Cost-effectiveness of a Motivational Intervention to Reduce Rapid Repeated Childbearing in High-Risk Adolescent Mothers: Title and subTitle BreakA Rebirth of Economic and Policy ConsiderationsChildbearing in High-Risk Adolescent Mothers FREE

Public concern about teenage childbearing has reemerged as a national issue, because after 14 years of steady decline, US teenage birth rates rose in both 2006 and 2007.^{1 - 2} Why this has occurred is not definitively known, but some speculate that rising economic disparities and decreased public funding for comprehensive pregnancy-prevention efforts may play a role.² Both first and subsequent births to US teenagers produce substantial detrimental health, social, and economic burdens.³ Such births disproportionately affect minorities. Compared with 11% of white girls giving birth during adolescence, 32% of African American and 24% of Hispanic youth bear at least 1 child before the age of 20 years.^{4 - 5} One-quarter of teenagers giving birth will bear another child within 2 years.^{6 - 7}

Until recently, public funds allocated to teenage pregnancy–prevention supported programs that were largely ineffective.⁸ In response to these disturbing trends, policymakers have called for renewed investment in effective, evidence-based prevention programs. The Institute of Medicine recently designated preventing unintended pregnancies in its top 25 priorities for Comparative Effectiveness Research.⁹ President Obama, in his fiscal year 2010 budget, proposed $178 million for “innovative, science-based teen and unintended pregnancy prevention initiatives.”¹⁰ Evidence-based pregnancy-prevention models have been developed, implemented, and studied, but most evaluations have measured effects on intermediate outcomes (eg, contraceptive use) rather than effects on actual pregnancies and births^{11 - 12} ; few have demonstrated effectiveness with rigorous study designs¹¹ ; and hardly any have included examination of costs.¹¹

Repeated childbearing during adolescence compounds the risk of academic failure for the teenage mother and increases the public costs associated with child welfare, criminal justice system involvement, and long-term poverty.^{3 ,6 ,13} Repeated births within 24 months of an index birth occur more commonly among African American (23%) and Hispanic (22%) girls than white adolescent mothers (17%).⁶ Rapid repeated childbearing among teenagers has been largely refractory to intervention, though many strategies for avoiding repeated teenage births are similar to those shown to be effective for prevention of a first birth.^{6 ,11} Effective programs provide individualized support, counseling, and education via nurturing relationships; promote educational achievement; and sponsor youth development through service learning.^{6 ,11} Examination of the cost-effectiveness of programs is sparse.¹² Cost-effectiveness studies may assist policymakers and funders in their quest for evidence-based models to prevent births to teenagers.

The current study contributes an economic analysis of computer-assisted motivational intervention (CAMI), our home-based intervention that successfully reduced repeated teenage childbearing among high-risk, low-income, predominantly African American teenage mothers. Evaluated experimentally, the multi-component, home-based, behavior-change counseling program achieved a 45% reduction in repeated births within 24 months of the index birth compared with usual care.¹⁴ In this article, we conduct a cost-effectiveness analysis of the program.

Study data came from a randomized controlled trial of an intervention to reduce rapid subsequent births in adolescent mothers.¹⁴ Between 2003 and 2005, pregnant adolescents were recruited from prenatal care sites in Baltimore, Maryland. Teenagers and their parent or guardian signed informed consent for participation in the study; the teenagers completed baseline assessment interviews and were randomly assigned to intervention or usual care control groups. Two home-based intervention groups were compared with usual care: (1) CAMI plus enhanced home visiting (CAMI+), an intensive multi-component intervention with monthly participant contact, and (2) CAMI only, a less intensive single-component intervention with quarterly participant contact. Intervention groups received services until the adolescent was 2 years post partum; study follow-up was conducted at that point. Methods were approved by the institutional review boards of the University of Maryland School of Medicine and the Maryland Department of Health and Mental Hygiene.

