Modifiable risk factors for diphtheria: A systematic review and meta-analysis

Objective To identify modifiable risk factors for diphtheria and assess their strengths of association with the disease. Methods This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement. Electronic databases and grey literature were searched from inception until January 2023. Studies had to report on diphtheria cases and estimates of association for at least one potential risk factor or sufficient data to calculate these. The quality of non-ecological studies was assessed using the Newcastle-Ottawa Scale (NOS), while the quality of evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria. Results The search yielded 37,705 papers, of which 29 were ultimately included. All the non-ecological studies were of moderate to high quality. Meta-analysis of 20 studies identified three factors increasing the risk of diphtheria: incomplete vaccination (<3 doses) (pooled odds ratio (POR) = 2.2, 95% confidence interval (CI) = 1.4–3.4); contact with a person with skin lesions (POR = 4.8, 95% CI = 2.1–10.9); and low knowledge of diphtheria (POR = 2.4, 95% CI = 1.2–4.7). Contact with a case of diphtheria; sharing a bed or bedroom; sharing utensils, cups, and glasses; infrequent bathing; and low parental education were associated with diphtheria in multiple studies. Evidence for other factors was inconclusive. The quality of evidence was low or very low for all the risk factors. Conclusions Findings from the review suggest that countries seeking to control diphtheria need to strengthen surveillance, improve vaccination coverage, and increase people's knowledge of the disease. Future research should focus on understudied or inconclusive risk factors.

Historically, diphtheria was a major cause of mortality, especially among children.In the 1970s, before the widespread use of vaccines, one million diphtheria cases (including 50,000-60,000 deaths) were estimated to occur annually in low-and middle-income countries (LMICs) [8].Following the launch in 1974 by the World Health Organization (WHO) of the Expanded Programme on Immunization, diphtheria incidence decreased by over 90% between 1980 and 2000 [8][9][10].
After a period of relative stability, there was a recrudescence of the disease, with 22,625 cases reported globally in 2019 -the highest number of cases since 1996, when 28,624 cases were notified [9].The resurgence of diphtheria has been fueled by large outbreaks in Bangladesh, Haiti, Venezuela, and Yemen [4,9,11].These outbreaks are a reminder of the continued threat posed by the disease to communities worldwide.The characteristics of diphtheria have been described thoroughly, including its prevalence, routes of transmission, and incubation period [4,8,10,12].However, to date, there has not been a comprehensive systematic review of its risk factors or meta-analysis of the strength of association between these factors and diphtheria.Increased knowledge on this subject would allow a more accurate detection of at-risk populations.It would also help develop more effective interventions, ultimately reducing the burden of disease.Therefore, we aimed to conduct a systematic review and meta-analysis that addresses the following two questions: what are the modifiable risk factors for diphtheria?What is the strength of association between these factors and the disease?

Methods
This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (Appendix 1) [13].Before starting the study, the protocol was registered with PROSPERO (CRD42019128741) [14].The authors received no specific funding for this study.

Search strategy
A literature search was conducted using EMBASE, MEDLINE, PubMed, and Web of Science.To identify grey literature, we consulted Gavi publications library, OpenGrey, and WHO Library Information System (WHOLIS).All databases were searched from inception until January 2023.No restrictions were applied.Native speakers translated studies in non-English languages.
The search strategy combined Medical Subject Headings (MeSH), free-text terms, and keywords related to diphtheria and risk factors.The strategy was developed for MEDLINE and adapted for EMBASE, PubMed, and Web of Science (Appendix 2).Only the term "diphtheria" was used for other databases.

Eligibility criteria
To be included, studies had to meet two criteria: report on diphtheria cases, as described by the authorsincluding individuals with a laboratory or clinical diagnosis of diphtheria [15]; and, present estimates of association for at least one potential risk factor or sufficient data to calculate these.In this study, a risk factor was defined as a modifiable variable associated with an increased likelihood of diphtheria.Eligible measures of association were odds ratios (ORs), risk ratios (RRs), alongside regression and correlation coefficients.
Studies that focused on non-modifiable risk factors for the disease (e. g., age, sex, and ethnicity) or on risk factors for diphtheria severity or mortality were deemed to be outside the scope of the review.Single case reports, qualitative studies, commentaries, editorials, economic analyses, letters, literature reviews, and animal studies were also excluded.

