We examined aggregate diphtheria surveillance data reported annually to WHO and the United Nations Children’s Fund (UNICEF) from each country through the Joint Reporting Form (JRF) for global and regional epidemiologic incidence trends during 2000–2017. JRF data include the aggregate number of cases of diphtheria reported by countries in a given year and do not provide information on case-patient age or vaccination status.

We obtained information on the age or vaccination status of diphtheria case-patients from accessible published or gray literature (publications produced by organizations outside traditional commercial or academic publishing and distribution channels), presentation reports, and regional case-based surveillance data. We performed systematic searches covering publication dates during January 2000–September 2018 (Table 1). Searches returned 1,080 unique abstracts; 2 researchers reviewed each abstract to determine relevance. Discrepancies were discussed until consensus was reached. To meet inclusion criteria, manuscripts had to contain data about the age or vaccination status of case-patients with respiratory diphtheria caused by Corynebacterium diphtheriae during 2000–2017, with availability of full text in English or Spanish. Twenty-nine abstracts were excluded for language, as were 779 abstracts not relevant to the review. Two additional articles were not retrievable in full text.

Of 107 manuscripts reviewed in full text, 28 met inclusion criteria for the analysis (Appendix references 1–28). The full text of each manuscript was reviewed by 2 investigators, and relevant data were compiled in an Excel (Microsoft, https://www.microsoft.com) database. An additional 19 published manuscripts were identified through the reference lists (Appendix references 29–47). A review of the gray literature resulted in 12 additional sources (Appendix references 48–59), and communications with colleagues resulted in access to 11 unpublished reports (Pan American Health Organization: T.S.P. Tiwari [2]; R. Kaiser; Centers for Disease Control and Prevention; WHO Punjab; J. Crucena; Republic of Philippines Epidemiology Bureau; T. Nguyen; A. Nihal; and L. Sangal).

Diphtheria data from the European Surveillance System were provided by Spain, Latvia, Germany, Italy, Lithuania, the Netherlands, the United Kingdom, Finland, Sweden, France, Austria, and Belgium and released by the European Centre for Disease Prevention and Control (Stockholm, Sweden) (Appendix reference 60). Similar case-based diphtheria data were not available from other regions. Because of multiple data sources, we conservatively excluded cases identified as potential duplicates when matching by age group, location, and year. The final dataset consisted of 15,380 cases of diphtheria (15,068 including age data and 7,242 including vaccination status data) from 34 countries.

We compared cases included in the final dataset with the number of cases in the aggregate JRF data for each country over the same period to cross-reference the completeness of the diphtheria data reported in the JRF. Because DTP 3 coverage is a major risk factor for disease transmission in a population, we took the average of the national WHO–UNICEF estimates of DTP3 coverage (16) for the 5 years previous to the cases for each set of reported cases. For analysis, we classified countries with data included in the review as either countries with higher case counts (defined as reporting >10 cases in each of >3 years of JRF incidence data during 2000–2017, or reporting >100 cases in a single year) or as countries with sporadic cases.

The age distribution analysis was complicated by the diverse ways in which age data were aggregated in different manuscripts. Our analysis used an age of 15 years for disaggregation of age data because this age was most frequently mentioned in the historical literature as a benchmark for the age shift in diphtheria incidence over time. However, on the basis of availability of age data from source documents, we made classifications by using heterogeneous age cutoffs in the initial analysis. To address this limitation, we compiled a more precise dataset (n = 9,334 cases from 32 countries) for sensitivity analyses that contained only data that classified case-patients as >15 or <15 years of age (± 1 year).

Sources also aggregated vaccination status data differently and had varying definitions for fully vaccinated depending on the vaccination schedule of the country or criteria of the investigators. For this review we defined fully vaccinated, at a minimum, as having received all 3 doses of the primary series. Cases with partial vaccination (>1 dose of infant DTP) were grouped with fully vaccinated cases in several sources; we conservatively designated these cases as partially vaccinated in the full dataset. Reports of cases with unknown or partial vaccination status were grouped with unvaccinated cases in other sources; we conservatively designated these cases as unvaccinated in the full dataset. To address this limitation, we compiled a dataset (1,534 cases from 27 countries) that contained only cases that were reported in >3 different groups (unvaccinated, partially vaccinated, or fully vaccinated). We also provide additional information on datasets compiled from the literature and gray literature review (Table 2).

We analyzed distribution of case-patients by age and vaccination status by using descriptive methods in Excel 2016. We performed sensitivity analysis to test for consistency of findings between those analyses performed by using the full dataset and those performed by using a dataset with enhanced precision around the variable of interest. Because cases from India represented >50% of cases in the full dataset, we conducted a final sensitivity analysis to check for consistency of trends for all analyses when data from India were excluded.

Finally, to examine the relationship between vaccination coverage with the primary series and age distribution, we combined the total number of cases for each country in the dataset. We compared the proportion of case-patients >15 years of age in each country with the average of DTP3 coverage for the 5 years preceding the year(s) when the cases were reported. We excluded countries with <5 cases in the dataset from this analysis, leaving 24 countries in the primary analysis and 23 countries in the sensitivity analysis.

It is clear from aggregate incidence data that progress in decreasing diphtheria incidence worldwide has stalled, and reported cases have recently increased. Larger societal factors, such as population migration or political instability, can create conditions favorable to an outbreak; the largest recent outbreaks (January 2016–February 2019) were reported in the Rohingya refugee population in Bangladesh (8,403 cases), as well as areas experiencing conflict or social disruption, such as Yemen (3,340 cases) and Venezuela (2,512 cases) (19). The South-East Asia region has reported most of the diphtheria cases since 2000, which might be caused by the large populations of several countries in the region that have endemic disease. However, incomplete data from other regions could be obscuring additional major foci of disease incidence. Global and regional trends can be shaped by incomplete data, including years of nonreporting or underreporting by populous and high-incidence countries.

