Survival of low birthweight neonates in Uganda: analysis of progress between 1995 and 2011

Survival of low birthweight neonates in Uganda: analysis of progress between 1995 and 2011 Background: Although low birthweight (LBW) babies represent only 15.5% of global births, it is the leading underlying cause of deaths among newborns in countries where neonatal mortality rates are high. In Uganda, like many other sub-Saharan African countries, the progress of reducing neonatal mortality has been slow and the contribution of low birthweight to neonatal deaths over time is unclear. The aim of this study is to investigate the association between low birthweight and neonatal mortality and to determine the trends of neonatal deaths attributable to low birthweight in Uganda between 1995 and 2011. Methods: Cross-sectional survey datasets from Uganda Demographic and Health Surveys between 1995 and 2011 were analyzed using binary logistic regression with 95% confidence interval (CI) and Kaplan-Meier survival analysis to examine associations and trends of neonatal mortalities with respect to LBW. A total of 5973 singleton last-born live births with measured birthweights were included in the study. Results: The odds of mortality among low birthweight neonates relative to normal birthweight babies were; in 1995, 6.2 (95% CI 2.3 −17.0), in 2000–2001, 5.3 (95% CI 1.7 −16.1), in 2006, 4.3 (95% CI 1.3 − 14.2) and in 2011, 3.8 (95% CI 1.3 − 11.2). The proportion of neonatal deaths attributable to LBW in the entire population declined by more than half, from 33.6% in 1995 to 15.3% in 2011. Neonatal mortality among LBW newborns also declined from 83.8% to 73.7% during the same period. Conclusion: Low birthweight contributes to a substantial proportion of neonatal deaths in Uganda. Although significant progress has been made to reduce newborn deaths, about three-quarters of all LBW neonates died in the neonatal period by 2011. This implies that the health system has been inadequate in its efforts to save LBW babies. A holistic strategy of community level interventions such as improved nutrition for pregnant mothers, prevention of teenage pregnancies, use of mosquito nets during pregnancy, antenatal care for all, adequate skilled care during birth to prevent birth asphyxia among LBW babies, and enhanced quality of postnatal care among others could effectively reduce the mortality numbers. Keywords: Low birthweight, Attributable neonatal mortality, Logistic regression, Kaplan-Meier survival analysis, Cross-sectional * Correspondence: arundamalachi@gmail.com Social Medicine and Global Health, Department of Clinical Sciences, Lund University, Jan Waldenströms gata 35, 205 02 Malmö, Sweden © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 2 of 12 Background 2006 retrospective demographic survey in Uganda esti- About 20 million low birthweight (LBW) babies are born mated that 60% of newborn deaths occurred at home [22]. every year, representing 15.5% of all births globally [1]. Over The Uganda roadmap for reducing neonatal mortality 95% of all LBW cases occur in low-income countries [1]. 2007–2015 fell short of incorporating LBW among the Of recent, Lawn et al. and the World Health Organization causes of neonatal deaths [21], possibly due to challenges (WHO) estimated that LBW contributes to 60–80% of in determining LBW-attributable deaths. No studies that all neonatal deaths (death within 28 days after birth) determined the national trends of LBW-attributable neo- worldwide [2, 3]. However, wider disparities in esti- natal mortality in Uganda were identified by our literature mates exist between countries. India, a low-to-middle search, despite being a key indicator of population and re- income country, contributes about 40% of global bur- productive health in a country [2]. However, in order for den of LBW babies [4], and in 2013, 48% of all neonatal Uganda to achieve the global Sustainable Development deaths in India were attributed to LBW and preterm Goal (SDG) target 3.2 that aims to drastically reduce neo- birth [5]. In comparison to Sweden, a high-income natal mortality by 2030 [23], the contribution of LBW to- country where neonatal mortality is very low (1.5 per wards neonatal mortality can no longer remain unclear. 1000 live births in 2014) [6], LBW babies constituted Although LBW is estimated to contribute about 80% of only 3.2% of national live birth in 2014, and barely 4.3% neonatal deaths in SSA [3], efforts to reduce neonatal of all neonatal deaths in 2014 were LBW cases [6]. mortality from the inception of the Millennium Devel- WHO defines LBW as birthweight of less than 2500 g opment Goals (MDGs) in 1990 to its end in 2015 in [1]. LBW is mainly a result of preterm births and re- Uganda have never been evaluated in terms of reduc- stricted fetal growth (resulting in small for gestational tion of LBW-attributable deaths. Further, there are no age (SGA) babies) or both [1]. The main risk factors national representative studies that have examined the leading to LBW include young mothers/short stature of contribution of LBW toward the overall neonatal mor- the mother [7], multiple births [8], poor nutrition before tality in Uganda. This present study thus aims to deter- conception and during pregnancy (poverty) [9], smoking mine both the association between LBW and neonatal [10], maternal HIV positivity, and malaria during preg- mortalityin Ugandaand to estimate thenationaltrends nancy [11, 12]. of LBW-attributable neonatal mortality between 1995 In sub-Saharan Africa (SSA), the general rate of de- and 2011. This period covered the entire MDG period cline in neonatal mortality (NM) has been slow com- except for the last 4 years to 2015. pared to infant or under-five mortality [13] and more than half of all births do not take place in health facil- Methods ities [14]. An individual participant level meta-analysis Study setting and maternal health situation study in four district projects within East Africa (EA) in With an annual population growth rate of about 3.2 and 2012 estimated that 52% of all neonatal deaths in Kenya, an overall fertility rate of 5.6, Uganda’s population rose Uganda, and Tanzania were attributable to preterm birth from about 17 million in the 1990s to about 34 million or small for gestational age, of which 99% were LBW in 2011 [24]. The sex ratio is 1:1 and the adolescence babies [15]. Several neonatal and infant mortality stud- fertility rate was about 131 per 1000 births in 2010 ies in SSA fall short of determining the contribution of [25]. Over 77% of the population live in rural areas. The LBW to neonatal deaths. Whereas LBW is the under- national poverty levels notably reduced from 38.8% in lying cause of majority of neonatal deaths, most studies 2002–2003 to about 20% in 2012–2013 [26]. However, have focused on other leading direct causes of neonatal povertylevelsdiffersignificantlybyregionand sub-regions. deaths such as birth asphyxia, infections, and preterm For instance, while incidence of poverty in the northern birth [16–18]. Another 5-year health facility-based study region in 2013 was 44%, it was only 5.1% in the central in Ghana estimated that LBW was a sole contributor of region [26]. In March 2001, Uganda abolished user fees 50% of neonatal deaths in the facility between 2008 and in first level government health facilities and this in- 2012 [19]. While LBW can be a result of preterm birth, it cluded maternal health services [27]. The proportion of is also a notable fetal risk factor for birth asphyxia and in- four or more antenatal care visits was still less than fections such as sepsis [17, 18]. 50% by 2011 [28]. Incrementally, by 2011 about 57% of total In Uganda, like in many SSA settings, apart from births took place in health facilities and the proportion of health system limitations such as inadequate resources, births that received post-natal care increased from less than paucity of data in hospital registries makes it difficult to 10% in 1995 to 26% in 2006 and to 32% in 2011 [28, 29]. determine the prevalence of LBW and associated mor- tality trends [20, 21]. The 2008 situation analysis report Study design and data source indicated that neonatal deaths were not registered in We obtained secondary data from repeated cross-sectional Uganda; no countrywide perinatal audit exists [20]. The surveys by the Demographic and Health Survey (DHS) Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 3 of 12 program. The datasets are independent and nationally rep- our sample in order to improve the statistical power of resentative. We used four datasets from the Uganda DHS our analyses. Records of the size of the babies registered birth recodes for the years 1995, 2000–2001, 2006, and as small or average among others were excluded from 2011. A total of 5973 singleton last-born live births with the study to minimize errors of misclassification due to birthweight measures were included in the study. This con- the unreliable subjective nature of the categorization cri- sisted of 1160 children in 1995 representing 25% of all the teria [35]. From the study’s selected samples, 72% of the last-born live births in the data sample for that year and 1160 selected sample in 1995 had birthweight from 1100 children for the year 2000–2001 representing 30% of mothers’ recall and the rest were from health cards. all the last-born live births in the sample for that year. Simi- Similarly, in 2000, 79% of the 1100 selected sample were larly, 1514 (35%) children were included for the year 2006 from recall. In 2006, 73.5% of the total 1514 were recall and 2199 (50%) for the year 2011. We targeted and utilized birthweights and in 2011, 67% of the total 2199 were re- the birth recode information for the last-born live births call birthweights. born within the 5-year period prior to each of the surveys. Preterm birth, LBW and birth asphyxia are highly cor- The Demographic and Health Survey (DHS) program em- related and it is difficult to determine their independent ploys standardized questionnaires and protocols that ensure contributions towards neonatal deaths. These three, to- that the participants remain anonymous [30, 31]. The DHS gether with infections, contribute to 80% of neonatal data collection procedure involves stratified two-stage clus- mortality as the highest cause of neonatal mortality, with ter sampling and collection of data countrywide using up- LBW being the underlying factor [36]. dated lists of enumeration areas for each of the surveys to avoid overlap and improve national representativeness of Maternal and socio-demographic variables thedata[32]. Further information on data sampling and In this study, independent variables that are known to collection criteria are detailed in the DHS field manuals be direct and indirect risk factors for neonatal mortality and methodology toolkits [30–32]. and LBW such as ‘young’ maternal age (7) and poor nu- trition (resulting from poverty and low or no education Variables (9) were investigated. Wealth status was determined as a Outcome variable composite cumulative living standard measured in terms of household asset inventory. These were investigated in Neonatal mortality This referred to death of newborn the univariate analysis to determine their distribution and within 28 days after birth. It was dichotomized into yes possible associations with birthweight and neonatal sur- (died) or no (alive). vival categories. Smoking was not examined due to lack of data. Figure 1 below shows a conceptual visualization Predictor variable of LBW as an overriding cause of the majority of neo- natal deaths. Low birthweight The variable low birthweight (LBW) Below (Table 1) is a summary of outcome and predictor was the predictor variable. Birthweight records were ob- variables and the covariates that influence the occurrence tained from the child’s health card or from the mother’s of low birthweight and the survival of neonates. verbal report of measured weight at birth. Birthweight was dichotomized into LBW (< 2500 g) or normal birth- Data analysis weight (NBW) ≥ 2500 g. Macrosomia (> 4000 g) [33] We used analytical software IBM SPSS version 24 and MS was eliminated in the univariate and logistic regression excel for analyses. Pearson’s chi square test of independ- analyses involving birthweights. The higher neonatal ence and association was used to examine the distribution mortality risks of macrosomia relative to NBW [34] of variables according to birthweight and neonatal mortal- would reduce the accuracy of our findings if they are in- ity for each survey. Survival plots of the birthweight cat- cluded among NBW numbers. At the hospital, newborns egories were generated using Kaplan-Meier’sestimator. are weighed and their birthweights recorded on the Binomial logistic regression analysis was used to determine child’s health card and is communicated. In contrast, for the odds ratios for the association between LBW and neo- births outside the health facility such as home births, natal mortality after adjusting for socio-demographic and birthweight is likely to be estimated by observing the maternal factors, cesarean births and check-ups for birth size of body parts, the accuracy of which is ques- pregnancy complications. The analysis was conducted tionable. To improve the accuracy of reported birth- at 5% significant level. In order to improve the validity weight, whether recall or from the health card, only of the results, the national representativeness of the hospital births were included in the study for the years data and to adjust for non-response, the complexity of 2000−2001, 2006, and 2011. For the 1995 dataset, how- DHS sampling design was taken into account, and data ever, we also included the very few home birth cases in sampling weights were applied to datasets for the years Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 4 of 12 Fig. 1 Conceptual visualization of potential risk factors leading to LBW and neonatal mortality. LBW – Low birthweight, SGA – Small for gestation age Table 1 Summary of variables Variables Categories Descriptions Outcome variable Neonatal mortality Yes (Dead) Died within age ≤ 1 month No (Alive) Alive at age ≥ 1 month Predictor variable Low birthweight Yes < 2500 g No ≥ 2500 g ≤ 4000 g Maternal and socio-economic variables Maternal age < 20 years 20–34 years 35–49 years Wealth status Poor Middle/rich Maternal education No education No formal education Primary < 9 years of education Secondary/higher ≥9 years of education Parity Primiparous First ever birth Para 2–32–3 children Para 4+ 4 or more children Marital status Single Never married, widowed, separated/divorce at delivery time, not living with the spouse Married Married or cohabiting Place of residence Rural Urban Cesarean birth No Yes Check-up for pregnancy complications No Yes Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 5 of 12 