Loss to Follow-Up in Early Hearing Detection and Intervention

Loss to Follow-Up

Many jurisdictions recognize any baby who did not receive or complete the recommended diagnostic or intervention process as lost to follow up (LTF). ^[1] However, perhaps the most obvious difficulty encountered in tracking such infants stems from the variability in how the term is defined from site to site, as well as on the state and national levels. In an effort to obtain a more comprehensive view of operational definitions used by states, an informal review was undertaken in February 2008 of 23 state definitions ( Hoffman & Farrell, 2008). The durational dimension of the definition varied among jurisdictions. For example, some states designated a child as LTF because of incorrect, incomplete, or expired contact information if not located within 2 months of age, while others attempted to follow a child until 3 years of age before classifying as LTF. Other jurisdictions, rather than using age of the child, declared a child as LTF according to the time transpired from the last EHDI system activity with the family—for example, 3–6 months with no known follow-up activity resulted in a designation of LTF. Inclusion of specific groups, such as babies who have expired, moved to another state, or whose parents refused the screening, in the LTF category varied among the states reporting.

In 2006, in an effort to provide more consistent reporting among EHDI jurisdictions, the CDC launched a national survey to better define the timelines for EHDI milestones based on recommendations of the Directors of Speech and Hearing Programs in State Health and Welfare Agencies (2004) and the CDC EHDI Data Committee ( CDC, 2008). This survey, designed to report data for Healthy People 2010 and the EHDI 1-3-6 goals on a national basis, requested data that could be documented by the EHDI jurisdictions and might differ from that which is defined and collected specific to their locale and reporting needs (e.g., adjusted or corrected age for preterm infants).

In that CDC survey, “no documented diagnosis” included infants (1) who are in process, (2) who died, (3) whose parents declined services, (4) who are nonresidents, (5) who moved out of the jurisdiction, (6) who cannot be contacted, (7) whose parents are unresponsive, and (8) who have no documented diagnosis for unknown reasons.

Clearly, there is not yet general consensus on how we record and report infants as LTF. This lack of consensus results in difficulty comparing old data to new data and comparing data across jurisdictions.

Return to Top

Loss to Documentation

The designation of lost to documentation (LTD) is becoming increasingly accepted as being distinct from LTF ( Mason, Gaffney, Green, & Grosse, 2008). LTD includes those infants who did not pass their hearing screening and whose diagnostic or intervention status has not been reported to the EHDI program; thus, their status remains unknown by the EHDI program despite the fact that they may have received services. There are multiple points in the EHDI process at which a child might become LTD (e.g., following screening, following diagnosis, following referral to intervention). Although the factors contributing to LTD in NHS programs have not been thoroughly delineated at this time, factors identified as contributing to LTD in the parallel newborn dried bloodspot screening programs included, primarily, name changes and, to a lesser degree, data mismatches. In addition, inadequate or erroneous parental contact information and/or unknown or changed primary care provider contribute to LTD ( Hoff et al., 2006). An additional factor contributing to LTD includes those infants who receive services in a jurisdiction different from the jurisdiction of the EHDI system in which they reside or in which they were born. These are often termed “border babies.”

Return to Top

Loss to System

Combining the designations LTF and LTD under the broad heading of lost to system (LTS) has been suggested as a viable option to reduce the confusion and variability in capturing this data set within and across EHDI systems and to increase the effectiveness of follow-up strategies ( Beauchaine & Hoffman, 2008).

Systems Issues

Return to Top

Family Issues

Possible family factors contributing to LTF examined by studies included in the ASHA systematic review of evidence (see the Appendix) addressed educational and literacy levels of mothers and the age of mothers, maternal marital status, maternal substance abuse, maternal smoking, number of children at home, insurance status, family history of hearing loss, receipt of prenatal care, and poverty level ( Folsom et al., 2000; Liu, Farrell, MacNeil, Stone, & Barfield, 2008; Oghalai, Chen, Brennan, Tonini, & Manolidis, 2002; Prince, Miyashiro, Weirather, & Heu, 2003). Child factors that might contribute to LTF that were examined by studies included in the review were severity of hearing loss, birth weight, race, gender, resident of neonatal intensive care or well-baby nursery, and neonatal surgery and ventilator status ( Davis & Wood, 1992; Folsom et al., 2000; Liu et al., 2008; Prieve et al., 2000; Prince et al., 2003; Shoup, Owen, Jackson, & Laptook, 2005; Stein, Jabaley, Spitz, Stoakley, & McGee, 1990; Stewart et al., 2000; Uus & Bamford, 2006).

