Survival Without Disability to Age 5 Years After Neonatal Caffeine Therapy for Apnea of Prematurity

Context Very preterm infants are prone to apnea and have an increased risk of death or disability. Caffeine therapy for apnea of prematurity reduces the rates of cerebral palsy and cognitive delay at 18 months of age.

Objective To determine whether neonatal caffeine therapy has lasting benefits or newly apparent risks at early school age.

Design, Setting, and Participants Five-year follow-up from 2005 to 2011 in 31 of 35 academic hospitals in Canada, Australia, Europe, and Israel, where 1932 of 2006 participants (96.3%) had been enrolled in the randomized, placebo-controlled Caffeine for Apnea of Prematurity trial between 1999 and 2004. A total of 1640 children (84.9%) with birth weights of 500 to 1250 g had adequate data for the main outcome at 5 years.

Main Outcome Measures Combined outcome of death or survival to 5 years with 1 or more of motor impairment (defined as a Gross Motor Function Classification System level of 3 to 5), cognitive impairment (defined as a Full Scale IQ<70), behavior problems, poor general health, deafness, and blindness.

Results The combined outcome of death or disability was not significantly different for the 833 children assigned to caffeine from that for the 807 children assigned to placebo (21.1% vs 24.8%; odds ratio adjusted for center, 0.82; 95% CI, 0.65-1.03; P = .09). The rates of death, motor impairment, behavior problems, poor general health, deafness, and blindness did not differ significantly between the 2 groups. The incidence of cognitive impairment was lower at 5 years than at 18 months and similar in the 2 groups (4.9% vs 5.1%; odds ratio adjusted for center, 0.97; 95% CI, 0.61-1.55; P = .89).

Conclusion Neonatal caffeine therapy was no longer associated with a significantly improved rate of survival without disability in children with very low birth weights who were assessed at 5 years.

Apnea of prematurity is a developmental disorder of respiratory control that may persist until a postmenstrual age of 44 weeks in very immature infants.¹ Caffeine is the respiratory stimulant of choice for the treatment of apnea.² A recent analysis of a large administrative data set found that caffeine is the most frequently used medication in preterm neonates.³ Caffeine has been called a “silver bullet in neonatology”⁴ because of its many proven benefits and few known risks: The international, randomized, placebo-controlled Caffeine for Apnea of Prematurity trial has shown that caffeine therapy reduces the duration of exposure to positive airway pressure and supplemental oxygen and the rates of neonatal morbidities, including bronchopulmonary dysplasia and severe retinopathy of prematurity; improves the rate of survival without neurodevelopmental impairment at 18 to 21 months of age; and is cost-effective.^5-7

However, outcomes up to 2 years after very preterm birth may not accurately predict function later in childhood.^8-12 We studied the children who were enrolled in the Caffeine for Apnea of Prematurity trial at a corrected age of 5 years to detect lasting or previously unrecognized consequences of caffeine therapy. In this article, we describe the effects of caffeine on motor function, cognition, behavior, hearing, vision, and general health when the study participants were old enough to enter school.

Between October 11, 1999, and October 22, 2004, we randomly assigned 2006 preterm infants to receive caffeine citrate or normal saline placebo until drug therapy for apnea of prematurity was no longer needed. Infants with a birth weight of 500 to 1250 g were eligible for this study if their clinicians considered them to be candidates for caffeine therapy during the first 10 days of life. The median age at randomization was 3 days. Indications for the use of caffeine were to prevent or treat apnea and to facilitate the removal of an endotracheal tube. Infants were excluded if they had congenital abnormalities, were unlikely to be available for long-term follow-up, or had received previous treatment with a methylxanthine.

