1. Epidemiology and Global Health
Download icon

Infection-exposure in infancy is associated with reduced allergy-related disease in later childhood in a Ugandan cohort

  1. Lawrence Lubyayi  Is a corresponding author
  2. Harriet Mpairwe
  3. Gyaviira Nkurunungi
  4. Swaib A Lule
  5. Angela Nalwoga
  6. Emily L Webb
  7. Jonathan Levin
  8. Alison M Elliott
  1. Immuno-modulation and Vaccines Programme, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine Uganda Research Unit, Uganda
  2. Division of Epidemiology and Biostatistics, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa, South Africa
  3. Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, United Kingdom
  4. Department of Infection Biology, London School of Hygiene and Tropical Medicine, United Kingdom
  5. Institute for Global Health, University College London, United Kingdom
  6. MRC International Statistics and Epidemiology Group, Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, United Kingdom
  7. Department of Clinical Research, London School of Hygiene and Tropical Medicine, United Kingdom
Research Article
  • Cited 0
  • Views 272
  • Annotations
Cite this article as: eLife 2021;10:e66022 doi: 10.7554/eLife.66022

Abstract

Background:

Lack of early infection-exposure has been associated with increased allergy-related disease (ARD) susceptibility. In tropical Africa, little is known about which infections contribute to development of ARDs, and at which time.

Methods:

We used latent class analysis to characterise the early infection-exposure of participants in a Ugandan birth cohort and assessed ARDs in later childhood.

Results:

Of 2345 live births, 2115 children (90%) had data on infections within the first year of life while 1179 (50%) had outcome data at 9 years. We identified two latent classes of children based on first-year infection-exposure. Class 1 (32% membership), characterised by higher probabilities for malaria (80%), diarrhoea (76%), and lower respiratory tract infections (LRTI) (22%), was associated with lower prevalence of wheeze, eczema, rhinitis, and Dermatophagoides skin prick test (SPT) positivity at 9 years. Based on 5-year cumulative infection experience, class 1 (31% membership), characterised by higher probabilities for helminths (92%), malaria (79%), and LRTI (45%), was associated with lower probabilities of SPT positivity at 9 years.

Conclusions:

In this Ugandan birth cohort, early childhood infection-exposure, notably to malaria, helminths, LRTI, and diarrhoea, is associated with lower prevalence of atopy and ARDs in later childhood.

Funding:

This work was supported by several funding sources. The Entebbe Mother and Baby Study (EMaBS) was supported by the Wellcome Trust, UK, senior fellowships for AME (grant numbers 064693, 079110, 95778) with additional support from the UK Medical Research Council. LL is supported by a PhD fellowship through the DELTAS Africa Initiative SSACAB (grant number 107754). ELW received funding from MRC Grant Reference MR/K012126/1. SAL was supported by the PANDORA-ID-NET Consortium (EDCTP Reg/Grant RIA2016E-1609). HM was supported by the Wellcome’s Institutional Strategic Support Fund (grant number 204928/Z/16/Z).

Introduction

Lack of early infection-exposure has been associated with increased allergy-related disease (ARD) susceptibility later in life, the so-called ‘hygiene hypothesis’ (Strachan, 1989; Bloomfield et al., 2006; Schaub et al., 2006) or ‘old friends hypothesis’ (Rook, 2010; Rook, 2011). Studies, mainly in high-income countries (HICs), have highlighted the inverse relationship between pathogen-exposure and atopy, specifically for hepatitis A virus (Strachan, 2000), gut flora (Björkstén et al., 1999), intestinal parasites (Yazdanbakhsh and Matricardi, 2004), mycobacteria (Shirakawa et al., 1997), malaria (Lell et al., 2001), total burden of infections (Martinez and Holt, 1999; Illi et al., 2001), and farm animal sheds (Loss et al., 2016).

Despite increasing evidence indicating inverse relationships between pathogen-exposure and atopy, other studies suggest no association (Jarvis et al., 2004; Cooper et al., 2003) or even increased risk of atopy following a combination of early infections (Seaton and Devereux, 2000; Bager et al., 2002). Therefore, it remains unclear which infections, and at which times, possibly contribute to reducing the risk of atopy or ARDs.

There is a paucity of data from countries in tropical Africa where the infectious diseases burden remains high, yet prevalence of ARDs remains low, despite reasonably high prevalence of atopy. Similar to findings from HICs (Bach, 2002), we have shown previously that ARDs (eczema, wheeze, and rhinitis) decline with age, despite increasing atopy (Lule et al., 2017).

We therefore investigated the role of common early childhood infections in the development of atopy and ARDs, using data from the Entebbe Mother and Baby Study (EMaBS) birth cohort (Webb et al., 2011; Elliott et al., 2007). We hypothesised that (i) there are latent classes (LCs) of children with similar early infection-exposure and (ii) specific profiles of early infection-exposure are associated with reduced ARDs in later childhood.

Materials and methods

Study design and population

Request a detailed protocol

The EMaBS has been described previously (Elliott et al., 2007). Briefly, the EMaBS, based at Entebbe General Hospital, Uganda, originated as a trial of anthelminthic treatment in pregnancy and early childhood, in a 2 × 2 (×2) factorial design. Between 2003 and 2005, women in their second or third trimester of pregnancy were randomised to single-dose albendazole (400 mg) vs. placebo and single-dose praziquantel (40 mg/kg) vs. placebo; their children were randomised to quarterly albendazole vs. placebo from age 15 months to 5 years (Webb et al., 2011Elliott et al., 2007). Cohort participants continue under follow-up and had detailed data on ARDs collected at age 9 years. Our earlier study showed no effect of the anthelminthic intervention during pregnancy and early childhood on allergy-related outcomes at 9 years (Namara et al., 2017). The geography of the study area has been described previously (Hillier et al., 2008) the main zones of maternal residence at enrolment included Kigungu (a relatively isolated fishing village, at the tip of the Entebbe peninsula, n = 267), Entebbe (a commercial town, n = 1018), Manyago and Kabale (peri-urban, n = 687), and Katabi (comprised of peri-urban and rural areas, n = 486) (Figure 1).

Map of Entebbe and Katabi showing the study area and the three main geographical zones of maternal residence at enrolment.

Ethical approval

Request a detailed protocol

The study was approved by ethics committees of the Uganda Virus Research Institute, London School of Hygiene and Tropical Medicine, and Uganda National Council for Science and Technology. Written informed consent and assent were obtained.

Exposure variables

Request a detailed protocol

The main exposures for this study were children’s most common clinical illnesses (malaria, diarrhoea, upper and lower respiratory tract infections [URTI and LRTI], intestinal helminths), and seroprevalence of herpes simplex virus (HSV), cytomegalovirus (CMV), and norovirus, in the first 5 years of life.

Malaria, diarrhoea, URTI, and LRTI were assessed and recorded by a doctor when a participant presented to the study clinic with an illness. Malaria was fever (temperature ≥37.5°C) with parasitaemia by thick film microscopy; diarrhoea was a child’s carer’s definition, with stool frequency recorded (Webb et al., 2011); URTI was the common cold while LRTI was cough, with difficulty in breathing, and fast breathing (defined by age), with or without abnormal breath sounds (Elliott et al., 2007).

At scheduled annual visits, a stool sample was collected and examined for the presence of helminth eggs using the Kato-Katz method (Katz et al., 1972). Two slides were examined for each sample, read within 30 min for hookworm and the following day for other worms including Schistosoma mansoni, T. trichiura, and Ascaris lumbricoides (Mpairwe et al., 2014).

Norovirus seropositivity was determined using immunoglobulin (Ig)G responses to human norovirus-like particles using enzyme-linked immunosorbent assays (ELISAs), as reported previously (Thorne et al., 2018). For 779 randomly selected children, plasma samples at 1 year were screened; any found negative were further tested at subsequent years until a positive response was detected (Thorne et al., 2018).

