Effectiveness of incidence thresholds for detection ... - Semantic Scholar

3 downloads 0 Views 207KB Size Report
Mar 3, 2000 - Health Organization (WHO) to control meningococcal meningitis epidemics is based upon therapeutic case management and mass vaccination ...
© International Epidemiological Association 2000

International Journal of Epidemiology 2000;29:933–940

Printed in Great Britain

Effectiveness of incidence thresholds for detection and control of meningococcal meningitis epidemics in northern Togo Anne-Valérie Kaninda,a François Belanger,a Rosamund Lewis,a Essosolem Batchassi,b Aristide Aplogan,a Yénou Yakouab and Christophe Paqueta

Background Early outbreak detection is necessary for control of meningococcal meningitis epidemics. A weekly incidence of 15 cases per 100 000 inhabitants averaged over 2 consecutive weeks is recommended by the World Health Organization (WHO) for detection of meningitis epidemics in Africa. This and other thresholds are tested for ability to predict outbreaks and timeliness for control measures. Methods

Meningitis cases recorded for 1990–1997 in health centres of northern Togo were reviewed. Weekly and annual incidences were determined for each district. Ability of different weekly incidence thresholds to detect outbreaks was assessed according to sensitivity, specificity, and positive and negative predictive values. The number of cases potentially prevented by reactive vaccination in 1997 was calculated for each threshold.

Results

Outbreaks occurred in 1995–1996 and in 1996–1997. The WHO-recommended threshold had good specificity but low sensitivity. Thresholds of 10 and 7 cases per 100 000 inhabitants in one week had sensitivity and specificity of 100% and increased the time available for intervention by more than one or two weeks, respectively. A maximum of 65% of cases could have been prevented during the 1997 epidemic, with up to 8% fewer cases prevented for each week of delay in achieving vaccine coverage.

Conclusions In northern Togo, thresholds of 7 or 10 cases per 100 000 inhabitants per week were excellent predictors of meningitis epidemics and allowed more time for a reactive vaccination strategy than current recommendations. Keywords

Meningitis, meningococcal, epidemic, threshold, vaccination, Africa, sensitivity, specificity

Accepted

3 March 2000

Meningococcal meningitis, caused by the bacterium Neisseria meningitidis, is a public health priority in the sub-Saharan region of Africa known as the ‘meningitis belt’.1 In this area, meningococcal meningitis is endemo-epidemic, with tens of thousands of cases and thousands of deaths reported during epidemic years.1–3 The strategy currently recommended by the World Health Organization (WHO) to control meningococcal meningitis epidemics is based upon therapeutic case management and mass vaccination with polysaccharide vaccine at the beginning of an epidemic.3 As seasonal variations in the number of meningitis cases are observed in the meningitis belt with an annual increase at the end of the dry season (December–April), it is necessary to identify a threshold which is generally exceeded at the beginning of an epidemic but not during endemic cycles. a Epicentre, 8 rue Saint-Sabin, 75011 Paris, France. b Direction Générale de la Santé, Ministère de la Santé, BP 336, Lomé, Togo.

Based on meningitis case data from Burkina Faso (1979– 1984), Moore et al.4 proposed the use of a weekly meningitis incidence rate of 15 cases per 100 000 inhabitants, averaged over 2 consecutive weeks, as a threshold to predict the subsequent occurrence of a meningitis epidemic and a lower ‘alert’ threshold of 5 cases per 100 000 inhabitants per week in districts contiguous to areas already declared epidemic. In 1995, WHO recommended application of these thresholds for control of epidemic meningococcal disease in the African meningitis belt3 and they have been used since then.5–7 Experience has shown, however, that under field conditions, the time required to decide upon and organize mass vaccination campaigns may delay interventions, resulting in a limited overall impact.6,7 The northern part of Togo, at the southern edge of the African meningitis belt, suffered a major group A meningococcal meningitis epidemic in 1996–19978 as a wave of epidemics swept through the region.9 This study, with data from northern Togo,

933

934

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

assesses the validity of the currently recommended threshold and studies other thresholds that might allow an earlier detection of meningococcal meningitis epidemics and thereby increase the impact of reactive mass vaccination campaigns.

