S U P P L E M E N T A R T I C L E
Influenza Sentinel Surveillance in Rwanda,
Thierry Nyatanyi,1Richard Nkunda,2Joseph Rukelibuga,3Rakhee Palekar,4Marie Aimée Muhimpundu,1
Adeline Kabeja,1Alice Kabanda,2David Lowrance,3Stefano Tempia,5,8Jean Baptiste Koama,3David McAlister,3
Odette Mukabayire,2Justin Wane,6Pratima Raghunathan,3Mark Katz,7and Corine Karema1
1Rwanda Biomedical Center, Ministry of Health,2National Reference Laboratory,3Centers for Disease Control and Prevention (CDC) Kigali, Rwanda;
4US Center for Disease Control and Prevention, (CDC) Influenza Division, Atlanta, Georgia;5CDC, Johannesburg, South Africa;6King Faisal Hospital,
Kigali, Rwanda;7CDC, Nairobi, Kenya; and8National Institute of Communicable Diseases, Johannesburg, South Africa
demiology of influenza and monitor for the emergence of novel influenza A viruses. We report surveillance results
from August 2008 to July 2010.
Methods. We conducted ISS by monitoring patients with influenza-like illness (ILI) and severe acute respira-
tory infection (SARI) at 6 hospitals. For each case, demographic and clinical data, 1 nasopharyngeal specimen,
and 1 oropharyngeal specimen were collected. Specimens were tested by real-time reverse-transcription polymer-
ase chain reaction for influenza A and B viruses at the National Reference Laboratory in Rwanda.
Results. A total of 1916 cases (945 ILI and 971 SARI) were identified. Of these, 29.2% (n =276) of ILI and
10.4% (n =101) of SARI cases tested positive for influenza. Of the total influenza-positive cases (n = 377), 71.8%
(n = 271) were A(H1N1) pdm09, 5.6% (n =21) influenza A(H1), 7.7% (n =29) influenza A(H3), 1.6% (n =6)
influenza A (unsubtyped), and 13.3% (n =50) influenza B. The percentage of positivity for influenza viruses was
highest in October–November and February–March, during peaks in rainfall.
Conclusions. The implementation of ISS enabled characterization of the epidemiology and seasonality of in-
fluenza in Rwanda for the first time. Future efforts should determine the population-based influenza burden to
inform interventions such as targeted vaccination.
In 2008, Rwanda established an influenza sentinel surveillance (ISS) system to describe the epi-
Influenza virus infection is a major public health
concern worldwide, resulting in an estimated 500 000
deaths annually [1–9]. The epidemiology of influenza
is well characterized in the temperate areas of the
Northern and Southern Hemispheres . However,
little is known about the epidemiology, seasonality,
and burden of influenza in tropical areas, especially in
sub-Saharan Africa, where the severity of infections
is likely compounded by malnutrition, limited supplies
of antibiotics to treat secondary bacterial infections
, poverty, and poor access to healthcare .
Rwanda is a landlocked developing country with an
area of 26338 km2, situatedin CentralAfrica
immediately south of the equator. It has an equatorial
climate with moderate temperatures and 2 rainy
seasons, from March through June and from October
through December. The average annual temperature
ranges from 18°C to 24°C. Given the small geographical
extent of the country, no climatic differences are ob-
served . The 2010 estimated population is approxi-
mately 10.4 million and the population density (350
persons/km2) is among the highest in sub-Saharan
Prior to 2008, there was no surveillance for influenza
in Rwanda. In July 2008 the Rwandan Ministry of
Health, in collaboration with the US Centers for
Disease Control and Prevention (CDC), established an
influenza sentinel surveillance (ISS) system with the
following objectives: to describe the epidemiology and
seasonality of influenza, to monitor for the emergence
of novel influenza viruses, to describe the circulating
influenza virus types and subtypes, and to detect influ-
enza outbreaks in a timely manner.
