International Measles Incidence and Immunization Coverage

Transcription

1 SUPPLEMENT ARTICLE International Measles Incidence and Immunization Coverage Robert Hall and Damien Jolley Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, The Alfred Centre, Melbourne, Victoria, Australia Measles is exquisitely sensitive to immunization programs. We investigated the decline in measles incidence after immunization with 1 or 2 doses of measles-containing vaccine (MCV), with or without supplementary immunization activities (SIAs). Using data from the World Health Organization, we modeled the impact of measles immunization using a negative binomial regression model. All countries offer measles immunization, and 192 of 193 countries offer a second dose of MCV (MCV2), using either a routine second dose, SIAs, or both. The incidence of measles fell from a median of 70.9 cases/100,000/year when coverage with a first dose of MCV (MCV1) was in the range of 0% 39% to a median of.9 cases/ 100,000/year when MCV1 coverage was 90% 100%, in both cases with no MCV2. Further reductions followed the introduction of MCV2 and SIAs. Modeling showed that each 1% increase in MCV1 coverage was followed by a 2.0% decrease in incidence in the same and following years (95% confidence interval [CI], 2.0% 1.9%, and 2.1% 1.9%, respectively). For a second dose, a rise of 1% in MCV2 coverage was followed by a decrease in measles incidence by.4% (95% CI,.3%.5%) in the same year and.3% (95% CI,.2%.5%) in the following year. SIAs were followed by decreases of measles incidence by 40.3% (95% CI, 46.3% 33.8%) in the same year and 45.2% (95% CI, 51.1% 48.7%) in the following year. A herd immunity effect was demonstrated with MCV1 coverage of.80%, and SIAs are an extraordinarily effective strategy for measles control. Historically, measles has been responsible for a vast amount of suffering and death among the world s children, and, despite great progress in measles control over the last 40 years, this remains true. The disease is characterized by both high mortality and morbidity and high transmissibility, especially in developing countries [1]. Complications of measles include viral and bacterial pneumonias, diarrhea, blindness, and encephalitis. Measles is one of the most transmissible of viral infections; before the widespread use of vaccine, 90% of children had contracted measles by the age of 10 years [2], and in a population that is completely immunologically Potential conflicts of interest: none reported. Correspondence: Robert Hall, MBBS MPH, School of Public Health and Preventive Medicine, Monash University, The Alfred Centre, 99 Commercial Rd, Melbourne, Victoria 3004, Australia (robert.hall@monash.edu). The Journal of Infectious Diseases 2011;204:S158 S163 Ó The Author Published by Oxford University Press on behalf of the Infectious Diseases Society of America. All rights reserved. For Permissions, please journals.permissions@oup.com (print)/ (online)/2011/204s1-0021$14.00 DOI: /infdis/jir124 naive, a case of measles, on average, results in 15 secondary cases [3]. An effective vaccine has been available since the 1960s, and all countries offer measles-containing vaccine (MCV) in their immunization programs. Protection is increased after 2 doses, and the World Health Organization (WHO) recommends a 2-dose immunization schedule [2]. Delivery of a second dose of MCV (MCV2) is problematic in many countries and is frequently achieved through supplementary immunization activities (SIAs). These SIAs target all children in a specified age range (typically 9 months to 5 years) in specified geographic regions (either subnational or national) and offer MCV irrespective of immunization history. SIAs are intended to ensure that as many children as possible receive $1 dose of MCV. This strategy can rapidly achieve a high level of immunity among the population and can cause dramatic falls in measles transmission [2]. Measles is consideredtofulfillthe criteria for eradicability [4]. In recognition of this, several countries and WHO regions have established S158 d JID 2011:204 (Suppl 1) d Hall and Jolley

