Risk Factors for Border Malaria in a Malaria Elimination Setting: A Retrospective Case-Control Study in Yunnan, China
Am. J. Trop. Med. Hyg.
Risk Factors for Border Malaria in a Malaria Elimination Setting: A Retrospective Case-Control Study in Yunnan, China
Jian-Wei Xu 0 1
Hui Liu 0 1
Yu Zhang 0 1
Xiang-Rui Guo 0 1
Jia-Zhi Wang 0 1
0 Diseases , 6 Xiyuan Road, Puer, 665000, Yunnan , China
1 Yunnan Institute of Parasitic Diseases, Yunnan Provincial Center of Malaria Research, Yunnan Provincial Collaborative Innovation Center for Public Health and Disease Prevention and Control, Yunnan Provincial Key Laboratory of Vector-Borne Diseases Control and Research , Puer 665000 , China; The Fourth Hospital of Baotou Municipality, Inner Mongolia , Baotou , China; Yingjiang County Center for Disease Control and Prevention, Yingjiang China; Tengchong County Center for Disease Control and Prevention , Tengchong , China
A retrospective case-control study was conducted to identify risk factors for border malaria in a malaria elimination setting of Yunnan Province, China. The study comprised 214 cases and 428 controls. The controls were individually matched to the cases on the basis of residence, age, and gender. In addition, statistical associations are based on matched analyses. The frequencies of imported, male, adult, and vivax malaria cases were respectively 201 (93.9%), 194 (90.7%), 210 (98.1%), and 176 (82.2%). Overnight stay in Myanmar within the prior month was independently associated with malaria infection (odds ratio [OR] 159.5, 95% confidence interval [CI] 75.1-338.9). In particular, stays in lowland and foothill (OR 5.5, 95% CI 2.5-11.8) or mid-hill (OR 42.8, 95% CI 5.1-319.8) areas, or near streamlets (OR 15.3, 95% CI 4.3-55.2) or paddy field or pools (OR10.1, 95% CI 4.4-55.8) were found to be independently associated with malaria. Neither forest exposure nor use of vector control measures was associated with malaria. In conclusion, travel to lowland and foothill or mid-hill hyperendemic areas, especially along the waterside in Myanmar, was found to be the highest risk factor for malaria. In considering the limitations of the study, further investigations are needed to identify the major determinants of malaria risk and develop new strategies for malaria elimination on China-Myanmar border.
Remarkable progress has been achieved in reducing
malaria burden globally over the past 10 years.1,2 Nowadays,
there are 99 endemic countries; of these, 65 are controlling
malaria and 34 are pursuing elimination. The 34 eliminating
countries have declared a national policy for malaria
elimination or are pursuing spatially progressive elimination within
their borders.2–4 Countries in the Asia Pacific region are mak
ing particularly strong progress toward eliminating malaria.5
As one of countries with a great reduction in malaria incidence,
China aims to eliminate malaria by 2020.6
In low transmission settings, malaria is increasingly imported,
caused by Plasmodium vivax, and clustered demographically in
adult men with shared epidemiological risk factors. Current
malaria control interventions and strategies are not likely to
be optimal, and novel strategies targeting high-risk groups are
urgently needed. In high endemic, control settings, risk factors
can be established through nationally representative
crosssectional surveys, such as malaria indicator surveys. However,
in areas where transmission is very low and malaria infections
are rare, these surveys are unlikely to adequately detect cases
or identify risk factors. Furthermore, such large-scale surveys
require substantial financial and operational resources.7,8
Case-control analysis is a well-established method to study
rare diseases and identify associated demographic,
behavioral, and clinical risk factors, being particularly appropriate
for rare diseases. As yet, this epidemiological tool has not to
be extensively applied to malaria, which has traditionally
been studied in high-endemic settings.7,9–11
In China, malaria has been eliminated from large parts of
the country. The remaining at-risk areas can be characterized
as low endemic and they are clustered in parts of central
China, southern China, and Tibet. The southern province of
Yunnan will be particularly challenging for elimination as it
shares a 4,061 km border with malarious countries including
Myanmar, Laos, and Vietnam.12,13 The epidemiology of malaria
in this area is presumed to be similar to other parts of
Southeast Asia, where reported malaria incidence and parasite
species have been identified. Malaria tends to concentrate among
migrant workers on border regions and in forest areas. For
example, men who travel into the deep forest for work become
infected with bites from Anopheles dirus (the primary vector in
the deep forest of Southeast Asia) and return to their home
villages on the forest fringe where they become reservoirs for
transmission of the parasites by Anopheles minimus (the
primary vector on the forest fringe of Southeast Asia) to their
young children.14 However, these risk factors have not been
evaluated in a controlled study. To improve understanding of
risk factors of border malaria infection in Yunnan and to
support the design and implementation of targeted interventions,
we conducted a retrospective case-control study.
