Vaccine 33 (2015) 3407–3414

Contents lists available at ScienceDirect

Vaccine
journal homepage: www.elsevier.com/locate/vaccine

Field evaluation of measles vaccine effectiveness among children in

the Democratic Republic of Congo
Reena H. Doshi a,∗ , Patrick Mukadi b , Calixte Shidi c , Audry Mulumba c , Nicole A. Hoff a ,
Sue Gerber d , Emile Okitolonda-Wemakoy e , Benoit Kebela Ilunga f ,
Jean-Jacques Muyembe g , Anne W. Rimoin a
a
Department of Epidemiology, UCLA Fielding School of Public Health, 650 S Charles E Young Drive, Los Angeles, CA 90095, USA
b
Department of Microbiology, Kinshasa School of Medicine, B.P. 127 Kinshasa, Lemba, Kinshasa, Democratic Republic of the Congo
c
Expanded Programme on Immunization, Ave de la Justice, Kinshasa, Democratic Republic of the Congo
d
Polio Program, Bill and Melinda Gates Foundation, 500 Fifth Avenue North, Seattle, WA 98109, USA
e
Kinshasa School of Public Health, B.P. 127 Kinshasa, Lemba, Kinshasa, Democratic Republic of the Congo
f
Division of Disease Control, Ministry of Public Health, Ave de la Justice, Kinshasa, Democratic Republic of the Congo
g
National Institute for Biomedical Research, Minister of Public Health, Avenue de la Democratie, Kinshasa, Democratic Republic of the Congo

a r t i c l e i n f o a b s t r a c t

Article history: Background: Large-scale measles outbreaks in areas with high administrative vaccine coverage rates sug-
Received 13 October 2014 gest the need to re-evaluate measles prevention and control in the Democratic Republic of Congo (DRC).
Received in revised form 10 April 2015 Monitoring of measles Vaccine Effectiveness (VE) is a useful measure of quality control in immunization
Accepted 17 April 2015
programs. We estimated measles VE among children aged 12–59 months in the Democratic Republic of
Available online 30 April 2015
Congo (DRC) using laboratory surveillance data from 2010–2012.
Methods: We used the case-based surveillance system with laboratory confirmation to conduct a case-
Keywords:
control study using the test negative design. Cases and controls were selected based on presence (n = 1044)
Measles
Vaccine effectiveness
or absence (n = 1335) of measles specific antibody IgM or epidemiologic linkage. Risk factors for measles
Immunization were assessed using unconditional logistic regression, stratified by age.
Democratic Republic of Congo Results: Among children 12–59 months, measles vaccination was protective against measles [aOR (95% C)],
Vaccine preventable diseases 0.20 (0.15–0.26) and estimated VE was 80% (95% CI 74–85%). Year of diagnosis, 2011: 6.02 (4.16–8.72)
and 2012; 8.31 (5.57–12.40) was a risk factor for measles when compared to 2010. Compared to Kin-
shasa, children in Bas-Congo, Kasai-Oriental, Maniema and South Kivu provinces all had higher odds of
developing measles. Measles VE was similar for children 12–23 months and 24–59 months (80% and 81%
respectively).
Conclusions: Repeated occurrences of measles outbreaks and lower than expected VE estimates suggest
the need to further evaluate measles vaccine efficacy and improve vaccine delivery strategies in DRC.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction availability of a safe and effective vaccine, measles continues to be a

An estimated 1.5 million deaths among children under five are
attributed to vaccine-preventable diseases [1,2]. Measles is among
the most infectious of diseases and is associated with complications
such as encephalitis, pneumonia, and blindness [3]. Despite the
∗ Corresponding author. Tel.: +1 310 825 2096
E-mail addresses: rhdoshi@ucla.edu (R.H. Doshi), patrickmukadi@gmail.com
(P. Mukadi), shidicalixte5@yahoo.fr (C. Shidi), audrymwk@hotmail.fr (A. Mulumba),
Nhoff84@ucla.edu (N.A. Hoff), Sue.gerber@gatesfoundation.org (S. Gerber),
okitow@yahoo.fr (E. Okitolonda-Wemakoy), kebelailunga@gmail.com (B.K. Ilunga),
muyembejj@gmail.com (J.-J. Muyembe), arimoin@ucla.edu (A.W. Rimoin).
major cause of vaccine-preventable disease mortality among chil-
dren under five in resource-limited countries, with up to ten out of
every hundred measles cases leading to death [3]. Globally, measles
immunization coverage has improved tremendously over the last
ten years; measles has been eliminated in most high and middle-
income countries. In 2012, there were 122,000 estimated measles
deaths worldwide, over 95% of which occurred in resource-limited
settings [4].
In the African region, increased routine vaccination cover-
age coupled with Supplementary Immunization Activities (SIAs)
has led to significant reductions in incidence and mortality [5].
Between 2000 and 2010, measles-related mortality was reduced
by 85% [6]. However, throughout the region, deficiencies in routine