2000−2001, 2006, and 2011. However, the 1995 dataset amounts of missing data. Birth complications were also was not subjected to weighting due to the need to not adjusted for in 1995 due to absence of data. maintain the statistical power of the data for that year, Figure 2 below shows the relationship between birth- the implication of which is a very minimal difference. A weight and time-to-death among neonatal mortality total of 5973 last-born live births with birthweights cases, combining all the study years. In conjunction with were included in the analyses. the survival table (not included in the paper), we ob- served that over 85% of all neonatal deaths in our study Estimation of LBW-attributable mortality risk fraction sample occurred in the first week of life. About 95% of among LBW neonates and in the population all the LBW (< 2500 g) neonatal deaths occurred within The LBW-attributable neonatal mortality risk fraction the first week of life. In comparison, about 82% of deaths (AF) and population-attributable mortality risk fraction among neonates with NBW (2500 g ≤ 4000 g) took place (PAF) were computed as proportion of prevalent deaths within in the first weeks. The rest died later, in the sec- that could be avoided if LBW was prevented or the ond, third, and fourth weeks. The figure also shows an death of LBW babies was eliminated. These were calcu- inverse proportionality relationship between weight and lated manually using eqs. (1) and (2) below. survival. With the exception of an outlier, the neonates with higher birthweights tended to survive longer, i.e. OR−1 beyond the first week. AF ¼  100; ð1Þ The LBW-attributable neonatal mortality in Uganda OR declined by more than half, from 33.6% (%) in 1995 The population attributable mortality risk fraction to 15.3% in 2011 as shown in Table 5 below. Similarly, PAF, expressed as a percentage (%) was computed using LBW-attributable neonatal mortality among LBW babies the eq. (2). also declined by 10.2% from 83.9% to 73.7% in the same period. OR−1 Figure 3 shows a non-uniform but continuous decline PAF ¼ P AF ¼ P   100; ð2Þ e e of LBW-attributable neonatal mortality in Uganda be- OR tween 1995 and 2011. OR is the odds ratio generated from binary logistic re- gression analysis and Pe is the proportion of deaths that Discussion have the exposure. Overall, the odds of neonatal mortality among LBW babies as compared to normal birthweight were re- Results duced by a third, from about 6 times higher in 1995 Table 2 shows birthweight and maternal and socio- to 3.8 times higher in 2011. The LBW-attributable demographic characteristics of last-born live births by neonatal mortality in the population declined by more neonatal survival status in Uganda. Overall, the average than half, from 33.6% in 1995 to 15% in 2011. This proportion of neonatal deaths among LBW babies be- present study is the first of its kind in Uganda and tween 1995 and 2011 was about 3.5% while the average perhapsthe wholeofeastAfricathatexaminesthe proportion of neonatal deaths among normal weight ba- trends of LBW-attributable mortality over the years. bies (≥2500 g ≤ 4000 g) during the same period was less The study reinforces the very few LBW-related studies in than 1 %. Cesarean birth was associated with neonatal Uganda and east Africa by providing new peer-reviewed mortality only in the year 2000−2001 (p <0.05). findings on the contribution of LBW towards neonatal Table 3 shows the distribution of the study variables mortality countrywide over a period of over 15 years. The by birthweight. Statistical significantly higher propor- study findings might be useful for auditing the causes of tions (p < 0.05) of mothers with no formal education neonatal deaths, and for evaluation, future health planning had LBW babies in almost all the years except 2011. and policy making aimed at improving neonatal survival. Similarly, maternal age < 20 years of age was associated The WHO emphasizes that auditing the causes of neo- with having higher proportions of LBW babies as shown natal deaths is paramount for effective monitoring and in the 1995 and 2006 findings (p < 0.01). improving mother and child health care [37]. In all surveys, LBW was significantly associated with The 3.8 times higher odds of deaths among LBW neo- neonatal mortality as shown in Table 4 below. The ad- nates in 2011 in the present study is consistent with the justed odds ratio (AOR) for the years in question were findings of a related study conducted by Kananura et al. as follows: in 1995, 6.2 (95% CI (2.3 − 17.0), in 2000−2001, in eastern Uganda in 2012–2013 that indicated a 3.51 5.3 (95% CI 1.7 − 16.1), in 2006, 4.3 (1.3 − 14.2), and in mortality odds ratio [36]. Comparable findings were also 2011, 3.8 (95% CI 1.3 − 11.2). The 1995 and 2000–2001 obtained in a follow-up study in western Uganda, com- data were not adjusted for wealth status due to large pleted in 2006 but analyzed by Marchant et al. in 2012 Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 6 of 12 Table 2 Distribution of birthweight, maternal and sociodemographic characteristics by neonatal survival status in Uganda, 1995–2011 Variables 1995 2000–2001 2006 2011 Survival, N = 1160 Survival, N = 1100 Survival, N = 1514 Survival, N = (2199) Died Lived P value Died Lived P value Died Lived P value Died Lived P value n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) Birthweight < 2500 g 4 (3.3) 118 (96.7) < 0.01 5 (4.6) 104 (95.4) < 0.01 5 (2.8) 175 (97.2) < 0.05 7 (2.9) 234 (97.1) < 0.05 ≥ 2500 g 6 (0.6) 1032 (99.4) 10 (1.0) 981 (99.0) 11 (0.8) 1323 (99.2) 22 (1.1) 1936 (98.9) Maternal age < 20 1 (0.6) 155 (99.4) > 0.05 1 (0.9) 111 (99.1) > 0.05 2 (1.4) 138 (98.6) > 0.05 2 (1.3) 154 (98.7) > 0.05 20–34 6 (0.7) 855 (99.3) 12 (1.4) 825 (98.6) 11 (1.0) 1105 (99.0) 15 (1.0) 1496 (99.0) 35–49 3 (2.1) 140 (97.9) 3 (2.0) 148 (98.0) 2 (0.8) 254 (99.2) 7 (1.6) 427 (98.4) b b Wealth index n = 392 n = 424 Poor 1 (0.7) 137 (99.3) > 0.05 1 (0.5) 187 (99.5) > 0.05 4 (0.9) 442 (99.1) > 0.05 7 (1.1) 652 (98.9) > 0.05 Middle / Rich 4 (1.6) 250 (98.4) 3 (1.3) 233 (98.7) 11 (1.0) 1056 (99.0) 17 (1.1) 1426 (98.9) Maternal education No education 2 (1.5) 132 (98.5) > 0.05 2 (1.6) 124 (98.4) > 0.05 3 (1.6) 179 (98.4) > 0.05 2 (1.2) 171 (98.8) > 0.05 Primary 6 (0.9) 653 (99.1) 8 (1.6) 605 (98.4) 7 (0.8) 857 (99.2) 12 (1.0) 1149 (99.0) Secondary higher 2 (0.5) 365 (99.5) 5 (1.4) 356 (98.6) 5 (1.1) 462 (98.9) 11 (1.4) 757 (98.6) Parity Primiparous 3 (1.0) 296 (99.0) > 0.05 4 (1.4) 278 (98.6) > 0.05 6 (1.7) 356 (98.3) < 0.05 3 (0.7) 424 (99.3) > 0.05 Para 2–3 3 (0.6) 532 (99.4) 5 (1.0) 483 (99.0) 7 (1.1) 622 (98.9) 11 (1.2) 945 (98.8) Para 4+ 4 (1.2) 322 (98.8) 6 (1.8) 323 (98.2) 2 (0.4) 520 (99.6) 10 (1.4) 709 (98.6) Marital status Single 1 (0.5) 199 (99.5) > 0.05 2 (1.0) 198 (99.0) > 0.05 2 (0.7) 277 (99.3) > 0.05 3 (0.8) 354 (99.2) > 0.05 Married 9 (0.9) 951 (99.1) 14 (1.6) 887 (98.4) 13 (1.1) 1221 (98.9) 22 (1.3) 1722 (98.7) Residence Rural 5 (1.0) 517 (99.0) > 0.05 11 (1.5) 737 (98.5) > 0.05 10 (0.9) 1051 (99.1) > 0.05 17 (1.1) 1493 (98.9) > 0.05 Urban 5 (0.8) 633 (99.2) 4 (1.1) 348 (98.9) 5 (1.1) 447 (98.9) 7 (1.2) 584 (98.8) Delivery mode Cesarean 1 (1.4) 71 (98.6) > 0.05 4 (4.4) 87 (95.6) < 0.05 1(0.8) 122 (98.2) > 0.05 5(2.1) 230 (97.9) > 0.05 Normal 9 (0.8) 1079 (99.2) 12 (1.2) 995 (98.8) 14 (1.0) 1372 (99.0) 24 (1.2) 1940 (98.8) Check-up No No data 11 (1.5) 742 (98.5) > 0.05 6 (0.7) 866 (99.3) > 0.05 13 (1.5) 843 (98.5) > 0.05 Yes 4(1.2) 332 (98.2) 9 (1.4) 613 (98.6) 14 (1.1) 1261 (98.9) P values were generated from Chi square analysis. Statistical significance (p < 0.05, two-sided) complications The separate totals(n) for wealth index in 1995 and 2000 shows a deviation from the total (N) due to missing data [15]. This study estimated the odds of neonatal mortality the health personnel interviewed about perinatal out- among LBW newborns relative to NBW newborns at comes in the health units indicated that LBW contrib- 3.45 [15]. Our findings of 15.3% LBW-attributable neo- uted to 16% of the total newborn deaths [38]. However, natal mortality in 2011 in the population are comparable the study also acknowledged the underreporting of LBW to the findings of a situation analysis study conducted by as a cause of death due to overlaps with infections and the Ministry of Health (MoH) in Uganda in 2008 [38]. breathing difficulties [38]. The MoH study combined both quantitative and qualita- The results indicated a significantly higher proportion tive methods and collected data from 10 districts cover- of deaths among LBW babies and this corroborates with ing the four conventional regions (Central, Eastern, findings of other studies [2, 3] that show higher mortalities Western and Northern) in Uganda. In this MoH study, among LBW newborns relative to their NBW counterparts. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 7 of 12 Table 3 Univariate analysis of maternal and sociodemographic characteristics of neonates by birthweight in Uganda, 1995–2011 Variables 1995, N = 1160 2000–2001, N = 1100 2006, N = 1514 2011, N = 2199 LBW (%) NBW (%) P value LBW NBW P value LBW NBW P value LBW NBW P value Maternal age < 20 26(16.7) 130(83.3) < 0.01 15(13.4) 97(86.6) > 0.05 27(19.1) 114(80.9) < 0.01 20(12.7) 137(87.3) > 0.05 20–34 81(9.4) 780(90.6) 77(9.2) 761(90.8) 112(10.0) 1004(90.0) 174(11.5) 1337(88.5) 35–49 15(10.5) 128(89.5) 17(11.2) 135(88.8) 41(16.0) 216(84.0) 39(9.0) 395(91.0) Wealth n = 392 n = 424 Poor 15(10.9) 123(89.1) > 0.05 19(10.1) 169(89.9) > 0.05 61(13.7) 385(86.3) > 0.05 72(10.9) 587(89.1) > 0.05 Middle/rich 26(10.2) 228(89.8) 25(10.6) 211(89.4) 118(11.1) 949(88.9) 161(11.2) 1282(88.8) Education level No education 24(17.9) 110(82.1) < 0.01 21(16.7) 105(83.3) < 0.01 29(15.9) 153(84.1) < 0.05 27(15.6) 146(84.4) > 0.05 Primary 67(10.2) 592(89.8) 60(9.8) 555(90.2) 101(11.7) 763(88.3) 121(10.4) 1040(89.6) Secondary 31(8.4) 336(91.6) 28(7.8) 332(92.2) 49(10.5) 418(89.5) 85(11.1) 684(88.9) Parity Primiparous 45(15.1) 254(84.9) < 0.01 27(9.6) 255(90.4) > 0.05 50(13.8) 312(86.2) > 0.05 58(13.6) 368(86.4) > 0.05 Para 2–3 48(9.0) 487(91.0) 51(10.5) 437(89.5) 69(11.0) 560(89.0) 98(10.3) 858(89.7) Para 4+ 29(8.9) 297(91.1) 31(9.4) 300(90.6) 60(11.5) 462(88.5) 77(10.7) 643(89.3) Place of residence Rural 67(12.8) 455(87.2) < 0.05 76(10.1) 674(89.9) > 0.05 134(12.6) 928(87.4) > 0.05 167(11.1) 1343(88.9) > 0.05 Urban 55(8.6) 583(91.4) 33(9.4) 319(90.6) 46(10.2) 406(89.8) 66(11.1) 526 (88.9) Marital status Single 25(12.5) 175(87.5) > 0.05 29(14.5) 171(85.5) < 0.05 39(13.9) 241(86.1) > 0.05 36(10.1) 321(89.9) > 0.05 Married 97(10.1) 863(89.9) 80(8.9) 821(91.1) 141(11.4) 1093(88.6) 197(11.3) 1547(88.7) Cesarean Yes 4(5.6) 68(94.4) > 0.05 11(11.8) 82(88.2) > 0.05 24(19.7) 98(80.3) < 0.01 29(12.3) 206(87.7) > 0.05 No 118(10.8) 970(89.2) 99(9.8) 909(9.2) 154(11.1) 1232(88.9) 212(10.8) 1752(89.2) Check-up No No data 72(9.5) 683(90.5) > 0.05 105(12.0) 767(88.0) > 0.05 87(10.2) 769(89.8) > 0.05 Yes 34(10.1) 302(89.9) 73(11.8) 1315(88.1) 143(11.2) 1132(88.8) LBW refers to low birthweight (< 2500 g), NBW refers to normal birthweight (≥2500 g – 4000 g). P values were obtained from chi square test The separate totals (n) for wealth index in 1995 and 2000 shows a deviation from the total (N) due to missing data Although cesarean births have been associated with mortal- Figure 2 showed that about 85% of neonatal deaths oc- ity as also shown by the findings (p < 0.05) for the year curred in the first week after birth. This is close to the 2000–2001 in Table 2, in 2006 and 2011 however, the find- estimate of a recent MoH report on maternal, perinatal ings (p > 0.05) indicated improvements in obstetric services and child death review that indicated about 75% neonatal that has enabled the survival of many cesarean birth babies. deaths in the first week [39]. The inverse proportional Table 4 Logistic regression analysis showing association between low birthweight and neonatal mortality in Uganda, 1995 − 2011 Adjusted odds ratios (95% confidence interval) Variable 1995 2000−2001 2006 2011 N = 1160 N = 1100 N = 1519 N = 2223 Birthweight b b a a Low birthweight 6.2 (2.3 − 17.0) 5.3 (1.7 − 16.1) 4.3 (1.3 − 14.2) 3.8 (1.3 − 11.2) Normal birthweight 1.0 1.0 1.0 1.0 LBW refers to low birthweight < 2500 g, NBW refers to normal birthweight (≥2500 g – 4000 g) Adjusted for all socio-demographic, maternal, pregnancy and birth related factors in Table 1 Adjusted for all socio-demographic (except wealth status), maternal, pregnancy and birth related factors in the study (Table 1). Complications were not adjusted for in 1995 Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 8 of 12 Fig. 3 Graphical representation of low birthweight-attributable Fig. 2 Kaplan-Meier survival curves by birthweight for neonates in neonatal mortality trends in Uganda between 1995 and 2011 Uganda between 1995 and 2011. Cum - cumulative relationship indicated by the trends of birthweight versus findings are comparable with the LBW-attributable mor- time-to-death among neonatal deaths in Fig. 2 concurs tality estimates among LBW babies in the whole country with findings from a hospital-based study in Dhaka, in this present study. Bangladesh [40]. The findings in Fig. 2 also implied that Neonatal mortality accounts for about 40% of global the risk of neonatal death is inversely proportional to under-five mortality [44]. In Uganda, in recent years, it birthweight and are in agreement with several other studies was estimated that about 45,000 neonates die every year [40–43]. However, our data on age at death (days) appeared [20]. By extension of our findings, this corresponds to to have been aggregated in terms of 7 days (weekly) and approximately 7000 (15.3%) neonatal deaths attributable not the actual mortality days. This slightly compromised to LBW in 2011. Although our findings could be a slight the accuracy of the Kaplan Meier’s survival curve in our underestimation given the many unrecorded births (about study in terms of days of survival. 