2 measles elimination targets [5], and several countries and the WHO Region of the Americas have claimed to have eliminated measles (Table 2 [6]). To achieve high-level control or elimination targets, national immunization programs are presented with a range of possible vaccine delivery strategies. We aimed to investigate the effectiveness of different strategies in achieving measles control. DATA AND METHODS All member states of the WHO report annual measles immunization coverage and incidence to the WHO and the United Nations Children s Fund using the Joint Reporting Form. Generally, immunization coverage is reported as the number of children in a defined target population who are recorded by the immunization program as receiving vaccine. Countries also report any SIAs and coverage achieved in those activities. Measles incidence data are reported from each country s disease surveillance system. The WHO collates these data, and many have been published on the internet [7]. We obtained data from the WHO for the 193 member states and areas for the period inclusive. The data set contained 34 variables for each country, including measles immunization schedule, reported annual coverage for a first dose of MCV (MCV1), reported annual MCV2 coverage for countries that conduct a 2-dose immunization schedule, SIAs for countries that conducted them, and average coverage achieved, where available. We organized and analyzed the data using Stata software (version 11; StataCorp); each observation was a record for a single country for 1 year. This data set contained 5404 observations. We grouped coverage data for MCV1 into 5 categories (0% 39%, 40% 59%, 60% 79%, 80% 89%, and 90% 100%) and for MCV2 into 6 categories (,1%, 1% 24%, 25% 49%, 50% 79%, 80% 89%, and 90% 100%). The,1% category generally represents absence of a second dose in the country s immunization schedule for the year concerned. Coverage categories were chosen to reflect observed coverage in the field and to enable differentiation of higher levels of coverage We investigated the relationships between coverage data and measles incidence reported in the same year and, because we hypothesized that the effects of immunization programs may be delayed until the following year, the relationship between immunization coverage achieved in a given year and the reported incidence of measles in the following year. We modeled the data using cross-sectional time-series regression models for count data. We used the number of reported measles cases in each year and each country as the outcome variable, and as predictors we included the rates of coverage for MCV1 and (where and when it was offered) MCV2inthesamecountryandsame year. We also included an indicator variable for SIAs in the same country-year record. We used negative binomial regression modeling, because this probability model accommodates overdispersed count data (such as disease incidence), and we used the country s population at the time as an exposure offset variable. Effectively, we modeled the population incidence for measles in each country for each year. Because of our interest in marginal gains in the presence of high MCV1 coverage, we included as a predictor variable in some models an indicator of MCV1 coverage of $80%. We included the interaction between this indicator and MCV2 coverage or SIAs. When these terms were included, we were able to estimate any change in effectiveness of the coverage with MCV1, MCV2, and/or SIAs when MCV1 coverage exceeded 80%. In models where the outcome was for the following year, the outcome variable (and exposure denominator) for a year was regressed on the coverage data for the previous year in the same country. There was a slight consequential reduction in data for the following year, because the coverage data in each country for the last year were excluded. RESULTS All 193 countries offered $1 dose of MCV. Schedules varied: 78 countries offered MCV1 at age #9 months, 1 country at 9 12 months, and 114 at $12 months. As of 2007, 132 countries offered MCV2. Supplementary activities have been conducted in 152 countries, with widely varying frequencies, target ages, geographic extent, and coverage achieved. Incidence in the Year Coverage Was Measured First, we present the incidence of measles observed in the same year as the recorded coverage with MCV. A total of 4163 usable country-year observations were available where only routine vaccination was carried out, with no SIAs, including countryyears where the immunization schedule recommended either 1 or 2 doses of MCV. For country-year observations with no MCV2 coverage, the median observed incidence in the same year was 70.9 cases/ 100,000/year for MCV1 coverage of 0% 39% (785 country-year