Infant Care Patterns at Epidemiologic Study of Cystic Fibrosis Sites That Achieve Superior Childhood Lung Function

METHODS

The ESCF database was used to establish 2 cross- sectional cohorts for this analysis: (1) an outcomes cohort of patients who were 6 to 12 years of age in 2003 and (2) an infant cohort of patients who were treated as infants between 1994 and 1999 at sites that were in the highest or lowest quartiles with respect to FEV₁ in 2003. CF sites that were included in the outcomes cohort reported on at least 20 patients who were aged 6 to 12 years in 2003; 24 sites that participated in ESCF II met these criteria. The best FEV₁ % predicted that was obtained in 2003 was determined for each patient, median FEV₁ was calculated for each site, and those in the highest and lowest quartiles were identified. The infant cohort, aged 0 to 3 years, was used to compare demographics, clinical characteristics, and care patterns identified at upper (UQ) and lower quartile (LQ) sites. All patients who enrolled and participated in ESCF and ESCF II provided informed consent.

Data that were collected for each patient reflected the period from enrollment into the ESCF to 12 months after enrollment. The following demographic characteristics were collected: age, gender, race (non-Hispanic white or other), genotype, and method of CF diagnosis. Clinical characteristics included mean weight-for-age and height-for-age percentiles (during the first year of enrollment), signs and symptoms of lung disease (cough was described as none, occasional, or daily; and the presence of crackles, wheezing, and clubbing was identified using a check box), medical conditions (eg, asthma, sinusitis, elevated liver function results), microbiology and related variables, and nutritional status. The presence of asthma or sinusitis was based on the diagnostic determination of the individual physician. Practice patterns were defined by number of visits and use of parenteral antibiotics and routine therapies.

Continuous variables are presented as means ± SD, and categorical variables are presented as number (%). Two-tailed χ² tests (for categorical variables) and t tests (for continuous variables) were performed to identify any differences between UQ and LQ sites. These variables were examined further by means of multivariable logistic regression modeling, controlling for quartile, genotype, non-Hispanic race, and age. All statistical analyses and summaries were performed using SAS 9.1 (SAS Institute, Cary, NC). P < .05 was considered to be significant.

RESULTS

Quartiles were established from an outcomes cohort of 837 children who were aged 6 to 12 years; 525 of these patients also were included in the infant cohort to evaluate clinical characteristics and practice patterns. A total of 1498 infants were included in the analysis of practice patterns: 755 patients from UQ sites and 743 patients from LQ sites. Figure 1 shows the distribution and overlap of infants and children who were evaluated in the analysis.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

This Venn diagram describes the relationship between the children from ESCF II, for whom outcome data were evaluated, and the infants from ESCF, for whom practice patterns were evaluated. Of the 837 patients who were in either a UQ or LQ site as children aged 6 to 12 years, 525 also were included in the practice pattern evaluation as infants. All but 16 of these 525 patients were at the same site in ESCF and ESCF II. The analysis of the infants included these 525 patients plus an additional 973 patients who were cared for at the same sites but who did not contribute data as children in ESCF II.

Nutritional Status

Infants at UQ sites exhibited both higher weight-for-age and height-for-age percentiles (P < .0001) compared with patients at the LQ sites (Fig 2), but these differences were not large. The mean weight-for-age percentiles at the UQ and LQ sites were 28.4 and 22.6, respectively, and the mean height-for-age percentiles at the UQ and LQ sites were 31.7 and 25.9, respectively.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

Nutritional status as determined by mean (±SD) percentiles for weight for age and height for age by outcomes quartile (based on FEV₁ in 2003).

Office Visits and Hospitalizations

Infants who received care at UQ sites had more frequent office visits (6.4 ± 3.3 vs 5.9 ± 3.1; P = .009) and sick visits (1.2 ± 1.9 vs 1.0 ± 1.7; P = .014) but a similar number of hospital admissions (0.5 ± 0.9 vs 0.5 ± 1.0) in the 12 months after enrollment (Fig 3).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3

Mean (±SD) number of office visits, hospitalizations, and sick visits by outcomes quartile (based on FEV₁ in 2003).

