Domestic Burning of Incense and Mosquito Coils on the Risk of Nasopharyngeal Carcinoma: A Systematic Review and Meta-Analysis

Abstract

Objectives: We performed this systematic review of epidemiological studies to clarify the association between incense and mosquito-coil burning and the risk of nasopharyngeal carcinoma (NPC). Methods: A search of studies published through October 2024 in the MEDLINE and EMBASE databases was performed, supplemented by searches of reference lists, recent reviews, and Chinese databases. The quality of the included studies was assessed, with special attention paid to exposure assessment. Random-effects meta-analysis estimated the pooled odds ratios (OR) and 95% confidence interval (CI) whenever applicable. Results: One cohort and twelve case–control studies were identified. Meta-analysis of one cohort study and four case–control studies with reasonable quality found an increased risk associated with incense burning during adulthood (pooled OR = 1.46; 95% CI 1.24–1.72). Five case–control studies assessed the association between exposure to incense smoke during childhood and NPC risk, and the pooled OR was 1.22 (95% CI 0.76–1.96) associated with incense burning at birth and was 1.37 (95% CI 1.10–1.71) for exposure at the age of 10 years. The pooled OR for mosquito-coil burning during adulthood was 1.31 (95% CI 0.99–1.74). None of the four previous case–control studies found an increased risk of NPC associated with mosquito-coil burning during childhood. Conclusions: Our findings suggest an increased NPC risk associated with burning incense and mosquito coils. More high-quality epidemiological studies with refined exposure assessments are warranted.

1. Introduction

Nasopharyngeal carcinoma (NPC) is a malignancy which has fascinated generations of epidemiologists for its distinctive geographical and racial variations in incidence. NPC is rare in most parts of the world, with age-adjusted incidence rates less being than 1 per 100,000 persons per year irrespective of sex, but high incidence rates of NPC are noted in certain areas, including Southern China and Southeast Asia [1,2,3,4,5].

Although the etiology of NPC has not been completely understood, it has become increasingly clear that NPC is a multi-factorial disease resulting from the joint effects of virus infection, environmental exposures, and genetic susceptibility [1,6]. The unique spatial and ethnic clustering and a decrease in incidence since the 1970s–1980s in Chinese populations in Hong Kong, Taiwan, and Singapore [1,7,8,9,10] strongly suggest that some environmental exposures related to traditional lifestyles in these high-risk populations could have played a role in the development of NPC.

Incense burning, a traditional daily practice in Chinese households, is a powerful producer of carcinogens such as polycyclic aromatic hydrocarbons (PAHs), benzene, and formaldehyde [11,12]. Incense burning has been reported to be the major contributor to PAHs in Hong Kong homes [13]. Toxicological studies have also shown mutagenic and genotoxic activities of incense smoke condensates in mammalian cells [14,15]. Burning of mosquito coils is a common practice at least several months per year in many households, especially in tropical and sub-tropical countries. Smoke from mosquito-coil burning contain considerably high levels of suspected carcinogens, including PAHs and formaldehyde [16,17]. Thus, domestic burning of incense and mosquito coils may be risk factor for respiratory tract carcinomas, including those in the upper respiratory tract, such as NPC.

Previous epidemiological studies have examined the associations between domestic burning of incense and mosquito coils and the risk of NPC, although the results remain inconsistent [18,19,20,21,22,23,24,25,26,27]. To clarify such associations and explore the reasons underlying such inconsistency, we conducted this systematic review of available epidemiological evidence with a comprehensive literature search, quality assessment, and statistical synthesis of the available evidence. The need for future research in light of the methodological issues in previous studies is also discussed hereby.

2. Methods

This study was conducted according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines and is registered in the PROSPERO registry (CRD420251179504). A PRISMA checklist is provided in Supplementary Table S1.

2.1. Search Strategy

A comprehensive search of studies published through October 2024 in the MEDLINE and Embase databases was performed using the interface provided by Ovid with no language restriction. These two databases were selected because they are the most widely used major biomedical literature databases and are considered the core sources for epidemiological and clinical research, providing broad and complementary global coverage of the biomedical literature. Briefly, a combination of key words for the exposures (incense or mosquito coils) and the outcome (NPC) were used to identify relevant publications. The full electronic search strategy is presented in Supplementary Table S2. The reference lists of the identified articles and four comprehensive reviews on the epidemiology of NPC [1,5,6,28] were also reviewed to identify additional studies. A flow chart of the study selection process is illustrated in Figure 1.