Methods Meningitis incidence The study took place in the Région des Savanes, in northern Togo, which is divided into four préfectures (districts)—Tone, Oti, Kpendjal and Tandjoaré (Figure 1). Clinic logbooks of health-care facilities of the study area were reviewed for cases of meningitis which occurred between 1 August 1990 and 31 July 1997. A case was defined as a person

diagnosed with meningitis as written in the clinic logbook by the consulting physician or nurse. The date of presentation to the health centre was considered to be the date of onset. Age, sex and residence were recorded for each case. Population figures were determined for each year 1990–1997 from a 1996 census conducted in the Région des Savanes (source: Direction Régionale de la Santé des Savanes) and retrospectively adjusted for a 3% annual growth rate.10 Meningitis incidence rates were calculated for each district by dividing the number of new cases occurring in the area per unit of time (week or year) by the annually adjusted population of the area. For this study, the year span was defined as 1 August through 31 July of the following year in order to include a complete meningitis season (December–May).

Figure 1 Districts of Tone, Kpendjal, Tandjoaré and Oti, Région des Savanes, northern Togo

MENINGOCOCCAL MENINGITIS IN TOGO

Definition of an epidemic An epidemic year was retrospectively defined in a district by a yearly meningitis incidence of >100 cases per 100 000 inhabitants.3 The peak of the epidemic was defined as the week with the maximum weekly incidence. Threshold analyses were repeated for other hypothetical definitions of an epidemic (70, 80 and 90 cases/100 000 inhabitants/year).

935

initiation of vaccination on the number of cases which may have been prevented was studied with a sensitivity analysis allowing the time to achieve vaccination objectives to vary from 2 to 8 weeks. Vaccine coverage was either the actual coverage achieved by the mass campaign in Togo or a theoretical coverage of 85% of the entire population.

Results Threshold study

Descriptive epidemiology

Weekly meningitis incidences between 2 and 25 cases/100 000 inhabitants were investigated as potential thresholds for predicting the emergence of an epidemic during the corresponding year. The WHO recommended threshold of 15 cases/100 000 inhabitants averaged over 2 consecutive weeks was also tested. Estimates of the sensitivity, specificity, positive predictive value and negative predictive value of these thresholds were calculated by determining which district crossed, at least once, the given threshold and experienced an epidemic during the same year. The sensitivity is the probability that the threshold was crossed during an epidemic year; the specificity is the probability that the threshold was not crossed during a non-epidemic year; the positive predictive value is the probability that there was an epidemic if the threshold had been crossed at least once; the negative predictive value is the probability that there was no epidemic if the threshold had never been crossed during a given year. The time available for intervention was also studied for the various thresholds. It was calculated as the number of weeks elapsed between the first week the threshold was crossed and the week of the epidemic peak. Thresholds were analysed for their ability to predict the occurrence of an epidemic within the year during which they were crossed.

The 1997 population of the study area was estimated to be 550 524 people (82 708 in Tandjoaré, 104 511 in Kpendjal, 127 567 in Oti and 235 738 in Tone). Mean population density is 63 inhabitants/km2. From August 1990 through July 1997, 5562 meningitis cases were counted in the clinic logbooks of the health facilities, among which 4952 (89%) reported a residence in the Région des Savanes. The weekly meningitis incidence in the region is presented in Figure 2. Annual meningitis incidence for each district is presented in Table 1. From 1990 to 1995, no epidemic was noted. In non-epidemic years, the incidence ranged from 4 to 86 cases/ 100 000 inhabitants/year with a mean of 35 and a median of 27. An epidemic occurred in early 1996 in Tone district with an annual incidence of 168 cases/100 000 inhabitants. During this epidemic, the maximum weekly incidence was 14.9 cases/ 100 000 inhabitants. In early 1997, another epidemic occurred in all four districts of the Région des Savanes, accounting for 3590 cases between 2 December 1996 and 1 June 1997. Overall annual incidence was 652 cases/100 000 inhabitants, varying from 154 in Oti district to 1135 in Tone. The epidemic threshold of 15 cases/100 000 inhabitants averaged over 2 weeks was exceeded in Tone during the first week of 1997 and the alert threshold of 5 cases/100 000 inhabitants/week was exceeded in Kpendjal during week 2, in Oti during week 5 and in Tandjoaré during week 6.

