Age-dependent acquisition of IgG antibodies to Shigella serotypes-a retrospective analysis of seroprevalence in Kenyan children with implications for infant vaccination

Background

Shigellosis mainly affects children under 5 years of age living in low- and middle-income countries, who are the target population for vaccination. There are, however, limited data available to define the appropriate timing for vaccine administration in this age group. Information on antibody responses following natural infection, proxy for exposure, could help guide vaccination strategies.

Methods

We undertook a retrospective analysis of antibodies to five of the most prevalent Shigella serotypes among children aged <5 years in Kenya. Serum samples from a cross-sectional serosurvey in three Kenyan sites (Nairobi, Siaya, and Kilifi) were analyzed by standardized ELISA to measure IgG against Shigella sonnei and Shigella flexneri 1b, 2a, 3a, and 6. We identified factors associated with seropositivity to each Shigella serotype, including seropositivity to other Shigella serotypes.

Results

A total of 474 samples, one for each participant, were analyzed: Nairobi (n = 169), Siaya (n = 185), and Kilifi (n = 120). The median age of the participants was 13.4 months (IQR 7.0-35.6), and the male:female ratio was 1:1. Geometric mean concentrations (GMCs) for each serotype increased with age, mostly in the second year of life. The overall seroprevalence of IgG antibodies increased with age except for S. flexneri 6 which was high across all age subgroups. In the second year of life, there was a statistically significant increase of antibody GMCs against all five serotypes (p = 0.01-0.0001) and a significant increase of seroprevalence for S. flexneri 2a (p = 0.006), S. flexneri 3a (p = 0.006), and S. sonnei (p = 0.05) compared with the second part of the first year of life. Among all possible pairwise comparisons of antibody seropositivity, there was a significant association between S. flexneri 1b and 2a (OR = 6.75, 95% CI 3-14, p < 0.001) and between S. flexneri 1b and 3a (OR = 23.85, 95% CI 11-54, p < 0.001).

Conclusion

Children living in low- and middle-income settings such as Kenya are exposed to Shigella infection starting from the first year of life and acquire serotype-specific antibodies against multiple serotypes. The data from this study suggest that Shigella vaccination should be targeted to infants, ideally at 6 or at least 9 months of age, to ensure children are protected in the second year of life when exposure significantly increases.

Decision-making process for introduction of maternal vaccines in Kenya, 2017-2018

BACKGROUND

Maternal immunization is a key strategy for reducing morbidity and mortality associated with infectious diseases in mothers and their newborns. Recent developments in the science and safety of maternal vaccinations have made possible development of new maternal vaccines ready for introduction in low- and middle-income countries. Decisions at the policy level remain the entry point for maternal immunization programs. We describe the policy and decision-making process in Kenya for the introduction of new vaccines, with particular emphasis on maternal vaccines, and identify opportunities to improve vaccine policy formulation and implementation process.

METHODS

We conducted 29 formal interviews with government officials and policy makers, including high-level officials at the Kenya National Immunization Technical Advisory Group, and Ministry of Health officials at national and county levels. All interviews were recorded and transcribed. We analyzed the qualitative data using NVivo 11.0 software.

RESULTS

All key informants understood the vaccine policy formulation and implementation processes, although national officials appeared more informed compared to county officials. County officials reported feeling left out of policy development. The recent health system decentralization had both positive and negative impacts on the policy process; however, the negative impacts outweighed the positive impacts. Other factors outside vaccine policy environment such as rumours, sociocultural practices, and anti-vaccine campaigns influenced the policy development and implementation process.

CONCLUSIONS

Public policy development process is complex and multifaceted by its nature. As Kenya prepares for introduction of other maternal vaccines, it is important that the identified policy gaps and challenges are addressed.

Respiratory syncytial virus seasonality in three epidemiological zones of Kenya

Understanding respiratory syncytial virus (RSV) circulation patterns is necessary to guide the timing of limited‐duration interventions such as vaccines. We describe RSV circulation over multiple seasons in three distinct counties of Kenya during 2006‐2018. Kilifi and Siaya counties each had consistent but distinct RSV seasonality, lasting on average 18‐22 weeks. Based on data from available years, RSV did not have a clear pattern of circulation in Nairobi. This information can help guide the timing of vaccines and immunoprophylaxis products that are under development.