2.2. Inclusion Criteria

Studies meeting the following criteria were included in this review: (1) case–control or cohort epidemiological studies published as original articles; (2) the studied outcome being NPC incidence rather than mortality; and (3) the association of incense or mosquito-coil burning with NPC risk being examined. In the case of multiple reports on the same population, only the most recent or informative ones were included. We only included studies which contained the minimum information necessary to estimate the odds ratio (OR) or relative risk (RR) with the reference group of non-users of incense or mosquito coils and a corresponding measure of statistical uncertainty (confidence interval [CI], standard error, variance, chi square and degree of freedom, or p value, etc.) into the meta-analysis. No additional exclusion criteria were applied.

2.3. Assessment of Study Quality

The quality of the included studies was assessed in terms of the major sources of selection bias, information bias, and confounding, with the help of checklists for appraising studies on risk factors developed by Yu and Tse [29,30,31]. The essential information regarding selection retrieved from case–control studies mainly included type (whether incident/newly diagnosed) and histological confirmation of cases, source and definition of controls, and response rate; the risk of selection bias in cohort studies was mainly assessed in terms of representativeness of the cohort and the completeness and duration of the follow-up. The retrieved information regarding exposure assessment included that on the exposure period, frequency, duration, intensity, and cumulative exposures. For cohort studies, risk of information bias on the outcome was assessed on the method of ascertainment (independent blind assessment, record linkage, or self-reported) and whether the follow-up was long enough for NPC to occur. The comparability of cases and controls (risk of confounding) was assessed by examining what potential confounding factors were considered and controlled for in the statistical analyses.

The Newcastle–Ottawa Scale (NOS) for assessing the quality of non-randomized studies in meta-analyses was also used to evaluate the quality of the selected studies [32]. The NOS contains eight items which are categorized into three domains, including selection of participants, comparability between groups, and assessment of exposure (case–control studies) or outcome (cohort studies). The NOS allocates stars, ranging from zero to nine, for the quality of non-randomized studies, such that studies with a higher quality are scored with more stars.

2.4. Meta-Analysis

The software Review Manager 5.4 was used for meta-analysis. Random-effects models were used to obtain pooled estimates of effect size and 95% CIs, combining data from studies of reasonable quality with NOS scores of at least 6. Heterogeneity across studies was assessed based on chi square tests and the I² index statistic. A low level of heterogeneity was defined as I² ≤ 25%, accompanied by p > 0.10 in the chi square test. Publication bias was assessed by visual inspection of funnel plots. Given the small number of studies available on mosquito coil burning during adulthood, we did not perform subgroup analysis or meta-regression to investigate heterogeneity. Sensitivity analysis was conducted by removing one study at a time to examine fluctuations in the pooled effect size. A two-sided p value less than 0.05 was considered statistically significant.

3. Results

A total of 507 articles were identified in MEDLINE and Embase, and 14 additional studies were identified from reference lists of reviews and articles of interest. After a careful review of titles and abstracts, and full texts if needed, 508 articles were excluded for duplication (n = 7), irrelevance (n = 496), or for not being a cohort or case–control study (n = 5). Finally, this systematic review included 13 individual studies, among which there were 12 case–control studies [18,19,20,21,22,23,24,25,27,33,34,35] and 1 cohort study [26]. Three case–control studies were not included in the further meta-analysis for associations between incense burning and NPC risk because two of them did not provide enough data to estimate the effect size and uncertainty [23,24], and the other one reported estimates of the effect size using participants in all of the other exposure groups instead of non-users of incense as a reference [34].

3.1. Cohort Study

Only one previous cohort study investigated the effect of incense burning on the risk of NPC [26]. This cohort study included 61,320 Singapore Chinese aged 45–74 years, including 6144 non-users, 7997 former users, and 47,179 current users of incense, who were recruited between 1993 and 1998 and followed up through 2005. Among current users, 92.7% were daily users and 83.9% had used incense for over 40 years. The study quality of this study was reasonably good, with an NOS quality score of 8, given the methodological strengths, including the representativeness of the original cohort (population-based including 85% of eligible individuals), complete follow-up (99.7%), comprehensive and prospective exposure assessment, objective assessment of outcome (91.3% histologically confirmed), and efforts to control for potential confounders (adjusting for age, gender, dialect group, education, body mass index, smoking, drinking, dietary items, and parity for women).

