C-Reactive Protein Is an Independent Predictor of Severity in Community-acquired Pneumonia

Methods 

We conducted a prospective study of all adult patients admitted between February 2005 and February 2007 with a primary diagnosis of community-acquired pneumonia. Ethical approval was obtained from the Lothian research ethics committee.

The main inclusion criteria were presentation to hospital with a diagnosis of community-acquired pneumonia within the study period (February 2005 to February 2007) and the absence of any exclusion criteria. Community-acquired pneumonia was defined as a history consistent with pneumonia (with 1 or more of the following: new-onset shortness of breath, cough, sputum production, haemoptysis, chest pain, new-onset confusion, or pyrexia) and new infiltrates on the chest radiograph.

In addition, the authors reviewed patient records at follow-up clinic review to ensure that the discharge diagnosis was considered to be pneumonia and that no exclusion criteria were present. Exclusion criteria are shown in Table 1.

Table 1. Study Exclusion Criteria

	Hospital-acquired pneumonia—development of symptoms >48 hours after admission to hospital or discharge from an acute care facility within 14 days of admission
	Active thoracic or extrathoracic malignancy
	Conditions likely to cause diagnostic confusion or where chest radiograph changes are equivocal (eg, pulmonary fibrosis, allergic bronchopulmonary aspergillosis)
	Chronic lung disease (chronic obstructive pulmonary disease, bronchiectasis, chronic asthma)
	Immunosuppression (iatrogenic or acquired)
	Solid organ transplant
	Hematological disorders including hematological malignancy
	Chronic liver disease or cirrhosis
	Other acute co-morbid illnesses leading to physiological or metabolic derangement such that pneumonia severity assessment would be inappropriate (eg, acute pulmonary embolism)
	Patients for whom active treatment is not considered appropriate (eg, palliative care)

Results 

There were 936 patients considered for inclusion during the study period; 570 patients met the criteria and were included in the study. Reasons for exclusion are shown in Table 2.

Table 2. Patients Excluded from Study

	Reason for Exclusion	n
	Hospital acquired pneumonia	31
	Active malignancy	54
	Immunosuppression	29
	Solid organ transplant	2
	Chronic liver disease	6
	Persistent shadowing on chest radiograph at follow-up	49
	Active treatment not appropriate	18
	Diagnosis other than pneumonia made at follow-up	20
	Diagnosis of lung cancer at follow-up	33
	Chronic lung disease	124

Baseline characteristics of the study population are shown in Table 3. Characteristics are shown for the whole population and for those patients who had repeat C-reactive protein measurements available during the first 4 days of treatment.

Table 3. Population Characteristics

	Baseline Characteristics	Study Population	Patients with Repeat CRP Measurement Up to Day 4
	Number	570	358
	Age (years)	62(44-76)	62(44-78)
	Sex - % male	48.8%	48%
	Duration of admission (days)	5(2-11)	6(3-11)
	Chronic cardiac disease	75(13.2%)	49(13.7%)
	Cerebrovascular disease	51(8.9%)	35(9.8%)
	Chronic renal failure	25(4.4%)	16(4.4%)
	Diabetes mellitus	38(6.6%)	18(5%)
	Current smokers	204(36%)	121(33.8%)
	Exsmokers	145(25%)	87(24.3%)
	Nonsmokers	221(39%)	150(41.9%)

CRP=C-reactive protein. Definitions for co-morbid conditions were as used in previous studies.4 Data presented as median (interquartile range) or % number of patients.

Of all patients, 20.7% were discharged within 24 hours of admission; 13.5% required invasive ventilation and/or inotropic support. Complicated pneumonia developed in 7.3% of patients. The 30-day mortality rate was 9.6%, similar to previous studies by Lim et al3 and Fine at al4 (Table 4).

