Michelle Hynes, Mani Sheik, Hoyt G. Wilson, Paul Spiegel. Reproductive Health Indicators and Outcomes Among Refugee and Internally Displaced Persons in Postemergency Phase Camps. JAMA. 2002;288(5):595–603. doi:10.1001/jama.288.5.595
Author Affiliations: Division of Reproductive Health, National Center for Chronic Disease (Ms Hynes and Dr Wilson) and International Emergency and Refugee Health Branch, Division of Emergency and Environmental Health Services, National Center for Environmental Health (Dr Spiegel), Centers for Disease Control and Prevention, Atlanta, Ga; and Center for Refugee and Disaster Studies, John Hopkins School of Public Health, Baltimore, Md (Mr Sheik).
Context Despite increasing awareness of the importance of reproductive health
programs and services for refugee and internally displaced populations, there
is a paucity of basic epidemiological data on reproductive health outcomes.
Objectives To collect data on reproductive health outcomes among refugees and internally
displaced persons in postemergency phase camps and compare these outcomes
with those of host country and country-of-origin populations. To determine
programmatic factors that may affect reproductive health outcomes.
Design, Setting, and Participants Retrospective study of data collected from August 1998 through March
2000 of 688 766 persons living in 52 postemergency phase camps in 7 countries.
Reproductive health outcomes of refugee and internally displaced populations
were compared with available data of reference populations within their respective
host country and country of origin.
Main Outcome Measures Crude birth rate (CBR), neonatal mortality rate (NNMR), maternal mortality
ratio (MMR), percentage of newborns with low birth weight (LBW), and incidence
of complications of unsafe or spontaneous abortions.
Results Six of 11 groups had lower CBRs than their country of origin and 5 of
9 groups had lower CBRs than their host country. Four of 5 had lower NNMRs
than their country of origin and 6 of 9 had lower NNMRs than the host country.
Four of 6 had lower MMRs than their country of origin, and 5 of 6 had lower
MMRs than their host country. Seven of 9 had lower percentages of LBWs than
in the country of origin and 5 of 9 had lower percentages of LBWs than the
host country. Higher CBRs were associated with more recently established camps
and higher numbers of local health staff per 1000 persons; and higher percentages
of LBW newborns were associated with rainy season, more recently established
camps, lower numbers of community health workers per 1000 persons, and camps
without supplementary feeding programs.
Conclusions Refugees and internally displaced persons in most postemergency phase
camps had better reproductive health outcomes than their respective host country
and country-of-origin populations.
There are approximately 37 million displaced persons worldwide, consisting
of refugees and internally displaced persons.1
An estimated 20%, or 7 million, of these displaced persons are women of reproductive
age (15-49 years).2 Women and children may
be particularly vulnerable in complex emergencies in which families are separated
and social support systems are destroyed.
During the emergency phase of a complex emergency, defined by a crude
mortality rate of 1 or more deaths/10 000 persons per day,3,4
infectious diseases such as diarrhea, measles, malaria, and respiratory tract
infections account for most of the excess deaths.3,5
In the postemergency phase of a complex emergency, crude mortality rates decrease,
primarily because the incidence of infectious diseases is reduced through
improved environmental conditions, public health measures, and health care
Although refugees and internally displaced populations may remain in camps
for many years during the postemergency phase, health organizations often
continue to focus their programs on diseases that contribute to excess deaths
during the emergency phase. Little attention is given to the evolving needs
of the population, such as chronic diseases, and psychosocial and reproductive
The 1994 International Conference on Population and Development initiated
a shift in reproductive health policy from a narrow focus on family planning
programs to more inclusive definitions of reproductive health that were adopted
by the humanitarian aid community.12 Even with
the expansion of reproductive health policies and programs, however, it is
unclear to what extent their services and policies are based on data, despite
increasing demands by donors for evidence-based programs.13
A survey in 1994 of reproductive health services in refugee populations
found few available data beyond antenatal and delivery services.14
In recent years, the international community has increasingly focused on providing
more comprehensive reproductive health services to refugee and internally
displaced populations, which include programs on family planning, gender-based
violence, and sexually transmitted infections (STIs) (including HIV/AIDS.)12 However, there is a paucity of basic epidemiological
data on reproductive health outcomes of refugee and internally displaced populations
in postemergency phase camps. A major reason for this dearth of knowledge
is because many health information systems do not include categories for the
recording of reproductive health data.