3.2. Case–Control Studies

The 12 identified case–control studies were published between 1966 and 2021, among which 11 were conducted in Chinese populations of different geographic areas, including Mainland China, Hong Kong, Taiwan, and Singapore, and the remaining one was conducted in a Filipino population. The number of cases ranged from 47 to 2533 across individual studies. Only four studies adjusted a set of various potential confounders in multivariate analyses. The majority of these studies obtained NOS quality scores ranging from 5 to 8, except for the earliest study published in 1966, which had a score of 2 (Supplementary Table S3). More detailed characteristics of the 12 case–control studies are summarized in

Table 1. Basic characteristics of identified case–control studies.

First Author	Year	Region	Ethnicity	Sex (% Male)	Age (Year)	Period	No. of Cases	No. of Controls
Sturton [22]	1966	Hong Kong	Chinese	Cases: 62%; controls: 35%	70% of cases and 35% of controls were under 50 years	Oct 1963–Jun 1965	47	217
Lin [23]	1973	Taiwan	Chinese	73%	Cases: 46.7 ± 11.5; controls: 46.2 ± 11.4	Dec 1969–May 1971	343	1017
Geser [35]	1978	Hong Kong	Chinese	70%	15–65+; Median: cases 49.1, controls 51.1	Jan 1973–Feb 1974	150	150
Shanmugaratnam [34]	1978	Singapore	Chinese	Cases: 70%; controls: 64%	Cases: 10–70+, median 47.4; controls: 0–70+, median 46.3	Mar 1966–Aug 1968	379	1639
Yu [19]	1986	Hong Kong	Chinese	64%	Under 35; cases: mean 29.2; controls: mean 28.6	Feb 1981–	250	250
Yu [18]	1988	Guangxi, China	Chinese	70%	Under 45; cases: mean 36.0; controls: mean 35.8	May 1984–Nov 1986	128	174
Yu [20]	1990	Guangzhou, China	Chinese	68%	Under 50	Mar 1983–Aug 1985	306	306
West [33]	1993	Philippine	Filipino	73%	11–83, median 46	Duration of 24 months	104	208
Lee [24]	1994	Singapore	Chinese	Cases: 73%; controls: 71%	Cases: median 37.9; controls: median 38.5	Mar 1988–Dec 1990	200	406
He [25]	2015	Guangdong	Chinese	Cases: 73.12%; controls: 70.77%	Cases: 46.11 ± 10.99; controls: 46.42 ± 11.74	Oct 2005–Oct 2007	1845	2275
Chen [27]	2021	Guangdong, Guangxi	Chinese	Cases: 73.4%; controls: 73.5%	20–74	2010–2014	2533	2597
Xie [21]	2014	Hong Kong	Chinese	Cases: 71.9%; controls: 73.4%	20–75	Jun 2010–Dec 2012	352	410

and

Table 2. Characteristics related to validity of identified case–control studies.

Study	Type of Cases	Histological Confirmation	Source of Controls	Response	Matched Factors	Adjusted Factors
Sturton 1966 [22]	N/A	N/A	Hospital (cancer patients)	N/A		Stratified by age and sex.
Lin 1973 [23]	Incident	95.90%	Community	Cases: 91.9%; controls: 85%	Sex, age, and neighborhood.
Geser 1978 [35]	Incident	N/A	Hospital	N/A	Sex and age.
Shanmugaratnam 1978 [34]	N/A	100%	Hospital	N/A		Age, sex, and interviewer.
Yu 1986 [19]	Incident	98%	Community (friend)	Cases: 94.0%; controls: 99.2%	Sex and age.	Salted fish intake during weaning and at age 10 years.
Yu 1988 [18]	Incident	100%	Community	Cases: 90.7%; controls: N/A	Sex, age, and race.
Yu 1990 [20]	Incident	100%	Community	100%	Sex, age, and residence.	Dietary factors.
West 1993 [33]	Incident	100%	Two series: hospital and community	Cases: 100%; hospital controls: 100%; community controls: 77%	Hospital controls: sex, age, and hospital ward type. Community controls: sex, age, and neighborhood.	Education, occupational exposures, smoking, fresh fish and processed meat intake, and herbal medicine use.
Lee 1994 [24]	Incident	N/A	Hospital	Cases: 88.6%; controls: 92.5%
He 2015 [25]	N/A	N/A	Hospital	Cases: 94.7%; controls: 95.5%	Sex and age.	Age, sex, education, housing type, smoking, salted fish, other dietary factors tea, herbal tea, and family history.
Chen 2021 [27]	Incident	100%	Community	Cases: 84%; controls: 83%	Sex, age, and geographic area distribution.	Age, sex, geographic area, education, occupation, family history, smoking, tea, oral hygiene and salted fish intake.
Xie 2014 [21]	Incident	100%	Hospital	Cases: 94.4%; controls: 93.8%	Age, sex, and residence.	Men: age, smoking, environmental tobacco smoke, education, housing type, family history, dietary factors, vitamins and supplements, occupational exposures, and cooking experience at home. Women: age, environmental tobacco smoke, education, and family history.

