Table 1.

Community-based surveillance systems in alphabetical order of country

	Disease or events	Purpose	Design, staffing, investigation, and response	Evaluation results
Bangladesh
13 subcamps in Cox's Bazaar refugee complex (n=548 739);41 May–November, 2019	Acute flaccid paralysis, acute watery diarrhoea, acute jaundice syndrome, diphtheria, measles, meningitis, dengue, mortality	Early warning for outbreaks; monitor community mortality trends (notably neonatal deaths; camp setting)	CHVs did active surveillance, covering on average 36 households per day; no information on training and supervision; CBS had integrated alert and response team and medical response team to launch rapid response to CBS alerts; the system was integrated into early warning alerts and response system	Comprehensive evaluation; multiple rapid diagnostic tests and cholera alerts triggered a targeted cholera response in surrounding households; high PPV for acute flaccid paralysis, acute watery diarrhoea, and measles (74–100%) and low PPV for meningitis and diphtheria (42–50%); resource intensive, requiring 354 staff, including alert and medical response teams
Central African Republic
24 villages, Boda, Boganangone, Boganda, and Gadzi subprefectures (n=158 000);21 March–December, 2010	Child and crude mortality	Monitor health trends in a crisis (rural setting)	24 CHVs did active surveillance once per week for crude mortality and migration in or out of the setting; 24 sentinel sites were randomly selected	Evaluation of sensitivity and data validity; there was low attrition and CBS was acceptable to communities; capture–recapture analysis showed that mortality data had >90% sensitivity and specificity; population data were difficult to estimate because of high migration in or out of the setting; sentinel site CBS is feasible in a rural crisis setting
80 villages, Paoua (Ouham-Pendé) and Markounda (Ouham) subprefectures (n=222 000);33 2009–14	Malaria (suspect cases)	Monitor malaria trends in a crisis (rural setting; acute phase)	80 CHWs did case management of malaria among children <5 years and pregnant women in the community and reported surveillance data once per month	No formal evaluation; CHWs maintained malaria surveillance during and after a crisis where no other sources of surveillance existed (near 100% reporting rates); CHWs migrated with villages; the assumption of constant population size is most likely untrue; high costs of supervision and training can only be supported by a non-governmental organisation
Chad
Five refugee or internal displacement camps in eastern Chad (ie, Farchana, Breidjing, Ade, Gassire, and Kerfi; n=13 000–30 000);19 2004–08	Child and crude mortality	Monitor mortality trends in a crisis (camp setting; acute phase)	CHWs (n=unknown) did active surveillance for births, deaths, and migration, collated once per week; CHWs assigned to 500 households each	Evaluation of validity, simplicity, flexibility, and timeliness; CBS helped to detect a diarrhoea outbreak that led to improvements in water quality; lessons learned included the need for improved population estimates, standardised reporting, and procedures for improved data quality and dissemination, the importance of a simple and flexible model for data collection, and supervising CHWs; CBS implementation too slow for emergency phase (2–5 months to implement)
Am Timam town, eastern Salamat region (n=65 000);35 September, 2016, to April, 2017	Hepatitis E (acute jaundice syndrome)	Active hepatitis E surveillance during an epidemic (periurban setting)	160 CHVs visited households twice a week to do surveillance of acute jaundice syndrome and immediate referral for people at risk of clinical complications	No formal evaluation; CBS made difficult by high mobility of population (seminomadic herders); might have underestimated true number of cases
Côte d'Ivoire
Biankouma, Danané, and Zouan-Hounien districts, Tonkpi region (n=992 565);40 March–May, 2015	Ebola virus disease	Active Ebola virus disease surveillance during an epidemic (rural setting)	110 CHWs reported events suggestive of Ebola virus disease every day, including deaths, animal contacts, and illness or deaths among health workers and visitors; FrontlineSMS was used to transmit data immediately	No formal evaluation