HSV and CMV seropositivity were determined using IgG responses, from standard commercially available ELISAs (DiaSorin, Saluggia, Italy). All plasma samples available at 2 and 5 years were assayed. Children whose samples were positive at 2 years had their 1-year samples tested. Those with negatives samples at 2 years were further assessed at years 3 and 4 until a positive response was detected.

ARD outcomes and atopy at 9 years

Request a detailed protocol

ARDs and atopy at 9 years were the outcomes for this analysis. This is because there was a specific effort to collect data on ARDs, skin prick test (SPT) reactivity, and allergen-specific immunoglobulin E (asIgE) during the 9-year annual visit. We used the International Study on Allergy and Asthma in Children (ISAAC) questionnaire (Asher et al., 1995) to collect data on recent (last 12 months) reported wheeze, eczema (recurrent itchy rash with typical flexural distribution), and rhinitis, classified according to responses from caregivers for their children (Namara et al., 2017).

Atopy was assessed by SPT and asIgE as previously reported (Lule et al., 2017; Namara et al., 2017). SPT reactivity to Dermatophagoides mix (D. farinae and D. pteronyssinus), Blomia tropicalis, German cockroach, cat, mould, grass pollen, Bermuda grass, and peanut (ALK-Abelló, Laboratory Specialities (Pty) Ltd, Randburg, South Africa) was assessed using standard methods (Heinzerling et al., 2013). A test was classified as positive if there was a papule of average size >3 mm, in the presence of saline (negative) and histamine (positive) controls. Plasma IgE to the commonest allergens in this setting (house dust mite [D. pteronyssinus] and German cockroach [Blatella germanica]) (Mpairwe et al., 2008) was measured using an in-house ELISA as previously described (Mpairwe et al., 2011; Nkurunungi et al., 2018). Because of the dynamic range of our in-house assay, it was not possible to use undiluted plasma, hence we optimised our assay to work with 20-fold diluted samples, with a lower detection limit of 15.625 ng/ml (equivalent to 312.5 ng/ml in undiluted plasma). We have previously shown that results from our ELISA were positively correlated with those from ImmunoCAP for both dust mite and cockroach (Sanya et al., 2019).

Statistical analysis

Request a detailed protocol

Analyses aimed to classify participants into LCs based on infection-exposure during the first 5 years of life and to study the association between early infection-exposure and ARDs and atopy at 9 years. Since infection experience is an unobservable construct which can be inferred from multiple observed infections, we used the LC analysis (LCA) framework which has been shown to flexibly divide a population into mutually exclusive subgroups (Goodman, 1974; Lanza et al., 2007; Nylund-Gibson and Choi, 2018; Velicer et al., 1996; Ryoo et al., 2018).

We fitted LCA models using infection-exposure variables as indicators. Models were fitted, successively increasing the number of classes up to a seven-class solution. Two models were fitted: one considering infection-exposure during the first year of life (infancy) and the other using cumulative infection experience over the first 5 years, with no a priori restrictions to find consistent classes across the two models. First-year infection-exposure was considered because of the importance of infancy in immune development following exposure to several pathogens for the very first time. We used the Bayesian information criterion (BIC), adjusted BIC (ABIC), consistent Akaike information criterion (CAIC), and entropy to guide the choice of optimal number of classes (Lanza et al., 2007; Swanson et al., 2012).

We conducted multiple-group LCA (Lanza et al., 2007) to establish whether measurement invariance across sex holds. Multiple-group LCA allows class membership and item response probabilities to vary across groups. Formal tests for measurement invariance across sex were conducted for the selected models. We incorporated the covariates maternal area of residence, maternal history of asthma or eczema, and maternal and household socio-economic status (SES) at enrolment, to establish whether these impacted the probability of class membership, through multinomial logistic regression (Lanza et al., 2007). Maternal SES was determined by level of education, personal income, and occupation while household SES was based on building materials and the number of rooms and items owned (Muhangi et al., 2007). SES scores were split into two: low and high. We predicted ARDs and atopy at 9 years, from LC membership, using a SAS macro (Dziak et al., 2017), implementing a model-based approach as previously described (Lanza et al., 2013).

All models were estimated using SAS version 9.4 (SAS Institute Inc, Cary, NC) via PROC LCA which handles missing data on LC indicators under the missing at random assumption (Lanza et al., 2007).

Results

Of 2345 live births in the EMaBS cohort, 2115 had data on illnesses during their first year of life, 1945 in the second, 1884 in the third, 1856 in the fourth, and 1838 in their fifth year (Figure 2).

Flowchart showing number of the Entebbe Mother and Baby Study participants with data on illnesses/infections during each year.

Numbers for each year (centre column of the flowchart) represent the children who presented at the study clinic at any point during that year. Numbers for malaria, diarrhoea, lower respiratory tract infection (LRTI) and upper respiratory tract infection (URTI) represent the children who were positively diagnosed (at least once during a given year) for each of those illnesses, on presentation at the study clinic. Numbers for any helminth, norovirus, cytomegalovirus (CMV), and herpes simplex virus (HSV) represent children samples assessed for any of those infections at the respective annual visit.

Maternal and infant characteristics of the cohort participants, during the first 5 years of life, have been described previously (Mpairwe et al., 2014). Briefly, 52% of the children were male, 90% were of normal birth weight, 57% were second to fourth born; maternal characteristics included 55% with none/primary education, 3% with history of asthma, 3% with history of eczema, 44% with hookworm infections, 21% with Mansonella perstans, 18% with S. mansoni, and 12% with Strongyloides stercoralis (Mpairwe et al., 2014); 1214 children were seen at 9 years, 52% of whom were male and 1179 had data collected on atopy and ARDs (Lule et al., 2017). Children seen at 9 years were similar to those not seen in terms of maternal marital status, BMI and worm infection at enrolment, and child’s sex, eczema diagnosis (before age 1 year), and atopy (3 years) (Lule et al., 2017). Compared to mothers whose children did not attend the 9-year visit, mothers of children seen at 9 years were older, more likely to be Baganda (the tribe traditionally based in central Uganda), multiparous, and with higher SES at enrolment (Lule et al., 2017; Namara et al., 2017).

Prevalence of childhood infections during the first five years of life

Diarrhoea and LRTI were common in the first year of life but declined by the fifth year, malaria was most common in the second year of life and then declined, URTI were consistently common throughout the first 5 years, helminth infections increased slightly, HSV increased steadily over the 5 years while CMV and norovirus seropositivity increased to over 85% by the second year (Figure 3).

Proportion of children with illnesses/infections, over time, in the Entebbe Mother and Baby Study cohort.

Considering cumulative experience of the various infections, over 95% of the cohort participants had been infected with URTI, noroviruses, and CMV by the end of the third year of life. Diarrhoea infections had been reported by over 91% by the third year increasing to 95% by the fifth year. HSV increased steadily over time reaching a level of 91% by the fifth year. The proportion of children with any recorded episode of malaria was 56% and 68% by the end of the third and fifth years, respectively. LRTI increased gradually reaching 32% and 41% by the end of the third and fifth years, respectively. Helminth infections increased slowly reaching 31% by the end of the fifth year. These cumulative figures are shown in Figure 4.

Cumulative infection experience, over time, in the Entebbe Mother and Baby Study cohort.

LCA models

We conducted a sequence of models, considering additional classes gradually up to a seven-class solution. Models were fitted separately considering infection-exposure during the first year of life (infancy) and then using cumulative experience of infections over all the first 5 years. Supplementary file 1a summarises LCA results for the two different sets of models. Based on BIC, CAIC, and ABIC, two-class solution models were favoured for both first-year and 5-year cumulative infection experience. Although the entropy, which measures certainty in classifying latent statuses, favoured seven-class and six-class solutions for the two sets of models, we chose the two-class solutions for easy interpretation in addition to being favoured by other statistics.