Impact study The number of cases which may have been prevented by a mass vaccination campaign during the 1996–1997 epidemic in the Région des Savanes was estimated using a method described by Pinner et al.11 The number of meningitis cases which would have occurred in the absence of a vaccination campaign was estimated on the basis of the weekly meningitis incidence, the meningitis vaccination coverage obtained during this epidemic and the vaccine effectiveness. The number of cases potentially prevented was then estimated through different scenarios, depending on the threshold and vaccination strategy used. This method assumes that vaccination does not interfere with meningococcal carriage or transmission,7,11 an assumption that seems valid for group A meningococcal polysaccharide vaccines in African populations.12 Vaccine coverage for the study area was assumed to be 0% prior to the 1996–1997 epidemic and was calculated after the epidemic by dividing the number of doses administered to individuals in each district during this epidemic by the 1997 population of the district. Vaccine coverage was stratified according to age (0–14 and >15 years). Vaccine efficacy was assumed to be 85%13 and time for seroconversion was set at one week after vaccination. After the threshold was crossed, a theoretical time-frame of a week was initially allocated for gathering and analysing data, ordering vaccines, organizing the campaigns and achieving vaccination objectives. The influence of delay in the

Performance of epidemic thresholds Crossing a threshold of 15 cases/100 000 inhabitants/week averaged over 2 consecutive weeks yielded both specificity and positive predictive value of 100% for the prediction of an epidemic (ù100 cases per 100 000 inhabitants) in a district during a given year. Sensitivity was 80% and negative predictive value was 96% (Table 2). Results were exactly the same when testing a threshold of 15 cases/100 000 inhabitants/week calculated over one week only. Thresholds between 7 and 10 cases/100 000 inhabitants/ week calculated over one week in a district resulted in sensitivity, specificity, positive and negative predictive values of 100%. Thresholds below 7 cases/100 000 inhabitants/week resulted in sensitivity and negative predictive value of 100%, but low to medium specificity and positive predictive value, whereas thresholds over 15 cases/100 000 inhabitants/week resulted in lower sensitivity and negative predictive value, but specificity and positive predictive value of 100%. Results were identical for an epidemic defined as >90 cases/ 100 000 inhabitants/year. Due to the high endemic rates of meningitis in Tone district in all years (Table 1), use of lower annual incidences to define an epidemic year resulted in lower sensitivity for all thresholds tested, as years are declared ‘epidemic’ in spite of low weekly incidences.

936

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Figure 2 Meningitis weekly incidencea in the Région des Savanes (a), and in each district (b, c, d, e), northern Togo, 1990–1997 Source: clinic logbooks from the Région des Savanes. a In the four districts (b, c, d, e) the incidence is truncated at 20 cases/100 000 inhabitants/week.

Time available for intervention The mean time elapsed between the threshold and the peak of the epidemic ranged from 2.3 weeks, after surpassing 25 cases/ 100 000 inhabitants/week, to a maximum of 8.6 weeks with a

threshold of 2 cases/100 000 inhabitants in one week. The threshold of 10 cases/100 000 inhabitants/week left 4.2 weeks for intervention before the epidemic peak whereas 7 cases/ 100 000 inhabitants/week left 5.4 weeks on average (Table 3).