Data for this study were obtained from study participants' baseline assessment interviews, program staff interviews, program administrative records, and Maryland birth certificate records. We defined a subsequent birth as one that occurred within 24 months of the index birth. Repeated births were ascertained from Maryland birth certificate records, which were successfully matched for the entire sample.¹⁴

The study sample was a cohort of 235 pregnant adolescents aged 12 to 18 years who were at 24 or more weeks of gestation at entry and predominantly African American and from low-income families. Participants were recruited from 5 urban medical clinics that provided prenatal care to women who were without insurance or insured under Medicaid. Following consent procedures, teenagers completed a baseline assessment interview and were randomly assigned to CAMI+, CAMI only, or the usual care group. By design, more teenagers were assigned to the intervention groups than the usual care group.

The home-based intervention used motivational interviewing, an empirically validated behavior change counseling method.¹⁵ Trained community outreach workers called CAMI counselors conducted the intervention. Most intervention activities took place in home settings, but some teenagers (eg, those with housing instability) preferred to meet with their CAMI counselor in other community-based locations. The CAMI counselors used laptop computers with customized software to measure each teenager's reproductive health risks, current behaviors, contraceptive-use intentions, motivation to remain nonpregnant, and short-term and long-term life goals.¹⁴ Computer-assisted motivational intervention algorithms measured the teenager's motivation to prevent a repeated pregnancy and computed her risk of pregnancy and sexually transmitted infections.^{16 - 17} The CAMI counselor then conducted a 20-minute motivational interview counseling session to enhance motivation to use contraception and avoid pregnancy.¹⁵ Motivational interviewing is an empirically validated behavior-change counseling style that facilitates an individual's motivation to change. That is, motivational interviewing differs from traditional counseling by acknowledging that merely telling a teenager to prevent repeated pregnancy or even providing her with access to medical care and contraception services is an ineffective strategy to promote behavior change in someone who is ambivalent or unmotivated.^{18 - 19} Motivational interviewing seeks to raise awareness of discrepancies between stated goals and current behaviors.¹⁵ For example, teenaged mothers entering CAMI frequently maintained that they did not want another pregnancy until adulthood, yet they continued risky sexual behaviors like intercourse without the use of contraception. Most expressed emphatically that they intended to complete school and get a good job, yet many had dropped out of school and enumerated multiple reasons that prevented their return to school. In CAMI sessions, the counselor offered the teenager factual information and gave personalized nonjudgmental feedback to promote self-efficacy for positive change. Collaboratively, the teenager and counselor established short-term and longer-term goals and reviewed them in subsequent sessions.¹⁴

Adolescents assigned to each intervention group received a CAMI session at quarterly intervals. Sessions began after delivery of the index child and were repeated every 3 months until the index child turned 2 years old, for a maximum of 9 sessions.

Adolescents allocated to the CAMI+ group received monthly home visits from their CAMI counselor, such that during a 3-month period 1 visit consisted of a CAMI session and 2 visits were spent on other activities. During non-CAMI visits, the counselor guided the teenager through a parenting curriculum²⁰ ; provided case management help with housing, day care, school, and health care issues; and sought to help the teenager to learn skills to negotiate daily challenges from adverse personal circumstances (eg, abusive relationships and drug trafficking within the home). The timing and frequency of CAMI sessions were identical in both intervention groups and began after delivery.

The main study describes how the adolescents were extremely challenging to find for CAMI sessions. Only 66% of teenagers in the CAMI+ group and 43% in the CAMI-only group completed our defined minimal set of intervention sessions.¹⁴ The most common reason for nonadherence in the CAMI+ group was the teenager's failure to keep confirmed home visit appointments. The most common reason in the CAMI-only group was our inability to locate the adolescent.

In the current study, we used intent to treat methods to compute our effectiveness estimations.^{21 - 22} That is, a teenager assigned to either intervention group was viewed as treated, whether or not she participated in intervention sessions. With similar methods, our original effectiveness study found that mothers in the CAMI+ group were significantly more likely to defer a subsequent birth beyond 24 months after their index birth (hazard ratio, 0.45; P = .047), but among mothers in the CAMI-only group, the observed lower risk of repeated births was not statistically different from controls. Additional models that controlled for actual receipt of the CAMI demonstrated that teenagers in either the CAMI+ or CAMI-only group experienced fewer births than controls.¹⁴

