pneumonia. Among them the predictive capability of CRP and PCT has been most widely studied and validated. The PCT has been widely studied for its usefulness in decision-making of whether to use antibacterial agents in patients with pneumonia or not (14). According to a meta-analysis conducted by Li et al, PCT-guided antibiotic therapy in patients with respiratory tract infections reduces antibiotic use without affecting overall mortality or LOS in the hospital (15). More recently, analysis of eight trials (n=3,492) addressed initiation and/or discontinuation of antibiotics in patients with acute upper and lower respiratory tract infections, which provided evidence that PCT guidance reduces antibiotic duration and prescription rates. There is also evidence that PCT guidance did not increase mortality, hospital LOS or ICU admission rates (14).
The most recent international consensus guidelines defined elevated serum PCT level as a warning sign of incipient severe sepsis and septic shock (16). It has also been shown that PCT levels predict bacteremia (17, 18). Johansson et al. found that median PCT levels were higher in bacteremic patients (than in those without bacteremia), in patients with non-bacteremic pneumococcal etiology (than in those infected with other bacteria) and in patients with pneumococcal etiology (as compared with viral etiology). They suggested high level of PCT to be a good marker for invasive disease and pneumococcal etiology (19).
Numerous studies have been conducted on the potential role of PCT as a prognostic biomarker (20–24). They found that PCT levels increase with increasing severity of sepsis and organ dysfunction. Several other studies have shown that PCT levels help predict the severity of pneumonia and may predict survival based on the magnitude of the result (9–11). It has been suggested that PCT should be regarded as a prognostic rather than a diagnostic factor in patients with CAP (25). Christ-Crain et al. found gradual increase of PCT levels with increasing severity of CAP, classified according to PSI score (P< 0.001) (26). Similarly, another study conducted by Albrich et al. showed its usefulness as a severity marker for pneumococcal pneumonia in HIV-infected adults (27). The prognostic value of PCT has also been demonstrated in patients with VAP. Abula et al. found that the increased PCT levels in VAP patients were associated with poor control of infection and subsequent deterioration (28). Also, PCT levels were associated with severity of pneumonia in our patients. We found that the PCT levels were significantly higher in patients with PSI class≥ 4, CURB-65≥ 3, SMART-COP≥ 3 and patients with severe pneumonia based on IDSA/ATS 2007 criteria. Menéndez et al. found that serum levels of CRP, IL6 and PCT were good predictors of early treatment failure, and that adding the PSI risk classes did not improve the sensitivity or negative predictive values (29). In another study conducted by Menéndez et al, among other biomarkers and cytokines including PCT, CRP, IL6, IL8, IL10 and TNFα, PCT had the highest positive correlation with PSI, CURB65 and CRB65 scales (30). However, while we calculated the power of PCT levels in predicting the outcome of pneumonia, the accuracy of PCT levels in predicting IHM and IVRS requirement based on the area under the ROC curve (AUC) was only 0.542 and 0.658, respectively; the best threshold was ≥ 2ng/mL for both of them. In a similar study, Menéndez et al. found AUC of only 0.66 (0.56–0.76) for PCT levels in patients with CAP (30). However, another study that included both out-patient and in-patient settings reported a comparable accuracy for PCT and CRB-65, based on AUC [0.80 (0.75–0.84) versus 0.79 (0.74–0.84)] (31). Berg and Lindhardt conducted a systematic review on the role of PCT in adult patients with CAP and suggested that although complications during admission, severity of disease measured with various scales (mostly PSI, CRB-65 and CURB-65) and death within a month all tend to correlate with higher PCT, no definite cut-off was found, and PCT should always be interpreted carefully (25). Based on the results of our study, in comparison with the defined severity scores, PCT level alone is a week predictor of pneumonia severity.