Study site. The study was conducted in Tengchong and
Yingjiang, two border counties at the Yunnan-Myanmar
border (Figure 1). There are 39 health facilities that are required
to report all suspected and laboratory-confirmed malaria cases
through the China Information System for Disease Control
and Prevention (CISDCP), the national internet-based
disease reporting system. Basic demographic data and contact
information are included in the reports. The epidemiology of
malaria in this area has been described.14–16 Briefly, these
sites were selected because they are formerly hyper-endemic
areas and the current hot spots of border malaria in China,
as a result there were enough cases for the study.
Study design and data collection. The study was a
retrospective case-control design and conducted from April to
October 2012. The laboratory-confirmed cases were reported
from all health facilities in Tengchong and Yingjiang from
September 2010 to December 2011. Li and Fan15 reported that
97.5% (2,556 of 3,646) of malaria patients were imported from
Myanmar and most of them were male adults in Tengchong.
To exclude the confounding effect of age and sex, controls were
matched by age and sex. The goal of malaria elimination was
to achieve zero incidence of locally contracted cases. It was
important to identify the risk factors of local malaria
infection,17,18 therefore community of residence was also selected
as one of matching criteria. Finally, the controls were matched
two to one with cases on the basis of residence, age (±2 years),
First, blood smears from malaria cases reported from
September 2010 to December 2011 were reread by an expert
microscopist for secondary confirmation. Using contact
information available in the case reporting system, the confirmed
cases were then visited at their homes by study staff and
approached for orally informed consent. A paper survey with
35 questions was administered in Mandarin to assess
demographics and potential risk factors including housing condition,
local ecology, socio-economic status, behavior, occupation,
activities, travel, malaria awareness and knowledge, and use of
malaria prevention measures. Information on activities,
housing, use of personal protective measures, and local ecology was
obtained on both the subject’s home in Yunnan and at the
locations where they had stayed 1 month before the date of
malaria attack of case patients.
Controls were then recruited from adjoining households of
cases.13 Subjects matching the case by age (±2 years) and
gender were invited to participate in the study. A finger prick
was performed in consenting subjects to generate a blood
slide for microscopy. Only subjects that were malaria free by
microscopy were recruited as controls. The same
questionnaire was then conducted with control subjects to collect data
on potential risk factors.
Ethical approval was obtained from the Ethics Committee
of Yunnan Institute of Parasitic Diseases, China. The Ethics
Committee approved a verbal consent procedure as sufficient
because the study was interview-based, and screening for
malaria by microscopy is one of the malaria control
components. Questionnaires were conducted by a team of seven
Defining wealth tertiles. Details were recorded on principal
components of housing characteristics (type of wall and roof),
household appliances, and transport tools; and then were used
to construct a wealth index. The wealth tertile was defined as 1)
lowest tertile: soil and wood walls and terracotta roofs, only
television sets or without any valuable electric items, only bikes
or without any transport tools; 2) middle tertile: brick walls and
terracotta roofs, television sets, motorcycles; 3) highest tertile:
brick or reinforced concrete walls and roofs, television sets, and
refrigerators, tractors or cars or trucks.19–22
Statistical analysis. Data were double entered and cleaned
in EpiData 3.1 (EpiData Association, Odense, Denmark),
and analyzed in Epi Info 2000 (Centers for Disease Control
and Prevention, Atlanta, GA). The statistical associations are
based on matched analyses. For the case and the control group,
the frequency and proportions of each predictor variable were
calculated and compared using a c2 test. Variables likely
associated with the outcome (P < 0.25) were selected as
candidate variables for matched univariate and multivariate logistic
regression.23 Missing data were excluded from these analyses.