http://dx.doi.org/10.1016/j.vaccine.2015.04.067
0264-410X/© 2015 Elsevier Ltd. All rights reserved.
3408 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414
immunization persist. Of the 28 countries reporting measles out-
breaks in 2009–2010, 18 reported <90% vaccine coverage with the
first dose of Measles Containing Vaccine (MCV) [7,8]. Addition-
ally, 13 had held SIAs with <90% coverage, less than 24 months
before the outbreak [7,8]. These occurrences of measles out-
breaks post campaigns have raised concerns about the loss of
vaccine effectiveness (VE) in areas where vaccine storage, hand-
ling, distribution, and cold chain requirements are difficult to
maintain. VE is of particular concern in resource-limited sett-
ings, where refrigeration and electricity are limited, and cold
chain maintenance represents a substantial economic and logistical
burden [9].
Since 2004, the Democratic Republic of Congo’s (DRC) effort to
reduce measles mortality has consisted of a 3-pronged approach:
(1) increasing routine immunization coverage of MCV1, given at
9–11 months of age, (2) implementing SIAs to provide second
dose of MCV, and 3) expanding epidemiologic surveillance [2,10].
Despite these efforts, in 2010, DRC saw a resurgence of measles
with large scale outbreaks throughout the country [10].
DRC suffers from inadequate roads and limited electricity
and water, coupled with a lack of human resources. Conse-
quently, DRC struggles to implement international vaccination
guidelines effectively, including cold chain maintenance. Despite
these problems, reported administrative vaccine coverage levels
are high, with 223 of 516 health zones reporting >90% coverage
in 2010 and SIA coverage reportedly above 100% in many health
zones [11].
Monitoring of measles vaccine effectiveness is a useful quality-
control measure for immunization programs. It can highlight areas
of weakness and inform policy decisions [12–14]. The effectiveness
of a vaccine depends on both potency and proper adminis-
tration [12,13]. In addition to serologic studies, which assess
vaccine uptake post-vaccination, laboratory-based techniques can
test measles vaccine potency; unfortunately, the complexity and
expense of these studies make implementation in resource-limited
countries infeasible [12–15].
The test-negative case-control design has proven an effec-
tive epidemiologic approach to estimate influenza and rotavirus
VE [14–19]. This design selects cases and controls from a pool
of subjects with “measles-like illness”, which are subsequently
laboratory-confirmed positive or negative. The design is non-
traditional as the marginal ratio of cases to controls is unknown
during enrollment; however, its efficient way of mitigating selec-
tion bias due to health seeking behaviors makes it useful for
measuring VE [16,20]. Using this approach, we estimated VE using
case-based measles surveillance data collected among children in
DRC between 2010 and 2012.
2. Methods
2.1. Case-control study
We utilized data from the Case-Based (CB) measles surveil-
lance system with laboratory confirmation collected from January
1, 2010 through December 31, 2012. The system was first imple-
mented in 2002 to coincide with the start of SIAs. Individuals
presenting to health centers with measles-like-illness (MLI), or
found through active case searching, are reported to the Integrated
Disease Surveillance and Response (IDSR) and CB surveillance sys-
tems, and a blood specimen is collected and tested for measles
specific antibodies (IgM).
Human Subjects Protection boards at both the Kinshasa School
of Public Health and UCLA approved the study protocol.
Case Definition: Individuals reported to the CB measles surveil-
lance system are persons presenting with MLI, i.e. any person with
fever and maculopapular rash and cough or coryza or conjunctivitis,
or a person in whom a clinician suspects measles [21]. Individuals
were confirmed as a recent infection through either (1) Siemens
Enzygnost® indirect enzyme immunoassay (EIA) for measles IgM
antibody; or (2) by epidemiologic linkage, i.e. a case meeting the
clinical case definition who had contact with a lab-confirmed case,
with rash onset within the preceding 30 days or living in same dis-
trict [21]. All laboratory analyses were conducted at the National
Institute for Biomedical Research in Kinshasa, part of the Global
Measles and Rubella Laboratory.
Case selection: Case patients were selected if they met the case
definition (confirmed infection) and were 12–59 months of age,
living in DRC and reported to the CB measles surveillance system,
with available vaccination history.
Controls: Individuals were eligible to be a control if they were
12–59 months of age, living in DRC, reported to the CB measles
surveillance system because of MLI, and testing IgM-negative for
measles virus with available vaccination history.
Vaccination status ascertainment: Measles vaccination history
was obtained through maternal recall when vaccination cards were
unavailable (Fig. 1). All subjects were considered vaccinated if a
vaccination date was recorded and occurred >1 month prior to dis-
ease onset. Children were also considered vaccinated if ≥2 doses of
vaccine were recorded. Children with one dose and no date of vac-
cination were excluded from the main analysis because we could
not confirm if vaccination occurred before disease onset.
Age calculation: We calculated age (in months) when birth-
date was available; date of birth was subtracted from the date
of specimen collection to calculate the age in months. If an exact
birthday was unavailable, the variable “age in years” was converted
to months and added to “age in months” to create overall age (in
months).
Statistical analyses: All statistical analyses were performed using
SAS 9.4 (SAS Institute, Cary NC). Unconditional multivariate logistic
regression was used to assess risk factors for recent measles infec-
tion among children aged 12–59 months. In multivariate analyses,
vaccination status, sex, age, year of diagnosis, province, and health
zone of residence (rural/urban) were included.
VE estimation: VE was estimated using the standard case-control
protocol, with the formula VE = (1–odds vaccinated/odds unvacci-
nated) × 100, where the odds vaccinated/odds unvaccinated was
the adjusted odds ratio for receiving ≥1 dose of measles vaccine
compared with no doses [12]. Children with unknown vaccination
history were excluded from multivariate models used to estimate
VE. Attack rates in each province were low (<10%), satisfying the
rare disease assumption; the odds ratio therefore approximates
the risk ratio [12,13,22]. A sub-analysis was performed to assess
year-to-year variation in VE estimates.
3. Results
Between 2010 and 2011, reported measles cases in DRC had
surged to 134,041, with >70% occurring in children under five years
of age [23]. The epidemic may have started in Katanga province and
then spread throughout the country, with the highest incidence
rates in province Orientale, Equateur and Kasai Orientale [10,23]
(Table 1).
Since 2004, 16,789 samples have been tested for measles spe-
cific antibodies, with 28.51% confirmed positive by measles IgM
and an additional 11.19% confirmed by epidemiologic linkage. Of
the 8650 samples tested between 2010 and 2012, 4208 (48.6%)
were considered measles-positive; 1734 (41.2%) were confirmed
through epidemiologic linkage and 4379 (50.6%) tested negative.
Negative samples were further tested for rubella IgM; 25.5% were
positive.
 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414 3409

