Assessment of seasonal and climatic effects on the incidence and species composition of malaria by using GIS methods

1
Assessment of seasonal and climatic effects on
the incidence and species composition of malaria
by using GIS methods

• Ali-Akbar Haghdoost
• Neal Alexander (supervisor)

2
Main objectives

• Assessment of the feasibility of an early warning
  system based on ground climate and remote sensing
  data
• Assessment of the interaction between Plasmodium
  spp from different points of view meta-analysis,
  modelling, and extended analysis of a large
  epidemiological dataset

3
Feasibility of the early warning (1)
The fitted values of models based on seasonality,
time trend and meteorological variables
classified by species, observed numbers (dashes)
and model estimated number (solid line)
4
Feasibility of the early warning (2)

• Main findings
• Ground climate data explained around 80 of P.
  vivax and 85 of P. falciparum variations one
  month ahead
• Comparing to the extrapolation of data from
  previous month, ground climate data improve the
  accuracies around 10 but remote sensing data
  does not improve
• The ground climate data are freely available in
  the filed therefore, it was concluded that the
  models based on ground climate data are feasible.

5
What is the interaction?

• The difference between the observed number of
  mixed infections in blood slides and the expected
  number if infection with one species is
  independent of infection with other species

6
Why the interaction is important?

• To know more about the pathogenesis of Plasmodium
  spp
• To know more about the immunity mechanisms
  against Plasmodium spp
• To estimate the impact of vaccine against one
  species on the other species

7
Positive interaction

• Similarity in transmission routes
• Higher susceptibility of a subgroup of people

8
Negative interaction

• Suppression
• Cross immunity
• Differences in the biology of Plasmodium spp
• Environmental factors
• Missed mixed infections in blood slides

9
Background
Howard (2001) showed that the logarithm of odds
ratio between P. falciparum and P. vivax changed
in a wide rage from 5.08 (in Bangladesh) to 2.56
(in Sierra Leone). He found that in Asian
countries, the associations were largely
negative however, positive associations were
seen in Tanzania, Papua New Guinea and USA.
10
Questions

• What is the overall association between species?
• How we can explain the differences between study
  findings?

11
Sections

• Meta-analysis
• To quantify the interaction between P. falciparum
  and P. vivax
• To assess the source of the heterogeneities
• Modelling the heterogeneity effect
• To measure the association between Plasmodium spp
  in the Garki region of Sudan Savanna of west
  Africa

12
Meta-analysis (1)

• Database number of citations
• Medline 1966-2001 395
• Embase 1980-2001 77
• CAB-Health 1973-2001 455
• Merged database (excluding repeated citation) 829

13
Meta-analysis (2)

• Reviewing abstracts (829)
• Non eligible papers 657 (72.2)
• Eligible papers 104 (12.5)
• Uncertain 68 ( 8.3)

• Reviewing full texts of papers (172)
• Eligible for meta-analysis 62 (36.1)
• Non eligible for meta-analysis 108 (63.3)
• Was not available (from China) 1 ( 0.6)

14
Meta-analysis (3)
Number of studies Percentage
Continent Asia Africa America 52 4 6 83.9 6.4 9.7
Spatial span Villages District Province or larger 36 16 10 58.1 25.8 16.1
Temporal span Month Season Year Greater than one year 26 12 5 19 41.9 19.3 8.1 30.7
Age group Children All age groups or adults 5 57 8.1 91.9
Samples Febrile Normal 26 36 41.9 58.1
15
Meta-analysis (4)
Minimum OR0.02 Maximum OR 10.9 Summary OR0.6
(0.49-0.79) Number of studies with ORlt141 Number
of studies with ORgt132
16
Meta-analysis (5)
Subgroup (number of studies) Odds ratio (95CI) Subgroup (number of studies) Odds ratio (95CI)
Continent Asia (52) South America (6) Africa(4) 0.62(0.46-0.83) 0.21(0.16-0.26) 1.76(0.47-6.6) Temporal Span Month(26) Season(12) Year or longer(24) 0.81(0.56-1.17) 0.97(0.52-1.79) 0.39(0.26-0.6)
Age group Children(5) Mixed(57) 1.38(0.31-6.08) 0.56(0.43-0.75) P. falciparum risk () lt10(23) 10-14.99(10) 15(29) 1.06(0.54-2.1) 0.75(0.42-1.35) 0.4(0.28-0.57)
Subjects Normal(36) Febrile(26) 0.9(0.65-1.24) 0.35(0.21-0.58) P. vivax risk () lt5(27) 5-9.99(18) 10(17) 1.43(0.98-2.1) 0.49(0.32-0.75) 0.25(0.13-0.5)
Spatial span A few villages(36) District(16) Larger than a district(10) 0.5(0.33-0.75) 0.99(0.591-1.63) 0.49(0.3-0.82) Both species risk () lt15(18) 15-29.99(22) 30(22) 2.51(1.66-3.8) 0.5(0.36-0.7) 0.32(0.22-0.47)
17
Meta-analysis (6)
The results of meta-reg analysis
Subgroup Tau square
Model 1 no explanatory variable 0.91
Model2 explanatory variables were age group, subjects (febrile or normal), spatial and temporal span of studies and continent 1.18
Model3 the only explanatory variable was the frequencies of all species (all Plasmodium species considered together) and temporal span of studies 0.72
a measure of between studies heterogeneity
18
Meta-analysis (7)

• Main findings
• The overall OR (between P. vivax and P.
  falciparum) was less than 1
• There were negative associations (weaker) between
  the prevalence of species and the overall OR
• There was a negative association between the
  temporal span of studies and the overall OR

19
Modelling (1)

• Positive associations between species mean that a
  subgroup of people, in terms of time or space,
  has higher infection risks for all species, i.e.,
  heterogeneity in infection risks within the
  population.
• Therefore, infection risk could be considered as
  a confounder.