Multiple-group LCA showed that measurement invariance by sex holds, based on the likelihood ratio difference test (p-value = 0.379 and p-value = 0.558 at year 1 and year 5, respectively) (Supplementary file 1b).

Definition of LCs

Figure 5 shows the probability of having been diagnosed with any of the infections, conditional on membership in either of the LCs, during the first year of life. The probabilities of membership in LC 1 and LC 2 are estimated as 32% and 68%, respectively. LC 1 was characterised by higher probabilities for malaria, diarrhoea, and LRTI (80% vs. 0.6%, 76% vs. 65%, and 22% vs. 12%, respectively) compared to LC 2. Conditional probabilities for other infections were similar between the LCs.

Probability of having infections conditional on latent class membership during the first year.

Figure 6 shows the probability of having been diagnosed with any of the infections, conditional on membership in either of the LCs, considering cumulative infection prevalence over all the first 5 years of life. The probabilities of membership in LC 1 and LC 2 are estimated as 31% and 69%, respectively. LC 1 was characterised by higher probabilities for helminth infections, malaria, and LRTI (92% vs. 2.4%, 79% vs. 64%, and 45% vs. 40%, respectively) compared to LC 2. Probabilities for other infections were similar between the LCs.

Probability of having infections conditional on latent class membership considering cumulative prevalence of infections over all the 5 years.

We labelled the class with higher probabilities for various infections as LC 1 and the other as LC 2. Membership of LC 1 was 32% in year 1 and 31% in year 5 (considering cumulative infection experience) (Table 1).

Table 1
Number (and proportion) of children in each latent class at year 1 and at year 5 considering cumulative infection experience.
LC 1LC 2
Number (%)Number (%)
Year 1 (n = 2077)665 (32)1412 (68)
Year 5 (n = 1668)517 (31)1151 (69)
  1. LC is latent class. LC 1 represents the class with higher probabilities for various infections as compared to LC 2, however, the profile of infections changes at each time period.

Effects of covariates on LC membership

At year 1, compared to children born in Entebbe town, those born in Kigungu fishing village were less likely to be in LC 1 [odds ratio (OR) = 0.33 (95% confidence interval (CI): 0.18, 0.62)], while those born in Katabi (peri-urban and rural areas) were more likely to be in LC 1 [OR = 2.43 (1.62, 3.65)] (Table 2).

Table 2
Odds ratio estimates for effects of covariates on membership in latent class 1 at year 1 and at year 5 considering cumulative infection experience.
Year 1 exposure5-Year cumulative exposure
CovariateOdds ratio(95% CI)Odds ratio(95% CI)
Mother’s social economic status
Higher vs. lower0.77(0.58, 1.02)0.51(0.31, 0.82)
Household’s social economic status
Higher vs. lower0.87(0.66, 1.14)0.58(0.25, 0.92)
Maternal history of asthma or eczema
History present vs. absent0.74(0.43, 1.27)0.50(0.24, 1.53)
Area of residence at birth
Kigungu vs. Entebbe0.33(0.18, 0.62)5.73(3.24, 8.12)
Manyago vs. Entebbe1.20(0.88, 1.63)1.08(0.88, 1.63)
Katabi vs. Entebbe2.43(1.62, 3.65)2.04(1.07, 3.54)
  1. CI is confidence interval. Bold values indicate statistically significant values at the 5% level.

Considering the LCA model for 5-year cumulative infection experience, children with higher maternal SES and those with higher household SES were less likely to be in LC 1 [OR = 0.51 (0.31, 0.82) and OR = 0.58 (0.25, 0.92), respectively] (Table 2). Children born in Kigungu and Katabi were more likely to be in LC 1 [OR = 5.73 (3.24, 8.12) for Kigungu and OR = 2.04 (1.07, 3.54) for Katabi] (Table 2). Maternal history of asthma or eczema was not associated with LC membership.

Prevalence of atopy and ARD outcomes at 9 years

Prevalence of atopy and ARDs at 9 years has been reported elsewhere (Lule et al., 2017; Namara et al., 2017). Briefly, out of 1179 children for whom data on ARDs was collected at 9 years, prevalence of reported recent wheeze was 3.8%, eczema 4.9%, allergic rhinitis 4.6%, SPT positivity for at least one allergen 25%, for Dermatophagoides 18%, for Blomia 15%, for German cockroach 12%, for peanut 1.4%, for Bermuda grass 1.2%, for cat 1.1%, for pollen 0.9%, and for mould 0.3% (Lule et al., 2017). Prevalence of detectable IgE to house dust mite was 29.4%, to cockroach 32.3%, or to any of the two 44.1% (Lule et al., 2017; Namara et al., 2017).

Associations between LC membership and atopy and ARD outcomes at 9 years

Based on covariate adjusted LCA of infection data collected during the first year of life, LC 1 membership was associated with lower probabilities of wheeze, eczema, rhinitis, and SPT positivity for Dermatophagoides; there was some suggestion of difference between the two LCs for SPT positivity overall and for Blomia (with SPT positivity less common for LC 1 children) but for cockroach and the IgE outcomes there was no suggestion of a difference (Table 3).

Table 3
Probability of allergy-related disease (ARD) outcomes and atopy at 9 years by latent class membership at year 1.
95% CI*
ARD outcomes at 9 yearsLatent classEstimateLower CIUpper CIp-Value
WheezeLC 10.020.0070.0430.030
LC 20.050.0380.071
EczemaLC 10.030.0130.0560.043
LC 20.060.0470.084
RhinitisLC 10.020.0070.0440.015
LC 20.060.0450.081
SPT-anyLC 10.220.1780.2780.201
LC 20.270.2350.304
SPT-DermatophagoidesLC 10.140.1030.1880.051
LC 20.200.1710.232
SPT-cockroachLC 10.110.0740.1490.760
LC 20.110.0910.141
SPT-BlomiaLC 10.120.0890.1690.144
LC 20.160.1380.196
IgE-DermatophagoidesLC 10.260.2140.3220.261
LC 20.310.2710.343
IgE-cockroachLC 10.300.2490.3610.429
LC 20.330.2950.370
IgE-any§LC 10.420.3570.4780.395
LC 20.450.4110.490
  1. *

    CI is confidence interval.

  2. LC is latent class. LC 1 represents the class with higher probabilities for various infections as compared to LC 2.

  3. SPT-any is skin prick test reactivity to any of Dermatophagoides mix, Blomia tropicalis, German cockroach, cat, mould, grass pollen, Bermuda grass, or peanut.

  4. §

    IgE-any is allergen-specific plasma IgE (asIgE) to any of house dust mite (HDM, Dermatophagoides pteronyssinus) or German cockroach.

Estimates are adjusted for maternal area of residence, maternal history of asthma or eczema, and maternal and household SES at enrolment.

Based on covariate adjusted LCA of cumulative infection-exposure data collected throughout the first 5 years of life, LC 1 membership was associated with lower probabilities of SPT positivity for Dermatophagoides and for any of the other allergens (any of Dermatophagoides mix, B. tropicalis, German cockroach, cat, mould, grass pollen, Bermuda grass or peanut); there was some suggestion of difference between the two LCs for eczema but for SPT positivity for Blomia, or cockroach, wheeze, rhinitis, and the IgE outcomes there was no suggestion of a difference (Table 4).