DISCUSSION

This analysis compares the demographics, clinical characteristics, and medical care received during the first 12 months after enrollment at ESCF sites in the highest and lowest quartiles for FEV₁ % predicted in 2003. We found that sites where patients had the best lung function at ages 6 to 12 years had fewer infants with risk factors (nonwhite race, symptomatic diagnosis, P aeruginosa airway infection) compared with sites with lower lung function and that infants who attended these sites had slightly better height and weight percentiles. Some of the risk factors, such as ethnicity and genotype, are not modifiable. Others, including symptomatic diagnosis, cannot be modified at the time of an individual patient's diagnosis but can be changed in future patient populations with specific interventions. Importantly, specific patterns of patient monitoring and treatment at sites in the highest quartile were significantly different from those at the lower quartile sites.

Comparisons of site outcomes must consider the baseline characteristics of patients who attended those sites, which may not be modifiable by treatment, and differences in treatment approach. Nonmodifiable differences between UQ and LQ sites were seen in this analysis. Fewer ethnic minorities were seen at UQ sites. Minority ethnicity is associated with a lower socioeconomic status in the United States, which, in turn, is associated with poorer pulmonary function in patients with CF.⁹ Because socioeconomic data were not available for this analysis, it is not possible to evaluate the impact of socioeconomic status separately. Fewer pancreatic-insufficient patients, as identified by pancreatic enzyme use, were seen at UQ sites. Pancreatic insufficiency is associated with increased disease severity.^10,11

Some of the risk factors that were seen at LQ sites can be modified in future populations of infants with CF. Infants from LQ sites demonstrated a greater severity of illness and a lower weight for age at enrollment. Both weight for age and severity of illness during the first year after diagnosis affect morbidity and mortality in children with CF.^12,13 These risk factors can be modified by widespread implementation of newborn screening programs. Observational and randomized studies have demonstrated that early diagnosis and treatment improve long-term nutritional status and lung health^12,14 and are the basis for the Centers for Disease Control and Prevention and the Cystic Fibrosis Foundation consensus for routine newborn screening for CF.¹⁵ In the United Kingdom, infants whose CF is diagnosed by newborn screening require less intensive therapies than those whose CF is diagnosed clinically.¹⁶ Our data show that ESCF sites whose 6- to 12-year-old patients have the highest pulmonary function had more infants whose CF was diagnosed by newborn screening or family history than sites with the lowest pulmonary function. These infants also seem to require less intensive therapy, because less frequent prescription of airway clearance therapy, inhaled bronchodilators, and enteral supplemental feedings were seen at UQ sites. There are no randomized, controlled studies of these therapies in CF, and the association between use of these therapies in infants and outcomes cannot be assessed adequately with the current analysis.

More patients from the LQ sites had sputum cultures positive for P aeruginosa, which is associated with greater airway inflammation and poorer lung function in ensuing years. In contrast, UQ sites performed more cultures per patient, which might be expected to allow earlier detection and treatment. Infection control practices and early intervention for P aeruginosa infection could reduce the frequency of positive P aeruginosa cultures in the population and further improve outcomes.

For achievement of the best health outcomes for infants whose CF is diagnosed without clinical symptoms and for improvement of outcomes for those whose CF is diagnosed with symptoms, it is essential to recognize and apply monitoring and treatment strategies that may improve subsequent health status. Although significant information indicates that important physiologic aberrations occur early in CF, very few data give the clinician insight into the best treatment strategies. For example, there is evidence for early airway infection and inflammation in infants and young children with CF,^17,18 and structural airway abnormalities are apparent on high-resolution computed tomography scan in infants and young children with CF.¹⁹ Abnormalities that are seen on high-resolution computed tomography are prevalent even in a cohort of children with normal lung function,²⁰ the conventional marker of lung health. Furthermore, early malnutrition has deleterious effects on cognitive function in children with CF; minimizing the duration of vitamin E deficiency may be associated with better cognitive function.²¹ Therefore, early diagnosis and prompt care, including close attention to growth and vitamin status, of infants with CF are expected to improve long-term outcomes. We do not know, however, which interventions might prevent bronchiectasis or slow its progression or whether specific strategies for nutrition in infants and young children could improve further later nutrition.