Correspondence: Joseph Rukelibuga, DVM, MSc, CDC Rwanda, US Embassy,
2657 Avenue de la Gendarmerie, BP 28 Kigali 2657, Rwanda (email@example.com.
The Journal of InfectiousDiseases2012;206(S1):S74–9
Published by Oxford University Press on behalf of the Infectious Diseases Society of
S74 • JID 2012:206 (Suppl 1) • Nyatanyi et al
by guest on November 10, 2015
From July 2008 to May 2009, sentinel surveillance was estab-
lished in 4 public hospitals. In May 2009, following the emer-
gence of influenza A(H1N1)pdm09 in North America in April
2009, the surveillance was extended to 2 additional sites
5 regions, and selection required capacity to collect and ship
samples to the National Reference Laboratory (NRL) in Kigali
and the site’s interest in participating in the ISS program. Each
hospital had pediatric, adult, and maternity inpatient wards and
ambulatory care services, and all services participated in the ISS.
One referral hospital and one district hospital were located in
the capital city of Kigali, while the other 3 district hospitals and
1 referral hospital were located in each of the country’s 4 prov-
inces. Each district hospital’s catchment population was estimat-
ed to be approximately 300000 persons . Unlike district
hospitals, the true catchment populations of referral hospitals
are unknown since people from all over the country attend re-
ferral hospitals for primary and secondary care.
The Pan American Health Organization/CDC Generic Proto-
col for Influenza Sentinel Surveillance  was adapted to
generate a country-specific protocol for the ISS. At each senti-
nel site, 2 syndromic case definitions were used—influenza-
like illness (ILI) and severe acute respiratory infection (SARI).
An ILI case was defined as an outpatient aged ≥2 months
with fever ≥38°C and cough or sore throat in the absence of
another diagnosis, with symptom onset within 72 hours of
presentation. A SARI case in persons ≥5 years of age was
defined as a hospitalized patient with fever ≥38°C, cough, and
shortness of breath or difficulty breathing, with the onset of
signs and symptoms within 7 days of presentation. The SARI
case definition for children 2 months through <5 years of age
was based on the World Health Organization’s Integrated
Management of Childhood Illness (IMCI) definition for pneu-
monia and severe pneumonia , and was defined as a hos-
pitalized patient with cough or difficulty breathing, and at
least 1 danger sign (unable to drink or breastfeed, lethargic or
unconscious [ensure patient is awake], vomits everything [not
only occasional], convulsions, nasal flaring, grunting, oxygen
saturation <90%, chest indrawing [retractions under ribcage
on inspiration], stridor in a calm child, tachypnea [2 months–
1 year: relative risk (RR) >50; 1 year–5 years: RR>40]), with
symptom onset within 7 days of presentation. Infants aged <2
months with fast breathing (≥60 breaths per minute) were ex-
cluded as they are referred for severe bacterial infection. Cases
were identified by dedicated surveillance officers (eg, trained
nurses) in collaboration with clinicians providing treatment. For
all eligible SARI cases and the first 2 ILI cases each day, a
questionnaire was completed that included demographic, clini-
cal, and epidemiological information. In addition, a nasopha-
ryngeal and an oropharyngeal swab were collected from each
patient and placed in the same cryovial containing 1 mL of viral
transport medium and stored at 4°C on site for a maximum of
72 hours until they could be shipped to the NRL.
At the NRL, RNA extractions were performed on specimens
stored at −70°C using the QIAamp Viral RNA Isolation Kit
(Qiagen). Extracted RNA was amplified using commercially
prepared master mix (Invitrogen Corp) and primers provided
by the CDC using standard real-time reverse-transcription po-
lymerase chain reaction (RT-PCR) procedures using 7500
standard real-time PCR system (Applied Biosystems) to detect
for the presence of human ribonucleoprotein (RNP) and influ-
enza type A and B viruses. Influenza A–positive specimens
were further subtyped with H1, H3, H5, and A(H1N1)pdm09
subtypes. The RNP was assessed to see whether or not the
samples contain sufficient human cells for internal quality
control purposes [14, 15].
We obtained data from the Rwandan meteorological service
for temperature, relative humidity, and rainfall during the sur-
veillance period from August 2008 through July 2010.
We assessed the demographic, clinical, and epidemiological char-
acteristics of influenza positive and negative ILI and SARI cases
as well as the temporal patterns of influenza virus circulation.