Indirect Effects of 10-Valent Pneumococcal Conjugate Vaccine Against Adult Pneumococcal Pneumonia in Rural Western Kenya

BACKGROUND

Data on pneumococcal conjugate vaccine (PCV) indirect effects in low-income countries with high human immunodeficiency virus (HIV) burden are limited. We examined adult pneumococcal pneumonia incidence before and after PCV introduction in Kenya in 2011.

METHODS

From 1 January 2008 to 31 December 2016, we conducted surveillance for acute respiratory infection (ARI) among ~12 000 adults (≥18 years) in western Kenya, where HIV prevalence is ~17%. ARI cases (cough or difficulty breathing or chest pain, plus temperature ≥38.0°C or oxygen saturation <90%) presenting to a clinic underwent blood culture and pneumococcal urine antigen testing (UAT). We calculated ARI incidence and adjusted for healthcare seeking. The proportion of ARI cases with pneumococcus detected among those with complete testing (blood culture and UAT) was multiplied by adjusted ARI incidence to estimate pneumococcal pneumonia incidence.

RESULTS

Pre-PCV (2008-2010) crude and adjusted ARI incidences were 3.14 and 5.30/100 person-years-observation (pyo), respectively. Among ARI cases, 39.0% (340/872) had both blood culture and UAT; 21.2% (72/340) had pneumococcus detected, yielding a baseline pneumococcal pneumonia incidence of 1.12/100 pyo (95% confidence interval [CI]: 1.0-1.3). In each post-PCV year (2012-2016), the incidence was significantly lower than baseline; with incidence rate ratios (IRRs) of 0.53 (95% CI: 0.31-0.61) in 2012 and 0.13 (95% CI: 0.09-0.17) in 2016. Similar declines were observed in HIV-infected (IRR: 0.13; 95% CI: 0.08-0.22) and HIV-uninfected (IRR: 0.10; 95% CI: 0.05-0.20) adults.

CONCLUSIONS

Adult pneumococcal pneumonia declined in western Kenya following PCV introduction, likely reflecting vaccine indirect effects. Evidence of herd protection is critical for guiding PCV policy decisions in resource-constrained areas.

Investigating the meat pathway as a source of human nontyphoidal Salmonella bloodstream infections and diarrhea in East Africa

Background

Salmonella Enteritidis and Salmonella Typhimurium are major causes of bloodstream infection and diarrheal disease in East Africa. Sources of human infection, including the role of the meat pathway, are poorly understood.

Methods

We collected cattle, goat, and poultry meat pathway samples from December 2015 through August 2017 in Tanzania and isolated Salmonella using standard methods. Meat pathway isolates were compared with nontyphoidal serovars of Salmonella enterica (NTS) isolated from persons with bloodstream infections and diarrheal disease from 2007 through 2017 from Kenya by core genome multi-locus sequence typing (cgMLST). Isolates were characterized for antimicrobial resistance, virulence genes, and diversity.

Results

We isolated NTS from 164 meat pathway samples. Of 172 human NTS isolates, 90 (52.3%) from stool and 82 (47.7%) from blood, 53 (30.8%) were Salmonella Enteritidis sequence type (ST) 11 and 62 (36.0%) were Salmonella Typhimurium ST313. We identified cgMLST clusters within Salmonella Enteritidis ST11, Salmonella Heidelberg ST15, Salmonella Typhimurium ST19, and Salmonella II 42:r:- ST1208 that included both human and meat pathway isolates. Salmonella Typhimurium ST313 was isolated exclusively from human samples. Human and poultry isolates bore more antimicrobial resistance and virulence genes and were less diverse than isolates from other sources.

Conclusions

Our findings suggest that the meat pathway may be an important source of human infection with some clades of Salmonella Enteritidis ST11 in East Africa, but not of human infection by Salmonella Typhimurium ST313. Research is needed to systematically examine the contributions of other types of meat, animal products, produce, water, and the environment to nontyphoidal Salmonella disease in East Africa.

Rotavirus group A genotype circulation patterns across Kenya before and after nationwide vaccine introduction, 2010-2018

BACKGROUND

Kenya introduced the monovalent G1P [8] Rotarix® vaccine into the infant immunization schedule in July 2014. We examined trends in rotavirus group A (RVA) genotype distribution pre- (January 2010-June 2014) and post- (July 2014-December 2018) RVA vaccine introduction.

