Acquisition of Pseudomonas aeruginosa in Children With Cystic Fibrosis
Objective. This study was pursued as an extension of a randomized clinical investigation of neonatal screening for cystic fibrosis (CF). The project included assessment of respiratory secretion cultures for pathogens associated with CF. The objective was to determine whether patients diagnosed through neonatal screening and treated in early infancy were more likely to become colonized with Pseudomonas aeruginosa compared with those identified by standard diagnostic methods.
Methodology. The design involved prospective cultures of respiratory secretions obtained generally by oropharyngeal swabs at least every 6 months and more often if clinically indicated. Patients were managed with a standardized evaluation and treatment protocol at the two Wisconsin certified CF centers; however, there were community and environmental variations associated with the follow-up period as described below.
Results. Overall, there were no differences in acquisition of respiratory pathogens between the screened and the control (standard diagnosis) groups. Evaluation of the data between and within the two centers, however, revealed significant differences with earlier acquisition of P aeruginosa in the center with the following distinguishing characteristics: urban location; following patients with the standard US approach in which newly diagnosed, young children were interspersed with older CF patients; and where there were more opportunities for social interactions with other CF patients. The differences were confined to the screened group followed in the urban center in which the median pseudomonas-free survival period was 52 weeks contrasted with 289 weeks in the other center. In addition, assessment of data for the entire CF populations followed at the two centers revealed that the urban center showed a significantly higher prevalence of P aeruginosa colonization in patients between the ages of 3 and 9 years.
Conclusions. These results present questions and generate hypotheses on risk factors for acquisition of P aeruginosain CF and suggest that clinic exposures and/or social interactions may predispose such patients to pseudomonas infections.
Despite many years of research, the epidemiology of respiratory pathogen acquisition in children with cystic fibrosis (CF) has not been adequately delineated, nor have the risk factors for colonization/infection been identified conclusively. This gap in knowledge is especially disconcerting with regard to Pseudomonas aeruginosa, an organism that is very difficult to manage and seemingly impossible to eradicate from the sputum of CF patients. Before 1946, the prevalence of P aeruginosa was low in CF patients.1 During the 1960s, observations from a variety of sources2 indicated that P aeruginosa became the pathogen most frequently isolated from the respiratory tract of such patients. Although the reasons for the emergence of P aeruginosa infections in CF patients are unclear, the appearance of this organism since the 1960s corresponds to that of center-based care, ie, the establishment of regional clinics and hospital facilities specialized in CF diagnosis and treatment. Use of regional CF clinics has been the standard approach for >25 years in the United States,4 but studies in Denmark have raised concerns about increased risks of acquiring P aeruginosa in such settings.5
Unfortunately, except for the Danish studies,5 most previous investigations of pseudomonas epidemiology in CF have been hampered by incomplete data and/or relatively short periods of follow-up. We were presented with a special opportunity to address some of the existing gaps as part of a longitudinal investigation of CF patients identified in a randomized clinical trial designed to assess the benefits and risks of early diagnosis through neonatal screening. This study has been underway since 1985 and is being conducted as a collaborative project involving the two CF centers in Wisconsin. One of the risks we have investigated, predisposition to early acquisition ofP aeruginosa, led us to culture respiratory secretions at routine intervals. When we analyzed the data, it became apparent that the screened and control groups overall were similar in this regard but that a striking center difference occurred, with one center showing a significantly younger age at which patients become pseudomonas-positive. This report describes data on the acquisition of respiratory pathogens in CF and presents results that are significant because of current concerns about risk factors for P aeruginosa colonization and potential value of segregated CF clinics as used in Denmark.8,9
Experimental Design and Methods