A computer-generated randomization scheme was used to assign infants to the 2 groups in a 1:1 ratio. Randomization was stratified by study center and balanced in random blocks of 2 or 4 patients.⁵ Infants who were part of multiple births were randomized individually. A loading dose of 20 mg of caffeine citrate per kilogram of body weight or an equivalent volume of normal saline was followed by a daily maintenance dose of 5 mg per kilogram. If apneas persisted, the daily maintenance dose could be increased to a maximum of 10 mg of caffeine citrate per kilogram. The drug was monitored according to its clinical effect only.¹³

The median duration of treatment with study drug was 37 days (interquartile range, 24 to 46 days) in the caffeine group and 36 days (interquartile range, 23 to 46 days) in the placebo group. The median postnatal age at which the last dose of study drug was administered was 34.4 weeks (interquartile range, 33.0 to 35.9 weeks) in the caffeine group and 34.7 weeks (interquartile range, 32.9 to 36.1 weeks) in the placebo group.⁵

The research ethics boards of all clinical centers approved the protocol. Written informed consent was obtained from a parent or guardian of each infant. An investigational new drug application was filed with Health Canada, clinical trial notification applications were filed in Australia, and appropriate regulatory approvals were obtained elsewhere. An external data and safety monitoring committee reviewed the study data regularly during the enrollment phase.⁵

The primary outcome was death before a corrected age of 18 months or survival with 1 or more of the following: cerebral palsy, cognitive delay, severe hearing loss, and bilateral blindness.⁶ Cognitive delay was defined as a Mental Development Index score of less than 85 (1 SD below the mean of 100) on the Bayley Scales of Infant Development II.¹⁴ We previously reported these outcomes, including the finding that caffeine improved the rate of survival without neurodevelopmental disability (odds ratio adjusted for center, 0.77; 95% CI, 0.64-0.93) by reducing the risks of cerebral palsy and cognitive delay.⁶

The children's families and all clinicians and researchers have remained unaware of the random assignments to neonatal caffeine or placebo treatment throughout this study and beyond the present 5-year follow-up. At the Data Coordinating and Methods Center, only the senior biostatistician (R.S.R.) had access to the study treatment code.

Four of the original 35 clinical centers did not participate in the 5-year follow-up. These 4 centers had enrolled 74 infants in the initial trial, leaving 1932 children eligible for the present study. The research ethics boards of all participating centers approved the follow-up protocol. Written informed consent was obtained from a parent or guardian of each child. We attempted to contact the families of all surviving children, even if no contact had been made at 18 months. The target date for the 5-year follow-up was corrected for the degree of prematurity by adding the difference between the child's day of birth and the term due date to the fifth birthday. Completing the assessment within 6 months of the target date was recommended, but efforts to locate and assess the children continued beyond this age when necessary. Race or ethnic group was self-reported and recorded according to predetermined categories that were appropriate for this international cohort of families.

The main outcome at 5 years was a composite of death before a corrected age of 5 years or survival with 1 or more of the following: motor impairment, cognitive impairment, behavior problems, poor general health, severe hearing loss, and bilateral blindness.¹⁵ Motor impairment was determined with the use of the Gross Motor Function Classification System (GMFCS).¹⁶ A child with normal motor function has a symmetric gait at 5 years and can run well, jump with both feet, and hop on 1 foot. GMFCS levels between 1 and 5 indicate increasing limitations of gross motor function. To be consistent with the GMFCS analyses that were performed at 18 months, the definition of motor impairment in the protocol for the 5-year follow-up was revised from a GMFCS level greater than 3 to one greater than 2 (GMFCS 3-5).¹⁵

Severe cognitive impairment was defined as a Full Scale IQ of less than 70 (2 SD below the mean of 100) on the Wechsler Preschool and Primary Scale of Intelligence III.¹⁷ Site investigators used their respective national test norms. If a national version of the Wechsler Preschool and Primary Scale of Intelligence III was not available or was no longer appropriate for the age of the child, Full Scale IQ values from similar standardized tests were used. This was the case for 45 children. The Full Scale IQ was assumed to be less than 70 if the child could not complete the testing because of severe developmental delay or severe autism.

Parents completed the Child Behavior Checklist.¹⁸ This questionnaire identifies a broad range of behavioral problems and can be used for children from diverse geographic and cultural backgrounds.¹⁹ A behavior problem was defined as a Total Problem T score (range, 28 to 100) of greater than 69 (2 SD above the mean of 50).¹⁸ Poor general health was defined as 1 or more of the following: need for supplemental oxygen, positive airway pressure, feeding through a tube or intravenously, seizures occurring more frequently than once per month, or a recent admission to an intensive care unit for complications resulting from a neonatal morbidity. Severe hearing loss was defined as the prescription of hearing aids or cochlear implants. Bilateral blindness was defined as a corrected visual acuity less than 20/200 in the better eye.