Study selection
A single reviewer (JI) exported all studies to EndNote X9 (Clarivate Analytics, Philadelphia, USA).The study selection then followed three stages.Firstly, one reviewer (JI) screened the titles of all papers to eliminate those that were clearly irrelevant to the review.Secondly, two reviewers (JI and BA) independently screened the abstracts of retained papers, excluding those that did not meet the eligibility criteria.Thirdly, both reviewers independently examined the full text of each study deemed potentially relevant based on the abstracts.References of included studies were screened to identify other relevant publications.Disagreements were resolved by discussion between the two reviewers and, when needed, referred to a third person (SL).

Data extraction
Two reviewers (JI and BA) independently extracted data from included studies using a standardized Excel spreadsheet (Microsoft, Redmond, United States).For each study, the following variables were recorded: last name of the first author, year of publication, country where the study was conducted, study design, number of cases, number of controls, definition of cases, definition of controls, age range of the sample, diagnostic method of diphtheria infection, considered confounders, data collection technique, and effect estimates.If studies did not report effect estimates, available data in those papers were used to calculate crude ORs from two-by-two tables.Where multiple estimates were provided, those adjusted for the most confounders were selected.The fully adjusted estimates were the most homogenous in terms of included confounders.This choice was made to reduce the impact of confounding as a source of heterogeneity.A sensitivity analysis was conducted to evaluate the impact of adjustment for confounding on the identified risk factor with the most evidence (i.e., incomplete vaccination).A sensitivity analysis was also performed to assess the impact of excluding the one preprint article on the pooled estimate for another risk factor (i.e., contact with a diphtheria case).

Quality assessment
Two reviewers (JI and WE) independently evaluated the methodological quality of case-control, cohort, and cross-sectional studies using the Newcastle-Ottawa Scale (NOS) [16,17].The evaluation was based on three main domains: selection of study groups, comparability between cases and controls, and ascertainment of the outcome of interest.For case-control and cohort studies, eight items were assigned one or two points, for a maximum score of nine.For cross-sectional studies, seven items received one or two points, for a maximum score of 10.Each score denoted a different level of quality: low (0-3), moderate (4-7), and high (≥8).
The quality of evidence for each risk factor across the studies was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria [18,19].The quality was rated very low, low, moderate, or high.In line with the GRADE approach, evidence from observational studies was initially rated by default as low quality due to residual confounding.Evidence could then be downgraded based on the consideration of five domains: risk of bias (as indicated by the NOS score), imprecision (if the confidence intervals were wide or the total population size was <400 -a threshold "rule-ofthumb" value), inconsistency (if there was variability in the direction or magnitude of effect across studies), indirectness (if there were factors relating to the study population that could limit generalizability), and publication bias (if there was a high probability of unreported studies).Criteria for upgrading evidence included a dose-response relationship or large effect size (OR ≥ 2 or OR ≤ 0.5).
For NOS and GRADE assessments, discrepancies between the two reviewers were resolved by consensus and, if needed, by adjudication of a third reviewer (SL).

Data analysis
Included studies were summarized in a qualitative synthesis.As diphtheria is a relatively rare event, ORs and RRs were considered equivalent [20,21].A meta-analysis was conducted if two or more studies captured the same risk factor but from different samples.The random effects model was chosen for the analysis due to variations in study methods and populations [22].Original data from the studies were combined to calculate pooled odds ratios (PORs) with 95% confidence intervals (CIs) using the Mantel-Haenszel method [23].If studies only presented adjusted data, PORs were estimated using the generic inverse weighted method [24].
Forest plots were generated to display individual and global estimates.Heterogeneity among studies was assessed using the I 2 statistic [25].Values of 25%, 50%, and 75% revealed low, medium, and high heterogeneity, respectively.All analyses were performed using Review Manager 5.3 (Cochrane Collaboration, Copenhagen, Denmark).

Results
After the initial search, 43 publications were selected for full-text review (Fig. 1).Fourteen were excluded as they did not provide measures of association and two because the investigated exposures were irrelevant.In addition to the remaining 27 publications, two other references were identified by hand searching.Twenty-nine papers were included in the review and 20 in the meta-analysis.Nine papers were excluded from the meta-analysis as they reported on risk factors that had not been investigated in other studies; therefore, their results could not be pooled with data from other publications.