This study is being conducted as a joint effort of the two certified Cystic Fibrosis centers in Wisconsin. The sites are the University of Wisconsin Hospital and Clinics, Madison (referred to as center A), and the Children's Hospital of Wisconsin, located in Milwaukee (center B). Protocols have been approved by the Human Subjects Committee at the University of Wisconsin and the Research and Publications Committee/Human Rights Board at Children's Hospital of Wisconsin. The experimental design and purpose of the Wisconsin CF Neonatal Screening Project have been described in detail elsewhere.10,11 In summary, a randomized clinical trial is being conducted to assess the potential benefits and risks of newborn screening for CF using measurement of either immunoreactive trypsinogen (IRT) from April 4, 1985 to June 30, 1991 or the combination of IRT assays and DNA analyses for theF508 mutation from July 1, 1991 to June 30, 1994.12,13 Health care providers at the two centers developed a screening plan and a standardized evaluation and treatment protocol in 1984 and have met regularly during the study to monitor implementation of protocols and ongoing results. Two cohorts of CF patients—an early diagnosis, or screened, group and a standard diagnosis, or control, group—have been generated randomly and followed concurrently on the same evaluation and treatment protocol. Randomization was achieved when Guthrie cards containing neonatal dried blood specimens were returned to the centralized state laboratory by predetermining that samples for which the terminal digit was odd would be assigned to the screened group. A variety of assessments during the trial demonstrated that this simple method produced satisfactory randomization. The standard diagnosis (control) group also had IRT or IRT/DNA analyses performed, but the data were computer-stored and the investigators kept blinded as to the results until children reached 4 years of age, unless parents requested the information before then (4 years was selected for unblinding, because it represented the approximate average age of CF diagnosis when this project was designed). The investigation was designed primarily to assess the value of early diagnosis through neonatal screening with longitudinal evaluation of patients followed at Wisconsin's two CF centers; thus, comparisons between the screened and control groups are of greatest interest. In addition to treatment effects within centers, we planned from the outset to compare other outcomes such as potential center effects within treatment groups and a variety of subgroups that might selectively show either benefits or risks. After randomization and acquisition of longitudinal data, unbiased analysis of results within centers of subgroups can be accomplished with epidemiologic and statistical validity.
The assessment of potential benefits was designed in 1984 to focus on nutritional and pulmonary outcome variables, which are still under investigation. The risk assessment component has included three areas of concern: 1) missed diagnoses attributable to laboratory or human errors; 2) adverse psychosocial impact;14 and 3) iatrogenic medical risks. The principal medical risk included in the design was early acquisition of P aeruginosa in the respiratory tract of screened patients. This issue was targeted because of a prevailing hypothesis in the 1970s and early 1980s that continuous oral antibiotic therapy might predispose CF patients to pseudomonas colonization/infection;15 more recent observations from a controlled trial of antistaphylococcal oral antibiotic treatment support this hypothesis.16 Consequently, the study protocol included respiratory secretion cultures every 6 months in the longitudinal evaluation protocol and focused on P aeruginosa, Burkholderia cepacia, andStaphylococcus aureus. Additional cultures (not specified in the protocol) were obtained from patients at the discretion of the treating physician. Respiratory secretion cultures were obtained on nonexpectorating children by performing oropharyngeal swabs using the culturette collection and transport system (Becton Dickinson, Cockeysville, MD). During the experimental design phase of this project, it was recognized that young children with CF would not expectorate sputum and that repetitive bronchoscopies were unacceptable to the institutional review boards. Therefore, an aggressive method of obtaining respiratory secretions by oropharyngeal swabbing was implemented in the two centers. Specifically, research nurses routinely obtained the specimens by using a tongue depressor in infants and young children who could not cough on instruction, and then swabbing aggressively until the child gagged; when children were able to cough when asked to do so, this occurred first, and then the vigorous swabbing of the oropharynx followed until the child gagged. The specimens on swabs were placed in culturette ampules, and the liquid bacterial transport medium was released/activated to keep the swabs wet. Cultures on expectorated sputum (<10% of the specimens) were obtained from patients who could produce such samples. Culture methods were similar at the two centers.
The protocols of this project specified referral patterns by county to each center for follow-up of positive newborn screening tests; procedures for sweat testing; methods and intervals for blood tests, respiratory secretions cultures, chest radiographs, and pulmonary function tests; and guidelines for nutritional and pulmonary management (including oral antibiotic therapy and hospitalizations). As new therapies have been recommended during this investigation (eg, prednisone, ibuprofen, and aerosolized DNase), the study group in association with a policy and data monitoring board has reviewed available information and reached conclusions on amendments to the protocol and conditions in which new agents could be used. This led to restricted use of only two new therapies (ibuprofen and DNase, 9% and 15%, respectively, of enrolled patients). When the project was designed in 1984, the use of aerosol therapy was rare for young children with CF in each center and, therefore, no specifications/expectations were included concerning this treatment modality.