MENINGOCOCCAL MENINGITIS IN TOGO Table 1 Annual meningitis incidence, Région des Savanes, northern Togo, 1990–1997

Table 3 Time elapsed between threshold and peak during the 1996–1997 meningitis epidemic, Région des Savanes, northern Togo

Meningitis incidence per district (no. of meningitis cases/100 000 inhabitants/year) Year

Kpendjal

Oti

937

Tandjoaré

Tone

Epidemic threshold (cases/100 000 inhabitants/week)

Average time between threshold and epidemic peak (weeks)

95% CIa

1990–1991

15

8

26

58

2

8.6

(5.0, 12.2)

1991–1992

22

7

29

70

5

5.4

(3.3, 7.5)

1992–1993

23

4

50

72

7

5.4

(3.3, 7.5)

1993–1994

27

9

28

81

10

4.2

(1.2, 7.2)

1994–1995

23

17

51

86

15

4.0

(0.1, 7.9)

1995–1996

33

20

54

168

15b

3.0

(–0.9, 6.9)

1996–1997

347

154

197

1135

20

2.5

(–1.3, 6.3)

25

2.3

(–1.7, 6.2)

Source: clinic logbooks from the Région des Savanes.

a Confidence interval calculated with the t distribution. b Averaged over 2 consecutive weeks.

Table 2 Test characteristics of weekly meningitis incidence as predictor of meningitis epidemics, northern Togo, 1990–1997 Epidemic threshold (cases/100 000 inhabitants/week) 2

No. of times the threshold was crossed in a districta

Sensitivity (%) (95% CI)

24

5

15b 20 25

Negative predictive value (%) (95% CI)

100

17

21

100

(5.7, 39.5)

(7.9, 42.7)

(39.6, 100.0)

100

87

62

100

(46.3, 100.0)

(65.3, 96.6)

(25.9, 89.8)

(80.0, 100.0)

5

10

Positive predictive value (%) (95% CI)

(46.3, 100.0) 8

7

Specificity (%) (95% CI)

100

100

100

100

(46.3, 100.0)

(82.2, 100.0)

(46.3, 100.0)

(82.2, 100.0)

5

100

100

100

100

(46.3, 100.0)

(82.2, 100.0)

(46.3, 100.0)

(82.2, 100.0)

4

80

100

100

96

(29.9, 98.9)

(82.2, 100.0)

(39.6, 100.0)

(76.9, 99.8)

4

80

100

100

96

(29.9, 98.9)

(82.2, 100.0)

(39.6, 100.0)

(76.9, 99.8)

3

60

100

100

92

(17.0, 92.7)

(82.2, 100.0)

(31.0, 100.0)

(72.5, 98.6)

a 28 units of observation: 4 districts ´ 7 years of observation per district. b Results are the same for an incidence of 15 cases/100 000 inhabitants/week during one week or averaged over 2 consecutive weeks, and when the 2-week

average is used for the first district in combination with the alert threshold of 5 cases/100 000 inhabitants/week for the remaining districts.

Impact of the 1997 mass vaccination campaign A serogroup A meningococcal meningitis epidemic occurred in the Région des Savanes8 from December 1996 through May 1997, accounting for 3590 cases. Mass vaccination campaigns were organized and a total of 346 469 doses of vaccine were administered in the four districts in February 1997.8 Vaccine coverage in the four districts ranged from 58% to 69%, yielding a weighted average administrative coverage for the region of 63%. The 1996–1997 epidemic curve and a projection of the curve, had there been no vaccination during this epidemic, are presented in Figure 3. According to the model, 49% (3419) of the projected cases were prevented by the vaccination campaign. If the WHO recommendations had been strictly followed (vaccination following a 2-week average of 15 cases/100 000 inhabitants/week for the first district and 5 cases/100 000 inhabitants/week for subsequent districts), 65% (4549) of the cases could have been prevented by a vaccination campaign covering 85% of the population and completed by the end of

