Epidemiology and demographics of head and neck cancer in Africa: A scoping review

Background Low- to middle-income countries account for 70% of global cancer deaths. Evidence of the changing prevalence of head and neck cancer in Africa in terms of gender, race and epidemiology will inform future research and health planning. Aim To synthesise epidemiological literature for head and neck cancer in Africa from 2010 to 2020. Method A scoping review was completed. The Joanna Briggs Institute Population, context and concept framework confirmed the inclusion criteria. Studies from Africa that included participant demographics, the types, stages, signs and symptoms of head and neck cancer were selected. Five databases were used. Descriptive statistics was completed. Results The Preferred Reporting Items for Systematic Reviews and MetaAnalysis guided the reporting of the findings. Of the 1891 articles downloaded, 66 were included in the final review. Nigeria produced the most studies and oral cancer at 74% was most prevalent. Substance abuse was the most prevalent cause. Diagnosis of head and neck cancers were in the late stage (stage IV) when signs and symptoms were severe. Males of lower socioeconomic status tended to have less health seeking behaviour. Conclusion Countries from North Africa produce the most research outputs on head and neck cancers. Gender differences were noted and may be linked to lifestyle choices. A range of head and neck cancers (HNCs) are prevalent however late diagnosis and severe symptomatology impact treatment options. Contribution Earlier diagnosis and intervention to prevent late-stage diagnosis is necessary. Awareness campaigns linked to evidence on causes, habits and lifestyle choices, signs and symptoms are needed.

With reference to Table 1, the distribution of HNC varies across anatomical regions. The most prevalent anatomical regions of the head and neck that are affected by cancer is the upper aerodigestive tract, which includes the oral, pharyngeal and laryngeal structures, but is also inclusive of the upper oesophageal region, that being up to the level of the cervical oesophagus. The primary anatomical structures include the oral, pharyngeal and laryngeal cavities. Each anatomical structure has sub-sites within the cavity that can be affected in isolation or if the HNC has spread to multiple sub-sites.
The following epidemiological data were considered: prevalence, pattern of disease (staging, anatomical sites involved and signs and symptoms experienced), distribution between population groups, lifestyle habits and populations most at risk. To understand the demographics of patients presenting with HNC, age, gender, race and socioeconomic status were considered.

Methods
A scoping review of the documented epidemiology of HNC in Africa in terms of presentation, types and most common anatomic sites, staging and age at the time of diagnosis as well as information pertaining to gender, race and socioeconomic status as reported in the studies were captured. The data were analysed quantitatively. An ethics waiver was obtained prior to the scoping review commencing, from the university Ethics Committee (Medical) (W-CBP-220218-02). The application for an ethics waiver was in line with the institution's policy for research involving review (scoping or systematic). Ethical clearance to conduct this study was obtained from the University of the Witwatersrand Human Research Ethics Committee (Medical) (No. R14/49).

Eligibility criteria as per the Joanna Briggs Institute, population, concept and context framework
Population: Patients presenting with head and neck squamous cell carcinoma (HNSCC), all types of nasopharyngeal carcinoma from the nasopharynx to the cervical oesophagus.
Concept: Epidemiology, types, stages, signs and symptoms of head and neck cancer, demographics including age, race, gender and socioeconomic status.

Selection process
A pilot study was conducted to refine the search terms and methods for the main study. Following the pilot study, MeSH terms and controlled vocabulary were not used because of it yielding irrelevant results. Primary keywords used for the main study were 'head and neck cancer' and 'Africa'. After a review of the six databases and 22 grey literature sources, a total of 1891 articles were obtained.
The Cochrane protocol guided the research process. 6 The data screening process followed the Preferred Reporting Items for Systematic Reviews and MetaAnalyses (PRISMA) flowchart. 7 This process is described below in Figure 1.
From the 1891 results, 399 duplicates were removed. The first level of screening (title and abstract) removed 1299 articles, leaving 269 articles for the second level of screening. After the second level screening of full articles, 203 articles were removed, leaving n = 66 articles in the final review. This is Oropharyngeal cancer (OPC) includes the mucosal lining of the tongue base, palatine tonsils, soft palate and oropharyngeal mucosa and constrictor muscles from the level of the palate to the hyoid bone.
Supraglottis cancer includes the mucosal lining of the epiglottis, aryepiglottic folds, false vocal folds and the deep pre-epiglottic and paraglottic space.
Nasopharyngeal carcinoma (NPC) includes squamous mucosa (NPSCC), lymphoid tissue (adenoids), the levator palatini muscle and the torus tubarius (the projecting posterior lip of the pharyngeal opening of the eustachian tube).
Glottis cancer includes the squamous mucosa of the vocal cords and thyroid cartilage.
Hypopharyngeal cancer (HPC) includes the squamous mucosa of: the pyriform sinuses, the pharyngeal walls and inferior and middle constrictors and the postcricoid region.
Subglottis cancer is the mucosal lining of the space between the glottis and trachea and includes the cricoid cartilage.  The data extracted from the reviewed articles followed a data extraction form which was self-developed. The extracted data were appraised according to its quality using the mixed method appraisal tool (MMAT) (Hong et al., 2018) to review the methodological quality of the data obtained from the studies used in the review. The quality of data was high as 68% (n = 45) of articles obtained a score of 100% on the MMAT. This can be seen in Appendix 2. All extracted data were synthesised and analysed quantitatively. These data were nominal in nature, and therefore the variables were grouped together, and the frequencies were calculated and compared.