Factors that were not examined by studies included in the systematic review but that have been considered in other documents include the following:

• Proximity to resources: In a survey of approximately 1,000 families whose babies received NHS in Massachusetts, those families who were compliant with follow-up revealed that a long travel time to their appointments could be a barrier to follow-up. Related to this concern was that the appointments might not be convenient and the time parents spent away from work for these appointments could be problematic ( MacNeil et al., 2007).

• Difficulty scheduling appointments: At least two surveys have suggested that difficulties scheduling appointments for follow-up have posed barriers for families ( Harrison, Roush, & Wallace, 2003; MacNeil et al., 2007). The difficulties included making the telephone calls, finding someone to care for other children, and transportation.

Return to Top

Quality Assurance

The CDC emphasizes the importance of an EHDI tracking and surveillance system that can minimize LTF ( CDC, 2003). Specifically, they set forth the following program objectives:

• Comprehensive system: Each state will have a computerized system that (1) maintains current information on all babies screened, including those who do not pass the screen, (2) contains diagnostic results for babies who were referred based on the results of their NHS test, and (3) documents interventions for those infants who were diagnosed with hearing loss.

• Policies and procedures: Each state will have written policies and procedures regarding their EHDI tracking and surveillance system.

• Privacy and confidentiality: Each state will develop policies, procedures, and informed-consent requirements regarding privacy and confidentiality of data in the EHDI tracking and surveillance system.

• Inclusion of all births: Each state will ensure that all live births in the state are included in the state EHDI tracking and surveillance system by matching with the state's birth certificate registry as allowed.

• Risk factors for hearing loss: The state EHDI tracking and surveillance system will ascertain risk factors for hearing loss for every infant by linkage with other state data systems.

• NHS results: The EHDI tracking and surveillance system will capture all hearing screening results at the birthing hospital within 1 week of discharge or transfer.

• Reporting mechanism for health care providers: Each state will provide a mechanism for hospitals, audiologists, and other health care providers to report NHS results, evaluations, and interventions.

• Identifying children who need screening and follow up: The state EHDI tracking and surveillance system will be able to identify, on a weekly basis, all infants and children who need initial hearing screening, rescreening, evaluation, follow-up, or intervention.

• Access to information: The state EHDI tracking and surveillance system will allow case managers and authorized health care providers to access relevant information about infants and children.

Methodology

A systematic search of the literature was conducted from September 2007 to January 2008. Studies were considered for the review if they were published in a peer-reviewed journal from 1990 to 2008, were written in English, and contained original data addressing one or more of the four clinical questions. Twenty-two electronic databases were searched in addition to all ASHA journals, the EHDI Publications Database, the National Center on Birth Defects and Developmental Disabilities Publications Database, and a full search of relevant references and authors. Databases searched included CINAHL, Cochrane Library, CRD database, DIMDI, EBM Guidelines, Education Abstracts, EMBASE, ERIC, Health Source: Nursing/Academic Edition, Linguistics and Language Behavior Abstracts, Medline, National Electronic Library for Health, National Rehabilitation Information Center, PsycARTICLES, PsycBITE, Psychology and Behavioral Sciences Collection, PubMed, Science Citation Index, Social Science Citation Index, SUMSearch, Teacher Reference Center, and TRIP Database. A total of 24 expanded search terms were used to review the UNHS and EHDI literature, including key words related to newborn hearing screening, early intervention, and hearing screening follow-up.

The search schematized in Figure 1 includes a total of 238 articles. Two authors (RM and TF), blind to one another's results, reviewed each abstract and identified 26 articles as meeting the initial inclusion criteria, with 91% agreement. Of those initially accepted, 12 were subsequently excluded after review of the full text article. A total of 14 studies were included in the final analysis.

