Growth monitoring and mortality risk in low birthweight infants: a birth cohort study in Burkina Faso

Background: Wasting and underweight in infancy is an increasingly recognised problem but consensus on optimum assessment is lacking. In particular, there is uncertainty on how to interpret anthropometry among low birth weight (LBW) infants who may be growing normally. This research aimed to determine growth of infants from birth to two months (around age of vaccination) and the mortality risk of underweight LBW infants compared to normal birth weight (NBW) infants at two and six months age. Methods: A secondary analysis of a birth cohort of 1103 infants in Burkina Faso was conducted. Anthropometry was performed monthly from 0 to 12 months. We assessed associations with mortality using Cox proportional hazards models and assessed discriminatory values using area under receiver operating characteristics curves. Results: Eighty-six (7.8%) children died by age one year, 26/86 (30%) and 51/86 (59%) within two and six months, respectively. At age two months, weight gain since birth did not better discriminate mortality risk than current weight-for-age (P=0.72) or mid-upper arm circumference (P=0.21). In total, 227 (21%) LBW infants had increased risk of mortality: adjusted hazards ratio (aHR) 3.30 (95%CI 2.09 to 4.90). Among infants who were underweight at two and six months, LBW infants (64% and 49%, respectively) were not at reduced risk of death compared to NBW infants (aHR 2.63 (95%CI 0.76 to 9.15) and 2.43 (95%CI 0.74 to 7.98), respectively). Conclusion: Assessing weight gain since birth does not offer advantages over immediate anthropometry for discriminating mortality risk. LBW infants who are later identified as underweight require care to help prevent mortality.


Introduction
Infancy is the period of fastest relative growth; on average, a normally growing infant more than doubles their birth weight in the first six months of life :The WHO Child Growth Standards 2006: [Available from: http://www.who.int/childgrowth/standards/en/].In early infancy, apparent wasting or underweight may occur due to growth faltering and/or as a result of having been born preterm, low birth weight (LBW) or small for gestational age 1 .
The global prevalence of LBW is 15%, representing more than 20 million infants, 91% of whom are in low and middleincome countries (LMICs) 2 .In LMICs, birthweight and subsequent anthropometry in infancy and childhood are predictive of both short and longer-term mortality 3 .Studies report ongoing extra-uterine weight and length restriction and reduced physical strength through to adolescence following LBW compared to normal birth weight (NBW) infants 4 .The rate of infant growth irrespective of birth weight is influenced by nutritional intake, absorption and assimilation of nutrients, nutrient losses due to infection, other acute or chronic diseases, and genetic or epigenetic predisposition 5 .
High-quality growth references standards exist for preterm infants from the INTERGROWTH-21 st study 6 .However, information on gestational age is often unavailable or unreliable in LMICs, hence WHO growth standards are usually applied to all infants irrespective of birth size.In growth monitoring or nutrition programs, LBW babies may be classified as underweight or wasted whilst having a normal (preterm) growth velocity tracking lower percentiles or 'catching up'.Anthropometric indicators such as weight-for-length Z score (WLZ) and mid-upper arm circumference (MUAC) identify a large proportion of infants with a history of LBW.For example, in Kenya, 43% of ill hospitalized malnourished infants under six months old had a history of small size at birth 1 , while in India, LBW was a strong predictor of severe wasting among infants below the age of six months 7 .
Thus, there is uncertainty among health workers regarding interpretation of anthropometry in LBW infants, with low expectations of growth in LBW infants common and "slow" and potentially "poor" growth in infants (underweight) regarded as acceptable 8 .Additionally, practitioners target catch up growth of LBW infants to that of peers and assume that mortality risk is resolved among LBW infants who are no longer classified as undernourished 9 .There are also concerns for potential future health risks associated with accelerated weight gain 10,11 .Whether and how to intervene on undernourished LBW infants has implications for health system workload and costs.Consequently, to evaluate risks associated with anthropometry and types of interventions needed, birthweight may need to be considered during anthropometric assessment in infancy.In practice, vaccination at around two months of age is an established opportunity to assess growth and to intervene.
Given this background, we examined data from a birth cohort to compare the discriminatory value for mortality for anthropometry at birth and changes from birth to the following timepoints in infancy: i) at two months of age which is around the time of infant immunisations; and ii) at six months of age.We also investigated overall mortality risk among LBW infants and whether among infants with low anthropometric values measured at two and six months of age, LBW was associated with lower risk of subsequent mortality.