Table 4
Probability of allergy-related disease (ARD) outcomes and atopy at 9 years by latent class membership across the first 5 years.
95% CI*
ARD outcomes at 9 yearsLatent classEstimateLower CIUpper CIp-Value
WheezeLC 10.020.0090.0650.272
LC 20.050.0320.066
EczemaLC 10.020.0040.0620.056
LC 20.070.0500.090
RhinitisLC 10.020.0080.0670.139
LC 20.060.0420.080
SPT-anyLC 10.170.1210.2400.012
LC 20.290.2490.325
SPT-DermatophagoidesLC 10.110.0690.1690.010
LC 20.210.1800.248
SPT-cockroachLC 10.080.0480.1350.184
LC 20.120.0990.154
SPT-BlomiaLC 10.110.0700.1690.096
LC 20.170.1430.206
IgE-DermatophagoidesLC 10.320.2560.3980.416
LC 20.280.2470.325
IgE-cockroachLC 10.340.2710.4140.658
LC 20.320.2790.359
IgE-any§LC 10.480.4030.5550.316
LC 20.430.3860.471
  1. *

    CI is confidence interval.

  2. LC is latent class. LC 1 represents the class with higher probabilities for various infections as compared to LC 2.

  3. SPT-any is skin prick test reactivity to any of Dermatophagoides mix, Blomia tropicalis, German cockroach, cat, mould, grass pollen, Bermuda grass, or peanut.

  4. §

    IgE-any is allergen-specific plasma IgE (asIgE) to any of house dust mite (HDM, Dermatophagoides pteronyssinus) or German cockroach.

Estimates are adjusted for maternal area of residence, maternal history of asthma or eczema, and maternal and household SES at enrolment.

Discussion

In this study, using LCA, we identified two distinguishable groups of children based on either first-year or 5-year cumulative infection experience. In the first year of life, LC 1 was characterised by higher probabilities of malaria, diarrhoea, and LRTI, and membership of this class was associated with lower proportions of wheeze, eczema, rhinitis, and SPT positivity for Dermatophagoides at 9 years. Considering cumulative infection experience over the first 5 years of life, LC 1 was characterised by higher probabilities of helminth infections, malaria, and LRTI, and membership of this class was associated with lower probabilities of SPT positivity for Dermatophagoides and SPT positivity overall. Our results imply a significant contribution of immuno-modulating parasitic infections in early life to protection against the development of atopy and ARDs in later childhood in this tropical, low-income setting.

A major strength of this study lies in the use of LCA, a novel person-centred analysis approach, to characterise infection-exposure – an unobservable construct inferred from multiple observed infections in a given time period. At both time periods, class membership was characterised by differences in the burden of infections; malaria, diarrhoea, and LRTI during the first year and helminths, malaria, and LRTI during the 5-year period. LCA for first-year infection-exposure demonstrated the strong inverse association between infectious illnesses in infancy and ARDs in later childhood. On the other hand, LCA for cumulative infection experience over the first 5 years only illustrated strong inverse associations with atopy (SPT positivity). This suggests that infancy is a critical period during which the risk of ARDs in later childhood is established.

The finding that LC 1 membership, during the first year, was associated with lower proportions of wheeze, eczema, rhinitis, and SPT positivity for Dermatophagoides is consistent with observations from elsewhere (Lell et al., 2001) and with our earlier results which showed that children with more frequent clinical malaria infections in their first 5 years of life had a lower incidence of eczema in the same period (Mpairwe et al., 2014), and supports the hygiene hypothesis that infections in early life are associated with a lower risk of atopy and ARDs in later childhood (Strachan, 2000; Björkstén et al., 1999; Yazdanbakhsh and Matricardi, 2004; Shirakawa et al., 1997; Lell et al., 2001; Martinez and Holt, 1999; Illi et al., 2001; Loss et al., 2016).

We also describe the life-course of common infections in the first 5 years of life, in a tropical African birth cohort. We show that diarrhoea, malaria, and LRTI were more common in infancy but gradually declined by the fifth year, URTI remained very common while helminth infections increased gradually throughout the first 5 years, HSV increased steadily over time while CMV and norovirus infections were common throughout the 5 years. This result is especially important when studying which infections, occurring at which time, contribute to the development of atopy and ARDs. We show that in as much as two LCs are favoured at both time periods, conditional probabilities for the infections varied between first-year and 5-year cumulative exposure periods. Malaria was the largest contributor for LC 1 membership during the first year while helminth infection was the largest contributor for the 5-year period.

Compared to children born in Entebbe town, those born in Kigungu fishing village were less likely to be in LC 1 in year 1, while those born in Katabi (peri-urban and rural areas) were more likely to be in LC 1. These findings accord with our earlier study which showed comparable geographical variations in the prevalence of malaria in these areas (Hillier et al., 2008).

Another intriguing result was that based on 5-year cumulative infection-exposure, LC 1 was associated with a higher probability of helminth infections and membership in this class was associated with lower probabilities of SPT positivity, but was not associated with clinical ARDs. These findings accord with our earlier study which showed inverse associations between SPT positivity and S. mansoni infection in the rural setting (Nkurunungi et al., 2019).

A limitation of our study could be in the potential influence of missing data. The LCA models employ a maximum-likelihood routine which operates under a missing at random assumption. This assumption is inherently untestable, however, we consider it reasonable since the amount of missing data for each indicator variable, during each time period, was relatively low. Another limitation is that it was not possible to account for multiple exposure to infections (under the LCA approach) due to data on different infections being collected through different approaches. Some infections (HSV, norovirus, CMV, and helminths) were only captured (once, by serology) at the scheduled annual visits, while others (malaria, diarrhoea, LRTI, and URTI) were recorded every time a participant was presented to the study clinic. Another possible limitation could be in the use of 2077 and 1668 participants for determining LC membership in infancy, and between years 1 and 5, respectively, while only 1179 participants had outcome data at 9 years. In the analysis plan, which was developed before data analysis, we specified that we would use all available data to define LCs. We felt that this was appropriate because it would describe the infection-exposure experience of the whole cohort, before going ahead to assess associations between this infection-exposure experience and the 9-year outcomes. We had detailed data on important potential confounders which we adjusted for in this analysis, however we cannot rule out the possibility of residual confounding.

Conclusion

In conclusion, we show that in this tropical African birth cohort children can be classified into two LCs based on their early childhood infection experience and that exposure to common infections, mainly malaria, during the first year of life, was associated with lower proportions of ARDs in later childhood. With cumulative infection experience over all the first 5 years, exposure to infections, mainly helminths, was associated with lower probabilities of SPT positivity in later childhood.

Data availability

Data is available on request via https://doi.org/10.17037/DATA.00002438. To gain access to the data please complete the application process via the website. Requests will be reviewed and assessed by the corresponding author, in consultation with the LSHTM's Research Data Manager and relevant LSHTM staff members responsible for research governance and data protection. Applications will be evaluated on the basis of their compatibility with the study's research objectives and the ability to provide de-identified data sufficient to meet the intended purpose, without breaching participant confidentiality or the study's ethical and legal commitments. Successful applicants will be asked to sign a Data Transfer Agreement prior to being provided with the data.

The following data sets were generated

References

  1. Book
    1. Dziak JJ
    2. Bray BC
    3. Wagner AT
    (2017)
    LCA_Distal_BCH SAS Macro Users’ Guide (Version 1.1). the Methodology Center, University Park, PA
    Penn State.
    1. Katz N
    2. Chaves A
    3. Pellegrino J
    (1972)
    A simple device for quantitative stool thick-smear technique in schistosomiasis Mansoni
    Revista Do Instituto de Medicina Tropical de Sao Paulo 14:397–400.
    1. Lell B
    2. Borrmann S
    3. Yazdanbakhsh M
    4. Kremsner PG
    (2001)
    Atopy and malaria
    Wiener Klinische Wochenschrift 113:927–929.

Decision letter

  1. Jos W Van der Meer
    Senior and Reviewing Editor; Radboud University Medical Centre, Netherlands

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The authors present an analysis of the potential effects of early life infections on allergy related outcomes at 9 years of age. To do this, they use large study population from Entebbe, Uganda, that was the subject of randomized controlled trails of the effects of anthelmintic treatment given during pregnancy to mothers and later to children during up to 5 years of age. The authors use these data to do a cohort analysis of the effects of early life infections on allergy outcomes later in childhood. What makes this study important and extremely unusual are the wide range of infectious diseases exposure measurements taken and the detailed data on allergic outcomes at 9 years.