A review of reproductive health research of displaced populations suggests
that improved access to and use of health services, as well as social and
demographic factors, may account for better health outcomes than those of
host country or country-of-origin populations.15
Better documentation of reproductive health outcomes in refugee and internally
displaced populations may lead to improved programs and policies for both
conflict-affected and developing country settings. Efforts to link relief
and development efforts through the integration of refugee and host communities
may help relieve some of the burden refugee populations can put on a host
community and increase benefits to both populations.16,17
The objectives of this study, which are part of a larger study of health
indicators among refugees and internally displaced persons in postemergency
phase camps, are 3-fold: (1) to provide baseline information on reproductive
health outcomes among refugee and internally displaced populations in postemergency
phase camps; (2) to compare these outcomes with those of populations in their
respective host country and country of origin; and (3) to identify important
policies and programs associated with reproductive health outcomes.
Fifty-two refugee and internally displaced persons postemergency phase
camps in Azerbaijan (n = 7 camps), Ethiopia (n = 11), Myanmar (n = 3), Nepal
(n = 7), Tanzania (n = 8), Thailand (n = 5), and Uganda (n = 11) were served
by a total of 10 different health organizations and were visited by 1 of 2
project investigators between November 1998 and March 2000. The following
inclusion criteria were met for camps in this study: (1) refugee or internally
displaced persons residing in a camp during the postemergency phase; (2) stable
camp population size (defined as <5% change in population size during the
3 months before data collection); (3) camp population at least partially dependent
on outside organizations for both food aid and health care; and (4) functioning
health information system (defined as those health organizations that recorded
morbidity and mortality in their health care units and produced a monthly
Using a standardized data collection form, investigators collected 3
months of retrospective record-based data on all health outcomes except maternal
deaths, which were collected for a 12-month period. Sources of data for all
health outcomes included mortality reports from the United Nations High Commissioner
for Refugees (UNHCR) and health organizations, camp registers from the inpatient
and outpatient departments in the camp health clinics, maternal and child
health reports, birth and death registers, and community health worker (CHW)
reports. To verify these data, investigators interviewed key informants, including
community leaders, camp health staff, ambulance drivers, and CHWs. Live birth
data were taken from monthly reports and checked against maternal ward registries.
Stillbirths or abortions mistakenly included in these figures were removed.
Because no camp in the study reported the number of complications due to unsafe
or spontaneous abortion, we systematically gathered the data from registers
kept by the camps' inpatient departments or referral hospitals. Project investigators
sought information on individual deaths from various sources available in
each camp and cross-checked the information to create as complete a list as
possible. Information collected for each death included the decedent's age,
sex, and cause of death. The ethical review boards at the Centers for Disease
Control and Prevention and Johns Hopkins University reviewed and approved
the research study proposal.
We chose the following reproductive health outcomes for our study on
the basis of available data and recommendations from the Reproductive Health
in Refugee Situations Interagency Field Manual18:
crude birth rate (CBR), defined as live births per 1000 population per year;
neonatal mortality rate (NNMR), defined as deaths among live-born infants
during the first 28 days of life per 1000 live births; maternal mortality
ratio (MMR), defined as maternal deaths during or within 42 days of pregnancy
from any cause related to or aggravated by the pregnancy or its management
per 100 000 live births; low–birth-weight (LBW) rate, defined as
the percentage of live-born infants weighing less than 2500 g; and incidence
of complications due to unsafe or spontaneous abortion (ICUSA), defined as
the number of women treated for complications of abortions per 1000 live births.
Unsafe abortion was defined as a procedure for terminating unwanted pregnancy
either performed by persons lacking the necessary skills, performed in an
environment lacking minimal medical standards, or both. Spontaneous abortion
was defined as termination of pregnancy before the 20th week with spontaneous
expulsion of the fetus. Because we recorded maternal deaths for a 12-month
period and live births only for a 3-month period within those 12 months, we
did Monte Carlo simulations to estimate the confidence intervals for the MMR.
