Description of the condition
Pneumonia is an acute inflammation of the lungs caused by pathogens such as bacteria, viruses, mycoplasma, chlamydia, fungi and parasites. It is a common illness worldwide and is a major cause of death among all age groups. Symptoms include cough, sputum production, fever, chills, fatigue, shortness of breath, night sweats and pleuritic chest pain. Other symptoms may include loss of appetite, skin discolouration, nausea, vomiting, mood swings and joint pains or muscle aches (Hoare 2006). Chemotherapy or radiation therapy can also cause non-infective pneumonitis.
Pneumonia can be divided into community-acquired pneumonia (CAP), nosocomial (hospital-acquired) pneumonia (HAP) and aspiration pneumonia. CAP is the most common type of pneumonia and occurs when one is infected by pathogens without being recently hospitalized. People of all ages can contract CAP and the most common causative agents are Streptococcus pneumoniae (S. pneumoniae), Mycoplasma pneumoniae (M. pneumoniae), Haemophilus influenzae (H. influenzae), viruses and atypical bacteria. The annual incidence of CAP in Europe and North America is 34 to 40 cases per 1000 in children (Ostapchuk 2004), six per 1000 in the 18 to 59 year old population and 20 to 34 per 1000 in those aged 60 and older. It occurs three to four million times per year in the United States, accounting for about 500,000 hospitalizations annually (Plouffe 1996).
Nosocomial pneumonia, also called hospital-acquired pneumonia (HAP), is defined as pneumonia occurring in the first 48 hours after hospital admission (Leroy 2004). Approximately one-half of HAP cases occur in patients admitted to medical or surgical wards. The remaining episodes occur in patients admitted to intensive care units (ICUs) and are related to mechanical ventilation (Leroy 2004). Individuals with HAP usually have an underlying illness and have been exposed to bacteria, therefore it tends to be more serious than CAP (Gerald 2004).
Aspiration pneumonia (or aspiration pneumonitis) comprises a small percentage of community-acquired and hospital-acquired pneumonia. Aspiration pneumonitis occurs rapidly and some cases may be associated with inhalation of a large volume of sterile acidic gastric contents or aspirating other foreign objects.
Description of the intervention
Many patients with pneumonia do not require hospitalization and usually recover fully with oral antibiotics, fluids and rest. However, older people, people who have difficulty in breathing, or those who have severe medical conditions are at greater risk (d'Escrivan 2005) and may require corticosteroids (Garcia-Vidal 2007). The natural steroid cortisol is produced by the adrenal glands. Steroids, also known as corticosteroids, are a class of steroid hormones. Synthetic derivatives of the natural steroid include prednisone, prednisolone, methylprednisolone, betamethasone, dexamethasone, triamcinolone and hydrocortisone. It is possible to administer corticosteroids by inhalation, orally or intravenously.
How the intervention might work
Corticosteroids influence immune regulation and also effect carbohydrate metabolism, protein catabolism, electrolyte balance and stress response. They are used for treating inflammatory diseases of the bowel (colitis), joints (arthritis), skin (dermatitis) and lungs (pneumonia or asthma). Corticosteroids act partly by inducing anti-inflammatory genes which repress inflammatory genes (Adcock 2000). A retrospective observational study suggests that systemic corticosteroids may reduce mortality in severe CAP cases (Garcia-Vidal 2007).
However, corticosteroids have side effects, most of which are related to the dose and duration of therapy (Seale 1986). The side effects only manifest after a long period of high-dose usage and the most common are disturbance of metabolism, immune depression, prolonged healing, growth retardation in children, hirsutism, diabetes, Cushing's Syndrome (also called hypercortisolism) and thinning of the bones (osteoporosis), particularly in women during and following menopause. In some cases they can cause emotional disturbances such as depression. Many of the adverse effects only occur over prolonged administration and most short-term adverse events are reversible when the drug is discontinued.