Epidemiology of head and neck cancer
Where were most studies conducted in Africa? 1. Most articles in the HNC category did not specify the sub-site, but rather grouped the HNC together. Oral or pharyngeal cancer was the most reported on as it occurred in 16 of the 24 articles and accounted for 1244/2805 of the participants (44%). The rise of OC and OPC is related to the use of tobacco and alcohol and in four countries, to HPV. The countries that reported an increase in HPV as a contributing factor included South Africa, Burkina Faso, Sudan and Ghana. 9,10,11,12 The results revealed that HPV is strongly related to obtaining OC and OPC, as opposed to other types of HNC. The study done by Ahmed et al. 10 aligned with this finding as it was stated that '… HPV infections were more frequently identified in the tumour tissues from oral (40%) followed by larynx and pharynx'.
2. Figure 3 indicates the articles that researched OC and OPC specifically. The researcher combined these two sub-sites, as many articles reported on these two areas together as they are anatomically proximal to each other.
Most OC and OPC articles did not identify the anatomic subsite but differentiated between OP and OPC (13 out of 23, 57%), of which most participants studied had OPC (71%). There were 10% of articles that did not differentiate a sub-site at all. As noted in Figure 3, the tongue is the most prevalent anatomic subsite (23%). Reasons for this were not specified in the articles. Khammisa et al. 13 noted that tongue cancer was more prevalent in black people in South Africa: 3. Nasopharyngeal carcinoma was the third most common subcategory found in the articles. The increasing number of NPC was confirmed in that 14 out of 66 (21%) made it the primary focus of the study. On analysis of the HNC subcategory, it was noted that 606 out of 2805 (22%) of the participants with a specified anatomic type had NPC. This is supported by the study done in the southern part of Nigeria (Ibadan) 14 and found that NPC was the most prevalent of the HNCs. In contrast, the most prevalent of HNCs in southwest Nigeria (Lagos) was of the oral cavity. 15 Reasons for this difference could not be determined but are interesting to note.
The rise of NPC, particularly in the northern part of Africa, was supported, 16 where there was an attempt to review the association of gene variants with the susceptibility of HNC (namely NPC) in Tunisia. The results concluded that there are gene variations that cause an increased susceptibility in obtaining NPC, and importantly, there is a '… possible role for these variants as biomarkers for early detection of HNC and especially the NPC subtype'. 16 4. Hypopharyngeal or laryngeal cancer was the fourth category and combined two subsites because of their anatomical proximity. The most reported subsite was the transglottic (48%), followed by the hypopharynx (33%) and the cervical oesophagus (10%). The remaining subsites attributed to 5% or less of the reported sites.

Stage of head and neck squamous cell carcinoma and nasopharyngeal carcinoma
In total, there were 19 out of 66 (29%) articles that reported on the stage of HNSCC and NPC, with a total of 2489 participants. Of the articles reporting stage, there were missing data for 16% of the participants (n = 404). Furthermore, 5 of the 19 articles only commented on early versus late-stage cancer. Figure 4 shows the articles that commented on the stages and is divided according to the anatomic regions that were affected. The majority of participants (65%) were diagnosed with stage III and stage IV cancer, with 19% presenting with early-stage cancer (stages I and II). Data relating to the stage of HNC were missing for 16% of participants.
With regard to the types of HNC, the largest proportion of stage IV cancers was found in the OC and OPC group (n = 272, 74%) and secondly the NPC group (n = 400, 69%). Hypopharyngeal cancer and laryngeal cancer was predominantly diagnosed at stage III (n = 64, 69%). The HNC with the highest proportion of stage I was OC and OPC (98%) (8%).

Signs and symptoms associated with head and neck squamous cell carcinoma and nasopharyngeal carcinoma
The distribution of symptoms was widespread. The top 10 most reported signs and symptoms amongst HNC are depicted in Figure 5.
The development of lumps or lymph nodes in the cervical area was, by far, the most reported sign, accounting for 572/2343 (24%) of all reported signs and symptoms. Most articles that reported cranial nerve damage did not specify which cranial nerves were damaged. The largest nasal symptom was nasal obstruction, which accounted for 53 (2%) of all signs and symptoms. The last most reported symptom or sign was epistaxis (nose bleeds) (n = 48, 2%).