To compare reproductive health outcomes among the refugee and internally
displaced persons with those of reference populations within their respective
host country and country of origin (the latter for refugees only), we aggregated
camp data into 9 refugee and 2 internally displaced persons groups according
to their country of origin and host country when applicable (Table 1). We also compared reproductive health outcomes of refugee
populations from the same country of origin residing in different host countries
and among refugee populations from different countries of origin residing
within the same host country. For example, reproductive health outcomes among
Sudanese refugees residing in Uganda were compared with outcomes of Sudanese
refugees living in Ethiopia as well as with outcomes of Rwandan refugees living
We performed descriptive analysis using SPSS software (Version 9.0,
SPSS Inc, Chicago, Ill). We calculated reproductive health outcomes and their
95% confidence intervals for camp residents and compared these with country-of-origin
(for both internally displaced and refugee groups) and host country (for refugees
only) outcomes. When reproductive health outcomes from specific areas within
the host country and country of origin were available, we used those for comparison;
such data from demographic and health surveys were available for Nepal, Rwanda,
Tanzania, and Uganda.19- 23
For other countries, comparison data were obtained from United Nations' statistics.24- 27
We performed multiple regression analyses to identify programmatic,
demographic, and policy factors that may be associated with the chosen reproductive
health outcomes. These independent variables were grouped into the 5 categories
Age of camp at the time of data collection was measured in years. Monthly
population figures for each camp were obtained from UNHCR and nongovernmental
organizations. The average population was determined by using the mean of
the 3 monthly populations. Because studies have shown that birth and death
rates can vary by season,28- 30
the season in which the data were collected for each camp was recorded and
categorized as rainy, dry, or cold.
Distance from the camp to the border or area of conflict was recorded
in kilometers. Time to referral hospital was provided by the health staff
and defined as the mean time it took for a camp resident to reach the referral
hospital from the camp using the most common mode of transportation.
Obstetric referral rates were defined as the number of documented obstetric
referrals by camp health staff to a referral hospital per 100 000 live
births. The percentage of births in health institutions was defined as the
percentage of all documented births occurring in either the health unit within
the camp or at a referral hospital.
Availability of supplemental food programs for pregnant women was defined
as presence of a program to provide food to pregnant women beyond the general
Number of local health care workers was defined as the total number
of physicians, nurses, and clinical officers from the refugee and internally
displaced population or host country working in the camp per 1000 persons.
Traditional birth attendants (TBAs) and CHWs were recorded as the number of
workers in each category per 1000 persons.
Whenever possible, continuous independent variables were categorized
according to recommendations from the Sphere Project4
and the Reproductive Health in Refugee Situations Interagency Field Manual,18 which provide minimum standards for the provision
of basic services to refugee and internally displaced populations in complex
emergencies. For example, because the Reproductive Health in Refugee Situations
Interagency Field Manual recommends 0.5 TBAs per 1000 persons, we categorized
the TBA variable into less than 0.5, 0.5 to 1.4, and 1.5 or more TBAs per
1000 persons. We categorized all other variables according to a logical division
of the variable that would result in an approximately equal number of camps
in each category.
We performed weighted multiple regressions using SAS software (Version
8.2, SAS Institute Inc, Cary, NC). For each of the 5 outcome variables, we
fit all possible models containing 1, 2, 3, or 4 independent variables and
retained only the model with the highest R2 among those models in which all included independent variables were
significant at the .05 level. Not all independent variables were considered
for each model to avoid models based on too few camps. For models with adjusted
CBR, the percentage of births in health institutions, obstetric referral rate,
and time to hospital were not considered. Models for NNMR excluded distance
to border and obstetric referral rate. The MMR models excluded distance to
border; LBW models excluded distance to border, obstetric referral rate, and
time to hospital; and ICUSA models excluded distance to border, percentage
of health institution births, TBAs, and CHWs per 1000 persons. For each outcome
variable, the subset of camps that we used to fit regression models consisted
of those camps for which there was complete data for the outcome variable
and all candidate independent variables. All models initially were adjusted
for region by means of 2 dummy variables. For this purpose, camps were categorized
as African, Asian, or Azerbaijani (Table
1). The camp region was not significant in any of the models for
LBW rate, so those models were not adjusted for region in the final analysis.
We did not test for interactions in any of the models.
Analysis weights were computed as the inverses of the estimated variances
of the outcome variables. Variances were estimated based on the Poisson distribution
for adjusted CBR, ICUSA, and MMR, and were based on the binomial distribution
for the NNMRs and LBW rates.