Study characteristics
Table 1 describes the characteristics of the 29 included studies.Overall, there were 14 case-control studies, eight cross-sectional studies, six ecological studies, and one cohort study.All non-ecological studies were of moderate to high quality (Appendix 3).The studies covered a period from 1907 to 2021, with most of the studies (55%) focusing on outbreaks that occurred prior to the year 2000.Only one study reported on an outbreak with cases identified both before and after 2000.The studies were conducted in five WHO regions: Europe (nine), Americas (eight), South-East Asia (six), Eastern Mediterranean (three), and Western Pacific (three).No study was done in Africa.
Twenty-seven studies were written in English, one in Portuguese, and one in Bahasa Indonesia.Sample sizes (cases and controls combined) ranged from 37 to 221,018 (median = 364).Nine studies involved only children (age range = 0-14 years); one study only involved adults (age range = 40-49 years); six studies involved both children and adults; and thirteen studies did not indicate the age range of the sample.Where reported, the median proportion of male participants was 51% (range = 30-93%).Infected individuals were diphtheria cases (20), asymptomatic carriers (six), or a combination of the two (three).Diphtheria was ascertained by isolation of C. diphtheriae in culture (12), clinical examination (six), and using either one of the two methods (five); six studies did not specify the method of diagnosis.
Meta-analysis of two studies conducted in Russia in the mid-1990s [29,39] found no increased risk of diphtheria associated with not having received a booster vaccination in the preceding five years (POR = 3.6, 95% CI = 0.5-24.0);nevertheless, the point estimate was relatively high and the studies had limited power.Two other studies [34,47] also reported no increased risk of diphtheria associated with time since last booster vaccination; however, these studies did not present effect estimates, nor did they specify the intervals used for the estimations [34,47].Finally, data from a study conducted in Brazil in 1969 pointed towards an increased risk of diphtheria associated with having a partially vaccinated sibling (OR = 4.1, 95% CI = 1.8-9.0)[31].
Meta-analysis of four studies (two from Yemen [33,35], one from  Georgia [36], one from Indonesia [37]) revealed no increased risk of acquiring diphtheria following contact with a diphtheria case (POR = 5.0, 95% CI = 0.8-31.8).However, the pooled estimate appeared to be skewed by one studya preprint article, which also reported outlier values for other risk factors (e.g., incomplete vaccination, low diphtheria knowledge).After removing this study from the meta-analysis, contact with a diphtheria case was associated with an increased risk of diphtheria (Appendix 6).Meta-analysis of two [33,36] of the four studies suggest that contact with a person with skin lesions (POR = 4.8, 95% CI = 2.1-10.9)does increase the risk of diphtheria.A fifth study from Vietnam [34] found no increased risk of infection related to the two exposures; nevertheless, the study was excluded from the metaanalysis as no effect estimates or data to calculate these were provided.

Underlying conditions
Meta-analysis of two studies from Georgia [36] and Greece (which did not adjust for potential confounders) [48] found that neither having tonsils (POR = 2.0, 95% CI = 0.4-10.0)or having a recent sore throat (POR = 1.8, 95% CI = 0.8-4.0)increased the risk of diphtheria.The study from Georgia also reported a doubling of the odds of infection related to having a chronic illness (OR = 2.1, 95% CI = 1.3-5.5)[36], while a study from Vietnam found no association with the exposure [34]; nonetheless, the latter study did not provide effect estimates or data to calculate these.Additionally, in the study from Georgia [36], both history of eczema (OR = 3.4, 95% CI = 1.2-9.9)and recent fever with myalgia (OR = 2.7, 95% CI = 1.3-5.5)were associated with diphtheria.

