The impact of Kangaroo Mother Care on Neonatal Mortality in Brazil
Melo Junior, AVP*; Huttly, SR; Victora, CG; Marshal, T
London School of Hygiene & Tropical Medicine, United Kingdom (Anisio.Veloso@lshtm.ac.uk)
Abstract
Objective
To reduce neonatal mortality and achieve the Millennium Development Goal for
Child Survival (MDG-4) poses a challenge for many low and middle-income
countries. In 2000, hospital-based Kangaroo Mother Care (KMC) - involving 24hour skin-to-skin contact between a mother and her stable low birth weight infant became national policy in Brazilian maternity hospitals. Although promoted as a
potential strategy for neonatal survival, there is little evidence for the impact of
KMC on mortality.
Methods: An ecological study was conducted with the units of analysis being
high-risk pregnancy hospitals in the state capital cities. Aggregated data for each
hospital on neonatal deaths and maternal education were compiled by linking
birth and death information databases for 2002. Information about the
implementation of KMC and other neonatal care facilities were obtained by postal
questionnaires.
Findings: 97 questionnaires were completed (88% response rate). The mean late
neonatal mortality rate was 3.75 per 1000 children with birth weight ranging from
1,250-2,000g. The partial correlation coefficient between mortality and KMC final
implementation score, adjusted for maternal education, region and technology
score, was -0.47 (95% CI: -0.53 to -0.23; p<0.01). The adjusted regression
coefficient was -0.42 (95% CI: -0.60 to -0.24; p< 0.01).
Interpretation: These results are compatible with an impact of KMC on late
neonatal mortality among low birth weight infants in Brazil, and with existing
information on the effect of KMC on severe morbidity. It provides evidence in
support of this national policy that should encourage other countries considering
implementation of KMC to achieve the MDG-4.
Background
The Millennium Development Goal for child survival (MDG-4) - to reduce
childhood mortality by two-thirds by 2015 - will not be met without substantial
reductions in neonatal mortality. The World Health Organization estimates that
each year four million children die in the first month of life, with three quarters of
these deaths occurring in the first week of life and nearly 99% occurring in
developing countries. Although the world has witnessed a fall in infant mortality
rates, in most developing countries neonatal mortality levels have declined less
quickly than post- neonatal rates (1).
Since the late 1970s, interest has grown in Kangaroo Mother Care (KMC) as an
alternative to incubator care for low birth weight babies. KMC keeps the low birth
weight infant in skin-to-skin contact between the mother’s breasts while in the
hospital and can continue after discharge. A large proportion of neonatal deaths
occur in infants born with low birth weight so KMC has the potential to contribute
to the reduction of neonatal mortality. Yet evidence for the effectiveness of KMC is
sparse, a Cochrane review by Conde-Agudelo et al. demonstrated the paucity of
evidence, particularly for its mortality impact. It concluded that the literature points
to a positive impact of KMC on severe morbidity (a reasonable proxy for mortality)
with no deleterious effects (2).
KMC is a good example of a policy that has a natural appeal to policymakers and
that is therefore implemented before it is fully tested in a randomised controlled
trial. Brazil is the only country where KMC has become a national health policy for
clinically stable low birth-weight infants as a hospital intervention and not at
community level (3). The national implementation of KMC in Brazil precluded a
randomized controlled trial to evaluate its impact. However, alternative
approaches for the evaluation of interventions have been advocated, and in this
paper we have explored such approaches to assess the impact of KMC in Brazil.
The study aimed to assess the impact of KMC by investigating the association
between the level of implementation of KMC and neonatal mortality among
children delivered in third level neonatal units in Brazil.
Methods
An ecological study was conducted with the units of analysis being high-risk
pregnancy hospitals in the 27 state capital cities. Aggregated data for each
hospital on neonatal deaths and maternal education were compiled by linking
birth and death information databases for 2002. Information about the
implementation of KMC and other neonatal care facilities were obtained by postal
questionnaires. The results of the questionnaire were used to develop various
scores for each hospital. The KMC implementation score comprised information
on provision and utilization and was calculated as set out in Table 1.
Table 1: Computation of the KMC implementation score
________________________________________________________________________________________
Provision 2 Components:
A) Number of Kangaroo Care Beds per 1000 low birth weight (LBW) infants born.
A score of 1 (low) to 5 (high) was assigned according to the quintile of the distribution of this variable across all
hospitals.
B) The percentage of maternity unit professionals trained in KMC
80-100%: 5
50-80%: 4
20-50%: 3
10- 20%: 2
<10%:
1
Provision Score = (A + B)
08-10 = 5
06-08 = 4
04-06 = 3
02-04 = 2
< 02 = 1
Utilization
C) The percentage of eligible LBW infants who received KMC
Utilization Score (C)
80-100% = 5
50-80% = 4
20-50% = 3
10- 20% = 2
<10%
=1
Implementation
Implementation Score = (Provision Score + Utilization Score)
2
________________________________________________________________________________________
A technology score was computed for each hospital which had 16 components
as set out in Table 2. This score constituted a potential confounding variable
because it might be associated both with KMC implementation and with neonatal
mortality.
