The World Health Organization (WHO) African Region set a goal for regional measles elimination by 2020; however, regional measles incidence was 125/1,000,000 in 2012. To support elimination efforts, the WHO and U.S. Centers for Disease Control and Prevention developed a tool to assess performance of measles control activities and identify high‐risk areas at the subnational level. The tool uses routinely collected data to generate district‐level risk scores across four categories: population immunity, surveillance quality, program performance, and threat assessment. To pilot test this tool, we used retrospective data from 2006 to 2008 to identify high‐risk districts in Senegal; results were compared with measles case‐based surveillance data from 2009 when Senegal experienced a large measles outbreak. Seventeen (25%) of 69 districts in Senegal were classified as high or very high risk. The tool highlighted how each of the four categories contributed to the total risk scores for high or very high risk districts. Measles case‐based surveillance reported 986 cases during 2009, including 368 laboratory‐confirmed, 540 epidemiologically linked, and 78 clinically compatible cases. The seven districts with the highest numbers of laboratory‐confirmed or epidemiologically linked cases were within the capital region of Dakar. All except one of these seven districts were estimated to be high or very high risk, suggesting that districts identified as high risk by the tool have the potential for measles outbreaks. Prospective use of this tool is recommended to help immunization and surveillance program managers identify high‐risk areas in which to strengthen specific programmatic weaknesses and mitigate risk for potential measles outbreaks.
Measles; risk assessment; Senegal
All six World Health Organization (WHO) regions have set measles elimination goals for 2020 or sooner.[
In 2011, the 46 countries of the WHO African Region (AFR) set a goal for regional measles elimination by 2020.[
Senegal, in western Africa, had an estimated population of 13 million people and gross domestic product of US$1,132 per capita in 2011.[
To assist program managers with measles elimination efforts, the WHO and U.S. Centers for Disease Control and Prevention (CDC), with funding from the Bill & Melinda Gates Foundation, developed a tool to assess performance of measles control activities and identify high‐risk areas within a country. This tool combines routinely collected immunization, surveillance, and demographic data to assign risk categories at the subnational level. Pilot testing was conducted using retrospective data from Senegal. District‐level‐risk categories were determined from 2006 to 2008 data and compared visually and statistically to the distribution of cases during the 2009 measles outbreak.
A complete description of the risk assessment tool is found elsewhere.[
I Maximum Risk Points by Component of the World Health Organization (WHO) Measles Programmatic Assessment Tool for Risk of Measles Virus Transmission
Components Possible Points Cut‐Off Criteria (Risk Points) Population Immunity (40) District MCV1 coverage 8 ≥95% (+0); 90–94% (+2); 85–89% (+4); 80–84% (+6); <80% (+8) Proportion of neighboring districts with <80% MCV1 4 <50% (+0); 50–74% (+2); >75% (+4) District MCV2 coverage 8 Same as MCV1 coverage Measles SIA conducted within the past 3 years 8 Yes: ≥95% coverage (+0); 90–94% coverage (+2); 85–89% coverage (+4); <85% coverage (+6); No coverage data (+6); No SIA (+8) Target age group of measles SIA conducted within the past 3 years 2 Wide age group (+0); Narrow age group (+2); No SIA (+2) Years since the last measles SIA 4 <1 year (+0); 2 years (+2); >3 years (+4) Proportion of suspected cases who are unvaccinated or have unknown vaccination status 6 <20% (+0); >20% (+6) Surveillance Quality (20) Nonmeasles discarded rate 8 ≥2 per 100,000 (+0); <2 per 100,000 (+4); <1 per 100,000 (+8) Proportion of measles cases with adequate investigation 4 ≥80% (+0); <80% (+4) Proportion of measles cases with adequate specimens collection 4 ≥80% (+0); <80% (+4) Proportion of measles cases with laboratory results available in a timely manner 4 ≥80% (+0); <80% (+4) Program Performance (16) Trends in MCV1 coverage 4 Increasing or same (+0); ≤10% decline (+2); >10% decline (+4) Trends in MCV2 coverage 4 Same as MCV 1 trend MCV1‐MCV2 dropout rate 4 ≤10% (+0); >10% (+4) DPT1‐MCV1 dropout rate 4 Same as MCV1‐MCV2 dropout rate Threat Probability Assessment (24) ≥1 measles case reported among children <5 years during the past 12 months 4 No (+0); Yes (+4) ≥1 measles case reported among persons 5–14 years during the past 12 months 3 No (+0); Yes (+3) ≥1 measles case reported among persons ≥15 years during the past 12 months 3 No (+0); Yes (+3) Population density 4 0–50/km2 (+0); 51–100/km2 (+1); 101–300/km2 (+2); 301–1000/km2 (+3); >1000/km2 (+4) ≥1 measles case reported in a bordering district within the past 12 months 2 No (+0); Yes (+2) Presence of vulnerable groups 8 No vulnerable groups (+0); 1 point for each vulnerable group present (up to max of +8) Total possible points 100
1 Note: DPT1 = first dose in series for diphtheria, pertussis, and tetanus vaccination; MCV1 = first dose in series for measles‐containing vaccination; MCV2 = second dose in series for measles‐containing vaccination; SIA = supplementary immunization activity.