the second week of 1997 in Tone, the third week in Kpendjal, the sixth week in Oti and the seventh week in Tandjoaré district. Using a single threshold of 7 cases/100 000 inhabitants/week in a district, 66% (4657) of the cases could have been prevented if 85% vaccine coverage had been achieved by the end of the first week of 1997 in Tone and Kpendjal, the sixth week in Oti and the seventh week in Tandjoaré district. The effect of time needed to achieve vaccination on the proportion of cases prevented is presented in Figure 4. For each week of delay, 3% to 8% fewer cases could be prevented and for each drop in vaccine coverage of 10%, approximately 5% fewer cases are prevented. In order to prevent a minimum of 60% of cases with an 85% vaccine coverage in the Région des Savanes, vaccination should be achieved within 3.5 weeks after the threshold of 7 cases/100 000 inhabitants/week is crossed or within 2.5 weeks with WHO guidelines. To achieve the same impact, the lower threshold would have provided one extra week to plan a response.

938

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Figure 3 Model of the impact of mass vaccination on the 1996–1997 meningitis epidemic in the Région des Savanes, northern Togo ——– actual meningitis cases recorded (source: Direction Régionale de la Santé, Région des Savanes) — — projected number of cases in the absence of vaccination ------ projected number of cases if 85% vaccine coverage achieved after passing threshold of 7 cases/100 000 inhabitants/week period when mass vaccination campaign was implemented

Figure 4 Potential impact of mass vaccination according to threshold strategy and lag time before vaccination. Meningitis epidemic, northern Togo, 1996–1997 Proportion of cases potentially prevented according to the number of weeks elapsed, if vaccination had been initiated after the incidence in a district reached or exceeded: ▲ 7 cases/100 000 inhabitants/week; vaccination coverage 85%

--

◆ 7 cases/100 000 inhabitants/week; vaccination coverage 63% (achieved in Togo in 1996–1997) 15 cases/100 000 inhabitants/week averaged over 2 weeks; vaccination coverage 85%

--

15 cases/100 000 inhabitants/week averaged over 2 weeks; vaccination coverage 63%

Discussion In the management of meningococcal meningitis epidemics, time is of the essence. It is imperative to detect an impending epidemic and institute control measures as early as possible if large numbers of cases and deaths are to be prevented. In Togo, the 1997 meningitis epidemic spread rapidly. This study shows that, following detection of an epidemic, the time available for intervention and the vaccination coverage achieved are critical to the success of the reactive vaccination strategy recommended for Africa by the WHO. Our findings confirm the validity of the threshold approach but questions are raised as to whether the currently recommended threshold is the most appropriate one. Meningitis incidence thresholds of 7 and 10 cases/100 000 inhabitants/week both achieved 100% sensitivity and specificity, accurately predicting all epidemics in the four districts of northern Togo over the period 1990–1997. In contrast, the threshold of 15 cases/100 000 inhabitants/week in a district, averaged over 2 consecutive weeks, predicted meningitis epidemics with excellent specificity and positive predictive value but failed to detect one epidemic in Tone district in 1995–1996 (population .225 000), resulting in a lower sensitivity and negative predictive value, as previously shown.4 This may serve to illustrate the lower sensitivity of this threshold in larger populations. In this study, we chose to work with the actual district population figures, rather than dividing them arbitrarily into smaller units, in order to test thresholds in the real-life situations faced by district health officers. Lower thresholds also allowed earlier detection of epidemics. In this study, thresholds of 7 and 10 cases/100 000 inhabitants/ week offer on average 2.4 and 1.2 weeks, respectively, more time to achieve vaccination coverage than current recommendations. The time elapsed between epidemic threshold and