The diphtheria data reported annually to WHO and UNICEF on the JRF have substantial limitations in terms of quality and reflect opportunities to improve disease surveillance. We found these data to be incomplete when cross-referenced with the literature, indicating a likely underestimate of incidence worldwide and a decreased understanding of the burden of disease. However, in some countries with lower laboratory capacity, only a small proportion of cases are laboratory confirmed, which could result in overreporting in some settings (19). The ability to use aggregate data for action is limited by the lack of key variables, including vaccination status, age, and subnational location.

Implementation of case-based surveillance for diphtheria, combined with availability of subnational coverage data, would result in improved understanding of diphtheria epidemiology and enhanced capability to prevent and respond to outbreaks. A 2017 gap analysis of diphtheria diagnostic capacity in the European Union found substantial gaps, including lack of sufficient laboratory systems with methods to determine toxigenicity, difficulty obtaining primary media culture, and challenges to obtaining diphtheria antitoxin for both laboratory diagnosis and clinical management of cases (20). Similar assessments are currently being conducted to more fully understand the scope of challenges in other regions (A. Efstratiou, National Infection Service, London, UK, pers. comm., 2019 Aug 1). Worldwide, only 55 countries report conducting national, case-based surveillance for diphtheria with laboratory confirmation (9). In response to the limitations of current disease surveillance systems, WHO has released new comprehensive surveillance recommendations (21). For diphtheria, immediate investigation and collection of case-based data are recommended for all outbreaks. These guidelines provide a set of recommended minimum data elements to collect, as well as recommended analyses and uses of data collected. These standards represent an opportunity to improve and standardize available data with widespread implementation but will require investments in both surveillance and laboratory capacity.

Accounts of outbreaks in the peer-reviewed literature, gray literature, presentations, and other reports were compiled as the best available sources of information on case-patient age and vaccination status. When examining these data, we found that most diphtheria cases occur in unvaccinated persons, particularly in countries with higher case counts where most disease is among children <15 years of age. Therefore, achieving adequate coverage with the primary series is urgently needed. Ensuring high primary series coverage is especially useful in the context of vaccine hesitancy, which has wide variability among countries and regions (22). DTP3 coverage worldwide has stagnated at 84%–85% since 2010 (23), and improving this coverage through increased equity of and access to routine immunization services is key to the efforts to combat diphtheria. Countries with sporadic incidence of diphtheria have a more even distribution of vaccination status among cases. Age data from these countries also reflect a higher proportion of cases in the adolescent and adult populations. The predominance of older case-patients in these countries, taken together with the higher proportion of case-patients who have received, at a minimum, 3 doses of diphtheria-containing vaccine in infancy, indicate that waning immunity is also a major issue. This issue can be addressed through widespread adoption of the 3 diphtheria toxoid–containing booster doses recommended in 2017 by WHO and recently approved as a future investment in the 2021–2025 Gavi Vaccine Investment Strategy (24).

In our dataset, as vaccination coverage in a country increased, the percentage of case-patients >15 years of age also increased. This increase in the proportion of older case-patients is not unique to diphtheria because similar changes have been seen in the epidemiology of other vaccine-preventable diseases as coverage increased (25). This finding indicates that the large proportion of case-patients >15 years of age in countries with sporadic incidence probably represents a proportional, rather than an absolute, increase because high vaccination coverage in childhood resulted in fewer susceptible persons in this age group. Many adults might have grown up during a time when no or fewer booster doses were given, although information on historical changes to vaccination schedules is incomplete. As countries implement or modify booster dose schedules according to the new WHO recommendation, data on booster dose coverage would improve understanding of susceptibility to disease in different age groups.

Limitations of this analysis include heterogeneous methods used by available sources to aggregate data on case-patient age and vaccination status. Because the available data might not be representative, findings might not be generalizable to all contexts. Because of lack of laboratory capacity in some settings, many cases in the literature are not confirmed by culture. Strengths of the analysis include the compilation of all known available data on the age and vaccination status of diphtheria case-patients, which highlight that data from published disease outbreaks are a useful resource for describing epidemiologic changes and to triangulate with other existing data sources. Limitations of the dataset were addressed as comprehensively as possible through sensitivity analyses of subsets of cases with more precise data to validate trends observed on analysis of the full dataset.

In light of a recent increase in reported cases, action is needed to make progress in combating diphtheria. However, many national immunization schedules lag behind current recommendations, and the lack of case-based diphtheria data limits the ability to take targeted action. Intensified efforts to improve routine immunization coverage with DTP3 and to implement recommended booster doses would help to decrease diphtheria cases by both decreasing the susceptibility of children and addressing the problem of waning immunity among adolescents and adults. Implementing new WHO guidelines (21) would result in case-based diphtheria surveillance data with laboratory confirmation that could be analyzed by using standardized age categories. This implementation would greatly enhance available data and highlights the need for enhanced laboratory capacity to provide these case confirmations, particularly in countries with endemic disease and other lower-middle-income and low-income countries. Increased availability of booster dose coverage data, subnational coverage data, and more complete historical records of changes to immunization schedules would lend context to data on incidence and epidemiologic trends. An improvement in the quality and consistency of data collected on diphtheria would create a stronger evidence base for future research, timely interventions, and recommendations to combat this deadly disease.