From the outset of this project, outpatient clinic arrangements differed at the two centers. Based on the standard approach and because of limitations in clinic space availability, the infants diagnosed with CF in center B were integrated/interspersed with older CF patients in the regular CF clinics. They used a common waiting room throughout the study, and it was relatively small and confined (10 × 11 feet with 10 chairs) for the first 5 years of the trial. Clinic space constraints in center A and other logistic considerations led us to establish a separate, newborn screening clinic on Mondays in which newly diagnosed patients have been segregated from the older CF patients for 91% of the total visits over 12 years. In center A, the flow of patients was nearly continuous and waiting room delay time generally did not occur. Personnel of the two CF centers used similar practices of handwashing routinely between patients; stethoscope cleaning between patients was not used regularly at either center.
When center differences in the acquisition of P aeruginosawere detected, a retrospective study reviewing the medical records of all CF patients at both centers for 10 years was performed. The goals included determination from chart review when a particular individual was in the clinic and the results of respiratory secretion cultures. The medical records of all CF patients seen between January 1, 1985 and June 15, 1994 at center A and center B were reviewed by a systematic process. A total of 163 patients' medical records at center B and 211 patients' medical records at center A were reviewed, representing 99% of CF patients followed in this state during the interval.
Baseline differences between centers were assessed with respect to the following variables: age at diagnosis, gender, race, total follow-up time, study group status (ie, control vs screened), genotype, meconium ileus status, marital status of parent, education of mother, presence of first-degree relatives with CF, residential area (ie, urbanized [based on the 1990 US census] vs other), population density, number of infections per patient year, number of days hospitalized per patient year, use of chronic maintenance antibiotics, and use of aerosol. For continuous variables, the mean differences between centers were tested by two-tailed t tests and the median differences by Wilcoxon rank-sum tests. For categoric variables, Fisher's exact test was used for testing center differences. Differences in prevalence of pseudomonas between groups or centers were also assessed via Fisher's exact test. Prevalence was determined from the number of patients who at any time during the study period (between January 1, 1985 and June 30, 1994) were culture-positive for pseudomonas. Incidence rates of pseudomonas per person-year were calculated to adjust for varying lengths of patient follow-up. Person-year was defined as the interval between birth and either the first positive pseudomonas culture or the last negative culture if no positive culture was detected. The differences between groups or centers were tested using a normal approximation in which the SE was calculated via a Poisson model.
The Kaplan–Meier estimator and log-rank statistic were used to assess differences in pseudomonas-free survival time between groups and between centers. Using survival analysis is particularly advantageous because this method measures and compares incidence in a precise manner by taking into account the number of patients at risk and the age of acquisition. After the analyses described above revealed a significant center difference, some characteristics before P aeruginosainfection between centers were examined, including presence of first-degree relatives with CF, number of infections per patient-year, days hospitalized per patient-year, and use of chronic maintenance antibiotics (generally trimethoprim-sulfamethoxazole or a cephalosporin such as cephalexin). The pseudomonas acquisition analyses were done for two data sets: 1) the one containing all protocol and nonprotocol cultures, in which the data set consisted of between-center differences in frequency of culturing patients (the mean intervals between cultures were 4.2 months for center A and 2.7 months for center B); and 2) the set consisting of cultures from protocol visits only, in which the intervals between cultures were similar at both centers (mean 6.1 months vs 6.2 months), so that the potential bias between centers attributable to unequal frequencies of cultures could be eliminated.
Descriptive data on the study population are provided in Table1. From 1985 through 1996, which included 9 years of randomly screening a total of 650 341 newborns, 88% of eligible CF patients were enrolled in the evaluation and treatment protocol. It has been possible to maintain 87% of subjects in the study. Table 1 shows that no between-center differences were observed for any of the variables analyzed except location of residence. Patients followed in center B tended to live in urbanized areas with significantly greater population per square mile.