Fig. 1. Age and vaccination status of MLI cases (a) confirmed positive (b) confirmed negative in measles case-based surveillance system, Democratic Republic of Congo,
January 1, 2010—December 31, 2012.

Table 1
Measles incidence rates (per 1000,000) by year among children aged 12–59 months1 and vaccine coverage estimates by province among children 12–23 months, Democratic
Republic of Congo, 2010–2012.

20102 20113 20124

Vaccine coverage (%) Incidence (95% CI) Vaccine coverage (%) Incidence (95% CI) Vaccine coverage (%) Incidence (95% CI)

Province
Bandundu 88 220 (31-1640) – 3280 (460-23300) 77 25762 (3630-182900)
Bas-Congo 78 380 (51-2810) – 6881 (970-48950) 85 6889 (970-49010)
Equateur 74 765 (110-5470) – 1108 (160-7910) 66 35102 (4940-249200)
Kasai-Occ 54 221 (30-1620) – 5260 (740-37390) 67 16904 (2380-120000)
Kasai-Ori 62 823 (120-5880) – 160539 (22610-1140000) 58 22866 (3220-162400)
Katanga 72 11804 (1660-83830) – 259541 (36560-184) 53 30670 (4320-217800)
Kinshasa 88 373 (52-2700) – 1952 (270-13900) 89 4805 (680 (34160)
Maniema 67 412 (55-3100) – 190752 (26870-1354000) 65 30086 (4230-213800)
North-Kivu 88 592 (82-4260) – 512 (71-3690) 85 3771 (530-26820)
Orientale 62 849 (120-6140) – 13349 (1880-94800) 67 119443 (16820-848000)
South-Kivu 86 73 (1030-52060) – 27540 (3880-195600) 88 4140 (580-29450)
1
Incidence rates are based on DRC’s Integrated Disease Surveillance and Response (IDSR) and do not allow for incidence estimation among 12–23 months of age.
2
Vaccine coverage estimates are based on the Kinshasa School of Public Health coverage survey.
3
No coverage survey was conducted in 2011.
4
Vaccine coverage estimated are based on the Demographic and Health Survey (DHS) completed in 2013–2014.

3.1. Case-control study were more likely to be vaccinated (OR [95% CI] 0.17 [0.14–0.21]
than cases (Table 2). Compared to 2010, children in 2011 had
In total, 1044 cases and 1335 controls were available for (OR [95% CI] 10.12 [7.60–13.47] times the odds of developing
the analysis (Fig. 2). Among all children 12–59 months, controls measles and in 2012, 6.46 [4.82–8.64] times the odds. Compared
 3410
Table 2
Characteristics associated with measles in children aged 12–59 months, stratified by age, case-control study, Democratic Republic of Congo, 2010–2012.

12–23 months 24–59 months 12–59 Months

Cases (n = 221) Controls OR (95% CI) Cases (n = 823) Controls OR (95% CI) Cases Controls OR (95% CI)
n (%) (n = 324) n (%) n (%) (n = 1011) n (%) (n = 1044) n (%) (n = 1335) n (%)

Measles vaccination
Unvaccinated 116 (67.05) 69 (23.71) Ref 364 (59.48) 173 (20.14) Ref 480 (61.15) 242 (21.04) Ref
Vaccinated1 57 (32.95) 222 (76.29) 0.15 (0.10–0.23) 248 (40.52) 686 (79.86) 0.17 (0.14–0.22) 305 (38.85) 908 (78.96) 0.17 (0.14–0.21)
Date of Vaccination 27 (47.30) 157 (70.72) 89 (35.89) 398 (58.02) 116 (38.03) 555 (61.12)