Study site
Data utilized for this secondary analysis was from a birth cohort within Barsalogho Health District, part of the Kaya Health Region in Central North Burkina Faso.It was collected between 1 st April and 31 st December of 2004 in four health centers, including Barsalogho, Basma, Dablo, and Foubé.Though old, the dataset is valuable because it contains follow-up data of an untreated infant cohort, which would be difficult to generate at the present time.An untreated cohort offer a more natural experience of growth patterns within the study population.Additional details on the study site can be found in a previous publication 3 .

Study population and design
The study cohort recruited pregnant women in their third trimester attending scheduled antenatal care visits.The objective of the original study was to compare survival in infancy of full-term LBW infants to that of full-term NBW infants 12,13 .In this secondary analysis, we included data from all live births within the cohort.Follow-up was from birth to 12 months of age through scheduled monthly clinic visits.

Variables
The main outcome of interest was mortality confirmed through hospital records or burial permits/death certificates.Deaths were included in this analysis if they occurred within the first year of life.Anthropometry was collected monthly from birth up to 12 months of age.Exposures examined were anthropometry at birth and ages two and six months (weight (kg), MUAC (cm) and length (cm)) and demographic factors.

Data source/measurements
Both the caregiver and infant demographics and anthropometric measurements were collected at birth, usually within two hours for health facility birth and 48 hours for births in the community by a trained community health worker.MUAC in centimetres were measured with a non-stretch measuring tape to the nearest one mm.An electronic scale (Seca 825 Birmingham, UK) was used to measure weight in kilograms.Length in centimetres were measured using an infantometer (Seca 416, Birmingham, UK).Anthropometric z-scores were calculated using WHO (2006) reference WHO; WHO growth standards STATA macro 2011 [Available from: http://www.who.int/childgrowth/software/en/]. Underweight was defined as weightfor-age z-score (WAZ) <-2.LBW was defined as <2.5kg.

Study size
The parent birth cohort recruited 1103 infants.This secondary analysis included all the 1103 infants.With 1103 infants provide a 7.8% probability of death in one year, a two-sided alpha level of 0.05, the study had power >90% to estimate adjusted hazard ratio of ≥2.25 of LBW associated with death and with power >80% to estimate similar hazard ratio from month two.
Even though the secondary analysis included all the 1103 infants, analyses at different time points used varying number of infants who were alive and in follow-up at the respective time-points.

Statistical analysis
Infant anthropometric measurements were summarised as means and standard deviation.Maternal age was reported as median and interquartile range.We calculated Medians (IQR) weight increments from birth to month 2 and 6 and compared with the WHO 2009 weight increment standards using Wilcoxon single sign rant test [Available from https://www.who.int/childgrowth/standards/velocity/tr3_velocity_report.pdf].To estimate hazards ratios for death associated with anthropometry, we used Cox proportional hazards regression adjusted for features presumed to have biological association with anthropometry such as sex, birth weight, prematurity and being a twin.We assessed and found evidence of unobserved heterogeneity between the four sites using likelihood ratio test (P=0.002) in a regression model assessing effects of LBW on one year mortality.To account for this unobserved heterogeneity, we used shared gamma frailty model.We assessed and found no evidence of violation of proportional hazard assumption using the Schoenfeld residuals.
To test the discrimination of mortality risks from month two to twelve months of age by anthropometric measurements at birth, at month two and the change between birth and month two measurements, we estimated the area under receiver operating characteristics curves (AUC).We used the STATA version 15.1 "roccomp" command to test the hypothesis that the AUCs were equal by comparing AUCs from a single time-point (month 2) with the change from birth to month two.We examined differences in WAZ, and proportion of infants underweight at month two stratified by birth weight and used an independent t-test to test for differences in WAZ between infants born NBW and LBW.

Ethical considerations
The original birth cohort was approved by the Ministry of Health of Burkina Faso (approval number: 1014) in 2003 in accordance with national procedure.All study participants provided written consent to take part in the original study.All data were anonymized before being shared for this analysis.
At two months of age, of 927 infants who had anthropometry assessed, 148 (16%) were underweight; 94 (64%) of whom were LBW.At six months of age, of the 968 infants who had anthropometry assessed, 236 (24%) were underweight; 92 (39%) of whom were LBW.Anthropometry by month and LBW is shown in Table 2.
During twelve months of follow-up, 86 (7.8%) infants died and 76 (6.9%) were lost to follow up after a median of 154 (IQR 91 to 247) days.The total period of observation was 1015 child/years (Figure 1).Twenty-six (30%) deaths occurred before two months of age.LBW was associated with an increased risk of death during the first year of life: hazards ratio (HR) 3.3 (95% CI 2.09 to 4.90) and P<0.001.
LBW and its association with underweight and mortality at two and six months of age LBW infants were persistently more underweight through the first 12 months of life; the proportion was highest at birth at 82% and lowest at age month 4 at 38% (Table 2).At two months, of the 148 (16%) underweight infants, 94 (64%) were LBW (Table 2).Being underweight was associated with mortality; adjusted hazard ratio (aHR) 1.75 (95% CI 1.04 to 2.79).Among underweight infants at two months, having been born LBW compared to NBW was associated with lower risk of mortality; aHR 2.63 (95% CI 0.76 to 9.15) (Figure 3).