Knowledge and behaviour
Meta-analysis of three recent studies from India [26] and Indonesia [32,37] showed that the odds of disease more than doubled for people with low knowledge of diphtheria (as assessed by a questionnaire) (POR = 2.4, 95% CI = 1.2-4.7).The study from India also reported an increased risk of diphtheria associated with believing that vaccines do not prevent diseases (OR = 4.0, 95% CI = 1.2-13.5)[26].
Meta-analysis of three studies from Georgia [36], Latvia (which did not adjust for potential confounders) [47], and Yemen [35] did not identify an increased risk of diphtheria associated with sharing utensils, cups, or glasses (POR = 1.7, 95% CI = 1.0-2.9).A fourth study from Vietnam [34] found no increased risk of infection related to this exposure; nevertheless, the study was excluded from the meta-analysis as it did not provide effect estimates or data to calculate these.Meta-analysis of four studies from Georgia [36], Laos [38], and Yemen [35] also found that sharing a bed or bedroom with other people did not increase the risk of diphtheria (POR = 1.3, 95% CI = 0.5-3.9).
Results relating to infrequent bathing were discordant.Two studies from Vietnam examined the association between diphtheria and bathing once a day or less (as opposed to twice a day or more).While one study found no association between the two variables (p = 0.40) [44], the other found that bathing once a day or less increased the odds of infection (OR = 1.7, 95% CI = 1.04-2.9)[34].Furthermore, a study from Georgia found that bathing less than once a week also increased the risk of diphtheria (OR = 2.6, 95% CI = 1.3-5.2) [36].
Meta-analysis of two studies from Indonesia [37] and Yemen [35] found no increased risk of infection associated with travel to an area with diphtheria cases (POR = 1.3, 95% CI = 0.5-3.9).A study conducted in Yemen in the 1980s reported an increase in the odds of diphtheria associated with obtaining water from a wheeled carrier (i.e., a person who carries water obtained from wells to people's houses) (OR = 28.4,p = 0.008) and consumption of factory-made yoghurt (OR = 14.9, p = 0.003) [33].
The study from Georgia also found that having ≥14 alcoholic drinks per week did not increase the risk of diphtheria (OR = 0.7, 95% CI = 0.3-1.7)[36]; by contrast, data from a study done in the 1970s in the United States (which did not adjust for potential confounders) indicated a higher risk of disease for "heavy consumers of alcoholic beverages" (OR = 48.8,95% CI = 27.2-87.6)[42].

Socioeconomic status
The only socioeconomic indicators that showed a potential association with diphtheria were related to the education level of parents.Metaanalysis of three studies from Indonesia [32], Georgia [36], and Yemen [35] suggested no increased odds of diphtheria associated with having a mother with primary education or less (POR = 1.5, 95% CI = 0.6-3.8).Similarly, meta-analysis of two studies from Indonesia [32] and Yemen [35] found no increased risk of diphtheria for children whose fathers had primary education or less (POR = 1.7, 95% CI = 0.2-14.9).

Population-level factors
Above-mentioned risk factors were assessed at the individual level.Instead, the following factors were examined at the population level in ecological studies comparing diphtheria incidence or case counts across different areas of the same country.
Four studies from Haiti [51], Indonesia [52], Ukraine [53], and Yemen [50] examined the association between diphtheria and vaccination coverage.The study from Indonesia identified a positive correlation between vaccination level and case count during the first year of a diphtheria outbreak (r = 0.42, p = 0.008); however, the following year, the two variables were no longer correlated (r = 0.22, p = 0.183).The study from Ukraine found a negative association between vaccination coverage and diphtheria incidence in six oblasts (ß = -0.04,p = 0.001).This result indicated that a unit increase in vaccination coverage was associated with a 4% average decrease in incidence rates, all else being equal.The study from Yemen showed that vaccination coverage affected the odds of a diphtheria outbreak (OR = 1.04; 95% CI =1.01-1.06,p = 0.002).In contrast with other articles, the study from Haiti did not find an association between diphtheria incidence and vaccination coverage (ß = 0.177; p > 0.05).The authors of the study explained that weaknesses in the country's immunization information systems and inaccuracies in the vaccination coverage estimates may have contributed to the observed lack of association between the two variables.
The association between diphtheria and population density was examined by four studies from Haiti [51], Indonesia [52], Ukraine [53], and the United States [54].Only the study from Ukraine found a positive association between diphtheria incidence and population density (ß = 0.04, p < 0.001) [53], whereas the other studies reported no association between the two variables.
A study that analyzed historical records from the early 20th century in the United States found a positive association between diphtheria case counts and reports of tuberculosis in four of five examined cities (β = 0.06-0.14,p < 0.05); no association was observed in the other city (β = 0.03, p = 0.41) [49].

Quality of evidence
The quality of evidence for the 32 identified risk factors varied from moderate to very low.Evidence for all risk factors was initially rated of low quality due to the observational nature of the included studies.Only the evidence for incomplete vaccination was upgraded from low to moderate given the strength of association with diphtheria observed in the meta-analysis and sensitivity analysis.Evidence for the other risk factors was often downgraded from low to very low due to concerns related to imprecision as most of the estimates came from a few studies, which in several instances had small sample sizes.Evidence for numerous risk factors was also downgraded due to inconsistencies in the direction of the effect across studies.The risk of methodological bias was considered serious only for population-level factors, whose evidence came from ecological studies.No risk of publication bias was detected for the 32 risk factors.Appendix 4 illustrates the full assessment of the quality of evidence.