National Birth certificate (SINASC) and National Mortality (SIM) registration
information systems was the source of data about neonatal mortality rates.
Information on maternal education, measured as years of education, was
collected from the SINASC database. In this database this variable is recorded
as 0, 1-3 years, 4-7 years, 8-11 years and 12 and more. For this study, these
categories were assigned values 0-4 and the mean score per hospital was
obtained.
The reliability of the two scales (implementation and technology) was assessed
through Cronbach’s alpha coefficient. Correlation and regression analyses were
used to assess the association between KMC implementation and neonatal
mortality, including adjustment for maternal education, availability of technology
in the hospital and region of the country.
Table 2: Computation of the technology score
________________________________________________________________________________________
16 components
A value of 1 was assigned for the presence of each of the following
1) Human Milk Bank
2) Baby Friendly Hospital
3) Infection committee
4) Paediatrician 24/7
5) Heated cot in labour room
6) Radiology service
7) Laboratory
8) Pathology facilities
9) Pharmacy
10) Teaching Hospital
11) The percentage of professionals trained in neonatal resuscitation:
80-100%: 5
50-80%: 4
20-50%: 3
10- 20%: 2
<10%:
1
A score of 1 (low) to 5 (high) was assigned according to the quintile of the distribution
of variables 12-15 across all hospitals.
12) The ratio of LBWI to high-risk obstetric beds
13) The ratio of LBWI infants to incubators
14) The ratio of LBWI to halogen photo therapy units
15) The ratio of LBWI to ventilators
16) Type of person responsible (first on-call basis) for neonatal resuscitation in the labour room
Only paediatrician:
5
Any doctor:
4
Any doctor / any nurse:
3
Only nurse:
2
Only students or others:
1
Technology Score (sum of all components score 32-40 = 5
24-32 = 4
16-24 = 3
08-16 = 2
< 08 = 1
________________________________________________________________________________________
Results
Questionnaires were sent to all 115 high-risk pregnancy maternity hospitals in the
country. Five of these had not yet received the training on KMC and were thus
ineligible for inclusion in the study. Another five hospitals provided incomplete
information and seven hospitals did not respond despite repeated attempts. Only
one hospital manager refused to participate in the study. Thus the response rate
among eligible hospitals was 88% (97/110).
The Cronbach’s alpha coefficients for the KMC final implementation and
technology scores were 0.81 and 0.79, respectively. The partial correlation
coefficient between mortality and KMC final implementation score, adjusted for
maternal education, region and technology score, was -0.47 (95% CI: -0.53 to 0.23; p<0.01) Table 3. The adjusted regression coefficient was -0.42 (95% CI: 0.60 to -0.24; p< 0.01) Figure 1.
Table 3: KMC implementation and neonatal mortality
______________________________________________________________________________________________
Implementation Technology Mother’s
Mortality rate Pearson’s Partial
R-square R-square
Score
Score
education
Mean (SD)
Coeff.
Correlation
partial
Mean (SD)
Mean (SD) Mean (SD) §
Coeff.**
Correlation**
______________________________________________________________________________________________
National
(97 units) 3.01 (1.25) 3.14 (1.23)
1.59 (0.15)
3.75 (1.04)
-0.64*
-0.47*
0.41
0.22
North
(10 units) 1.50 (0.55) 1.82 (0.75)
1.52 (0.83) 4.83 (0.57)
-0.25*
-0.23*
0.06
0.05
Northeast
(31 units) 2.98 (1.21) 3.52 (1.31)
1.58 (0.14) 3.74 (1.11)
-0.57*
-0.34*
0.33
0.12
Southeast
(30 units) 3.53 (1.01) 3.21 (1.04)
1.58 (0.16)
3.30 (0.86)
-0.47*
-0.37*
0.22
0.14
South
(11 units) 3.86 (0.78) 3.63 (1.21)
1.67 (0.18)
3.06 (0.86)
-0.60*
-0.29*
0.36
0.08
West
(15 units) 2.50 (1.30)
2.86 (1.09) 1.57 (0.87)
4.32 (0.66)
-0.37*
-0.20*
0.14
0.04
____________________________________________________________________________________________
§ 18 (18.6%) cases were missing
* Correlation is statistically significant at the 0.01 level (2-tailed)
** Adjusted for technology scores and, for the national score, adjusted for region and maternal education.
This study was funded by Pan American Health Organization, Brazilian CAPES Foundation and Brazilian Ministry of Health.
