[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.205.0.26. Please contact the publisher to request reinstatement.
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Article
January 2005

Factors That Influence Use of a Home Cardiorespiratory Monitor for InfantsThe Collaborative Home Infant Monitoring Evaluation

Author Affiliations

Author Affiliations: Department of Pediatrics, Rush Medical College of Rush University, Rush Children’s Hospital, Chicago, Ill (Drs Silvestri and Weese-Mayer and Ms Smok-Pearsall); Department of Pediatrics, The University of Texas Southwestern Medical School, Dallas (Dr Lister); Departments of Pediatrics and Epidemiology and Biostatistics, Boston University Schools of Medicine and Public Health, Boston, Mass (Drs Corwin and Cantey-Kiser); Departments of Pediatrics (Dr Baird) and Obstetrics and Gynecology (Ms Mendenhall and Dr Neuman), Case Western Reserve University School of Medicine, MetroHealth Medical Center, Cleveland, Ohio; Department of Pediatrics, John A. Burns School of Medicine, University of Hawaii at Manoa, Kapiolani Medical Center for Women and Children, Honolulu (Drs Crowell and Tinsley); Department of Pediatrics, Medical College of Ohio, Toledo (Dr Hunt); Department of Pediatrics and Neonatology, University of Southern California School of Medicine, Los Angeles; Los Angeles County and University of Southern California Medical Center, Women’s and Children’s Hospital, Los Angeles (Drs Palmer and Hoppenbrouwers); and Pregnancy and Perinatology Branch, Center for Research for Mothers and Children, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Md (Dr Willinger).

Arch Pediatr Adolesc Med. 2005;159(1):18-24. doi:10.1001/archpedi.159.1.18
Abstract

Background  As part of the Collaborative Home Infant Monitoring Evaluation, a home monitor was developed to record breathing, heart rate, other physiologic variables, and the time the monitor was used.

Objective  To determine the frequency of monitor use, factors that influence use, and validity of a model developed to predict use.

Design  We developed a model to predict monitor use using multiple linear regression analysis; we then tested the validity of this model to predict adherence for the first week of monitoring and for the subsequent 4-week period (weeks 2-5).

Setting  Clinical research centers in Chicago, Ill; Cleveland, Ohio; Honolulu, Hawaii; Los Angeles, Calif; and Toledo, Ohio.

Patients  Preterm infants, infants younger than 1 month with a history of autopsy-confirmed sudden infant death syndrome in a sibling, and infants with an idiopathic apparent life-threatening event were divided into 2 cohorts based on enrollment date.

Main Outcome Measure  Mean hours of monitor use per week.

Results  In cohort 1, the variables available before monitoring were only weakly associated with total hours of monitor use in weeks 2 to 5 (total model r2 = 0.08). However, when hours of monitor use in week 1 were included as a variable to predict monitor use in weeks 2 to 5, the r2 increased to 0.64 for hours of monitor use per week.

Conclusions  Our data show that monitor use in the first week was the most important variable for predicting subsequent monitor use. The study suggests that a major focus of home monitoring should be adherence in the first week, although it remains to be tested whether this adherence can be altered.

Home cardiorespiratory monitoring has frequently been prescribed for preterm infants with apnea, infants who have experienced an idiopathic apparent life-threatening event (ALTE), infants with a sibling who died of sudden infant death syndrome (SIDS), and infants who depend on a technological resource for support.1 Although the efficacy of home cardiorespiratory monitoring for the prevention of SIDS has not been evaluated in clinical trials1 and Collaborative Home Infant Monitoring Evaluation (CHIME) data have not substantiated the utility of monitoring for some groups of infants,2 clinically indicated monitoring clearly cannot be effective unless families are willing to use the equipment. Furthermore, there are still many infants for whom monitoring is recommended.3

Monitors equipped with memory have been shown to be useful, not only to distinguish true from false alarms and document the nature of cardiorespiratory events that occur in the home47 but also to document the hours the monitor was in use.79 As part of a multicenter collaborative study (CHIME), a state-of-the-art home monitor (Non-Invasive Monitoring Systems Inc, Miami Beach, Fla) was developed to assess breathing, heart rate, hemoglobin saturation, body position, and the precise time when the monitor was in use.10 To understand and if possible predict caregiver adherence, we determined (1) the frequency of monitor use among infants enrolled in CHIME, (2) the sociodemographic factors that influence monitor use, and (3) the validity of a model developed to predict monitor use. Although the CHIME monitor is more complicated for parents to use than the current conventional impedance-electrocardiographic monitor, these findings may provide a strategy and a standard for evaluating the extent of use of any home monitor. Moreover, the ability to predict monitor use and caregiver adherence for this population of infants provides insight into strategies that may prove valuable in other clinical and research settings where medical technological devices are used at home. The rationale for our approach was based on 2 hypotheses: (1) information derived before monitoring would be predictive of monitor use and hence valuable for directing strategies for caregiver support and (2) monitor use within the first week would yield patterns predictive of subsequent adherence.

