Chlamydia pneumoniae has been implicated as a cause of community-acquired pneumonia (CAP) in several studies. However, there has been no comprehensive study of the role of Chlamydia species (C. pneumoniae, C. psittaci (avian and feline strains) and C. pecorum) as a cause of CAP. The aim of the present study was to determine the role of C. pneumoniae, C. psittaci and C. pecorum as causes of CAP.
A prospective cohort observational study of CAP was conducted at 15 teaching centres in eight Canadian provinces between January 1996–October 1997. Acute (n=539) and convalescent (n=272) serum samples were obtained for determination of antibody titres to C. pneumoniae, C. psittaci, C. pecorum, C. trachomatis, Mycoplasma pneumoniae, Legionella pneumophila serogroups I–VI, Streptococcus pneumoniae and various respiratory viruses.
Twelve of 539 (2.2%) patients had acute C. pneumoniae pneumonia and an additional 32 (5.9%) had possible acute infection. C. pneumoniae was the sole pathogen in 16 of 42 (38.1%) of these patients. The most common copathogens were S. pneumoniae, respiratory syncytial virus and influenza virus type A. C. pneumoniae pneumonia patients were older and more likely to show congestive heart failure compared to bacteraemic S. pneumoniae patients. The latter had a lower mean diastolic blood pressure, a higher white blood cell count and a lower arterial carbon dioxide tension. Two patients had antibody titres suggestive of recent infection with the feline strain of C. psittaci.
Although numerically Chlamydia pneumoniae is an important cause of community-acquired pneumonia, no distinctive clinical features associated with this pathogen were detected in the present study. Feline Chlamydia psittaci may cause a few cases of community-acquired pneumonia. Avian Chlamydia psittaci should be considered only if there is a compatible epidemiological history.
This study was supported by a research grant from Pfizer Canada, Inc.
There are four species within the genus Chlamydia, C. pneumoniae, C. psittaci, C. pecorum and C. trachomatis 1. C. psittaci has been recognised as a cause of community-acquired pneumonia (CAP) following exposure to C. psittaci-infected parrots since 1879 2. In 1986, Grayston and co-workers 3, 4 described a new species of Chlamydia, C. pneumoniae, as a cause of respiratory tract infections. Subsequent studies have shown that this microorganism is frequently implicated as a cause of CAP 5. It is unclear whether C. pecorum, which was isolated from ruminants and named in 1992, is a human pathogen 6.
The purpose of the present study was to define the role of C. pneumoniae, C. psittaci (avian and feline strains) and C. pecorum as causes of CAP.
Patient population and study design
The present prospective study of CAP began on January 11, 1996 and was completed on October 31, 1997. It was conducted at 15 teaching hospitals in eight Canadian provinces. The study was approved by the local research ethics boards. All patients who were admitted to hospital with a diagnosis of pneumonia were screened by a study nurse.
Patients who met the following eligibility criteria were approached for enrolment in the study: 1) age ≥16 yrs; 2) evidence of a new pulmonary infiltrate on chest radiography not attributable to an aetiology other than that of pneumonia (e.g. congestive heart failure); 3) findings suggestive of bacterial pneumonia (chest pain, cough, rales, rhonchi and/or signs of consolidation) or an oral temperature of <36.5°C or >38.5°C; and 4) absence of hospitalisation for ≥14 days. Sputum production was not a requirement for study entry; however, every attempt was made to collect a sputum sample for Gram stain and culture.
Patients were excluded if they had nosocomial pneumonia.
A study nurse completed the data collection forms and made daily visits to each patient (for a maximum of 7 days follow-up while in hospital) to record progress in terms of the progression of the pneumonia and any complications that may have occurred during the hospital stay. All data were verified against the original chart by a study monitor during a site visit.
Acute and 4–6-week convalescent serum samples were collected as part of the study protocol. In addition, a urine sample from each patient was tested for Legionella pneumophila serogroup-I antigen.
Blood, sputum and/or respiratory samples were processed for culture according to the methods in place at each centre's laboratory. All serum samples were tested for antibodies directed against Mycoplasma pneumoniae, influenza virus types A and B, parainfluenza virus types 1, 2 and 3, adenovirus, and respiratory syncytial virus using a standard complement fixation technique in microtitre plates. The adenovirus and respiratory syncytial virus antigens were purchased from Flow Laboratories (McLean, VA, USA); influenza virus types A and B, parainfluenza virus types 1, 2 and 3, and M. pneumoniae antigens were purchased from M.A. Bioproducts (Walkersville, MD, USA).