Discussion
We set out to determine if LBW infants who are underweight at month two, which coincides with the age of first vaccination, could be assumed to be growing normally and at a lower risk of mortality than non-LBW underweight infants.We found that being underweight at two or six months of age is associated with a significantly increased risk of mortality irrespective of birth weight.Anthropometric changes observed since birth to two months were no better than a single measure at the point of vaccination in discriminating mortality risk.
LBW was associated with an increased risk of death during the first year of life.This risk is well-documented 14 .WHO guidelines recommend feeding LBW infants mother's breastmilk for the first six months of life 15 , as this is associated with lower incidence of infections and necrotizing enterocolitis than those fed with infant formula 16 .There is also strong evidence for supplementation with Vitamin D, Iron and Calcium among very low-birth weight (VLBW) infants who are fed on mother's breastmilk within the first six months of life 16 .In VLBW and LBW infants, vitamin D supplementation resulted in increases in height, weight, and MUAC in two randomised control trials 17,18 .These specific interventions for LBW infants need to be implemented; however, the impact of growth monitoring programmes on growth and mortality of LBW infants is less clear 15 .A LBW infant tracking below the reference line may be inappropriately considered as "protected" from the risk of mortality because they were born small and apparently growing 'normally'.Mothers of LBW infants where poor growth is recognised may be advised or initiate supplemental feeds before six months in an attempt to facilitate accelerated growth 19 , which may actually increase risk.There may also be maternal nutritional and health factors that influence feeding and care, which may usually be missed if the wellbeing of the mother is not assessed with that of her infant.Growth monitoring does present an opportunity if accompanied by informed assessment and appropriate action.
Emerging evidence suggests that proactive peer support to mothers of LBW infants improves compliance to exclusive breastfeeding in and out of the hospital environment 20 .Our results indicate that at two months, infants identified to be underweight, irrespective of birth weight, are at increased risk of mortality and should receive targeted support.Underweight infants with a history of LBW should receive micronutrient supplementation as currently recommended, noting the existing gaps in guidance 16 .
We found that a single MUAC, WAZ or LAZ measure taken at two months of age discriminate mortality risk better than WLZ and that a single measure was better than change from birth.This is an important finding given that WLZ is the currently recommended criterion for intervention and among LBW infants, health worker may consider change in anthropometry more important than the single measure taken at growth monitoring.Our results concur with studies of community infants in The Gambia 21 and BukinaFaso 3 , and from hospital infant cohort in Kenya 22 and India 23 where among infants u6m, WLZ is not reliably measured 24,25 possibly partly explaining its poor prediction of subsequently mortality.Although using growth velocity may be better at identifying risk, in practice repeated measures may be more complex to implement 26 .Current evidence suggests that using a MUAC cut-off of <11.0cm and WAZ<-3 when applied at two months (vaccination point) will effectively identify infants with a high risk of subsequent mortality 21,22 .As a simple tool, MUAC may be applied at home by either community health workers or family members to help identify at risk infants early 27 .
In many LMIC settings it is not feasible to distinguish LBW and NBW infants and so there is a need to identify early growth failure in infancy and determine intervention strategies appreciating that LBW infants will comprise a considerable proportion of these.To guide interventions, the Management of At risk Mothers and Infants (MAMI) care pathway approach leverages and connects existing services with active case identification and holistic management of the mother-infant dyad (Emergency Nutrition Network MAMI Tool 2018 available from: https://www.ennonline.net/c-mami).
A strength of this study was the large birth cohort with systematically collected monthly infant anthropometry up to one year of age and their vital status.However, the analysis at month two and six only used data from children with a measured anthropometry.Those excluded because of missing anthropometry could have differed with those included in this analysis.
In the absence of ultrasound in pregnancy we were not able to distinguish risks from being preterm versus small for gestational age.A further limitation is that risks and care provision may now differ from those when the data were collected.The data used in this analysis was collected in 2004 and may not reflect improvements in infants care practices in the health system such as improvements in vaccination, community management of acute malnutrition.