Odienné, Touba, and Minignan districts, Kabadougou-Bafing-Folon region (n=501 328);24, 43 April, 2016, to December, 2017	Polio, cholera, measles, meningitis, yellow fever, and unusual health events	Early warning of outbreaks in a border area where Ebola virus disease was present	541 CHWs and a key informant network visited homes and were notified of signals; they used FrontlineSMS to immediately transmit signals; triage nurse received signal and carried out initial investigation	Evaluation of effectiveness; large-scale increases in reporting of suspect cases after implementation; low yield of nurse-verified suspect cases (11% [420 of 3734]) and confirmed cases (5% [23 of 420]); highest proportion of suspect cases for polio (49% [33 of 68]) and yellow fever (29% [166 of 568]); cholera produced no verified suspect cases but 1857 signals, producing a large burden on the system; large burden of human resources, supervision, and costs
Democratic Republic of the Congo
Rwandan refugee camps, Goma (n=90 000);22 April–May, 1992	Cholera	Active cholera surveillance during an epidemic in a crisis (camp setting)	CHWs (n=unknown) did active case finding of cholera cases (defined as people with sudden onset of watery diarrhoea resulting in dehydration)	No formal evaluation
39 health areas, Katanga Province;30 population unknown; April–May, 2011	Measles and measles-related mortality	Active measles surveillance during an epidemic (rural setting)	CHVs (n=unknown) interviewed community leaders about suspected measles-related deaths (within 30 days of onset of rash and not due to an obvious other cause, such as trauma) and collected information on measles deaths once per week	No formal evaluation; CBS reported higher numbers of measles-related deaths (n=376) vs facility surveillance (n=27); authors reported minimal time and resources were put into CBS in this context
34 health centre catchment areas, Fizi Health Zone, South Kivu Province (n=2926);26 October, 2011, to September, 2012	Crude mortality	Monitor mortality trends in a crisis (rural setting)	CHVs (n=unknown) did active surveillance once per month at preselected households on deaths (crude) and births in the past month while accounting for population changes; 34 sentinel sites were selected non-randomly from Fizi Health Zone and cluster sampling was used to select 15 households in each cluster	Evaluation assessed data validity, sensitivity, specificity, PPV, simplicity, and flexibility; by use of a mortality survey and resolution of differences in recorded births and deaths between surveillance and the survey, the evaluation reported improved sensitivity (87% vs 71%), specificity (>99% vs 99%), and PPV (>91% vs 28%) for surveillance over the survey; data collection was simple and flexible, as it was reduced to few data elements and visits could be delayed due to insecurity; surveillance needed maintenance and multiple layers of data checking
Ethiopia
Three South Sudanese refugee camps, Gambella (n=approximately 150 000);20 April, 2014, to January, 2015	Hepatitis E	Active hepatitis E surveillance during an epidemic in a crisis (camp setting)	CHWs (n=unknown) detected and referred acute jaundice syndrome cases to health facilities	No formal evaluation
Guinea
Three subprefecture of Guéckédou prefecture (Guéckédou city, Tékoulo, and Guendembou; n=43 000);37 2011–14	Child and crude mortality and causes (malaria and Ebola)	Monitor trends in malaria mortality (rural setting)	46 CHVs did passive surveillance of deaths and suspected causes (using a simple algorithm to attribute to malaria, Ebola, or other cause) in 46 sentinel sites; one CHW to every 12 500 people	Evaluation assessed validity of mortality data; CBS can capture information on mortality in areas where surveillance is weak or patients do not present to facilities (eg, during Ebola virus disease outbreak); establishing causes of death is challenging; CBS of mortality is useful for outbreak detection if timeliness of data collection and reporting facilitate real-time data analysis
Haiti
Internal displacement camps in Delmas (n=43 930–54 890) and Champs de Mars (n=approximately 23 500), Port au Prince;31 epidemic weeks 12–32, 2010	Child and crude mortality	Monitor health trends during a crisis (urban setting; acute phase)	CHVs (n=unknown) assessed child and crude mortality, births, and migration in or out of the setting in households in a week	No formal evaluation; CBS recorded lower than anticipated mortality rates (with the assumption that mortality was very high following the earthquake) and shifted focus toward other needs; communities were frustrated that visits once per week did not yield immediate benefits; threats toward home visitors caused closure of CBS; high migration in or out of the setting; denominators were difficult to establish