Acute phase serum samples from 539 patients and convalescent phase samples from 272 patients were tested for immunoglobulin (Ig)G and IgM directed against C. pneumoniae strain AR39; C. psittaci strains 6BC (avian), FP (feline pneumonitis), TT3 (turkey) and CP3 (pigeon); the C. pecorum ovine polyarthritis strain; and pooled antigens of C. trachomatis serovars BED, CJHI and FGK using a microimmunofluoresence assay 7.
Serum specimens were screened at 1:16 dilution for IgG and IgM directed against C. pneumoniae and titrated to end-point (some samples were tested for IgA); a four-fold rise in antibody titre between acute and convalescent serum samples or an IgM antibody titre of ≥1:16 was classified as indicating acute C. pneumoniae infection 5. An IgG titre of ≥1:512 was indicative of possible acute infection 5. Seropositive patients were defined as those with IgG titres of 1:16–1:256 and seronegative patients as those whose titres were <1:16. Seropositivity to other Chlamydia species was defined as a four-fold rise in antibody titre between acute and convalescent serum samples or a stable IgG titre of ≥1:64 8.
Antibiotic therapy was categorised into three groups: monotherapy, concurrent combination therapy and sequential combination therapy. Monotherapy was defined as administration of one antibiotic class throughout the course of therapy regardless of method of administration or overlap in time of use. Concurrent combination therapy was defined as administration of two or more classes of antibiotic overlapping in time by ≥1 day. Sequential combination therapy constituted two or more drug classes administered sequentially with no overlap in time.
Differences between patient subpopulations were tested using the Pearson Chi-squared test for categorical variables and an unpaired t‐test or one-way analysis of variance for continuous variables. Predictors of seropositivity to C. pneumoniae were tested in a logistic regression model using C. pneumoniae antibody status (negative or positive) as the dependent variable. Clinically relevant patient characteristics and factors found to be significantly associated with C. pneumoniae seropositivity on bivariate analysis were entered into the logistic model 9.
Antibodies directed against C. pneumoniae were present at a titre of ≥1:16 in 404 of 539 (75%) acute phase serum samples tested. Table 1⇓ shows the seropositivity of the test population according to age in decades.
Significantly more males than females were seropositive (54.7 versus 45.3%, p=0.039) (table 2⇓). A greater proportion of seropositive than seronegative patients were current smokers (32.1 versus 25.6%, p=0.005) and had a history of chronic obstructive pulmonary disease (COPD) (35.1 versus 22.2%, p=0.02). The mean body mass index (BMI) of the seropositive subjects was significantly higher than that of the seronegative subjects (26.0±6.4 versus 23.4±4.9, p=0.000). A significantly higher systolic and diastolic blood pressure was observed among seropositive subjects (p=0.024 and p=0.045, respectively).
The results of the logistic regression analysis for predictors of C. pneumoniae seropositivity are also presented in table 2⇓. Significant risk factors included greater BMI, nonwhite race and current smoking status. No other significant associations were found between patient characteristics and risk of seropositivity.
Chlamydia pneumoniae pneumonia
Forty-two (7.8%) of the patients studied met the criteria for acute or possible acute C. pneumoniae infection. Eight of the 42 (19%) had an IgM titre of ≥1:16, four (9.5%) demonstrated a four-fold rise in antibody titre and 31 (73.8%) had a stable IgG titre of ≥1:512. No differences were found when the 12 patients with acute infection were compared with the 32 patients with possible acute infection. Multivariate analysis revealed no features predictive of acute infection. Twenty-two of the 42 (52%) patients with acute or possible acute C. pneumoniae infection who were tested for IgA had a titre of ≥1:64. Sixteen of the 42 (38%) patients with acute or possible acute C. pneumoniae pneumonia had no other pathogen identified as a cause of the pneumonia. Sixteen of the remaining 26 (61.5%) had one other pathogen identified and 10 (38.5%) had two or more pathogens implicated in the aetiology of the pneumonia. Table 3⇓ details these copathogens. It is noteworthy that S. pneumoniae, influenza A and respiratory syncytial virus accounted for 33.0, 17.9 and 12.8% of the copathogens respectively.
In order to more clearly define the clinical manifestation of C. pneumoniae pneumonia, the 16 patients with C. pneumoniae as the sole pathogen were compared with the 26 patients who had C. pneumoniae plus copathogens (table 4⇓).