Table 4. Mortality and Existing Severity Scores (CURB65 and Pneumonia Severity Index)

	CURB65 Score	n	30-Day Mortality (All Patients)	National Studies (Lim et al)3	PSI Class	n	30-Day Mortality (All Patients)	National Studies (Fine et al)4
	0	105	0%	0.7%	1	81	0%	0.1-0.4%
	1	128	2.3%	3.2%	2	131	1.5%	0.6-0.7%
	2	142	3.5%	3%	3	108	2.8%	0.9-2.8%
	3	110	15%	17%	4	159	10.7%	4-10%
	4	70	31%	41.5%	5	91	35.2%	27%
	5	15	47%	57%
	Total	570	9.6%	9%	Total	570	9.6%	10.6%

PSI=pneumonia severity index.

Admission C-Reactive Protein as a Predictor of 30-day Mortality, Need for Invasive Ventilation and/or Inotropic Support, and Complicated Pneumonia 

Thirty-day mortality, need for invasive ventilation and/or inotropic support, and rates of complicated pneumonia increase with increasing levels of CRP on admission (Table 5).

Table 5. Admission C-Reactive Protein (CRP) and Adverse Outcomes

	C-Reactive Protein (mg/L)	n	Mortality (30-Day)	Invasive Ventilation and/or Inotropic Support	Complicated Pneumonia
	CRP ≥400	70	18.8%	33.3%	21.7%
	CRP 300-399	82	18.8%	20%	7.5%
	CRP 200-299	106	9.6%	13.5%	15.4%
	CRP 100-199	132	10.7%	15.2%	3.1%
	CRP <100	180	1.1%	2.2%	0.6%
	CRP <50	115	0.9%	0.9%	0%
	CRP <10	39	0%	0%	0%

On multivariate logistic regression, a CRP <100 mg/L was independently associated with a reduced risk of:

•30-day mortality (odds ratio [OR] 0.18; 0.04-0.85; P=.03)

•Need for invasive ventilation and/or inotropic support (OR 0.21; 0.14-0.4; P=.002)

•Complicated pneumonia (OR 0.05; 0.01-0.35; P=.003)

Predictive Value of C-Reactive Protein for Outcome in Community-acquired Pneumonia 

C-reactive protein had high negative predictive values for excluding 30-day mortality up to levels <100 mg/L. The performance of CRP using a cut-off of 100 mg/L was compared with the existing severity criteria: CURB653 and the Pneumonia Severity Index scores4 (Table 6).

Table 6. C-Reactive Protein as a Predictor of Severity Compared with CURB65 and Pneumonia Severity Index Scores

		PPV	NPV	Sensitivity	Specificity	AUC
	Prediction of 30-day mortality
	CRP ≥100 mg/L	13.3%	98.9%	96.3%	34.5%	0.70(0.66-0.74)
	CURB65 score ≥3	23.6%	97.6%	85.2%	71.1%	0.83(0.80-0.87)
	Pneumonia Severity Index ≥3	14.5%	99.1%	96.3%	40.7%	0.83(0.80-0.87)
	Prediction of mechanical ventilation and/or inotropic support
	CRP ≥100 mg/L	18.7%	97.8%	94.8%	35.7%	0.71(0.67-0.74)
	CURB65 score ≥3	22.0%	95.9%	81.6%	59.9%	0.77(0.74-0.81)
	Pneumonia Severity Index ≥3	20.7%	99%	97.4%	41.8%	0.79(0.75-0.82)
	Prediction of complicated pneumonia
	CRP ≥100 mg/L	10.5%	99.4%	97.6%	33.9%	0.76(0.71-0.80)
	CURB65 score ≥3	5.7%	94.8%	40.6%	61.5%	0.54(0.48-0.59)
	Pneumonia Severity Index ≥3	7.4%	97.1%	81.3%	37.9%	0.60(0.55-0.65)

PPV= positive predictive value; NPV= negative predictive value; AUC= area under the receiver operator characteristic curve.