The Movement Assessment Battery for Children was used to examine motor performance, including manual dexterity, ball skills, and static and dynamic balance.²⁰ Standardized total impairment scores range from 0 to 40, with lower scores indicating better performance. Visual-motor integration, visual perception, and motor coordination were tested with the Beery-Buktenica Developmental Test of Visual-Motor Integration.²¹ Standard scores have a mean of 100 and a standard deviation of 15, with higher scores indicating better performance.

Height, weight, and head circumference were measured during the clinic visit. Because the results of all psychometric tests and of the Child Behavior Checklist were corrected for the degree of prematurity, individual percentiles for height and weight were computed in the data center according to the corrected age at the assessment.²²

This 5-year follow-up study was designed in 2003, more than 3 years before the participants' outcomes at 18 months for the Caffeine for Apnea of Prematurity trial were known. In our sample size estimation for the composite outcome at 18 months, we had assumed a control group event rate between 20% and 30%, although the observed event rate at 18 months was higher. To retain similar statistical power with the available fixed sample size, we chose the criteria for abnormality at 5 years such that our estimated control group event rate for the composite main outcome was in the same range. Assuming an incidence of death or impairment of 25% and an ascertainment rate of 85%, we had a statistical power of 80% to detect a 22% relative reduction in the risk of the combined main outcome at 5 years.

Because randomization was stratified by study center, the analyses of all dichotomous outcomes were adjusted with the use of a logistic regression model that included terms for treatment and center (smaller centers were combined). The regression coefficient associated with treatment in the fitted model yielded a point estimate and 95% CI for the treatment effect expressed as an odds ratio. The quotient of the estimated coefficient and its standard error provided a z-test statistic for the test of the null hypothesis of no treatment effect. An ordinal logistic model was used in a secondary analysis of the GMFCS data. The mean differences between the 2 groups for quantitative outcomes were adjusted for center with multiple linear regression. The relationship between the gain in cognitive function at 5 years and the Mental Development Index score at 18 months was estimated with a linear regression model.

All P values were 2-sided and considered significant if P < .05. SAS version 9.1 was used for all statistical analyses. We chose a composite primary outcome to protect against the problems of multiple testing. Additional adjustments for multiple comparisons were not made.

The numbers of infants who were randomly assigned to receive neonatal caffeine or placebo therapy and the numbers assessed at a corrected age of 5 years are shown in Figure 1. Follow-up assessments began on March 12, 2005, and ended on January 26, 2011. Adequate data for an analysis of the primary composite outcome were available for 1640 (84.9%) of the 1932 children who were eligible for this study.

The characteristics of these 1640 children and of their maternal caregivers and families were broadly similar in the 2 groups (Table 1). The eTable shows the characteristics of participants and nonparticipants in the 5-year follow-up. As in the initial study, the rates of death or disability at 18 months and of cerebral palsy were reduced by caffeine therapy in the present study cohort. Treatment with caffeine compared with placebo significantly reduced the incidence of both moderate and severe cognitive delay at 18 months, defined as a Bayley II Mental Development Index score less than 85 and less than 70, respectively (Table 1).

The results for the primary composite outcome and for its components are shown in Table 2. Of the 833 children who had been assigned to caffeine, 176 (21.1%) died or survived with at least 1 impairment compared with 200 of the 807 children (24.8%) assigned to placebo (odds ratio adjusted for center, 0.82; 95% CI, 0.65-1.03; P = .09). The rates of motor impairment, severe cognitive impairment, behavior problems, poor general health, blindness, and deafness were not significantly different between the 2 groups (Table 2). Only 2 children in each of the 2 groups died between 18 months and 5 years.

A secondary analysis using the full ordinal scale of the GMFCS showed evidence of an improvement in gross motor function associated with caffeine therapy. The common odds ratio adjusted for center for the 5 levels of impairment was 0.64 (95% CI, 0.47-0.88; P = .006; Table 3). Better mean scores for total impairment and manual dexterity on the Movement Assessment Battery for Children and better mean standard scores for visual perception and motor coordination on the Beery-Buktenica Developmental Test were likewise associated with random assignment to neonatal caffeine therapy (Table 4).