Western area;39 population unknown; 2014–15	Cholera	Active cholera surveillance during an epidemic (rural setting)	239 Red Cross CHVs did active surveillance for cholera (acute watery diarrhoea) every day by use of Magpi or SMS on a mobile phone; suspect cases were referred immediately and logged	Evaluation of sensitivity and specificity; CBS appeared sensitive (ie, it compared well over time with facility data) and detected cholera in geographical areas not covered well by facility surveillance; CBS had high sensitivity with high numbers of false positives; CBS should be used with facility surveillance (low sensitivity, high specific) to show the overall perspective
India
37 Tibetan refugee settlements and Delhi (South, Doon Valley, Central, Ladakh, Himachal, and North East; n=54 537);18 1994–96	Child and crude mortality, causes of death	Monitor health trends during a crisis (urban setting; rural setting, long-term)	CHWs (n=unknown) collected monthly data for mortality and cause of death from households of refugees in India; CHWs sent data by paper to a central office in Dharamsala	Evaluation of various aspects (largely qualitative); CBS provided a profile of cause of death but unrealistically low death rates; CHW visits once per month were not always possible as CHWs were overworked; therefore, collection of monthly morbidity data was not possible, as was intended
Liberia
43 districts in Grand Cape Mount, Gbarpolu, Lofa, Bong, Nimba, Grand Gedeh, River Gee, and Maryland (n=2 million);44 February–October, 2016	Acute flaccid paralysis, measles, rabies, acute bloody diarrhoea, meningitis, viral haemorrhagic fever, acute watery diarrhoea, neonatal tetanus, death (neonatal and maternal), and unexplained cluster (disease or death)	Early warning of outbreaks in a border area where Ebola virus disease is present	2972 surveillance volunteers (1:100 population) were notified of potential cases or events that would be referred to the health facility; integrated disease surveillance and response guidelines were followed to investigate and notify the suspect case	Evaluation of effectiveness; 24% (885 of 3746) of alerts were suspect cases according to the community case definition; 32% of non-Ebola virus disease cases of epidemic disease were signalled by CBS; PPV was highest for neonatal deaths (70% [50 of 71]), maternal deaths (82% [27 of 33]), and unexplained deaths (43% [50 of 117]) and low for viral haemorrhagic fever (8% [40 of 505]), meningitis (10% [13 of 125]), and acute flaccid paralysis (14% [10 of 71]); coverage among CBS and health facilities was highest for acute watery diarrhoea, neonatal tetanus, acute flaccid paralysis, and unexplained death; CBS was financially unsustainable (19% of government health expenditure)
Malawi
11 Mozambican refugee camps (n=269 859);45 1987–89	Child and crude mortality, causes of death	Monitor mortality and cause of death trends in a crisis (camp setting; acute phase); relied on health posts for morbidity surveillance	CHWs in 11 sections counted deaths each week in sections of up to 2000 persons; causes of death were investigated by use of health-facility registries and interviews with family members, where case definitions were used to assign a cause of death; the goal was to identify diseases related to the highest cause of mortality in the refugee camp	Qualitative evaluation of simplicity, flexibility, and adaptability; a basic evaluation was done of both CBS and health facility-based surveillance (morbidity); the entire surveillance system was assessed to be simple and acceptable on the basis of the few steps and sources (ie, CHWs, health posts) used, use of the data collection infrastructure of the ministry of health, and rapid weekly reporting
Nepal
Six Bhutanese refugee camps, Teraj region (n=73 500);38 July, 1992, to January, 1993	Child and crude mortality, causes of death	Monitor mortality and cause of death trends during a crisis (camp setting; acute phase); relied on health posts for morbidity surveillance	One CHV specialising in mortality per camp did active surveillance every day of deaths and causes of deaths (by use of a simple algorithm for malaria, measles, acute respiratory illness, diarrhoea, death in childbirth, injury, or other or unknown); the data collection approach was not described; on the basis of mortality data, CHWs later focused on active case finding of acute respiratory illness and diarrhoea to provide immediate treatment and referral	No formal evaluation; CBS led to public health actions and rapid detection and response to cholera, Shigella dysenteriae, and meningoencephalitis; provision of burial expenses encouraged reporting from the community