R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414

≥2 doses 30 (52.70) 65 (29.28) 159 (64.11) 288 (41.98) 189 (61.97) 353 (38.88)
Sex
Male 114 (51.18) 172 (53.09) Ref 436 (52.98) 550 (54.40) Ref 550 (52.68) 722 (54.08) Ref
Female 107 (48.42) 152 (46.91) 1.06 (0.75–1.50) 387 (47.02) 461 (45.60) 1.06 (0.88–1.27) 494 (47.32) 613 (45.92) 1.06 (0.90–1.24)
Age (months) at diagnosis
Median age (IQR) 13 (12–17) 16 (14–20) 0.89 (0.85–0.93) 36 (24–48) 36 (31–48) 0.97 (0.96–0.98) 30 (24–36) 36 (24–48) 0.99 (0.98–1.00)
Year of diagnosis
2010 19 (8.60) 130 (40.12) Ref 47 (5.71) 347 (34.32) Ref 66 (6.32) 477 (35.73) Ref
2011 111 (50.23) 108 (33.33) 7.03 (4.06–12.18) 474 (57.59) 310 (30.66) 11.29 (8.06–15.81) 585 (56.03) 418 (31.31) 10.12 (7.60–13.47)
2012 91 (41.18) 86 (26.54) 7.24 (4.12–12.73) 302 (36.70) 354 (35.01) 6.30 (4.48–8.86) 393 (37.64) 440 (32.96) 6.46 (4.82–8.64)
Province
Kinshasa 16 (7.24) 53 (16.36) Ref 38 (4.62) 205 (20.28) Ref 54 (5.17) 258 (19.33) Ref
Bandundu 6 (2.71) 11 (3.40) 1.81 (0.58–5.66) 13 (1.58) 41 (4.06) 1.71 (0.84–3.49) 19 (1.82) 52 (3.90) 1.75 (0.96–3.19)
Bas-Congo 13 (5.88) 21 (6.48) 2.05 (0.84–5.00) 34 (4.13) 38 (3.76) 4.83 (2.71–8.60) 47 (4.50) 59 (4.42) 3.81 (2.35–6.17)
Equateur 42 (19.00) 16 (4.94) 8.70 (3.90–19.40) 110 (13.37) 50 (4.95) 11.87 (7.33–19.20) 152 (14.56) 66 (4.94) 11.00 (7.29–16.61)
Kasaï-Occidental 12 (5.43) 11 (3.40) 3.61 (1.34–9.73) 24 (2.92) 56 (5.54) 2.31 (1.28–4.17) 36 (3.45) 67 (5.02) 2.57 (1.56–4.23)
Kasai-Oriental 15 (6.79) 16 (4.94) 3.10 (1.26–7.63) 54 (6.56) 46 (4.55) 6.33 (3.75–10.69) 69 (6.61) 62 (4.64) 5.32 (3.39–8.35)
Katanga 34 (15.38) 128 (39.51) 0.88 (0.45–1.73) 130 (15.80) 364 (36.00) 1.93 (1.29–2.87) 164 (15.71) 492 (36.85) 1.59 (1.13–2.24)
Maniema 7 (3.17) 16 (4.94) 1.45 (0.51–4.14) 36 (4.37) 21 (2.08) 9.25 (4.88–17.53) 43 (4.12) 37 (2.77) 5.55 (3.27–9.42)
North-Kivu 4 (1.81) 8 (2.47) 1.66 (0.44–6.22) 8 (0.97) 34 (3.36) 1.27 (0.54–2.95) 12 (1.15) 42 (3.15) 1.36 (0.67–2.76)
Orientale 61 (27.60) 38 (11.73) 5.32 (2.67–10.61) 345 (41.92) 132 (13.06) 14.10 (9.45–21.03) 406 (38.89) 170 (12.73) 11.41 (8.09–16.09)
South-Kivu 11 (4.98) 6 (1.85) 6.07 (1.94–19.01) 31 (3.77) 24 (2.37) 6.97 (3.69–13.15) 42 (4.02) 30 (2.25) 6.69 (3.85–11.62)
Health Zone
Urban 15 (11.00) 41 (13.18) Ref 44 (9.76) 192 (19.71) Ref 59 (10.03) 233 (18.13) Ref
Rural 122 (89.05) 270 (86.82) 1.24 (0.66–2.32) 407 (90.24) 782 (80.29) 2.27 (1.60–3.22) 529 (89.97) 1052 (81.87) 1.99 (1.46–2.69)
1
Vaccinated is defined as 2 or more doses reported OR a date of vaccination more than 30 days before disease onset.
 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414
Fig. 2. Study population selection, children 12–59 months, case-control study, Democratic Republic of Congo.
to Kinshasa, children in Bas-Congo (OR [95% CI] 3.81 [2.35–6.17],
Kasai-Oriental 5.32 [3.39–8.35], Maniema 5.55 [3.27–9.42], Orien-
tale 11.41 [8.09–16.09] and South Kivu 6.69 [3.85–11.62] provinces
all had higher odds of developing measles (Table 2). Those living
in rural areas had (OR [95% CI] 1.99 [1.46–2.69] times the odds of
developing measles compared to those living in urban areas. Trends
were similar for children 12–23 months and 24–59 months of age.
In the multivariate analysis (Table 3), measles vaccination (aOR
[95% CI] 0.20 [0.15–0.26]) and year of diagnosis, 2011 6.02 [95% CI,
4.16–8.72] and 2012; 8.31 [95% CI, 5.47–12.40] remained signifi-
cant. Compared to Kinshasa, children in Bas-Congo, Kasai-Oriental,
Maniema, and South Kivu all had higher odds of developing
measles.
3.2. Vaccine effectiveness (VE)
We calculated measles vaccine VE for all children. VE was 80%
(95% CI, 74–85%) for children 12–59, 80% (95% CI, 66–88%) for chil-
dren 12–23 months of age and 81% (95% CI, 74–86%) for children
24–59 months. VE was also calculated individually for 2010, 2011
3411
and 2012, and was 86% (95% CI, 70–93%), 76% (95 CI, 65–84%), and
75% (95% CI, 61–84%) respectively.
3.3. Sensitivity analysis
Sensitivity analyses were conducted to measure the impact of
vaccination history on VE estimates (Table 4). For children 12 to 59
months, VE was 82% (95% CI, 75–87%) when including only children
with exact vaccination dates, and 77% (95% CI, 71–82%) when also
including children with missing vaccination dates, but reporting 1
or more doses of MCV.
To assess the impact of the type of measles confirmation on VE
estimates, we excluded children confirmed positive by epidemio-
logic linkage (Table 4). For children 12 to 59 months, VE was 80%
(95% CI, 74–85%). VE was 81% (95% CI, 67–89) and 81% (95% CI,
74–86%) among children 12 to 23 months and 24 to 59 months
respectively.
When we only included children lab-confirmed positive with
an exact date of last vaccination, VE was 82% (95% CI, 75–87%) for
children 12 to 59 months. VE was 83% (95% CI, 69–91%) and 81%
3412 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414