- 2 Vaccination coverage estimates from surveys if conducted within past three years and includes birth cohorts of recent three years can be used to replace administrative coverage.
- 3 Outbreak response immunization (ORI) campaign coverage data can be considered if an SIA was not conducted within the past 3 years and if the ORI targeted a geographical area that included the entire district.
- 4 Presence of vulnerable groups includes any of the following: (
1 ) migrant population, internally displaced population, slums, or tribal communities; (2 ) communities resistant to vaccination (i.e., religious, cultural, philosophical reasons, etc.); (3 ) security and safety concerns; (4 ) areas frequented by calamities/disasters; (5 ) poor access to health services due to terrain/transportation issues; (6 ) lack of local political support; (7 ) high‐traffic transportation hubs/major roads or bordering large urban areas (within and across countries); (8 ) areas with mass gatherings (i.e., trade/commerce, fairs, markets, sporting events, high density of tourists).
The population immunity category allocated points based on administrative coverage data for MCV1, MCV2, and measles SIAs. It also included the proportion of suspected measles cases that were unvaccinated or had unknown vaccination status according to the national measles case‐based surveillance data. Indicators in the surveillance quality category were calculated using the case‐based surveillance data and included the nonmeasles discard rate as well as the proportions of suspected measles cases with adequate investigations, adequate specimen collection, and timely laboratory results. The indicators for program performance were calculated from administrative data and included trends in MCV1 and MCV2 coverage, as well as dropout rates from MCV1 to MCV2 and from first dose of diphtheria, pertussis, and tetanus vaccine (DPT1) to MCV1. The threat assessment indicators accounted for factors that may influence the risk for measles virus transmission and were calculated from census data, case‐based surveillance data, and knowledge of vulnerable populations by staff at the Ministry of Health and Social Action (MOH). These indicators included population density, measles cases reported within specific age groups, measles cases reported in a bordering district, and presence of vulnerable groups (Table [NaN] ).
The distribution of all possible combinations of scores from the indicators was calculated, and the 50th, 75th, and 90th percentiles of this distribution were used as cut‐off points for four risk categories. Districts with scores below the 50th percentile (≤47 points) were defined as “low risk,” 50th–74th percentile (48–54 points) as “medium risk,” 75th–89th percentiles (55–60 points) as “high risk,” and 90th percentile or higher (≥61 points) as “very high risk.”
The 2009 measles case‐based surveillance data used for this assessment were existing data that had been collected according to the WHO AFR measles case‐based surveillance guidelines. The case definition for suspected measles was presence of maculopapular rash and fever plus one or more of cough, coryza, or conjunctivitis, or where a clinician suspected measles. All suspected cases reported were investigated and classified as laboratory‐confirmed, epidemiologically linked, clinically compatible, or discarded. Laboratory‐confirmed cases had a positive laboratory test result for measles‐specific immunoglobulin M (IgM) antibodies. Epidemiologically linked cases lacked laboratory results but had contact with or lived in the same district as laboratory‐confirmed case whose rash onset was within the preceding 30 days. Clinically compatible cases were defined as suspected measles cases without a laboratory test result or established epidemiological link. Suspected cases with a negative measles‐specific IgM result were discarded. We classified all laboratory‐confirmed and epidemiologically linked measles cases as confirmed measles cases. Measles incidence was calculated for each district using the number of confirmed measles cases divided by the estimated annual population, multiplied by 1,000,000. We used shape files provided by the MOH to create maps of risk categories, reported measles cases, and incidence. EPI data from 2006 to 2008 and national case‐based measles surveillance data from 2008 were used to assign a risk score to each district. Data were managed using Excel (Microsoft Corporation) and mapped using ArcGIS version 10.1 (ESRI). Assessment of the correlation between risk categories and confirmed measles incidence was made by the Kruskal‐Wallis test, after exclusion of districts with poor surveillance. Districts with ≥12 out of a possible 20 risk points for surveillance quality were excluded in the statistical analysis as their confirmed measles incidences were potentially highly inaccurate due to poor surveillance. All data were analyzed using SAS (version 9.3, SAS Institute, Cary, NC). Differences were considered significant when p < 0.05.