Demographic data of head and neck cancer in Africa Gender
A total of 60 out of 66 articles documented the gender of each participant, which included 33 602 participants. Of this, 28% were female (n = 9375), and 72% were male (n = 24 227). Within the subcategories of OC and OPC, NPC and LCC, all had a higher percentage of males when compared to females. For the OC and OPC subgroup, 25% were females (n = 6214), versus 75% being males (n = 18 879), which is a ratio of 1:3 (females to males). For the NPC group, there was a predominance of 66% (n = 2287) males versus females, accounting for 34% (n = 1156), resulting in a ratio of 1:1.9. Finally, for the HPC and LC group, there was 78% males (n = 188) to 12% females (n = 52).

Age
There were 56 articles that documented the mean age of the group, which comprised of 26 818 participants. The overall age mean range was from 19 to 69.9 years, with the largest distribution range being from 36 to 60 years. The largest mean age frequency range is between 56 and 60 years, accounting for 30% of the articles (n = 17). There is also a peak noted from 41 to 45 years (n = 12) and from 51 to 55 years (n = 10). The results indicate a lower overall mean age of     people with NPC and then OC and OPC with the highest overall mean age for people with HPC and LC. A potential reason for a lower mean age for the NPV group could be that HIV was prevalent in two of the three studies in Ghana, 17 Ethiopia 18 and South Africa, 19 indicating that people with HIV and NPC presented at a younger age.

Socioeconomic status
Socioeconomic status (SES) was discussed in only 11 of the articles, with a total of 1468 participants across the four categories. That included four articles in the HNC category (n = 979), four in the OC and OPC group (n = 408), two in the NPC group (n = 50) and two in the HPC and LC group (n = 31). Socioeconomic status was described according to high versus low; level of education, income and/or type of employment for each of the articles. The majority of the participants had a low SES, characterised by a low SES, low education level (primary level or no formal education), low employment level (labourers, peasants, artisans; unemployed or students) or a low-income level. Low income was defined by each article, according to the national breadline, which included earning less or more than R688.00 and R3000.00 in Uganda and Nigeria, respectively. Both Nigeria and Uganda used their own currency; however, this was converted into South African Rands for ease of comparison. Education level was the only SES marker to be greater in the high SES group (54.87%), compared to the low SES group (45.13%). The higher percentage is because secondary level education was included in the high-level group, which accounted for 68% of the high group. Only 32% of the participants had a high level of education or tertiary level of education.
Furthermore, it was found that those of a lower SES tended to have more advanced stage HNC, and a potential reason for this was '… because of their inability to afford treatment, ignorance, incorrect diagnosis and reliance on native medications'. 20 This is in keeping with the primary theme of poor access to healthcare and resources, suggesting that those of a lower SES are at a greater disadvantage.
There was no significant difference of SES between the types of HNC noted. No comparison between regions could be made, as SES was demonstrated differently between countries. Furthermore, limited associations could be made between SES and the risk factors, as the results within the articles did not include associations, and there was no access to the raw data.

Race
Race was mentioned in eight studies, comprising of 24 462 participants. Three of the eight studies were case reports of single participants. There were six that were conducted in South Africa, of which three were done in the city of Johannesburg. The other two were conducted in Ethiopia (n = 1) and Malawi (n = 1). Of the articles mentioning race, seven were in the OC and OCP subcategory and one case study for NPC. Data pertaining to race were missing for 407 participants. A predominance of black people is shown, accounting for 56% (n = 13 355) of the participants. There were two studies 21,22 that investigated the race with the highest prevalence when compared to the population distribution. It was found in both studies that mixed race males are the most prevalent race with OC and OPC in South Africa and are therefore most at risk, even though the mixed race population only accounted for 12% of the participants (n = 2984). Possible reasons for an increased risk within the mixed race population were not conclusive, although both articles suggested an increase in smoking, in conjunction with high alcohol use. 21,22