Table 3
Results of multivariate analyses of risk factors associated with measles in children aged 12–59 months of age, stratified by age, case-control study, Democratic Republic of
Congo.

12–23 months 24–59 months 12–59 months

aOR (95% CI) aOR (95% CI) aOR (95% CI)

Measles vaccination
Unvaccinated Ref Ref Ref
Vaccinated1,2 0.20 (0.12–0.34) 0.19 (0.14–0.26) 0.20 (0.15–0.26)
Sex
Male Ref Ref Ref
Female 0.90 (0.54–1.51) 0.90 (0.68–1.21) 0.91 (0.72–1.17)
Age (months) at diagnosis
Median age (IQR) 1.02 (0.95–1.09) 0.96 (0.95–0.98) 0.99 (0.98–1.00)
Year of diagnosis
2010 Ref Ref Ref
2011 5.22 (2.53–10.78) 6.69 (4.30–10.43) 6.02 (4.16–8.72)
2012 8.25 (3.71–18.37) 8.52 (5.31–13.68) 8.31 (5.57–12.40)
Province
Kinshasa Ref Ref Ref
Bandundu 1.53 (0.33–7.14) 0.44 (0.17–1.13) 0.66 (0.31–1.43)
Bas-Congo 1.18 (0.32–4.33) 4.29 (2.04–9.01) 2.83 (1.51–5.31)
Equateur 1.13 (0.25–5.18) 1.96 (0.89–4.31) 1.79 (0.90–3.56)
Kasaï-Occidental 3.21 (0.77–13.37) 1.03 (0.47–2.26) 1.43 (0.74–2.79)
Kasai-Oriental 2.89 (0.73–11.47) 4.50 (2.11–9.58) 4.02 (2.11–7.67)
Katanga 1.36 (0.46–4.04) 1.73 (1.00–3.01) 1.69 (1.04–2.73)
Maniema 0.43 (0.08–2.19) 4.27 (1.86–9.80) 2.44 (1.23–4.86)
North-Kivu 2.99 (0.57–15.75) 1.34 (0.45–3.94) 1.77 (0.73–4.29)
Orientale 0.98 (0.28–3.37) 1.52 (0.82–2.81) 1.47 (0.86–2.51)
South-Kivu 14.64 (2.52–84.93) 4.52 (1.97–10.34) 6.07 (2.93–12.58)
Health zone
Urban Ref Ref Ref
Rural 0.90 (0.30–2.69) 1.48 (0.90–2.41) 1.26 (0.81–1.94)
1
Vaccinated is defined as 2 or more doses of vaccine reported OR a date of vaccinated greater than 30 days before disease onset.
2
VE = (1 − aOR) × 100.