Bivariate analysis was performed using χ2and Fisher exact tests
with P values ≤.05. Multivariate logistic models were used to
assess the association of influenza positivity with age group, sex,
rainfall, temperature, humidity, and 7 other risk factors
(Table 1). Model selection was performed using a backward se-
lection (probability of entry [pe]=0.05 and pe=0.10). The final
model is an additive model that retains a total of 7 covariates.
For this analysis, interaction among different covariates was not
assessed. EpiInfo version 3.5.1 (CDC) and Stata software version
10 (StataCorp) were used for the analysis.
The ISS protocol was reviewed and deemed to be nonresearch
by the Rwandan Ministry of Health and the CDC. Verbal
consent was obtained from all patients prior to data and speci-
men collection. For children aged <15 years, verbal consent
was obtained from a parent or legal guardian.
Demographic and Exposures Among ILI and SARI Cases
From August 2008 through July 2010, we enrolled 1916 cases
(ILI, n=945; SARI, n=971). Among SARI cases, 71.8% (697/
Influenza Sentinel Surveillance in Rwanda • JID 2012:206 (Suppl 1) • S75
by guest on November 10, 2015
971) were <5 years of age whereas among ILI, 72% (681/945)
were >5 years of age. The ratio of males to females was 0.83
for ILI and 1.02 for SARI cases. The median age was 19 years
(range, 2 months–86 years) for ILI and 3 years (range, 2
months–93 years) for SARI cases. Among ILI and SARI cases,
29.2% (276/945) and10.4% (101/971), respectively, tested posi-
tive for influenza. Influenza was associated with age and the
influenza detection rate was higher in those aged 5–14 years
(P<.01; odds ratio [OR], 8.145) and 15–49 years (P<.01; OR,
4.89) for both ILI and SARI cases followed by those aged 6–23
months (P< .05; OR, 3.266) for SARI cases (Tables 1 and Sup-
plementary Table 2). In all categories, few cases reported preg-
nancy (1.0%), office worker (2.1%), or healthcare worker
status (6.2%). Risk factor analysis showed that student status
(12.2%) (OR, 1.667; 95% confidence interval [CI], .164–2.287)
and having children at home (27.3%) (OR, 1.842; 95% CI,
.406–2.414) were significantly more frequent exposures in in-
fluenza-positive ILI cases compared to influenza-negative ILI
For all case types, and for influenza-positive and influenza-
negative cases, fever, cough, headache, sore throat, and lethar-
gy were the most common symptoms reported at illness onset
(Supplementary Table 1). Nausea and headache were signifi-
cantly more frequent in influenza-positive compared with in-
fluenza-negative cases (P<.05).
Among all SARI- and ILI-positive influenza cases, the
median temperature on presentation was similar, ranging
from 38.2°C to 38.7°C (Supplementary Table 1). The median
pulse oximetry on presentation was 97% (range, 90–100) for
influenza-positive ILI cases, 95% (range, 72–100) for influen-
za-positive SARI cases (Supplementary Table 1). There was no
difference in pulse oximetry on presentation between influen-
za-positive and influenza-negative cases. The median time
from symptom onset to presentation was also similar among
influenza-positive and influenza-negative cases and was 3 days
(range, 1–14 days) for influenza-positive ILI cases, 3.5 days
(range, 2–15 days) for influenza-positive SARI cases. Of all
Rwanda (August 2008–July 2010)
Demographic and Epidemiological Characteristics of Influenza-like Illness and Severe Acute Respiratory Infection Cases,
ILI (n=945) SARI (n=971)b
ParametersInfluenza Negative Influenza Positive P Value Influenza Negative Influenza Positive P Value
Median, y(95% CI)
Children at home
Close contact with similar illness
in last 3 wk
Close contact with a person who died of
undiagnosed acute respiratory illness
669 (70.8) 276 (29.2)870 (89.6) 101 (10.4)
2 mo–86 y
3 mo–62 y
.02 1 (1–1)
2 mo–93 y
2 mo–77 y
375 (56.1) 142 (51.4) .196434 (49.9) 47 (46.5).524
5 (0.8) 3 (1.1).7033 (0.4) 0 (0.0) .723
Data are presented as no. (%) unless otherwise specified.