Why it is important to do this review
The benefits of corticosteroids for severe pneumonia remain unclear, although they are sometimes used in clinical practice. There is therefore a need to systematically review the effectiveness of corticosteroids for pneumonia.
To assess the efficacy and safety of corticosteroids in the treatment of pneumonia. In particular, we aim to answer the following questions.
- Do corticosteroids reduce mortality and the incidence of pneumonia complications such as severe sepsis or acute respiratory distress syndrome (ARDS) in severe pneumonia?
- Do corticosteroids shorten symptoms in mild pneumonia?
- Is the occurrence of relative adrenal insufficiency a possible rationale for using corticosteroids in pneumonia?
- Are there any dose-effect relationships between corticosteroids and pneumonia?
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs) assessing the effectiveness of corticosteroids for pneumonia.
Types of participants
Participants of any age or sex with pneumonia. The diagnosis of pneumonia is usually made from taking a medical history, physical examination and a chest X-ray. Occasionally sputum cultures assist diagnosis. Diagnosis can be difficult, especially in HAP and pneumonia of immunocompromised people. Computerised tomography (CT) scanning of the lungs is sometimes used to make the diagnosis (Wipf 1999).
We included people with pneumonia associated with chronic obstructive pulmonary disease (COPD) but excluded cases associated with immunosuppression, HIV, tuberculosis, acute schistosomiasis, fungal or parasitic infections, or chemotherapy and radiotherapy-associated lung changes.
Types of interventions
Trials comparing the following.
- Corticosteroids with antibiotics versus antibiotics alone.
- Corticosteroids with antibiotics versus corticosteroids of a different dose with antibiotics.
Types of outcome measures
- Time to resolution of symptoms or time to clinical stability. Symptoms include fever, cough, positive chest X-ray, elevation of white blood cell count, difficulty in breathing, etc.
- Relapse of pneumonia.
- Proportion of patients requiring either ventilatory or inotropic support, or both.
- Rate of admission to intensive care unit (ICU).
- Time to discharge from ICU.
Serious adverse events have been defined as any event that leads to death, is life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent or significant disability, and any reported serious event that might jeopardise the patient or require an intervention to prevent it (ICHEWG 1997). All other adverse events are not considered to be serious.
Search methods for identification of studies
We searched the Cochrane Central Register of Controlled Clinical Trials (CENTRAL) (The Cochrane Library 2010, Issue 11) which contains the Cochrane Acute Respiratory Infections Group's Specialised Register, MEDLINE (1966 to December, 2010), EMBASE (1974 to December 2010), the China National Knowledge Infrastructure (CNKI) (1978 to December 2010) and VIP (1986 to December 2010).
We used the following search strategy to search MEDLINE and CENTRAL. We combined the search strategy with the Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE: sensitivity- and precision-maximising version (2008 revision); Ovid format (Lefebvre 2009). We adapted the search strategy for EMBASE (see Appendix 1), the China National Knowledge Infrastructure and VIP (Figure 1).
|Figure 1. Search strategy used in the Chinese databases|
1 exp Pneumonia/
5 Respiratory Distress Syndrome, Adult/
6 adult respiratory distress syndrome.tw.
7 acute respiratory distress syndrome.tw.
10 exp Steroids/
12 exp Adrenal Cortex Hormones/
22 9 and 21
Searching other resources
We did not impose any language or publication restrictions. We searched the Chinese Journals Full Text Database (1979 to February 2010), Chinese Journals Full Text Database Century Journals (1979 to February 2010), Chinese Doctoral Degree Thesis Full Text Database (1979 to February 2010), Chinese Outstanding Master Degree Thesis Full Text Database (1979 to February 2010) and WANFANG Database (1993 to February 2010).
We searched the WHO ICTRP Search Portal (http://www.who.int/ictrp/network/en/index.html) for ongoing trials.