observations) and.9 cases/100,000/year for MCV1 coverage of 90% 100% (828 observations) (Table 1). This demonstrates that high MCV1 coverage has a dramatic effect on the incidence of measles. For 90% 100% coverage with both MCV1 and MCV2 (301 country-year observations), the median incidence in the same year was.2 cases/100,000/year. Generally, the incidence of measles was low when rates of coverage were high for both MCV1 and MCV2 (Table 1). Countries carrying out $1 SIA during the year of observation generally reported lower incidences of measles than countries that did not conduct SIAs. Where SIAs were carried out (368 country-year observations with usable data), median incidences were generally lower when MCV1 and MCV2 coverage rates were higher. In countries conducting SIAs, the median reported incidence in the same year was 5.7 cases/100,000/year when MCV1 coverage was in the range of 0% 39% and MCV2 was Measles Incidence and Immunization Coverage d JID 2011:204 (Suppl 1) d S159

3 Table 1. Median Reported Measles Incidence in the Same Year by MCV1 and MCV2 Coverage and Supplementary Immunization Activities MCV1 coverage, % No SIAs $1 SIA MCV2 coverage, % , (785) 38.3 (549) 19.9 (750) 5.5 (762).9 (828) 5.7 (6) 5.7 (38) 4.9 (84) 2.3 (72).5 (69) (1) 0 (1) 0 (1).5 (2) 7.7 (2).. 0 (1) 0 (1) (2) 5.2 (4).6 (8).4 (3) 5.6 (5). 3.5 (1) 14.6 (4).1 (3) 43.2 (1) (12).4 (32) 0 (25). 0 (1) 23.3 (5).1 (7) 0 (6) (1).2 (25) 0 (44) (2) 4.4 (6) 0 (10) (3).2 (17).2 (301) (3) 4.1 (8).7 (40) NOTE. Incidences are for the same country and year as the coverage data; data in cells represent median reported cases/100,000/year (number of country-year observations). MCV1, first dose of measles-containing vaccine (MCV); MCV2, second dose of MCV; MCV2 coverage,1% generally means that MCV2 was not offered; SIA, supplementary immunization activity. not offered (6 country-year observations) and.5 cases/100,000/ year when MCV1 coverage was in the range of 90% 100% (69 observations). When SIAs were conducted and both MCV1 and MCV2 coverage rates were in the range of 90% 100% (40 country-year observations), the median reported incidence in the same year was.7 cases/100,000/year (Table 1). Incidence in the Year After Coverage Was Measured The impact of measles immunization coverage is not immediate, so we investigated the incidence of reported measles in the year after reported immunization with MCV. In this analysis, each country-year observation is for the year after vaccination. When MCV2 was not offered, the median observed incidence in the year after MCV1 was 59.4 cases/100,000/year for the 812 country-year observations with recorded MCV1 coverage in the range of 0% 39% and.7 cases/100,000/year for the 814 observations with MCV1 coverage of 90% 100%. The 254 country-year observations with 90% 100% coverage for both MCV1 and MCV2 were followed by a median incidence of.2 cases/ 100,000/year (Table 2). There were 320 country-year observations with usable data where SIAs were carried out. For countries conducting SIAs, the median reported incidence in the following year was 2.6 cases/ 100,000/year for the 6 country-years where MCV1 coverage was in the range of 0% 39% and MCV2 was not offered and.2 cases/ 100,000/year for the 32 country-years where both MCV1 and MCV2 coverage was 90% 100%.(Table 2). Negative Binomial Model We modeled measles incidence as a negative binomial distribution, using MCV1 coverage, MCV2 coverage, and SIAs as predictor variables. In the first model, we included these variables without interaction. For each 1% increase in coverage with MCV1, there was a 2.0% fall in reported measles incidence in both the same year and the following year. This result was significant (95% CI, 2.0% 1.9% for the same and 2.1% 1.9% for the following year). The impact of MCV2 coverage was smaller, with decreases in incidence of.4% (95% CI,.3%.5%) in the same year and.3% (95% CI,.2%.5%) in the following year for each 1% increase in coverage. SIAs had a marked effect, with the Table 2. Median Reported Measles Incidence in the Following Year by MCV1 and MCV2 Coverage and Supplementary Immunization Activities MCV1 coverage, % No SIAs $1 SIA MCV2 coverage, % , (812) 34.3 (546) 19.4 (735) 4.5 (747).7 (814) 2.6 (6) 4.4 (35) 2 (74).5 (64).1 (62) (1). 0 (1).5 (2) 11.8 (2) (2) (2).1 (8).3 (2).9 (5). 4.8 (1) 4.3 (4).1 (3) (10).3 (28) 0 (20). 0 (1) 20.1 (5).1 (5) 2.2 (3) (2).1 (23) 0 (34) (2).1 (4) 0 (9) (3).1 (16).2 (254) (3) 5.4 (7).2 (32) NOTE. Incidences are for the same country and the year after the coverage data; data in cells represent median reported cases/100,000/year (number of country-year observations). Table 2 available at viewed 1 February MCV1, first dose of measlescontaining vaccine (MCV); MCV2, second dose of MCV; MCV2 coverage,1% generally means that MCV2 was not offered; SIA, supplementary immunization activity. S160 d JID 2011:204 (Suppl 1) d Hall and Jolley

