Table 1.

Overview on manuscripts on NCD surveillance systems selected for the literature review

Authors & year	Country (region)	Surveillance approach	Data source/ reporting unit	Diseases under surveillance	Time	Lessons learned
Birtwhistle 2009 [17]	Canada	Longitudinal passive assisted sentinel surveillance system of NCDs (Pan-Canadian Primary Care Sentinel Surveillance Network, ongoing)	General practitioners (7 regional networks with ten practices in each network)	Hypertension, diabetes, depression, chronic obstructive lung disease, osteoarthritis	2008 (7 months) (first phase)	Primary care sentinel surveillance for NCDs is possible; major challenges are inclusion of risk factors and social variables, estimating practice denominators and ensuring representativeness of sentinel sites.
Bollag 2009 (BAG 2014) [18]	Switzerland	Longitudinal passive assisted sentinel surveillance system (Swiss Sentinel Surveillance Network, ongoing)	General practitioners, internists and paediatricians (total: 150–250 GPs)	Asthma, different communicable diseases	1989 – 2005	Sentinel surveillance on primary care level is a valid research instrument to analyse asthma incidence and seasonality. Denominator problems occurred since age and sex were only recorded for asthma cases, not all consultations.
Boydell et al. 1995 [19]	Northern Ireland	Cross-sectional pilot study on an active sentinel surveillance system (General Practice Data Retrieval Project)	General practitioners (n = 81) in 23 general practices (study population: 132,975)	33 chronic and acute conditions (results presented in paper: diabetes, myocardial infarction and depression)	1992–93	The accuracy of the diagnosis varied; validity of data needs to be explored in relation to the purpose for which it is to be used.
Deprez et al. 2002 [20]	USA (Maine)	Pilot study for a sentinel surveillance system using hospital data (passive), population-based phone survey and physician survey (active)	Secondary data: hospital admissions, emergency department/hospital outpatient data, physician survey (n = 59), population phone survey (n = 627)	Asthma	1994/95 (secondary data), 1997 (both surveys)	Data were useful to estimate the prevalence and to identify high risk groups; survey data provided otherwise unobtainable data on asthma symptoms; methods were not useful to identify environmental risks or the severity of asthma. The physician survey yielded useful information about diagnostic and treatment practices.
Fleming et al. 2003 [28]	Europe	Cross sectional survey, questionnaire based evaluation of sentinel surveillance systems (Health Monitoring in Sentinel Practice Networks Project)	Primary health care sector	33 sentinel practice networks, mainly on influenza, some also on diabetes	12/1998 – 12/2000	The primary care sector is an appropriate source for diabetes surveillance; if based on EMRs the costs of the system are very low; diagnostic validity of data has been demonstrated.
Klompas et al. 2012 [21]	USA (Massachusetts and Ohio)	Passive assisted sentinel surveillance system using the Electronic Medical Record Support for Public Health (ESP) surveillance platform	Primary health care sector (2 mixed provider groups: a multi provider multi-speciality ambulatory care provider group and a mixed inpatient and ambulatory provider group)	Diabetes, influenza, notifiable diseases	06/2006 – 07/2011	EMR based surveillance can provide timely and rich primary care data to public health departments on broad population and wide sets of health indicators; challenges include availability of sufficient electronic data, inclusion of contextual data, initial installation and activation of EMR based systems (financing) and electronic infrastructure to receive EMR-based reports.
Namusisi et al. 2011 [22]	Uganda (Mbarara district)	Pilot study on a passive assisted sentinel surveillance of NCDs	Regional referral hospital (n = 1) (1383 patient records)	Diabetes	01/2005 –04/ 2010	Use of hospital data is a valuable first step in setting up NCD surveillance systems, risk factor data are important for disease prevention and intervention. Incompleteness of records was a major limitation in the study.
Saran et al. 2010 [26]	USA	Pilot study for a passive national surveillance system	Various secondary data sets	Chronic kidney diseases (CKD)	10/2006 – 09/2008	Six broad themes, several measures for CKD and several data sources were identified for a pilot phase; active surveillance methods might be integrated in the future. Identification and acquisition of data sets and integration with other NCD surveillance systems were identified as some of the challenges.
Szeles et al. 2005 [23]	Hungary (4 counties)	Cross-sectional pilot study on a passive assisted sentinel surveillance system	General practitioners (n = 73) in four counties (Cohort size: 138,088)	Cardiovascular diseases, diabetes, liver cirrhosis, 4 malignant diseases	1998	Sentinel stations at primary care level are feasible and sustainable, data provide important information for health policy and health service planning, regular contact to reporting units is important.
Trepka et al. 2009 [24]	USA (Miami-Dade County)	Longitudinal pilot study for a passive assisted sentinel surveillance system (Miami Asthma Incidence Surveillance System)	Outpatient paediatric, allergy and pulmonary clinics (n = 18), emergency departments of hospitals (n = 3), standardized interviews with patients (n = 669)	Incident asthma	07/2002–06/2006	The pilot was useful in evaluating the case definition, in describing participants’ characteristics and health care use patterns; without mandatory laws, the system is not feasible.
Westert et al. 2005 [25]	Netherlands	Cross sectional surveillance study based on sentinel sites, health interviews with patients and census data (Dutch National Survey of General Practice 2)	General practitioners (n = 195) in 104 general practices (cohort size: 385,461), health interviews with Dutch speaking patients (n = 12,699) and non-native patients (n = 1339)	16 chronic conditions, e.g., BP, asthma/COPD, cancer, diabetes, myocardial infarction, vascular disorders, stroke, migraine	04/2000 –01/2002	Socioeconomic disparities existed in the 16 chronic conditions with higher prevalence rates in lower groups; the differences were similar using self-assessed health and practitioner data; this shows that accessibility to primary care is sufficient for all strata in the Netherlands.
Yiannakoulias et al. 2009 [27]	Canada (Alberta)	Study on passive surveillance using secondary data with special focus on spatial surveillance of NCDs using GIS	IPD hospital data, medical claims system (electronic public health insurance registry), hospital outpatient system	Asthma	1998–2005	The study reveals spatial differences in the asthma prevalence in Alberta. Disease distribution depends on case ascertainment algorithms and is aggravated through information inequity. Spatial data in surveillance are important to inform policy makers about disease patterns; however availability of spatial data is a limiting factor in many countries.