Discussion
This systematic review revealed several factors associated with an increased risk of diphtheria: incomplete vaccination, contact with a person with skin lesions, and low knowledge of diphtheria.Other factors showed potential associations with the disease in multiple studies, including no booster vaccination in the last five years; contact with a case of diphtheria; sharing a bed or bedroom; sharing utensils, cups, and glasses; and infrequent bathing.The quality of evidence for the identified risk factors was rated as low or very lowexcept for incomplete vaccination, whose evidence was judged of moderate quality.Such a finding is not uncommon.Many systematic reviews following Cochrane or WHO guidelines are based on observational evidence rated as low quality [56,57]; however, these studies are very important in publichealth decision-making where randomized controlled trials are not available or ethical.
While GRADE offers a viable approach for evaluating the quality of evidence, it may not be completely suitable for assessing the quality of evidence relating to nonmodifiable risk factors, such as those identified in this systematic review.This is because evidence for this type of risk factor generally comes from observational studies, whose evidence is always initially rated as low-quality using GRADEirrespective of the study design or the methodological quality of the studies.When the entirety of the evidence comes from observational studies, it may be advisable to not consider this as a criterion for quality assessment.
In the sensitivity analysis, incomplete vaccination remained associated with an increased risk of diphtheria after removing studies that did not adjust for potential confounders.This finding was consistent with studies of other vaccine-preventable diseases [58][59][60][61], which showed a higher risk of infection for partially vaccinated individuals.This finding was also in line with results of another review [62], which reported that two primary vaccine doses produced lower protective levels of diphtheria antibodies compared with three doses.The importance of vaccines was further corroborated by other studies included in the present review [53,63], which pointed towards an association between diphtheria and vaccination coverage.
In the meta-analysis, the POR for not having received a booster vaccination in the last five years suggested a potential increased risk of diphtheria.The result might have been inconclusive due to low power and heterogeneity of the study populations.Five years might have also been insufficient to identify time since the last booster as a risk factor, especially in children.Among adults, for which there were fewer studies, this may be a strong risk factor, as suggested by one of the included studies [29].Previous studies [64][65][66][67] found that immunity to diphtheria decreases as age increases.Further research could provide valuable evidence to inform vaccination guidelines for adults, which currently vary extensively worldwide [68,69].
Exposure to an infected person is an established risk factor for various vaccine-preventable diseases [58,60,[70][71][72][73].Hence, we anticipated to find that individuals who had contact with a diphtheria case or a person with skin lesions (a potential proxy for cutaneous diphtheria) had a higher risk of infection compared to those who did not.These results support current surveillance guidelines that recommend the prompt identification, monitoring, and implementation of preventive measures for close contacts of diphtheria cases [3,5,6].
The association between diphtheria and low knowledge of the disease was unsurprising.The increased risk of diphtheria associated with believing that vaccines do not prevent diseases, as reported by one of the included studies, was also expected.Past studies have shown that knowledge of vaccine-preventable diseases and attitudes towards vaccination influence vaccine uptake and adoption of other protective measures [74][75][76].It is unclear whether antivaccination attitudes have been increasing in countries that are reporting diphtheria cases.Nonetheless, the recent rise of the internet and social media has certainly created unprecedented opportunities for antivaccination messages and false health information to spread virally, potentially shaping people's knowledge, attitudes, and actions [77,78].
The increased risk of diphtheria associated with sharing utensils, cups, and glasses; bathing infrequently; and obtaining water from a wheeled carrier (a possible indicator of a lack of access to clean water and sanitation) suggested that poor hygiene may play a role in contracting the disease.While the exact mechanism through which such practices heighten the likelihood of disease is not fully understood, it is conceivable that they might increase susceptibility to C. diphtheriae colonization [36].Our results support those of previous systematic reviews, which concluded that adherence to good hygiene practices lowers the risk of respiratory infections [79,80].
Incomplete vaccination has often been associated with low parental education [81,82].This review also revealed studies indicating a potential relation between education and diphtheria, with children whose parents had low educational levels appearing to be at an increased risk of infection.The pathway behind this association is unclear.Less educated parents may have lower literacy skills that makes them less receptive to health information [83][84][85].Less educated parents may also have lower communication abilities, which decrease their capacity to navigate the healthcare system to have their children vaccinated [83,84,86].Finally, less educated parents may simply live in poorer areas where the risk of acquiring diphtheria is higher due to reduced access to healthcare, lower vaccination coverage, and worse sanitary conditions [83,87,88].This last hypothesis seems to be supported by the link identified in one of the studies included in this review between the share of people living below the poverty line and diphtheria incidence.