C-reactive protein <100 mg/L showed comparable negative predictive values for excluding 30-day mortality, need for mechanical ventilation, and/or inotropic support and complicated pneumonia. The area under the ROC curve for CRP was lower for 30-day mortality and need for mechanical ventilation and/or inotropic support, but was higher for predicting complicated pneumonia.

Admission CRP Combined with Day 4 CRP as a Predictor of Outcome 

Repeat CRP measurements at day 4 were available in 268 patients (223 were discharged in <4 days, 26 died or were admitted to the Intensive Care Unit within 4 days with no repeat measurement available, and data were missing in 53 patients).

A C-reactive protein level that fails to fall by 50% or more within 4 days is associated with increased 30-day mortality, increased rates of mechanical ventilation and/or need for inotropic support, and a higher incidence of complicated pneumonia (Table 7).

Table 7. Admission C-Reactive Protein with Repeat Measurement at Day 4

	C-Reactive Protein (mg/L)	n	Day 4 CRP	Mortality (30 Days)	Invasive Ventilation/Inotropic Support	Complicated Pneumonia
	All patients	570	n/a	9.6%	13.5%	7.3%
	All patients	175	Decreased by ≥50%	0.5%	1.7%	2.3%
	All patients	93	Increased/decreased by <50%	18.3%⁎	22.6%⁎	19.4%⁎
	CRP <100	83	Decreased by ≥50%	0%	0%	0%
	CRP <100	13	Increased/decreased by <50%	15.4%⁎	7.7%†	15.4%⁎
	CRP ≥100	92	Decreased by ≥50%	1.1%	3.3%	4.3%
	CRP ≥100	80	Increased/decreased by <50%	18.8%⁎	26.3%⁎	20%⁎

Comparison between groups where CRP decreased by 50% or more compared with groups where CRP decreased by less than 50% or increased. n/a=not applicable.

⁎ P <.001. † P=.01.

On multivariate logistic regression, a CRP that failed to fall by 50% or more at day 4 was associated with increased:

•30-day mortality (OR 24.5; 6.4-93.4; P <.0001)

•Need for invasive ventilation and/or inotropic support (OR 7.1; 2.8-17.8; P <.0001)

•Complicated pneumonia (OR 15.4; 6.32-37.6; P <.0001)

Discussion 

In this study, low CRP levels (<100 mg/L) have high negative predictive values in excluding 30-day mortality, requirement for invasive ventilation and/or inotropic support, and complicated pneumonia. In addition, repeat measurement of CRP at day 4 is shown to be a powerful marker of treatment response—patients in whom the CRP level falls by 50% or more in 4 days have low rates of 30-day mortality, requirement for invasive ventilation and/or inotropic support, and development of complicated pneumonia. CRP may therefore be a useful adjunct to clinical judgement in identifying low-risk patients.

C-reactive protein is an acute phase protein synthesized by the liver in response to tissue damage. Interleukin-6 (IL-6) is thought to be the primary trigger of CRP release,12 although tumor necrosis factor (TNF)-alpha, IL-1, and other cytokines are thought to be involved. Studies of cytokines and inflammatory markers in community-acquired pneumonia have not translated into clinically useful tests, in part because TNF-alpha and interleukin-6 are detectable in only a minority of patients.5, 13 Evidence of a relationship with severity has been conflicting, with some studies showing that TNF-alpha,14 IL-6, and soluble interleukin-2 receptor (IL-2R) do not correlate with severity.13 Others, however, have shown that IL-6 levels correlate with BTS severity score15 and that IL-6 and IL-10 (an anti-inflammatory cytokine) correlate with Apache II score and are higher in patients fitting the systemic inflammatory response syndrome criteria compared with patients who do not.16 TNF alpha, IL-6, and IL-1 beta levels appear also to be higher in patients admitted to the Intensive Care Unit with pneumonia compared with those with less severe pneumonia.17