4 Table 3. Incidence Rate Ratios for Measles Predictors in a Negative Binomial Model of the Impact of Increasing Coverage With Measles-Containing Vaccine (MCV), Without Interaction Terms Covariate Incidence rate ratio (95% CI) Same year (N ) Following year (N ) MCV1 coverage.980 ( ).980 ( ) MCV2 coverage.996 ( ).997 ( ) $1 SIA.597 ( ).584 ( ) NOTE. Data from 193 countries, ; CI, confidence interval; MCV1, first dose of MCV; MCV2, second dose of MCV; N, number of country-year observations; SIA, supplementary immunization activity. incidence of measles falling by 40.3% (95% CI, 46.3% 33.8%) in the same year and by 41.6% (95% CI, 38.7% 51.1%) in the following year (Table 3). There was little change in these values when we included interaction between predictor variables in the model. The main effect for MCV1 coverage was a 1.9% (95% CI, 2.0% 1.8%) decrease in incidence of measles for each 1% increase in MCV1 coverage, for both the same year and the following year. The effect of MCV2 disappeared, with a reduction in incidence of 0% (95% CI, 2.4% to.4%) in the same year and an increase of.1% (95% CI, 2.4% to.5%) in the following year for each 1% increase in coverage (Table 4). There was an additional effect at high levels of MCV1 coverage, with a reduction in the reported incidence of measles in both the same year and the following year. In the absence of SIAs, when coverage with MCV1 reached 80%, each 1% increase in coverage led to decreases in incidence of 11.3% (95% CI, 3.38% 18.2%) in the same year and 14.6% (95% CI, 7.2% 21.3%) in the following year. When the effect of MCV1 coverage is taken into account, there is no further decline in incidence with increases in MCV2 coverage in the same year and a small increase (.1%; 95% CI, -.4%.5%) in the following year. For MCV1 coverage,80%, SIAs are associated with decreases in measles incidence of 35% (95% CI, 24.7% 44.6%) in the same year and 45% (95% CI, 35.3% 55.4%) in the following year. MCV1 coverage of #80% combined with $1 SIA, without interaction, would be expected to be associated with a 43% (1 2 [ ]) decline in reported incidence in the same year,but the interaction with higher MCV1 coverage (.80%)reduces this still further to a51%decreaseinthesameyear. DISCUSSION Our results show that immunization with a single dose of MCV is highly effective in controlling measles. Our model predicts that for each 1% increase in coverage with MCV1 without SIAs, a 2% decrease in reported disease incidence will be seen, in both the same year and the following year. There is an additional effect of MCV1 coverage when levels of immunization coverage reach 80%, perhaps because of a herd immunity effect, where the smaller pool of susceptible individuals in a population results in lower rates of transmission [8]. Vaccination with MCV2 is followed by a further dramatic decrease in median reported measles incidence. The model predicts a reduction in incidence of.4% in the same year with each additional 1% increase in MCV2 coverage, and virtually the same reduction in the following year (.3% decrease in incidence with each 1% increase in MCV2 coverage). SIAs are followed by further decreases in median reported measles incidence. Our model predicted a 40% reduction in measles incidence in the same year as a SIA and 45% reduction in the following year (Figure 1). SIAs are exceptionally good means of controlling measles. Our analysis has a number of limitations. Measles epidemiology varies greatly in the world. In some countries there is intense transmission, with very high incidence in young children. Our analysis did not differentiate between countries with different intensities of transmission. We did not analyze the effect of age at immunization, but during the period of our analysis many different schedules were used by countries in their immunization Table 4. Incidence Rate Ratios for Measles Predictors in a Negative Binomial Model of the Impact of Increasing Coverage With Measles-Containing Vaccine, With Interaction Terms Incidence rate ratio (95% CI) Covariate Same year (N ) Following year (N ) MCV1 coverage.981 ( ).981 ( ) MCV1 coverage.80%.887 ( ).854 ( ) MCV2 coverage ( ) ( ) $1 SIA.646 ( ).549 ( ) $1 SIA and MCV1.80% (interaction).851 ( ).970 ( ) NOTE. Data from 193 countries, ; CI, confidence interval; MCV, Measles-Containing Vaccine; MCV1, first dose of MCV; MCV2, second dose of MCV; N, number of country-year observations; SIA, supplementary immunization activity. Measles Incidence and Immunization Coverage d JID 2011:204 (Suppl 1) d S161