There were 214 malaria cases and 428 matched controls
recruited to participate in the study. Of the 214 confirmed
For cases and controls, N = 214 and N = 428, respectively, unless otherwise indicated.
ITNs = insecticide-treated nets; LLINs = long-lasting insecticide nets.
N = 213
N = 426
malaria case patients, 176 (82.2%) were vivax malaria and
38 (17.8%) were falciparum malaria. There were no mixed
infections. Based on the CISDCP, 201 (93.9%) cases were
imported from other countries, 12 (8.0%) locally acquired,
Characteristics of cases and controls, Yunnan, China, 2012
Activities, housing conditions, and local ecology in Myanmar, data
from cases and controls that stayed overnight in Myanmar before
the malaria attack in the case
Adjusted OR (95% CI)
nets including ITNs or LLINs, using measures against
mosquito bites including sleeping under a net did not differ
significantly between case patients and matched controls. In
Myanmar, activities and housing conditions of Chinese
migrants, and vegetation nearby the staying site did not differ
significantly between case patients and matched controls too.
The study revealed that most case-patients were imported
from the neighboring districts of Myanmar (93.9%), male
(90.7%) and 16–50 years of age (90.7%), and vivax malaria
(82.2%). Independent risk factors associated with malaria
infection were overnight in the lowland, foothill, and half-hill
areas of Myanmar. This coincides with comments in the
literature: malaria cases are increasingly male, adult, clustered
geographically, imported among migrant and other
hard-toreach groups, and caused by P. vivax in eliminating settings.7
This is also similar in Cambodia, the males and adults had an
increased risk of Plasmodium spp. infection.24 However, local
malaria infection still exists, especially for patients < 15 years
of age, and the youngest one only 3 years of age, who had
never been in the neighboring districts of Myanmar, showed
that local malaria transmission has not been interrupted
In the Greater Mekong Subregion (GMS), forests are
commonly considered as a major determinant of malaria
risk.14,25–29 In Vietnam a large proportion of all malaria cases
and deaths occurred in the central mountainous and forested
areas.28 In Myanmar, about 60% of the total malaria cases
occurred in forest or forest fringe areas; forest workers
(loggers, gem miners, etc.) were at high risk of malaria.29 The
explanation is that forests and forest fringes are mountainous
and hilly areas, difficult of access, sparsely inhabited by the
ethnic minorities, and limited economic development and
infrastructure. Men who travel into the deep forest for work
become infected with bites from An. dirus and return to their
home villages on the forest fringe where they become
reservoirs for transmission of the parasites by An. minimus to their
young children.14,29 However, the study found that malaria
was associated with Chinese migrants staying in the lowland
and foothill or mid-hill areas, especially along the waterside.
The study results showed that the border crossers of the
control group reported staying in forests more often than that of
the case group, and the percentage of loggers among controls
(86.5%) was higher than that among case patients (68.2%).
One of the reasons might be that the study site was next to
Kachin State of Myanmar, where malaria burden is
particularly high and malaria outbreaks occur frequently, the
malaria-related mortality rate was as high as 7.8 deaths per
1,000 people in 200529–31; Liu and others32 reported a mean
parasite rate 16.3% (95% confidence interval [CI]: 12.9–
20.1%) among residents of Kachin State in 2010. In most parts
of Southeast Asia, the year-round high rainfall and
temperatures, and malaria vectors lead to persistent and intense
malaria transmission.25 However, in northern Myanmar the
mountainous areas border with China, the high altitude leads
to its difference from most parts of Southeast Asia; the hilly
forested areas have low temperatures, fewer mosquito
breeding sites, less malaria vector, and less parasite reservoirs
caused by sparse population.