Information on isolates from respiratory secretion cultures were examined in the screened and control groups annually, and no significant differences were detected during the 9-year screening period. Because >90% of the specimens were oropharyngeal swabs, results of these cultures and those of expectorated sputum were combined. Analysis using the Kaplan–Meier survival time method on the data from the overall population, ie, weeks of pseudomonas-free status in the screened compared with the control group, also revealed no differences. In addition, colonization with S aureus was similar in prevalence (78% and 75%, respectively, in the screened and control groups). Only one patient (0.8%) at center B had a positive culture for B cepacia. Table 2summarizes the prevalence and incidence rates of P aeruginosa for all subjects as well as by group and by center. No significant differences between the screened and control groups were observed for prevalence or incidence rate of P aeruginosa or for weeks of P aeruginosa-free status. At the conclusion of randomized neonatal screening, 62.7% of patients in the screened group and 53.6% in the control group showed at least one P aeruginosa-positive culture (P = .360). The incidence rates were .300 and .209, respectively, per person-year for the screened and control groups (P = .130).
Stratification by center and subgroup, however, revealed significant differences in acquisition of P aeruginosa. The prevalence of P aeruginosa was higher at center B than at center A (69.5% vs 48.4%; P = .028). The incidence rate at center B was .101 per person-year higher than at center A. Pseudomonas-free survival analysis (Fig 1) also indicated a shorter time to acquisition of P aeruginosain the CF patient population at center B. In analyzing all culture results, we found a difference of 42.2 weeks in median time to acquisition of P aeruginosa between the CF patients at the two centers. The center difference was analyzed further in the screened and control groups separately (Table 2, Fig2). In the control group, no significant differences were detected in prevalences or incidence rates of P aeruginosa between centers, nor were there any differences in weeks of P aeruginosa-free status. In the screened group, however, prevalence and incidence rates of P aeruginosa-positive cultures were significantly higher at center B than at center A (Table 2). The median duration of the P aeruginosa-free period was much shorter at center B than at center A (52.1 weeks vs 289.4; P = .0002) (Fig 2). A similar analysis using only the scheduled protocol visits, which eliminated possible bias attributable to difference in frequency of culture visits between centers, confirmed all of the above findings. There were significant differences in P aeruginosa prevalence (P = .011), incidence (P= .036), and median pseudomonas-free survival interval (P = .032) between centers (Table 2, Fig 2). Again, these differences were seen only in the screened group.
The self-reported number of infections, days hospitalized, and use of chronic maintenance antibiotics before acquisition of P aeruginosa, as well as the percent of subjects with CF first-degree relatives, were compared between the centers in each group. The number of infections per patient per year was similar at both centers regardless of group. When data regarding use of chronic maintenance antibiotics and days of hospitalization were examined over the total follow-up time, there were no significant differences between centers (Table 3). In addition, there were no significant differences between the screened and control groups within centers. There were no differences in use of intravenous or aerosolized antipseudomonas antibiotics between centers or for groups within centers; only 21% of the study patients ever used aerosolized antibiotics. Although patients at center B had a higher percentage of first-degree relatives with CF than patients at center A, the differences between centers in both groups were not statistically significant. Analysis of pseudomonas-free survival curves by center in screened subjects without CF first-degree relatives revealed a much shorter period at center B (52.0 weeks) than at center A (289.4 weeks;P = .0003) for all cultures and for protocol cultures (Fig 3).
Table 4 provides data on aerosol therapy use in each center. Overall, 63.7% of subjects were ever on aerosol therapy during the course of the study, with 66.3% at center A and 60.8% at center B (P = .563). In each group, the proportion of subjects on aerosol therapy was similar between the centers. On the other hand, 49% of subjects were on aerosol therapy for more than 1 year at center B compared with 32% at center A (P = .084). This trend persisted in both the screened and the control groups, although none of the observed differences were statistically significant.
Table 5 summarizes prevalence data for the entire CF populations of the two CF centers. The overall prevalence ofP aeruginosa was higher at center B than at center A (74.9% vs 64.5%; P = .033). The prevalences were examined further by the age group at the last culture visit. The prevalences were very similar between centers in patients 10 to 19 years of age, and >19 years of age. The prevalence was much higher at center B than at center A in the two younger age groups, especially for 3 to 9 years of age (P = .029). The prevalences of P aeruginosa were significantly different among age groups at both centers (P < .001). The prevalence increases as age increases. More than 90% of the patients older than 19 years were colonized with P aeruginosa. We also examined the month and season of pseudomonas acquisition and could not identify any pattern for either center.