The rates of moderate cognitive impairment, defined as a Full Scale IQ of less than 85 (Table 3); the mean scores for the Full Scale IQ, Verbal IQ, Performance IQ, and Total Problem Behavior T score; and the mean weight, height, and head circumference measures were not significantly different between the 2 groups (Table 4).

We performed a post hoc analysis of the composite 5-year outcome that included children who were blind, were profoundly deaf, or had GMFCS levels 3 to 5 at 18 months but who were not seen at 5 years. This imputation added 2 and 6 children, respectively, to the caffeine and placebo groups. The odds ratio adjusted for center for this comparison was 0.80 (95% CI, 0.63-1.01; P = .06).

A total of 1553 children were assigned a GMFCS level at both ages. Of the 30 children with a GMFCS level greater than 2 at 18 months, 26 (86.7%) had a level greater than 2 at 5 years. Of the remaining 1523 children with a GMFCS level of 2 or less at 18 months, 1515 (99.5%) remained unimpaired or mildly impaired at 5 years.

A total of 1419 children underwent successful cognitive testing at both points. Rates of moderate and severe cognitive impairments were much lower at 5 years than at 18 months. Only 33 of the 178 children (18.5%) with a Bayley II Mental Development Index less than 70 at 18 months had a Wechsler Preschool and Primary Scale of Intelligence III Full Scale IQ less than 70 at 5 years. Figure 2 shows the gain in cognitive scores between 18 months and 5 years. The fitted regression line suggests that, on average, a child with a Mental Development Index of 70 at 18 months had a Wechsler Preschool and Primary Scale of Intelligence III Full Scale IQ that was almost 20 points higher at 5 years.

We performed the Caffeine for Apnea of Prematurity trial to resolve the longstanding uncertainty about the benefits and risks of neonatal caffeine therapy.^24,25 Previously, we reported that caffeine improved the rate of survival without neurodevelopmental disability at 18 to 21 months in infants with very low birth weight by reducing the incidence of cerebral palsy and cognitive delay.⁶ The present report summarizes the results of detailed assessments of motor function, cognition, behavior, general health, hearing, and vision after the study participants had reached a corrected age of 5 years. Although the observed rates of the combined main outcome of death or disability continued to favor caffeine over placebo treatment, the difference between the groups was not statistically significant. The rates of cognitive impairment were much lower at 5 years than at 18 months and similar in the caffeine and placebo groups. The rates of death, behavior problems, poor general health, severe hearing loss, and bilateral blindness did not differ significantly between the 2 groups.

We conducted the present study because outcomes up to 2 years are not reliable end points for follow-up studies of very preterm infants.^8,12 We observed a substantial decline in the rate of cognitive impairment in this large, international cohort of children with very low birth weights who were raised in mostly middle-class families. Further examination of the relationship between the individual Mental Development Index scores at 18 months and change in cognitive scores at 5 years regardless of group assignment showed that children with very low scores at 18 months had improved cognitive scores at 5 years, whereas children with Mental Development Index scores above 100 at 18 months had slightly lower Full Scale IQ scores at 5 years. Although this phenomenon may be partly statistical because retesting would be expected to show some regression to the mean, the observed overall improvement in cognition between 18 months and 5 years was striking and likely real.^8,12 A low Bayley II Mental Development Index score at 18 months may reflect transient developmental delay rather than lasting cognitive impairment. This pattern of change in cognitive performance between 18 months and 5 years has important implications for all outcome studies after very preterm birth and for the future evaluation of the long-term effects of common neonatal therapies.

Very preterm infants are at risk of motor impairment even if they do not develop cerebral palsy.^26,27 Our supportive analysis of the full ordinal scale of the GMFCS suggested that gross motor impairment was less severe in caffeine-treated children than in controls. Caffeine therapy was also associated with improved scores on the Movement Assessment Battery for Children. Last, children in the caffeine group had better motor coordination and visual perception, although the magnitude of the differences was small (less than 0.2 SD). However, “subtle motor problems are likely to increase when greater demands are put on these vulnerable children” as they progress through school.²⁸ Further follow-up of our study cohort is in progress at 11 to 12 years and will enable us to examine whether motor performance and visual perception at age 5 years are strong predictors of academic success.