In the current study, we assessed CAMI's overall effectiveness using logistic regression to calculate the odds of a teenager experiencing a repeated birth (yes/no) within 24 months of her index birth. We computed 2 models: model 1 assessed the effectiveness of any CAMI (any CAMI) by combining data from the CAMI+ and CAMI-only groups to maximize power to detect intervention effects (CAMI+ or CAMI only vs usual care); model 2 assessed the effectiveness of each intervention separately (CAMI+ vs usual care and CAMI only vs usual care). Both models controlled for confounders that prior analyses showed to be associated with repeated birth or that differed significantly among the groups at baseline, thereby creating an imbalance between groups despite randomization.¹⁴ These variables included whether the teenager was already a parent (had a prior birth before this study's index pregnancy), had ever been diagnosed with a sexually transmitted infection prior to the current study, intended to use reliable contraception after delivery of the index child, and became newly insured (ie, with Medicaid) during the period of the index pregnancy or she was continuously insured in the 12 months prior to the pregnancy. We believe this insurance variable is a marker of social risk rather than reflecting differences in insurance eligibility due to parity because (1) we found no association between prior birth and insurance status, and (2) virtually all teenagers in our sample were eligible for Medicaid coverage or Maryland's Child Health Insurance Program prior to their pregnancy, but for whatever reason were not enrolled. Unadjusted and adjusted odds ratios (AORs) and 95% confidence intervals (CIs) were estimated for all variables. We conducted sensitivity analyses with age categories (12-15, 16-17, and 18-19 years) and for interactions of age with intervention groups.²³

We calculated the intervention costs, including start-up and implementation, but excluding research costs. Costs were determined by program personnel (B.B. and M.D.), who reviewed program administrative records. Aggregate costs were calculated by summing the individual components and personnel costs for each intervention arm separately. We estimated costs per teenager for the 80 teenagers assigned to the CAMI+ group and for the 87 teenagers assigned to the CAMI-only group, and we computed a weighted mean cost per teenager for the 167 teenagers in the any-CAMI group. All costs were adjusted to 2009 US dollars.

The incremental cost-effectiveness ratio (ICER) is a measure that compares the relative expenditures (costs) and outcomes (effects) of 2 or more types of treatment.²⁴ We estimated ICERs normalized to a sample of 100 adolescent mothers for our 2 models: (1) any CAMI vs usual care and (2) CAMI+ vs usual care or CAMI only vs usual care. To calculate the cost per prevented repeated birth, a specific type of ICER, we divided the costs attributable to the intervention by the difference in number of repeated births between the intervention and control groups. For each model, we calculated unadjusted ICERs as well as ICERs that adjusted for covariates.

To translate our ICER findings into more meaningful estimates for clinical and policy decision-makers, we present our findings about the heterogeneity of results as case scenarios rather than showing statistical distributions and cost-effectiveness accessibility curves.²⁵ Each point in a scenario analysis represents different assumptions within a variety of clinical scenarios. For example, we created scenarios for specific subpopulations of teenage mothers who differed with respect to intervention group, age, insurance status, and parity. We examined and compared each subgroup's costs per prevented repeated birth. All statistical analyses were performed using the Stata statistical package, version 9.2 (Stata Corp, College Station, Texas).

Baseline data depict the sample's high level of social and economic disadvantage (Table 1). The mean age was 17.0 years (SD, 1.2 years); 97% were African American; 42% had dropped out of school; and while 86% reported current insurance coverage by Medicaid during the index pregnancy, 39% had become newly insured because of their pregnancy.

Weighted mean costs for any CAMI were $2064 per teenager; costs per teenager for CAMI+ and CAMI only were $2735 and $1449, respectively (Table 2). On average, the intervention lasted between 24 and 27 months for each teenager (prenatal enrollment through 24 months post partum). Compared with the CAMI-only arm, CAMI+ costs were higher per teenager because each CAMI+ interventionist carried a smaller caseload than the CAMI-only interventionist (20 vs 60 teenagers) to provide the greater contact frequency of the enhanced home-visiting components.

Effectiveness of CAMI is shown in Table 3. Teenagers in the group receiving any CAMI (model 1) and those in the CAMI+ group (model 2) were significantly less likely to experience a subsequent birth compared with teenagers in the usual care group (any CAMI AOR, 0.47; 95% CI, 0.22-0.97; CAMI+ AOR, 0.38; 95% CI, 0.16-0.92). While teenagers in the CAMI-only group exhibited lower odds of subsequent birth compared with teenagers in the usual care group, differences were not statistically significant (CAMI-only AOR, 0.56; P = .17). Of note, newly insured teenagers experienced twice the odds of repeated birth compared with continuously insured teenagers (AOR, 2.05; 95% CI 1.03-4.10) (Table 3).