METHODS

Frequency of monitor use and variables that were presumed to influence monitor use (including race, marital status, and rationale for monitoring) were examined in all patients. Our strategy was to develop a model for prediction of monitor use with data from patients enrolled in cohort 1, followed by validation of this model with use of patients enrolled in cohort 2. To understand patterns of use, we chose to perform this modeling to predict total hours of use in the first week of monitoring and in the subsequent 4-week period.

PATIENT POPULATION

Patients were infants enrolled in the CHIME study from May 1, 1994, to August 31, 1997, at 5 clinical sites (Chicago, Ill; Cleveland, Ohio; Honolulu, Hawaii; Los Angeles, Calif; and Toledo, Ohio). Informed consent was obtained before enrollment of each infant according to institutional guidelines at each site. To be eligible, infants had to meet 1 of the following criteria: (1) younger than 1 month with a sibling who died of SIDS as confirmed by autopsy; (2) idiopathic ALTE experienced between age 12 hours and 6 months that included color or muscle tone change and required intervention, with study enrollment within 2 weeks of the event; and (3) preterm age younger than 34 weeks of gestation and birth weight of 1750 g or less, study enrollment within 2 weeks of hospital discharge, and a postnatal age of younger than 120 days. Exclusion criteria included the following: current caregiver drug use, language barrier, or no telephone in the home; current pneumonia, congenital heart disease other than asymptomatic left to right shunts, congenital and chromosomal anomalies, inborn errors of metabolism, brain anomalies, or ventriculoperitoneal shunt; or home treatment with anticonvulsants, bronchodilators, diuretics, steroids, respiratory stimulants other than methylxanthine, or continuous oxygen. For data analyses, the infants enrolled from May 1, 1994, to April 30, 1996 (n = 527), were identified as cohort 1 and those enrolled from May 1, 1996, to August 31, 1997 (n = 248), as cohort 2.

The information obtained at enrollment included family history and maternal and infant history related to pregnancy, labor, delivery, and hospital course. As part of the family interview at enrollment, caretakers were asked whether the family reported problems getting along before the infant was born.

MONITORING METHODS

Home monitoring was performed with the CHIME monitor, which included electrocardiographic electrodes, conventional impedance sensors on the chest, and respiratory bands (Respibands; Ambulatory Monitoring, Ardsley, NY) on the rib cage and abdomen to monitor breathing by respiratory inductance plethysmography, a pulse oximeter probe on the foot, and a position sensor on the diaper.10 In addition, the time the monitor was in use was recorded. All data were stored on a removable hard drive that was transferred to a central site (Data Coordinating and Analysis Center, Boston, Mass) for analysis. Caregivers at all sites received standardized training in the use of the monitor.

The first week of monitoring was defined as beginning with the first full day of monitoring in the home. Depending on the indication, monitoring was prescribed for variable durations. For purposes of this article, monitor use was assessed during the first 5 weeks following enrollment because all infants were expected to use the prescribed monitor for at least this period. All participants were instructed to use the monitor for all sleep and any unattended awake time.

FAMILY CONTACT AND SUPPORT

The staff was available 24 hours a day for questions related to the study protocol or monitor. Caregivers were contacted at least weekly to complete a scripted interview regarding alarms, infant health status, monitor use, reasons for nonuse, and any equipment problems. If the review of previous monitor use revealed that it was not being used at least 8 hours per day for 80% of the days, barriers to use were examined and additional support or intervention on behalf of enhanced monitor use was provided as appropriate. This included but was not restricted to home visits, retraining, and referrals to community resources. The CHIME staff or an affiliated home care vendor scheduled home visits that included downloading of monitor data within the first week of monitoring, at 1 month, and then every 4 weeks thereafter. Barriers to monitor use were again examined, and additional support and education regarding monitor use were provided. At each of these contacts, caregivers were provided support, training, and encouragement to use the monitor.

