Skip to main content

Table 1 GAPPS quality assessment of epidemiological parameters in global estimates using adapted version of GRADE: PRETERM BIRTH

From: Global report on preterm birth and stillbirth (1 of 7): definitions, description of the burden and opportunities to improve data

  Prevalence Preterm birth as a direct/indirect cause-of-death   Impairment following preterm birth  
Epidemiological Parameters Preterm Birth Prevalence Rate Preterm Birth as Direct Cause of Neonatal Death 2000 Preterm Birth as Direct Cause of Neonatal Death 2005 Preterm Birth as Risk Factor for Neonatal Death Retinopathy of Prematurity Chronic Lung Disease Multi-Domain Impairment
Definition The proportion of babies born at less than 37 weeks of completed gestation, per 100 live births Preterm birth direct co006Dplications as the cause-of-death as defined in ICD ie RDS/HMD, necrotizing enterocolitis, intraventricular hemorrhage and other direct complications of preterm birth, or very early neonatal death in newborn of gestational age <32 weeks Neonatal death in a preterm baby where preterm birth is indirect - for example a baby who is moderately preterm and dies of infection a few days after birth. A vasoproliferative disorder due to aberrant vascular proliferation of the immature retina. There are 5 stages of severity. The risk of ROP is increased for preterm babies exposed to hyper-oxygenation. CLD is defined as persistent need for oxygen therapy in order to maintain oxygen saturation above 88% to 36 weeks of postmenstrual age (note some variation in case definitions, also sometimes called Bronchopulonary dysplasia) Impairment aff ecting more than one domain of function as used in GBD assessments (cognitive, motor, vision, hearing, seizure disorder)
Systematic global estimates available (source and date) WHO and Child Health Epidemiology Reference Group (CHERG) for Global Burden of Disease (GBD), in process CHERG for WHO/UN [4] CHERG and WHO in process CHERG new grant - not yet started CHERG for GBD, in process CHERG for GBD, in process (review by Maneesh Batra) CHERG for GBD, in process
Countries with VR data used (used as reported, new analysis or adjusted) Not in vital registration 45 (~96,797 neonatal deaths) 61 (~142,000 neonatal deaths) Work to be done 2009 - 2012 Not reported consistently from ICD data although ICD codes allocated Several ICD codes allocated. Not reported consistently from ICD data Not in vital registration
Countries with survey data used (used as is, new analysis or adjusted) Not in current national surveys Not in current national surveys Not in current national surveys Work to be done 2009 - 2014 Not in current national surveys Not in current national surveys Not in current national surveys
Countries where modelled estimates used (basis of model) (in process) 138 (27 from VR model and 111 from study based model) Multinomial model to simultaneously estimate 7 causes. 132 (32 from VR model and 100 from study-based model). Multinomial model to simultaneously estimate 7 causes. Work to be done 2009 - 2014 (in process) (in process) (in process)
Types of data inputs for modelling (in process) VR, published studies from community (verbal autopsy) and facilities, unpublished datasets (eg DSS) VR, published studies from community (verbal autopsy) and facilities, unpublished datasets (eg DSS) Work to be done 2009 - 2014 Often hospital registries Very few comparable studies even in HICs Varied and no population based studies cohort from LICs identified
No of studies/ datasets included in global estimate modelling (in process) Searches of 6820 hits, 46 studies and 10 unpublished datasets included Searches of 5591 hits, 71 studies/datasets included Work to be done 2009 - 2014 (in process) (in process) (in process)
Total with measure of parameter (e.g., SBs, preterm births) (in process) VR (~96,797 neonatal deaths) Study data (13,685 neonatal deaths) VR (~130,000 neonatal deaths) Study data (23,638 neonatal deaths) Work to be done 2009 - 2014 (in process) (in process) (in process)
Median yr of input data (in process) VR median year 1999 Study data median year 1991 VR median year 2004 Study data median year 1991 Work to be done 2009 - 2014 (in process) (in process) (in process)
Variability in outcome measurement methods (in process) Yes – death certificates/ ICD codes, clinical assess, med records, verbal autopsy Yes - death certificates/ ICD codes, clinical assess, med records, verbal autopsy Work to be done 2009 - 2014 (in process) varied case definitions Very varied case definitions, varied age at assessment and multiple assessment tools
Limitations re population representativeness (in process) VR data representative for 46 countries. Study based data often from non representative populations VR data representative for 69 countries. Study based data often from non- representative populations Work to be done 2009 - 2014 Hospital studies and in LIC/MIC from referral centres, few data re moderate gestational age/BWT babies Hospital studies and only in high income countries LIC/MIC from referral centers and limited cohort data
Generalizability to Population of Interest (i.e., geographic match of data to burden) (in process) Important gaps in the input data—especially in China, west and central Africa and central Asia Much improved data for China and India. Still limited for central Africa and central Asia Work to be done 2009 - 2014 Limited Very limited Very limited
Is there systematic equity assessment No No No No No No No
Global estimate (in process) 1.12 million (27.9% of 4 million) 1.23 million (33.1% of 3.72 million) Work to be done 2009 - 2014 (in process) (in process) (in process)
Range (in process) 0.74 to 1.38 million 0.84 to 1.52 million Work to be done 2009 - 2014 (in process) (in process) (in process)
Consistency between estimates if more than one set To compare to WHO global and regional estimates 2009 First set of systematic estimates In comparison with 2000 estimates, a reduction in neonatal tetanus deaths, and apparent increase in deaths due to preterm birth None to compare with None to compare with None to compare with None to compare with
Overall summary of quality of data input (in process) Moderate for high income countries, low for low income countries Moderate for high income countries, low for low income countries (in process) (in process) (in process) (in process)
Overall quality of estimates according to standards for global estimates (in process) Moderate to high - transparent methods, advances in multi-cause modelling now the std for multi-cause work but limited by input data and especially consistency in cause-of-death attribution (in process) (in process) (in process) (in process)  
Priority areas to improve measurement now 1. Increase quality and quality of gestational age data in routine data sources
2. Test feasibility/ validity of gestational age data collection through household surveys
3. Test feasibility/ validity of simplified gestational age clinical assessment
4. Increase dissemination of country level best estimates once completed
1. Disseminate case definition and hierarchies for preterm complications as a direct cause-of-death vs preterm birth as a risk factor,
2. Increase quality and quality of cause-of-death data in vital registration and national audit data
3. Increase quality and quality of SBR data in verbal autopsy data applying and a standard hierarchy
4. Increase dissemination of country level best estimates
Same inputs as required to improve measurement of gestational age and causes of death 1. Agree on case definitions and measurement tools and ages at which measurement should occur
2. Increase quality and quality of impairment data especially in transitional countries with increasing neonatal survival
3. Disseminate case definitions well so that future studies result in more comparable data, and ideally standard assessment guidelines
4. Advocate for more funding for cohort studies tracking impairment outcomes from preterm birth and other neonatal morbidities