45% in 2011) [43] and unregistered neonatal deaths, they According to a facility-based study by Hedstrom et al. provide comparable national estimates that can be used in central Uganda that admitted neonates born between for advocacy and countrywide public health planning to December 2005 and September 2008, 89% of neonatal reduce LBW-attributable neonatal deaths. For instance, deaths among LBW neonates weighing under 1000 g the successful Kangaroo Mother Care project for prema- could be attributable to LBW [43]. Another study by ture and LBW newborns initiated by Uganda Newborn Marchant et al. [15] that utilized data collected in 2006 in Study project (UNEST) in 2007–2011 in Iganga and western Uganda also estimated a 71% LBW-attributable Mayuge district [45] could be implemented countrywide. neonatal mortality among LBW neonates. Both of these The greatest national decline of LBW-attributable mor- tality estimated in 2011 in our study is a notable finding that could be attributed to the efforts of the inter-agency Table 5 Low birthweight-attributable neonatal mortality risk national Newborn Steering Committee (NSC) [46]. The proportions in Uganda between 1995 and 2011 NSC, which was initiated in 2006, ensured rapid policy Year of survey Attributable risk fraction (%) adaptation and implementations both at the health facility Among LBW neonates (AF) 1995 83.9 and community levels in the few years to 2011 [46]. It was mandated by the MoH to spearhead comprehensive ser- 2000–2001 81.1 vice delivery and community-and health facility-based 2006 76.7 training [46, 47]. Our findings thus reveal that the policy 2011 73.7 changes and its implementation may have had a profound In the entire population (PAF) 1995 33.6 positive impact on the survival of LBW newborns during 2000–2001 27.0 this period. The findings indicate that it is possible to 2006 24.0 eliminate unnecessary neonatal deaths due to LBW and make significant contributions towards achieving the SDG 2011 15.3 3.2 target that aims to lower neonatal death rate to 12 per LBW low birthweight, AF Attributable Fraction, PAF Population Attributable Fraction 1000 live births by 2030 [23]. Further, both the present Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 9 of 12 study findings and the NSC initiative could be of keen There was no statistically significant association be- interest to similar countries (with high neonatal mortalities) tween place of residence, maternal education, marital for policy making and study replications with the aim of status, wealth status, maternal age, and neonatal mortality, improving LBW neonatal survival, for instance, in the (P > 0.05) (Table 2). Although studies vary in their Philippines, where the decline of neonatal deaths has findings concerning the association between these stagnated [48]. socio-demographic and maternal factors (including parity) Also, the Uganda Newborn Study (UNEST) Project and neonatal mortality [53], many study findings have in- partly contributed to the decline in mortality of LBW dicated an association between single motherhood [54], and preterm newborns in parts of eastern Uganda and teenage maternal age [55–57], lack of education [56], rural consequently contributed to the overall national decline residence [57] and neonatal mortality. A systematic review during this period [45]. of 17 studies up to the year 2013 in SSA [55]indicated The survival analysis indicated that the rate of decline that socio-demographic and maternal risk factors are in LBW-attributable mortality in the 5-year periods in- much more prevalent among teenage mothers as com- creased from 6.6% between 1995 and 2000–2001 to 8.7% pared to adult mothers [55]. With the decentralized sys- between 2006 and 2011 in the population (Table 5). tem in Uganda, further analytical research at the districts However, between the two periods, there was a signifi- or regional levels on the effect of socio-demographic fac- cant deceleration in the decline to 3.0% between 2000 tors on birthweight and neonatal deaths would provide and 2001 and 2006 (Fig. 3 and Table 5). This could po- more robust findings for monitoring, policy making and tentially be due to the 20% decline in the use of family interventions. However, at the national level, comprehen- planning methods among < 20 years old sexually active sive measurement and recording of birthweight need to girls during this period as noted by the analytical over- be made possible, irrespective of whether a child is born view of the Ugandan child report [49]. This could have at home or at the hospital. As a national policy driven ini- led to increased teenage pregnancies. LBW are common tiative, the provision of weighing scales to health volun- among teenage mothers (< 20 years) [7] and the mortal- teers and midwives at the community level, even on a ity among babies born to younger mothers in Uganda shared basis based on proximity and locality, is feasible was also notably high between 1995 and 2005 [22]. and could be very effective for monitoring neonatal health Nevertheless, our findings in Table 2 did not show any sig- countrywide. Apart from improving accuracy on birth- nificant higher mortality numbers among the < 20 years weight data collection, the availability of weighing scales old mothers, perhaps because of the few number of births could also be a profound campaign tool for lowering LBW in this age-group in our sample selection. However, statis- incidences by highlighting preventive measures. Afford- tically reasonable numbers in 2006 showed a significant able and easy to maintain mechanical weighing scales have association between primipara mothers (most of whom previously been used at the community level in over 400 were younger mothers (Table 3)) and neonatal mortality. villages in western Kenya [58]. Although it was on a small A study conducted by Andualem et al. in western Uganda scale, the initiative was profoundly successful, as shown by between 2005 and 2008 revealed that over 82% of female an increase in the birthweight measurements of newborns students had unmet sexual/reproductive health counseling of about 54%, from 43% to 97% [58]. The current study needs [50]. Lack of knowledge about the signs of pregnancy could thus give the impetus to communities and local or- complications has been linked to birth unpreparedness in ganizations to take initiatives and improve the survival of Uganda [51], a consequent risk factor for neonatal deaths, LBW neonates. Further, as LBW is an underlying cause of including LBW deaths. A comparative development study 60–80% of all neonatal deaths globally (2,3) and about by Kevin Croke [52] also highlighted the decline in the 15% of neonatal deaths in Uganda (present study 2011 health system gains in Uganda between 2001 and 2006 findings), continuous data collection on birthweights that due to political shocks related to removal of presiden- supports research, monitoring, and strategic preventive in- tial term limits. Financing of the health care system was terventions could be a formidable approach to curbing negatively affected. This could partly account for the neonatal deaths and overall health systems strengthening rise in LBW-neonatal deaths during this period. The both globally and in Uganda. specialized care of LBW babies requires extra financing Although our study largely indicated no significant as- compared to NBW. The direct impact of the decline in sociations between cesarean birth, pregnancy complica- health system gains on survival of LBW detected by the tions and neonatal mortality for most of the years, a present study is consistent with WHO/UNICEF obser- number of studies have found associations between vations that survival of LBW neonates, a high-risk in- cesarean births [57, 59], pregnancy complications [59] fant group, is among the most sensitive indicators to and neonatal deaths. There were inconsistencies in our assess the progress of maternal and child health status findings with regard to the significant associations be- in a country [2]. tween socio-demographic factors and LBW across all the Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 10 of 12 study years (p < or > 0.05) (Table 3). However, there were based on the physical size of the body parts such as foot higher proportions of LBW babies among teenage and length, chest or head [63]. A study in Uganda compared uneducated mothers in all the survey years. Teenage preg- the accuracy of a proxy measure of LBW by midwives in a nancy was associated with LBW only in 1995 and 2006. hospital-based setting showed an accuracy of over 80%. These findings corroborate study findings elsewhere in However, the study also noted the limitation that the find- rural India [60] and in several SSA countries [7, 61]that ings may not reflect the actual situation in the communi- strongly indicate that young maternal age is associated ties where less skilled community volunteers assist in with LBW. A study in Brazil, however, found an associ- most births, and their estimates of cut-offs are prone to ation between teenage pregnancy and LBW only when bias [63]. Elimination of macrosomic newborns improved marital partners (an economic factor) were lacking [62]. the validity of our findings. Although the 1995 data included both home and hos- Methodological considerations pital births, which undermined the consistency of the The random sampling of data across the entire country study methodology across years, preliminary analysis in- and the standardized nature of data collection method dicated that among the selected sample of newborns of the DHS strengthen the external validity of our study with birthweight measures in 1995, only 3.5% of the and enable global comparability among countries. births were home births (or perhaps on the way to the Weighting the data for the years 2000, 2006 and 2011en- hospital). The 1995 data thus has a reasonable degree of abled us to adjust for disproportionate sampling and consistency with other survey years. However, the se- non-response. This improved the national representa- lection of only hospital births in other survey years tiveness and validity of the study estimates. The 1995 improved the quality and validity of the findings for dataset was not weighted and the results for that year those years. are slightly less representative. However, the results are The recording of neonatal survival data from day 0 to still valid, due to the fact that there was only a small dif- 30 by the DHS allowed us to clearly categorize our out- ference when weighted and unweighted results of all the come variable and investigate risk factors across all the other years were compared. The national representative- survey years with consistency. Given the large number ness of the 1995 data was only dependent on the random of home births (about 50%) in all the surveys, both the sampling across the entire country and the standardized LBW and neonatal deaths were likely underreported. nature of DHS data collection for its reliability. The birthweight data are prone to rounding-off or ag- The repeated findings of significant associations be- gregation into 500 g-weight intervals which could have tween LBW and neonatal mortality across all surveys slightly compromised the accuracy of Kaplan-Meier’s confirm the existing evidence of association and the in- survival analysis in this study. This aggregation of data ternal validity of this present study. Nonetheless, our was observed in a study by Channon et al. [64]. How- study could not confirm the causal association because ever, the fact that over 90% of LBW neonatal deaths in the exact causes of newborn deaths were not ascertained our study occurred in the first week is quite consistent medically. The in-depth use of the nationally representa- with global WHO findings that 75% of neonatal deaths tive DHS datasets in this study has revealed the need to occur in the first week [65], given the high-risk group of improve data collection techniques and to include other LBW in a low-income country. similarly important variables such as diagnostic causes of death among individual children, for example, birth Conclusion asphyxia. Low birthweight is associated with neonatal mortality Another limitation of our study was that although hos- and contributes to a substantial proportion of neonatal pital births recorded and/or communicated birthweights, deaths in Uganda. Although significant progress has over 65% were from mothers` recall and the rest from been made to reduce newborn deaths attributed to the health card, and we cannot therefore completely dis- LBW, by 2011, about 74% of all LBW neonates died in miss the possibility of recall bias. This also applies to the the neonatal period. This implies that the health system 1995 data that included both hospital and home births. in place has been inadequate to meet the challenge of Nevertheless, child birth is a significant event in a ensuring LBW survival. There is also profound need to mother’s life and with our study selection of the most re- strengthen both birth and neonatal death registration ir- cent birth experience, there is a very high possibility that respective of whether the infants are born at home or at the mothers recalled correct birthweights. Moreover, for the health centers. The decentralized health system in the years 2000 to 2011, birthweight data concerned solely Uganda can enable community health workers (CHW) information regarding hospital born babies because these and the village health teams (VHT) in liaison with the were measured birthweights and not estimated weights as sub-counties and the districts to close the existing gaps in-home births, where birthweights are mainly estimated concerning neonatal birth and death audits. This will Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 11 of 12 enable robust and continuous research and monitoring Department of Health and Human Services regulations for the protection of human subjects. of the progress of LBW neonatal survival. Our study presents national estimates of risks and mortality trends Competing interests that provide national basis for continual evaluation and The authors declare that they have no competing interests. policy recommendations to prevent LBW and minimize risks of neonatal deaths. A holistic approach to reduce the incidence of preventable LBW babies could be fos- Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in tered to reduce these mortality rates. Viable fronts that published maps and institutional affiliations. could be strengthened include sexual education in schools to prevent teenage pregnancies, complementing Received: 26 April 2017 Accepted: 15 May 2018 nutritional diet of pregnant mothers, HIV testing, ensur- ing that all pregnant mothers use mosquito nets, training References of health workers, and promoting antenatal care visits and 1. United Nations Children’s Fund and World Health Organization. Low hospital births. Enhancing the quality of postnatal care birthweight: country, regional and global estimates. New York: UNICEF; 2004. could also reduce mortality incidence of LBW newborns. 2. World Health Organization. Care of the preterm and/or low-birth-weight newborn. [Updated 2017]. Available from: http://www.who.int/maternal_ Abbreviations child_adolescent/newborns/prematurity/en/. Accessed July 2017. AF: Attributable fraction; AOR: Adjusted odds ratio; CHW: Community health 3. Lawn J, Blencowe H, Oza S, You D, Lee A, Waiswa P, et al. Every newborn: workers; DHS: Demographic and health survey; EA: East Africa; LBW: Low progress, priorities, and potential beyond survival. Lancet. 