5 Figure 1. Modeled incidence of measles reported in the following year by immunization coverage with first and second doses of measlescontaining vaccine and the implementation of supplementary immunization activities (SIAs). Upper panels show first-dose coverage rates of 0% 39%, 40% 59%, 60% 79%, 80% 89%, and 90% 100%, all without SIAs; lower panels show the same first-dose coverage rates, all with $1 SIA. programs, with MCV1 given at ages ranging from before 9 months to after 18 months, and MCV2 at widely varying ages. We also did not account for differences between the scheduled and actual ages at immunization. The scheduled age for the second dose varied from 12 months to.7 years [9], with variations in protection conferred. Our analysis also did not differentiate between SIAs. SIAs had varying targets by age and geographic extent. For example, SIAs that targeted a single province of a country were analyzed in the same way as nationwide SIAs. We made no distinction between SIAs achieving high levels of coverage and those for which coverage was low. Data quality and comparability are also major issues, because different countries use different methods for estimating population, immunization coverage, and measles incidence. Routinely collected data on immunization coverage often suffer from population underestimates and coverage overestimates. Communicable disease surveillance data are commonly believed to be underestimates. It is likely that SIAs with low coverage and a narrow geographic or target-age focus have less effect than high-coverage SIAs, and we consider our estimates of SIA effect to be underestimates. Although the model was a poor fit to the data (owing to these and other factors, all contributing to the extreme dispersion evident in incidence rates for 193 nations and 28 years), we are confident that the log-likelihood tests and estimates for the nested negative binomial models remain valid. Our results show that measles is exquisitely sensitive to immunization programs, and even moderate coverage is followed by a low incidence of measles. High coverage with 2 doses, or SIAs, at any level of routine coverage, is followed by a very low incidence of measles. In particular, SIAs have a spectacular impact on measles transmission and they should be the preferred strategy. Cessation of SIAs and the addition of a second dose to the routine schedule should be considered only when high levels of MCV2 coverage can be achieved and maintained. Funding This work was supported by a contract from the World Health Organization. References 1. Progress in global measles control and mortality reduction, Wkly Epidemiol Rec 2008; 83: Measles vaccines: WHO position paper. Wkly Epidemiol Rec 2009; 84: Gay NJ. The theory of measles elimination: implications for the design of elimination strategies. J Infect Dis 2004; 189(Suppl 1):S Meeting of the International Task Force for Disease Eradication, June Wkly Epidemiol Rec 2009; 84: World Health Organization Regional Committee for the Western Pacific. Measles elimination, hepatitis B control and poliomyelitis eradication. Resolution WPR/RC56.R en/rc56/rc_resolutions/wpr_56_r08.htm. Accessed 4 December S162 d JID 2011:204 (Suppl 1) d Hall and Jolley

6 6. Heywood AE, Gidding HF, Riddell MA, Mcintyre PB, MacIntyre CR, Kelly HA. Elimination of endemic measles transmission in Australia. Bull World Health Organ 2009; 87: volumes/87/1/ table-t2.html. Accessed 4 December World Health Organization. Immunization surveillance, assessment and monitoring Available at: monitoring/data/data_subject/en/index.html. Accessed 4 December John TJ, Samuel R. Herd immunity and herd effect: new insights and definitions. Eur J Epidemiol 2000; 16: UNICEF. Immunization summary: the 2007 edition. New York: UNI- CEF Measles Incidence and Immunization Coverage d JID 2011:204 (Suppl 1) d S163