The significance of P aeruginosa in the respiratory tract of patients with CF appears to be well established. Wilmott et al17 studied the survival rates of 117 such patients, according to P aeruginosa status, and found that the presence of this organism was strongly associated with more mortality. More specifically, survival to 16 years of age was 53% in the P aeruginosa-positive patients compared with 84% in patients without this organism. Subsequently, Kerem et al18 found that at 7 years of age, CF patients colonized with P aeruginosa had a significantly lower forced expiratory volume in 1 second compared with uninfected patients. Other data19,20also indicate a relationship between pulmonary dysfunction and pseudomonas-positive cultures. In addition, recent observations by McCubbin et al21 have identified an association between pseudomonas infections and the development of bronchiectasis in young children with CF.
Speculation has existed regarding the transmission of pseudomonas species among patients with CF. Studies have been published suggesting a low risk of person-to-person transmission of P aeruginosa,22 but Danish investigators have reported that CF patients can contaminate the environment with this organism during coughing and that there is a risk of cross-infection.5 In Denmark, an epidemic of a multiresistant strain of P aeruginosa spread among CF patients in 1983. This epidemic was managed by isolating patients with the resistant strain. Subsequently, the incidence of chronic pseudomonas infections in Danish CF patients was dramatically decreased after culture- positive and -negative patients were segregated from each other.8,9 Of the patients treated at the Danish Regional CF center, 57% were colonized with P aeruginosacompared with only 27% in noncenter-treated patients. In addition to person-to-person transmission, a variety of other risk factors has been considered potentially contributory to pseudomonas colonization/infection. These include use of oral antibiotics, mist tent exposure, nebulized water treatment as part of bronchial drainage, aerosol therapy, and CF sibling interactions.3,5,7,25
Substantial evidence indicates that person-to-person transmission ofB (Pseudomonas) cepacia occurs in CF. During the early 1980s, this pathogen was recognized as a major threat to patients with CF.26 In Cleveland, the incidence of B cepaciainfection decreased after separation of colonized from noncolonized patients and initiation of separate summer camp sessions and facilities for the two groups.27 More recent direct evidence has been published demonstrating person-to-person transmission of B cepacia at CF summer camps.28 As a result of these findings, in 1993 the Cystic Fibrosis Foundation recommended the discontinuation of all CF summer camps in the United States. It should be emphasized that the Wisconsin CF camp was closed in 1993, before any of the patients followed in our randomized trial reached an age when they could attend summer camp. If B cepacia can be transmitted between patients, this suggests that P aeruginosa could also be communicable between CF patients. Observations supporting this hypothesis based on molecular typing ofP aeruginosa isolates from CF patients have been reported recently,29,30 including genomic fingerprinting results that provide strong evidence that person-to-person spread occurred during an outbreak of pseudomonas infections in a British CF clinic.30
We were presented with a unique opportunity to investigate acquisition of respiratory pathogens as an extension of our randomized investigation of CF neonatal screening. When the project was designed, we recognized that a comprehensive investigation including both potential benefits and risks was necessary to reach a conclusion about the efficacy of early diagnosis. The major medical risk we defined at the outset was early acquisition of P aeruginosa, which we regarded in 1984 as potentially attributable to more frequent use of oral antibiotics in young children with respiratory infections.25 As we monitored the data obtained from respiratory secretion cultures every 6 months, no differences were detected between the screened and control groups. On the other hand, near the end of the randomization period, comparisons of subgroups indicated a center difference that was significant by Kaplan–Meier survival analysis of the pseudomonas-free period. More extensive assessment of the data in subsequent analyses revealed that the difference was confined to the screened group diagnosed early in infancy through CF neonatal screening. Because they are at the highest risk for initial acquisition of P aeruginosa, these subjects provided us with the greatest potential to identify risk factors. Examination of the special characteristics of the two centers was revealing. First, the patients followed at center B tended to be from urban areas, an outcome predetermined by arrangements for referrals of newborns with positive screening tests to obtain diagnostic sweat tests. Our impression is that children with CF in an urban area have more social contacts with other CF patients, and indeed center B organized special events of short duration such as “Breakfast with Santa” that were well attended by patients and their families. Neither CF center recommended summer camp for these study patients, and we determined that none of the study participants ever attended a summer camp. Most significantly, in our judgment, study patients at center B were integrated and interspersed with older CF patients followed in the center's regular clinics according to the standard practice in the regional CF centers of the United States.