METHODS

Stool samples were collected from children aged < 13 years from four surveillance sites across Kenya: Kilifi County Hospital, Tabitha Clinic Nairobi, Lwak Mission Hospital, and Siaya County Referral Hospital (children aged < 5 years only). Samples were screened for RVA using enzyme linked immunosorbent assay (ELISA) and VP7 and VP4 genes sequenced to infer genotypes.

RESULTS

We genotyped 614 samples in pre-vaccine and 261 in post-vaccine introduction periods. During the pre-vaccine introduction period, the most frequent RVA genotypes were G1P [8] (45.8%), G8P [4] (15.8%), G9P [8] (13.2%), G2P [4] (7.0%) and G3P [6] (3.1%). In the post-vaccine introduction period, the most frequent genotypes were G1P [8] (52.1%), G2P [4] (20.7%) and G3P [8] (16.1%). Predominant genotypes varied by year and site in both pre and post-vaccine periods. Temporal genotype patterns showed an increase in prevalence of vaccine heterotypic genotypes, such as the commonly DS-1-like G2P [4] (7.0 to 20.7%, P < .001) and G3P [8] (1.3 to 16.1%, P < .001) genotypes in the post-vaccine introduction period. Additionally, we observed a decline in prevalence of genotypes G8P [4] (15.8 to 0.4%, P < .001) and G9P [8] (13.2 to 5.4%, P < .001) in the post-vaccine introduction period. Phylogenetic analysis of genotype G1P [8], revealed circulation of strains of lineages G1-I, G1-II and P [8]-1, P [8]-III and P [8]-IV. Considerable genetic diversity was observed between the pre and post-vaccine strains, evidenced by distinct clusters.

CONCLUSION

Genotype prevalence varied from before to after vaccine introduction. Such observations emphasize the need for long-term surveillance to monitor vaccine impact. These changes may represent natural secular variation or possible immuno-epidemiological changes arising from the introduction of the vaccine. Full genome sequencing could provide insights into post-vaccine evolutionary pressures and antigenic diversity.

Integrated point-of-care testing (POCT) of HIV, syphilis, malaria and anaemia in antenatal clinics in western Kenya: A longitudinal implementation study

Background

In sub-Saharan Africa, HIV, syphilis, malaria and anaemia are leading preventable causes of adverse pregnancy outcomes. In Kenya, policy states women should be tested for all four conditions (malaria only if febrile) at first antenatal care (ANC) visit. In practice, while HIV screening is conducted, coverage of screening for the others is suboptimal and early pregnancy management of illnesses is compromised. This is particularly evident at rural dispensaries that lack laboratories and have parallel programmes for HIV, reproductive health and malaria, resulting in fractured and inadequate care for women.

Methods

A longitudinal eight-month implementation study integrating point-of-care diagnostic tests for the four conditions into routine ANC was conducted in seven purposively selected dispensaries in western Kenya. Testing proficiency of healthcare workers was observed at initial training and at three monthly intervals thereafter. Adoption of testing was compared using ANC register data 8.5 months before and eight months during the intervention. Fidelity to clinical management guidelines was determined by client exit interviews with success defined as ≥90% adherence.

Findings

For first ANC visits at baseline (n = 529), testing rates were unavailable for malaria, low for syphilis (4.3%) and anaemia (27.8%), and near universal for HIV (99%). During intervention, over 95% of first attendees (n = 586) completed four tests and of those tested positive, 70.6% received penicillin or erythromycin for syphilis, 65.5% and 48.3% received cotrimoxazole and antiretrovirals respectively for HIV, and 76.4% received artemether/lumefantrine, quinine or dihydroartemisinin–piperaquine correctly for malaria. Iron and folic supplements were given to nearly 90% of women but often at incorrect doses.

Conclusions

Integrating point-of-care testing into ANC at dispensaries with established HIV testing programmes resulted in a significant increase in testing rates, without disturbing HIV testing rates. While more cases were detected and treated, treatment fidelity still requires strengthening and an integrated monitoring and evaluation system needs to be established.

Evaluation of case definitions to detect respiratory syncytial virus infection in hospitalized children below 5 years in Rural Western Kenya, 2009-2013

Background

In order to better understand respiratory syncytial virus (RSV) epidemiology and burden in tropical Africa, optimal case definitions for detection of RSV cases need to be identified.