Table 4 measles vaccine coverage was only 73%, well below the WHO
Sensitivity analysis of vaccine effectiveness by measles case definition and vaccina-
measles mortality reduction strategy calling for ≥90% coverage
tion status, stratified by age, Democratic Republic of Congo.
[24,25]. Additionally, at the health zone level, between 25–40%
Age Vaccination Vaccination Vaccination of health zones were below the WHO recommended 80% for rou-
(months) dates date or ≥2 date or ≥1
tine immunization [26]. Provincial SIAs, follow-up immunization
doses1 doses
activities, and catch-up campaigns were too infrequent to provide
Case definition: laboratory-confirmed (IgM) or Epi-linked VE% (95% CI) sustained immunity to those not accessing routine health services
12–232 83 (69–91) 81 (67–89) 77 (63–86)
and therefore failed to prevent the recent measles epidemics [10].
24–59 80 (72–87) 81 (74–86) 78 (71–83)
12–59 82 (75–87) 80 (74–85) 77 (71–82) While our analyses for VE were restricted to a younger age group, a
large proportion of the confirmed measles cases were seen in older
Case definition: laboratory-confirmed (IgM) only VE% (95% CI)
12–23 83 (69–91) 81 (67–89) 77 (63–86)
age categories, suggesting a history of immunization deficiencies
24–59 81 (72–87) 81 (74–86) 78 (70–83) in DRC.
12–59 82 (75–87) 80 (74–85) 77 (71–82) Measles vaccine efficacy is expected to be 85% following the
1
This was the vaccination and case definition use for our case-control study. first dose at 9 to 11 months of age and 95% after the second dose at
2
There were no epi-linked children in the age group 12–23 months. ≥12 months [25,27]. Given the study population of children aged
12 to 59 months; the subjects were more likely to be exposed to
more than one dose of MCV either through routine immunization
(95% CI, 72–87%) among children 12 to 23 months and 24 to 59 and SIAs, suggesting that failure rates should be low. We performed
months respectively. additional sub-analyses assessing the change in VE when restricting
When we excluded children with samples taken less than 3 the analysis to children with an exact vaccination date. We also
days or more than 28 days after disease onset, VE was 77% (95 CI, assessed the VE with the inclusion of children reporting one or more
70–87%). doses of vaccine. VE estimates were generally stable regardless of
our definition of vaccination history ranging from 77–81%. While
4. Discussion we were able to control for provincial differences, our VE estimates
represent national averages. With the large diversity seen across
Despite the fact that DRC has a country specific measles control the country, additional stratified studies are needed, to tease out
strategy and target elimination date of 2020, large-scale measles differences between provinces and health zones.
outbreaks have persisted throughout the country [10]. This sug- Calculated VE estimates are lower than expected regardless of
gests the need to re-evaluate DRC’s immunization program in terms year. Extensive rain, inadequate transport, impassable roads, pop-
of immunization strategy, control and prevention. In settings where ulation movement, and insecure regions make a large majority of
accessibility to health services is limited, efforts to estimate the VE the country inaccessible. Furthermore, improper staff training and
of delivered vaccine can have an important impact. fuel shortages have resulted in an unreliable cold chain in many
Routine immunization coverage remains sub-optimal through- areas. The measles vaccine requires a 4–8 ◦ C cold chain until use;
out the country. In 2012, WHO/UNICEF estimated that national then after reconstitution, the vaccine faces the risk of bacterial
 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414
contamination and loss of potency with exposure to high temper-
atures and light after only 6 h [7]. Using a vaccine with reduced
effectiveness, coupled with a 15% vaccination failure rate at 9 to 11
months of age, will lead to lower vaccine-induced immunity in the
population [28,29]. Additionally, past SIA’s have included children
as young as 6 months and vaccine failure may occur due to decay-
ing maternal antibodies and the lack of maturity of the immune
system [24,30,31].
Our analyses are subject to a number of limitations. The measles
surveillance system has not been fully implemented in DRC. Case
reporting occurs at the health center level and is then transmit-
ted to the health zone, where it is aggregated and sent to the
national level. The CB surveillance system and the IDSR are not
fully integrated. The CB surveillance system represents just a small
percentage (3.7%) of those cases reported via the IDSR. Further-
more, since confirmed outbreaks only require three1 IgM positive
cases, it is probable that a health zone may not have sent addi-
tional samples for laboratory confirmation. Despite the fact that
reporting “0” cases is part of the country’s surveillance protocol,
inconsistencies are common and 17–42% of health zones remain
“silent” (do not report) each year [10]. Depending on their true
measles and vaccination statuses, these missed cases may have
led to biased VE estimates. A majority of the vaccination informa-
tion was likely obtained through maternal recall and it is unclear
which children had available vaccination records or how the vac-
cination was obtained (RI or SIA). While SIAs are planned over
shorter periods of time, RI services require consistent monitor-
ing and maintenance and may result in differential VE estimates. A
number of children (7.03%) were recorded as receiving 3 or more
measles doses, suggesting possible confusion with polio campaigns.
However, bias is likely to be non-differential because both cases and
controls were tested based on measles-like symptoms, leading to a
possible underestimate of vaccine effectiveness. The high propor-
tion of subjects with unknown vaccination status (43.3%) may have
led to biased VE estimates.
We did not have data on important confounders, particularly
the nutritional status of each child. Malnourishment is strongly
associated with more severe cases of measles and therefore more
likely to be reported [31–33]. Additionally, nutritional status and
