Some Say Goodbye and Some Disappear: Patterns of Service Departure in Long-Term Home Visiting

Background and Purpose: Most evidence-based home visiting (HV) models are designed to support families from a woman’s pregnancy through her child’s second or third birthday. Yet programs often struggle to retain families for this long. Previous studies have found caregiver age, community resources, and high caseloads associated with early departure, but findings related to other demographic and risk factors have been inconsistent. Complicating matters, early program exits happen for different reasons, but these reasons have largely been ignored in HV retention studies. Therefore, this study focuses on two types of early leavers: those who withdraw from services and those who “disappear,” i.e.,  who persistently no-show or are unable to be located. The study aims to answer two research questions:

1. Does service exit timing differ between clients who withdraw and those who disappear?

2. Do the demographic, risk and service factors associated with early exit differ between clients who withdraw and those who disappear?

Methods: The study uses a sample of 836 low-income women served by 138 home visitors in 26 HV programs across geographically diverse regions in Wisconsin. These programs implement evidence-based HV models that are designed to serve families for two or more years. Client demographic and risk indicators included participant age, race/ethnicity, education, employment, cohabitation/marriage, primiparous status, prenatal enrollment, prenatal smoking, recent drug use, and transportation challenges. Measures of service experience included completed visit rate for the first three months of service, home visitor’s average caseload during client’s service episode, and home visitor’s job satisfaction. Research question 1 employed a Cox regression model comparing hazard rates of clients who withdrew and disappeared from services. Research question 2 used a random-intercept multinomial regression model to compare participants who stayed to those who either disappeared or withdrew from services. Participants were nested within home visitors. Demographic, risk, and service measures were entered as fixed effects.

Results:  Results from the Cox regression suggest no significant differences in retention survival curves between women who disappeared and those who withdrew from services. Multinomial regression results indicated that first-time mothers were consistently more likely to exit early regardless of departure reason (withdraw OR= 3.16; disappear OR = 1.60). However, women who withdrew from services were significantly more likely to have home visitors with lower caseloads (OR = 1.10). In contrast, American Indian, White , and Latina women were significantly less likely to disappear than African American women (OR = 0.19, 0.26, 0.37 respectively). Women who disappeared were significantly more likely to be younger (OR =1.06), unemployed (OR = 1.77), smoke while pregnant (OR = 1.84), and have a lower completed visit rate during the first three months (OR = 1.02).

Conclusions and Implications: Women who withdraw and disappear tend to leave services at similar times, yet findings suggest they are distinct groups whose exits may result from different factors. Although service exit categories require further refinement, the findings suggest that treating early leavers as a homogenous group masks important exit patterns. Implications for targeted retention strategies and program model design will be discussed.