Unadjusted and adjusted ICERs are presented in Table 4. In this sample normalized to 100 teenagers, compared with no intervention and adjusted for covariates, any CAMI prevented 12 repeated teenage births at a cost per prevented repeated birth of $17 388. The CAMI+ prevented a greater number of repeated teenage births (14 births prevented), but was less cost-effective (ICER of $19 247). Scenario analyses demonstrated significant variability in the intervention's cost-effectiveness that depended on teenage subgroup characteristics (Figure). Costs per prevented repeated birth were highest for the youngest teenage mothers in the CAMI+ group with continuous insurance (ICER = $27 187) and lowest for older teenage mothers in the CAMI-only group who were newly insured (ICER = $6822). The teenager's parity, whether or not she had experienced a prior birth, did not affect the ICER appreciably (data not shown). In general, newly insured teenagers represented the group for whom the intervention was most cost-effective.

This article provides one of the first cost-effectiveness analyses of an evidence-based intervention that successfully reduced rapid subsequent births in adolescent mothers. Overall, the mean cost per adolescent was $2064, with unadjusted and adjusted ICERs of $21 895 and $17 388, respectively, per prevented repeated birth. Specific personal characteristics moderated CAMI's cost-effectiveness, thereby producing variability in the ICERs for subgroups of teenagers. For example, the ICER for a 12-year-old newly insured mother in the CAMI+ group was $15 199 compared with $27 187 for a continuously insured teenager of the same age and intervention group. Similarly, the ICER for a 15-year-old newly insured mother in the CAMI+ group was $16 516, compared with $11 233 for a continuously insured mother in the CAMI-only group (Figure).

Cost-effectiveness ratios were most favorable for newly insured teenagers—teenagers who were not enrolled in insurance prior to their pregnancy, though they were eligible for public insurance coverage. This newly insured subset of teenagers may be at especially high risk. Studies have shown that children who are eligible but uninsured are more likely to live in single-parent homes or without a parent at all and are among the most disadvantaged.²⁶ Not surprisingly, this vulnerable group with fewer support resources is less attuned to preventive health care,²⁷ and among teenagers with disproportionate risk, the CAMI may have mitigated some of the effects of social disadvantage. The CAMI aimed to promote teenagers' self-efficacy for achieving future goals as well as for pregnancy-prevention behaviors. For those with the greatest risk of repeated birth, CAMI produced greater impact at a lower cost. These findings provide useful data for program planners and policymakers trying to decide whom to target with similar, evidence-based interventions to reduce risk of teenage childbearing.

Only a handful of experimentally evaluated teenage pregnancy–prevention interventions have been subject to examination of costs and benefits.^{28 - 30} One such program, the Children's Aid Society Carrera Program, an intensive, multi-component, 3-year intervention, found that program participants demonstrated significantly lower odds of pregnancy compared with controls at a cost of $4000 per teenager per year (2002 dollars); cost-effectiveness was not calculated.³¹ Another experimentally evaluated, intensive program provided services to teenage mothers for 2 years via home visits from nurses that were aimed to reduce repeated births. It demonstrated significant reductions in repeated teenage births for first-time teenage mothers at a cost of $7681 per family (1999 dollars).³² Policy analysts have singled out these evidence-based programs as meriting further investment.²⁸ Our findings suggest that CAMI is at least as cost-effective as these programs and warrants replication in larger samples for consideration in that group.