Concerns have been lingering that neonatal caffeine therapy may cause long-term harm.^24,25 The methylxanthines inhibit 2 adenosine receptors that have been linked to anxious and aggressive behavior in mice with targeted deficiencies of each of these receptors.^29,30 The absence of any adverse effects in children who were randomly assigned to neonatal caffeine therapy on the incidence of behavior problems or on any other outcomes is reassuring.

In summary, this 5-year follow-up study of participants in the international Caffeine for Apnea of Prematurity trial showed that the benefits of neonatal caffeine therapy on the rate of survival without disability at 18 months were attenuated during child development. The rates of cognitive impairment were much lower at 5 years than at 18 months, suggesting that cognitive delay during the second year of life may not be a lasting outcome after very preterm birth.

Corresponding Author: Barbara Schmidt, MD, MSc, Division of Neonatology, Hospital of the University of Pennsylvania, Ravdin 8, 3400 Spruce St, Philadelphia, PA 19104 (barbara.schmidt@uphs.upenn.edu or schmidt@mcmaster.ca).

Author Contributions: Mr Roberts had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Schmidt, Anderson, Doyle, Dewey, Grunau, Davis, Tin, Moddemann, Ohlsson, Solimano, Roberts.

Acquisition of data: Schmidt, Anderson, Doyle, Grunau, Asztalos, Tin, Moddemann, Ohlsson, Barrington.

Analysis and interpretation of data: Schmidt, Dewey, Solimano, Roberts.

Drafting of the manuscript: Schmidt, Roberts.

Critical revision of the manuscript for important intellectual content: Anderson, Doyle, Dewey, Grunau, Asztalos, Davis, Tin, Moddemann, Solimano, Ohlsson, Barrington.

Statistical analysis: Roberts.

Obtained funding: Schmidt, Dewey, Ohlsson, Roberts.

Administrative, technical, or material support: Schmidt, Anderson, Dewey, Davis, Moddemann.

Study supervision: Schmidt, Anderson, Grunau, Asztalos, Tin, Solimano.

Funding/Support: This follow-up study was supported by the Canadian Institutes of Health Research (MCT 13288).

Role of the Sponsors: Neither the Canadian Institutes of Health Research, Chiesi Pharmaceuticals, nor any other manufacturer of caffeine citrate had a role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.