Studies pertaining to intervention were evaluated for methodological quality. The authors (RM and TF) independently appraised each study on eight quality indicators:

• adequate description of protocol for replication

• adequate description of subjects (within-subject design) or groups comparable at baseline (between-subjects design)

• assessors blinded

• random sample adequately described

• evidence of treatment fidelity

• p value reported or calculable

• effect size and confidence interval reported or calculable

• analyzed by intention to treat (controlled trials only).

Reliability of appraisal ratings between authors was assessed, and any discrepancies were resolved via consensus. Studies received one point for each quality marker that met the above criteria. Studies incorporating controlled trials could obtain a maximum quality score of eight. All other study designs where intention to treat analysis was not applicable could obtain a maximum quality score of seven.

Return to Top

Results

Question 1. What characteristics of individuals, families, or populations have been found to be associated with increased risk of delays or loss to follow-up for children with a positive initial screen for hearing loss and in need of an audiologic evaluation by 3 months of age?

The 11 studies included in this section of the review reported findings on a total of 19 different risk factors. As shown in Table 1, 6 of the studies ( Davis & Wood, 1992; Prieve et al., 2000; Shoup, Owen, Jackson, & Laptook, 2005; Stein, Jabaley, Spitz, Stoakley, & McGee, 1990; Stewart et al., 2000; Uus & Bamford, 2006) compared follow-up in infants who had been in neonatal intensive care units (NICUs) with infants who had been exclusively in well-baby nurseries (WBNs).

In four of the six studies ( Prieve et al., 2000; Shoup et al., 2000; Stewart et al., 2000; Uus & Bamford, 2006), findings were reported in terms of rates of follow-up during the study period (see Table 2). Prieve et al. found a significantly lower rate of follow-up in WBN infants, whereas in the other three studies no statistically significant differences were found between the populations. Interpretation of the findings across studies is difficult because none of the studies specified the time frame or cut-off for designating an infant as lost to follow-up. For example, in the Uus and Bamford study data are presented for the infant's age on the date of the evaluation, and the ages ranged up to 31 weeks. Data on age are not reported in the other studies, and one cannot tell whether an infant evaluated at 31 weeks would have been within or outside the parameters for designation as followed-up or lost in other studies.

In addition to Uus and Bamford (2006), two other studies ( Davis & Wood, 1992; Stein et al., 1990) reported data on the infant's age at evaluation (see Table 3). In those two studies, the authors found that the age at evaluation for WBN infants was significantly higher than for NICU infants, whereas Uus and Bamford found a statistically significant difference in the opposite direction.

The Folsom et al. (2000); Prince, Miyashiro, Weirather, and Heu (2003); and Liu, Farrell, MacNeil, Stone, and Barfield (2008) studies reported data on maternal age. Folsom et al. found a mean maternal age of 27.5 years among mothers whose infants were followed up, compared to 26.0 years among mothers of infants lost to follow-up, a statistically significant difference. Liu et al. found significantly lower follow-up in mothers aged younger than 20. In contrast, Prince et al. (2003) found the lowest rates of follow-up among infants whose mothers were age 35 or older, but none of the differences in that study's follow-up rates by maternal age were statistically significant.

The same three studies also reported data on the infants' race, again with conflicting results. Folsom et al. (2000) found significantly higher rates of follow-up in non-Hispanic White infants (64.8%) than in infants of other racial/ethnic groups (51.5%). Liu et al. (2008) also found significantly higher follow-up in non-Hispanic White infants (92% vs. 85%). Prince et al. (2003) found significantly lower rates of follow-up in White infants (75.9%) than in infants of other races (85.6%).

The Davis and Wood (1992) and Uus and Bamford (2006) studies both looked at follow-up by severity of hearing loss. Davis and Wood found statistically significant differences in the mean age at evaluation: 9 months for infants with a hearing loss of 80+ dB, 17 months for hearing loss between 50 db and 79 dB, and 35 months for <50 dB. Uus and Bamford found no significant differences by degree of loss.

Folsom et al. (2000) and Liu et al. (2008) both looked at follow-up by insurance status. Folsom et al. found significantly higher follow-up in infants with health insurance (60.7%) than in those with no insurance (42.4%). Liu et al. compared private insurance, public insurance, and no insurance, and found that follow-up of infants with private insurance (94% follow-up) was similar to that of infants with no insurance (92% follow-up), and both were significantly higher than in infants with public health insurance (81% follow-up).