Abbreviations: CI, confidence interval; ILI, influenza-like illness; SARI, severe acute respiratory infection.
aAmong cases in women of childbearing age (15–49 y).
bSARI definition differed for <5 y.
S76 • JID 2012:206 (Suppl 1) • Nyatanyi et al
by guest on November 10, 2015
positive influenza ILI cases, 68.1% and 98.9% presented
within 3 days and 7 days of symptom onset, respectively. For
all positive influenza SARI cases, 50% and 95% presented
within 3 days and 7 days, respectively, of symptom onset. The
mean duration of hospitalization was 5 days for 58 influenza-
positive SARI cases with medical records available. The dura-
tion of hospitalization ranged from 2 to 15 days except for 1
patient who had influenza A(H3) coinfection with hepatitis C
and tuberculosis for whom the hospital stay was 42 days.
Influenza Types and Subtypes
Of all influenza cases detected during the surveillance period,
86.7% (n =327) were influenza A viruses and 13.3% (n =50)
were influenza B viruses (Supplementary Table 2). The major-
ity of influenza in the 2008 season (July–December) was
caused by influenza A(H3) viruses. In the 2009 season
(January–December), initially, influenza A(H3) viruses pre-
dominated, with some cocirculation of influenza A(H1)
viruses, but in October, 2009 influenza A(H1N1)pdm09
largely replaced all other influenza viruses. In 2010 (January–
July), influenza A(H1N1) pdm09 continued to predominate
even though it cocirculated with influenza A(H3) and influen-
The peaks in the percent positivity for influenza were seen
in October–November 2008, February–March 2009, October–
November 2009, and February–March 2010, roughly coincid-
ing with the peaks in rainfall (Figure 1). During these same
periods, overall numbers of SARI and ILI cases were higher.
Overall, the temperature during this period was very stable but
became slightly cooler with the onset of the rainy seasons.
Analysis showed that there was significant association of influ-
enza positivity with average monthly rainfall (P<.001) and
average monthly relative humidity (P< .05). An increasing
trend was found between rainfall and percent positivity of in-
fluenza (OR, 1.197; 95% CI, .996–1.439) (Supplementary
This is the first description of the seasonality and epidemiology
of influenza in Rwanda. Key findings from this 24-month sur-
veillance period relate to seasonality and influenza A and B
viruses type and subtypes for the ultimate purpose of vaccine
development, monitoring trends in virus spread and virus vari-
ations compared to annually recommended vaccines strains.
Although influenza viruses were detected year-round, there
appear to be 2 distinct periods of increased transmission of
influenza viruses annually, in February–March and October–
November, which match the peak rainy seasons. These find-
ings are consistent with patterns observed in some other
tropical countries, including Bangladesh, Kenya, Lao People’s
Democratic Republic (Lao PDR), Singapore, and Thailand,
where increased influenza activity was associated with in-
creased rainfall [5, 7, 16–20]; also, in some of these countries,
increased influenza activity was associated with higher relative
humidity and higher temperatures.
The virologic surveillance of influenza during these 2 years
in Rwanda revealed that influenza A virus circulation predom-
inated over influenza B virus circulation: in 2008, influenza A
(H3) subtype viruses predominated; in 2009 and 2010, influ-
enza A(H1N1)pdm09 viruses predominated. These findings
are consistent with circulation patterns documented in South
Africa by the viral watch program, where in 2009, influenza A
(H3) was the dominant influenza virus subtype before influen-
za A(H1N1)pdm09 was detected . In Kenya, similar pat-
terns were observed with a predominance of influenza A(H3)
subtype in 2008 .
Influenza virus infection was detected in all age groups, but
the relative proportion of positive influenza cases among SARI
cases <5 years of age was low compared to the high proportion
of SARI cases <5 years of age. This could represent the fact
that other respiratory pathogens, including respiratory syncy-
tial virus, play a more important role in respiratory syndromes
Rwanda (August 2008–July 2010)
Influenza Type and Subtypes Among Influenza-like Illness and Severe Acute Respiratory Infection Cases by Age Group in
All data are presented as no. (%).