20
Modelling (2)

• Main question
• Can the confounding effect of the heterogeneity
  in infection risks explain OR as large as 11 by
  its own?

21
Modelling (3)

• Model specification
• Population has been divided into low and high
  risk strata
• The OR between species in each stratum was 1
• The risk ratio of infection with species i in
  high risk versus low risk stratum (k1) was varied
  from 1 up to its maximum possible values
• The ratio of the populations in low and high risk
  strata (m) was varied in a wide range (0.2-5)
• The prevalence of species were varied in a wide
  range from 0.05 to 0.8

22
Modelling (4)

• The impact of ki on the overall OR in the whole
  population

23
Modelling (5)
The impact of m on the overall OR in the whole
population
24
Modelling (6)

• Greatest ORs were observed when the prevalence of
  species were equal
• By increasing the prevalence of species in low
  risk stratum, the overall OR was decreased

25
Modelling (7)

• Conclusion
• Just heterogeneity in infection risk can explain
  an OR as large as 11

26
Garki (1)

• The Garki project was one of the largest
  epidemiological studies on malaria, with data
  comprised from more than 12,000 people in 23
  surveys.
• It was conducted in a highly endemic area in
  northern Nigeria from 1969 to 1976 by
  co-operation between the World Health
  Organisation (WHO) and the Nigerian government.

27
Garki (2)

• The published results of the Garki data had not
  thoroughly explored the interactions between
  Plasmodium species, and that too had only
  approached this issue cross-sectionally using
  very simple methods.

28
Garki (3)

• Objectives
• To measure the associations between Plasmodium
  spp cross-sectionally and longitudinally and
  assess the effects of
• repeated infections (i.e., within subject
  clustering)
• Age
• spatial and temporal distribution of individual
  species

29
Garki (4)

• Cross-sectional analysis the presence of
  P. falciparum in each survey was considered as a
  risk factor for the presence of the other species
  in the same survey
• Longitudinal analysis the presence of one
  species in each survey was considered as a risk
  factor for the presence of the other species in
  the following survey

30
Garki (5)
Frequencies of single and mixed Plasmodium spp in
118,346 blood slides
31
Garki (6)
Annual variation of Plasmodium spp prevalence,
based on 6 years data
32
Garki (7)

• Multi-level models showed that the risk of
  P.falciparum had the largest within
  person-variation, and also within and between
  village variations

33
Garki (8)
The risk of infection with Plasmodium spp
classified by age
Age group lt4 months Number () 4-7 months Number () 8-12 months Number () 1-9.9 year Number () 10 year Number ()
P. falciparum OR (95 CI) 0.75 (0.64-0.9) 2.52 (2.2-2.9) 3.9 (3.41-4.56) 11.68 (11.13-12.27) 1 -
P. falciparum OR (95 CI) OR for the whole first year 2.1 (1.8-2.4) OR for the whole first year 2.1 (1.8-2.4) OR for the whole first year 2.1 (1.8-2.4) 11.68 (11.13-12.27) 1 -
P. malariae OR (95 CI) 0.56 (0.39-0.8) 1.31 (1.04-1.65) 1.95 (1.6-2.37) 5.9 (5.63-6.2) 1 -
P. malariae OR (95 CI) OR for the whole first year 1.3 (1.1-1.5) OR for the whole first year 1.3 (1.1-1.5) OR for the whole first year 1.3 (1.1-1.5) 5.9 (5.63-6.2) 1 -
P. ovale OR (95 CI) 1 0.47-2.12 2.59 (1.68-4) 2.2 (1.38-3.49) 4.2 (3.72-4.75) 1 -
P. ovale OR (95 CI) OR for the whole first year 4.2 (3.6-5.0) OR for the whole first year 4.2 (3.6-5.0) OR for the whole first year 4.2 (3.6-5.0) 4.2 (3.72-4.75) 1 -
34
Garki (9)
The associations of P. falciparum (as risk
factor) with other species adjusted for
intra-person clustering effect in cross-sectional
analysis
P. malariae OR (95 CI) P. ovale OR (95 CI)
All subjects Age (year) lt1 1-9 gt10 Season Dry and cool Dry and hot Wet Rho0.34 3.64(3.4-3.9) 6.25(2.63-14.82) 2.32(1.70-3.16) 3.97(3.24-4.85) 4.02(3.7-4.35) 6.32(5.48-7.29) 3.58(3.3-3.9) Rho0.25 5.1 (4.33-6.0) 6.26(2.64-14.83) 2.19(1.59-3.03) 3.95(3.23-4.84) 5.53(4.6-6.68) 3.94(2.18-7.12) 3.76(2.78-5.07)
35
Garki (10)
The associations between P. falciparum in a
former survey with species in the latter survey,
adjusted for intra-person clustering effect
P. falciparum OR (95 CI) P. malariae OR (95 CI) P. ovale OR (95 CI)
All subjects Age (year) lt1 1-9 gt10 Season Dry and cool Dry and hot Wet Rho0.73 1.9(1.9-2) 9.3(7.6-11.5) 3.1(2.7-3.6) 1.5(1.4-1.6) 4.3(3.9-4.6) 9.8(9-10.6) 4.3(3.9-4.6) Rho0.44 2.7(2.5-2.9) 11.6(6.8-20) 2(1.7-2.3) 1.8(1.7-2) 4.1(3.7-4.5) 5.5(4.8-6.2) 3.6(3.2-4.1) Rho0.34 3.6(3-4.4) 6.9(2.7-17.7) 2.0(1.4-2.7) 2.7(2.2-3.4) 2.6(2-3.5) 4(2.8-5.7) 4.7(3.5-6.4)
36
Garki (11)