Data collection and analysis
Selection of studies
Four review authors (YC, KL, HP, TW) undertook the selection of studies. Two review authors (YC, HP) independently scanned the titles and abstracts of all the articles identified by the literature search. The same review authors independently assessed whether the trials met the inclusion criteria. We resolved disagreements by discussion. We contacted trial authors of the original reports to obtain further information if the search results contained insufficient information to make a decision about eligibility. We excluded those trials which had not developed a research protocol, or in which coin tossing or card shuffling was used in the presence of the allocated participants, because there was high risk of selection bias. We recorded the trial ID, name and contact details of trialists, date of query to trialist, method of query (for example, telephone) and response of trialist for each trial, whether it was included or excluded. We recorded all of the information in an additional table.
Data extraction and management
Three review authors (YC, KL, HP) independently extracted the methodological details and data from publications. Data for extraction included study title; design; study population size; duration; number of drop-outs, withdrawals and loss to follow up and participants analysed in the different treatment groups; inclusion and exclusion criteria; intervention (route and dosage); and outcomes (Higgins 2009). There was no disagreement among the review authors.
Assessment of risk of bias in included studies
We assessed risk of bias using the following criteria described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009) and Wu 2007, and we graded the quality of evidence using GRADEprofiler.
A - adequate sequence generation was reported using one of the following approaches: we considered a random numbers table or computer-generated random numbers; coin tossing; or shuffling used for generating the allocation sequence before the trial as having a low risk of selection bias.
B - does not specify one of the adequate methods outlined in (A) but only mentions 'random' - we considered this to have a moderate risk of selection bias.
C - we excluded other methods of allocation, for example quasi-randomisation, that appeared to have a high risk of bias.
Allocation concealment process
A - adequate measures to conceal allocation and the allocation sequences, such as central randomisation, and the use of sealed opaque envelopes - we considered this a low risk for selection bias.
B - unclear: concealed allocation where the author does not report the allocation concealment method used - we considered this a moderate risk for selection bias.
C - inadequately-concealed allocation, where an approach was reported that does not fall into one of the categories in (A).
D - does not conceal allocation.
C and D can be considered as high risks for selection bias.
Level of blinding
A - double-blinding: participants and results assessor were masked - we considered this a low risk for both performance and detection bias.
B - single-blinding of results assessor - we considered this a moderate risk for both performance and detection bias. If single-blinding was performed for participants but not the results assessor, we considered the study as having a high risk of detection bias.
C - we considered non-blinding as a high risk for both performance and detection bias.
Use of blinding for mortality was not required.
Incomplete outcome data
Low risk of bias - we considered number of drop-outs or loss to follow less than 10% as low risk of bias due to incomplete outcome data.
Moderate risk of bias - we considered this to be a rate of drop-out or loss to follow up between 10% and 15%.
High risk of bias - number of drop-outs or loss to follow higher than 15%.
Selective reporting bias
We expected to assess selective reporting bias by comparing the protocol and published trial reports. However, we were unable to locate the protocol and so were unable to identify whether reporting bias existed or not.
Other potential sources of bias
We expected to address potential sources of bias, for example, different doses of corticosteroids, length of follow up, and characteristics of participants (for example, age, stage of disease). However, we did not find any potential sources of bias in the included studies.
Measures of treatment effect
We expected both dichotomous and continuous data. We analysed different comparisons separately. We used the risk ratio (RR) with 95% confidence intervals (CI) and control events rates for reporting dichotomous data. We expressed continuous data as mean differences (MD) with 95% CI.
Dealing with missing data
We assessed incomplete outcome data for potential bias from exclusions and attrition.
- Low risk of bias: trials where few exclusions and attrition are noted and an intention-to-treat (ITT) analysis is possible.
- Moderate risk of bias: trials which report the rate of exclusion, attrition or both to be about 10%, whatever ITT analysis is used.
- High risk of bias: the rate of exclusion, attrition or both is higher than 15%, or wide differences in exclusions between groups, whatever ITT analysis is used.