The Prince et al. (2003) and Liu et al. (2008) studies both looked at follow-up by birth weight. Prince et al. found significantly higher follow-up in infants weighing at least 2,500 g at birth (83.6%) than in infants weighing less than 2,500 g (72.0%). The Liu et al. study found no statistically significant differences by birth weight.

The Prince et al. (2003) and Liu et al. (2008) studies also looked at follow-up by mother's marital status. Prince et al. found no significant difference in follow-up by marital status, whereas Liu et al. found significantly higher follow-up in married versus unmarried mothers (93% vs. 83%).

Finally, the same two studies looked at follow-up by maternal education. Both studies found significantly higher follow-up in mothers who had completed high school or beyond. The Prince et al. (2003) study found 84.1% follow-up in high school graduates compared to 74.5% in nongraduates. The rates reported in the Liu et al. (2003) study were 92% in graduates and 79% in nongraduates.

Data on 12 other risk factors were included in only one study each. Delayed or no follow-up was found to be statistically significant in families with the following characteristics:

• more than two children at home ( Folsom et al., 2000)

• mother had no or late prenatal care ( Folsom et al., 2000)

• mother was a substance abuser (type of substance not specified; Folsom et al., 2000)

• mother smoked during pregnancy ( Liu et al., 2008)

• infant underwent surgery during the neonatal period ( Folsom et al., 2000)

• infant had not been on a ventilator ( Folsom et al., 2000)

• no family history of hearing loss ( Folsom et al., 2000)

• infant was born in a community hospital rather than an obstetric Level 2 or 3 medical center ( Prince et al., 2003).

Single studies reported no significant differences in follow-up associated with these factors:

• infant's risk for hearing loss ( Parving, 1993)

• laterality of hearing referral ( Liu et al., 2008)

• gender ( Prince et al., 2003)

• poverty ( Oghalai, Chen, Brennan, Tonini, & Manolidis, 2002).

Question 2. What characteristics of individuals, families, or populations have been found to be associated with increased risk of delays or loss to follow-up for children with a positive diagnosis of hearing loss and in need of an audiologic intervention by 6 months of age?

Three studies reported data on the NICU versus WBN comparison (see Table 4). The Davis and Wood (1992) and Stein et al. (1990) studies both found 100% follow-up from evaluation to hearing aid fitting in both groups, and both studies found that the significant differences in mean age that had been observed at the time of diagnosis (see Table 3) were no longer evident at the time of habilitation. Davis and Wood reported a mean age at habilitation of 19.7 months in NICU infants versus 25.0 months in WBN infants. Stein et al. reported means of 20.0 and 20.8 months, respectively. Uus and Bamford (2006) reported a mean age of 25.7 weeks in NICU infants and 19.8 weeks in WBN infants (statistical significance was not reported).

Davis and Wood (1992) also reported findings on the relationship between the severity of hearing loss and follow-up. Here, the significant differences in follow-up observed at the time of evaluation continued to be reflected in age at time of habilitation. Mean age was 12 months in children with hearing loss of 80+ dB, 25 months with hearing loss between 50 dB and 79 dB, and 37 months with hearing loss of <50 dB. However, the gap between diagnosis and habilitation was not significantly different by severity of hearing loss. Uus and Bamford (2006) also reported findings of delayed habilitation in moderate, compared with profound, hearing loss, although data were not presented.

The Oghalai et al. (2002) study, which found no impact of poverty on age at diagnosis, found that infants born at a hospital serving a primarily indigent population had a significantly later mean age at time of habilitation (49.6 weeks vs. 28.2 weeks; p = .041).

The Liu et al. (2008) study reported findings on the same nine variables it addressed in Question 1. Loss to follow-up from diagnosis to early intervention services was significantly higher among the following groups:

• infants weighing <2,500 g at birth

• infants with unilateral hearing loss

• infants with mild/moderate hearing loss.