2014;384(9938):189–205. birthweight; MDG: Millennium development goals; MoH: Ministry of Health; 4. Ministry of Health and Family Welfare (India). India Newborn Action Plan: NBW: Normal birthweight; NSC: Newborn Steering Committee; Government of India. New Delhi: 2014. PAF: Population attributable fraction; SDG: Sustainable development goal; 5. Office of the Registrar General and Census Commissioner (India). Causes of SGA: Small for gestational age; SSA: Sub-Saharan Africa; UNAP: Uganda death in India 2010–2013 [data table online]. c2010–2013. Available from: nutrition action plan; UNEST: Uganda newborn study; VHT: Village health http://www.censusindia.gov.in/2011-common/Sample_Registration_System. teams html. Accessed 11 Nov 2016. 6. Official Statistics of Sweden. Statistics – Health and Medical Care (Sweden), Acknowledgements 2015. Pregnancies, Deliveries and Newborn Infants. The Swedish Medical Special gratitude to the Demographic and Health Survey program for Birth Register 1973–2014 Assisted Reproduction, treatment 1991–2013. making the datasets available for this study. 7. Chibwesha C, Zanolini A, Smid M, Vwalika B, Phiri Kasaro M, Mwanahamuntu M, et al. Predictors and outcomes of low birth weight in Availability of data and materials Lusaka, Zambia. Int J Gynecol Obstet. 2016;134(3):309–14. Authorization to use the secondary, unrestricted Uganda DHS data was 8. Nkyekyer K. Twin and singleton births in Ghana — a case-control study. obtained from MEASURE DHS after a requisition was made through their Twin Res. 2002;5(04):265–9. webpage by registration, selection of country and submission of a brief 9. Kayode G, Amoakoh-Coleman M, Agyepong I, Ansah E, Grobbee D, description of the study. The DHS program is authorized by the country, Klipstein-Grobusch K. Contextual risk factors for low birth weight: a Uganda to distribute data for research aimed at improving public health. multilevel analysis. PLoS One. 2014;9(10):e109333. Procedures and questions for standard DHS surveys have been approved by 10. Juárez S, Merlo J. Revisiting the effect of maternal smoking during both ICF Institutional Review board and the review board in the host country pregnancy on offspring birthweight: a quasi-experimental sibling analysis in Uganda. Both of which ensure that DHS complies with the protection of the Sweden. PLoS One. 2013;8(4):e61734. dignity and rights of human participants and the norms of the country. 11. Xiao P, Zhou Y, Chen Y, Yang M, Song X, Shi Y, et al. Association between maternal HIV infection and low birth weight and prematurity: a meta- Authors’ contributions analysis of cohort studies. BMC Pregnancy Childbirth. 2016;16:243. https:// MOA Conceptualized the research study, developed the study design, doi.org/10.1186/s12884-016-1044-3. acquired data, analyzed and drafted the manuscript. AA Contributed to the 12. Guyatt H, Snow R. Impact of malaria during pregnancy on low birth weight interpretation of data and to the writing of the manuscript. BOA Involved in in sub-Saharan Africa. Clin Microbiol Rev. 2004;17(4):760–9. developing the study design, interpretation and contributed to the writing 13. Registrar General of India and the Centre for Global Health Research. Causes of the manuscript. MOA, AA, BOA Read and approved the final draft of the of Death Statistics 2010-2013. [Updated 2015]. Available from: http://www. manuscript. cghr.org/2015/12/causes-of-death-in-india-2010-2013/. 14. Lawn JE, Cousens SN, Wilczynska K. Estimating the causes of four million Ethics approval and consent to participate neonatal deaths in the year 2000: statistical annex—the world health report Approval for utilization of Uganda DHS data for this study was obtained 2005. Geneva: World Health Organization; 2005. from ICF Macro Int. Data was accessible from: https://dhsprogram.com/Data/. 15. Marchant T, Willey B, Katz J, Clarke S, Kariuki S, Ft K, et al. Neonatal mortality Uganda National guidelines for secondary (third party) research that use risk associated with preterm birth in East Africa, adjusted by weight for unrestricted, publicly available and unlinked data that does not identify gestational age: individual participant level meta-analysis. PLoS Med. 2012; individuals or communities are exempted from ethical approval requirement. 9(8):e1001292. During data collection process by the primary data collector, ICF Macro Int, a 16. Welaga P, Moyer C, Aborigo R, Adongo P, Williams J, Hodgson A, et al. Why written informed consent was sought for each willing participant by are babies dying in the first month after birth? A 7-year study of neonatal MEASURE DHS before data collection commenced. The statement of mortality in northern Ghana. PLoS One. 2013;8(3):e58924. consent emphasize to the participant that they are at liberty to terminate or 17. Aslam H, Saleem S, Afzal R, Iqbal U, Saleem S, Shaikh M, et al. Risk factors of decline to respond at any stage of the interview process. Those unwilling birth asphyxia. Ital J Pediatr. 2014;40:94. https://doi.org/10.1186/s13052-014- were excluded from the study. The participants remained anonymous and 0094-2. the DHS approved procedures ensure that future identification of any 18. Shah G, Budhathoki S, Das B, Mandal R. Risk factors in early neonatal sepsis. individual participant or community is practically impossible. These measures Kathmandu Univ Med J (KUMJ). 2006;4(2):187–91. are thus in adherence with both the international guidelines for 19. Siakwa M, Kpikpitse D, Laryea T, Ankobil A, Dare S, Ebu N. A Five-year epidemiological studies of the Council of International Organizations of neonatal mortality trend in a Ghanaian teaching hospital after the Medical Sciences (CIOMS) and the Uganda National guidelines for research implementation of strategies to achieve the millennium development goal. involving human subjects. Further, the measures abide by the United States Int J Pediatrics Child Health. 2014;2(1):043–9. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 12 of 12 20. Ministry of Health. Situation analysis of newborn health in Uganda: current 45. Waiswa P, Peterson S, Namazzi G, Ekirapa E, Naikoba S, Byaruhanga R, et al. status and opportunities to improve care and survival. Kampala: The Uganda newborn study (UNEST): an effectiveness study on improving government of Uganda. Save the children, UNICEF, WHO; 2008. newborn health and survival in rural Uganda through a community-based 21. The Republic of Uganda: Roadmap for Accelerating the Reduction of intervention linked to health facilities - study protocol for a cluster Maternal and Neonatal Mortality and Morbidity in Uganda: Government of randomized controlled trial. Trials. 2012;13:213. https://doi.org/10.1186/1745- Uganda; 2007–2015. 6215-13-213. 46. Newborn Steering Committee. Kampala. Uganda: Save the Children, Healthy 22. Uganda Bureau of Statistics (UBOS) and Macro International Inc. Uganda Child Uganda, UNICEF, on behalf of the national Newborn Steering Demographic and Health Survey 2006. Calverton: UBOS and Macro Committee: 2011. Newborn Health Implementation Framework Standards International Inc.; 2007. for Newborn Health Services: Health Facility Assessments. 23. Sustainable development goals - United Nations. United Nations Sustainable 47. Mbonye A, Sentongo M, Mukasa G, Byaruhanga R, Sentumbwe-Mugisa O, Development. 2016. Available from: http://www.un.org/sustainabledevelopment/ Waiswa P, et al. Newborn survival in Uganda: a decade of change and sustainable-development-goals/. Accessed 01 Dec 2016. future implications. Health Policy Plan. 2012;27(suppl 3):iii104–17. 24. World Population Review. Uganda Population 2016. [Updated 2016]. 48. Kraft A, Nguyen K, Jimenez-Soto E, Hodge A. Stagnant neonatal mortality Available from: http://worldpopulationreview.com. Accessed 20 Nov 2016. and persistent health inequality in middle-income countries: a case study of 25. The World Bank. World Development Indicators (WDI). [Updated 2016]. the Philippines. PLoS One. 2013;8(1):e53696. Available from: http://databank.worldbank.org/data/reports.aspx?source= 49. Walakira E.J.D., Muhangi S.. Munyuwiny F,. Matovu E,. Awich I., Ddumba world-development-indicators. Accessed 21 Nov 2016. Nyanzi J. et al. (2016). The state of the Ugandan child – an analytical 26. Uganda Bureau of Statistics (UBOS) (Uganda), Republic of Uganda; 2014. overview. Kampala/Washington DC: USAID/QED. Statistical Abstract. 50. Boltena A, Khan F, Asamoah B, Agardh A. Barriers faced by Ugandan 27. Xu K, Evans DB, Kadama P, Nabyonga J, Ogwal PO, Nabukhonzo P, et university students in seeking medical care and sexual health counselling: a al. Understanding the impact of eliminating user fees: utilization and cross-sectional study. BMC Public Health. 2012;12:986. https://doi.org/10. catastrophic health expenditures in Uganda. Soc Sci Med. 1186/1471-2458-12-986. 2006;62(4):866–76. 51. Sandberg J, Odberg Pettersson K, Asp G, Kabakyenga J, Agardh A. 28. Uganda Bureau of Statistics (UBOS) and ICF International Inc. 2012. Uganda Inadequate knowledge of neonatal danger signs among recently delivered Demographic and Health Survey 2011. Kampala: UBOS and Calverton: ICF women in southwestern rural Uganda: a community survey. PLoS One. International Inc. 2014;9(5):e97253. 29. Statistics Department/Uganda and Macro International. Uganda 52. Croke K. The political economy of child mortality decline in Tanzania and Demographic and Health Survey, 1995. Calverton: Statistics Department/ Uganda, 1995–2007. Stud Comp Int Dev. 2012;47(4):441–63. Uganda and Macro International; 1996. 53. Berhan Y, Berhan A. A meta-analysis of socio-demographic factors for 30. Measure DHS. Guide to demographic and health survey statistics: perinatal mortality in developing countries: a subgroup analysis of methodology for the measure DHS phase III project, implemented from the National Surveys and small-scale studies. Ethiop J Health Sci. 2008–2013. 2006. 2014;24(0):41. 31. ICF International. MEASURE DHS Biomarker Field Manual. Calverton: ICF 54. Izugbara C. Single motherhood and neonatal and infant mortality in Sierra International; 2012. Leone, Burkina Faso and Burundi. Public Health. 2016;135:122–30. 32. Macro international Inc. Sampling manual. DHS-III basic documentation no. 55. Ramaiya A, Kiss L, Baraitser P, Mbaruku G, Hildon Z. A systematic review of 6. Calverton: Macro international Inc; 1996. risk factors for neonatal mortality in adolescent Mother’s in sub Saharan 33. Ota E, Haruna M, Suzuki M, Anh D, Tho L, Tam N, et al. Maternal body mass Africa. BMC Res Notes. 2014;7(1):750. index and gestational weight gain and their association with perinatal 56. Mekonnen Y, Tensou B, Telake D, Degefie T, Bekele A. Neonatal mortality in outcomes in Viet Nam. Bull World Health Organ. 2011;89(2):127–36. Ethiopia: trends and determinants. BMC Public Health. 2013;13:483. https:// 34. Said A, Manji K. Risk factors and outcomes of fetal macrosomia in a doi.org/10.1186/1471-2458-13-483. tertiary Centre in Tanzania: a case-control study. BMC Pregnancy 57. Ezeh O, Agho K, Dibley M, Hall J, Page A. Determinants of neonatal mortality in Childbirth. 2016;16:243. https://doi.org/10.1186/s12884-016-1044-3. Nigeria: evidence from the 2008 demographic and health survey. BMC Public 35. Blanc AK, Wardlaw T. Monitoring low birth weight: an evaluation of Health. 2014;14:521. https://doi.org/10.1186/1471-2458-14-521. international estimates and an updated estimation procedure. Bull World 58. Gisore P, Shipala E, Otieno K, Rono B, Marete I, Tenge C, et al. Community Health Organ. 2005;83(3):178–85. based weighing of newborns and use of mobile phones by village elders in 36. Kananura R, Tetui M, Mutebi A, Bua J, Waiswa P, Kiwanuka S, et al. The neonatal rural settings in Kenya: a decentralised approach to health care provision. BMC mortality and its determinants in rural communities of eastern Uganda. Reprod Pregnancy Childbirth. 2012;12:15. https://doi.org/10.1186/1471-2393-12-15. Health. 2016;13:13. https://doi.org/10.1186/s12978-016-0119-y. 59. Arunda M, Emmelin A, Asamoah B. Effectiveness of antenatal care services 37. World Health Organization. Making every baby count: audit and review of in reducing neonatal mortality in Kenya: analysis of national survey data. stillbirths and neonatal deaths, 2016. Geneva: World Health Organization; 2016. Glob Health Action. 2017;10(1):1328796. 38. Ministry of Health. Situation analysis of newborn health in Uganda. Kampala: 60. Borah M, Agarwalla R. Maternal and socio-demographic determinants of Government of Uganda; 2008. low birth weight (LBW): a community-based study in a rural block of Assam. 39. Ministry of Health. Situation analysis of newborn health in Uganda: current J Postgrad Med. 2016;62(3):178. status and opportunities to improve care and survival. Kampala: 61. Mombo-Ngoma G, Mackanga J, González R, Ouedraogo S, Kakolwa M, Government of Uganda. Save the Children, UNICEF, WHO; 2008. Manego R, et al. Young adolescent girls are at high risk for adverse 40. Yasmin S, Osrin D, Paul E, Costello A. Neonatal mortality of low-birth-weight pregnancy outcomes in sub-Saharan Africa: an observational multicountry infants in Bangladesh. Bull World Health Organ. 2001;79(7):608–14. study. BMJ Open. 2016;6(6):e011783. 41. Golestan M, Fallah R, Akhavan Karbasi S. Neonatal mortality of low birth 62. Guimarães A, Bettiol H, Souza L, Gurgel R, Almeida M, Ribeiro E, et al. Is weight infants in Yazd — Iran. Early Hum Dev. 2008;84:S26. adolescent pregnancy a risk factor for low birth weight? Rev Saude Publica. 42. Gaiva M, Fujimori E, Sato A. Neonatal mortality in infants with low 2013;47(1):11–9. birthweight. Revista da Escola de Enfermagem da USP. 2014;48(5):778–86. 63. Elizabeth N, Christopher O, Patrick K. Determining an anthropometric 43. Hedstrom A, Ryman T, Otai C, Nyonyintono J, McAdams R, Lester D, et al. surrogate measure for identifying low birth weight babies in Uganda: a Demographics, clinical characteristics and neonatal outcomes in a rural hospital-based cross-sectional study. BMC Pediatr. 2013;13:54. https://doi. Ugandan NICU. BMC Pregnancy Childbirth. 2014;14:327. https://doi.org/10. org/10.1186/1471-2431-13-54. 1186/1471-2393-14-327. 64. Channon A, Padmadas S, McDonald J. Measuring birth weight in 44. Wang H, Liddell CA, Coates MM, Mooney MD, Levitz CE, Schumacher developing countries: does the method of reporting in retrospective AE, et al. Global, regional, and national levels of neonatal, infant, and surveys matter? Matern Child Health J. 2010;15(1):12–8. under-5 mortalities during 1990-2013: a systematic analysis for the 65. Lawn JE, Cousens S, Zupan J. 4 million neonatal deaths: when? Where? global burden of disease study 2013. The Lancet (London, England). Why? Lancet. 2005;365:891–900. https://doi.org/10.1016/S0140- 2014;384(9947):957–79. 6736(05)71048-5. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Pregnancy and Childbirth Springer Journals

Survival of low birthweight neonates in Uganda: analysis of progress between 1995 and 2011

Free
12 pages

Loading next page...
