BACKGROUND: International infant mortality rates vary widely. This variation has been attributed to many factors, including differential reporting. In the US, American Indians and Alaska Natives (AI/AN), who generally have low socioeconomic status, have a low neonatal mortality rate. One possible explanation is underregistration of very low birthweight (VLBW, < 1,500 g) births. We hypothesized that underregistration may occur disproportionately among AI/AN residing on or near reservations (areas controlled by an American Indian group). We estimated infant mortality in these areas. METHODS: Linked birth-infant death files for 1989-1991 were used to compare VLBW and neonatal mortality among AI/AN infants in counties with reservations with those in non-reservation counties. The VLBW rates for non-reservation counties were applied to the reservation risk distribution to calculate directly adjusted VLBW and neonatal mortality rates for reservation counties. This method assumes that greater registration in non-reservation counties yields a more accurate estimate of the relationship between risk factors and outcomes. RESULTS: Despite a higher prevalence in reservation counties of risk factors, the reported VLBW rate was 0.84% in reservation and 1.17% in non-reservation counties. The neonatal mortality rate was 5.4 per 1,000 in reservation counties and 6.0 in non-reservation counties. Direct adjustment yielded a VLBW rate of 1.28% (95% CI: 1.14-1.39) and a neonatal mortality rate of 6.7-9.8 per 1,000 in reservation counties. CONCLUSIONS: Reported neonatal mortality among AI/AN may understate the true rate due to underregistration of VLBW births. Direct adjustment may be useful in estimating infant mortality rates for populations with incomplete vital registration.
BACKGROUND: International infant mortality rates vary widely. This variation has been attributed to many factors, including differential reporting. In the US, American Indians and Alaska Natives (AI/AN), who generally have low socioeconomic status, have a low neonatal mortality rate. One possible explanation is underregistration of very low birthweight (VLBW, < 1,500 g) births. We hypothesized that underregistration may occur disproportionately among AI/AN residing on or near reservations (areas controlled by an American Indian group). We estimated infant mortality in these areas. METHODS: Linked birth-infant death files for 1989-1991 were used to compare VLBW and neonatal mortality among AI/AN infants in counties with reservations with those in non-reservation counties. The VLBW rates for non-reservation counties were applied to the reservation risk distribution to calculate directly adjusted VLBW and neonatal mortality rates for reservation counties. This method assumes that greater registration in non-reservation counties yields a more accurate estimate of the relationship between risk factors and outcomes. RESULTS: Despite a higher prevalence in reservation counties of risk factors, the reported VLBW rate was 0.84% in reservation and 1.17% in non-reservation counties. The neonatal mortality rate was 5.4 per 1,000 in reservation counties and 6.0 in non-reservation counties. Direct adjustment yielded a VLBW rate of 1.28% (95% CI: 1.14-1.39) and a neonatal mortality rate of 6.7-9.8 per 1,000 in reservation counties. CONCLUSIONS: Reported neonatal mortality among AI/AN may understate the true rate due to underregistration of VLBW births. Direct adjustment may be useful in estimating infant mortality rates for populations with incomplete vital registration.