Abbreviations: ILI, influenza-like illness; SARI, severe acute respiratory infection.
Influenza Sentinel Surveillance in Rwanda • JID 2012:206 (Suppl 1) • S77
by guest on November 10, 2015
in young children [1, 23]. Finally, in Rwanda, influenza was
associated with 10.4% of SARI cases. The proportion of influ-
enza positivity among hospitalized pneumonia cases is consis-
tent with findings from Lao PDR, Peru, and Argentina and
offers further evidence that a considerable proportion of pneu-
monia is associated with seasonal influenza [1, 4, 9, 16, 17, 19,
23]. The proportion of 29.2% influenza-positive (influenza A
and B) among ILI cases is consistent with 34.8% found in Peru
from samples collected during routine surveillance ; however,
it is higher than results from national hospital-based surveillance
in Bangladesh, where 10% of outpatient ILI case-patients had
influenza . The median length of hospital stay is consistent
with 4 days (range, 0–42 days) across all age groups reported in
Thailand . The higher proportion of influenza among
school-aged children suggests that school-based interventions
(eg, school closures, vaccination) may be worth considering for
preventing or mitigating influenza outbreaks in Rwanda.
There are several important limitations to the data gathered
through this surveillance system. First, while we were able to
obtain epidemiologic data, we were not able to estimate the
burden of influenza because our surveillance system currently
does not capture denominator data. Understanding the pro-
portion of influenza-associated outpatient and inpatient visits
and influenza virus infection rates in the population could
inform policy decisions about interventions such as use of in-
fluenza vaccine. Second, our surveillance system does not
perform viral culture and isolation of RT-PCR–positive
influenza viruses, and subsequently does not perform any
genetic and antigenic characterization of the circulating influ-
enza. Therefore our data are limited to identifying influenza
virus types and influenza A virus subtypes, and we do not
have information on circulating strains and antiviral sensitivi-
ties. Finally, there are only 24 months of data available, in-
cluding the period of initial implementation of the system,
which limits the ability to establish clear temporal and season-
al influenza trends across the surveillance sites.
In conclusion, we found that influenza causes inpatient and
outpatient disease year-round in Rwanda, with seasonal peaks
during rainy periods. The molecular diagnostic methods
allowed the NRL to promptly detect the influenza A(H1N1)
pdm09 and to timely respond appropriately.
Future efforts should focus on continuing to strengthen the
surveillance system, including evaluating the case definitions,
ensuring that the system is capturing cases at high-risk of
severe disease, adding viral culture, and introducing popula-
tion-based methods to estimate disease burden.
Supplementary materials are available at The Journal of Infectious Diseases
online (http://jid.oxfordjournals.org/). Supplementary materials consist of
data provided by the author that are published to benefit the reader. The
posted materials are not copyedited. The contents of all supplementary
data are the sole responsibility of the authors. Questions or messages
regarding errors should be addressed to the author.
Percentage of positivity of influenza by influenza types and subtypes, rainfall by month, Rwanda (August 2008–July 2010).
S78 • JID 2012:206 (Suppl 1) • Nyatanyi et al
by guest on November 10, 2015
Notes Download full-text
tion of surveillance officers at sentinel surveillance hospitals: University
Teaching Hospital of Kigali, University Teaching Hospital of Butare,
Kibungo District Hospital, Gihundwe District Hospital, Kibagabaga Dis-
trict Hospital, and Ruhengeri District Hospital. We are also grateful to Dr
George Rutherford of the University of California, San Francisco, for pro-
viding valuable training in the process of manuscript preparation. We ac-
knowledge the contributions of Dr Bassirou Chitou to the statistical
The Rwandan ISS system is supported by a cooper-
ative agreement between the Rwandan Ministry of Health and the CDC.
This work was supported by the CDC (cooperative agreement number
IP00158). Its contents are solely the responsibility of the authors and do
not necessarily represent the view of CDC.
Potential conflicts of interest.
All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential
Conflicts of Interest. Conflicts that the editors consider relevant to the
content of the manuscript have been disclosed.