 
/lp/springer_journal/survival-of-low-birthweight-neonates-in-uganda-analysis-of-progress-pywqmXIdWC
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Medicine & Public Health; Reproductive Medicine; Maternal and Child Health; Gynecology
eISSN
1471-2393
D.O.I.
10.1186/s12884-018-1831-0
Publisher site
See Article on Publisher Site

Abstract

Background: Although low birthweight (LBW) babies represent only 15.5% of global births, it is the leading underlying cause of deaths among newborns in countries where neonatal mortality rates are high. In Uganda, like many other sub-Saharan African countries, the progress of reducing neonatal mortality has been slow and the contribution of low birthweight to neonatal deaths over time is unclear. The aim of this study is to investigate the association between low birthweight and neonatal mortality and to determine the trends of neonatal deaths attributable to low birthweight in Uganda between 1995 and 2011. Methods: Cross-sectional survey datasets from Uganda Demographic and Health Surveys between 1995 and 2011 were analyzed using binary logistic regression with 95% confidence interval (CI) and Kaplan-Meier survival analysis to examine associations and trends of neonatal mortalities with respect to LBW. A total of 5973 singleton last-born live births with measured birthweights were included in the study. Results: The odds of mortality among low birthweight neonates relative to normal birthweight babies were; in 1995, 6.2 (95% CI 2.3 −17.0), in 2000–2001, 5.3 (95% CI 1.7 −16.1), in 2006, 4.3 (95% CI 1.3 − 14.2) and in 2011, 3.8 (95% CI 1.3 − 11.2). The proportion of neonatal deaths attributable to LBW in the entire population declined by more than half, from 33.6% in 1995 to 15.3% in 2011. Neonatal mortality among LBW newborns also declined from 83.8% to 73.7% during the same period. Conclusion: Low birthweight contributes to a substantial proportion of neonatal deaths in Uganda. Although significant progress has been made to reduce newborn deaths, about three-quarters of all LBW neonates died in the neonatal period by 2011. This implies that the health system has been inadequate in its efforts to save LBW babies. A holistic strategy of community level interventions such as improved nutrition for pregnant mothers, prevention of teenage pregnancies, use of mosquito nets during pregnancy, antenatal care for all, adequate skilled care during birth to prevent birth asphyxia among LBW babies, and enhanced quality of postnatal care among others could effectively reduce the mortality numbers. Keywords: Low birthweight, Attributable neonatal mortality, Logistic regression, Kaplan-Meier survival analysis, Cross-sectional * Correspondence: arundamalachi@gmail.com Social Medicine and Global Health, Department of Clinical Sciences, Lund University, Jan Waldenströms gata 35, 205 02 Malmö, Sweden © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 2 of 12 Background 2006 retrospective demographic survey in Uganda esti- About 20 million low birthweight (LBW) babies are born mated that 60% of newborn deaths occurred at home [22]. every year, representing 15.5% of all births globally [1]. Over The Uganda roadmap for reducing neonatal mortality 95% of all LBW cases occur in low-income countries [1]. 2007–2015 fell short of incorporating LBW among the Of recent, Lawn et al. and the World Health Organization causes of neonatal deaths [21], possibly due to challenges (WHO) estimated that LBW contributes to 60–80% of in determining LBW-attributable deaths. No studies that all neonatal deaths (death within 28 days after birth) determined the national trends of LBW-attributable neo- worldwide [2, 3]. However, wider disparities in esti- natal mortality in Uganda were identified by our literature mates exist between countries. India, a low-to-middle search, despite being a key indicator of population and re- income country, contributes about 40% of global bur- productive health in a country [2]. However, in order for den of LBW babies [4], and in 2013, 48% of all neonatal Uganda to achieve the global Sustainable Development deaths in India were attributed to LBW and preterm Goal (SDG) target 3.2 that aims to drastically reduce neo- birth [5]. In comparison to Sweden, a high-income natal mortality by 2030 [23], the contribution of LBW to- country where neonatal mortality is very low (1.5 per wards neonatal mortality can no longer remain unclear. 1000 live births in 2014) [6], LBW babies constituted Although LBW is estimated to contribute about 80% of only 3.2% of national live birth in 2014, and barely 4.3% neonatal deaths in SSA [3], efforts to reduce neonatal of all neonatal deaths in 2014 were LBW cases [6]. mortality from the inception of the Millennium Devel- WHO defines LBW as birthweight of less than 2500 g opment Goals (MDGs) in 1990 to its end in 2015 in [1]. LBW is mainly a result of preterm births and re- Uganda have never been evaluated in terms of reduc- stricted fetal growth (resulting in small for gestational tion of LBW-attributable deaths. Further, there are no age (SGA) babies) or both [1]. The main risk factors national representative studies that have examined the leading to LBW include young mothers/short stature of contribution of LBW toward the overall neonatal mor- the mother [7], multiple births [8], poor nutrition before tality in Uganda. This present study thus aims to deter- conception and during pregnancy (poverty) [9], smoking mine both the association between LBW and neonatal [10], maternal HIV positivity, and malaria during preg- mortalityin Ugandaand to estimate thenationaltrends nancy [11, 12]. of LBW-attributable neonatal mortality between 1995 In sub-Saharan Africa (SSA), the general rate of de- and 2011. This period covered the entire MDG period cline in neonatal mortality (NM) has been slow com- except for the last 4 years to 2015. pared to infant or under-five mortality [13] and more than half of all births do not take place in health facil- Methods ities [14]. An individual participant level meta-analysis Study setting and maternal health situation study in four district projects within East Africa (EA) in With an annual population growth rate of about 3.2 and 2012 estimated that 52% of all neonatal deaths in Kenya, an overall fertility rate of 5.6, Uganda’s population rose Uganda, and Tanzania were attributable to preterm birth from about 17 million in the 1990s to about 34 million or small for gestational age, of which 99% were LBW in 2011 [24]. The sex ratio is 1:1 and the adolescence babies [15]. Several neonatal and infant mortality stud- fertility rate was about 131 per 1000 births in 2010 ies in SSA fall short of determining the contribution of [25]. Over 77% of the population live in rural areas. The LBW to neonatal deaths. Whereas LBW is the under- national poverty levels notably reduced from 38.8% in lying cause of majority of neonatal deaths, most studies 2002–2003 to about 20% in 2012–2013 [26]. However, have focused on other leading direct causes of neonatal povertylevelsdiffersignificantlybyregionand sub-regions. deaths such as birth asphyxia, infections, and preterm For instance, while incidence of poverty in the northern birth [16–18]. Another 5-year health facility-based study region in 2013 was 44%, it was only 5.1% in the central in Ghana estimated that LBW was a sole contributor of region [26]. In March 2001, Uganda abolished user fees 50% of neonatal deaths in the facility between 2008 and in first level government health facilities and this in- 2012 [19]. While LBW can be a result of preterm birth, it cluded maternal health services [27]. The proportion of is also a notable fetal risk factor for birth asphyxia and in- four or more antenatal care visits was still less than fections such as sepsis [17, 18]. 50% by 2011 [28]. Incrementally, by 2011 about 57% of total In Uganda, like in many SSA settings, apart from births took place in health facilities and the proportion of health system limitations such as inadequate resources, births that received post-natal care increased from less than paucity of data in hospital registries makes it difficult to 10% in 1995 to 26% in 2006 and to 32% in 2011 [28, 29]. determine the prevalence of LBW and associated mor- tality trends [20, 21]. The 2008 situation analysis report Study design and data source indicated that neonatal deaths were not registered in We obtained secondary data from repeated cross-sectional Uganda; no countrywide perinatal audit exists [20]. The surveys by the Demographic and Health Survey (DHS) Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 3 of 12 program. The datasets are independent and nationally rep- our sample in order to improve the statistical power of resentative. We used four datasets from the Uganda DHS our analyses. Records of the size of the babies registered birth recodes for the years 1995, 2000–2001, 2006, and as small or average among others were excluded from 2011. A total of 5973 singleton last-born live births with the study to minimize errors of misclassification due to birthweight measures were included in the study. This con- the unreliable subjective nature of the categorization cri- sisted of 1160 children in 1995 representing 25% of all the teria [35]. From the study’s selected samples, 72% of the last-born live births in the data sample for that year and 1160 selected sample in 1995 had birthweight from 1100 children for the year 2000–2001 representing 30% of mothers’ recall and the rest were from health cards. all the last-born live births in the sample for that year. Simi- Similarly, in 2000, 79% of the 1100 selected sample were larly, 1514 (35%) children were included for the year 2006 from recall. In 2006, 73.5% of the total 1514 were recall and 2199 (50%) for the year 2011. We targeted and utilized birthweights and in 2011, 67% of the total 2199 were re- the birth recode information for the last-born live births call birthweights. born within the 5-year period prior to each of the surveys. Preterm birth, LBW and birth asphyxia are highly cor- The Demographic and Health Survey (DHS) program em- related and it is difficult to determine their independent ploys standardized questionnaires and protocols that ensure contributions towards neonatal deaths. These three, to- that the participants remain anonymous [30, 31]. The DHS gether with infections, contribute to 80% of neonatal data collection procedure involves stratified two-stage clus- mortality as the highest cause of neonatal mortality, with ter sampling and collection of data countrywide using up- LBW being the underlying factor [36]. dated lists of enumeration areas for each of the surveys to avoid overlap and improve national representativeness of Maternal and socio-demographic variables thedata[32]. Further information on data sampling and In this study, independent variables that are known to collection criteria are detailed in the DHS field manuals be direct and indirect risk factors for neonatal mortality and methodology toolkits [30–32]. and LBW such as ‘young’ maternal age (7) and poor nu- trition (resulting from poverty and low or no education Variables (9) were investigated. Wealth status was determined as a Outcome variable composite cumulative living standard measured in terms of household asset inventory. These were investigated in Neonatal mortality This referred to death of newborn the univariate analysis to determine their distribution and within 28 days after birth. It was dichotomized into yes possible associations with birthweight and neonatal sur- (died) or no (alive). vival categories. Smoking was not examined due to lack of data. Figure 1 below shows a conceptual visualization Predictor variable of LBW as an overriding cause of the majority of neo- natal deaths. Low birthweight The variable low birthweight (LBW) Below (Table 1) is a summary of outcome and predictor was the predictor variable. Birthweight records were ob- variables and the covariates that influence the occurrence tained from the child’s health card or from the mother’s of low birthweight and the survival of neonates. verbal report of measured weight at birth. Birthweight was dichotomized into LBW (< 2500 g) or normal birth- Data analysis weight (NBW) ≥ 2500 g. Macrosomia (> 4000 g) [33] We used analytical software IBM SPSS version 24 and MS was eliminated in the univariate and logistic regression excel for analyses. Pearson’s chi square test of independ- analyses involving birthweights. The higher neonatal ence and association was used to examine the distribution mortality risks of macrosomia relative to NBW [34] of variables according to birthweight and neonatal mortal- would reduce the accuracy of our findings if they are in- ity for each survey. Survival plots of the birthweight cat- cluded among NBW numbers. At the hospital, newborns egories were generated using Kaplan-Meier’sestimator. are weighed and their birthweights recorded on the Binomial logistic regression analysis was used to determine child’s health card and is communicated. In contrast, for the odds ratios for the association between LBW and neo- births outside the health facility such as home births, natal mortality after adjusting for socio-demographic and birthweight is likely to be estimated by observing the maternal factors, cesarean births and check-ups for birth size of body parts, the accuracy of which is ques- pregnancy complications. The analysis was conducted tionable. To improve the accuracy of reported birth- at 5% significant level. In order to improve the validity weight, whether recall or from the health card, only of the results, the national representativeness of the hospital births were included in the study for the years data and to adjust for non-response, the complexity of 2000−2001, 2006, and 2011. For the 1995 dataset, how- DHS sampling design was taken into account, and data ever, we also included the very few home birth cases in sampling weights were applied to datasets for the years Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 4 of 12 Fig. 1 Conceptual visualization of potential risk factors leading to LBW and neonatal mortality. LBW – Low birthweight, SGA – Small for gestation age Table 1 Summary of variables Variables Categories Descriptions Outcome variable Neonatal mortality Yes (Dead) Died within age ≤ 1 month No (Alive) Alive at age ≥ 1 month Predictor variable Low birthweight Yes < 2500 g No ≥ 2500 g ≤ 4000 g Maternal and socio-economic variables Maternal age < 20 years 20–34 years 35–49 years Wealth status Poor Middle/rich Maternal education No education No formal education Primary < 9 years of education Secondary/higher ≥9 years of education Parity Primiparous First ever birth Para 2–32–3 children Para 4+ 4 or more children Marital status Single Never married, widowed, separated/divorce at delivery time, not living with the spouse Married Married or cohabiting Place of residence