In theory, people who have a better knowledge of malaria
transmission and methods of prevention, awareness and using
measures for malaria prevention, should have reduced malaria
infection. The reported investigation in the literature
documented that use of bed nets provide useful protection against
malaria infection.33,34 However, the results of the study did
not show that the knowledge of malaria transmission and
prevention was associated with the reduced risk of malaria
infection; furthermore, it was not documented that the
ownership and the use of bed nets or repellents was related to the
reduced risk of malaria infection. On the contrary, the
percentage of case patients (54.0%) who knew chemoprophylaxis
was higher than that of controls (46.0%); and the percentage
of case patients (43.1%) who used measures of bed nets or
repellents against mosquito bites in Myanmar was higher than
that of controls (13.9%). When malaria is rare to see in the
eliminating setting of China, the better knowledge and
behavior might be attributable to malaria infection. People who had
malaria attacks might have more experience with malaria and
they might actively seek knowledge and take action for
malaria prevention and control.
Overmatching may be one of the main limitations in this
study. In the eliminating setting, malaria cases cluster on
border areas, are increasingly male, adult, imported, among
migrant and other hard-to-reach groups.7 Li and Fan reported
that 97.5% (2,556 of 3,646) of malaria patients were imported
from Myanmar and most of them were male adults in the
study site.15 To exclude the confounding effect of age and
sex, controls were matched by age and sex. The goal of
malaria elimination was to interrupt local mosquito-borne
malaria transmission in a defined geographical area, i.e., zero
incidence of locally contracted cases; but imported cases will
continue to occur, and continued intervention measures are
required to prevent reintroduction.17,18 Therefore,
community of residence was also selected as one of the matching
criteria to identify the risk factors of local malaria infection.
One of the purposes of using age, gender, and community of
residence as matching criteria was to explore methods for
reduction of the confounding effect from age and gender,
and to identify risk factors of local malaria infection.
However, we could not ensure whether this design might lead to
overmatching and limiting assessment on these as
independent risk factors. The study results did not identify any risk
factors of local malaria infection because of only 12 (8.0%)
locally acquired cases, therefore the purpose has not been
achieved by using residence as one of the matching criteria.
This failure showed that use of residence as a matching
criterion may be inappropriate in a malaria elimination setting.
Despite among 214 case-patients, 194 (90.7%) were male and
210 (98.1%) adult (³ 16 years), this study could not identify
whether they are an independent risk or confounding factors
because of using age and gender as the matching criteria.
The study also had several other limitations. The first was
the potential bias in capture of case patients. The cases were
from records of public health facilities. The malaria patients
who sought treatment from the private sector were excluded
from this study. Some Chinese migrants carry anti-malarial
drugs with them for self-saving during staying in Myanmar;
however, the case-patients were from parasite-based diagnosis
in China. In cases where the self-medication cured their
malaria infection, they might not seek diagnosis and
treatment from the public health service in China, which led to
exclusion of them from the study too. These might lead to
selection bias. The second is that some people declined
to answer certain questions that they thought of as sensitive,
and might cause responding bias. The third is that the
financial support for this study only permitted an operation period
of < 1 year, unable to recruit enough case patients in such a
short period, therefore a design of retrospective case-control
study was used. A weakness of this design was that the data
were collected after the fact, and there are major challenges
with recall bias, particularly in the control subjects who would
have a hard time remembering the time before a date that is
not relevant to them.
In conclusion, imported malaria is particularly prominent,
and most malaria patients are male, adult, and infection of
P. vivax in the malaria elimination setting of Yunnan, China.
The independent risk factors for malaria infection were
staying in the lowland, foothill, and half hill, and availability
of streamlets and paddy fields or pools nearby their staying
sites of Myanmar. However, the study identified neither forest
exposure nor use of vector control measures to be
independently associated with malaria infection. In considering the
limitations of the study, further research is needed to identify
the major determinants of malaria risk on the China-Myanmar
border, and to develop new strategies for malaria elimination;
for example, how to promptly find and treat imported malaria
infection in China, reduce malaria transmission and incidence
in Myanmar through effective cooperation mechanism between
China and Myanmar, and promote vigilance and use of
personal protection among Chinese border crossers through
effective behavior change in communication.