Overall scores for the 69 districts in Senegal in 2009 ranged from 35 to 66 points out of a possible 100 (Appendix A). The greatest variability between districts was in the categories of population immunity (18–38 point range) and surveillance quality (0–20 points). Threat assessment points ranged from 0–16 and program performance had the narrowest spread, ranging from 8 to 16 points. Seventeen districts (25%) were classified as either high risk (13 districts) or very high risk (four districts) (Figs. [NaN] and [NaN] ). The remaining 52 districts (75%) were classified as medium (27 districts) or low risk (25 districts) (Fig. [NaN] ). Six (75%) of eight districts in the region of Dakar were high or very high risk, including the districts with the two highest scores: Dakar‐Sud (66 points) and Pikine (64 points) (Fig. [NaN] ). The district containing Senegal's second largest city, Touba, was also high risk and the remaining high‐ and very high‐risk districts were clustered in areas of southern Senegal (Fig. [NaN] ).
We examined the underlying categories driving the overall risk scores. Within the Dakar region, high‐ and very high‐risk classifications were driven by poor population immunity scores (32–38 points) and high threat assessment scores (9–16 points) (Fig. [NaN] ). In all of these districts, the average MCV1 administrative coverage during 2006–2008 was ≤ 80%, contributing to the poor population immunity scores. High threat assessment scores in this area were primarily driven by high population densities, the presence of vulnerable populations, and bordering districts with measles cases in the prior 12 months. In other parts of the country, high‐ and very‐high‐risk scores tended to be driven by poor surveillance quality (8–20 points) and program performance (12–16 points) as well as population immunity (22–30 points) (Fig. [NaN] ). Several high‐ and very‐high‐risk districts (Dioffor, Passy, Guinguineo, and Makakoulibantang) had the maximum score for surveillance quality (20 points) because no suspected measles cases were reported in 2008, resulting in poor surveillance performance indicators and a discard rate of zero. Higher risk scores attributed to program performance typically resulted from decreasing trends in MCV1 coverage and substantial DPT1‐MCV1 dropout rates.
A total of 986 measles cases were reported during 2009. Among these, 368 were laboratory‐confirmed, 540 were epidemiologically linked, and 78 were clinically compatible cases. The seven districts with the highest number of reported confirmed (laboratory‐confirmed or epidemiologically linked) measles cases (N = 763) were all within Dakar Region: Dakar‐Sud (218 cases; confirmed measles incidence = 853/1,000,000), Dakar Centre (172 cases; incidence = 510/1,000,000), Pikine (160 cases; incidence = 261/1,000,000), Guediaway (78 cases; incidence = 253/1,000,000), Dakar Nord (73 cases; incidence = 194/1,000,000), Mbao (40 cases; incidence = 129/1,000,000), and Dakar‐Ouest (22 cases; incidence = 144/1,000,000) (Fig. [NaN] ). All of these districts were identified by the assessment tool as having high or very high risk except Mbao district, which was classified as medium risk, primarily because of lower risk points for population immunity than the other districts. The other 145 cases were spread out over an additional 39 districts with risk categories that varied from low to very high (Fig. [NaN] ). Nine districts reported incidence rates higher than 100/1,000,000 (Dakar Centre, Dakar Nord, Dakar Ouest, Dakar Sud, Guediawaye, Pikine, Mbao, Khombole, and Popenguine) and all except Mbao, Khombole, and Popenguine were considered high or very high risk. While Khombole and Popenguine had high incidence rates, their confirmed case counts (18 in Khombole and four in Popenguine) were lower than the districts within the Dakar region. The Kruskal‐Wallis test was carried out across all four risk assessment categories; however, eight districts were excluded from this analysis due poor surveillance quality, which may have resulted in an underreporting of cases (Appendix [NaN] ). There was a statistically significant association between risk assessment categories and confirmed measles incidence (H = 8.27, 3 d.f., p = 0.04).
To achieve regional measles elimination goals, weaknesses in Senegal's measles control programs should be identified regularly so that appropriate interventions can be implemented. The assessment tool uses routinely collected EPI, surveillance, and demographic information to identify high‐risk areas within a country and explain the specific types of weaknesses identified. Using retrospective data from Senegal, we identified 13 high‐risk and four very‐high‐risk districts in 2009. Of the seven districts with the highest measles case counts in 2009, six had been identified by the tool as high or very high risk, demonstrating the utility of the tool and suggesting that at‐risk districts have the potential for measles outbreaks. In addition, statistical comparisons of risk categories with historical outbreak data showed correlation between at‐risk districts and the occurrence of measles transmission during the following year.