Although this comparison is limited by the small sample sizes, noteworthy illness-associated manifestations (between-group differences) emerged. The mean duration of symptoms (in days) prior to hospital admission among patients with C. pneumoniae as the sole pathogen was shorter at 4.6 versus 12.9 days. The overall mean duration of hospital stay was longer in the C. pneumoniae alone group than in the C. pneumoniae plus copathogens group (9.00±7.30 versus 7.58±3.49 days).
The unusually long mean hospital stay was the result of four outliers with hospital stays of 39, 43, 93 and 172 days. All four outliers had a history of chronic illness and underlying respiratory disease. Only one of the four subjects had complications during the first 7 days of hospitalisation. However, three of the four outliers had their discharge delayed due to other medical conditions.
The C. pneumoniae alone group was more likely to exhibit asthma, nausea and vomiting and less likely to show COPD, purulent sputum and shaking chills. A higher respiratory frequency, pulse rate and systolic and diastolic blood pressure was more common among C. pneumoniae plus copathogen subjects.
In table 5⇓, all patients with C. pneumoniae pneumonia are compared with those with bacteraemic pneumococcal pneumonia. The C. pneumoniae group was significantly older, had a lower pulse rate, a higher diastolic blood pressure and a higher arterial carbon dioxide tension. The white blood cell count was significantly higher in the bacteraemic S. pneumoniae group. The mortality rates were not significantly different at 4.9% for C. pneumoniae and 5.4% for the bacteraemic pneumococcal pneumonia group. The mortality rate of 9.4% for the remaining cohort (excluding the C. pneumoniae and bacteraemic S. pneumoniae patients) was also not significantly different from the mortality rates for the C. pneumoniae and the bacteraemic pneumococcal pneumonia patients.
Chlamydia psittaci cases
There were no cases of pneumonia due to C. pecorum or avian strains of C. psittaci. Two patients had IgG titres suggestive of pneumonia due to feline C. psittaci.
Patient 1 was a 77-yr-old female with shaking chills, nausea, vomiting, chest pain and a cough productive of yellow sputum. The oral temperature was 39°C, and crackles and wheezes were present on bilateral auscultation. The chest radiograph revealed bilateral interstitial basilar opacities. The patient was treated intravenously with cefuroxime (750 mg t.i.d.) and erythromycin (500 mg q.i.d.) for 4 days followed by cefuroxime axetil (500 mg b.i.d.) for 11 days and made an uneventful recovery. Acute and convalescent serum samples had IgG titres of 1:256 directed against C. psittaci strain FP and titres of 1:64 directed against strain CP3. Unfortunately data on pet exposure were not collected as part of the present study.
Patient 2 was a 60-yr-old female who had received cefuroxime axetil (500 mg b.i.d.) at home for 4 days prior to being admitted with cough, pleuritic chest pain, and shortness of breath. The oral temperature was 38.2°C, crackles and wheezes were present on auscultation over the right lung and chest radiography showed patchy opacities involving the right lung. Cefuroxime (750 mg t.i.d.) and erythromycin (500 mg q.i.d.) were administered intravenously for 2 days, followed by clarithromycin (500 mg b.i.d.) orally for 13 days, which led to a complete recovery. An acute phase serum sample had an IgG titre of 1:1024 directed against strain FP and 1:256 against CP3 and C. pneumoniae. The IgA titre directed against strain FP was 1:256 and against CP3 1:128.
The present study highlights several features of C. pneumoniae seroepidemiology and pneumonia due to this microorganism. Wang and Grayston 11, using a titre of 1:32 as significant or positive, found that antibody directed against C. pneumoniae was rare in young children but increased considerably in the 15–19-yr-old age group in Seattle and amongst 10–14-yr-olds in Denmark. The rates continued to rise until age 30–39 yrs and remained high into old age. Because of the difference in criteria for a positive test, these data are not directly comparable with the present data; however, the same general trend in seropositivity is shown in the present study.
O'Neill et al. 12 tested 400 serum samples from 400 randomly selected adults in Northern Ireland using the microimmunofluoresence assay at a single dilution of 1:64. They found that 70% of patients had IgG directed against C. pneumoniae. Using a cut-off of ≥1:16 it was found that 75% of the Canadian adults in the present study had IgG directed against C. pneumoniae, whereas 57.3% were positive if a cut-off of ≥1:64 was used, and 67.8% if a cut-off of ≥1:32 was used.
Saikku 13 tested 16,328 patients with respiratory symptoms at the University of Helsinki between 1986–1991. Using a cut-off of 1:32 as positive, it was found that seropositivity continues to increase with age up to age 90 yrs. The rate among males was higher than that among females after age 40 yrs. In this study, by the age of 90 yrs, 90% of the males and 75% of the females were seropositive. Others have also reported higher seropositivity rates in males 14–16.
