A Psychometric Study of the Modified Bridges for Newborns Screening Tool

Welcome Baby, a Los Angeles County-based program that First 5 Los Angeles (First 5 LA) sponsors, provides free services during pregnancy and through a child's first nine months. It currently serves 14 hospitals and approximately 9 percent of families living in high-risk Los Angeles communities.

To identify mothers in need of more-focused support, First 5 LA employs hospital liaisons who administer a maternal risk assessment tool, the Modified Bridges for Newborns screening tool, during bedside, postpartum interviews of mothers. The Modified Bridges consists of 26 questions, which receive weights that correspond to the severity of risk that each question assesses, across three subscales (Medical, Psychosocial, and Demographics/Basic Needs).

First 5 LA requested that the RAND Corporation undertake a psychometric study of the Modified Bridges to assess whether the tool is accurately assessing risk among mothers in the participating hospitals. The research reported here aimed to provide information relating to both the reliability and validity of the Modified Bridges and whether the hospital liaisons can accurately and precisely score patients via the Modified Bridges. In particular, we examined the following research questions at the request of First 5 LA:

1. Do hospital liaisons assign similar Modified Bridges scores (overall and subscores) to mock patient interviews?
2. Does the Modified Bridges identify a similar level of risk among the same women as other well-validated measures of maternal/infant risk?
3. Does the Modified Bridges accurately distinguish between low-/moderate-risk women and high-risk women?

Findings

1. Do Hospital Liaisons Assign Similar Scores When Using the Modified Bridges?
   • Liaisons were found to provide similar risk scores (inter-rater reliability), and their assessments were similar over time (test-retest reliability).
     • In particular, liaison inter-rater reliabilities are highest for the Demographics/Basic Needs subscale and somewhat lower for the Medical subscale.
   • Overall, liaisons perform at acceptable levels. However, experience as a liaison and use of liaison training materials, such as the Modified Bridges protocol, improved performance.
2. Does the Modified Bridges Identify a Similar Level of Risk Among the Same Women as Other Measures of Psychosocial and/or Medical Risk?
   • The Modified Bridges is most comparable to maternal risk assessments sponsored by the Florida Department of Health, which assess similar content to that of the Modified Bridges.
   • Given the current scoring protocol, the Modified Bridges might overidentify mothers as being high risk.
   • The Modified Bridges' ability to distinguish between high and low/moderate levels of risk would improve if a higher risk threshold were implemented.
3. Does the Modified Bridges Accurately Distinguish Between Low-/Moderate-Risk Mothers and High-Risk Mothers?
   • Because nearly half of mothers qualify as being at risk, the current risk threshold score of 50 cannot distinguish between the highest levels of risk.
   • The current reliability of the Modified Bridges indicates, in practical terms, that First 5 LA can have only limited confidence that the given risk score for a mother reflects that mother's level of risk.
     • The current scoring protocol and item weights used for the Modified Bridges have a moderate level of reliability (0.65), below the widely used reliability threshold of 0.70, while scoring the tool without the current item weights would improve its reliability to a more acceptable level (0.74).
   • Given the lower reliability of the Modified Bridges, we considered potential revisions to the item weights in order to improve the Modified Bridges' ability to distinguish between high- and low-risk mothers.
     • Given the relationship between the item weight and items' ability to distinguish between risk levels, we propose increasing the weight of seven items and decreasing the weight of three items.
     • In addition to these changes, we recommend removing two items: infant medical problems, which is redundant with another item on the Modified Bridges that performs better, and mother's use of English, which does not distinguish between maternal risk levels.
   • Implementing these changes would improve the Modified Bridges' ability to distinguish between risk levels and increase the tool's reliability from 0.65 to 0.73 (standardized reliability = 0.75).