Because the number of women of reproductive age in the camps was often
estimated, we chose not to use these estimates as denominators for the adjusted
CBRs. Most of the health information systems in the camps distinguished only
between persons younger than 5 years and 5 years or older as well as between
males and females. We used the number of females 5 years or older for each
camp as the denominator of the adjusted CBR (number of live births per 1000
females aged ≥5 years per year) in the regression analyses. Eliminating
girls younger than 5 years and all males produced a more specific denominator
than total population for adjusted CBR. However, this estimate was still imperfect
since it included some females outside the reproductive age range.
Of the 52 camps in 7 countries with a total population of 688 766
persons, 7 (13.5%) were internally displaced persons camps with a population
of 26 923 (range, 836-4606) and 45 (86.5%) were refugee camps with a
population of 661 843 (range, 2569-40 474). Most camps existed for
4 or more years (63.5%) and had 10 000 or more persons (55.8%). Based
on populations in 51 camps, mean household size was 4.8 persons per household
(range, 2.7-6.7). Based on 44 camps, the mean male-to-female ratio was 1.01
(range, 0.69-1.47). The mean percentage of children younger than 5 years was
15.7 (range, 4.5-26.2).
The mean percentage of births occurring in health institutions was 26.6
(range, 0-100). Only 1 camp's health facilities had the ability to perform
cesarean deliveries; all other camps sent women to referral hospitals, most
of which were located less than 60 minutes from the camp (51.9%). Family planning
services were provided in all but 1 of the 52 camps, with the remaining camp
using services in a neighboring camp. Of the 51 camps with such services,
all offered condoms, 44 (86.3%) injectable contraceptives, 41 (80.4%) oral
contraceptives, and 13 (25.5%) intrauterine devices. Descriptions of other
independent variables from the 5 major categories included in the multiple
regression analysis are shown in Table 2.
Comparisons of reproductive health outcomes in refugee and internally
displaced groups with those in their respective host country and country of
origin are presented in Table 1.
Because the Myanmars and Azerbaijanis were internally displaced within their
own countries, host country comparisons are not applicable. Therefore, comparisons
with host countries are limited to 9 of the 11 groups of refugees and their
respective countries of origin. Missing data either from our data set or country
comparison data for NNMR, MMR, LBW, and ICUSA led to fewer comparisons. Differences
between the outcomes of groups refer to statistically significant differences
at the .05 level.
A total of 5476 live births from a population of 688 766 refugee
and internally displaced persons in 52 camps were reported during the 3-month
data collection periods. The CBR data for comparison with country of origin
were available for all 11 groups: 6 had lower unadjusted CBRs than their country
of origin, 4 had higher CBRs, and 1 had no difference. Comparison data with
host country rates were available for all 9 groups: 5 had lower unadjusted
CBRs than their host country, 2 had higher CBRs, and 2 had no difference.
A total of 47 neonatal deaths out of 5476 live births in 52 camps were
reported during the 3-month periods. Of the 5 of 11 groups for which comparison
country-of-origin data were available, 4 had a lower NNMR than their country
of origin. Of the 7 of 9 groups for which comparison host country data were
available, 6 had a lower NNMR than the host country.
Because of the low number of maternal deaths reported in the Asian and
Azerbijan camps (3 deaths reported for 10 Asian camps, missing data for 5
Asian camps and all Azerbaijan camps), we calculated MMRs only for the sub-Saharan
African camps (6 refugee groups). Maternal mortality data were not available
for 2 camps in the region. One camp was in Uganda and the other was in Tanzania.
Thirty-two maternal deaths were reported for 6 refugee groups in 28 camps
during the 1-year data collection period in which we estimated 4350 live births.
Four of 6 groups had lower MMRs than their country of origin, and 5 of 6 had
lower MMRs than their host country.
Three hundred eighty-nine (7.5%) of the 5173 live births recorded for
9 refugee groups from 39 camps for which data were available had LBW rates.
No data were available for Azerbaijan, Myanmar, and 2 Ethiopian camps. Seven
of 9 groups had lower percentages of LBW infants than in the country of origin
and 5 of 9 had lower percentages of LBW infants than the host country.
A total of 201 abortions requiring a medical procedure, such as vacuum
extraction or dilatation and curettage, occurred among 7 refugee and internally
displaced groups in 34 camps that reported such procedures. Data on abortions
were unavailable from health organizations in Thailand, Myanmar, and 3 camps
in Tanzania. We excluded Azerbaijan camps from our analysis because abortion
is legal, and thus we could not determine whether complications were due to
intentional or spontaneous abortion. Medical procedures occurred both in the
camp and at referral hospitals. These same 34 camps reported 3845 live births
over the same period. Overall, the ICUSA for camps in sub-Saharan Africa was
54.8/1000 live births (range, 47.2-64.5; 188 complications). No country-of-origin
or host country data were available to make comparisons.