Received May 23, 2014. Accepted for publication December 7, 2014.
Published online January 19, 2015.
Acknowledgments: We thank Michelle S. Hsiang and Hugh Sturrock
from Global Health Group, University of California San Francisco
for their comments and copyediting.
Financial support: This study was funded by a grant to China National
Strategy Application of the Global Fund to fight AIDS, Tuberculosis
and Malaria (GFATM/CHN-S10-G13-M).
1. World Health Organization , 2011 . World Malaria Report. Geneva: WHO.
2. Bousema T , Griffin JT , Sauerwein RW , Smith DL , Churcher TS , 2012 . Hitting hotspots: spatial targeting of malaria for control and elimination . PLoS Med 9 : e1001165 .
3. Feachem RG , Phillips AA , Hwang J , 2010 . Shrinking the malaria map: progress and prospects . Lancet 376 : 1566 - 1578 .
4. Tatem AJ , Smith DL , Gething PW , Kabaria CW , Snow RW , Hay SI , 2010 . Ranking of elimination feasibility between malariaendemic countries . Lancet.
5. Gosling RD , Whittaker M , Gueye CS , Fullman N , Baquilod M , Kusriastuti R , Feachem RG , 2012 . Malaria elimination gaining ground in the Asia Pacific . Malar J 11 : 346 .
6. Ministry of Health (MOH), People's Republic of China , 2010 . Malaria Elimination Action Plan 2010 - 2020 . (in Chinese). Beijing: The Ministry , 1 - 2 .
7. Cotter C , Sturrock HJ , Hsiang MS , Liu J , Phillips AA , Hwang J , Gueye CS , Fullman N , Gosling RD , Feachem RG , 2013 . The changing epidemiology of malaria elimination: new strategies for new challenges . Lancet 382 : 900 - 916 .
8. Hsiang MS , Hwang J , Kunene S , 2012 . Surveillance for malaria elimination in Swaziland: a national cross-sectional study using pooled PCR and serology . PLoS ONE 7 : e29550 .
9. Cousens SN , Mertens TE , Kirkwood BR , Smith PG , Feachem RG , 1995 . Case-Control Studies of Common Childhood Diseases: The Example of Diarrhea . London: Macmillan Education.
10. Hayes RJ , Marsh K , Snow RW , 1992 . Case-control studies of severe malaria . Trop Med Hyg 95 : 157 - 166 .
11. Chalwe V , Geertruyden VJ , Mukwamataba D , Menten J , Kamalamba J , Mulenga M , Alessandro U , 2009 . Increased risk for severe malaria in HIV-1infected adults, Zambia . Emerg Infect Dis 15 : 749 - 755 .
12. Xu JW , Liu H , 1997 . Border malaria in Yunnan, China. Southeast Asian J Trop Med Public Health 28 : 456 - 459 .
13. Xu JW , Liu H , 2012 . The challenges of malaria elimination in Yunnan Province , People's Republic of China. Southeast Asian J Trop Med Public Health 43 : 819 - 824 .
14. Roll Back Malaria Mekong, 2003 . Mekong Malaria II: update of malaria, multi-resistance and economic development in the Mekong Region of Southeast Asia . Southeast Asian J Trop Med Public Health 34 (Suppl): 68 - 70 .
15. Li PS , Fan DH , 2005 . Analysis of 3646 malaria cases in Tengchong County of Yunnan Province 1999-2003 (in Chinese) . Parasitoses Infectious Dis 3 : 34 - 35 .
16. Chen GW , Wei C , Li HX , Yang LX , Huang Q , 2011 . Malaria epidemic trend and characteristics at monitoring sites in Yunnan Province in 2008 (in Chinese) . Chin J Parasitol Parasit Dis 29 : 54 - 57 .
17. World Health Organization , 2008 . Global Malaria Control and Elimination: Report of a Technical Review. Geneva: WHO.