In general, all districts had a substantial number of risk points for population immunity and program performance. One substantial factor contributing to the population immunity and program performance scores was the fact that Senegal had not yet introduced MCV2 into its routine schedule in 2006–2008. Thus all districts received a minimum of 8 points for population immunity and 8 points for program performance. A prospective risk assessment should be conducted for Senegal in 2015 to account for the introduction of MCV2 into the EPI program in 2013. Several additional districts had maximum risk scores for surveillance quality and high scores for threat assessment.
Application of the tool in Senegal provided valuable lessons about the risk for measles outbreaks in high population densities, especially in areas with historically low population immunity. The risk assessment tool identified six of eight districts in Dakar Region as high or very high risk and these districts reported most of the measles cases that occurred in 2009. These districts had very high population densities (threat assessment category) and poor population immunity indicated primarily by < 80% MCV1 administrative coverage. If the tool had been available and used before 2009, the results might have facilitated interventions to mitigate risk and potentially reduce the scale of the 2009 outbreak, especially in the Dakar Region. Several high‐risk districts in Senegal, but outside of Dakar, were identified as having weaknesses in surveillance. Training providers and district‐level staff on case definitions and surveillance guidelines might have also improved rapid case detection and reporting of epidemiological data to guide a timely outbreak response.
This tool can be used prospectively to monitor implementation of measles elimination strategies and to guide necessary corrective actions. For example, results could be used as advocacy to mobilize resources to strengthen RI services and SIA activities in high‐risk districts with poor population immunity. Additionally, districts that scored poorly in program performance could be targeted for additional supervisory visits to strengthen micro‐plans to help reach every child and reduce dropout rates. Awareness of risk identified in the threat assessment category could help program managers identify areas to target with intensified activities to protect vulnerable populations and prevent further spread of measles. An electronic version of the tool is expected to be available in 2016.
This risk assessment tool has limitations. First, the accuracy of results depends on the quality of data used. For example, if administrative data overestimate measles vaccination coverage through RI or SIAs in a district, then the risk scores will be biased to underestimate the overall risk in that district. Efforts should be made to provide high‐quality data to ensure optimal estimates of risk within each district. For example, data from coverage surveys, if available at the district level, could be used as a substitute for administrative coverage data. Another limitation is that districts with high risk scores driven by poor surveillance may not identify and report cases, presenting what appear to be incongruous results between their risk category and actual burden of disease.
This pilot test of the measles risk assessment tool in Senegal identified at‐risk districts that bore the greatest burden of the 2009 measles outbreak. With the information obtained through prospective use of this tool, program managers could tailor interventions and recommendations to the specific needs of subnational areas and help countries achieve measles elimination goals.
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the U.S. Centers for Disease Control and Prevention.