STATISTICAL ANALYSES

When baseline characteristics of cohort 1 and cohort 2 were assessed, continuous variables that were normally distributed (expressed as mean ± SD) were compared by analysis of variance, and χ2 tests were used to compare categorical variables. To assess the variables associated with monitor use during weeks 2 to 5 for cohort 1, multiple linear regression analyses were performed as follows. First, study group (sibling with SIDS, idiopathic ALTE, or preterm delivery at <34 weeks’ gestational age) and race (white, black, Hispanic, Asian, or other) were forced into the model. Next, a forward stepwise linear regression analysis was performed in which candidate independent variables were retained only if P<.05 was achieved. The list of candidate independent variables included those potentially associated with the amount of monitor use and included the following maternal factors: age, married or not married, education, parity, smoking or alcohol use during pregnancy, and self-reporting that the family was or was not getting along well before the infant was born. We included the following infant factors: sex, gestational age, birth weight, and age that monitoring began. Following performance of the linear regression analysis on the entire cohort 1 population, a second linear regression analysis was performed but restricted to the cohort 1 patients who were in the preterm group. For this linear regression analysis, the following additional candidate variables were also included: number of ventilator and oxygen days, the presence or absence of apnea in the 5 days before discharge, and whether infants were discharged home with a prescription to continue methylxanthine treatment. Finally, each of the 2 linear regression analyses was repeated with inclusion of the candidate variable: hours of monitor use in the first week of monitoring.

To ensure reliability of the model identified in each stepwise regression, we performed cross-validation analysis11 using data obtained in cohort 2. We first determined the r2 for each multiple regression analysis by computing the squared correlation of the observed and predicted outcomes. We then applied the coefficients from each model based on data obtained from cohort 1 to the data for the independent variables observed in cohort 2 and computed predicted values for each outcome. The correlation coefficient (r2) of these predicted values to the observed outcomes in cohort 2 is called the cross-validation correlation. The difference between this correlation and the observed r2 for the data from cohort 1 is the shrinkage of cross-validation, a rough rule of thumb used to indicate that the reliability of a model is shrinkage of less than 10%.

RESULTS
VARIABLES OBTAINED AT ENROLLMENT AND BEFORE MONITORING

Maternal, infant, family, and monitoring characteristics for each cohort are given in Table 1. Characteristics listed for cohort 1 were not significantly different from those in cohort 2 with the exception of higher parity in cohort 2. Monitoring for siblings of infants with SIDS began in the first month of life (mean ±SD, 9.0 ± 18.7 days from birth), monitoring for ALTE began within 2 weeks of the event (mean ±SD, 49.3 ± 40.8 days from birth), and monitoring for preterm infants began at discharge from the neonatal intensive care unit (mean ±SD, 50.5 ± 21.9 days from birth).

Table 1. 
Maternal and Infant Characteristics by Cohort
Maternal and Infant Characteristics by Cohort
VARIABLES ASSOCIATED WITH MONITOR USE

Patterns of monitor use by patient characteristics in the first week of monitoring and weeks 2 to 5 are given in Table 2 for cohort 1. For cohort 1, mean hours of monitor use during week 1 varied significantly depending on race, marital status, and infant groups. Mean hours of monitor use during weeks 2 to 5 varied significantly depending on race, marital status, education groups, and week 1 monitor use in hours. Table 3 gives the linear regression analysis using cohort 1, including the relation between monitor use for weeks 2 to 5 to variables obtained at enrollment (before monitoring) and the prediction for a model that includes hours of monitor use in week 1. The important finding is that variables available before monitoring was initiated were not highly associated with monitor use in weeks 2 to 5 (total model, r2 = 0.08). However, when total hours of monitor use in week 1 were added, the r2 increased to 0.64, demonstrating that week 1 monitor use was the most important factor in predicting subsequent use in weeks 2 to 5. This point is well illustrated by Figure 1A and B, which shows a scatterplot that compares the actual hours of monitor use for weeks 2 to 5 (y-axis) with the predicted hours (x-axis); the model of predicted hours is derived using only variables available before monitoring was started (Figure 1A) and includes the data from week 1 (Figure 1B).

Figure 1.
Actual monitor use vs predicted monitor use for 527 infants. A, Actual use during weeks 2 to 5 is plotted against predicted use based on a model using only data available before monitor use. B, Predicted use in weeks 2 to 5 is also based on use in week 1. The solid line represents the regression line; the dashed lines, the 95% confidence interval for the points.