Methods

We used data collected between September 2009 - August 2013 from children aged <5 years hospitalized with acute respiratory Illness at Siaya County Referral Hospital. We evaluated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of individual signs, symptoms and standard respiratory disease case definitions (severe acute respiratory illness [SARI]; hospitalized influenza-like illness [hILI]; integrated management of childhood illness [IMCI] pneumonia) to detect laboratory-confirmed RSV infection. We also evaluated an alternative case definition of cough or difficulty breathing plus hypoxia, in-drawing, or wheeze.

Results

Among 4714 children hospitalized with ARI, 3810 (81 %) were tested for RSV; and 470 (12 %) were positive. Among individual signs and symptoms, cough alone had the highest sensitivity to detect laboratory-confirmed RSV [96 %, 95 % CI (95–98)]. Hypoxia, wheezing, stridor, nasal flaring and chest wall in-drawing had sensitivities ranging from 8 to 31 %, but had specificities >75 %. Of the standard respiratory case definitions, SARI had the highest sensitivity [83 %, 95 % CI (79–86)] whereas IMCI severe pneumonia had the highest specificity [91 %, 95 % CI (90–92)]. The alternative case definition (cough or difficulty breathing plus hypoxia, in-drawing, or wheeze) had a sensitivity of [55 %, 95 % CI (50–59)] and a specificity of [60 %, 95 % CI (59–62)]. The PPV for all case definitions and individual signs/symptoms ranged from 11 to 20 % while the negative predictive values were >87 %. When we stratified by age <1 year and 1- < 5 years, difficulty breathing, severe pneumonia and the alternative case definition were more sensitive in children aged <1 year [70 % vs. 54 %, p < 0.01], [19 % vs. 11 %, p = 0.01] and [66 % vs. 43 %, p < 0.01] respectively, while non-severe pneumonia was more sensitive [14 % vs. 26 %, p < 0.01] among children aged 1- < 5 years.

Conclusion

The sensitivity and specificity of different commonly used case definitions for detecting laboratory-confirmed RSV cases varied widely, while the positive predictive value was consistently low. Optimal choice of case definition will depend upon study context and research objectives.

Risk of Injection-Site Abscess among Infants Receiving a Preservative-Free, Two-Dose Vial Formulation of Pneumococcal Conjugate Vaccine in Kenya

There is a theoretical risk of adverse events following immunization with a preservative-free, 2-dose vial formulation of 10-valent-pneumococcal conjugate vaccine (PCV10). We set out to measure this risk. Four population-based surveillance sites in Kenya (total annual birth cohort of 11,500 infants) were used to conduct a 2-year post-introduction vaccine safety study of PCV10. Injection-site abscesses occurring within 7 days following vaccine administration were clinically diagnosed in all study sites (passive facility-based surveillance) and, also, detected by caregiver-reported symptoms of swelling plus discharge in two sites (active household-based surveillance). Abscess risk was expressed as the number of abscesses per 100,000 injections and was compared for the second vs first vial dose of PCV10 and for PCV10 vs pentavalent vaccine (comparator). A total of 58,288 PCV10 injections were recorded, including 24,054 and 19,702 identified as first and second vial doses, respectively (14,532 unknown vial dose). The risk ratio for abscess following injection with the second (41 per 100,000) vs first (33 per 100,000) vial dose of PCV10 was 1.22 (95% confidence interval [CI] 0.37–4.06). The comparator vaccine was changed from a 2-dose to 10-dose presentation midway through the study. The matched odds ratios for abscess following PCV10 were 1.00 (95% CI 0.12–8.56) and 0.27 (95% CI 0.14–0.54) when compared to the 2-dose and 10-dose pentavalent vaccine presentations, respectively. In Kenya immunization with PCV10 was not associated with an increased risk of injection site abscess, providing confidence that the vaccine may be safely used in Africa. The relatively higher risk of abscess following the 10-dose presentation of pentavalent vaccine merits further study.

Associations between presence of handwashing stations and soap in the home and diarrhoea and respiratory illness, in children less than five years old in rural western Kenya

OBJECTIVE

We tested whether soap presence in the home or a designated handwashing station was associated with diarrhoea and respiratory illness in Kenya.

METHODS

In April 2009, we observed presence of a handwashing station and soap in households participating in a longitudinal health surveillance system in rural Kenya. Diarrhoea and acute respiratory illness (ARI) in children < 5 years old were identified using parent-reported syndromic surveillance collected January-April 2009. We used multivariate generalised linear regression to estimate differences in prevalence of illness between households with and without the presence of soap in the home and a handwashing station.