The following factors did not reflect significantly different rates of follow-up in the Liu et al. study:

• maternal age

• race

• mother's marital status

• mother smoked during pregnancy

• mother's education

• insurance status

Question 3. What has been the effectiveness of initiatives designed to increase follow-up for appropriate audiologic evaluation by 3 months of age?

No studies were found related to this question.

Question 4. What has been the effectiveness of initiatives designed to increase follow-up for appropriate audiologic intervention by 6 months of age?

No studies were found related to this question.

Because no intervention studies directly relevant to Questions 3 or 4 were found, we made the decision to amend the inclusion criteria to include studies of interventions designed to reduce loss to follow-up from initial to repeat screen. We thought that the issues might be sufficiently closely related that the findings from those studies could have implications for follow-up from screening to evaluation.

Three studies ( Françozo, Fernandes, Marconi Pinheiro Lima, & Rossi, 2007; Korres et al., 2005; Wittmann-Price & Pope, 2002) investigated the effectiveness of interventions to improve follow-up rates from initial to repeat screens. The three studies (see Table 5) examined three different types of interventions.

Françozo et al. (2007) compared the effectiveness of small-group versus individual counseling of mothers by a social worker. The results were not statistically significant. In the small-group cohort, follow-up was 60.2% (95% confidence interval [CI] = 52.5%–67.6%), and in the individual counseling cohort, follow-up was 74.3% (95% CI = 57.8%–86.0%).

The Korres et al. (2006) study was the only one of the three to report statistically significant findings. In one cohort, parents received a written report of the screening results and recommendation for rescreening. That cohort had follow-up of 27.8% (95% CI = 24.6%–31.3%). The second cohort received, in addition to the written report, written comments in their discharge papers, personal contact, and monitoring of their compliance with rescreening appointments. Follow-up in that cohort was 41.8% (95% CI = 37.7%–46.0%). The changes to the protocol were bundled, and no data on the relative impact of the three individual changes were reported.

The Whittman-Price and Pope (2002) study examined the effectiveness of adding a 20-minute educational module on UNHS to a hospital-based prenatal class. In the cohort that did not receive the module, 7 of the 9 infants were rescreened. It was not reported whether the parents of those children attended the prenatal class before the addition of the UNHS module, whether they attended a different class, or whether they did not attend a class at all. In the cohort that did receive the module, all 5 infants were rescreened. These results were not statistically significant.

Return to Top

Culturally and Linguistically Diverse Populations

Questions 1 and 2, that is, those that dealt specifically with the characteristics of individuals or populations, were developed in part in anticipation that cultural and linguistic factors might be an important influence on follow-up. These factors, including educational level, poverty, race, and others were addressed by a small number of studies, and with little consistency among results. Table 6 describes the context in which these studies were conducted.

Table 6. Patient populations by study.

Nine of the 14 studies were conducted in the United States. In only one instance would separate consideration of the findings result in somewhat different conclusions. Of the six studies comparing follow-up from screening to evaluation in NICU versus WBN infants, five were conducted in the United States and one in England. The English study found better follow-up in WBN infants, whereas of the five U.S. studies, three found better follow-up in NICU infants and two found no differences. This raises at least the possibility that the contrasting findings between these studies could have been influenced by the different health care systems.

The diversity of study populations may have also influenced another aspect of the overall findings. As noted under Question 1, the three studies examining the rates of follow-up by infant race reached different conclusions. The Folsom et al. (2000) and Liu et al. (2008) studies found higher follow-up in non-Hispanic White infants, whereas the Prince et al. (2003) study found significantly lower follow-up in non-Hispanic Whites. There is some indication, though, that the non-Hispanic White populations may not have been comparable across the three studies. Although all three studies were conducted in the United States, Prince et al. pointed out that “a small percentage of White women [in the outer islands of Hawaii] may choose ‘alternative lifestyles’ preferring nontraditional health care rather than traditional services” (p. 1205).

Return to Top

Discussion

To affect clinical practice and/or policy, it is desirable to have an evidence base of multiple studies of high methodological quality with generally consistent findings. The available evidence on issues surrounding loss to follow-up in newborn hearing screening programs fails on all three counts.

Return to Top

Return to Top

References

References marked with an asterisk indicate studies included in the systematic review.