Actual monitor use vs predicted monitor use for 527 infants. A, Actual use during weeks 2 to 5 is plotted against predicted use based on a model using only data available before monitor use. B, Predicted use in weeks 2 to 5 is also based on use in week 1. The solid line represents the regression line; the dashed lines, the 95% confidence interval for the points.

Table 2. 
Mean Hours of Monitor Use per Week for Cohort 1 by Maternal and Infant Characteristics
Mean Hours of Monitor Use per Week for Cohort 1 by Maternal and Infant Characteristics
Table 3. 
Multiple Linear Regression Analysis Using Cohort 1 (527 Infants) for Prediction of Mean Number of Hours of Monitor Use per Week in Weeks 2 to 5
Multiple Linear Regression Analysis Using Cohort 1 (527 Infants) for Prediction of Mean Number of Hours of Monitor Use per Week in Weeks 2 to 5

Table 4 presents a separate analysis of the preterm infants from cohort 1 to examine predictive factors for use of the monitor. Use of methylxanthines at discharge was included with the other variables provided in Table 3. Owing to missing data in 6 preterm patients, this analysis was performed on only 322 of the 328 infants from cohort 1. Results were similar to those of the entire group when examining the preterm infants with the multiple linear regression model, with an increase in r2 from 0.13 to 0.66 with the addition of week 1 hours of monitor use; use of a methylxanthine and married caregiver were associated with increased monitor use.

Table 4. 
Multiple Linear Regression Analysis Using Cohort 1 (322 Preterm Infants) for Prediction of Mean Number of Hours of Monitor Use per Week in Weeks 2 to 5
Multiple Linear Regression Analysis Using Cohort 1 (322 Preterm Infants) for Prediction of Mean Number of Hours of Monitor Use per Week in Weeks 2 to 5
VALIDATION OF PREDICTION MODEL

To assess the validity of the models derived from cohort 1, we prospectively applied the predictive model to cohort 2. When the model developed for prediction of total hours of use during weeks 2 to 5 was applied to cohort 2, we observed an r2 of 0.64 in the total analysis and 0.65 in the analysis restricted to the preterm group. The resultant shrinkage was only 0.01 from that observed in cohort 1.

COMMENT

We recognize the controversy in the utility of routine home monitoring for many infants. However, monitors are still used for specific research and clinical indications.3 If home monitoring is to be a practical tool to evaluate infants at risk for cardiorespiratory compromise, it is essential to know if patients actually use the monitor during the prescribed period. In addition, it is valuable to determine in advance which families are least likely to use the monitor and therefore most in need of additional support. The development of a monitor with a recording device satisfies the first aim, but the ability of the physician to predict who is likely to use the monitor has been far from certain. Thus, in this study we tested the hypotheses that demographic data and information related to early monitor use were predictive of caregiver adherence. From the first cohort of 527 infants, we found that monitor use in hours during the first week of monitoring was the most important predictor of monitor use in the subsequent month. We then validated this observation in another group of 248 infants. Therefore, we cannot determine a priori who will use the monitors based on factors assessed before monitor use. In contrast, however, early monitor use was associated with subsequent use.

Before the availability of home memory monitoring, assessment of monitor use had been based on caretaker report,12 which is known to be unreliable. Some more recent reports document actual use of a monitor, although none of these studies has determined what factors are predictive of that use. In a previous study by Silvestri et al,8 using home memory monitoring with transthoracic impedance and electrocardiography, the median monitor use was 15.5 hours per day among 67 infants in families who were instructed to use the monitor for all sleep time and unattended awake time. However, no factors were identified that were predictive of the extent of monitor use. Although the current study examined a population of infants who met the same enrollment criteria as this earlier study,8 some important differences limit the comparison: the population in the former study did not have a similar racial distribution (70% white), there were few noncompliant families, and the sample size was much smaller.