Rural Urban Cesarean birth No Yes Check-up for pregnancy complications No Yes Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 5 of 12 2000−2001, 2006, and 2011. However, the 1995 dataset amounts of missing data. Birth complications were also was not subjected to weighting due to the need to not adjusted for in 1995 due to absence of data. maintain the statistical power of the data for that year, Figure 2 below shows the relationship between birth- the implication of which is a very minimal difference. A weight and time-to-death among neonatal mortality total of 5973 last-born live births with birthweights cases, combining all the study years. In conjunction with were included in the analyses. the survival table (not included in the paper), we ob- served that over 85% of all neonatal deaths in our study Estimation of LBW-attributable mortality risk fraction sample occurred in the first week of life. About 95% of among LBW neonates and in the population all the LBW (< 2500 g) neonatal deaths occurred within The LBW-attributable neonatal mortality risk fraction the first week of life. In comparison, about 82% of deaths (AF) and population-attributable mortality risk fraction among neonates with NBW (2500 g ≤ 4000 g) took place (PAF) were computed as proportion of prevalent deaths within in the first weeks. The rest died later, in the sec- that could be avoided if LBW was prevented or the ond, third, and fourth weeks. The figure also shows an death of LBW babies was eliminated. These were calcu- inverse proportionality relationship between weight and lated manually using eqs. (1) and (2) below. survival. With the exception of an outlier, the neonates with higher birthweights tended to survive longer, i.e. OR−1 beyond the first week. AF ¼  100; ð1Þ The LBW-attributable neonatal mortality in Uganda OR declined by more than half, from 33.6% (%) in 1995 The population attributable mortality risk fraction to 15.3% in 2011 as shown in Table 5 below. Similarly, PAF, expressed as a percentage (%) was computed using LBW-attributable neonatal mortality among LBW babies the eq. (2). also declined by 10.2% from 83.9% to 73.7% in the same period. OR−1 Figure 3 shows a non-uniform but continuous decline PAF ¼ P AF ¼ P   100; ð2Þ e e of LBW-attributable neonatal mortality in Uganda be- OR tween 1995 and 2011. OR is the odds ratio generated from binary logistic re- gression analysis and Pe is the proportion of deaths that Discussion have the exposure. Overall, the odds of neonatal mortality among LBW babies as compared to normal birthweight were re- Results duced by a third, from about 6 times higher in 1995 Table 2 shows birthweight and maternal and socio- to 3.8 times higher in 2011. The LBW-attributable demographic characteristics of last-born live births by neonatal mortality in the population declined by more neonatal survival status in Uganda. Overall, the average than half, from 33.6% in 1995 to 15% in 2011. This proportion of neonatal deaths among LBW babies be- present study is the first of its kind in Uganda and tween 1995 and 2011 was about 3.5% while the average perhapsthe wholeofeastAfricathatexaminesthe proportion of neonatal deaths among normal weight ba- trends of LBW-attributable mortality over the years. bies (≥2500 g ≤ 4000 g) during the same period was less The study reinforces the very few LBW-related studies in than 1 %. Cesarean birth was associated with neonatal Uganda and east Africa by providing new peer-reviewed mortality only in the year 2000−2001 (p <0.05). findings on the contribution of LBW towards neonatal Table 3 shows the distribution of the study variables mortality countrywide over a period of over 15 years. The by birthweight. Statistical significantly higher propor- study findings might be useful for auditing the causes of tions (p < 0.05) of mothers with no formal education neonatal deaths, and for evaluation, future health planning had LBW babies in almost all the years except 2011. and policy making aimed at improving neonatal survival. Similarly, maternal age < 20 years of age was associated The WHO emphasizes that auditing the causes of neo- with having higher proportions of LBW babies as shown natal deaths is paramount for effective monitoring and in the 1995 and 2006 findings (p < 0.01). improving mother and child health care [37]. In all surveys, LBW was significantly associated with The 3.8 times higher odds of deaths among LBW neo- neonatal mortality as shown in Table 4 below. The ad- nates in 2011 in the present study is consistent with the justed odds ratio (AOR) for the years in question were findings of a related study conducted by Kananura et al. as follows: in 1995, 6.2 (95% CI (2.3 − 17.0), in 2000−2001, in eastern Uganda in 2012–2013 that indicated a 3.51 5.3 (95% CI 1.7 − 16.1), in 2006, 4.3 (1.3 − 14.2), and in mortality odds ratio [36]. Comparable findings were also 2011, 3.8 (95% CI 1.3 − 11.2). The 1995 and 2000–2001 obtained in a follow-up study in western Uganda, com- data were not adjusted for wealth status due to large pleted in 2006 but analyzed by Marchant et al. in 2012 Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 6 of 12 Table 2 Distribution of birthweight, maternal and sociodemographic characteristics by neonatal survival status in Uganda, 1995–2011 Variables 1995 2000–2001 2006 2011 Survival, N = 1160 Survival, N = 1100 Survival, N = 1514 Survival, N = (2199) Died Lived P value Died Lived P value Died Lived P value Died Lived P value n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) Birthweight < 2500 g 4 (3.3) 118 (96.7) < 0.01 5 (4.6) 104 (95.4) < 0.01 5 (2.8) 175 (97.2) < 0.05 7 (2.9) 234 (97.1) < 0.05 ≥ 2500 g 6 (0.6) 1032 (99.4) 10 (1.0) 981 (99.0) 11 (0.8) 1323 (99.2) 22 (1.1) 1936 (98.9) Maternal age < 20 1 (0.6) 155 (99.4) > 0.05 1 (0.9) 111 (99.1) > 0.05 2 (1.4) 138 (98.6) > 0.05 2 (1.3) 154 (98.7) > 0.05 20–34 6 (0.7) 855 (99.3) 12 (1.4) 825 (98.6) 11 (1.0) 1105 (99.0) 15 (1.0) 1496 (99.0) 35–49 3 (2.1) 140 (97.9) 3 (2.0) 148 (98.0) 2 (0.8) 254 (99.2) 7 (1.6) 427 (98.4) b b Wealth index n = 392 n = 424 Poor 1 (0.7) 137 (99.3) > 0.05 1 (0.5) 187 (99.5) > 0.05 4 (0.9) 442 (99.1) > 0.05 7 (1.1) 652 (98.9) > 0.05 Middle / Rich 4 (1.6) 250 (98.4) 3 (1.3) 233 (98.7) 11 (1.0) 1056 (99.0) 17 (1.1) 1426 (98.9) Maternal education No education 2 (1.5) 132 (98.5) > 0.05 2 (1.6) 124 (98.4) > 0.05 3 (1.6) 179 (98.4) > 0.05 2 (1.2) 171 (98.8) > 0.05 Primary 6 (0.9) 653 (99.1) 8 (1.6) 605 (98.4) 7 (0.8) 857 (99.2) 12 (1.0) 1149 (99.0) Secondary higher 2 (0.5) 365 (99.5) 5 (1.4) 356 (98.6) 5 (1.1) 462 (98.9) 11 (1.4) 757 (98.6) Parity Primiparous 3 (1.0) 296 (99.0) > 0.05 4 (1.4) 278 (98.6) > 0.05 6 (1.7) 356 (98.3) < 0.05 3 (0.7) 424 (99.3) > 0.05 Para 2–3 3 (0.6) 532 (99.4) 5 (1.0) 483 (99.0) 7 (1.1) 622 (98.9) 11 (1.2) 945 (98.8) Para 4+ 4 (1.2) 322 (98.8) 6 (1.8) 323 (98.2) 2 (0.4) 520 (99.6) 10 (1.4) 709 (98.6) Marital status Single 1 (0.5) 199 (99.5) > 0.05 2 (1.0) 198 (99.0) > 0.05 2 (0.7) 277 (99.3) > 0.05 3 (0.8) 354 (99.2) > 0.05 Married 9 (0.9) 951 (99.1) 14 (1.6) 887 (98.4) 13 (1.1) 1221 (98.9) 22 (1.3) 1722 (98.7) Residence Rural 5 (1.0) 517 (99.0) > 0.05 11 (1.5) 737 (98.5) > 0.05 10 (0.9) 1051 (99.1) > 0.05 17 (1.1) 1493 (98.9) > 0.05 Urban 5 (0.8) 633 (99.2) 4 (1.1) 348 (98.9) 5 (1.1) 447 (98.9) 7 (1.2) 584 (98.8) Delivery mode Cesarean 1 (1.4) 71 (98.6) > 0.05 4 (4.4) 87 (95.6) < 0.05 1(0.8) 122 (98.2) > 0.05 5(2.1) 230 (97.9) > 0.05 Normal 9 (0.8) 1079 (99.2) 12 (1.2) 995 (98.8) 14 (1.0) 1372 (99.0) 24 (1.2) 1940 (98.8) Check-up No No data 11 (1.5) 742 (98.5) > 0.05 6 (0.7) 866 (99.3) > 0.05 13 (1.5) 843 (98.5) > 0.05 Yes 4(1.2) 332 (98.2) 9 (1.4) 613 (98.6) 14 (1.1) 1261 (98.9) P values were generated from Chi square analysis. Statistical significance (p < 0.05, two-sided) complications The separate totals(n) for wealth index in 1995 and 2000 shows a deviation from the total (N) due to missing data [15]. This study estimated the odds of neonatal mortality the health personnel interviewed about perinatal out- among LBW newborns relative to NBW newborns at comes in the health units indicated that LBW contrib- 3.45 [15]. Our findings of 15.3% LBW-attributable neo- uted to 16% of the total newborn deaths [38]. However, natal mortality in 2011 in the population are comparable the study also acknowledged the underreporting of LBW to the findings of a situation analysis study conducted by as a cause of death due to overlaps with infections and the Ministry of Health (MoH) in Uganda in 2008 [38]. breathing difficulties [38]. The MoH study combined both quantitative and qualita- The results indicated a significantly higher proportion tive methods and collected data from 10 districts cover- of deaths among LBW babies and this corroborates with ing the four conventional regions (Central, Eastern, findings of other studies [2, 3] that show higher mortalities Western and Northern) in Uganda. In this MoH study, among LBW newborns relative to their NBW counterparts. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 7 of 12 Table 3 Univariate analysis of maternal and sociodemographic characteristics of neonates by birthweight in Uganda, 1995–2011 Variables 1995, N = 1160 2000–2001, N = 1100 2006, N = 1514 2011, N = 2199 LBW (%) NBW (%) P value LBW NBW P value LBW NBW P value LBW NBW P value Maternal age < 20 26(16.7) 130(83.3) < 0.01 15(13.4) 97(86.6) > 0.05 27(19.1) 114(80.9) < 0.01 20(12.7) 137(87.3) > 0.05 20–34 81(9.4) 780(90.6) 77(9.2) 761(90.8) 112(10.0) 1004(90.0) 174(11.5) 1337(88.5) 35–49 15(10.5) 128(89.5) 17(11.2) 135(88.8) 41(16.0) 216(84.0) 39(9.0) 395(91.0) Wealth n = 392 n = 424 Poor 15(10.9) 123(89.1) > 0.05 19(10.1) 169(89.9) > 0.05 61(13.7) 385(86.3) > 0.05 72(10.9) 587(89.1) > 0.05 Middle/rich 26(10.2) 228(89.8) 25(10.6) 211(89.4) 118(11.1) 949(88.9) 161(11.2) 1282(88.8) Education level No education 24(17.9) 110(82.1) < 0.01 21(16.7) 105(83.3) < 0.01 29(15.9) 153(84.1) < 0.05 27(15.6) 146(84.4) > 0.05 Primary 67(10.2) 592(89.8) 60(9.8) 555(90.2) 101(11.7) 763(88.3) 121(10.4) 1040(89.6) Secondary 31(8.4) 336(91.6) 28(7.8) 332(92.2) 49(10.5) 418(89.5) 85(11.1) 684(88.9) Parity Primiparous 45(15.1) 254(84.9) < 0.01 27(9.6) 255(90.4) > 0.05 50(13.8) 312(86.2) > 0.05 58(13.6) 368(86.4) > 0.05 Para 2–3 48(9.0) 487(91.0) 51(10.5) 437(89.5) 69(11.0) 560(89.0) 98(10.3) 858(89.7) Para 4+ 29(8.9) 297(91.1) 31(9.4) 300(90.6) 60(11.5) 462(88.5) 77(10.7) 643(89.3) Place of residence Rural 67(12.8) 455(87.2) < 0.05 76(10.1) 674(89.9) > 0.05 134(12.6) 928(87.4) > 0.05 167(11.1) 1343(88.9) > 0.05 Urban 55(8.6) 583(91.4) 33(9.4) 319(90.6) 46(10.2) 406(89.8) 66(11.1) 526 (88.9) Marital status Single 25(12.5) 175(87.5) > 0.05 29(14.5) 171(85.5) < 0.05 39(13.9) 241(86.1) > 0.05 36(10.1) 321(89.9) > 0.05 Married 97(10.1) 863(89.9) 80(8.9) 821(91.1) 141(11.4) 1093(88.6) 197(11.3) 1547(88.7) Cesarean Yes 4(5.6) 68(94.4) > 0.05 11(11.8) 82(88.2) > 0.05 24(19.7) 98(80.3) < 0.01 29(12.3) 206(87.7) > 0.05 No 118(10.8) 970(89.2) 99(9.8) 909(9.2) 154(11.1) 1232(88.9) 212(10.8) 1752(89.2) Check-up No No data 72(9.5) 683(90.5) > 0.05 105(12.0) 767(88.0) > 0.05 87(10.2) 769(89.8) > 0.05 Yes 34(10.1) 302(89.9) 73(11.8) 1315(88.1) 143(11.2) 1132(88.8) LBW refers to low birthweight (< 2500 g), NBW refers to normal birthweight (≥2500 g – 4000 g). P values were obtained from chi square test The separate totals (n) for wealth index in 1995 and 2000 shows a deviation from the total (N) due to missing data Although cesarean births have been associated with mortal- Figure 2 showed that about 85% of neonatal deaths oc- ity as also shown by the findings (p < 0.05) for the year curred in the first week after birth. This is close to the 2000–2001 in Table 2, in 2006 and 2011 however, the find- estimate of a recent MoH report on maternal, perinatal ings (p > 0.05) indicated improvements in obstetric services and child death review that indicated about 75% neonatal that has enabled the survival of many cesarean birth babies. deaths in the first week [39]. The inverse proportional Table 4 Logistic regression analysis showing association between low birthweight and neonatal mortality in Uganda, 1995 − 2011 Adjusted odds ratios (95% confidence interval) Variable 1995 2000−2001 2006 2011 N = 1160 N = 1100 N = 1519 N = 2223 Birthweight b b a a Low birthweight 6.2 (2.3 − 17.0) 5.3 (1.7 − 16.1) 4.3 (1.3 − 14.2) 3.8 (1.3 − 11.2) Normal birthweight 1.0 1.0 1.0 1.0 LBW refers to low birthweight < 2500 g, NBW refers to normal birthweight (≥2500 g – 4000 g) Adjusted for all socio-demographic, maternal, pregnancy and birth related factors in Table 1 Adjusted for all socio-demographic (except wealth status), maternal, pregnancy and birth related factors in the study (Table 1). Complications were not adjusted for in 1995 Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 8 of 12 Fig. 3 Graphical representation of low birthweight-attributable Fig. 2 Kaplan-Meier survival curves by birthweight for neonates in neonatal mortality trends in Uganda between 1995 and 2011 Uganda between 1995 and 2011. Cum - cumulative relationship indicated by the trends of birthweight versus findings are comparable with the LBW-attributable mor- time-to-death among neonatal deaths in Fig. 2 concurs tality estimates among LBW babies in the whole country with findings from a hospital-based study in Dhaka, in this present study. Bangladesh [40]. The findings in Fig. 2 also implied that Neonatal mortality accounts for about 40% of global the risk of neonatal death is inversely proportional to under-five mortality [44]. In Uganda, in recent years, it birthweight and are in agreement with several other studies was estimated that about 45,000 neonates die every year [40–43]. However, our data on age at death (days) appeared [20]. By extension of our findings, this corresponds to to have been aggregated in terms of 7 days (weekly) and approximately 7000 (15.3%) neonatal deaths attributable not the actual mortality days. This slightly compromised to LBW in 2011. Although our findings could be a slight the accuracy of the Kaplan Meier’s survival curve in our underestimation given the many unrecorded births (about study in terms of days of survival. 