In the second arm of the study, we added the PCT level factor to each scoring system as a risk factor for IHM and IVRS requirement and calculated the AUC for each new model. Previously, Huang et al, in 2008 demonstrated that adding PCT levels to the assessment of high clinical risk patients significantly improved the possibility to rule out the likelihood of death. They also demonstrated that simply adding PCT test results to the PSI and CURB-65 in all subjects led to only minimal improvement in test performance. In their study, the AUC of PSI score increased from 0.83 to 0.85 (21). We also demonstrated that adding the PCT levels to the defined scoring systems resulted in improvement of the AUC for each score. Menéndez et al also calculated the AUCs of the different logistic regression models with combinations of markers and cytokines added to the prognostic scales. They found that CRP significantly improved the diagnostic value of both PSI and the CURB65/CRB65 scales. The best AUC (0.88) in their study was achieved with the PSI together with CURB65 and CRP; whereas the AUC of prognostic scales in prediction of 30-day mortality after adding PCT level to PSI, CRB-65 and CURB-65 was 0.83 (0.77–0.89), 0.83 (0.76–0.89), and 0.84 (0.77–0.90), respectively (30). In our study, the best accuracy for prediction of IHM was obtained for PSI plus PCT and IDSA/ATS 2007 plus PCT with AUC of 0.869 and 0.868, respectively. The best accuracy for predicting IVRS requirement was obtained for SMART-COP plus PCT with AUC of 0.881.
Schuetz et al. assessed the need for recalibration of well-established pneumonia severity prediction scores in a special population. Accordingly, because management strategies of patients with CAP depend on cut-off values of absolute predicted mortalities and the observation of discordance between the reported mortality rates of patients with CAP and original studies, it is essential that predicted risks agree with observed risks in the studied population. They referred to it as calibration. Miscalibration may lead to risk underestimation or overestimation (32). We recalculated the best threshold of predicting IHM and IVRS requirement for each validated scoring system alone and after addition of PCT levels. The analysis resulted in only slightly different values compared with original studies.
CONCLUSION
Our study is one of the first to prospectively analyze the power of different scoring systems and PCT level alone and in combination for prediction of both IHM and IVRS requirement. Upon calculating the power of PCT levels in predicting the outcome of pneumonia, we found that PCT level alone was a week predictor of pneumonia severity. Therefore, we do not suggest using PCT level alone as a predictor for mortality and IVRS requirement in patients who present with CAP. We suggest PSI plus PCT and IDSA/ATS 2007 plus PCT as accurate predictors for IHM and SMART-COP plus PCT for IVRS requirement in hospitalized CAP patients (Figure 2). We also suggest using CRB-65 plus PCT instead of CURB-65 for predicting IHM and IVRS requirement with better accuracy (0.882 and 0.865 compared with 0.712 and 0.806, respectively). However, the present study has also some limitations, such as the limited number of patients, single-center design, assessment of hospitalized patients only and semi-quantitative technique for measuring PCT levels. In addition, because of the limited number of ICU beds, the number of patients admitted to the ICU was far less than the actual number of patients who needed intensive care. The difference in the efficacy and quality of intensive care among patients with severe illness could be a potential confounder in our study. Further large-scale, randomized controlled trials are recommended.
Acknowledgment
This article has been derived in part from the thesis of the corresponding author for a specialty degree in Infectious Diseases.
Footnotes
Conflict of Interests
None.
Author Contributions
HamidReza Naderi: study conception and design, drafting article, critical revision of article, and final approval of article; Fereshte Sheybani: study conception and design, data acquisition, data analysis and interpretation, drafting article, critical revision of article, and final approval of article; MohammadReza Sarvghad: study conception and design, drafting article, critical revision of article, and final approval of article; Mehdi Jabbari Nooghabi: study conception and design, data analysis and interpretation, statistics, drafting article, critical revision of article, and final approval of article.