RESULTS

Among 2547 children, prevalence of diarrhoea and ARI was 2.3 and 11.4 days per 100 child-days, respectively. Soap was observed in 97% of households. Children in households with soap had 1.3 fewer days of diarrhoea/100 child-days (95% CI -2.6, -0.1) than children in households without soap. ARI prevalence was not associated with presence of soap. A handwashing station was identified in 1.4% of households and was not associated with a difference in diarrhoea or ARI prevalence.

CONCLUSIONS

Soap presence in the home was significantly associated with reduced diarrhoea, but not ARI, in children in rural western Kenya. Whereas most households had soap in the home, almost none had a designated handwashing station, which may prevent handwashing at key times of hand contamination.

Epidemiology of respiratory syncytial virus infection in rural and urban Kenya

BACKGROUND

Information on the epidemiology of respiratory syncytial virus (RSV) infection in Africa is limited for crowded urban areas and for rural areas where the prevalence of malaria is high.

METHODS

At referral facilities in rural western Kenya and a Nairobi slum, we collected nasopharyngeal/oropharyngeal (NP/OP) swab specimens from patients with influenza-like illness (ILI) or severe acute respiratory illness (SARI) and from asymptomatic controls. Polymerase chain reaction assays were used for detection of viral pathogens. We calculated age-specific ratios of the odds of RSV detection among patients versus the odds among controls. Incidence was expressed as the number of episodes per 1000 person-years of observation.

RESULTS

Between March 2007 and February 2011, RSV was detected in 501 of 4012 NP/OP swab specimens (12.5%) from children and adults in the rural site and in 321 of 2744 NP/OP swab specimens (11.7%) from those in the urban site. Among children aged <5 years, RSV was detected more commonly among rural children with SARI (odds ratio [OR], 2.0; 95% confidence interval [CI], 1.2-3.3), urban children with SARI (OR, 8.5; 95% CI, 3.1-23.6), and urban children with ILI (OR, 3.4; 95% CI, 1.2-9.6), compared with controls. The incidence of RSV disease was highest among infants with SARI aged <1 year (86.9 and 62.8 episodes per 1000 person-years of observation in rural and urban sites, respectively).

CONCLUSIONS

An effective RSV vaccine would likely substantially reduce the burden of respiratory illness among children in rural and urban areas in Africa.

Caretakers' perception towards using zinc to treat childhood diarrhoea in rural western Kenya

Zinc treatment for diarrhoea can shorten the course and prevent future episodes among children worldwide. However, knowledge and acceptability of zinc among African mothers is unknown. We identified children aged 3 to 59 months, who had diarrhoea within the last three months and participated in a home-based zinc treatment study in rural Kenya. Caretakers of these children were enrolled in two groups; zinc-users and non-users. A structured questionnaire was administered to all caretakers, inquiring about knowledge and appropriate use of zinc. Questions on how much the caretakers were willing to pay for zinc were asked. Proportions were compared using Mantel-Haenszel test, and medians were compared using Wilcoxon Rank Sum test. Among 109 enrolled caretakers, 73 (67%) used zinc, and 36 (33%) did not. Sixty-four (88%) caretakers in zinc-user group reported satisfaction with zinc treatment. Caretakers in the zinc-user group more often correctly identified appropriate zinc treatment (98%-100%) than did those in the non-user group (64-72%, p<0.001). Caretakers in the zinc-user group answered more questions about zinc correctly or favourably (median 10 of 11) compared to those in the non-user group (median 6.3 of 11, p<0.001). Caretakers in the zinc-user group were willing to pay more for a course of zinc in the future than those in the non-user group (median US$ 0.26, p<0.001). Caretakers of children given zinc recently had favourable impressions on the therapy and were willing to pay for it in the future. Active promotion of zinc treatment in clinics and communities in Africa could lead to greater knowledge, acceptance, and demand for zinc.

Influenza and malaria coinfection among young children in western Kenya, 2009-2011

BACKGROUND

Although children <5 years old in sub-Saharan Africa are vulnerable to both malaria and influenza, little is known about coinfection.