peak is critical in a strategy based on reactive mass vaccination.6 The reactive strategy must allow time to collect and analyse field data (usually through a surveillance system), decide on mass vaccination, order the vaccines, and organize and implement the vaccination campaign. Our baseline model allowed just one week for these activities but field experience shows that it takes at least 2–3 weeks to mount such a response. In our study, each week of delay in completing the vaccination campaign resulted in a 3–8% decrease in the number of cases prevented, similar to what was found using meningitis data from Ghana.7 Therefore any strategy that increases the time available for a response should be seriously considered. Woods et al. have demonstrated that knowledge of meningitis epidemics in neighbouring countries could provide an earlier alert.7 Although the Région des Savanes borders Bénin, Burkina Faso and Ghana, data from these countries were not available to us at the time of the study. The potential impact of reactive mass vaccination is inherently limited by vaccination coverage achieved, as this study illustrates, and by the efficacy of the polysaccharide vaccine.14–16 In the model presented here, even with a highly sensitive threshold and ‘ideal’ conditions where 85% of the population can be vaccinated within one week of crossing the threshold, a maximum of 65% of the meningitis cases could have been prevented during the 1996–1997 epidemic in Togo. In reality, an estimated 49% or 3419 cases of meningitis were prevented, a figure which still fully justifies the vaccination campaign. Figure 4 shows that, for a given vaccination coverage, the use of the lower threshold (7 cases/100 000 inhabitants/week) allows one more week for intervention than the currently recommended threshold to achieve the same impact (proportion of cases

MENINGOCOCCAL MENINGITIS IN TOGO

prevented). Therefore, the additional time provided by a lower threshold can make a real difference in the potential impact of a vaccination campaign. These findings illustrate the difference between a theoretical approach based only on sensitivity and specificity and a pragmatic approach that accounts for the time necessary to implement epidemic control measures. Although our results suggest that strict adherence to WHO guidelines would allow the prevention of a similar proportion of meningitis cases as use of a lower threshold, the delay engendered by waiting for that higher (more specific and less sensitive) threshold may actually set back the whole reponse to the epidemic, thus reducing the number of cases prevented. The results of our study are based on the number of cases recorded in the clinic logbooks at the peripheral level, giving the best available representation of the actual number of cases. In an operational context, however, decisions must usually be made using information from surveillance systems. This will inevitably be less efficient, as the timeliness and sensitivity of case detection can vary widely.4 If reporting from health facilities is incomplete or delayed, thresholds will be apparently crossed later in the epidemic and interventions based on surveillance data will be delayed. This underlines the need for awareness of surveillance system performance, for improving the quality and timeliness of meningitis surveillance and for undertaking field investigation whenever an outbreak is suspected. Incomplete or delayed reporting of cases to district surveillance officers is a further argument for using as sensitive a threshold as possible for detection of meningitis outbreaks. This study is limited by the small area studied. Despite the large amount of data collected, the number of observations available for the threshold study was 28 district-years (four districts over 7 years), resulting in wide confidence intervals on the estimates of performance characteristics. The small data set precluded the comparison of performance characteristics for epidemic and non-epidemic years. The study area is located at the southern edge of the meningitis belt and may not be representative of the entire region. As characteristics of meningitis epidemics may vary from one area to another, further studies are needed in other sub-Saharan Africa countries. In addition, study of thresholds at a sub-district level may provide further useful information. Recommendations regarding detection of meningitis outbreaks in small populations using absolute numbers of cases (e.g. doubling of number of cases over 3 weeks or comparison with previous years)3 could be tested and validated. At the present time, the reactive vaccination strategy is the only one possible for most regions of the meningitis belt due to limited resources.14–16 Preventive mass vaccination with the polysaccharide vaccine has been discussed as an alternative17 but this approach would suffer from the same flaws as the reactive strategy, i.e. the absence of herd immunity12 and a moderate vaccine efficacy. In addition, the short duration of protection offered by the polysaccharide vaccine13 and the unpredictability of meningitis epidemics has so far limited its potential usefulness. Where resources permit, a preventive strategy could be considered for areas with very high endemic rates of disease. Preventive vaccination would be a suitable alternative with a vaccine inducing long-term immunity in children and herd immunity in the populations vaccinated. Meningococcal

939

conjugate vaccines currently under development seem to have these characteristics and represent the best hope for the improvement of meningitis epidemic control in Africa, provided that they are affordable for countries with limited resources. In the meantime, efforts must continue to improve the impact of mass vaccination in the face of devastating meningitis epidemics. Avenues to pursue include improving surveillance systems and encouraging international collaboration. Use of lower, more sensitive, epidemic thresholds should be preferred over specific ones to enable quicker action and the time required to mount an appropriate and comprehensive response should be accounted for.18 New recommendations concerning meningitis epidemic thresholds have been drafted, taking into consideration these and other important elements.19

Acknowledgements This study has been funded by Médecins Sans Frontières. We wish to thank Dr Kombate, Dr Aleki (Togo Ministry of Health), the health workers of the Région des Savanes and Médecins Sans Frontières for their collaboration. We also wish to thank Dr Brian Greenwood for reviewing the manuscript.