malaria infection has been associated with reduced vaccine uptake
and may partially explain lower VE estimates in a malaria endemic
country where a majority of children are malnourished [34–41].
Children in Bas-Congo, Kasai-Oriental, Maniema, and Orientale all
had higher odds of developing measles and a higher prevalence of
malaria (44.1–49.1%) [42].
The indirect ELISA used to confirm measles generally works well
and has high sensitivity (89.9–97.4%; higher after the first week of
rash onset) and specificity (92.8–100%) for samples collected within
3–28 days after rash onset [43,44]. However, even a test that is
90% specific can result in a large number of non-measles MLI cases
being attributed to measles and in equal proportions among the
vaccinated and unvaccinated, resulting in an underestimate of VE
[15]. Our analysis on a subset of children with samples collected
and tested within 3–28 days of rash onset (as a proxy for rash onset)
resulted in a slight decrease in VE. VE estimates are generally biased
towards the null when the diagnosis is less specific, however our VE
results were fairly stable across both case definitions. While, there
was a slight reduction in VE when we modified the definition of
vaccinated to include one or more doses, this is likely explained by
a reduced vaccine efficacy at younger ages.
While the test-negative design represents a simple approach to
VE estimation, the design is sensitive to confounding by disease
1 tion data. WHO; 2014.
Health zones must send a minimum of five samples, however an outbreak is
confirmed when 3 samples are IgM positive.
3413
severity [16]. The disease is less likely to be severe in children with
a history of prior measles vaccination. Moreover, severity is likely
to affect the probability of an individual seeking care and subse-
quently the probability of recruitment into the study. If severity
varies by measles and non-measles cases and changes the probabil-
ity of health seeking behaviors, VE estimates may be substantially
confounded and overestimated [16].
The validity of any case control study is dependent on the con-
trol group representing the exposure distribution of the source
population. This design provides some reassurance that our con-
trols did emerge from the source population that gave rise to the
cases, considering that all subjects present with a measles-like ill-
ness [14]. However, vaccination status is likely to be associated with
improved health seeking behaviors and obtaining care at a health
center, thus limiting the generalizability of our results to the entire
population of DRC.
Large-scale outbreaks in DRC and other African countries
demonstrate the virus’s ability to easily re-enter communities lack-
ing sustained levels of population immunity [5]. These outbreaks
underscore the importance of re-evaluating measles dynamics
and determining the elements necessary to attain global targets
for reducing transmission and mortality. Accurate immunization
coverage data must be collected and reasons why vaccination
was not received should be assessed in order to inform new
and improved vaccination strategies. Additional studies assessing
measles VE should be conducted, specifically in inaccessible areas
to determine if vaccine distribution and cold chain maintenance
are reasons for reduced estimates. This information can assist in
refining prevention and control strategies. New interventions such
as a remote temperature monitoring system, computerized immu-
nization information systems, expanding the use of solar freezers
to community-level clinics, and creating new supply chain mod-
els should be considered for use in DRC [45,46]. Heavily monitored
program scale-up and an increased financial commitment will be
essential as we push forward toward measles elimination goals.
Conflict of interest statement
The authors do not have a commercial or other association that
might pose a conflict of interest.
There was no funding source for this study. The corresponding
author had full access to all the data in the study and had final
responsibility for the decision to submit for publication.
Acknowledgements
Many individuals contributed to the design of this study and
manuscript. We would especially like to give thanks to Heather
Scoobie and James Goodson from the Global Immunization Divi-
sion at Centers for Disease Control and Prevention (GID-CDC) for
their help in preparing the dataset. Additionally, we would like to
thank the Faucett Family Foundation for their support through-
out this project. We also wish to acknowledge the DRC Ministry of
Health, World Health Organization, and UNICEF country offices in
Kinshasa and National Institute for Biomedical Research for their
organizational and technical laboratory support.
References
[1] World Health Organization. Measles. World Health Organization; 2012.
[2] Alberti KP, King LA, Burney M, Illunga BK, Grais RF. Reactive vaccination as
an effective tool for measles outbreak control in measles mortality reduction
settings, Democratic Republic of Congo, 2005–2006. Int Health 2011;2:65–8.
[3] World Health Organization. Measles. Fact Sheet No. WHO. 286; 2015.
[4] World Health Organization. United Nations Children’s Fund. Global immuniza-
[5] Moss WJ, Griffin DE. Measles. Lancet 2012;379:153–64.
3414 R.H. Doshi et al. / Vaccine 33 (2015) 3407–3414
[6] Simons E, Ferrari M, Fricks J, Wannemuehler K, Anand A, Burton A, et al. Assess-
ment of the 2010 global measles mortality reduction goal: results from a model
of surveillance data. Lancet 2012;379:3.
[7] Goodson JL, Chu SY, Rota PA, Moss WJ, Featherstone DA, Vijayaraghavan M,
et al. Research priorities for global measles and rubella control and eradication.
Vaccine 2012;30:4709–16.
[8] Centers for Disease Control and Prevention. Measles outbreaks and progress
toward measles preelimination—African region, 2009–2010. MMWR Morb
Mortal Wkly Rep 2011;60:374–8.
[9] Zhang J, Pritchard E, Hu X, Valentin T, Panilaitis B, Omenetto FG, et al. Stabiliza-
tion of vaccines and antibiotics in silk and eliminating the cold chain. Proc Natl
Acad Sci USA 2012;109:11981–6.
[10] Programme Elargi De Vaccination. Plan Stratégique d’élimination de la Rouge-
ole en RDC: 2012–2020. Democratic Republic of Congo; 2012.