• Why the ORs were greater in infants?
• Heterogeneity in infection risk (as the source of
  positive associations depends on
• The heterogeneity in exposure to mosquitoes
• The heterogeneity in acquired protective immunity
• It is reasonable to assume a positive association
  between the strength of acquired immunity and
  exposure to mosquitoes in adults. Therefore,
  these two factors somehow decreased their impacts
  on the heterogeneity in infection risk in adults.

37
Garki (12)

• The relationship between P. falciparum density
  and the risk of other species based on
  cross-sectional data

Density 0 1-50 gt50
P. malariae 1 4.05 8.66
P. Ovale 1 4.05 8.73
number of positive filed in 200 examined fields
38
Garki (13)
The association between Plasmodium spp adjusted
for intra-person clustering effect in
cross-sectional analysis
Latter survey Latter survey Latter survey Latter survey Latter survey Latter survey
P. falciparum P. falciparum P. malariae P. malariae P. ovale P. ovale
Former Survey P. falciparum P. falciparum P. malariae P. malariae P. ovale P. ovale
P. falciparum
OR(95 CI) Rho 1.9(1.9-2) 0.73 1.9(1.9-2) 0.73 2.63(2.5-2.9) 0.44 2.63(2.5-2.9) 0.44 3.6(3-4.4) 0.34 3.6(3-4.4) 0.34
P. malariae
OR(95 CI) Rho 1.7(1.5-2) 0.22 1.7(1.5-2) 0.22 2.7(2.5-2.9) 0.33 2.7(2.5-2.9) 0.33 2.6(2.2-3.0) 0.03 2.6(2.2-3.0) 0.03
P. ovale
OR(95 CI) Rho 1.9(1.3-2.8) 0.22 1.9(1.3-2.8) 0.22 2.8(2.3-3.4) 0.29 2.8(2.3-3.4) 0.29 5.3(3.9-7.2) 0.17 5.3(3.9-7.2) 0.17
39
Garki (14)
Daily conversion rates in
Estimated daily clearance and acquisition rates
of P. malariae and P. ovale classified by the
presence of P. falciparum in the former survey
logarithmic scale
1
0.1
0.01
0.001
0.0001
gt1
1-9
gt10
gt1
1-9
gt10
age group (year)
Plasmodium malariae
Plasmodium ovale
pf negative acquisition rate
pf negative clearance rate
pf positive acquisition rate
pf positive clearance rate
40
Garki (15) conclusion

• Cross-sectional analysis
• Suppression decreases the association between
  species
• Longitudinal analysis
• Cross immunity, suppression and changing ones
  behaviour (such as the exposure risk to
  mosquitoes) after contracting the first infection
  decrease the association between species

41
Garki (16) conclusion

• P. falciparum suppress other species particularly
  P. malaria
• The suppression is not just due to the
  competition for host cells or nutrients. It is
  most probably due to heterologous immunity
• Low level of acquired immunity suppresses the
  other species stronger immunity increases the
  clearance rate, and very strong immunity
  decreases the acquisition rate as well.

42
Summary (1)

• A very wide range of associations between
  Plasmodium spp was observed in meta-analysis
  which was partly explained by the prevalence of
  species and the temporal span of studies
• The heterogeneity in infection risk (due to
  heterogeneity in exposure risk or immunity) can
  explain the observed high ORs in meta-analysis

43
Summary (2)

• The ORs in longitudinal analysis of the Garki
  data was smaller than those in cross-sectional
  analysis
• The ORs in infants were less than others which
  can be explained by the heterogeneity in
  infection risk theory
• P. falciparum suppresses other species, probably
  via immunological pathways
• People obtained protective immunity after many
  infections therefore, the frequency of species
  had direct association with the variation of
  infection risk within and between subjects and
  villages

44
Time for your comments

• Thanks for you kind attention