Assessment of heterogeneity
We assessed clinical and methodological heterogeneity before pooling. If data were similar enough, we carried out an assessment for statistical heterogeneity using the Chi
Assessment of reporting biases
We did not investigate potential publication bias by funnel plot (Egger 1997) due to only a few studies being included, although we assessed reporting bias according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008).
- No - low risk of reporting bias: all of the outcomes were reported in detail.
- Probably yes - moderate risk of reporting bias: at least one of the outcomes were mentioned but not in detail.
- Yes - high risk of reporting bias: at least one of the outcomes were not reported.
We carried out meta-analyses using the Review Manager 5 software program (RevMan 2008). We used a random-effects model for pooled data analysis. We did not combine results of trials with different comparator drugs.
Subgroup analysis and investigation of heterogeneity
We performed subgroup analyses based on different types of corticosteroids and different comparators (Higgins 2009).
We did not perform sensitivity analyses to test the robustness of the evidence because of the small number of included studies. However, we did test the following:
- excluded studies with inadequate concealment of allocation;
- excluded studies in which the outcome evaluation was not blinded; and we
- compared the difference between pooling analysis results by using a fixed-effect model and a random-effects model because robust evidence should not be transposed by changing the effect model.
Description of studies
We identified 50 controlled trials. Of these, we excluded 35 trials and 10 trials are awaiting for classification due to difficultly in retrieving the article or contacting the original trial authors. These data will be analysed when we update the review. The remaining six trials involving 437 participants met the inclusion criteria and were included in this review.
Results of the search
125 relevant studies appeared in our search.
Participants in the included studies were of any age or sex with pneumonia. Two trials (Cao 2007; van Woensel 2003) mainly focused on children, while the other four (Confalonieri 2005; Marik 1993; McHardy 1972; Mikami 2007) focused on adults or the elderly.
One study was conducted in China (Cao 2007) with 120 children aged from three months to 14 years infected by M. pneumoniae. They were divided into four groups: 30 in the control group, 30 in the budesonide (Pulmicort) group, 30 in the clarityne (Loratadine tablets) group, and 30 in the thymosin or transfer factor (TF) injection group.
The Confalonieri 2005 study involved 48 participants: 24 in the hydrocortisone infusion group and 24 in the placebo group.
There were 30 patients in the Marik 1993 study: 14 in the hydrocortisone group and 16 in the placebo group.
One hundred and twenty six participants were included in McHardy 1972 and were divided to four groups: 43 in group 1, 20 in group 2, 43 in group 3 and 20 in group 4.
The Mikami 2007 study involved 31 patients: 15 received prednisolone intravenously for three days and 16 in the control group did not receive prednisolone.
There were 82 participants in the van Woensel 2003 study: 37 in the steroid group (including 18 participants with bronchiolitis and 17 with pneumonia) and 45 in the control group (including 21 participants with bronchiolitis and 22 with pneumonia).
In all trials, the interventions were antibiotics with corticosteroids versus antibiotics with placebo or antibiotics alone. The corticosteroids included hydrocortisone (Confalonieri 2005; Marik 1993), prednisolone (McHardy 1972; Mikami 2007), budesonide (Cao 2007) and dexamethasone (van Woensel 2003). In Confalonieri 2005, hydrocortisone was given as an intravenous 200 mg loading bolus followed by an infusion (hydrocortisone 240 mg in 500 cc 0.9% saline) at a rate of 10 mg/hour for seven days and protocol-guided antibiotic treatment.
In Marik 1993, the participants of the treatment group received 10 mg/kg of hydrocortisone intravenously 30 minutes prior to starting antibiotic therapy.
In McHardy 1972, group 1 and group 3 received 1 g ampicillin daily and 2 g ampicillin daily, respectively; group 2 received 1 g ampicillin plus 20 mg prednisolone and group 4 received 2 g ampicillin plus 20 mg prednisolone, respectively. In Mikami 2007, the corticosteroids group received 40 mg of prednisolone intravenously for three days plus intravenous antibiotics within eight hours of hospital admission. This was then modified, based on culture results.
