Levels and trends in fertility rates among adolescents aged 15-19 in 21 States and Union Territories in India from 1990 to 2050: A Bayesian Modelling Study

ObjectivesAdolescent fertility in India remains underexplored at the state level. This study estimates age-specific fertility rates (ASFRs) for females aged 15-19 across all major Indian states and union territories (UTs) from 1990 to 2025 and generates probabilistic projections of state-level ASFRs for females aged 15-19 through 2050. DesignBayesian modelling based on fertility data from five phases of the India National Family Health Surveys (NFHSs) in 1992-93, 1998-99, 2005-06, 2015-16, and 2019-21 and the Sample Registration System from 1971 to 2007. SettingIndia. Participants(1) 1992-93 NFHS: 7,815 ever-married women aged 15-19, (2) 1998-99 NFHS: 7,041 ever-married women aged 15-19, (3) 2005-06 NFHS: 23,955 women aged 15-19, (4) 2015-16 NFHS: 124,878 women aged 15-19, (5) 2019-21 NFHS: 122,480 women aged 15-19 years, and (6) Sample Registration System (SRS) from 1971 to 2007: 313 state-level ASFR 15-19 records. Outcome measuresIn 2025, Jammu and Kashmir had the lowest ASFR estimates at 19.5 births with a 95% credible interval [14.2, 26.4] per 1,000 females aged 15-19, while West Bengal had the highest rate at 79.4 [54.0, 116.8] births per 1,000. High 15-19 ASFR estimates concentrated in north-western states. We project adolescent fertility to eight decline across states/UTs or remain stable in those estimated at low levels prior 2025. By 2050, the 15-19 ASFRs are projected to range from 18.5 [12.0; 28.3] to 64.7 [33.9; 116.8] per 1,000 across Indian states/UTs. ConclusionsThis study provides the first estimates and probabilistic projections of ASFR for ages 15-19 at the state level in India, including uncertainty estimates based on reproducible models. These outcomes can assist in region-specific policy planning to address child marriage, expand contraceptive access among adolescents, and strengthen educational and gender-equity initiatives. ARTICLE SUMMARYO_ST_ABSStrengths and limitations of this studyC_ST_ABSO_LIWe developed a Bayesian hierarchical modelling framework that incorporates measurement uncertainty and differences in data quality across sources, and enables information sharing across states/UTs to improve estimates in states/UTs that lack data. C_LIO_LIThe study is based on an extensive database of state-level ASFR aged 15-19, including five India National Family Health Surveys and the Sample Registration System, totalling 1,510 fertility observations. C_LIO_LIBy modelling key covariates such as marital status, education attainment, and total fertility at the state level, the approach captures subnational demographic variations in India. Particularly, we allow potential non-linear effect from the marital status covariate in model setting. C_LIO_LIThe modelling choice of assuming fixed effects for selected covariates may not fully capture state-level dynamics or shifts over time. C_LIO_LIThe probabilistic projections assume that the relationships between covariates and adolescent fertility remain stable, which may be inadequate if major behavioural or policy changes occur in the future. C_LI