Between refugee and internally displaced populations from the same country
of origin living in different host countries, the only different reproductive
health outcomes were percentages of LBWs among Sudanese in Ethiopia and Uganda.
However, refugee and internally displaced populations from different countries
of origin living in the same host country (eg, Somalis and Sudanese living
in Ethiopia) showed differences in all reproductive health outcome variables
Table 3 shows the final
regression models for adjusted CBR, NNMR, and LBW. We report no models for
MMR or ICUSA because there were no single-variable models in which the independent
variable was significant and no multivariable models in which all independent
variables were significant.
Regression coefficients in Table 3 indicate the change in the outcome variable associated with a change
in the independent variable from the lowest to the middle or highest category.
In all 3 of the models, the coefficients for all categories of the independent
variables were monotonically increasing or decreasing, so that the direction
of association was consistent across all levels of the independent variable.
Because the lowest category was the referent in all models (Table 3), its regression coefficient is always zero by definition.
We regard this as an exploratory analysis indicative of the independent variables
that seem to be associated with specific outcomes.
The coefficients of the first model listed in Table 3 indicate that camps in the middle category of camp age (4-6
years from Table 2) had (on average)
8.4 live births/1000 per year fewer than camps in the low category of camp
age (0-3 years); and camps in the high category of camp age (≥7 years)
had 31.3 live births/1000 per year fewer than camps in the low category of
camp age. Camps in the middle category of local health staff (0.5-0.9 local
staff/1000) had 73.1 live births/1000 per year more than camps in the low-staff
category (0-0.4 staff/1000); and camps in the high category of local health
staff (≥1 staff/1000) had 77.7 live births/1000 per year more than camps
in the low category of local health staff.
Camps in the middle category of TBAs (0.5-1.4 TBAs/1000) had 0.6 neonatal
deaths/1000 more than camps in the low category of TBAs (0-0.4 TBAs/1000);
and camps in the high category of TBAs (≥1.5 TBAs/1000) had 8.3 more neonatal
deaths/1000 than camps the low category of TBAs.
The cold season was associated with 1.3/1000 LBWs fewer than the rainy
season, and the dry season with 4.3/1000 LBWs fewer than the rainy season.
Camps in the high category of camp age (≥7 years) had 4.2 LBWs/1000 fewer
than camps in the lowest category of camp age (0-3 years), but there was no
difference between camps in the low and middle categories of camp age. Camps
in the middle category of CHWs (1.0-1.9 CHWs/1000) had 3.5 LBWs/1000 fewer
than camps in the low category of CHWs (0-0.9 CHWs/1000); and camps in the
high category of CHWs (≥2/1000) had 5.8 LBWs/1000 fewer than camps in the
low category of CHWs. Camps that had supplemented food programs for pregnant
women had 20.5/1000 LBWs fewer than camps that did not have such programs.
We assessed the statistical power to detect relationships in models
containing only 1 independent variable. For this purpose, we determined the
partial R2 corresponding to 80% power
for each outcome variable. Partial R2
in this context is the increase in R2
produced by adding an independent variable (consisting typically of 2 dummy
variables) to a model whose independent variables are only the 2-region dummy
variables. The partial R2 corresponding
to a power level of 80% ranged from 0.17 to 0.27 for the 5 outcome variables.
Overall, reproductive health outcomes in refugee and internally displaced
populations in postemergency phase camps are better than those of their respective
host country and country-of-origin populations. Our results are similar to
those of a 1998 UNHCR report, which examined data from 1996 to 1998 of 42
camps in 7 countries.26 Better access by camp
residents to preventative and curative health care services, and to food and
nonfood items, as well as improvements in water supply and sanitation,6- 9 may account
for better reproductive health outcomes among postemergency phase camp populations
than those in their respective host country and country-of-origin populations.
The camp populations in our study had access to health services within a maximum
of 5 km from where they lived, a much smaller distance than for most rural
populations in developing countries.31 Furthermore,
the quality of health care services for refugees is generally assumed to be
better than that for the surrounding populations.7,9
The CBRs were generally lower than those for host country and country-of-origin
populations, although there were exceptions for Myanmar and Burundi camp populations.