Decision letter after peer review:

Thank you for submitting your article "Infection-exposure in infancy is associated with reduced allergy-related disease in later childhood in a Ugandan cohort" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jos van der Meer as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and we have drafted this to help you prepare a revised submission.

Essential revisions:

The authors investigated the association between infection exposure in early life and susceptibility to allergy-related diseases (ARD) in the Entebbe Mother and Baby Study (EMaBS) birth cohort, using the latent class analysis (LTA). Most common clinical illnesses such as malaria, diarrhea, upper and lower respiratory tract infections, intestinal helminths, herpes simplex virus infection, cytomegalovirus and norovirus infections are analyzed in the first 5 years of life, as well as ARD through skin prick test and allergen-specific Immunoglobulin E (asIgE) at 9 years of age. At each year of life, children were divided into two groups (latent classes) defined by their infection-exposure history in that year and probabilities of ARDs outcomes in these groups were compared. Authors concluded that early childhood infection-exposure to malaria and diarrhoea was associated with lower prevalence of ARDs in later childhood.

Strengths:

1. The cohort with a large number of patients originating from 4 different geographical areas and followed throughout the first 5 years of age, with a detailed follow-up at 9 years of age.

2. The amount of data obtained and available for each patient, which allows an extended analysis and comparison.

Weaknesses:

The major weaknesses of the paper concern the statistics applied.

1. Analysis of the exposure: It is not the best approach to define exposure classes for each year and separately analyze their associations with outcomes at 9 years. Outcomes could be affected by any/all exposures in years 1-5, so analysis of a single year is likely to be confounded by exposure in the other years and provides biased estimates. Additionally, many children were likely to be exposed to some infections multiple numbers of time. It is unclear how this was handled in the analysis. Not accounting for the multiple exposure could also bias the results as, under the study hypothesis, children with multiple exposure would be those mostly protected from the ARDs.

2. Analysis of the association with outcome: The association between exposure and ARDs is studied in separated from other possible risk factors for ARDs, such as genetic atopic predisposition, early childhood allergen exposure and sensitization, maternal smoking during pregnancy, poor dietary factors, lack of breast-feeding, childhood obesity, having a certain immunologic predisposition, air pollution, and frequent immunizations in childhood. These risk factors as well as ARDs status at 1-5 years are crucial in understanding the association between infection exposure and development of ARDs. This needs to be discussed.

3. Interpretation of results: Possible bias in results aside, they are very difficult to interpret. Results of the analyses are inconsistent across years. Differences between classes with respect to exposure patterns and probabilities of outcomes are rather small, therefore it is impossible to draw any sound conclusions on the association (or lack of it) between exposures and outcomes. Authors' strong conclusion is in contradiction to the above, but also to their own assessment of weaknesses of the study presented in the discussion. The interpretation of the results was primary based on p-values, when multiple testing was certainly an issue with 10 outcomes and repeated analysis for each of 5 years.

As the statistical analysis is considered the major weakness these issues are listed first.

1. We think that the data should have been analyzed together, not split by year. We do not understand why single LCA exploring exposure over 5 years could not be performed. We think LTA, which authors were planning to conduct in the first place, would not be the right approach as it explores changes in the exposure over time – while the objective of this analysis was to relate the exposure in the first years to ARDs at nine years of age. Latent classes defined over 5 years would account for any changes in exposures.

2. The standard procedure for conducting LCA is to conduct a sequence of models, starting with a one-class model and then specifying models with one additional class at a time. Restriction of the analysis to {greater than or equal to}2 classes is very likely to produce some "significant" associations, especially when multiple testing is performed.

3. The analysis cohort is not described at all, which is important especially in the context of risk factors for ARDs and the history of ARDs.

4. Exposure to an infection is presented as a proportion of children with that infection per year, but not presented per child. For each infection, what was the incidence rate and its variability between children? It is unclear how exposure was captured for each child in the analysis – as an incidence or as `at least one infection'? Incidence could have been a better representation of the exposure.

5 Other risk factors for ARDs should have been considered in the analysis otherwise the association between exposure and outcome could be seriously biased.

6. It would be worth summarizing the effects of the RCTs on allergy outcomes to 9 years in the introduction, so that the reader is informed of any potential effects of treatment arms on study outcomes, given that the analysis does not take into account the original design. It might also be worth doing sub-group analyses among control-arms of both RCTs to see if the results of those, albeit likely underpowered analyses, are consistent with the whole-group analysis such that treatment allocations can be safely ignored. It would be useful to provide some discussion in the limitations section of the potential effects, if any, of these RCTs on study findings.

7. The cohort is composed of 2345 children, but only 1179 have data on ARD at nine years of age. As the goal is to investigate a possible association between infection exposure in early life and susceptibility to ARDs, shouldn't the number of analyzed subjects be 1179 in total? The authors should either repeat the LTA or discuss the difference in population samples (2345 vs 1179) as a weakness/limitation of their study.

8. The authors describe the ARD information methodology and state (page 6, lines 122-124) that they used 20-fold diluted plasma, thus significantly reducing the sensitivity of their test. IgE tests are normally done on undiluted serum. The reason for diluting plasma 20x should be given. Moreover, information concerning validation of this assay against a gold standard (e.g. immunoCAP, TFS) should be at least mentioned here (just a reference is not enough). Finally, what SPT wheal (not papule!) diameter corresponds to the threshold of 312.5 ng/ml? Is the IgE test sensitive enough?

9. Plasma samples were only tested against Dermatophagoides pteronyssinus and Blatella germanica. Please give information why this choice has been made. Why were only these two allergen extract examined with IgE assays?

10. Please insert more information to help the interpretation of data (the favored models) of table 1 (page 11) and consider to move the table in an electronic repository, considering that this table is occupying one page and the respective information is not the major focus of the paper.

11. It seems that children from Entebbe were compared against the other 3 areas, but these were not compared among themselves. Please provide a brief explanation for this choice.

12. The LCA model is fitted separately at each time point: 1-5 years of age. This provides a lot of information, but can also originate some confusion: children born in Kigungu are less likely to be in LC 1 – and therefore have a higher probability for various infections – when they are 2 years of age, but the opposite occurs when they are 4 years of age (page 14, lines 242-247). This age dependency of LTA categorization needs to be discussed: is it a limitation of the study?

13. SPT or IgE positivity is not equivalent of ARD. Please reformulate the definition of atopic and/or allergic outcome at 9y of age in the whole paper.

14. A table showing the prevalence of SPT/IgE positivity for each allergen extract should be given somewhere in the results. In particular, only a few extracts (blomia, mites and cockroach) are used in the presentation of the table 4. However, many more allergen extracts have been examined. Probably, the prevalence of e.g. pollen sensitization was low. However, this is not shown in the paper.

Reviewer #1:

The authors present an analysis of the potential effects of early life infections on allergy and allergic diseases at 9 years of age. To do this, they use large study population from Entebbe, Uganda, that was the subject of previous randomized intervention studies. The authors use these data to do a cohort analysis of the effects of early life infections on allergy later in childhood. What makes this study important and extremely unusual are the wide range of infectious diseases exposure measurements taken and the detailed data on allergic outcomes at 9 years. Because many of these infections are strongly associated, the authors used a statistical technique to infer how underlying patterns of infectious diseases might affect later allergy. The authors show that having more types of different infections during the first year of life appears to protect the children from having allergy at school-age.