As these cytokines are the primary stimulus for C-reactive protein release, it would seem to follow that CRP also should be higher in patients with severe pneumonia. Despite this, no large studies have examined CRP and mortality in community-acquired pneumonia. Studies have shown that elevated CRP in community-acquired pneumonia is independently associated with requirement for inpatient care,18 and that higher CRP levels result in longer duration of hospital stay and poorer clinical and radiological recovery.10 Elevated CRP also has been shown to be associated with increased mortality in lower respiratory tract infection in primary care.19

This is the first study of this size to examine whether CRP levels predict 30-day mortality in community-acquired pneumonia. Severity assessment is an important initial step in the management of community-acquired pneumonia. The difficulty of this initial assessment has led to development of a number of severity scores and indices to assist clinical decision-making. In the UK, the CURB65 score has become quickly established as an excellent tool for predicting patients at high risk of mortality. The CURB65 score was designed and is validated to predict 30-day mortality. It is not intended or validated to predict other outcomes, and studies suggest its performance is reduced when other outcome measures are considered.20

In the present study, elevated CRP≥100 mg/L was associated not only with increased 30-day mortality, but also was a marker of requirement for invasive ventilation and/or inotropic support and development of complicated pneumonia. The high negative predictive value of CRP levels below 100 mg/L for each of these outcomes can reassure clinicians and has the potential to aid the initial decision to admit or discharge patients from hospital.

This study has shown that CRP <100 mg/L on admission has similar negative predictive values to the CURB65 and Pneumonia Severity Index scores for predicting 30-day mortality, need for mechanical ventilation and/or inotropic support, and the development of complicated pneumonia. The ROC curves demonstrate that CURB65 and Pneumonia Severity Index had superior performance for predicting 30-day mortality and need for mechanical ventilation and/or inotropic support. The area under the ROC curve for CRP appears to be reduced primarily due to its lower specificity at high levels of CRP (≥ 200 mg/L). The authors feel that the negative predictive value of a test is the most clinically relevant. C-reactive protein had, however, superior performance for predicting complicated pneumonia. The authors are not aware of any previous study that examined the predictive value of CURB65 or the Pneumonia Severity Index scores for complicated pneumonia, and the finding that neither CURB65 nor the Pneumonia severity index can reliably predict complicated pneumonia requires further study.

C-reactive protein <100 mg/L has a high negative predictive value for 30-day mortality, requirement for invasive ventilation and/or inotropic support, and for development of complicated pneumonia, suggesting that a low CRP can predict patients who may be safely discharged on clinical grounds. This study found that around 30% of patients presenting with community-acquired pneumonia will fall into this category. This is supported in the literature, where 25% of patients presenting with community-acquired pneumonia have a CRP <100 mg/L.5 C-reactive protein has the potential to accurately identify a significant proportion of patients presenting with community-acquired pneumonia as low risk.

One of the major advantages of CRP is that serial measurements can be taken as a marker of treatment response. Our study suggests that by using repeat measurements on day 4, as recommended by the British Thoracic Society, patients can be reclassified as low or high risk, leading to a significant improvement in risk stratification. Currently, objective severity assessment criteria are available only for patients on admission. There is no reason why severity assessment should end at the hospital “front door,” and the finding in this study that patients in whom CRP falls by 50% or more in 4 days have low 30-day mortality, need for mechanical ventilation and/or inotropic support, and complicated pneumonia should be of value to clinicians.

Conclusion 

Low admission C-reactive protein levels <100 mg/L effectively excludes severe community-acquired pneumonia and can be used as an adjunct to clinical judgment to identify low-risk patients who may be safely discharged. C-reactive protein <100 mg/L provides a high negative predictive value comparable with CURB65 and PSI severity rules. In patients admitted to the hospital, a CRP level that falls by 50% or more in 4 days indicates a low risk of 30-day mortality, need for mechanical ventilation and/or inotropic support, or the development of complicated pneumonia.