Although our analysis showed that sites with the highest FEV₁ had a lower proportion of high-risk infants, it is important and most useful to examine the differences in treatment approaches that were used during infancy. In older children, significant differences in care patterns and outcomes have been noted at ESCF sites.⁷ More visits to the site, more frequent respiratory tract cultures, and more frequent use of intravenous antibiotics have been documented at UQ versus LQ sites. The present analysis shows that these specific care patterns also are seen in the care of the youngest patients at sites that later demonstrate the best pulmonary function outcomes. Despite lesser severity of illness, infants in the UQ sites were evaluated more frequently (office and sick visits), cultured more frequently, and treated more often with intravenous antibiotics (at home), oral corticosteroids, mast cell stabilizers, and mucolytics than those at LQ sites. Increased use of anti-inflammatory therapies in young children in UQ sites suggests consideration of evaluating the benefits of these agents in prospective, controlled clinical trials. Also surprising was the finding that patients at UQ sites received supplemental oxygen more frequently than those at LQ sites; given that UQ patients had a lower risk for severe lung disease, this finding might represent a variability in prescribing practices. The more frequent diagnosis of sinusitis at the UQ sites (P < .001) may be a marker for more frequent antibiotic therapy, leading to better lung function in subsequent years. The significance of this finding is difficult to determine because case report forms captured physician-diagnosed sinusitis but did not give specific criteria for this diagnosis.

It is intriguing that airway clearance and the administration of bronchodilators, interventions that the majority of CF care providers initiate early and consider to be effective, were used more frequently at the LQ than the UQ sites. It is possible that providers at UQ sites do not prescribe these therapies in asymptomatic or minimally symptomatic infants. It also is possible that UQ sites more frequently diagnose CF by newborn screening and that these patients receive less intense treatment.¹⁶ Alternatively, this finding may be explained by the variability in defining actual versus prescribed care. The case report forms were designed to capture prescribed therapy, but because case reports are extracted from the medical chart, it is possible that intended treatment, rather than actual treatment, has been captured and that parents of asymptomatic or minimally symptomatic infants are less likely to adhere to airway clearance regimens. This issue might be investigated further through a detailed survey of CF care provider practices.

It is impossible to determine fully the influence of care in previous years versus more recent or current care on pulmonary function outcomes. However, given the early onset of inflammation and bronchiectasis in CF and the progressive nature of CF lung disease, it is compelling to consider the benefits of frequent monitoring and more frequent use of antibiotics in infants and young children as strategies for improving long-term health outcomes. Our study design is limited by the use of 2 cross-sectional cohorts at identical sites rather than a true longitudinal assessment of individual patients. However, 64% of the 6- to 12-year-old patients who were used to define the outcomes cohorts were part of the infant cohort at the same care site. Nonetheless, we believe that our findings are valid. A longitudinal cohort study would be confirmatory, and multivariable analysis then would allow the ascertainment of the relative contribution of patient risk factors and treatment patterns in relation to pulmonary outcomes.

It is important to realize that an epidemiologic study can identify only associations and, as such, does not define cause and effect but rather points out opportunities for future evaluation. In the present study, observation of better lung function at ESCF sites in patients who were aged 6 to 12 years is associated with specific patient characteristics and clinical care patterns in infants in previous years. Although some patient characteristics are not modifiable, others can be modified through currently available strategies such as widespread implementation of newborn screening. Higher FEV₁ is associated with specific practice patterns, including more frequent visits, more sputum cultures, and more frequent antibiotic therapy. The correlations between these practices and improved outcomes generate hypotheses for prospective research and suggest specific strategies that can be implemented in quality improvement initiatives. Frequent monitoring and use of antibiotics may preserve long-term lung health in CF.