18. World Health Organization , 2007 . A Field Manual for Low and Moderate Endemic Countries . Geneva: The World Organization.
19. Vyas S , Kumaranayake L , 2006 . Constructing socio-economic status indices: how to use principal components analysis . Health Policy Plan 21 : 459 - 468 .
20. Filmer D , Pritchett LH , 2001 . Estimating wealth effects without expenditure data-or tears: an application to educational enrolments in states of India . Demography 38 : 115 - 132 .
21. Smith L , Bruce J , Gueye L , Helou A , Diallo R , Gueye B , Jones C , Webster J , 2010 . From fever to anti-malarial: the treatmentseeking process in rural Senegal . Malar J 9 : 333 .
22. Gikandi P , Noor A , Gitonga C , Ajanga A , Snow R , 2008 . Access and barriers to measures targeted to prevent malaria in pregnancy in rural Kenya . Trop Med Int Health 13 : 208 - 217 .
23. Ward KA , Spokes PJ , McAnulty JM , 2011 . Case-control study of risk factors hospitalization caused by pandemic (H1N1) 2009 . Emerg Infect Dis 17 : 1409 - 1416 .
24. Incardona S , Vong S , Chiv L , Lim P , Nhem S , Sem R , Khim N , Doung S , Puijalon OM , Fandeur T , 2007 . Large-scale malaria survey in Cambodia: novel insights on species distribution and risk factors . Malar J 6 : 37 .
25. Prothero RM , 1999 . Malaria, forests and people in Southeast Asia . Singap J Trop Geogr 20 : 76 - 85 .
26. Guerra CA , Snow RW , Hay SR , 2006 . A global assessment of closed forests, deforestation and malaria risk . Ann Trop Med Parasitol 100 : 189 - 204 .
27. Trung HD , Van Bortel W , Sochantha T , Keokenchanh K , Quang NT , Cong LD , Coosemans M , 2004 . Malaria transmission and major malaria vectors in different geographical areas of Southeast Asia . Trop Med Int Health 9 : 230 - 237 .
28. Erhart A , Thang ND , Ky PV , Tinh TT , Overmeir CV , Speybroeck N , Obsomer V , Le XH , Le KT , Coosemans M , Dalessandro U , 2005 . Epidemiology of forest malaria in central Vietnam: a large scale cross-sectional survey . Malar J 4 : 58 .
29. Delacollette C , Souza C , Christophe E , Thimasarn K , Abdur R , Bell D , Dai TC , Gopinath D , Lu SH , Mendoza R , Ortega L , Rastogi R , Tantinimitkul C , Ehrenberg J , 2009 . Malaria trends and challenges in the greater Mekong Subregion . Southeast Asian J Trop Med Public Health 40 : 674 - 691 .
30. Cui L , Yan G , Sattabongkot J , Cao Y , Chen B , Chen X , Fan Q , Fang Q , Jongwutiwes S , Parker D , Sirichaisinthop J , Kyaw MP , Su XZ , Yang H , Yang Z , Wang B , Xu J , Zheng B , Zhong D , Zhou G , 2012 . Malaria in the Greater Mekong Subregion: heterogeneity and complexity . Acta Trop 121 : 227 - 239 .
31. WHO, 2008 . Malaria in the Greater Mekong Subregion: Regional and Country Profiles . Available at: http://www.whothailand.org/ EN/Section3/section. Accessed August 17 , 2014 .
32. Liu H , LI X , Chen H , Deng C , Long K , Kong L , 2011 . A crosssectional malaria survey in Mangwen District , Kachin State , Myanmar (in Chinese) . J Pathogen Bio 6 : 55 - 59 .
33. Xu JW , Wang LB , Liu XZ , Long SW , Yang GC , Yang XW , 2002 . Logistic analysis for the bed net role in malaria prevention . China Trop Med 2 : 90 - 93 .
34. Bermejo A , Veeken H , 1992 . Insecticide-impregnated bed nets for malaria control: a review of the field trials . Bull World Health Organ 70 : 293 - 296 .