Reviewer #2:

The authors investigate the association between infection exposure in early life and susceptibility to allergy-related diseases (ARD) in the Entebbe Mother and Baby Study (EMaBS) birth cohort, using the latent class analysis (LTA). Most common clinical illnesses such as malaria, diarrhoea, Upper and Lower Respiratory Tract Infections, intestinal helminths, Herpes Simplex Virus, Cytomegalovirus and norovirus are analysed in the first 5 years of life, as well as ARD through skin prick test and allergen-specific Immunoglobulin E (asIgE) at 9 years of age. Using LTA, children were allocated into two latent classes: LC1 and LC2, being LC1 the one with higher probabilities for various infections. The Authors concluded that early childhood infection-exposure is associated with lower prevalence of ARDs in later childhood.

General comments: This article provides interesting new data within a topic of high interest and importance.

Strengths:

– The cohort, with a large amount of patients originating from 4 different geographical areas and followed throughout the first 5 years of age, with a detailed follow-up at 9 years of age.

– The amount of data obtained and available for each patient, which allows an extended analysis and comparison, therefore leading to new knowledge and further hypothesis.

– The outcome of the work.

Reviewer #3:

Lubyayi, Lawrence et al., conducted a secondary data analysis to explore association between early childhood infections and development of atopy and allergy related diseases (ARDs) in later childhood. Data from the Entebbe Mother and Baby Study birth cohort were used. Information on exposure to children's most common clinical illnesses and seroprevalence of common viruses was available in the first five years of life; allergy-related disease outcomes at nine years of age were collected using allergy tests and standard questionnaires. Multiple latent class analyses were conducted. At each year of life, children were divided into two groups (latent classes) defined by their infection-exposure history in that year and probabilities of ARDs outcomes in these groups were compared. Authors concluded that early childhood infection-exposure to malaria and diarrhoea was associated with lower prevalence of ARDs in later childhood.

The study has a number of serious weaknesses:

1. Analysis of the exposure: It is not the best approach to define exposure classes for each year and separately analyse their associations with outcomes at 9 years. Outcomes could be affected by any/all exposures in years 1-5, so analysis of a single year is likely to be confounded by exposure in the other years and provides biased estimates. Additionally, many children were likely to be exposed to some infections multiple number of times. It is unclear how this was handled in the analysis. Not accounting for the multiple exposure could also bias the results as, under the study hypothesis, children with multiple exposure would be those mostly protected from the ARDs.

2. Analysis of the association with outcome: The association between exposure and ARDs is studied in a complete separation from other possible risk factors for ARDs, such as genetic atopic predisposition, early childhood allergen exposure and sensitization, maternal smoking during pregnancy, poor dietary factors, lack of breast-feeding, childhood obesity, having a certain immunologic predisposition, air pollution, and frequent immunizations in childhood. These risk factors as well as ARDs status at 1-5 years are crucial in understanding the association between infection exposure and development of ARDs.

3. Interpretation of results: Possible bias in results aside, they are very difficult to interpret. Results of the analyses are inconsistent across years. Differences between classes with respect to exposure patterns and probabilities of outcomes are rather small, therefore it is impossible to draw any sound conclusions on the association (or lack of it) between exposures and outcomes. Authors' strong conclusion is in contradiction to the above, but also to their own assessment of weaknesses of the study presented in the discussion. The interpretation of the results was primary based on p-values, when multiple testing was certainly an issue with 10 outcomes and repeated analysis for each of 5 years.

https://doi.org/10.7554/eLife.66022.sa1

Author response

Essential revisions:

The authors investigated the association between infection exposure in early life and susceptibility to allergy-related diseases (ARD) in the Entebbe Mother and Baby Study (EMaBS) birth cohort, using the latent class analysis (LTA). Most common clinical illnesses such as malaria, diarrhea, upper and lower respiratory tract infections, intestinal helminths, herpes simplex virus infection, cytomegalovirus and norovirus infections are analyzed in the first 5 years of life, as well as ARD through skin prick test and allergen-specific Immunoglobulin E (asIgE) at 9 years of age. At each year of life, children were divided into two groups (latent classes) defined by their infection-exposure history in that year and probabilities of ARDs outcomes in these groups were compared. Authors concluded that early childhood infection-exposure to malaria and diarrhoea was associated with lower prevalence of ARDs in later childhood.

Strengths:

1. The cohort with a large number of patients originating from 4 different geographical areas and followed throughout the first 5 years of age, with a detailed follow-up at 9 years of age.

2. The amount of data obtained and available for each patient, which allows an extended analysis and comparison.

Weaknesses:

The major weaknesses of the paper concern the statistics applied.

1. Analysis of the exposure: It is not the best approach to define exposure classes for each year and separately analyze their associations with outcomes at 9 years. Outcomes could be affected by any/all exposures in years 1-5, so analysis of a single year is likely to be confounded by exposure in the other years and provides biased estimates. Additionally, many children were likely to be exposed to some infections multiple numbers of time. It is unclear how this was handled in the analysis. Not accounting for the multiple exposure could also bias the results as, under the study hypothesis, children with multiple exposure would be those mostly protected from the ARDs.

We thank the reviewers for this comment. We have now analysed exposure over 5 years altogether and present these findings in the updated manuscript. We also report results for first year exposure, considering the importance of infancy in immune development following exposure to several pathogens for the very first time.

Accounting for multiple exposure to infections in the study participants is challenging under the LCA approach. Since we are using several indicator infections with some only captured based on data from the scheduled annual visits (HSV, norovirus, CMV and helminths), while others were recorded every time a participant was presented to the study clinic (malaria, diarrhoea, LRTI and URTI), it was not possible to account for multiplicity for some and not for others in a consistent way. We have added this as a potential limitation in the Discussion section (lines 450 – 455).

2. Analysis of the association with outcome: The association between exposure and ARDs is studied in separated from other possible risk factors for ARDs, such as genetic atopic predisposition, early childhood allergen exposure and sensitization, maternal smoking during pregnancy, poor dietary factors, lack of breast-feeding, childhood obesity, having a certain immunologic predisposition, air pollution, and frequent immunizations in childhood. These risk factors as well as ARDs status at 1-5 years are crucial in understanding the association between infection exposure and development of ARDs. This needs to be discussed.

We did not analyse associations between exposure and ARDs separately from other risk factors. The LCA approach allows for incorporation of covariates by way of establishing the effect of covariates on latent class membership. In our original analysis, we incorporated the covariates maternal area of residence, and maternal and household socio-economic status (SES) at enrolment. We have now additionally included maternal history of asthma and eczema as a proxy for genetic atopic predisposition. We appreciate that the other factors highlighted by the reviewer could be relevant, however, we did not have data on genetic atopic predisposition (other than as maternal history), early childhood allergen sensitisation was only assessed for a subset of participants at 3 years, and some factors (e.g. maternal smoking, lack of breast-feeding, childhood obesity) were universally uncommon in this setting, other factors (e.g. childhood immunisations which were conducted at the study clinic) were universally common. Air pollution may have differed by maternal area of residence, which was considered, but was not otherwise assessed. Effects of covariates on LC membership are highlighted in the manuscript (lines 299 – 309).

3. Interpretation of results: Possible bias in results aside, they are very difficult to interpret. Results of the analyses are inconsistent across years. Differences between classes with respect to exposure patterns and probabilities of outcomes are rather small, therefore it is impossible to draw any sound conclusions on the association (or lack of it) between exposures and outcomes. Authors' strong conclusion is in contradiction to the above, but also to their own assessment of weaknesses of the study presented in the discussion. The interpretation of the results was primary based on p-values, when multiple testing was certainly an issue with 10 outcomes and repeated analysis for each of 5 years.

We thank the reviewers for this comment. We believe that since we have now re-analysed the exposure data over 5 years, the results should now be simpler to interpret.

Regarding the issue of multiple testing, there are fewer tests now since we have restricted the analysis to only the first year and then cumulative infection experience over the five years. Secondly, we employed the LCA approach because it helps in dealing with the issue of multiple related exposures. Thirdly, regarding the 10 atopy and ARD outcomes, we think that it would be overly conservative if we consider adjustment for multiplicity because these outcomes are not a random collection, they were selected because there is prior evidence for their associations with infection exposure and they are mostly correlated with each other.