Conclusions
In the first year of life, LBW infants are more likely to be underweight and continue to be at higher risk of mortality than NBW infants.

Ekawaty Lutfia Haksari
Gadjah Mada University, Yogyakarta, Special Region of Yogyakarta, Indonesia Please give inclusion and exclusion criteria Are there any confounding variables in this study?If yes please give an explanation.Discussion: This section contains the author's findings, which are also presented in the Result section and are linked to the study questions.It should be compared with the result of the previous study (or studies) and use the most recent references.Also, it should be explained whether the result supports that of the previous study (or studies) or is in contrast to, or is brand new.The statements should, however, be elaborated and supported by adequate references.In regards to the above matters, the second paragraph should contain the results obtained, and then the results of the previous study/studies are subsequently presented.The investigators conducted a secondary analyses of an old dataset from Burkina Faso to address the question of whether mortality risk differed among underweight LBW infants compared to normal birth weight infants at 2 to 6 months.While this analysis attempts to address an important public health question, there are substantial limitations to the methods, inherent in the older data set (which the authors acknowledge).First, the data is from 2004, and there have been advances in anthropometric measurements since then.Women were only recruited in the 3rd trimester of pregnancy and did not have ultrasound-confirmed GA dating, so we have no information about growth restriction or potential confounders.Also the weight was measured in kg (not g) so there is likely substantial rounding issues and subtle differences may have been lost.Finally, less than half were delivered at a health facility, and the quality of the birth weight for home births is likely to be lower.It would be helpful to include an enrollment flow diagram to help interpret the lost to follow-up at each stage (i.e., from screening through 12 mos follow-up).Overall, while the analytic methods appear to be solid and the research question important, there are substantial limitations with the methods of the original study that reduce the validity of the study findings.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? Partly
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Perinatal epidemiology I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
6. Overall, while the analytic methods appear to be solid and the research question important, there are substantial limitations with the methods of the original study that reduce the validity of the study findings.
Response: It is important to remember that much as an old cohort may come with some limitations (duly acknowledged), we feel that its value far outweighs the limitations.Some of these advantages include; It would be impossible to replicate an untreated birth cohort in this day and age and hence impossible to observe nutritional trends as the current ethics restrictions would not allow.

○
To date, not many birth cohorts collect MUAC measurement at birth and in the first 6 months of life.The value of MUAC in early infancy is challenging to establish partly because of lack of data.A dataset like this gives an opportunity to explore the potential of expanding the use of MUAC among infants under 6 months.

○
Competing Interests: I am the author of the article providing a point by point response to the reviewer's comments Do the authors have information on breastfeeding practices, morbidity, care-seeking behaviors, and maternal nutrition status?If not, the authors should mention these in the limitations.If yes, I would suggest presenting this information to better understand the cohort.

3.
Why did the authors not use diagnostic accuracies to discriminate the mortality for anthropometry at different time points?4.
I would suggest the authors provide a flow diagram to better understand the loss to follow up at different stages.

Is the work clearly and accurately presented and does it cite the current literature? Partly
Is the study design appropriate and is the work technically sound?Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?Partly Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Drivers of child growth I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Figure 1 .
Figure 1.Flow diagram of study participants from recruitment to 12 months follow-up.

Figure 2 .
Figure 2. Month two medians of: A) WAZ, B) WLZ, C) MUAC and D) birth to month two weight difference.The red bars are the medians, for panel A and B, the dashed line is the cut-off of -2 and panel C MUAC=11 cm.

Figure 3 .
Figure 3. Cumulative hazard of death from two months of age among underweight infants by low birth weight.

Publisher Full Text | Free Full Text 27
To reduce risk of mortality among infants, research should focus on interventions to prevent LBW and on effective comprehensive interventions to reduce risks of mortality and promote neurodevelopment.Since LBW infants who are underweight have at least the mortality risk of non-LBW infants, all underweight infants need identification during screening or growth monitoring, to have individual nutritional, health and family assessment, and actions to address the risks associated with being underweight. .Ale FG, Phelan KP, Issa H, et al.:

the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? Yes If applicable, is the statistical analysis and its interpretation appropriate? Yes Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes Is the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? Yes If applicable, is the statistical analysis and its interpretation appropriate? Yes Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes
Reviewer Expertise: Drivers of child growthI confirm that I

have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard. Version 1
This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.