[11] WHO/UNICEF. Immunization summary—a statistical reference containing data
through 2011. WHO/UNICEF; 2013. p. 200.
[12] Orenstein WA, Bernier RH, Dondero TJ, Hinman AR, Marks JS, Bart KJ, et al. Field
evaluation of vaccine efficacy. Bull World Health Organ 1985;63:1055–68.
[13] Orenstein WA, Bernier RH, Hinman AR. Assessing vaccine efficacy in the field:
further observations. Epidemiol Rev 1988;10:38.
[14] De Serres G, Skowronski D, Wu X, Ambrose C. The test-negative design: valid-
ity, accuracy and precision of vaccine efficacy estimates compared to the gold
standard of randomised placebo-controlled clinical trials. Euro Surveill 2013;
18:9.
[15] Orenstein EW, De Serres G, Haber MJ, Shay DK, Bridges CB, Garguillo P,
et al. Methodologic issues regarding the use of three observational study
designs to assess influenza vaccine effectiveness. Int J Epidemiol 2007;36:
623–31.
[16] Foppa IM, Haber M, Ferdinands JM, Shay DK. The case test-negative design for
studies of the effectiveness of influenza vaccine. Vaccine 2013;31:3104–9.
[17] Barret AS, Donnell JO, O’Hora A, Collins C, Coughlan S, Joyce M, et al.
Effectiveness of 2010/2011 seasonal influenza vaccine in Ireland. Ir Med J
2012;105:39–42.
[18] Staat MA, Payne DC, Donauer S, Weinberg GA, Edwards KM, Szilagyi PG, et al.
Effectiveness of pentavalent rotavirus vaccine against severe disease. Pediatrics
2011;128:e267–75.
[19] Donauer S, Payne DC, Edwards KM, Szilagyi PG, Hornung RW, Weinberg GA,
et al. Determining the effectiveness of the pentavalent rotavirus vaccine against
rotavirus hospitalizations and emergency department visits using two study
designs. Vaccine 2013;31:2692–7.
[20] Jackson ML, Nelson JC. The test-negative design for estimating influenza vaccine
effectiveness. Vaccine 2013;31:2165–8.
[21] World Health Organization. WHO-recommended standards for surveillance of
selected vaccine-preventable disease. World Health Organization; 2003. p. 62.
[22] Greenland S, Thomas DC. On the need for the rare disease assumption in case-
control studies. Am J Epidemiol 1982;116:547–53.
[23] Mancini S, Coldiron ME, Ronsse A, Ilunga BK, Porten K, Grais RF. Description of
a large measles epidemic in Democratic Republic of Congo, 2010–2013. Confl
Health 2014;8:9.
[24] World Health Organization. Measles vaccines: WHO position paper. Releve epi-
demiologique hebdomadaire/Section d’hygiene du Secretariat de la Societe des
Nations, 84; 2009. p. 349–60.
[25] Kidd S, Ouedraogo B, Kambire C, Kambou JL, McLean H, Kutty PK, et al. Measles
outbreak in Burkina Faso, 2009: a case-control study to determine risk factors
and estimate vaccine effectiveness. Vaccine 2011;30:8.
[26] Masresha BKR, Eshetu M, Katsande R, Luce R, Fall A, Dosseh A, et al. Progress
toward measles preelimination—Africa Region, 2011–2012. Morb Mortal Wkly
2014;63:285–91.
[27] Sudfield C, Navar A, Halsey N. Effectiveness of measles vaccination and vitamin
A treatment. Int J Epidemiol 2010;39(Suppl. 1):i48–55.
[28] Grout LM, Hurtado A, Francois N, Fermon G, Chatelain F, Harczi A, et al. Measles
in Democratic Republic of Congo: an outbreak description from Katanga,
2010–2011. BMC Infect Dis 2013;13:6.
[29] World Health Organization. Measles vaccines: WHO position paper. Weekly
epidemiology record, vol. 84; 2009. p. 349–60.
[30] Gans HA, Arvin AM, Galinus J, Logan L, DeHovitz R, Maldonado Y. Deficiency of
the humoral immune response to measles vaccine in infants immunized at age
6 months. JAMA: J Am Med Assoc 1998;280:527–32.
[31] Halsey NA, Boulos R, Mode F, Andre J, Bowman L, Yaeger R, et al. Response to
measles vaccine in Haitian infants 6 to 12 months old: influence of mater-
nal antibodies, malnutrition, and concurrent illness. N Engl J Med 1985;
313:5.
[32] Reddy V, Bhaskaram P, Raghuramulu N, Milton RC, Rao V, Madhusudan J,
et al. Relationship between measles, malnutrition, and blindness: a prospective
study in Indian children. Am J Clin Nutr 1986;44:7.
[33] Bhaskaram P. Micronutrient malnutrition, infection and immunity: an
overview. Nutr Rev 2002;60:5.
[34] Kizito D, Tweyongyere R, Namatovu A, Webb EL, Muhangi L, Lule SA, et al.
Factors affecting the infant antibody response to measles immunisation in
Entebbe-Uganda. BMC Public Health 2013;13:619.
[35] McMurray DN, Loomis SA, Casazza LJ, Rey H. Influence of moderate malnutrition
in morbidity and antibody response following vaccination with live,attenuated
measles virus vaccine. Bull Pan Am Health Organ 1979;13:52–7.
[36] Rytter MJ, Kolte L, Briend A, Friis H, Christensen VB. The immune sys-
tem in children with malnutrition—a systematic review. PLoS ONE 2014;9:
e105017.
[37] Idris S, El Seed AM. Measles vaccination in severely malnourished Sudanese
children. Ann Trop Paediatr 1983;3:63–7.
[38] Powell GM. Response to live attenuated measles vaccine in children with severe
kwashiorkor. Ann Trop Paediatr 1982;2:143–5.
[39] Hafez M, Aref GH, Mehareb SW, Kassem AS, El-Tahhan H, Rizk Z, et al. Antibody
production and complement system in protein energy malnutrition. J Trop Med
Hyg 1977;80:36–9.
[40] Williamson WA, Greenwood BM. Impairment of the immune response to vac-
cination after acute malaria. Lancet 1978;1:1328–9.
[41] Greenwood BM, Bradley-Moore AM, Palit A, Bryceson ADM. Immunosuppres-
sion in children with malaria. Lancet 1972;299(7743):169–72.
[42] Minister of Plan, Minister of Public Health, ICF International. Demographic
and Health Survey, Democratic Republic of Congo 2013–2014. Rockville, MD:
Minister of Plan, Minister of Public Health, ICF International; 2014.
[43] Bellini WJ, Helfand RF. The challenges and strategies for laboratory diagnosis
of measles in an international setting. J Infect Dis 2003;187:8.
[44] Sampedro A, Rodriguez-Granger J, Gomez C, Lara A, Gutierrez J, Otero A. Com-
parative evaluation of a new chemiluminiscent assay and an ELISA for the
detection of IgM against measles. J Clin Lab Anal 2013;27:477–80.
[45] PATH, World Health Organization. Collaborating with countries to improve
supply chain. PATH, World Health Organization; 2014.
[46] Path, World Health Organization. Stimulating innovation in the vaccine cold
chain equipment industry. In: Optimize: immunization systems and technolo-
gies for tomorrow. Path, World Health Organization; 2013.