The association of lower CBRs with camps that have existed longer may be due
to access to family planning services combined with low-mortality rates during
the postemergency phase,6 both of which take
time to develop and may allow families to better control the timing of pregnancies
and influence the number of children they desire. Social and demographic factors
as well as improved access to health services may influence CBRs in older
camps more than refugee status.15
A major factor in preventing maternal deaths is the availability of
emergency obstetric care and the population's access to such care.32 The risk for maternal deaths is most adversely affected
by delayed treatment due to a delay in the decision to seek medical care,
delay in arriving at a health facility, or the delay of the facility in providing
adequate care.33 Unlike many of the host country
and country-of-origin populations, all refugee and internally displaced populations
that we studied had access to health care, including emergency obstetric services,
at little or no cost. All camps had access to referral hospitals providing
emergency obstetric care within 12 hours, an indicator recommended by the
World Health Organization.34 Free transportation
and support to the referral hospital was provided by most nongovernmental
organizations. The lack of association between MMR and independent variables
in our study may be due to lack of statistical power because maternal deaths
are rare. Furthermore, we may have underestimated the number of maternal deaths,
since most camps did not have a recording system solely established for these
Low birth weight is a major risk factor in neonatal and postneonatal
deaths.35 Our study suggests an association
between the existence of maternal supplemental food programs in postemergency
phase camps and a lower percentage of LBW newborns. However, we do not know
whether all pregnant women in the camps, regardless of their nutritional status,
received supplemental food rations. Furthermore, the quantity of rations differed
according to the nongovernmental organization in charge of those services.
Whether an association exists between maternal nutrition and LBW remains controversial,
with several studies supporting such an association34- 36
and others finding little evidence.37 Because
there are no conclusive data, neither the World Health Organization nor the
UNHCR supports providing a blanket supplemental food program to all pregnant
women in refugee and internally displaced populations, regardless of nutritional
status.38,39 However, a blanket
supplemental food program for pregnant women may provide an incentive for
women to receive antenatal care. The association between higher rates of LBW
and the rainy season may be due to increased malarial infection in pregnant
women, because a strong association has been found between placental malarial
infection and LBW.40,41 The association
found between lower rates of LBW and older camps may be a reflection of better
access to health services as well as better health status in general than
refugee and internally displaced populations that have recently arrived to
a camp from a conflict situation.
The associations between increased staff and both higher CBRs and poorer
outcomes of NNMRs and LBWs may be due to a recording bias. More staff working
in the camps are more likely to capture and record these events. The reverse
associations with lower numbers of CHWs and higher rates of LBW, while making
intuitive sense, is difficult to explain given the direction of the other
The ICUSA reproductive health indicator is relatively new, but it allows
researchers in complex emergencies to determine the magnitude of the problem
and possible underlying causes. A high ICUSA may indicate inadequate family
planning coverage for women who wish to delay or avoid pregnancy.18 A 1998 UNHCR report included ICUSA data for 5 of
8 postemergency phase refugee sites, documenting a range of 12 to 114 per
1000 live births.26 We documented a narrower
ICUSA range of 31 to 96 per 1000 live births. Despite the lack of comparison
data for ICUSA in the host country and country of origin, we present baseline
data for the refugee and internally displaced populations to allow for comparisons
in future studies.
We were unable to distinguish from the records whether an abortion requiring
medical intervention was spontaneous or induced, possibly because of the sensitive
nature of the distinction or poor record keeping. Such a distinction, however,
has important programmatic implications, and therefore supervisors of health
information systems should implement and enforce a standardized case definition
Because of the poor quality of data collected or reported for antenatal
coverage, STIs, and gender-based violence in the camps, we did not include
these variables in our model. Although all 52 camps reported antenatal care
coverage, most reported more than 100% coverage. Possible reasons included
recording more than first-time visits, not distinguishing between local and
displaced women using antenatal services, or undercounting births. Although
49 (94.2%) of 52 camps collected STI data, the same case was often recorded
at different points in the health care system. Conversely, health care workers
believed that STIs were underreported because of the stigma of such a diagnosis.
Furthermore, STIs were often not disaggregated into their various syndromes
using standardized case definitions, making interpretation of the data difficult.