Implications
Given the strength of association between diphtheria and incomplete vaccination alongside the consistency of our findings with the existing body of knowledge on vaccine-preventable diseases, additional observational studies on this risk factor would have limited valueexcept for better characterizing its association with diphtheria.Further research on this subject should, therefore, focus on other factors presenting a possible link with diphtheria and on factors overlooked by previous studies.Overlooked factors include those related to healthcare (e.g., availability of health workers, proximity to health centres, possession of health insurance), which have shown associations with vaccination status [82,89,90].Finally, the risk factors presented in this paper could be used to adjust for confounding in future studies that investigate potential risk factors for diphtheria.
By identifying the main modifiable diphtheria risk factors, this review provided a basis for detecting those most vulnerable to the disease and an agenda for public health action.The observed high risk of infection following exposure to a person with skin lesions, alongside evidence in some of the included studies of an increased risk of infection after contact with a diphtheria case, underscored the importance of early case finding and contact tracing.These activities entail the existence of adequate means for laboratory testing.Recent studies highlighted major gaps in diphtheria diagnostics in Europe and in the Western Pacific, including a lack of capacity for molecular typing, insufficient testing equipment, and inadequate training of laboratory personnel [91,92].These challenges are likely present in other parts of the world.Tackling them is crucial for controlling diphtheria.
The overwhelming evidence of a link between diphtheria and incomplete vaccination reinforced the need for countries to achieve timely vaccination with a complete primary series followed by booster doses, as recommended by the WHO [8].In the 1990s, countries of the former Soviet Union succeeded in controlling widespread diphtheria epidemics by vaccinating >140 million adults and adolescents alongside millions of children [93].Since such mass interventions might be difficult for some of the countries currently affected by diphtheria, other solutions should be explored.The recent experience of many LMICs (some of which are reporting high diphtheria rates; e.g., Haiti and India) with delivering the human papillomavirus vaccine in schools showed that large numbers of children can be reached through school-based vaccination [94,95].
Efforts should be made to increase people's knowledge of diphtheria and improve understanding of how to prevent the disease.As healthcare professionals are generally considered the most trusted source of health information [96][97][98][99], they should be encouraged to inform patients about diphtheria and the importance of personal hygiene and vaccines.Furthermore, activities should be directed towards implementing effective social mobilization strategies.Polio eradication initiatives in Afghanistan, India, Nigeria, and Pakistan have demonstrated that community health workers can help raise public awareness about vaccination and support surveillance activities [100][101][102][103].
In this paper, we identified a lack of studies from Africa, despite some countries in this region having reported high rates of diphtheria in recent years.This might reflect the fact that diphtheria is a relatively rare disease that mainly affects LMICs, where research capacity is sometimes limited [104,105].Furthermore, previous studies have shown that the fees that journals charge to publish articles open access can also pose a barrier for researchers and research institutions in LMICs [106,107].The dearth of studies on diphtheria might also be partly due to disparities in research priorities, with higher attention given to diseases perceived to have greater public health impact, such as HIV/AIDS, malaria, and tuberculosis.Governments and donor agencies should allocate adequate funding for the implementation of research and dissemination of findings on diphtheria to strengthen the evidence base and inform disease control efforts in countries affected by the disease.Such research is needed more than ever given the impact of the COVID-19 pandemic, which has caused the largest global disruption to routine immunization services in recent history, leaving millions of children unvaccinated or under-vaccinated against diphtheriaincluding in countries that had previously controlled the disease [108,109].