In some of the studies, treatment of pneumonia with corticosteroids was just one part of the trial. In the study by Cao 2007 for example, the control group were given azithromycin or erythromycin for seven days, and in the budesonide (Pulmicort) group the participants also had budesonide (Pulmicort) inhalation 250 to 500 g/day for seven days in the control group. The study aimed to compare the effectiveness of budesonide and azithromycin and erythromycin for M. pneumonia in children. We only focused on the results of the effectiveness of budesonide.
The van Woensel 2003 study looked at the effectiveness of corticosteroids for pneumonia and bronchiolitis. The trial medication was intravenous dexamethasone (0.15 mg/kg six-hourly for 48 hours) or placebo and had to have been started within 24 hours of mechanical ventilation.
These trials mainly measured improvements in oxygenation (Confalonieri 2005), time to resolution of symptoms or time to clinical stability (Cao 2007; Confalonieri 2005; Mikami 2007) and the length of hospital stay or ICU stay (Marik 1993; Mikami 2007; van Woensel 2003). One study measured mortality (Marik 1993). The outcomes of each trial are noted in the Characteristics of included studies table.
Most of the excluded studies were retrospective. Some did not use random allocation and were excluded because the trial author refused to give details of the study. The reasons for exclusion can be found in the Characteristics of excluded studies table.
Risk of bias in included studies
Two studies were of higher quality (Confalonieri 2005; van Woensel 2003) and four were of poorer quality (Cao 2007; Marik 1993; McHardy 1972; Mikami 2007). The summary of quality assessment can be found in Figure 2 and Figure 3.
|Figure 2. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
All of the participants were randomly allocated. Randomisation methods for two studies (Confalonieri 2005; van Woensel 2003) involved generation in blocks of 10 for each participating site by a randomisation centre. The Cao 2007 study did not provide any information about the method of randomisation. We telephoned Dr. Cao who told us that the allocation sequence was generated by computer but we still judged the randomisation method as unclear and the allocation procedure was not concealed because the author mentioned that the participants "were selected by sampling from the in-patients". In Marik 1993, participants were randomized by a random number generator to receive either hydrocortisone or placebo. In McHardy 1972, the method of generating and concealing the allocation sequence was not mentioned. In Mikami 2007, the study did not reveal the details of the method of randomisation, it just mentioned that they divided the participants randomly but the allocation concealment in this trial was well done, because the investigators were not actively involved in the treatment of the participants and the physicians who took care of the participants were not informed of the outcome parameters. In fact, most of the included trials performed adequate allocation concealment, either by sealed envelopes (Cao 2007; Confalonieri 2005) or by keeping the concealed randomisation list in a separate location (van Woensel 2003). However, in the trial by Marik 1993, allocation concealment was not clearly stated.
There was no blinding in McHardy 1972 and Mikami 2007 as they were open-label trials. In Cao 2007, blinding was not clearly stated in the article, so we contacted the trial author and discovered that it was a single-blinded trial, because the doctor knew the treatment assignment while the participants did not. The other trials are double-blinded, placebo-controlled studies.
Incomplete outcome data
Incomplete outcome data was clearly stated in the included studies if they had drop-outs or losses to follow up (Confalonieri 2005; van Woensel 2003). There was no risk of attrition bias for the other four studies.
There was no evidence of selective reporting.
Other potential sources of bias
This was not clearly stated in the included studies.
Effects of interventions
We collected data on the basis of corticosteroid treatment plus antibiotic therapy or antibiotics only. The corticosteroid treatment included hydrocortisone, prednisolone, budesonide and dexamethasone. The baseline characteristics were clearly stated in the studies and the groups were comparable with regards to sex and age distribution and severity of the disease, except for Cao 2007 and van Woensel 2003. In Cao 2007, there were 32 more boys than girls. In van Woensel 2003 the proportion of males was higher and patients were significantly younger in the dexamethasone group than in the placebo group.