The strategy of our CHIME study was to document adherence with monitor use recommendations among all infants who were discharged with a home monitor, which might have resulted in much lower adherence than that reported by others. In 2 prior studies,7,9 for example, some nonadherent families may have been eliminated before the study group was analyzed; these selection biases would have the effect of increasing the mean time for monitor use of the remaining infants who were described in the respective studies. Gibson et al9 reported data from 114 infants on home memory monitoring. Parents were instructed to use the monitor all the time, and mean monitor use was again 15 hours per day. However, approximately 20% of the data downloads were not included for analysis. In a study by Cote et al,7 193 patients were discharged home with a memory monitor because of an ALTE: persistent apnea and bradycardia in preterm infants, history of SIDS in a sibling, or parental anxiety. Forty-six patients were eliminated from analysis because they had a “nonrecording monitor for some part of or the whole recording period,”7(p785) which would have raised the mean adherence of the remaining group; for the 147 infants whose data were reported, 81% of parents used the monitor for at least 12 hours per day, and 53% of parents used the monitor for at least part of each day of recommended use. Although the findings provided by these 2 studies provide a valuable contrast, in neither study were the data analyzed to predict which caregivers used the monitor; moreover, there were not sufficient data presented to compare sociodemographic characteristics with our patients and determine whether this was a factor that contributed to differences in adherence. Important factors that could have caused our reported compliance to be lower than previous reports include the large broad-based patient population, the use of a more complex monitor, and inclusion of all patients in the data analysis, including those with early attrition. Despite the complex nature of the CHIME study and monitor, however, it is noteworthy that nearly 70% of the families were using this monitor in the home for at least 10 hours per week and 30% for at least 70 hours per week.

The utility of our findings and the potential to generalize our interpretation are substantiated further by the consistency with other studies that predict adherence to use of medical regimens or devices.13 Perhaps most relevant to our study is the reported experience with home continuous positive airway pressure treatment for sleep apnea in adults; Weaver et al14 studied 32 patients, half of whom applied positive airway pressure for more than 90% of the nights for a mean of 6.2 hours, whereas the other half used it only intermittently and averaged 3.4 hours on the nights when it was applied. Adherence with treatment recommendations during several months was best predicted by adherence during the first week, whereas no other factors were reliable predictors.

Regardless of the complexity of the CHIME monitor and specific differences in adherence compared with other reports, our observation that adherence to monitor use recommendations 1 month after initiation of use is predicted only by adherence in the first week is a valuable finding and consistent with other studies of adults who use diverse treatments. Our data should provide a reference for the minimum extent of adherence that can be expected with other and perhaps less complicated home-recording devices in a large population with diverse sociodemographic characteristics. Although there is no proof that knowledge of adherence will permit successful intervention, our findings demonstrate that adherence with monitor use recommendations3 should be assessed very early. This conclusion may also be applicable for adherence with use of other medical technological devices at home.

Article
The CHIME Study Group Participants

CLINICAL SITES

Department of Pediatrics, Case Western Reserve University School of Medicine, Cleveland, Ohio: MetroHealth Medical Center: Terry M. Baird, MD* (currently at Rainbow Babies and Children’s Hospital); Rainbow Babies and Children’s Hospital: Richard J. Martin, MD; Lee J. Brooks, MD (currently at The Children’s Hospital of Philadelphia, Philadelphia, Pa); Roberta O’Bell, RN.** Department of Pediatrics, Medical College of Ohio, Mercy Children’s Hospital, and Children’s Medical Center, Toledo: Carl E. Hunt, MD* (currently at National Center on Sleep Disorders Research, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md); David R. Hufford, MD; Mary Ann Oess, RN.** Department of Pediatrics, Division of Respiratory Medicine, Rush Medical College of Rush University, Chicago, Ill: Rush Children’s Hospital at Rush-Presbyterian-St Luke’s Medical Center: Debra E. Weese-Mayer, MD*; Jean M. Silvestri, MD; Sheilah M. Smok-Pearsall, RN.** Department of Pediatrics, John A. Burns School of Medicine, University of Hawaii at Manoa, Honolulu: Kapiolani Medical Center for Women and Children: David H. Crowell, PhD*; Larry Tinsley, MD (currently at Pediatric Critical Care Medical Group, Encino, Calif); Linda E. Kapuniai, DrPH.** Department of Pediatrics and Neonatology, University of Southern California School of Medicine, Los Angeles: Los Angeles County and University of Southern California Medical Center, Women’s and Children’s Hospital, Los Angeles: Toke T. Hoppenbrouwers, PhD*; Rangasamy Ramanathan, MD; Paula H. Palmer, PhD.** Children’s Hospital of Los Angeles, Los Angeles: Thomas G. Keens, MD; Sally L. Davidson Ward, MD; Daisy B. Bolduc, BA, technical coordinator.