Our study has several limitations. First, participants were predominantly urban, African American teenagers living within a relatively circumscribed geographic area, which raises concerns of generalizability. However, because African American teenagers experience high teenage birth rates⁶ and Baltimore's teenage birth rate is among the nation's highest,³³ it makes sense to develop culturally appropriate, effective, evidence-based interventions for this group. Second, we do not know whether the CAMI effects persisted beyond the 2-year study follow-up period throughout the teenage mother's adolescence. Third, the greater effectiveness of CAMI+ raises the question of whether components within the enhanced home visits, rather than the CAMI, prevented repeated births. We do not believe this to be the case because (1) our prior trial of home visiting without CAMI showed no impact on repeated births to teenaged mothers,³⁴ and (2) statistical models that accounted for whether a teenager actually received a portion of the CAMI sessions demonstrated reductions in repeated birth in the CAMI-only group.¹⁴ We attribute the greater effectiveness of CAMI+ to a combination of more frequent contact with a broader scope of activities, which resulted in superior participant engagement. However, this intensity came with a cost. Fourth, the relatively small sample size produced wide CIs for our ICER estimates. Finally, while we believe CAMI can be applied to primary pregnancy-prevention efforts, this study cannot answer whether CAMI is effective or cost-effective for preventing first births to teenagers.

The CAMI effectively reduced repeated births in adolescent mothers. Its costs compare favorably with other effective teenage pregnancy-prevention programs, but it was not low cost. Cost-effectiveness will be enhanced by clarifying the target population for whom CAMI is most effective. Nevertheless, evidence-based prevention efforts represent sound fiscal policy, and this economic analysis makes a case for further investment in CAMI with replication in a larger sample.

Nearly 1 in 5 (18%) girls experiences a birth by age 20 years,^{35 - 36} and one-quarter of these teenaged mothers will bear another child within 24 months. They face increased risk of delivering premature and low-birth-weight infants,⁴ dropping out of school and never graduating,³⁷ bearing children who experience academic and behavioral problems,³⁸ and living in poverty.³⁹ The daughters of teenage mothers commonly become teenage parents themselves,^{39 - 40} while the sons frequently end up in prison.⁴¹ Public spending for families begun by teenagers costs the US taxpayer about $9.2 billion per year in public assistance, child welfare, health care costs, incarceration, and lost tax revenue,⁴² highlighting the continued need for effective programs that reduce the number of children born into the disadvantaged circumstances of teenage parenthood. Expenditures per teenage mother per year are estimated at $4080 nationally (for teenagers ≤17 years old) and $5150 in Maryland.³³ Findings from this study provide evidence for the value of investing in evidence-based interventions, particularly for young people who are most vulnerable.

Reducing births to teenagers will improve the well-being of children, adolescents, families, and communities. Fewer teenage births will lower taxpayers' burden and benefit national and state economies, enabling investment in other priority areas. The National Conference of State Legislatures, in its recent anti-poverty policy agenda, calls for federal policies that foster and enable state-based initiatives and strategies that reach across government, business, and community-based efforts.⁴³ The National Conference of State Legislatures urges innovation in program development, a focus on meaningful outcomes, and accountability in the measurement of outcomes. Given the current spirit of health care reform, personalized medicine, and the need to identify what works in slowing current health care spending, this study offers an effective model for reducing births to teenagers, particularly in high-risk subgroups, and is worthy of consideration for replication by state and national policy leaders.

Correspondence: Beth Barnet, MD, University of Maryland Family Medicine, 29 S Paca St Lower Level, Baltimore, MD 21201 (bbarnet@som.umaryland.edu).

Accepted for Publication: December 5, 2009.

Author Contributions:Study concept and design: Barnet, Rapp, DeVoe, and Mullins. Acquisition of data: Barnet and DeVoe. Analysis and interpretation of data: Barnet, Rapp, and Mullins. Drafting of the manuscript: Barnet, Rapp, and DeVoe. Critical revision of the manuscript for important intellectual content: Barnet, Rapp, and Mullins. Statistical analysis: Rapp and Mullins. Obtained funding: Barnet. Administrative, technical, and material support: Barnet and DeVoe. Study supervision: Barnet, DeVoe, and Mullins.

Financial Disclosure: Dr Mullins has received consulting income or honoraria from Amylin Pharmaceuticals, AHIMA Foundation, Bayer Pharmaceuticals, Bristol-Myers Squibb, Genentech, GlaxoSmithKline, Eli Lilly, Merck, Novartis, Pfizer, and Sanofi-Aventis. He has also received grant support from Elan, GlaxoSmtihKline, Novartis, Pfizer, and Sanofi-Aventis.

Funding/Support: This research was supported by grant APRPA006010 from the Department of Health and Human Services, Office of Population Affairs, Office of Adolescent Pregnancy Programs.