References 1 Lapeyssonie L. La méningite cérébro-spinale en Afrique. Bull World

Health Organ 1963;28(Suppl.):1–114. 2 Moore PS. Meningococcal meningitis in sub-Saharan Africa: a model

for the epidemic process. Clin Infect Dis 1992;14:515–25. 3 World Health Organization. Control of Epidemic Meningococcal Disease

—WHO Practical Guidelines. Lyon: Ed. Fondation Marcel Mérieux, 1995. 4 Moore PS, Plikaytis BD, Bolan GA et al. Detection of meningitis

epidemics in Africa: a population-based analysis. Int J Epidemiol 1992; 21:155–62. 5 Varaine F, Caugant DA, Riou J-Y et al. Meningitis outbreaks and

vaccination strategy. Trans R Soc Trop Med Hyg 1997;91:3–7. 6 Veeken H, Ritmeijer K, Hausman B. Priority during a meningitis

epidemic: vaccination or treatment? Bull World Health Organ 1998; 76:135–41. 7 Woods CW, Armstrong G, Sackey SO et al. Emergency vaccination

against epidemic meningitis in Ghana: implications for the control of meningococcal disease in West Africa. Lancet 2000;355:30–33. 8 Aplogan A, Batchassi E, Yakoua Y et al. Une épidémie de méningite

à méningocoque dans la région des Savanes au Togo en 1997: investigation et stratégies de contrôle. Santé 1997;7:384–90. 9 WHO. Meningitis in the WHO African Region—Update, January-

April 1997. Wkly Epidemiol Rec 1997;72:131. 10 Palmore JA, Gardner RW. Measuring Mortality, Fertility and Natural

Increase. A Self Teaching Guide to Elementary Measures, 5th Edn. Honolulu: East West Center Ed., 1994. 11 Pinner RW, Onyango F, Perkins BA et al. Epidemic meningococcal

disease in Nairobi, Kenya, 1989. The Kenya/Centers for Disease Control (CDC) meningitis study group. J Infect Dis 1992;166:359–64. 12 Hassan-King MKA, Wall RA, Greenwood BM. Meningococcal

carriage, meningococcal disease and vaccination. J Infect 1988;16: 55–59. 13 Reingold AL, Hightower AW, Bolan GA et al. Age-specific differences

in duration of clinical protection after vaccination with meningococcal polysaccharide A vaccine. Lancet 1985;2:114–18.

940

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

14 Kaninda AV, Varaine F, Henkens M, Paquet C. Meningococcal vaccine

in sub-Saharan Africa [comment]. Lancet 1997;350:1708. 15 Perkins BA, Broome C, Rosenstein NE, Schuchat A, Reingold AL.

Meningococcal vaccine in sub-Saharan Africa [comment]. Lancet 1997;350:1708. 16 Wenger J, Tikhomirov E, Barakamfitiye D, Bele O, Heymann DL.

Meningococcal vaccine in sub-Saharan Africa [comment]. Lancet 1997;350:1709.

17 Robbins JB, Towne DW, Gotschlich EC, Schneerson R. ‘Love’s labours

lost’: failure to implement mass vaccination against group A meningococcal meningitis in sub-Saharan Africa. Lancet 1997;350:880–82. 18 Lewis R, Varaine F, Belanger F, Nathan N, Diarra L. Control of

meningococcal disease in West Africa (Letter). Lancet 2000;355: 1185–86. 19 Consensus meeting on epidemic thresholds for meningococcal

meningitis in Africa. Wkly Epidemiol Rec 2000;(In Press).