As the statistical analysis is considered the major weakness these issues are listed first.

1. We think that the data should have been analyzed together, not split by year. We do not understand why single LCA exploring exposure over 5 years could not be performed. We think LTA, which authors were planning to conduct in the first place, would not be the right approach as it explores changes in the exposure over time – while the objective of this analysis was to relate the exposure in the first years to ARDs at nine years of age. Latent classes defined over 5 years would account for any changes in exposures.

We thank the reviewers for this comment. We have now analysed exposure over 5 years altogether. We also report results for first year LCA, considering the importance of infancy in immune development following exposure to several pathogens for the very first time. We have also removed all the text relating to LTA from the paper.

2. The standard procedure for conducting LCA is to conduct a sequence of models, starting with a one-class model and then specifying models with one additional class at a time. Restriction of the analysis to {greater than or equal to}2 classes is very likely to produce some "significant" associations, especially when multiple testing is performed.

We thank the reviewers for this comment. Indeed, we conducted a sequence of models, considering additional classes gradually and using Bayesian Information Criterion (BIC), adjusted BIC (ABIC), consistent Akaike Information Criterion (CAIC) and entropy to guide the choice of optimal number of classes (without reference to the outcome variables). We considered models with 2 classes or more because we needed to account for the effects of covariates on latent class membership, otherwise this is not possible with one-class models which again would not require the use of LCA analysis.

3. The analysis cohort is not described at all, which is important especially in the context of risk factors for ARDs and the history of ARDs.

We thank the reviewers for this comment. We have now included more information about the analysis cohort (lines 198 – 209).

4. Exposure to an infection is presented as a proportion of children with that infection per year, but not presented per child. For each infection, what was the incidence rate and its variability between children? It is unclear how exposure was captured for each child in the analysis – as an incidence or as `at least one infection'? Incidence could have been a better representation of the exposure.

We describe exposure as at least one infection reported to the study clinic during the given period for malaria, diarrhoea, LRTI and URTI. For HSV, CMV, Norovirus and any helminths infections, samples collected at the annual visits were assessed. The age of seroconversion was determined for HSV, CMV and norovirus. Exposure to helminth infections was determined based on testing positive at any annual visit. As such, we could not use incidence for representation of the exposure, hence the use of prevalence estimates indicating cumulative infection at any time during the five years.

5 Other risk factors for ARDs should have been considered in the analysis otherwise the association between exposure and outcome could be seriously biased.

The LCA approach allows for incorporation of covariates by way of establishing the effect of covariates on latent class membership. In our original analysis, we incorporated the covariates maternal area of residence, and maternal and household socio-economic status (SES) at enrolment. We have now additionally included maternal history of asthma and eczema as a proxy for genetic atopic predisposition. Indeed, inclusion of covariates in these models leads to different results as compared to models without covariates; we cannot rule out the possibility of residual confounding, which we have now highlighted in the discussion (lines 462 – 463).

6. It would be worth summarizing the effects of the RCTs on allergy outcomes to 9 years in the introduction, so that the reader is informed of any potential effects of treatment arms on study outcomes, given that the analysis does not take into account the original design. It might also be worth doing sub-group analyses among control-arms of both RCTs to see if the results of those, albeit likely underpowered analyses, are consistent with the whole-group analysis such that treatment allocations can be safely ignored. It would be useful to provide some discussion in the limitations section of the potential effects, if any, of these RCTs on study findings.

Thank you for this comment. The effects of the RCT on allergy outcomes at 9 years were described in our earlier report (Namara et al., Pediatric Allergy and Immunology 2017; 28(8): 784-792). We have now briefly mentioned these results in this manuscript (lines 75 – 76). We can confirm that because treatment allocation had no effect on the outcomes at nine years, it can be safely ignored for this paper.

7. The cohort is composed of 2345 children, but only 1179 have data on ARD at nine years of age. As the goal is to investigate a possible association between infection exposure in early life and susceptibility to ARDs, shouldn't the number of analyzed subjects be 1179 in total? The authors should either repeat the LTA or discuss the difference in population samples (2345 vs 1179) as a weakness/limitation of their study.

Thank you for this comment. In the analysis plan which was developed before data analysis, we specified that we would use all available data to define latent classes. We felt that this was appropriate because it would describe the infection exposure experience of the whole cohort, before going ahead to assess associations between this infection-exposure experience and the 9 year outcomes. We have included a few sentences in the Discussion section discussing this and whether it could have introduced bias (lines 455 – 461). Here we have referenced two earlier papers which reported that characteristics of those seen at 9 years were similar to those not seen in terms of maternal marital status, BMI and worm infection at enrolment, and child's sex, eczema diagnosis (before age 1 year) and atopy (at 3 years) (lines 204 – 209).

8. The authors describe the ARD information methodology and state (page 6, lines 122-124) that they used 20-fold diluted plasma, thus significantly reducing the sensitivity of their test. IgE tests are normally done on undiluted serum. The reason for diluting plasma 20x should be given. Moreover, information concerning validation of this assay against a gold standard (e.g. immunoCAP, TFS) should be at least mentioned here (just a reference is not enough). Finally, what SPT wheal (not papule!) diameter corresponds to the threshold of 312.5 ng/ml? Is the IgE test sensitive enough?

It is true that standard IgE tests such as ImmunoCAP normally use undiluted serum/plasma. For logistical reasons, we used an in-house ELISA assay to detect IgE, using native IgE from human myeloma plasma as a standard. Because of the dynamic range of our in-house assay, it was not possible to use undiluted plasma, hence we optimised our assay to work with 20-fold diluted samples, with a lower detection limit of 15.625 ng/ml. As our assay is not a standard assay, we do not use it to classify sensitised vs non-sensitised individuals, rather, we use it to define detectable vs undetectable IgE levels. We also do not use it to determine what SPT wheal size corresponds to the lower detection limit. We have previously shown that results from our in-house ELISA were positively correlated with ImmunoCAP results for both dust mite and cockroach (Sanya et al., Clinical infectious diseases 2019; 68(10): 1665-1674). We have now added some of the above explanation to the manuscript text (lines 129 – 133).

9. Plasma samples were only tested against Dermatophagoides pteronyssinus and Blatella germanica. Please give information why this choice has been made. Why were only these two allergen extract examined with IgE assays?

We have previously shown that house dust mites (Dermatophagoides spp, Blomia tropicalis) and cockroach allergens are the commonest in this study setting (Mpairwe et al., Trans R Soc Trop Med Hyg. 2008;102(4):367-37). Our recent studies using the ISAC allergen microarray confirm this (Nkurunungi et al., Allergy. 2020;76(1):233-246.) Therefore, taking into account the budgetary constraints, we chose to test dust mite- and cockroach-specific IgE. We have now added this further explanation to the manuscript text (lines 126 – 127).

10. Please insert more information to help the interpretation of data (the favored models) of table 1 (page 11) and consider to move the table in an electronic repository, considering that this table is occupying one page and the respective information is not the major focus of the paper.

We thank the reviewers for this comment. We have now moved the former table 1 to supplementary material (Supplementary table S1). We have also added some more information to aid the interpretation of this result (lines 236 – 245).

11. It seems that children from Entebbe were compared against the other 3 areas, but these were not compared among themselves. Please provide a brief explanation for this choice.

We chose to have Entebbe as the reference group because it was the largest group (and also the most urban location).

12. The LCA model is fitted separately at each time point: 1-5 years of age. This provides a lot of information, but can also originate some confusion: children born in Kigungu are less likely to be in LC 1 – and therefore have a higher probability for various infections – when they are 2 years of age, but the opposite occurs when they are 4 years of age (page 14, lines 242-247). This age dependency of LTA categorization needs to be discussed: is it a limitation of the study?