Risk Category Scores District Population Immunity Surveillance Quality Program Delivery Threat Assessment Total Risk Points (max = 100) Risk Category Confirmed Cases (2009)a Population (2009) ConfirmedIncidence per 1,000,000 (2009) Sud 38 4 8 16 66 VHR 218 255,622 853 Pikine 36 4 12 12 64 VHR 160 613,756 261 Dioffior 24 20 14 3 61 VHR 1 58,897 17 Guinguineo 24 20 14 3 61 VHR 0 94,603 0 Touba 28 12 14 6 60 HR 11 645,334 17 Nord 38 4 8 9 59 HR 73 376,397 194 Ouest 38 4 8 9 59 HR 22 152,738 144 Passy 24 20 12 3 59 HR 2 79,956 25 Kolda 26 12 14 6 58 HR 1 246,412 4 Birkilane 30 8 14 5 57 HR 0 91,310 0 Centre 32 0 12 13 57 HR 172 337,178 510 Makakoulibantang 22 20 14 1 57 HR 0 80,961 0 MedinaYoro Foulah 26 12 14 5 57 HR 0 93,894 0 Kaffrine 28 8 14 6 56 HR 2 278,216 7 Guediawaye 34 4 8 9 55 HR 78 308,004 253 MalemHodar 28 8 16 3 55 HR 1 88,769 11 Ndoffane 28 8 12 7 55 HR 0 149,583 0 Koungheul 28 8 16 2 54 MR 1 133,158 8 Mbao 26 0 12 16 54 MR 40 308,938 129 Dahra 28 8 16 1 53 MR 1 135,068 7 Joal Fadiouth 26 4 14 9 53 MR 1 76,407 13 Saraya 26 4 14 9 53 MR 0 27,511 0 Sedhiou 28 8 14 3 53 MR 5 154,792 32 Matam 30 8 14 0 52 MR 5 264,120 19 Kaolack 24 8 12 7 51 MR 9 254,118 35 Thies 30 4 8 9 51 MR 0 356,149 0 Bignona 30 4 14 2 50 MR 2 129,034 15 Bounkiling 26 8 14 2 50 MR 0 108,456 0 Pout 30 4 12 4 50 MR 5 80,555 62 Rufisque 20 8 14 8 50 MR 4 336,025 12 Thiadiaye 28 4 14 4 50 MR 2 161,964 12 Bakel 28 8 12 1 49 MR 0 86,595 0 Goudomp 28 4 14 3 49 MR 0 161,380 0 Kedougou 26 4 12 7 49 MR 0 74,270 0 Mbour 22 4 14 9 49 MR 4 302,494 13 Mekhe 26 12 8 3 49 MR 1 141,453 7 Salamata 28 4 12 5 49 MR 0 19,803 0 Ziguinchor 32 0 14 3 49 MR 0 203,608 0 Goudiry 28 8 12 0 48 MR 0 61,028 0 Khombole 26 4 14 4 48 MR 18 132,672 136 Kidira 28 4 14 2 48 MR 0 54,919 0 Nioro 26 4 14 4 48 MR 1 305,882 3 Pete 30 4 14 0 48 MR 4 180,098 22 Podor 30 4 14 0 48 MR 1 209,011 5 Darou Mousty 26 4 14 3 47 LR 2 79,291 25 Diouloulou 32 0 14 1 47 LR 0 59,818 0 Ranerou 30 4 12 1 47 LR 4 57,943 69 Saint Louis 24 4 12 7 47 LR 1 244,339 4 Tambacounda 28 4 12 3 47 LR 0 186,340 0 Dianke Makha 28 4 12 2 46 LR 0 43,761 0 Fatick 28 0 14 4 46 LR 6 222,380 27 Oussouye 32 4 10 0 46 LR 0 37,596 0 Tivaoune 24 4 12 6 46 LR 5 210,001 24 Linguere 24 4 16 1 45 LR 3 105,299 28 Mbacke 22 4 14 5 45 LR 7 151,504 46 Popenguine 32 0 8 5 45 LR 4 39,480 101 Thionk Essyl 32 0 12 1 45 LR 0 49,804 0 Kanel 30 0 14 0 44 LR 1 230,915 4 Gossas 24 4 12 3 43 LR 1 98,025 10 Velingara 18 4 14 7 43 LR 0 227,482 0 Louga 24 0 12 5 41 LR 4 325,553 12 Richard Toll 18 4 16 3 41 LR 5 146,471 34 Diourbel 18 4 12 6 40 LR 11 235,736 47 Foundiougne 26 0 12 1 39 LR 0 40,957 0 Bambey 20 4 10 4 38 LR 1 263,504 4 Kebemer 18 0 14 6 38 LR 0 148,199 0 Sokone 22 0 14 1 37 LR 1 125,492 8 Dagana 24 0 12 0 36 LR 6 83,634 72 Koumpentoum 22 0 12 1 35 LR 1 115,670 9
5 Includes laboratory‐confirmed and epidemiologically linked cases.
Graph: Point distributions for districts with high‐ and very‐high‐risk scores on measles risk assessment tool—Senegal, 2009.
Graph: District‐level point distributions on measles risk assessment tool—Dakar Region, Senegal, 2009.
Graph: Comparison of (a) measles risk level categories, (b) measles cases reported, and (c) measles incidence—Dakar, Senegal, 2009.
Graph: Comparison of (a) measles risk level categories, (b) measles cases reported, and (c) measles incidence—Senegal, 2009.
Graph: B1 Lines in the boxes denote median values; diamonds indicate the mean. VHR=Very high risk; HR=High risk; MR=Medium risk; LR=Low risk. Districts with ≥ 12 out of a possible 20 risk points for surveillance quality were excluded due to high likelihood of inaccurate measles incidence. Districts excluded were: Touba, Dioffior, Passy, Guinguineo, Kolda, Medina Yoro Foulah, Makacoulibantang, and Mekhe.
By Jennifer B. Harris; Ousseynou Badiane; Eugene Lam; Jennifer Nicholson; Ibrahim Oumar Ba; Aliou Diallo; Amadou Fall; Balcha G. Masresha and James L. Goodson