METHODS

This retrospective, cross-sectional study in rural western Kenya examined outpatient visits and hospitalizations associated with febrile acute respiratory illness (ARI) during a 2-year period (July 2009-June 2011) in children <5 years old.

RESULTS

Across sites, 45% (149/331) of influenza-positive patients were coinfected with malaria, whereas only 6% (149/2408) of malaria-positive patients were coinfected with influenza. Depending on age, coinfection was present in 4%-8% of outpatient visits and 1%-3% of inpatient admissions for febrile ARI. Children with influenza were less likely than those without to have malaria (risk ratio [RR], 0.57-0.76 across sites and ages), and children with malaria were less likely than those without to have influenza (RR, 0.36-0.63). Among coinfected children aged 24-59 months, hospital length of stay was 2.7 and 2.8 days longer than influenza-only-infected children at the 2 sites, and 1.3 and 3.1 days longer than those with malaria only (all P < .01).

CONCLUSIONS

Coinfection with malaria and influenza was uncommon but associated with longer hospitalization than single infections among children 24-59 months of age.

The burden of common infectious disease syndromes at the clinic and household level from population-based surveillance in rural and urban Kenya

BACKGROUND

Characterizing infectious disease burden in Africa is important for prioritizing and targeting limited resources for curative and preventive services and monitoring the impact of interventions.

METHODS

From June 1, 2006 to May 31, 2008, we estimated rates of acute lower respiratory tract illness (ALRI), diarrhea and acute febrile illness (AFI) among >50,000 persons participating in population-based surveillance in impoverished, rural western Kenya (Asembo) and an informal settlement in Nairobi, Kenya (Kibera). Field workers visited households every two weeks, collecting recent illness information and performing limited exams. Participants could access free high-quality care in a designated referral clinic in each site. Incidence and longitudinal prevalence were calculated and compared using Poisson regression.

RESULTS

INCIDENCE RATES RESULTING IN CLINIC VISITATION WERE THE FOLLOWING: ALRI--0.36 and 0.51 episodes per year for children <5 years and 0.067 and 0.026 for persons ≥ 5 years in Asembo and Kibera, respectively; diarrhea--0.40 and 0.71 episodes per year for children <5 years and 0.09 and 0.062 for persons ≥ 5 years in Asembo and Kibera, respectively; AFI--0.17 and 0.09 episodes per year for children <5 years and 0.03 and 0.015 for persons ≥ 5 years in Asembo and Kibera, respectively. Annually, based on household visits, children <5 years in Asembo and Kibera had 60 and 27 cough days, 10 and 8 diarrhea days, and 37 and 11 fever days, respectively. Household-based rates were higher than clinic rates for diarrhea and AFI, this difference being several-fold greater in the rural than urban site.

CONCLUSIONS

Individuals in poor Kenyan communities still suffer from a high burden of infectious diseases, which likely hampers their development. Urban slum and rural disease incidence and clinic utilization are sufficiently disparate in Africa to warrant data from both settings for estimating burden and focusing interventions.

Evaluation of the optimal recall period for disease symptoms in home-based morbidity surveillance in rural and urban Kenya

Background In African settings with poor access to health care, surveillance and surveys of disease burden are often done through home visits. The optimal recall period to capture data on symptoms and health utilization is unknown.

Methods We collected illness data among 53 000 people during fortnightly home visits in rural and urban Kenya. Dates of cough, fever and diarrhoea in the past 2 weeks and health-seeking behaviour were recorded. Incidence rates were modelled using Poisson regression for data collected from 1 July 2006 to 30 June 2007.

Results Incidence rates were higher in days 0–6 before the home visit than in days 7–13 before the home visit for all three symptoms, for the rural and urban sites, for children and adults, for self- and proxy-reported symptoms and for severe and non-severe illness in children. Recall decay was steeper in the rural than the urban sites, and for proxy- than self-reported symptoms. The daily prevalence of symptoms fell <80% of the maximum prevalence when asking about symptoms >3 days before the home visit for children and >4 days for persons ≥5 years of age. Recall of previously documented clinic visits, and prescriptions of antimalarials and antibiotics also declined by ∼7, 15 and 23% per week, respectively, in children aged <5 years, and 6, 20 and 16%, respectively, in older persons (P < 0.0001 for each decline).

Conclusions A 2-week recall period underestimates true disease rates and health-care utilization. Shorter recall periods of 3 days in children and 4 days in adults would likely yield more accurate data.