CLINICAL TRIALS OPERATIONS CENTER

Department of Obstetrics and Gynecology, Case Western Reserve University School of Medicine, MetroHealth Medical Center, Cleveland, Ohio: Michael R. Neuman, MD, PhD* (currently at the University of Memphis, Memphis, Tenn); Rebecca S. Mendenhall, MS.**

DATA COORDINATING AND ANALYSIS CENTER

Departments of Pediatrics and Epidemiology and Biostatistics, Boston University Schools of Medicine and Public Health, Boston, Mass: Michael J. Corwin, MD*; Theodore Colton, ScD; Sharon M. Bak, MPH**; Mark Peucker, technical coordinator; Howard Golub, MD, PhD, physiologic data biostatistician; Susan C. Schafer, RNC, MS, clinical trials coordinator.

STEERING COMMITTEE CHAIRMAN

Formerly Department of Pediatrics, Yale University School of Medicine, New Haven, Conn; currently Department of Pediatrics, The University of Texas Southwestern Medical School, Dallas: George Lister, MD.

NATIONAL INSTITUTES OF HEALTH

Pregnancy and Perinatology Branch, Center for Research for Mothers and Children, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Md: Marian Willinger, PhD.

*Principal investigator.

**Study coordinator.

Back to top
Article Information

Correspondence: George Lister, MD, Department of Pediatrics, The University of Texas Southwestern Medical School, 5323 Harry Hines Blvd, Dallas, TX 75390-9063 (george.lister@utsouthwestern.edu).

Accepted for Publication: April 14, 2004.

Funding/Support: The CHIME study is supported by grants 29067, 29071, 28971, 29073, 29060, 29056, and 34625 from the National Institute of Child Health and Human Development, Bethesda, Md.

References
1.
 National Institutes of Health Consensus Development Conference on Infantile Apnea and Home Monitoring, Sept 29 to Oct 1, 1986. Pediatrics 1987;79292- 299
PubMed
2.
Ramanathan  RCorwin  MJHunt  CE  et al. Collaborative Home Infant Monitoring Evaluation (CHIME) Study Group, Cardiorespiratory events recorded on home monitors: comparison of healthy infants with those at increased risk for SIDS. JAMA 2001;2852199- 2207
PubMedArticle
3.
Committee on Fetus and Newborn; American Academy of Pediatrics, Apnea, sudden infant death syndrome, and home monitoring. Pediatrics 2003;111914- 917
PubMedArticle
4.
Weese-Mayer  DEBrouillette  RTMorrow  ASConway  LPKlemka-Walden  LMHunt  CE Assessing validity of infant monitor alarms with event recording. J Pediatr 1989;115702- 708
PubMedArticle
5.
Weese-Mayer  DEMorrow  ASConway  LPBrouillette  RTSilvestri  JM Assessing clinical significance of apnea exceeding fifteen seconds with event recording. J Pediatr 1990;117568- 574
PubMedArticle
6.
Nathanson  IO’Donnell  JCommins  MF Cardiorespiratory patterns during alarms in infants using apnea/bradycardia monitors. AJDC 1989;143476- 480
PubMed
7.
Cote  AHum  CBrouillette  RTThemens  M Frequency and timing of recurrent events in infants using home cardiorespiratory monitors. J Pediatr 1998;132783- 789
PubMedArticle
8.
Silvestri  JMHufford  DR  JrDurham  J  et al.  Assessment of compliance with home cardiorespiratory monitoring in infants at risk of sudden infant death syndrome: Collaborative Home Infant Monitoring Evaluation (CHIME). J Pediatr 1995;127384- 388
PubMedArticle
9.
Gibson  ESpinner  SCullen  JAWrobel  HASpitzer  AR Documented home apnea monitoring: effect on compliance, duration of monitoring, and validation of alarm reporting. Clin Pediatr (Phila) 1996;35505- 513
PubMedArticle
10.
Neuman  MRMendenhall  RSZoldak  JT  et al. Collaborative Home Infant Monitor Evaluation (CHIME) Study Group, Cardiopulmonary monitoring at home: the CHIME monitor. Physiol Meas 2001;22267- 286
PubMedArticle
11.
Kleinbaum  DGKupper  LLMuller  KE Applied Regression Analysis and Other Multivariate Methods.  Boston, Mass PWS-Kent1988;
12.
Ahmann  EMeny  RGFink  RJ Use of home apnea monitors. J Obstet Gynecol Neonatal Nurs 1992;21394- 399
PubMedArticle
13.
Fraser  SNSpink  KS Examining the role of social support and group cohesion in exercise compliance. J Behav Med 2002;25233- 249
PubMedArticle
14.
Weaver  TEKribbs  NBPack  AI  et al.  Night-to-night variability in CPAP use over the first three months of treatment. Sleep 1997;20278- 283
PubMed
×