In one study, seropositivity was unrelated to sex, age or smoking, but there was an inverse trend between infection and socioeconomic status, as measured by educational level achieved 12. In contrast, in the present study of adults with CAP, current tobacco smoking, nonwhite race and higher BMI were significantly associated with C. pneumoniae seropositivity. Non-White subjects had a risk of 2.91 (95% confidence interval (CI) 1.04–8.12) relative to White subjects. Furthermore, subjects with a higher BMI had a risk of 1.09 (95% CI 1.04–1.15). O'Neill et al. 12 also found that seropositive subjects weighed more than seronegative ones. Increased body mass is a known risk factor for coronary artery disease and recent seroepidemiological studies have associated coronary artery disease and atherosclerosis with antibody to C. pneumoniae 17.
The present study demonstrates that current smokers have a risk of 2.24 (95% CI 1.18–4.24) relative to patients who have never smoked. COPD was associated with C. pneumoniae seropositivity in the present study; however, it was not considered as a confounding factor. Two studies have found higher seropositivity rates in smokers 15, 18, whereas Mendall et al. 19 did not.
In an outbreak of C. pneumoniae pneumonia in a nursing home, there was no difference in incidence between smokers and nonsmokers but the smokers showed an earlier onset of illness 20. Among smokers, the median day of onset of illness after the index case resident was 16 days compared with 22 days for nonsmokers. In a population-based study, Almirall et al. 21 found that smoking any type of tobacco had an odds ratio for CAP of 1.00 for never smokers and 1.88 for current smokers. Since the present authors studied a population of patients with pneumonia, the effect of current smoking on the rate of C. pneumoniae seropositivity may be confounded by the increased risk of pneumonia in this population.
Twelve of 539 (2.2%) patients had acute C. pneumoniae pneumonia and an additional 32 (5.9%) had possible acute infection. No differences between the two groups were found and multivariate analysis did not suggest features predictive of C. pneumoniae infection.
Thirty-eight per cent (16 of 42) of the present patients with acute or possible acute C. pneumoniae pneumonia had this microorganism as the sole infecting pathogen. This is similar to the 30.6% reported by Lieberman et al. 22. The copathogens in the study of Lieberman et al. 22 and in the present study were also similar. In both studies, S. pneumoniae was the most common copathogen. In the present study, influenza virus type A and respiratory syncytial virus were the second and third most common copathogens, whereas, in the study of Lieberman et al. 22, they were M. pneumoniae and Legionella sp. McConnell et al. 23, in a study of 62 patients with C. pneumoniae pneumonia, divided the patients into those with primary infection (n=17) and those with recurrent infection (n=38). Seven (41%) with primary infection had a second respiratory pathogen present and seven (18%) in the recurrent infection group also had more than one respiratory pathogen present. S. pneumoniae and Haemophilus influenzae were most commonly implicated. Almirall et al. 24 studied 105 patients with CAP and found that 16 had C. pneumoniae pneumonia. Seven of the 16 had a copathogen demonstrated, adenovirus in three, parainfluenza virus in two, L. pneumophila in one and respiratory syncytial virus in one. File et al. 25 found that 21 of 47 (44.6%) patients with C. pneumoniae pneumonia had a copathogen. The consistency of these reports leaves little doubt that, in most instances, C. pneumoniae infection is accompanied by a copathogen.
The observation that C. pneumoniae induces ciliostasis in human ciliated bronchial epithelial cells suggests that this pathogen impairs normal clearance mechanisms in the respiratory tract and hence makes the host more susceptible to infection by other, more virulent, pathogens 26.
In a recent meta-analysis of C. pneumoniae pneumonia, mortality was 9% and associated with underlying factors or secondary infections 27. C. pneumoniae mortality in the present study was low at 4.9% but this was not significantly different from the 5.4% in the bacteraemic pneumococcal pneumonia group and the 9.4% for the remainder of the cohort.
In a comparison of C. pneumoniae with M. pneumoniae and viral infections as a cause of acute respiratory disease in a university student health clinic population, Thom et al. 28 found that the mean number of days from onset of symptoms until enrolment was longer in patients with C. pneumoniae infections than in the other two groups (12.8 versus 7.9 and 7.3 days respectively).