REFERENCES
- 1. Mandell LA, Wunderink RG, Anzueto A, Bartlett JG, Campbell GD, Dean NC, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007; 44 Suppl 2: S27– 72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997; 336 ( 4): 243– 50. [DOI] [PubMed] [Google Scholar]
- 3. Lim WS, van der Eerden MM, Laing R, Boersma WG, Karalus N, Town GI, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003; 58 ( 5): 377– 82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Bauer TT, Ewig S, Marre R, Suttorp N, Welte T, CAPNETZ Study Group CRB-65 predicts death from community-acquired pneumonia. J Intern Med 2006; 260 ( 1): 93– 101. [DOI] [PubMed] [Google Scholar]
- 5. Riley PD, Aronsky D, Dean NC. Validation of the 2001 American Thoracic Society criteria for severe community-acquired pneumonia. Crit Care Med 2004; 32 ( 12): 2398– 402. [DOI] [PubMed] [Google Scholar]
- 6. Charles PG, Wolfe R, Whitby M, Fine MJ, Fuller AJ, Stirling R, et al. SMART-COP: a tool for predicting the need for intensive respiratory or vasopressor support in community-acquired pneumonia. Clin Infect Dis 2008; 47 ( 3): 375– 84. [DOI] [PubMed] [Google Scholar]
- 7. Chalmers JD, Singanayagam A, Hill AT. C-reactive protein is an independent predictor of severity in community-acquired pneumonia. Am J Med 2008; 121 ( 3): 219– 25. [DOI] [PubMed] [Google Scholar]
- 8. Iwata K, Kagawa H. C-reactive protein is an independent predictor of severity in community-acquired pneumonia: what does it add to? Am J Med 2008; 121 ( 12): e7; author reply e9. [DOI] [PubMed] [Google Scholar]
- 9. Kim JH, Seo JW, Mok JH, Kim MH, Cho WH, Lee K, et al. Usefulness of plasma procalcitonin to predict severity in elderly patients with community-acquired pneumonia. Tuberc Respir Dis (Seoul) 2013; 74 ( 5): 207– 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Masiá M, Gutiérrez F, Shum C, Padilla S, Navarro JC, Flores E, et al. Usefulness of procalcitonin levels in community-acquired pneumonia according to the patients outcome research team pneumonia severity index. Chest 2005; 128 ( 4): 2223– 9. [DOI] [PubMed] [Google Scholar]
- 11. Boussekey N, Leroy O, Alfandari S, Devos P, Georges H, Guery B. Procalcitonin kinetics in the prognosis of severe community-acquired pneumonia. Intensive Care Med 2006; 32 ( 3): 469– 72. [DOI] [PubMed] [Google Scholar]
- 12. Yang Y, Xu F, Shi LY, Diao R, Cheng YS, Chen XY, et al. Efficacy and significance of various scores for pneumonia severity in the management of patients with community-acquired pneumonia in China. Chin Med J (Engl) 2012; 125 ( 4): 639– 45. [PubMed] [Google Scholar]
- 13. Buising KL, Thursky KA, Black JF, MacGregor L, Street AC, Kennedy MP, et al. A prospective comparison of severity scores for identifying patients with severe community acquired pneumonia: reconsidering what is meant by severe pneumonia. Thorax 2006; 61( 5): 419– 24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Soni NJ, Samson DJ, Galaydick JL, Vats V, Pitrak DL, Aronson N. Procalcitonin-Guided Antibiotic Therapy [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2012. Oct. (Comparative Effectiveness Reviews, No. 78.) Summary and Discussion. Available from: http://www.ncbi.nlm.nih.gov/books/NBK114999/ [PubMed] [Google Scholar]
- 15. Li H, Luo YF, Blackwell TS, Xie CM. Meta-analysis and systematic review of procalcitonin-guided therapy in respiratory tract infections. Antimicrob Agents Chemother 2011; 55 ( 12): 5900– 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Angus DC, van der Poll T. Severe sepsis and septic shock. N Engl J Med 2013; 369 ( 9): 840– 51. [DOI] [PubMed] [Google Scholar]
- 17. Müller F, Christ-Crain M, Bregenzer T, Krause M, Zimmerli W, Mueller B, et al. Procalcitonin levels predict bacteremia in patients with community-acquired pneumonia: a prospective cohort trial. Chest 2010; 138 ( 1): 121– 9. [DOI] [PubMed] [Google Scholar]