Results
Figure 1. Linear regression of neonatal mortality on KMC implementation score,
with adjustment for technology score.
Discussion
This study provides some of the first evidence on the potential impact of Kangaroo Mother
Care on neonatal mortality. Our results suggest that higher levels of implementation of KMC
in Brazilian maternity hospitals in major metropolitan settings are associated with lower
mortality. This association persisted after adjustment for three potential confounding variables
- maternal education, hospital technology score and region of the country.
Trials comparing mortality rates between 24-hour KMC versus conventional care (incubator
care) are scarce in the literature. In Ecuador, a trial was stopped after six months because of
a highly significant impact on severe morbidity, however no difference in mortality was
reported at this stage(4). A trial in Colombia and a multi-centre one in Ethiopia, Indonesia and
Mexico also showed no impact on mortality (5,6). In Ethiopia, the effectiveness of early KMC before stabilization of low birth weight infants – was compared with conventional care. The
results
. suggest better survival of those infants under KMC but comparison with our study is
limited as the study population differs (pre-stabilisation infants) (7). A common feature of all
these trials is their very limited sample sizes for estimating mortality impact. Data on mortality
impact from other types of study are also limited. In a cohort study in Zimbabwe, following the
introduction of KMC in a mission hospital, survival of babies born under 1,500g improved from
10% to 50%, whereas that of babies 1,500-1,999g improved from 70% to 90% (8). Facilities in
the neonatal unit were scarce and therefore the pre-KMC ‘conventional care’ was not
comparable with our study. Also in Zimbabwe, a quasi experimental pilot study found that 0/37
KMC infants died compared to 3/37 in conventional care (9). The existing evidence has been
summarised in a Cochrane review (2) and in the Lancet Neonatal Survival Series, both of
which that there was evidence of substantial benefit (1).
Discussion
Our study design enabled mortality rates to be based on adequate numbers. The limitations
of ecological analyses are well established, but the approach used here makes adjustment for
regional differences, a socio-economic factor (maternal education) and technology levels
between hospitals. Also, because the units of implementation of KMC under routine
conditions are the hospitals, individual level analyses are inappropriate. Although a definitive
statement on causality is not possible due to the nature of our study, the evidence is
nevertheless quite convincing. The association remained after taking account of differences
between hospitals in the availability of other resources which might impact on mortality and
was also seen within each region of the country. Although mortality data are not totally
reliable, it is unlikely that there is substantial bias in the results, mainly because vital
registration in the capital cities is quite uniform within each region. Furthermore, the study
achieved a high level of coverage of the study population with complete information on 88%
of eligible hospitals. The missing data were not clustered in any particular region nor were of
a particular type of hospital.
References
In conclusion this study suggests that Kangaroo Mother Care, implemented under routine
conditions, may reduce later neonatal mortality among infants with birth weight of 1,2502,000g in Brazil. It provides evidence in support of this national policy that should provide
encouragement to other countries considering implementation of KMC to achieve the MDG-4.
References
1. Lawn JE, Cousens S, Zupan J. 4 million neonatal deaths: when? Where? Why? Lancet 2005;365(9462):891-900; 2.Conde-Agudelo A, Diaz-Rossello JL, Belizan JM. Kangaroo
mother care to reduce morbidity and mortality in low birthweight infants. Cochrane Database Syst Rev 2003(2):CD002771; 3.Ministry of Health of Brazil. Humanized Care
Assistance To The Low Birth Weight Newborn Baby: Kangaroo Mother Method. Brasilia: MoH, 2004. 282p; 4.Sloan NL, Camacho LW, Rojas EP, Stern C. Kangaroo mother
method: randomised controlled trial of an alternative method of care for stabilised low-birthweight infants. Maternidad Isidro Ayora Study Team. Lancet 1994;344(8925):782-5.; 5.
Charpak N, Ruiz-Pelaez JG, Figueroa de CZ, Charpak Y. Kangaroo mother versus traditional care for newborn infants </=2000 grams: a randomized, controlled trial. Pediatrics
1997;100(4):682-8; 6. Cattaneo A, Davanzo R, Worku B, et al. Kangaroo mother care for low birthweight infants: a randomized controlled trial in different settings. Acta Paediatr
1998;87(9):976-85; 7. Worku B, Kassie A. Kangaroo mother care: a randomized controlled trial on effectiveness of early kangaroo mother care for low birth weight infants in
Addis Ababa, Ethiopia. J Trop Pediatr 2005 Apr;51(2):93-7; 8. Bergman NJ, Jurisoo LA. “The Kangaroo-Method” for treating low birth weight babies in a developing country. Trop
Doct 1994; 24(2):57-60; 9. Kambarami RA, Chidede O, Kowo DT. Kangaroo care versus incubator care in the management of well preterm infants- a pilot study. Ann Trop
Paediatr 1998 Jun;18(2):81-86.