45% in 2011) [43] and unregistered neonatal deaths, they According to a facility-based study by Hedstrom et al. provide comparable national estimates that can be used in central Uganda that admitted neonates born between for advocacy and countrywide public health planning to December 2005 and September 2008, 89% of neonatal reduce LBW-attributable neonatal deaths. For instance, deaths among LBW neonates weighing under 1000 g the successful Kangaroo Mother Care project for prema- could be attributable to LBW [43]. Another study by ture and LBW newborns initiated by Uganda Newborn Marchant et al. [15] that utilized data collected in 2006 in Study project (UNEST) in 2007–2011 in Iganga and western Uganda also estimated a 71% LBW-attributable Mayuge district [45] could be implemented countrywide. neonatal mortality among LBW neonates. Both of these The greatest national decline of LBW-attributable mor- tality estimated in 2011 in our study is a notable finding that could be attributed to the efforts of the inter-agency Table 5 Low birthweight-attributable neonatal mortality risk national Newborn Steering Committee (NSC) [46]. The proportions in Uganda between 1995 and 2011 NSC, which was initiated in 2006, ensured rapid policy Year of survey Attributable risk fraction (%) adaptation and implementations both at the health facility Among LBW neonates (AF) 1995 83.9 and community levels in the few years to 2011 [46]. It was mandated by the MoH to spearhead comprehensive ser- 2000–2001 81.1 vice delivery and community-and health facility-based 2006 76.7 training [46, 47]. Our findings thus reveal that the policy 2011 73.7 changes and its implementation may have had a profound In the entire population (PAF) 1995 33.6 positive impact on the survival of LBW newborns during 2000–2001 27.0 this period. The findings indicate that it is possible to 2006 24.0 eliminate unnecessary neonatal deaths due to LBW and make significant contributions towards achieving the SDG 2011 15.3 3.2 target that aims to lower neonatal death rate to 12 per LBW low birthweight, AF Attributable Fraction, PAF Population Attributable Fraction 1000 live births by 2030 [23]. Further, both the present Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 9 of 12 study findings and the NSC initiative could be of keen There was no statistically significant association be- interest to similar countries (with high neonatal mortalities) tween place of residence, maternal education, marital for policy making and study replications with the aim of status, wealth status, maternal age, and neonatal mortality, improving LBW neonatal survival, for instance, in the (P > 0.05) (Table 2). Although studies vary in their Philippines, where the decline of neonatal deaths has findings concerning the association between these stagnated [48]. socio-demographic and maternal factors (including parity) Also, the Uganda Newborn Study (UNEST) Project and neonatal mortality [53], many study findings have in- partly contributed to the decline in mortality of LBW dicated an association between single motherhood [54], and preterm newborns in parts of eastern Uganda and teenage maternal age [55–57], lack of education [56], rural consequently contributed to the overall national decline residence [57] and neonatal mortality. A systematic review during this period [45]. of 17 studies up to the year 2013 in SSA [55]indicated The survival analysis indicated that the rate of decline that socio-demographic and maternal risk factors are in LBW-attributable mortality in the 5-year periods in- much more prevalent among teenage mothers as com- creased from 6.6% between 1995 and 2000–2001 to 8.7% pared to adult mothers [55]. With the decentralized sys- between 2006 and 2011 in the population (Table 5). tem in Uganda, further analytical research at the districts However, between the two periods, there was a signifi- or regional levels on the effect of socio-demographic fac- cant deceleration in the decline to 3.0% between 2000 tors on birthweight and neonatal deaths would provide and 2001 and 2006 (Fig. 3 and Table 5). This could po- more robust findings for monitoring, policy making and tentially be due to the 20% decline in the use of family interventions. However, at the national level, comprehen- planning methods among < 20 years old sexually active sive measurement and recording of birthweight need to girls during this period as noted by the analytical over- be made possible, irrespective of whether a child is born view of the Ugandan child report [49]. This could have at home or at the hospital. As a national policy driven ini- led to increased teenage pregnancies. LBW are common tiative, the provision of weighing scales to health volun- among teenage mothers (< 20 years) [7] and the mortal- teers and midwives at the community level, even on a ity among babies born to younger mothers in Uganda shared basis based on proximity and locality, is feasible was also notably high between 1995 and 2005 [22]. and could be very effective for monitoring neonatal health Nevertheless, our findings in Table 2 did not show any sig- countrywide. Apart from improving accuracy on birth- nificant higher mortality numbers among the < 20 years weight data collection, the availability of weighing scales old mothers, perhaps because of the few number of births could also be a profound campaign tool for lowering LBW in this age-group in our sample selection. However, statis- incidences by highlighting preventive measures. Afford- tically reasonable numbers in 2006 showed a significant able and easy to maintain mechanical weighing scales have association between primipara mothers (most of whom previously been used at the community level in over 400 were younger mothers (Table 3)) and neonatal mortality. villages in western Kenya [58]. Although it was on a small A study conducted by Andualem et al. in western Uganda scale, the initiative was profoundly successful, as shown by between 2005 and 2008 revealed that over 82% of female an increase in the birthweight measurements of newborns students had unmet sexual/reproductive health counseling of about 54%, from 43% to 97% [58]. The current study needs [50]. Lack of knowledge about the signs of pregnancy could thus give the impetus to communities and local or- complications has been linked to birth unpreparedness in ganizations to take initiatives and improve the survival of Uganda [51], a consequent risk factor for neonatal deaths, LBW neonates. Further, as LBW is an underlying cause of including LBW deaths. A comparative development study 60–80% of all neonatal deaths globally (2,3) and about by Kevin Croke [52] also highlighted the decline in the 15% of neonatal deaths in Uganda (present study 2011 health system gains in Uganda between 2001 and 2006 findings), continuous data collection on birthweights that due to political shocks related to removal of presiden- supports research, monitoring, and strategic preventive in- tial term limits. Financing of the health care system was terventions could be a formidable approach to curbing negatively affected. This could partly account for the neonatal deaths and overall health systems strengthening rise in LBW-neonatal deaths during this period. The both globally and in Uganda. specialized care of LBW babies requires extra financing Although our study largely indicated no significant as- compared to NBW. The direct impact of the decline in sociations between cesarean birth, pregnancy complica- health system gains on survival of LBW detected by the tions and neonatal mortality for most of the years, a present study is consistent with WHO/UNICEF obser- number of studies have found associations between vations that survival of LBW neonates, a high-risk in- cesarean births [57, 59], pregnancy complications [59] fant group, is among the most sensitive indicators to and neonatal deaths. There were inconsistencies in our assess the progress of maternal and child health status findings with regard to the significant associations be- in a country [2]. tween socio-demographic factors and LBW across all the Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 10 of 12 study years (p < or > 0.05) (Table 3). However, there were based on the physical size of the body parts such as foot higher proportions of LBW babies among teenage and length, chest or head [63]. A study in Uganda compared uneducated mothers in all the survey years. Teenage preg- the accuracy of a proxy measure of LBW by midwives in a nancy was associated with LBW only in 1995 and 2006. hospital-based setting showed an accuracy of over 80%. These findings corroborate study findings elsewhere in However, the study also noted the limitation that the find- rural India [60] and in several SSA countries [7, 61]that ings may not reflect the actual situation in the communi- strongly indicate that young maternal age is associated ties where less skilled community volunteers assist in with LBW. A study in Brazil, however, found an associ- most births, and their estimates of cut-offs are prone to ation between teenage pregnancy and LBW only when bias [63]. Elimination of macrosomic newborns improved marital partners (an economic factor) were lacking [62]. the validity of our findings. Although the 1995 data included both home and hos- Methodological considerations pital births, which undermined the consistency of the The random sampling of data across the entire country study methodology across years, preliminary analysis in- and the standardized nature of data collection method dicated that among the selected sample of newborns of the DHS strengthen the external validity of our study with birthweight measures in 1995, only 3.5% of the and enable global comparability among countries. births were home births (or perhaps on the way to the Weighting the data for the years 2000, 2006 and 2011en- hospital). The 1995 data thus has a reasonable degree of abled us to adjust for disproportionate sampling and consistency with other survey years. However, the se- non-response. This improved the national representa- lection of only hospital births in other survey years tiveness and validity of the study estimates. The 1995 improved the quality and validity of the findings for dataset was not weighted and the results for that year those years. are slightly less representative. However, the results are The recording of neonatal survival data from day 0 to still valid, due to the fact that there was only a small dif- 30 by the DHS allowed us to clearly categorize our out- ference when weighted and unweighted results of all the come variable and investigate risk factors across all the other years were compared. The national representative- survey years with consistency. Given the large number ness of the 1995 data was only dependent on the random of home births (about 50%) in all the surveys, both the sampling across the entire country and the standardized LBW and neonatal deaths were likely underreported. nature of DHS data collection for its reliability. The birthweight data are prone to rounding-off or ag- The repeated findings of significant associations be- gregation into 500 g-weight intervals which could have tween LBW and neonatal mortality across all surveys slightly compromised the accuracy of Kaplan-Meier’s confirm the existing evidence of association and the in- survival analysis in this study. This aggregation of data ternal validity of this present study. Nonetheless, our was observed in a study by Channon et al. [64]. How- study could not confirm the causal association because ever, the fact that over 90% of LBW neonatal deaths in the exact causes of newborn deaths were not ascertained our study occurred in the first week is quite consistent medically. The in-depth use of the nationally representa- with global WHO findings that 75% of neonatal deaths tive DHS datasets in this study has revealed the need to occur in the first week [65], given the high-risk group of improve data collection techniques and to include other LBW in a low-income country. similarly important variables such as diagnostic causes of death among individual children, for example, birth Conclusion asphyxia. Low birthweight is associated with neonatal mortality Another limitation of our study was that although hos- and contributes to a substantial proportion of neonatal pital births recorded and/or communicated birthweights, deaths in Uganda. Although significant progress has over 65% were from mothers` recall and the rest from been made to reduce newborn deaths attributed to the health card, and we cannot therefore completely dis- LBW, by 2011, about 74% of all LBW neonates died in miss the possibility of recall bias. This also applies to the the neonatal period. This implies that the health system 1995 data that included both hospital and home births. in place has been inadequate to meet the challenge of Nevertheless, child birth is a significant event in a ensuring LBW survival. There is also profound need to mother’s life and with our study selection of the most re- strengthen both birth and neonatal death registration ir- cent birth experience, there is a very high possibility that respective of whether the infants are born at home or at the mothers recalled correct birthweights. Moreover, for the health centers. The decentralized health system in the years 2000 to 2011, birthweight data concerned solely Uganda can enable community health workers (CHW) information regarding hospital born babies because these and the village health teams (VHT) in liaison with the were measured birthweights and not estimated weights as sub-counties and the districts to close the existing gaps in-home births, where birthweights are mainly estimated concerning neonatal birth and death audits. This will Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 11 of 12 enable robust and continuous research and monitoring Department of Health and Human Services regulations for the protection of human subjects. of the progress of LBW neonatal survival. Our study presents national estimates of risks and mortality trends Competing interests that provide national basis for continual evaluation and The authors declare that they have no competing interests. policy recommendations to prevent LBW and minimize risks of neonatal deaths. A holistic approach to reduce the incidence of preventable LBW babies could be fos- Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in tered to reduce these mortality rates. Viable fronts that published maps and institutional affiliations. could be strengthened include sexual education in schools to prevent teenage pregnancies, complementing Received: 26 April 2017 Accepted: 15 May 2018 nutritional diet of pregnant mothers, HIV testing, ensur- ing that all pregnant mothers use mosquito nets, training References of health workers, and promoting antenatal care visits and 1. United Nations Children’s Fund and World Health Organization. Low hospital births. Enhancing the quality of postnatal care birthweight: country, regional and global estimates. New York: UNICEF; 2004. could also reduce mortality incidence of LBW newborns. 2. World Health Organization. Care of the preterm and/or low-birth-weight newborn. [Updated 2017]. Available from: http://www.who.int/maternal_ Abbreviations child_adolescent/newborns/prematurity/en/. Accessed July 2017. AF: Attributable fraction; AOR: Adjusted odds ratio; CHW: Community health 3. Lawn J, Blencowe H, Oza S, You D, Lee A, Waiswa P, et al. Every newborn: workers; DHS: Demographic and health survey; EA: East Africa; LBW: Low progress, priorities, and potential beyond survival. Lancet. 