The present authors found that patients with C. pneumoniae pneumonia were older, had a lower white blood cell count and were more likely to have congestive heart failure than patients with S. pneumoniae pneumonia. Furthermore, the C. pneumoniae pneumonia patients exhibited a higher diastolic blood pressure and arterial carbon dioxide tension than the comparison group. There were no differences in the radiographic patterns of the pneumonia or outcomes such as admission to intensive care and mortality. However, the bacteraemic S. pneumoniae pneumonia subjects had a shorter duration from onset of symptoms to admission and a significantly higher pulse rate and white blood cell count on admission.
File et al. 25 compared seven patients with C. pneumoniae primary infection (IgM-positive) with 18 reinfected patients. The mean age of those with primary infection was significantly less and mean temperature was higher for those with primary infection than for those with reinfection.
No cases of avian psittacosis were found and only two patients with possible feline C. psittaci pneumonia. The present data suggest that a diagnostic work-up for these pathogens should be limited to patients with epidemiological data that suggest such an infection.
The strengths of the present study are the large sample size, inclusion of patients with various underlying factors, comprehensiveness of the serological diagnostic work-up, prevalence of key CAP pathogens and identification of clinical manifestations, observation of the clinical course of CAP infection while hospitalised and the multiple-centre Canadian sites.
Some limitations exist to this study. First, the requirements for informed consent and follow-up of serological studies curtailed enrolment. Secondly, the study was performed exclusively in Canada, which may limit the general applicability of the findings. Another limitation of the study was reliance on serological techniques alone for the diagnosis of C. pneumoniae infection.
It is concluded that Chlamydia pneumoniae is an important pathogen in community-acquired pneumonia. Coinfection with other respiratory pathogens is common in patients with Chlamydia pneumoniae pneumonia. Avian and feline Chlamydia psittaci are uncommon causes of community-acquired pneumonia and should only be considered in the appropriate epidemiological setting.
The authors would like to thank G. McClarty (University of Manitoba), J. Papp (University of Guelph) and A.A. Andersen (United States Dept of Agriculture, National Animal Centre) for kindly providing the strains of Chlamydia psittaci. The authors would also like to thank J.A. Inverso (Pfizer, Inc.) and L. Dillon (Laboratory Centre for Disease Control) for their assistance in this study. The authors are grateful to S. Ioannou and N.G. Moghaddam (Pfizer Canada, Inc.) and the following clinical study coordinators: M. Dambreville (Centre Universitaire de Sante de l'Estrie, University of Sherbrooke, Sherbrooke, Quebec, Canada), H. Stalts (Foothills Medical Centre, University of Calgary, Calgary, Canada), J. Clark-DiPrata (Toronto General Hospital, University of Toronto, Toronto, Canada), B. Peters (St Royal University Hospital, University of Saskatchewan, Saskatoon, Canada), K. Heney and J. Graham (Ottawa Civic Hospital, University of Ottawa, Ottawa, Canada), F. Brisebois (Hôpital du St Sacrement, University of Laval, Quebec City, Canada), H. Patil and G. Patrick (Queen Elizabeth II Hospital, Dalhousie University, Halifax, Canada), T. Muir (Peter Lougheed Medical Centre, University of Calgary, Calgary, Canada), F. Habel (Hôpital Maisonneuve-Rosemont, University of Montreal, Montreal, Canada), E. Condon (The General Hospital, Memorial University, St John's, Canada), S. Roberts and A. Lindemulder (University Hospital, University of Alberta, Edmonton, Canada), D. Paget-Dellio (Mount-Sinai Hospital, Toronto, Canada), M. Jones (Vancouver Hospital, University of British Columbia, Vancouver, Canada), C. Hammerberg (St Joseph's Health Centre, University of Western Ontario, London, Canada) and H. Duckworth (Health Science Centre, University of Manitoba, Winnipeg, Canada).
Investigators. M. Gribble, University of British Columbia, Vancouver; S. Field and T. Louie, University of Calgary, Calgary; S. Houston, University of Alberta, Edmonton; K. Williams, University of Saskatchewan, Saskatoon; L. Nicolle, University of Manitoba, Winnipeg; R. Grossman and I. Salit, University of Toronto, Toronto; R. Saginur, University of Ottawa, Ottawa; D. Gregson, University of Western Ontario, London; R. Duperval, University of Sherbrooke, Sherbrooke; M. Laverdière; University of Montreal, Montreal; M. Rouleau, Université Laval, Sainte-Foy; and J. Hutchinson, Memorial University of Newfoundland, St John's, Canada.
↵ For editorial comments see page 741.
- Received October 18, 2002.
- Accepted December 20, 2002.
- © ERS Journals Ltd