In three studies participants receiving hydrocortisone or prednisolone displayed no statistically significant differences compared with placebo (Peto odds ratio (OR) 0.26; 95% CI 0.05 to 1.37; and Peto OR1.41, 95% CI 0.39 to 5.04, respectively). One participant died in the hydrocortisone group and five in the control group (Confalonieri 2005; Marik 1993) ( Analysis 1.1), nine died in ampicillin alone groups and three in ampicillin plus prednisolone groups, respectively (McHardy 1972) ( Analysis 2.1).
Improvement in oxygenation
There are several ways to measure oxygenation, including as partial pressure of oxygen in arterial blood (PaO
Time to pneumonia resolution
There were no reports of time to pneumonia resolution.
Time to resolution of symptoms or time to clinical stability
Resolution of symptoms and clinical stability can be calculated as improvement in chest radiograph and normalisation of body temperature, respiratory rate, white blood cell (WBC) and C reactive protein. In Confalonieri 2005, participants receiving prednisolone displayed significantly lower chest radiograph scores (MD -1.50; 95% CI -2.10 to -0.90) and significantly higher improvement in chest radiograph scores from day 1 to day 8 (RR 4.20; 95% CI 1.91 to 9.21) ( Analysis 1.4, Analysis 1.5).
In McHardy 1972 the duration of treatment, change of treatment, resolution of temperature, clearance of pathogens from sputum or laryngeal swabs and maximum radiological clearance were reported. None of the outcomes identified important differences between ampicillin groups and ampicillin plus prednisolone groups ( Analysis 2.2 and Analysis 2.3).
In Mikami 2007, participants receiving prednisolone showed faster normalisation of body temperature (MD -3.60; 95% CI -6.28 to -0.92) ( Analysis 2.7), respiratory rate (MD -3.40; 95% CI -5.52 to -1.28) ( Analysis 2.8) and normalisation of C reactive protein (MD -4.10; 95% CI -7.15 to -1.05) ( Analysis 2.9) compared to the placebo group. There was no significant difference between prednisolone and placebo in terms of WBC (MD 2.00; 95% CI -0.61 to 4.61) ( Analysis 2.10). The Cao 2007 study showed clinical symptoms disappearing faster in the budesonide treatment group than in the control group (MD -8.00; 95% CI -9.25 to -6.75) ( Analysis 3.1).
Relapse of the disease
Proportion of patients requiring either ventilatory or inotropic support, or both
Participants receiving hydrocortisone in two trials (Confalonieri 2005; Marik 1993) displayed less need for mechanical ventilation (RR 0.43; 95% CI 0.22 to 0.85) than participants receiving placebo ( Analysis 1.6). In van Woensel 2003 the dexamethasone group had significantly longer mechanical ventilation (MD 0.80; 95% CI 0.15 to 1.45) and duration of supplemental oxygen (MD 1.70; 95% CI 0.75 to 2.65) compared to the placebo group ( Analysis 4.1 and Analysis 4.2).
Rate of admission to intensive care unit (ICU)
There were no reports of admissions to ICU.
Time to discharge from ICU
In Marik 1993, participants receiving hydrocortisone displayed no difference in length of stay in the ICU (MD -0.30; 95% CI -3.81 to 3.21) compared to the placebo group ( Analysis 1.7). In van Woensel 2003 there was no difference between the dexamethasone group and the placebo group in terms of length of stay in the pediatric ICU (MD 0.20; 95% CI -0.50 to 0.90) ( Analysis 4.3).
Arrhythmia (one case) and upper gastrointestinal bleeding (one case) were reported in Confalonieri 2005. One participant developed malignant hypertension three days after medication (van Woensel 2003). No other severe adverse events were reported.