2014;384(9938):189–205. birthweight; MDG: Millennium development goals; MoH: Ministry of Health; 4. Ministry of Health and Family Welfare (India). India Newborn Action Plan: NBW: Normal birthweight; NSC: Newborn Steering Committee; Government of India. New Delhi: 2014. PAF: Population attributable fraction; SDG: Sustainable development goal; 5. Office of the Registrar General and Census Commissioner (India). Causes of SGA: Small for gestational age; SSA: Sub-Saharan Africa; UNAP: Uganda death in India 2010–2013 [data table online]. c2010–2013. Available from: nutrition action plan; UNEST: Uganda newborn study; VHT: Village health http://www.censusindia.gov.in/2011-common/Sample_Registration_System. teams html. Accessed 11 Nov 2016. 6. Official Statistics of Sweden. Statistics – Health and Medical Care (Sweden), Acknowledgements 2015. Pregnancies, Deliveries and Newborn Infants. The Swedish Medical Special gratitude to the Demographic and Health Survey program for Birth Register 1973–2014 Assisted Reproduction, treatment 1991–2013. making the datasets available for this study. 7. Chibwesha C, Zanolini A, Smid M, Vwalika B, Phiri Kasaro M, Mwanahamuntu M, et al. Predictors and outcomes of low birth weight in Availability of data and materials Lusaka, Zambia. Int J Gynecol Obstet. 2016;134(3):309–14. Authorization to use the secondary, unrestricted Uganda DHS data was 8. Nkyekyer K. Twin and singleton births in Ghana — a case-control study. obtained from MEASURE DHS after a requisition was made through their Twin Res. 2002;5(04):265–9. webpage by registration, selection of country and submission of a brief 9. Kayode G, Amoakoh-Coleman M, Agyepong I, Ansah E, Grobbee D, description of the study. The DHS program is authorized by the country, Klipstein-Grobusch K. Contextual risk factors for low birth weight: a Uganda to distribute data for research aimed at improving public health. multilevel analysis. PLoS One. 2014;9(10):e109333. Procedures and questions for standard DHS surveys have been approved by 10. Juárez S, Merlo J. Revisiting the effect of maternal smoking during both ICF Institutional Review board and the review board in the host country pregnancy on offspring birthweight: a quasi-experimental sibling analysis in Uganda. Both of which ensure that DHS complies with the protection of the Sweden. PLoS One. 2013;8(4):e61734. dignity and rights of human participants and the norms of the country. 11. Xiao P, Zhou Y, Chen Y, Yang M, Song X, Shi Y, et al. Association between maternal HIV infection and low birth weight and prematurity: a meta- Authors’ contributions analysis of cohort studies. BMC Pregnancy Childbirth. 2016;16:243. https:// MOA Conceptualized the research study, developed the study design, doi.org/10.1186/s12884-016-1044-3. acquired data, analyzed and drafted the manuscript. AA Contributed to the 12. Guyatt H, Snow R. Impact of malaria during pregnancy on low birth weight interpretation of data and to the writing of the manuscript. BOA Involved in in sub-Saharan Africa. Clin Microbiol Rev. 2004;17(4):760–9. developing the study design, interpretation and contributed to the writing 13. Registrar General of India and the Centre for Global Health Research. Causes of the manuscript. MOA, AA, BOA Read and approved the final draft of the of Death Statistics 2010-2013. [Updated 2015]. Available from: http://www. manuscript. cghr.org/2015/12/causes-of-death-in-india-2010-2013/. 14. Lawn JE, Cousens SN, Wilczynska K. Estimating the causes of four million Ethics approval and consent to participate neonatal deaths in the year 2000: statistical annex—the world health report Approval for utilization of Uganda DHS data for this study was obtained 2005. Geneva: World Health Organization; 2005. from ICF Macro Int. Data was accessible from: https://dhsprogram.com/Data/. 15. Marchant T, Willey B, Katz J, Clarke S, Kariuki S, Ft K, et al. Neonatal mortality Uganda National guidelines for secondary (third party) research that use risk associated with preterm birth in East Africa, adjusted by weight for unrestricted, publicly available and unlinked data that does not identify gestational age: individual participant level meta-analysis. PLoS Med. 2012; individuals or communities are exempted from ethical approval requirement. 9(8):e1001292. During data collection process by the primary data collector, ICF Macro Int, a 16. Welaga P, Moyer C, Aborigo R, Adongo P, Williams J, Hodgson A, et al. Why written informed consent was sought for each willing participant by are babies dying in the first month after birth? A 7-year study of neonatal MEASURE DHS before data collection commenced. The statement of mortality in northern Ghana. PLoS One. 2013;8(3):e58924. consent emphasize to the participant that they are at liberty to terminate or 17. Aslam H, Saleem S, Afzal R, Iqbal U, Saleem S, Shaikh M, et al. Risk factors of decline to respond at any stage of the interview process. Those unwilling birth asphyxia. Ital J Pediatr. 2014;40:94. https://doi.org/10.1186/s13052-014- were excluded from the study. The participants remained anonymous and 0094-2. the DHS approved procedures ensure that future identification of any 18. Shah G, Budhathoki S, Das B, Mandal R. Risk factors in early neonatal sepsis. individual participant or community is practically impossible. These measures Kathmandu Univ Med J (KUMJ). 2006;4(2):187–91. are thus in adherence with both the international guidelines for 19. Siakwa M, Kpikpitse D, Laryea T, Ankobil A, Dare S, Ebu N. A Five-year epidemiological studies of the Council of International Organizations of neonatal mortality trend in a Ghanaian teaching hospital after the Medical Sciences (CIOMS) and the Uganda National guidelines for research implementation of strategies to achieve the millennium development goal. involving human subjects. Further, the measures abide by the United States Int J Pediatrics Child Health. 2014;2(1):043–9. Arunda et al. BMC Pregnancy and Childbirth (2018) 18:189 Page 12 of 12 20. Ministry of Health. Situation analysis of newborn health in Uganda: current 45. Waiswa P, Peterson S, Namazzi G, Ekirapa E, Naikoba S, Byaruhanga R, et al. status and opportunities to improve care and survival. Kampala: The Uganda newborn study (UNEST): an effectiveness study on improving government of Uganda. Save the children, UNICEF, WHO; 2008. newborn health and survival in rural Uganda through a community-based 21. The Republic of Uganda: Roadmap for Accelerating the Reduction of intervention linked to health facilities - study protocol for a cluster Maternal and Neonatal Mortality and Morbidity in Uganda: Government of randomized controlled trial. Trials. 2012;13:213. https://doi.org/10.1186/1745- Uganda; 2007–2015. 6215-13-213. 46. Newborn Steering Committee. Kampala. Uganda: Save the Children, Healthy 22. Uganda Bureau of Statistics (UBOS) and Macro International Inc. Uganda Child Uganda, UNICEF, on behalf of the national Newborn Steering Demographic and Health Survey 2006. Calverton: UBOS and Macro Committee: 2011. Newborn Health Implementation Framework Standards International Inc.; 2007. for Newborn Health Services: Health Facility Assessments. 23. Sustainable development goals - United Nations. United Nations Sustainable 47. Mbonye A, Sentongo M, Mukasa G, Byaruhanga R, Sentumbwe-Mugisa O, Development. 2016. Available from: http://www.un.org/sustainabledevelopment/ Waiswa P, et al. Newborn survival in Uganda: a decade of change and sustainable-development-goals/. Accessed 01 Dec 2016. future implications. Health Policy Plan. 2012;27(suppl 3):iii104–17. 24. World Population Review. Uganda Population 2016. [Updated 2016]. 48. Kraft A, Nguyen K, Jimenez-Soto E, Hodge A. Stagnant neonatal mortality Available from: http://worldpopulationreview.com. Accessed 20 Nov 2016. and persistent health inequality in middle-income countries: a case study of 25. The World Bank. World Development Indicators (WDI). [Updated 2016]. the Philippines. PLoS One. 2013;8(1):e53696. Available from: http://databank.worldbank.org/data/reports.aspx?source= 49. Walakira E.J.D., Muhangi S.. Munyuwiny F,. Matovu E,. Awich I., Ddumba world-development-indicators. Accessed 21 Nov 2016. Nyanzi J. et al. (2016). The state of the Ugandan child – an analytical 26. Uganda Bureau of Statistics (UBOS) (Uganda), Republic of Uganda; 2014. overview. Kampala/Washington DC: USAID/QED. Statistical Abstract. 50. Boltena A, Khan F, Asamoah B, Agardh A. Barriers faced by Ugandan 27. Xu K, Evans DB, Kadama P, Nabyonga J, Ogwal PO, Nabukhonzo P, et university students in seeking medical care and sexual health counselling: a al. Understanding the impact of eliminating user fees: utilization and cross-sectional study. BMC Public Health. 2012;12:986. https://doi.org/10. catastrophic health expenditures in Uganda. Soc Sci Med. 1186/1471-2458-12-986. 2006;62(4):866–76. 51. Sandberg J, Odberg Pettersson K, Asp G, Kabakyenga J, Agardh A. 28. Uganda Bureau of Statistics (UBOS) and ICF International Inc. 2012. Uganda Inadequate knowledge of neonatal danger signs among recently delivered Demographic and Health Survey 2011. Kampala: UBOS and Calverton: ICF women in southwestern rural Uganda: a community survey. PLoS One. International Inc. 2014;9(5):e97253. 29. Statistics Department/Uganda and Macro International. Uganda 52. Croke K. The political economy of child mortality decline in Tanzania and Demographic and Health Survey, 1995. Calverton: Statistics Department/ Uganda, 1995–2007. Stud Comp Int Dev. 2012;47(4):441–63. Uganda and Macro International; 1996. 53. Berhan Y, Berhan A. A meta-analysis of socio-demographic factors for 30. Measure DHS. Guide to demographic and health survey statistics: perinatal mortality in developing countries: a subgroup analysis of methodology for the measure DHS phase III project, implemented from the National Surveys and small-scale studies. Ethiop J Health Sci. 2008–2013. 2006. 2014;24(0):41. 31. ICF International. MEASURE DHS Biomarker Field Manual. Calverton: ICF 54. Izugbara C. Single motherhood and neonatal and infant mortality in Sierra International; 2012. Leone, Burkina Faso and Burundi. Public Health. 2016;135:122–30. 32. Macro international Inc. Sampling manual. DHS-III basic documentation no. 55. Ramaiya A, Kiss L, Baraitser P, Mbaruku G, Hildon Z. A systematic review of 6. Calverton: Macro international Inc; 1996. risk factors for neonatal mortality in adolescent Mother’s in sub Saharan 33. Ota E, Haruna M, Suzuki M, Anh D, Tho L, Tam N, et al. Maternal body mass Africa. BMC Res Notes. 2014;7(1):750. index and gestational weight gain and their association with perinatal 56. Mekonnen Y, Tensou B, Telake D, Degefie T, Bekele A. Neonatal mortality in outcomes in Viet Nam. Bull World Health Organ. 2011;89(2):127–36. Ethiopia: trends and determinants. BMC Public Health. 2013;13:483. https:// 34. Said A, Manji K. Risk factors and outcomes of fetal macrosomia in a doi.org/10.1186/1471-2458-13-483. tertiary Centre in Tanzania: a case-control study. BMC Pregnancy 57. Ezeh O, Agho K, Dibley M, Hall J, Page A. Determinants of neonatal mortality in Childbirth. 2016;16:243. https://doi.org/10.1186/s12884-016-1044-3. Nigeria: evidence from the 2008 demographic and health survey. BMC Public 35. Blanc AK, Wardlaw T. Monitoring low birth weight: an evaluation of Health. 2014;14:521. https://doi.org/10.1186/1471-2458-14-521. international estimates and an updated estimation procedure. Bull World 58. Gisore P, Shipala E, Otieno K, Rono B, Marete I, Tenge C, et al. Community Health Organ. 2005;83(3):178–85. based weighing of newborns and use of mobile phones by village elders in 36. Kananura R, Tetui M, Mutebi A, Bua J, Waiswa P, Kiwanuka S, et al. The neonatal rural settings in Kenya: a decentralised approach to health care provision. BMC mortality and its determinants in rural communities of eastern Uganda. Reprod Pregnancy Childbirth. 2012;12:15. https://doi.org/10.1186/1471-2393-12-15. Health. 2016;13:13. https://doi.org/10.1186/s12978-016-0119-y. 59. Arunda M, Emmelin A, Asamoah B. Effectiveness of antenatal care services 37. World Health Organization. Making every baby count: audit and review of in reducing neonatal mortality in Kenya: analysis of national survey data. stillbirths and neonatal deaths, 2016. Geneva: World Health Organization; 2016. Glob Health Action. 2017;10(1):1328796. 38. Ministry of Health. Situation analysis of newborn health in Uganda. Kampala: 60. Borah M, Agarwalla R. Maternal and socio-demographic determinants of Government of Uganda; 2008. low birth weight (LBW): a community-based study in a rural block of Assam. 39. Ministry of Health. Situation analysis of newborn health in Uganda: current J Postgrad Med. 2016;62(3):178. status and opportunities to improve care and survival. Kampala: 61. Mombo-Ngoma G, Mackanga J, González R, Ouedraogo S, Kakolwa M, Government of Uganda. Save the Children, UNICEF, WHO; 2008. Manego R, et al. Young adolescent girls are at high risk for adverse 40. Yasmin S, Osrin D, Paul E, Costello A. Neonatal mortality of low-birth-weight pregnancy outcomes in sub-Saharan Africa: an observational multicountry infants in Bangladesh. Bull World Health Organ. 2001;79(7):608–14. study. BMJ Open. 2016;6(6):e011783. 41. Golestan M, Fallah R, Akhavan Karbasi S. Neonatal mortality of low birth 62. Guimarães A, Bettiol H, Souza L, Gurgel R, Almeida M, Ribeiro E, et al. Is weight infants in Yazd — Iran. Early Hum Dev. 2008;84:S26. adolescent pregnancy a risk factor for low birth weight? Rev Saude Publica. 42. Gaiva M, Fujimori E, Sato A. Neonatal mortality in infants with low 2013;47(1):11–9. birthweight. Revista da Escola de Enfermagem da USP. 2014;48(5):778–86. 63. Elizabeth N, Christopher O, Patrick K. Determining an anthropometric 43. Hedstrom A, Ryman T, Otai C, Nyonyintono J, McAdams R, Lester D, et al. surrogate measure for identifying low birth weight babies in Uganda: a Demographics, clinical characteristics and neonatal outcomes in a rural hospital-based cross-sectional study. BMC Pediatr. 2013;13:54. https://doi. Ugandan NICU. BMC Pregnancy Childbirth. 2014;14:327. https://doi.org/10. org/10.1186/1471-2431-13-54. 1186/1471-2393-14-327. 64. Channon A, Padmadas S, McDonald J. Measuring birth weight in 44. Wang H, Liddell CA, Coates MM, Mooney MD, Levitz CE, Schumacher developing countries: does the method of reporting in retrospective AE, et al. Global, regional, and national levels of neonatal, infant, and surveys matter? Matern Child Health J. 2010;15(1):12–8. under-5 mortalities during 1990-2013: a systematic analysis for the 65. Lawn JE, Cousens S, Zupan J. 4 million neonatal deaths: when? Where? global burden of disease study 2013. The Lancet (London, England). Why? Lancet. 2005;365:891–900. https://doi.org/10.1016/S0140- 2014;384(9947):957–79. 6736(05)71048-5.

Journal

BMC Pregnancy and ChildbirthSpringer Journals

Published: May 30, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off