- 18. Julián-Jiménez A, Timón Zapata J, Laserna Mendieta EJ, Parejo Miguez R, Flores Chacartegui M, Gallardo Schall P. Ability of procalcitonin to predict bacteremia in patients with community acquired pneumonia. Med Clin (Barc) 2014; 142 ( 7): 285– 92. [DOI] [PubMed] [Google Scholar]
- 19. Johansson N, Kalin M, Backman-Johansson C, Larsson A, Nilsson K, Hedlund J. Procalcitonin levels in community-acquired pneumonia - correlation with aetiology and severity. Scand J Infect Dis 2014; 46 ( 11): 787– 91. [DOI] [PubMed] [Google Scholar]
- 20. Kibe S, Adams K, Barlow G. Diagnostic and prognostic biomarkers of sepsis in critical care. J Antimicrob Chemother 2011; 66 Suppl 2: ii33– 40. [DOI] [PubMed] [Google Scholar]
- 21. Huang DT, Weissfeld LA, Kellum JA, Yealy DM, Kong L, Martino M, et al. Risk prediction with procalcitonin and clinical rules in community-acquired pneumonia. Ann Emerg Med 2008; 52 ( 1): 48– 58.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Masiá M, Gutiérrez F, Shum C, Padilla S, Navarro JC, Flores E, et al. Usefulness of procalcitonin levels in community-acquired pneumonia according to the patients outcome research team pneumonia severity index. Chest 2005; 128 ( 4): 2223– 9. [DOI] [PubMed] [Google Scholar]
- 23. Okimoto N, Hayashi Y, Ishiga M, Nanba F, Kishimoto M, Yagi S, et al. Procalcitonin and severity of community-acquired pneumonia. J Infect Chemother 2009; 15 ( 6): 426– 7. [DOI] [PubMed] [Google Scholar]
- 24. Lee M, Snyder A. The role of procalcitonin in community-acquired pneumonia: a literature review. Adv Emerg Nurs J 2012; 34 ( 3): 259– 71. [DOI] [PubMed] [Google Scholar]
- 25. Berg P, Lindhardt BØ. The role of procalcitonin in adult patients with community-acquired pneumonia--a systematic review. Dan Med J 2012; 59 ( 3): A4357. [PubMed] [Google Scholar]
- 26. Christ-Crain M, Morgenthaler NG, Stolz D, Müller C, Bingisser R, Harbarth S, et al. Pro-adrenomedullin to predict severity and outcome in community-acquired pneumonia [ISRCTN04176397]. Crit Care 2006; 10 ( 3): R96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Albrich WC, Madhi SA, Adrian PV, van Niekerk N, Telles JN, Ebrahim N, et al. Pneumococcal colonisation density: a new marker for disease severity in HIV-infected adults with pneumonia. BMJ Open 2014; 4 ( 8): e005953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Abula A, Wang Y, Ma L, Yu X. The application value of the procalcitonin clearance rate on therapeutic effect and prognosis of ventilator associated pneumonia. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue 2014; 26 ( 11): 780– 4. [DOI] [PubMed] [Google Scholar]
- 29. Menéndez R, Cavalcanti M, Reyes S, Mensa J, Martinez R, Marcos MA, et al. Markers of treatment failure in hospitalised community acquired pneumonia. Thorax 2008; 63 ( 5): 447– 52. [DOI] [PubMed] [Google Scholar]
- 30. Menéndez R, Martínez R, Reyes S, Mensa J, Filella X, Marcos MA, et al. Biomarkers improve mortality prediction by prognostic scales in community-acquired pneumonia. Thorax 2009; 64 ( 7): 587– 91. [DOI] [PubMed] [Google Scholar]
- 31. Krüger S, Ewig S, Marre R, Papassotiriou J, Richter K, von Baum H, et al. Procalcitonin predicts patients at low risk of death from community-acquired pneumonia across all CRB-65 classes. Eur Respir J 2008; 31 ( 2): 349– 55. [DOI] [PubMed] [Google Scholar]
- 32. Schuetz P, Koller M, Christ-Crain M, Steyerberg E, Stolz D, Müller C, et al. Predicting mortality with pneumonia severity scores: importance of model recalibration to local settings. Epidemiol Infect 2008; 136 ( 12): 1628– 37. [DOI] [PMC free article] [PubMed] [Google Scholar]