Summary of main results
In statistical terms, corticosteroids did not significantly reduce mortality in participants with pneumonia. However, there was obvious clinical value: in two small studies (Confalonieri 2005; Marik 1993) one participant the corticosteroid group died compared to five deaths in the placebo group. Corticosteroids can accelerate the resolution of symptoms and clinical stability and reduce relapse rates of the disease. Improvement in oxygenation, need for ventilatory or inotropic support and time to discharge from the intensive care unit (ICU) differed between the studies, therefore subgroup analyses were necessary. There were no data, or insufficient data, to examine the time to pneumonia resolution and rate of admission to ICU. Typical adverse events associated with corticosteroid therapy were infrequent.
The use of corticosteroids showed significantly better oxygenation in terms of PaO
In Confalonieri 2005 and Marik 1993, the application of hydrocortisone also reduced the need for mechanical ventilation and length of stay in the ICU, while in van Woensel 2003 the need for mechanical ventilation and supplemental oxygen increased, as did the length of stay in ICU after using dexamethasone. Although the three studies focused on participants with severe pneumonia, the difference was that participants in the van Woensel 2003 study were respiratory syncytial virus (RSV) infected children, including both the bronchiolitis and pneumonia subgroups. RSV is a common cause of lower respiratory tract infection in infants and children. The results of the van Woensel 2003 trial showed that corticosteroids benefited the bronchiolitis subgroup only. In other participants with severe pneumonia, the reduced need for mechanical ventilation was related to oxygenation improvement and inflammation relief from corticosteroids (Confalonieri 2005; Marik 1993).
In Cao 2007, the main focus was M. pneumoniae in children; it detected the duration of clinical symptoms and the rate of relapse. The results of this study showed that the a corticosteroid (budesonide) can decrease the duration of clinical symptoms and the rate of relapse. Mycoplasma pneumoniae (M. pneumoniae) infection is a common cause of pneumonia, especially in children, accounting for 10% to 40% of all pneumonia in children and it is characterized by a long clinical course and repeated infection (Esposito 2001). M. pneumoniae is a superantigen which can activate the macrophages and induce secretion of cytokine, so the inflammation response is intense. Corticosteroids can act as an anti-inflammatory agent for children with M. pneumoniae.
We would have liked to have answered three important questions:
- whether the incidence of complications of pneumonia (such as severe sepsis or acute respiratory distress syndrome (ARDS)) could be reduced by corticosteroids;
- whether the occurrence of relative adrenal insufficiency is a possible rationale for using corticosteroids in pneumonia; and
- whether there are any dose-effect relationships between corticosteroids and pneumonia?
Overall completeness and applicability of evidence
On the whole, corticosteroids can improve clinical symptoms and reduce the rate of relapse of mild pneumonia. In severe pneumonia corticosteroids can improve oxygenation and clinical symptoms and reduce mechanical ventilation and length of stay in ICU. Further trials would help clarify the validity of the findings of this review and could determine more clearly the role of corticosteroids in patients with pneumonia in comparison with other therapies.
Quality of the evidence
In general, the quality of the evidence is still weak (Figure 4; Figure 5; Figure 6; Figure 7; Figure 8; Figure 9) due to the fact that the results were taken from small, single studies. We collected the information by telephoning trial authors rather than relying on published articles. In addition, the blinding procedures were not adequate (Cao 2007). The Mikami 2007 study was an open-label study and the detailed method of randomisation was not clearly stated. Allocation concealment was not clear in Marik 1993 so there is also a risk of selection bias. The two other higher quality studies included only small numbers of participants (Confalonieri 2005; van Woensel 2003).
Potential biases in the review process
Some of the trials that we encountered claimed to be randomized controlled trials. We contacted the trial authors and found out that many of the studies were either retrospective or incorrect in their method of randomisation. We allocated studies where we could not contact the trial author or we could not find the full text to Studies awaiting classification.
Agreements and disagreements with other studies or reviews
We were unable to find other reviews or meta-analyses of corticosteroids for pneumonia, but a review of corticosteroid therapy for sepsis and related syndromes (Annane 2009) found that prolonged low-dose corticosteroid therapy could have a beneficial effect on short-term mortality.
