Department of Pediatrics, Stanford University School of Medicine, Stanford, California, USA.
Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, USA.
J Glob Health. 2020 Dec;10(2):021004. doi: 10.7189/jogh.10.021004.
Geographical variations in the levels and trajectory of health indicators at local level can inform the adaptation of interventions and development of targeted approaches for efficient scale-up of intervention impact. We examined the hypothesis that time trends of a set of reproductive, maternal, newborn, and child health and nutrition (RMNCHN) indicators varied at block-level during the statewide scale-up phase of the program in Bihar, India.
We used data on 22 selected indicators from four rounds of the Community-based Household Survey carried out between 2014 and 2017. Indicator levels at each round were estimated for each block. We used hierarchical Bayesian spatiotemporal modelling to smooth the raw estimates for each block with the estimates from its neighbouring blocks, and to examine space-time interaction models for evidence of variations in trends of indicators across blocks. We expressed the uncertainty around the smoothed levels and the trends with 95% credible intervals.
There was evidence of variations in trends at block level in all but three indicators: facility delivery, public facility delivery, and age-appropriate initiation of complementary feeding. Fifteen indicators showed trends in opposite directions (increases in some blocks and declines in others). All blocks had at least 97.5% probability of a rise in immediate breastfeeding, early pregnancy registration, and having at least four antenatal care visits. All blocks had at least 97.5% probability of a decline in seeking care for pregnancy complications.
The findings underscore the value of monitoring and evaluation at local level for targeted implementation of RMNCHN interventions. There is a need for identifying systematic factors leading to universal trends, or variable contextual or implementation factors leading to variable trends, in order to optimise primary health care program impact.
ClinicalTrials.gov number NCT02726230.
在地方层面上，健康指标的水平和轨迹存在地域差异，这可以为干预措施的调整和有针对性方法的制定提供信息，以有效扩大干预措施的影响。我们检验了这样一个假设，即在印度比哈尔邦实施该计划的全州范围扩大阶段，一系列生殖、孕产妇、新生儿和儿童健康与营养（RMNCHN）指标在街区层面的时间趋势存在差异。
我们使用了 2014 年至 2017 年期间进行的四轮基于社区的家庭调查中收集的 22 项选定指标的数据。每一轮每个街区的指标水平都进行了估计。我们使用分层贝叶斯时空模型对每个街区的原始估计值进行平滑处理，同时利用其相邻街区的估计值进行平滑处理，并对指标在街区之间的趋势进行时空交互模型检验，以发现趋势变化的证据。我们用 95%可信区间表示平滑水平和趋势的不确定性。
除了三个指标（分娩地点、公共分娩地点和适当年龄开始补充喂养）外，所有指标在街区层面都存在趋势变化的证据。15 个指标显示出相反的趋势（一些街区上升，另一些街区下降）。所有街区都有至少 97.5%的可能性会增加立即母乳喂养、早期妊娠登记和至少进行四次产前护理。所有街区都有至少 97.5%的可能性会减少因妊娠并发症寻求医疗护理的行为。
这些发现强调了在地方层面进行监测和评估对于有针对性地实施 RMNCHN 干预措施的重要性。需要确定导致普遍趋势的系统性因素，或导致趋势变化的可变背景或实施因素，以优化初级卫生保健计划的影响。
ClinicalTrials.gov 编号 NCT02726230。