The authors acknowledge the effort and contribu-
1. Viboud C, Alonso WJ, Simonsen L. Influenza in tropical regions. PloS
Med 2006; 3:e89.
2. Simonsen L. The global impact of influenza on morbidity and mortal-
ity. Vaccine 1999; 17:S3–10.
3. Cohen C, Simonsen L, Kang JW, et al. Elevated influenza-related
excess mortality in South African elderly individuals, 1998–2005. Clin
Infect Dis 2010; 51:1362–1369 doi: 10.1086/657314
4. Berkley JA, Munywoki P, Ngama M, et al. Viral etiology of severe pneu-
monia among Kenyan infants and children. JAMA 2010; 304:963.
5. Finkelman BS, Viboud C, Koelle K, et al. Global patterns in seasonal
activity of influenza A/H3N2, A/H1N1, and B from 1997 to 2005:
viral coexistence and latitudinal gradients. PLoS One 2007; 2:e1296.
6. Thompson WW, Comanor L, Shay DK. Epidemiology of seasonal in-
fluenza: use of surveillance data and statistical models to estimate the
burden of disease. J Infect Dis 2006; 194(Suppl 2):S82–91.
7. Gachara G, Ngeranwa J, Magana JM. Influenza virus strains in
Nairobi Kenya. J Clin Virol 2006; 35:117–8.
8. Zaman RU, et al. Influenza in outpatient ILI case-patients in national
hospital-based surveillance, Bangladesh, 2007–2008. PLoS One 2009;
9. Poehling KA, et al. The underrecognized burden of influenza in
young children. N Engl J Med 2006; 355:31–40.
10. Rwanda Ministry of Health. National Institute of Statistics, ICF
Macro, 2009. Rwanda Interim Demographic and Health Survey 2007–
11. National Institute of Statistics of Rwanda. Rwanda general population
and housing census (RGPH). 2002. http://statistics.gov.rw/survey/
12. Pan American Health Organization/Centers for Disease Control and
Prevention. Generic protocol for influenza surveillance. 2006. http
13. World Health Organization. Integrated management of childhood
illness chart booklet. WHO, 2008. http://www.who.int/maternal_child_
of influenza A and B viruses. Influenza Other Respi Viruses 2007;
15. Cox NJ, Bai ZS, Kendal AP. Laboratory-based surveillance of influenza
A(H1N1) and A(H3N2) viruses in 1980–81: antigenic and genomic
analyses. Bull World Health Organ 1983; 61:143–52.
16. Simmerman JM, et al. Incidence, seasonality and mortality associated
with influenza pneumonia in Thailand: 2005–2008. PLoS One 2009;
17. Katz MA, et al. Who gets hospitalized for influenza pneumonia
in Thailand? Implications for vaccine policy. Vaccine 2007; 25:
18. Vongphrachanh P, et al. An early report from newly established labo-
ratory-based influenza surveillance in Lao PDR. Influenza Other Respi
Viruses 2010; 4:47–52.
19. Suntarattiwong P, et al. Influenza-associated hospitalization in urban
Thai children. Influenza Other Respi Viruses 2007; 1:177–82.
20. Chow A, et al. Influenza-associated deaths in tropical Singapore.
Emerg Infect Dis 2006; 12:114–21.
21. McAnerney J, Cohen C, Besselaar T, et al. Twenty-six years (1984–
2009) of influenza surveillance in South Africa. Options for the
Control of Influenza VII, 2010. Oral Abstract.
22. Katz M, Lebo E, Emukule G, et al. The epidemiology, seasonality and
burden of influenza and influenza-like illness in urban and rural
Kenya, 2007–2010. J Inf Dis. In press.
23. Chiu SS, Lau YL, Chan KH, et al. Influenza related hospitalization
among children in Hong Kong. N Engl J Med 2002; 347:26.
24. Laguna-Torres VA, et al. Influenza-like illness sentinel surveillance in
Peru. PLoS One 2009; 4:e6118.
Influenza Sentinel Surveillance in Rwanda • JID 2012:206 (Suppl 1) • S79
by guest on November 10, 2015