American Speech-Language-Hearing Association. (2007). Year 2007 position statement: Principles and guidelines for early detection and intervention programs. Available from www.asha.org/policy/.

Centers for Disease Control and Prevention. (2005). Final summary of 2005 national EHDI data (Version 6). Retrieved from www.cdc.gov/ncbddd/ehdi/documents/Nat_EHDI_Summ_2005_Web_v6.pdf.

*Davis, A., & Wood, S. (1992). The epidemiology of childhood hearing impairment: Factor relevant to planning of services. British Journal of Audiology, 26, 77–90.

*Folsom, R. C., Widen, J. E., Vohr, B. R., Cone-Wesson, B., Gorga, M. P., Sininger, Y. S., & Norton, S. J. (2000). Identification of neonatal hearing impairment: Recruitment and follow-up. Ear and Hearing, 21, 462–470.

*Françozo, M., Fernandes, J., Marconi Pinheiro Lima, C. M., & Rossi, T. (2007). Improvement of return rates in a neonatal hearing screening program: The contribution of social work. Journal of Health Care Social Work, 44, 179–190.

Joint Committee on Infant Hearing. (2000). Year 2000 position statement. Principles and guidelines for early hearing detection and intervention programs. Pediatrics, 106, 798–817.

*Korres, S., Balatsouras, D. G., Vlachou, S., Kastanioudakis, I. G., Ziavra, N. V., & Ferekidis, E. (2005). Overcoming difficulties in implementing a universal newborn hearing screening program. Turkish Journal of Pediatrics, 47, 203–212.

*Lui, C., Farrell, J., MacNeil, J., Stone, S., & Barfield, W. (2008). Evaluating loss to follow-up in newborn hearing screening in Massachusetts. Pediatrics, 121, e335–e343.

National Center for Hearing Assessment and Management. (2008). EHDI legislation. Available from www.infanthearing.org/legislative/index.html.

*Oghalai, J. S., Chen, L., Brennan, M. L., Tonini, R., & Manolidis, S. (2002). Neonatal hearing loss in the indigent. Laryngoscope, 112, 281–286.

*Parving, A. (1993). Congenital hearing disability—epidemiology and identification: A comparison between two health authority districts. International Journal of Pediatric Otorhinolaryngology, 27, 29–46.

*Prieve, B., Dalzell, L., Berg, A., Bradley, M., Cacace, A., Campbell, D., et al. (2000). The New York State universal newborn hearing screening demonstration project: Outpatient outcome measures. Ear and Hearing, 21, 104–117.

*Prince, C. B., Miyashiro, L., Weirather, Y., & Heu, P. (2003). Epidemiology of early hearing loss detection in Hawaii. Pediatrics, 111, 1202–1206.

*Shoup, A., Owen, K., Jackson, G., & Laptook, A. (2005). The Parkland Memorial Hospital experience in ensuring compliance with universal newborn hearing screening follow-up. Journal of Pediatrics, 146, 66–72.

*Stein, L. K., Jabaley, T., Spitz, R., Stoakley, D., & McGee, T. (1990). The hearing-impaired infant: Patterns of identification and habilitation revisited. Ear and Hearing, 11, 201–205.

*Stewart, D. L., Mehl, A., Hall, J. W., Thomson, V., Carroll, M., & Hamlett, J. (2000). Universal newborn hearing screening with automated auditory brainstem response: A multi-site investigation. Journal of Perinatology, 20, S128–S131.

U.S. Department of Health and Human Services. (2000). Healthy People 2010. Washington, DC: U.S. Public Health Service.

*Uus, K., & Bamford, J. (2006). Effectiveness of population-based newborn hearing screening in England: Ages of interventions and profile of cases. Pediatrics, 117, e887–e893.

*Wittmann-Price, R., & Pope, K. (2002). Universal newborn hearing screening: One hospital's project improved follow-up rates for infants who didn't pass the initial test. American Journal of Nursing, 102, 71–77.

Yoshinaga-Itano, C., Sedey, A., Coulter, D. K., & Mehl, A. L. (1998). Language of early and later identified children with hearing loss. Pediatrics, 102, 1161–1171.

Return to Top

Figures and Tables

Return to Top

Appendix A: Quality Scores for Intervention Studies

Figure 8.