The CAP Trial Investigators

The following investigators and research staff contributed to the 5-year follow-up of CAP trial participants. Study sites are listed according to the number of infants they enrolled. McMaster University Medical Center, Hamilton, Canada: Barbara Schmidt, Judy D’Ilario, Janice Cairnie, Joanne Dix, Beth Adams, Erin Warriner, Mee-Hai Marie Kim; Royal Women's Hospital, Melbourne, Australia: Peter Anderson, Peter Davis, Lex Doyle, Brenda Argus, Kate Callanan, Noni Davis, Julianne Duff, Marion McDonald; Sunnybrook Health Sciences Center, Toronto, Canada: Elizabeth Asztalos, Denise Hohn, Maralyn Lacy; Women's and Children's Hospital, Adelaide, Australia: Ross Haslam, Christopher Barnett, Louise Goodchild, Rosslyn Lontis; Mercy Hospital for Women, Melbourne, Australia: Simon Fraser, Julie Keng, Kerryn Saunders, Gillian Opie, Elaine Kelly, Heather Woods, Emma Marchant, Anne-Marie Turner, Noni Davis, Emma Magrath, Amanda Williamson; Centre Hospitalier Universitaire de Quebec, Quebec City, Canada: Aida Bairam, Sylvie Bélanger, Annie Fraser; Ottawa Hospital, Ottawa, Canada: Brigitte Lemyre, Jane Frank; Children's & Women's Health Centre of British Columbia, Vancouver, Canada: Alfonso Solimano, Anne Synnes, Ruth E. Grunau, Philippa Hubber-Richard, Marilyn Rogers, Margot Mackay, Julianne Petrie-Thomas, Arsalan Butt; Academic Medical Center, Amsterdam, the Netherlands: Aleid van Wassenaer, Debbie Nuytemans, Bregje Houtzager, Loekie van Sonderen; Meir General Hospital, Kfar-Saba, Israel: Rivka Regev, Netter Itzchack, Shmuel Arnon, Adiba Chalaf; Mount Sinai Hospital, Toronto, Canada: Arne Ohlsson, Karel O’Brien, Anne-Marie Hamilton, May Lee Chan; Royal University Hospital, Saskatoon, Canada: Koravangattu Sankaran, Sheila Morgan, Pat Proctor; Soroka University Medical Center, Beer Sheva, Israel: Agneta Golan, Esther Goldsch-Lerman; The Canberra Hospital, Canberra, Australia: Graham Reynolds, Barbara Dromgool, Sandra Meskell, Vanessa Parr, Catherine Maher, Margaret Broom, Zsuzsoka Kecskes, Cathy Ringland; Foothills Hospital and Alberta Children's Hospital, Calgary, Canada: Douglas McMillan, Debbie Schaab, Elizabeth Spellen, Reginald S. Sauve, Heather Christianson, Deborah Anseeuw-Deeks, Dianne Creighton, Jennifer Heath; St. Boniface, Winnipeg, Canada: Ruben Alvaro, Aaron Chiu, Ceceile Porter, Gloria Turner, Diane Moddemann, Naomi Granke, Karen Penner, Jane Bow; University Hospital Maastricht, Maastricht, the Netherlands: Antonius Mulder, Renske Wassenberg, Markus van der Hoeven; Kingston General Hospital, Kingston, Canada: Maxine Clarke, Judy Parfitt, Kevin Parker; Windsor Regional Hospital, Windsor, Canada: Chukwuma Nwaesei, Heather Ryan, Cory Saunders; Ludwig Maximilian University, Munich, Germany: Andreas Schulze, Inga Wermuth, Anne Hilgendorff, Andreas W. Flemmer; Astrid Lindgren Children's Hospital, Stockholm, Sweden: Eric Herlenius, Lena Legnevall, Hugo Lagercrantz; Victoria General Hospital, Victoria, Canada: Derek Matthew, Wendy Amos, Suresh Tulsiani, Cherrie Tan-Dy, Marilyn Turner, Constance Phelan; Kaplan Medical Center, Rehovot, Israel: Eric Shinwell, Michael Levine, Ada Juster-Reicher; Royal Victoria Hospital, Montreal, Canada: May Khairy, Patricia Grier, Julie Vachon, Larissa Perepolkin, Keith J. Barrington; James Cook University Hospital, Middlesbrough, United Kingdom: Sunil Sinha, Win Tin, Susan Fritz; University of Sherbrooke, Sherbrooke, Canada: Herve Walti, Diane Royer; Royal Maternity Hospital Belfast, Northern Ireland: Henry Halliday, David Millar, Anne Berry, Clifford Mayes, Christopher McCusker, Olivia McLaughlin; Basel Children's Hospital, Basel, Switzerland: Hubert Fahnenstich, Bettina Tillmann, Peter Weber; Royal Victoria Infirmary, Newcastle upon Tyne, United Kingdom: Unni Wariyar, Nicholas Embleton, Ravi Swamy; University Hospital Zurich, Zurich, Switzerland: Hans U. Bucher, Jean-Claude Fauchere, Vera Dietz; Northern Neonatal Initiatives, Middlesbrough, United Kingdom: Chidambara Harikumar, Win Tin, Susan Fritz.

Steering Committee for 5-Year Follow-up: Barbara Schmidt (Chair), Peter J. Anderson, Elizabeth V. Asztalos, Keith J. Barrington, Peter G. Davis, Deborah Dewey, Lex W. Doyle, Ruth E. Grunau, Diane Moddemann, Arne Ohlsson, Robin S. Roberts, Alfonso Solimano, Win Tin.

External Data and Safety Monitoring Committee: Michael Gent (Chair), William Fraser, Edmund Hey, Max Perlman, Kevin Thorpe.

Consultant Pharmacist: Shari Gray.

Coordinating and Methods Center in Hamilton, Canada: Robin S. Roberts, Carole Chambers, Lorrie Costantini, Wendy Yacura, Erin McGean, Lori Scapinello.