Sierra Leone
Bo, Kailahun, Kambia, Kenema, Kono, Moyamba, Pujehun, and Tonkolili Districts (nine of 14 districts; n=3 981 665);25, 32, 36 Feb 27 to Sept 30, 2015	Ebola virus disease (events)	Active Ebola virus disease surveillance during an epidemic (mostly rural setting)	CBS based on events was rapidly scaled at a national level by a network of non-governmental organisations during a health emergency; 7416 CHWs and CHVs and 137 supervisors did active surveillance every day of six events suggestive of Ebola virus transmission, including community deaths, by use of simple phones	Evaluation of sensitivity and timeliness; of the 12 126 reports, 287 reports (2%) met the suspected case definition, 16 reports were confirmed positive (detecting 30% [16 of 53] of Ebola virus disease cases identified during the study period); consistent surveillance data was produced from districts reporting few or no cases; CBS detection was faster than facility detection and new chains of transmission were found; CBS cost US$1·3 million at start-up with approximately $129 000 monthly costs; to sustain performance, event definitions should be refined and integrated into the surveillance system
South Sudan
National scale, population not provided;27 22 000 villages; 2006–12	Guinea worm	Eradication (case containment) in a crisis (mostly rural setting)	CHWs (n=unknown) did active surveillance every day at village level and implemented case containment	No formal evaluation
Yusuf Batil, Jamam, Gendrassa, and Doro refugee camps, Maban County, Upper Nile (n=110 000);23 July, 2012, to January, 2013 (7 months)	Hepatitis E (acute jaundice syndrome)	Active hepatitis E surveillance during an epidemic in a crisis (camp setting)	CHWs (n=unknown) detected and referred cases of acute jaundice syndrome to health facilities	No formal evaluation
34 counties, Jonglei, eastern Equatoria, Unity State, and Upper Nile states (n=3·7 million);29 October, 2015, to September, 2017	Acute flaccid paralysis	Eradication (case containment) in a crisis (mostly rural setting)	3228 community key informants passively reported cases to 230 payam assistants; key informants given training in case identification; payam assistants visited key informants once per week; suspect case triggered WHO field team to do specimen collection	Evaluated functionality, sensitivity, and effectiveness; counties that regularly reported increased from 64% (16 of 25) to 92% (23 of 25); increase in acute flaccid paralysis reporting from CBS (12·5 cases per month) vs health facilities (2·2 cases per month); CBS incurred a low stool sampling rate (51–63%) compared with 100% in health facilities due to difficulties tracing suspect cases and a scarcity of specimen sampling materials and transport; higher proportions of cases reported within 24 h and investigated within 48 h with CBS than with health facilities, but the denominators used were unclear
Tanzania
Four refugee camps in Kibondo District (n=279 455);28 March, 2000, to May, 2001 (1 year, 3 months)	Measles	Active measles surveillance during an epidemic in a crisis (camp setting)	CHWs (n=unknown) did active case finding in households on five diseases of epidemic potential, including measles; two surveillance focal people reconciled CHW-identified suspect cases and health facility registers to identify missed measles cases	No formal evaluation; authors state that “this epidemic was remarkable for its scarcity of reported measles-associated deaths despite rigorous community-based surveillance”28
Uganda
East Moyo South Sudanese camp (n=30 000);34 February, 1994, to March, 1995 (1 year)	Meningococcal meningitis	Active meningitis surveillance during an epidemic in a crisis (camp setting)	CHWs (n=unknown) did active surveillance and referral for meningitis cases	No formal evaluation
Yemen
Al Hawak and Al Mena districts, Hodeidah City (n=400 000);46 October, 2016, to February, 2017 (4 months)	Cholera	Active cholera surveillance during an epidemic in a crisis (urban setting)	Community teams monitored households in neighbourhoods where cholera suspect cases came from to do active case finding for cholera (acute watery diarrhoea); mild cases were given oral rehydration solution and complicated cases were referred	No formal evaluation

CBS=community-based surveillance. CHV=community health volunteer. CHW=community health worker. PPV=positive predictive value.