Thanks for raising this issue, we have now re-analysed all the 5 years’ data together, but also considered the first year analysis. We believe that this approach has simplified the results and eliminated the confusion.

13. SPT or IgE positivity is not equivalent of ARD. Please reformulate the definition of atopic and/or allergic outcome at 9y of age in the whole paper.

We thank the reviewers for this observation. We have now distinguished ARDs and atopy (SPT or asIgE) throughout the paper.

14. A table showing the prevalence of SPT/IgE positivity for each allergen extract should be given somewhere in the results. In particular, only a few extracts (blomia, mites and cockroach) are used in the presentation of the table 4. However, many more allergen extracts have been examined. Probably, the prevalence of e.g. pollen sensitization was low. However, this is not shown in the paper.

We thank the reviewers for this observation. Prevalence of SPT/IgE positivity for each allergen extract has been reported before in our earlier paper (Lule et al., 2017. Life-course of atopy and allergy-related disease events in tropical sub-Saharan Africa: A birth cohort study). We have briefly reported the same results again, in this manuscript (lines 326 – 333), and have included the prevalences of SPT positivity for peanut, Bermuda grass, cat, pollen and mould which we had initially skipped (lines 331 – 332). Indeed, prevalence of SPT positivity for each of these allergen extracts was less than 1.5%.

https://doi.org/10.7554/eLife.66022.sa2

Article and author information

Author details

  1. Lawrence Lubyayi

    1. Immuno-modulation and Vaccines Programme, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine Uganda Research Unit, Entebbe, Uganda
    2. Division of Epidemiology and Biostatistics, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa, Johannesburg, South Africa
    Contribution
    Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing - original draft, Writing – review and editing
    For correspondence
    lawrencelby@gmail.com
    Competing interests
    none
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5286-6488
  2. Harriet Mpairwe

    1. Immuno-modulation and Vaccines Programme, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine Uganda Research Unit, Entebbe, Uganda
    2. Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    None
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1199-4859
  3. Gyaviira Nkurunungi

    1. Immuno-modulation and Vaccines Programme, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine Uganda Research Unit, Entebbe, Uganda
    2. Department of Infection Biology, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    none
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4062-9105
  4. Swaib A Lule

    Institute for Global Health, University College London, London, United Kingdom
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    none
  5. Angela Nalwoga

    Immuno-modulation and Vaccines Programme, Medical Research Council/Uganda Virus Research Institute and London School of Hygiene and Tropical Medicine Uganda Research Unit, Entebbe, Uganda
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    none
  6. Emily L Webb

    MRC International Statistics and Epidemiology Group, Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Contribution
    Conceptualization, Investigation, Methodology, Supervision, Writing – review and editing
    Competing interests
    none
  7. Jonathan Levin

    Division of Epidemiology and Biostatistics, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa, Johannesburg, South Africa
    Contribution
    Conceptualization, Methodology, Supervision, Writing – review and editing
    Competing interests
    none
  8. Alison M Elliott

    Department of Clinical Research, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Contribution
    Conceptualization, Funding acquisition, Investigation, Supervision, Writing – review and editing
    Competing interests
    none
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2818-9549

Funding

African Academy of Sciences (DELTAS Africa Initiative SSACAB PhD Fellowship 107754)

  • Lawrence Lubyayi

Wellcome Trust (Senior Fellowship 064693 Senior Fellowship 079110 Senior Fellowship 95778)

  • Alison M Elliott

Medical Research Council (MR/K012126/1)

  • Emily L Webb

PANDORA-ID-NET Consortium (RIA2016E-1609)

  • Swaib A Lule

Wellcome Trust (Institutional Strategic Support 204928/Z/16/Z)

  • Harriet Mpairwe

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the study participants of the EMaBS, the staff of the Immunomodulation and Vaccines (I-Vac) Programme at the MRC/UVRI and LSHTM Uganda Research Unit, the midwives of the Entebbe General Hospital Maternity Department and the community field workers in Entebbe municipality and Katabi sub county.

Ethics

Human subjects: Parents or guardians of the children provided written informed consent, and children eight years or older provided written informed assent. This consent was to participate in the study, and to publish anonymous results. The study was approved by ethics committees of the Uganda Virus Research Institute [reference number GC/127], London School of Hygiene and Tropical Medicine [application number 790] and Uganda National Council for Science and Technology [reference number MV 625].

Senior and Reviewing Editor

  1. Jos W Van der Meer, Radboud University Medical Centre, Netherlands

Publication history

  1. Received: December 22, 2020
  2. Accepted: September 16, 2021
  3. Version of Record published: September 22, 2021 (version 1)

Copyright

© 2021, Lubyayi et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 272
    Page views
  • 23
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Epidemiology and Global Health
    2. Microbiology and Infectious Disease
    Paul Z Chen et al.
    Research Advance Updated

    Background:

    Previously, we conducted a systematic review and analyzed the respiratory kinetics of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (Chen et al., 2021). How age, sex, and coronavirus disease 2019 (COVID-19) severity interplay to influence the shedding dynamics of SARS-CoV-2, however, remains poorly understood.

    Methods:

    We updated our systematic dataset, collected individual case characteristics, and conducted stratified analyses of SARS-CoV-2 shedding dynamics in the upper (URT) and lower respiratory tract (LRT) across COVID-19 severity, sex, and age groups (aged 0–17 years, 18–59 years, and 60 years or older).

    Results:

    The systematic dataset included 1266 adults and 136 children with COVID-19. Our analyses indicated that high, persistent LRT shedding of SARS-CoV-2 characterized severe COVID-19 in adults. Severe cases tended to show slightly higher URT shedding post-symptom onset, but similar rates of viral clearance, when compared to nonsevere infections. After stratifying for disease severity, sex and age (including child vs. adult) were not predictive of respiratory shedding. The estimated accuracy for using LRT shedding as a prognostic indicator for COVID-19 severity was up to 81%, whereas it was up to 65% for URT shedding.

    Conclusions:

    Virological factors, especially in the LRT, facilitate the pathogenesis of severe COVID-19. Disease severity, rather than sex or age, predicts SARS-CoV-2 kinetics. LRT viral load may prognosticate COVID-19 severity in patients before the timing of deterioration and should do so more accurately than URT viral load.

    Funding:

    Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant, NSERC Senior Industrial Research Chair, and the Toronto COVID-19 Action Fund.

    1. Epidemiology and Global Health
    Audrie Lin et al.
    Research Advance Updated

    Background:

    Previously, we demonstrated that a water, sanitation, handwashing, and nutritional intervention improved linear growth and was unexpectedly associated with shortened childhood telomere length (TL) (Lin et al., 2017). Here, we assessed the association between TL and growth.

    Methods:

    We measured relative TL in whole blood from 713 children. We reported differences between the 10th percentile and 90th percentile of TL or change in TL distribution using generalized additive models, adjusted for potential confounders.

    Results:

    In cross-sectional analyses, long TL was associated with a higher length-for-age Z score at age 1 year (0.23 SD adjusted difference in length-for-age Z score [95% CI 0.05, 0.42; FDR-corrected p-value = 0.01]). TL was not associated with other outcomes.

    Conclusions:

    Consistent with the metabolic telomere attrition hypothesis, our previous trial findings support an adaptive role for telomere attrition, whereby active TL regulation is employed as a strategy to address ‘emergency states’ with increased energy requirements such as rapid growth during the first year of life. Although short periods of active telomere attrition may be essential to promote growth, this study suggests that a longer overall initial TL setting in the first 2 years of life could signal increased resilience against future telomere erosion events and healthy growth trajectories.

    Funding:

    Funded by the Bill and Melinda Gates Foundation.

    Clinical trial number:

    NCT01590095