Implications for practice
In patients with pneumonia, corticosteroids may relieve symptoms but the evidence is weak. In severe pneumonia, corticosteroids can also be used to improve oxygenation and reduce the use of mechanical ventilation. However, there is insufficient evidence to confirm whether they can reduce mortality and resolve pneumonia. We do not recommend the use of steroids for respiratory syncytial virus-infected children with pneumonia because there is no significant benefit for the patient. However, we do recommend corticosteroids for M. pneumoniae infected children because corticosteroids can significantly relieve clinical symptoms and prevent relapse of the disease. Our included studies did not compare the effects of different doses of corticosteroids and so we are unable to make any recommendations with regards to the dosage of corticosteroids in clinical practice. Due to the limitations of the included studies and the review process, the quality of evidence is low.
Implications for research
More large clinical trials of corticosteroids for pneumonia are needed to enhance the body of evidence. These trials need to be designed for patients with severe pneumonia and patients with mild pneumonia, and children and adults, covering the different causes of pneumonia. We encourage trialists to measure changes in pneumonia resolution, symptom relief, oxygenation, length of stay in ICU and rate of admission to ICU. In the meantime, more data concerning adverse events need to be reported to guide our application of corticosteroids. Future studies should explore whether the occurrence of relative adrenal insufficiency is a possible rationale for using corticosteroids in pneumonia or not, and whether prolonged low-dose corticosteroid therapy can benefit mortality or not.
The review authors wish to thank Elizabeth Dooley (Managing Editor), Clare Dooley (Assistant Managing Editor) and Sarah Thorning (Trials Search Co-ordinator) of the Cochrane ARI Group. We would also like to thank the following people for commenting on the draft protocol: Anne Lyddiatt, Chanpen Choprapawon, Marco Confalonieri, Sree Nair and Allen Cheng; and Amanda Young, Marco Confalonieri, Robert Ware and Allen Cheng for commenting on the draft review.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. Embase.com search strategy
#16. #12 AND #15
#15. #13 OR #14
#14. random*:ab,ti OR placebo*:ab,ti OR factorial*:ab,ti OR crossover*:ab,ti OR 'cross-over':ab,ti OR 'cross over':ab,ti OR assign*:ab,ti OR allocat*:ab,ti OR volunteer*:ab,ti OR ((singl* OR doubl*) NEAR/2 (blind* OR mask*)):ab,ti
#13. 'randomized controlled trial'/exp OR 'single blind procedure'/exp OR 'double blind procedure'/exp OR 'crossover procedure'/exp
#12. #6 AND #11
#11. #7 OR #8 OR #9 OR #10
#10. prednisone*:ab,ti OR prednisolone*:ab,ti OR methylprednisolone*:ab,ti OR betamethasone*:ab,ti OR dexamethasone*:ab,ti OR triamcinolone:ab,ti OR hydrocortisone*:ab,ti
#9. steroid*:ab,ti OR corticosteroid*:ab,ti
#6. #1 OR #2 OR #3 OR #4 OR #5
#5. 'adult respiratory distress syndrome':ab,ti OR 'acute respiratory distress syndrome':ab,ti OR ards:ab,ti
#4. 'adult respiratory distress syndrome'/de
#3. cap:ab,ti OR hap:ab,ti
Last assessed as up-to-date: 31 December 2010.
Protocol first published: Issue 2, 2009
Review first published: Issue 3, 2011
Contributions of authors
All four review authors, Yuanjing Chen (YC), Ka Li (KL), Hongshan Pu (HP) and Taixiang Wu (TW) were responsible for developing and writing the protocol and review.
Declarations of interest
Sources of support
- West China Hospital, Sichuan University, China.
- Cochrane Acute Respiratory Infections Group, Australia.
Differences between protocol and review
We were unable to conduct sensitivity analyses and funnel plots for analysing publication bias due to the small number of included studies.
Medical Subject Headings (MeSH)
MeSH check words