Thirty-six (97.3%) of 37 camps queried had gender-based violence programs
and collected data. However, the small number of documented cases suggested
that few women used the services or that the information was not accurately
recorded. In camps with well-established gender-based violence programs, more
cases were recorded per capita, suggesting a reporting bias.
Our study has several limitations. First, underestimation of deaths
and overestimation of camp populations may have led to underestimated mortality
in the camps visited. We attempted to minimize the former problem by limiting
the reporting period to 3 months before the camp visit to allow for a thorough
collection and verification of all data collected; we identified 20% more
deaths than the health organizations actually reported.42
Both factors would likely have occurred in the same direction for most camps,
thus minimizing any potential biases in the analysis. Second, the number of
live births may have been underestimated because of underreporting. However,
families have an incentive to report births to receive additional food rations.
Third, the data were collected retrospectively and therefore poor recording
and recall bias could have resulted in less accurate data. To improve the
reliability of the data, we collected only 3 months of retrospective data
(except for annual maternal mortality data), introducing a potential seasonality
bias. However, camps in different countries were visited in different seasons,
thus distributing this potential bias across all seasons. Furthermore, the
season in which data collection occurred was a candidate independent variable
for all 5 outcomes. The season variable was significant in several of the
models for LBW, suggesting an important seasonal effect for that outcome.
Season was not significant in models for any of the other 4 outcomes. To the
extent that season was correlated with other independent variables, our models
may have missed other important seasonal effects. Fourth, certain independent
variables that may affect reproductive health outcomes, such as antenatal
care coverage and the incidence of STIs, were not included in the models because
of unreliable data. Thus, we could not account for their potential to confound
relationships with other variables that were included in the study. Fifth,
there is a selection bias. We were unable to visit all postemergency phase
camps in every country with postemergency phase camps that met our inclusion
criteria because of logistical constraints or lack of government or nongovernmental
organization authorization. Lastly, the disparities between reproductive health
indicators of camp populations and those of the local populations in the host
country and country of origin may actually be greater than demonstrated in
this study. Refugee and internally displaced populations often come from and
settle in marginal areas. The countrywide and world regional reproductive
health indicators used for comparison are likely better than those of the
local populations surrounding the camps. Comparison data from host countries
were collected during times of relative stability, similar to situations in
which we collected data for the postemergency phase camps (one of the study
criteria). However, countries of origin may have been in conflict at the time
of data collection and thus may have had worse health outcomes due to the
instability than the refugee and internally displaced persons in our study.
Specifically, the countries with potential disparities are Burundi, DRC, Somalia,
and Sudan. Since both the periods and methods of data collection used in determining
comparison reproductive health outcomes differ from those we used in the camps,
comparisons should be interpreted with caution.
To make reliable programmatic decisions, creators and supervisors of
health information systems must systematically include reproductive health
indicators and improve their recording, reporting, and interpretation. We
refer specifically to maternal mortality, STIs, spontaneous and induced abortions,
antenatal coverage, and gender-based violence. Delineation of populations
into more specific age categories, such as women of reproductive age and adolescents,
rather than younger and older than age 5 years could improve the interpretation
and consequently the targeting of reproductive health programs.
Our data suggest that refugee and internally displaced persons in most
postemergency phase camps had better reproductive health outcomes than their
respective host country and country-of-origin populations. This result is
a testament to the skill and dedication of the organizations providing reproductive
health care services and of the progress made during the past decade in the
field of refugee reproductive health. Lessons learned in the developing world
context, such as the need for emergency obstetric care, are being applied
in camps during postemergency periods. However, the differences in reproductive
health outcomes among these different populations raise important issues regarding
the equity of humanitarian aid. The reasons for these disparities need further
examination, and solutions for making the health outcomes more equitable between
groups need to be found and implemented. We recommend prospective studies
to examine these issues as well as to determine the interaction between birth
and mortality rates during the different phases of complex emergencies and
to examine the relationships between CBRs and the implementation of various
reproductive health programs. Furthermore, the effects of supplemental food
distribution to pregnant women, and season and socioeconomic patterns should
also be examined.
Health programs in postemergency phase camps may provide guidance to
improving reproductive health outcomes among the local host country population
living near refugee and internally displaced communities. The data from this
study as well as from future studies may help health professionals and their
organizations develop evidence-based reproductive health programs. Lessons
learned from such programs may in turn have applications for program development
and implementation in developing countries.