Limitations and strengths
Identified risk factors emerged from a limited number of single studies, resulting in evidence of mostly low quality.The small sample sizes of many of the included studies also meant that studies were not adequately powered to reach definitive conclusions.All studies were observational; as such, they could not demonstrate causality.Certain studies did not report effect estimates for some of the investigated risk factors or data to calculate these; therefore, it was not always possible to conduct a meta-analysis.Due to the paucity of studies, we could not stratify the analysis by potential confounders (e.g., age, socioeconomic status).Included studies often varied substantially in methodology.For instance, only a minority of studies (46%) consistently used laboratory tests to diagnose diphtheria.Thus, other studies may have included individuals who did not contract the infection, possibly leading to an underestimation of the risk estimates.Differences in the methods used for the ascertainment of the exposures (e.g., surveys vs. review of medical records) may have also affected the estimates.Moreover, half of the studies did not adjust for confounders, which might have contributed to the observed inconsistencies across studies on the same risk factors.This methodological heterogeneity may partly explain the high I 2 for some of the pooled estimates.
The high I 2 might also be partly explained by the inclusion of studies from various contexts.The use of an extensive search strategy with clearly defined search terms and no date, geographical, or language restrictions allowed the identification and inclusion of studies from several countries spanning almost all continents, encompassing different time periods and stages of diphtheria epidemics, and covering thousands of individuals across different populations.Such diversity provided more validity to our results and enhanced the generalizability of findings.The employed search strategy helped to minimize reporting bias in our systematic review, although publication bias may still exist in the literature.Furthermore, as with every systematic review, there is a possibility that not all relevant studies were included.All included nonecological studies were of moderate-to-high quality.The assessment of study characteristics was conducted meticulously to ensure that only papers with similar exposure definitions would be combined in a metaanalysis.As such, this review did not only provide a comprehensive synthesis of evidence on risk factors for diphtheria, but also offered summary estimates of the association between these exposures and the disease.These findings are critical since evidence-based policy and practice rely on objective measures of risk, which until now were fragmented and incomplete for diphtheria.

Conclusions
This study identified several risk factors for diphtheria.We found moderate to low quality evidence that incomplete vaccination, contact with a person with skin lesions, and low knowledge of diphtheria increased the likelihood of disease.Contact with a case of diphtheria; sharing a bed or bedroom; sharing utensils, cups, and glasses; infrequent bathing; and low parental education also appeared to be associated with diphtheria in multiple studies.Future research should focus on risk factors that have previously been neglected or for which evidence is inconclusive.
Most of the factors identified in this review are difficult to modify.Achieving substantial reductions in the incidence of diphtheria will, therefore, require sustained efforts by countries to strengthen their laboratory capacity, improve vaccination coverage, and increase people's knowledge of the disease and prevention methods with support from health workers.While these interventions have been advocated in the past, the current resurgence of diphtheria makes their implementation as critical as ever.

Contributions
Juniorcaius Ikejezie conceptualized and designed the study, screened articles, collected data, performed the quality assessment, carried out the statistical analysis, and wrote and revised the manuscript.Busola Adebusoye screened articles and collected data.Winifred Ekezie performed the quality assessment.Revati Phalkey, Tessa Langley, and Sarah Lewis conceptualized and designed the study, provided content expertise and methodological guidance, and reviewed the manuscript.All authors approved the final manuscript as submitted.

Summary of evidence
Summarise the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (such as health care providers, users, and policy makers) 8-9 Limitations Discuss limitations at study and outcome level (such as risk of bias), and at review level (such as incomplete retrieval of identified research, reporting bias) 9

Conclusions
Provide a general interpretation of the results in the context of other evidence, and implications for future research 9 Funding

Funding
Describe sources of funding for the systematic review and other support (such as supply of data) and role of funders for the systematic review 9 Appendix 2. Search strategy

9 Fig. 1 .
Fig. 1.Search and study selection process.*Additional papers included following hand searching of reference lists.

Fig. 2 .
Fig. 2. Risk of diphtheria associated with specific risk factors.

Table 2
Factors investigated for which at least one study found an association with diphtheria.

Newcastle-Ottawa Scale assessment scores Case-control studies
* OR ecologic* OR factor* OR probabilit* OR frequenc* OR characteristic* OR criteri* OR prognos* case-control OR cohort* OR cross-sectional OR ecologic* OR retrospective OR prospective OR longitudinal OR risk* OR predict* OR correlat* OR etiol* OR aetiol* OR prevalence* OR incidence* OR rate* OR determinant* OR odds ratio* OR associat* OR stratif* OR ROC Curve* OR discriminat* OR c statistic* OR area under the curve* OR AUC OR outcome OR indice* OR algorithm OR multivaria* OR model*) Appendix 3.