The original purpose of our prospective surveillance of P aeruginosa was to determine whether CF patients diagnosed through screening would have a greater risk of acquiring this organism in early childhood. Although the results of the overall study with the combined subgroups (screened vs control populations) of the two centers were regarded as negative, it is clear that the conditions associated with center B increased the potential risk for patients diagnosed through newborn screening. Although our results do not definitely identify a causative factor, the different characteristics associated with center B are of concern, particularly the clinic and social exposures and greater number of first-degree CF relatives—all of which suggest person-to-person transmission, in keeping with recent evidence from the British study.30 Our results indicate that center differences persist when survival analyses are performed excluding screened subjects with first-degree CF relatives. Other potential risk factors include urban location and a tendency toward longer use of aerosol therapy.
It should be emphasized that a limitation in this study is the use of oropharyngeal cultures. Results published by Armstrong et al31 and Ramsey et al32 suggest that the positive predictive value of oropharyngeal cultures in CF patients ranges from 57% to 83%. Nevertheless, in a longitudinal study with infants and young children, such as in this trial, there is no other option because neither expectorated sputum nor repetitive bronchoscopies can be obtained readily. Repeated assessment of patients followed in our protocol after the first positive culture for any bacterial pathogen demonstrated that ∼80% of the subsequent cultures were positive for S aureus (36%), Haemophilus influenzae (37%), and/or P aeruginosa (40%). Thus, although our results are imperfect with regard to determining lower airway colonization, the data obtained represent the best effort that can be made to examine this issue. Furthermore, it is unlikely that the imperfect sensitivity of oropharyngeal cultures would alter the wide separation of pseudomonas-free survival curves of the screened patients followed at the two centers.
To settle the question of risks associated with various exposures, another study is needed with prospective monitoring of person-to-person interactions. In addition, it will be important to determine whether the acquisition of P aeruginosa simply represents colonization with the organism or whether it predisposes to a greater risk for infection and chronic lung disease. In the meantime, we recommend that CF screening programs take these observations into account in their follow-up systems. Based on our observations in center A and the Denmark experiences,8,9 it is possible that segregated clinics coupled to CF neonatal screening might provide an opportunity to delay acquisition of P aeruginosa. If so, this would represent another potential benefit, albeit unanticipated, of organizing neonatal screening programs for this disease.
This work was supported by National Institutes of Health Grants DK 34108 and RR03186 and by Cystic Fibrosis Foundation Grant A001 5-01.
We thank Drs W. Theodore Bruns, Lee Rusukow, and William Gershon for their essential role in the enrollment and management of study patients; Lynn Feenen, Audrey Tluczek, Miriam Block, Catherine McCarthy, Mary Ellen Freeman, and Holly Colby for expert research nursing support; Ronald Laessig, PhD, Ronald Gregg, PhD, David Hassemer, and Gary Hoffman for development and operation of the IRT and IRT/DNA neonatal screening programs; and Elizabeth Colby and Unchu Ko for their assistance in data management.
- Received December 9, 1996.
- Accepted June 11, 1997.
Reprint requests to (P.M.F.) University of Wisconsin Medical School, 1300 University Ave, 1217 MSC, Madison, WI 53706.
- CF =
- cystic fibrosis •
- IRT =
- immunoreactive trypsinogen
- Pier GB
- Pederson SS,
- Koch C,
- Høiby N,
- Rosendal K
- van Egmond AWA,
- Kosook MR,
- Koscik R,
- Laxova A,
- Farrell PM
- Rock MJ,
- Mischler EH,
- Farrell PM,
- et al.
- ↵Farrell PM, Aronson RA, Hoffman G, Laessig RH. Newborn screening for cystic fibrosis in Wisconsin: first application of population-based molecular genetics testing. Wis Med J. 1994;415–421
- ↵Marks M, Stutman HR. Antibiotic prophylaxis in CF. Presented at the Ninth Annual North American Cystic Fibrosis Conference; Dallas, TX; October 11–15, 1995
- Wilmott RW,
- Tyson SL,
- Matthew DJ
- Winnie GB,
- Cowan RG
- McCubbin MM,
- Ahrens R,
- Kao S,
- Seidel G,
- Teresi M
- Ojeniyi B,
- Steen Petersen U, Høiby N
- Copyright © 1997 American Academy of Pediatrics