Discussion
With only 66 articles included in the scoping review, there is a clear dearth of information in the field of epidemiology of HNC in Africa. Data reflected findings from 19 of the 54 African countries. Literature concedes that there is limited clinical research across Africa generally, with only 1% of the world's research being done in Africa, even though it accounts for 12% of the world's population. 23 A high number of full-text articles were excluded from the study (n = 203), as they did not specifically study HNSCC or NPC (n = 99). Many articles did not answer the objectives of the study (n = 63), whilst 16 articles were not the primary source of information.
Head and neck cancer is on the rise in low-to middleincome countries, whilst some developed countries have declining numbers, particularly of OC. 24 Simard et al. 24 reviewed HNC across five continents excluding Africa, showing declining OC incidence in Asian countries, Canada and the United States. Oropharyngeal cancer rates, however, increased in some European countries but decreased in Asian countries. The rise of HPV was a contributory factor. 25 The rise of HNC is similar to Africa, even though overall prevalence was not documented; however, this has been confirmed in other studies. 26 As a result, prioritising HNC as a field of research, policymaking and education programmes is indicated.
The most studied and reported sub-site of HNC in this review was OC and OPC. Results showed that OPC was reported more than OC. This correlates with international findings where OC is higher in countries where tobacco is still a large risk factor (Slovak Republic, Estonia, Finland and Japan) and OPC being high in countries with decreasing tobacco use trends, but increasing HPV (Belarus, Czech Republic, Denmark, Finland, Iceland, Latvia, Norway and the United Kingdom). 25 In terms of histology, SCCs were noteworthy for being the most common globally and in Africa. 27 The histology of SCC is aggressive and fast spreading. In countries such as the United States and the United Kingdom, the few cases of NPCs are typically SCC in nature. 28 In recent years, however, NPC type III is of epidemic standards in Southeast Asia, Southern China and North African countries and is largely associated with EBV. 27 The systematic review supports this notion, as most NPCs were found in Northern Africa, as well as some northeast African countries (Sudan) and are mostly type III in histology. Nasopharyngeal carcinoma in Northern Africa is typically associated with a comorbidity (unspecified). 29 The systematic review also supports a worse outcome with the later stage diagnosis.
Late-stage HNC was common in the studies in Africa. Interestingly both developed and developing have late-stage diagnoses. 30,31,32 This is likely related to lifestyle habits in those of a lower SES and in males, regardless of country. Individuals from a lower SES and males tend to have less health-seeking behaviours and are diagnosed at a later stage. 33,34 The current review concurred. Importantly, HNCs tend to have a high metastatic rate and are aggressive in nature and can be difficult to diagnose as they mimic a common cold or flu 5 such as swollen lymph nodes, headaches, pain in the affected area (nose, mouth or throat), coughing, ulcer development, loss of taste or smell, nasal obstruction or congestion and nasal discharge. 35 Males over 40 years and from a lower SES are most at risk in Africa. A South African study 36 revealed that 70.4% of those diagnosed with HNC were black people (that being 1131 out of 1605 files). This aligned with international studies that found that tobacco and alcohol use by males versus females contributed to this. 4 The highest proportion of females found in the systematic review was in the NPC group, accounting for 34% of the NPCs. The higher number of females and higher proportion of NPC in women in Africa, versus other HNCs was not explained in the articles. However, it is likely that the association of EBV and a genetic predisposition causing females to have NPC is possible, but this warrants further research for confirmation. The OC and OPC group had a higher proportion of women when compared to other HNCs, and when compared to OC and OPC in previous years because of the growing rate of HPV. The effect of HIV on gender differences for HNC still needs exploring.
The average age of HNC effects was from 56 to 60 years, similar to the study from South Africa. 36 This is slightly less than the average age in developed countries, for example, the United States with an average age range of 50-70 years. 37 Age seems to not be a significant demographic factor for developing OC and OPC in Africa. Rather, HPV and HIV have an impact on the age of OC and OPC diagnoses, of which Africa is at a high risk. 1,38 Most people with HNC had a low SES, characterised by low income, being unemployed or labourers and/or having no formal education or only primary education. No associations could be made between the SES to various regions and to the associated risk factors, and this needs to be further explored in Africa.
Race was documented in South Africa only. Africa in comparison used ethnicity differences. Literature concedes that HNC is not particularly race related, but rather depends on social and cultural practices within different ethnic groups. 39

Conclusion
Nigeria conducted the most studies on HNC in Africa. Oral squamous cell carcinomas or oropharyngeal squamous cell carcinomas were the most prevalent types of HNC, with a rise of nasopharyngeal carcinoma (type III) in Northern Africa. Diagnosis was late (stage IV). Substance abuse, namely tobacco use, was the most prevalent cause, with a rise of HPV-related HNC and EBV-related nasopharyngeal cancer across Africa. Males and those of lower SES were most at risk.
This scoping review has limitations. The first is the limitation to the data obtained. No raw data were used; therefore, it is a summary and analysis of other researcher's data collection. Because of missing data in some of the articles, sample sizes were small. Causal relationships were therefore difficult to make. Another challenge was the various foci across the epidemiological studies, across a large space. As a result, it was difficult to generalise findings, as each country had a unique epidemiological pattern. This needs to be further explored in each country. J.S. and K.C. were responsible for the conceptualisation, article composition and overall supervision of the project. They were responsible for ensuring the validity of the analysis and overseeing the project. E.V. conducted the review and did the original write-up of the study findings. She conducted the original analysis of data.

Funding information
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Data availability
Data sharing is not applicable to this article as no new data were created or analysed in this study.

Disclaimer
The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any affiliated agency of the authors.