Table 1. Key findings from the evaluation of the influenza surveillance system in Madagascar, 2009–2014.
| Attribute, issue Data quality |
Indicator | Key findings | Scorea |
|---|---|---|---|
| Does the information submitted contain all mandatory and/or requested data items and are the data recorded valid? | Proportion of expected SMS messages and CRF that were received | 93.0% (IQR:70.2–98.1) of expected SMS and 89.5% (IQR: 40.9–95.3) of expected CRF | 3 |
| Proportion of SMS and CRF without missing or inconsistent value for selected key variablesb | 99.9% (44 203/44 252) of SMS and 96.6% (117 397/121 543) of CRF | 3 | |
| Proportion of ILI cases that met the case definition | 94.9% (24 490/25 809) | 3 | |
| Proportion of sampled ILI cases that met the sampling criteriac | 99.5% (9251/9293) | 3 | |
| Proportion of sampled ILI cases with available laboratory results | 98.9% (9192/9293) | 3 | |
| Proportion of collected variables included in the WHO recommended minimum data collection for influenza sentinel surveillance | Data on antiviral treatment and underlying medical conditions were not collected | 2 | |
| Timeliness | |||
| Are the data and samples from the surveillance system collected and dispatched without delay? | Proportion of SMS texts sent within 48 hours of reference day | 69.8% (IQR: 59.8–77.1) | 2 |
| Proportion of data collection forms received by IPM within 7 days of data collection | 90.3% (IQR: 81.2–98.1) | 3 | |
| Proportion of samples received by IPM within 48 hours of collection | 45.9% (IQR: 29.9–72.7) | 1 | |
| Proportion of weekly reports issued within the target date | 90.1% (281/312) | 3 | |
| Representativeness | |||
| Are the data collected on influenza by the surveillance system representative of the general population in Madagascar? | Geographical coverage | Surveillance sites located in all provinces | 3 |
| Inclusion of all age groups | Although all age groups were eligible, median age was only 4 years (range: 1 day–91 years) | 2 | |
| Simplicity | |||
| Do the surveillance staff find the system easy to implement? | Surveillance staff’s perceptions of how easy certain surveillance activities are to use – categorized as very difficult, difficult or easy | Of 50 respondents, the collection of aggregated data, the completion of CRF and SMS-based data transfer were reported to be easy by 47, 50 and 50, respectively | 3 |
| Performance of the courier in collecting CRF from sentinel sites | Of 50 respondents, 27 reported that they had rarely or never experienced delays in the collection of CRF | 1 | |
| Performance of the courier in collecting samples from sentinel sites | All the 18 respondents from sites where samples were collected reported that they had rarely or never experienced delays in the collection of samples | 3 | |
| Acceptability | |||
| Do the surveillance staff and key stakeholders find the system acceptable? | Proportion of surveillance staff that were satisfied with reports and follow-ups | Of the relevant staff interviewed, 17/18, 42/50 and 49/50 reported being satisfied with the virological reports, quarterly bulletins and telephone follow-ups, respectively | 3 |
| Proportion of work time devoted to surveillance activities | 37% and 25% for the 50 respondents from the sentinel sites and 17 respondents from the IPM, respectively | 2 | |
| Mean annual cost of the surveillance system, for ILI surveillance | US$ 94 364 | 2 | |
| Flexibility | |||
| Could the system be easily adapted to cover illnesses other than influenza? | Number of syndromes surveyed under the fever surveillance system | Four: arboviruses, diarrhoea, influenza and malaria | 3 |
| Number of pathogens surveyed under the ILI component of the fever surveillance system | The system can detect up to 14 respiratory viruses | 3 | |
| Stability | |||
| Does the system function smoothly and does it appear sustainable? | Proportion of evaluated weeks during which all sentinel sites were functional | 93.3% (291/312) | 3 |
| Proportion of data queries successfully resolved | 93.2% (137/147) | 3 | |
| Availability and use of SOP for surveillance | Of 50 respondents, 29, 46 and 44 reported making regular use of sample collection, decision tree and surveillance procedures, respectively | 3 | |
| Frequency of interruptions in supplies | Of 50 respondents, 28, 9 and 36 reported no interruptions in the supplies of CRF, sampling materials and telephone credit for SMS, respectively | 2 | |
| Proportion of sentinel sites with at least one member of staff trained in sentinel surveillance procedures | 71.9% (23/32) | 2 | |
| Proportion of surveillance staff trained in sentinel surveillance procedures | Training had been received by 66.7% (18/27) of respondents with primary responsibility for surveillance activities and 34.8% (8/23) of respondents who were supporting staff | 1 | |
| Utility | |||
| Does the system provide information that is useful for public health authorities and communities? | Number of ILI alerts detected | In 2014, 38 alerts were detected in 16 sentinel sites | 3 |
| Proportion of sentinel sites – other than those that collected samples routinely – that initiated collection of samples after local ILI alert | 72.7% (8/11) | 2 | |
| Isolation and sharing of circulating seasonal influenza strains | NIC shared circulating isolates with WHO Collaborating Centres 11 times – out of the 12 requested by WHO | 3 | |
| Identification capacity for emerging influenza strains with pandemic potential | NIC successfully passed nine external quality assessments, with an overall score of 98.9% | 3 | |
| Proportions of surveillance staff that receive the virological reports, the influenza surveillance reports and influenza bulletins | 12 (66.7%) of 18 respondents working at biological sites, 35 (70.0%) of all 50 respondents and 27 (54.0%) of all 50 respondents had reportedly received the virological reports, influenza surveillance reports and the influenza bulletins, respectively | 2 |
CRF: case report forms; ILI: influenza-like illness; IPM: Institut Pasteur de Madagascar; IQR: interquartile range; NIC: national influenza centre; SMS: short message service; SOP: standard operating procedures; US$: United States dollars; WHO: World Health Organization.
a Each quantitative indicator with a percentage value of < 60%, 60–79% and ≥ 80% were given scores of 1, 2 and 3, respectively – representing weak, moderate and good performance, respectively. The qualitative indicators were also scored 1, 2 or 3 but these scores were based on the consensus opinions of 10 surveillance experts – i.e. three virologists, three public health specialists, two epidemiologists and two surveillance officers – who worked at the Institut Pasteur de Madagascar or the Malagasy Ministry of Public Health.
b The key variables evaluated for SMS data were code of sentinel site, date of patient visit and numbers of fever cases, ILI cases and patients. Those evaluated for CRF were absence/presence of fever with cough and/or sore throat, code of sentinel site and dates of patient visit and symptom onset.
c Any of the following were considered to be failures in meeting the sampling criteria: specimen collection tube vial left open; sample without patient identification; no corresponding CRF; no identification of the patient; time between date of onset and date of sampling either ≥ 7 days or not available; time between date of sampling and sample receipt at IPM either ≥ 7 days or not available; sampling kit used after its stated expiry date; no diagnosis of influenza.