N/A: not available.

.

Eleven out of the twelve studies assessed exposure to incense or mosquito-coil smoke during adulthood based on usual or recent practice (four for incense burning only, one for mosquito-coil burning only, and six for both). Three of the twelve studies, all by research group of Yu et al., also assessed exposures at the age of 10 years and at birth [18,19,20]. One recent study assessed exposures at the ages of 10, 18, and 30 years [27]. One study assessed exposures to incense smoke during childhood, but did not further clarify the specific period [34]. Seven studies reported frequency data, but only one out of the twelve case–control studies considered other aspects of exposure, such as intensity and duration (

Table 3. Exposure assessment in the included studies.

Study	Exposure Periods		Frequency	Intensity	Duration	Cumulative Exposure
	Incense Burning	Mosquito-Coil Burning
Cohort study
Friborg 2008 [26]	at baseline	not assessed	yes	yes	yes	yes
Case–control studies
Sturton 1966 [22]	adulthood	not assessed	no	no	no	no
Lin 1973 [23]	adulthood	not assessed	no	no	no	no
Geser 1978 [35]	adulthood	not assessed	no	no	no	no
Shanmugaratnam 1978 [34]	adulthood, childhood	adulthood	yes	no	no	no
Yu 1986 [19]	3 years ago, age 10, at birth	3 years ago, age 10, at birth	no	no	no	no
Yu 1988 [18]	at birth, age 10	at birth, age 10	yes	no	no	no
Yu 1990 [20]	3 years ago, age 10, at birth	3 years ago, age 10, at birth	yes	no	no	no
West 1993 [33]	not assessed	adulthood	yes	no	no	no
Lee 1994 [24]	adulthood	not assessed	no	no	no	no
He 2015 [25]	adulthood	adulthood	yes	no	no	no
Chen 2021 [27]	ages of 10, 18 and 30 in the last 10 years	ages of 10, 18, and 30 in the last 10 years	yes	no	no	no
Xie 2014 [21]	adulthood	adulthood	yes	yes	yes	no

) [21].

3.2.1. Incense Burning During Adulthood

In the only cohort study, a total of 175 NPC cases were diagnosed during the follow-up; however, in the follow-up period (12 years at most), the association between incense burning and NPC risk were examined based on multiple exposure indicators, and no statistically significant associations were found. The adjusted hazard ratios (HRs) for NPC were 0.7 (95% CI: 0.4–1.4) in former users and 0.8 (95% CI 0.5–1.3) for current uses of incense, using never-users as the reference group. No exposure-response relations were observed. The HRs were 0.7 (95% CI: 0.3–1.5) for less than daily current use, 1.0 (95% CI: 0.6–1.7) for daily use for 40 years or less, and 1.0 (95% CI: 0.7–1.4) for daily use for over 40 years.

Ten previous case–control studies evaluated the associations between incense burning during adulthood and NPC risk. Among these, three studies simply reported the results as negative or not statistically significant [20,23,24], and one study reported estimates of RRs without using non-users of incense as a reference [34].

Meta-analysis combining five studies with NOS scores of at least 6 generated a pooled OR of 1.46 (95% CI: 1.24, 1.72) in random-effects model (Figure 2a). The heterogeneity tests suggest moderate heterogeneity across studies (p = 0.10, I² = 49%). The funnel plot with a limited number of studies suggests no substantial publication bias (Supplementary Figure S1), and the pooled OR did not substantially change in the sensitivity analysis excluding one study at a time (Supplementary Figure S2).

3.2.2. Incense Burning During Childhood

Five previous case–control studies, including three studies from the research group of Yu et al., examined the association between incense burning at early ages and NPC risk in Chinese populations in different areas. None of them found statistically significant associations. For exposure at birth, one of these three studies only reported the results as non-significant (p = 0.14). The pooled OR for NPC risk associated with incense burning at birth combining two studies was 1.22 (95% CI 0.76–1.96) (Figure 2b). For exposure at the age of 10 years, two studies described the results as non-significant (p = 0.34 and p = 0.23). The pooled OR for NPC associated with incense burning at the age of 10 years combining the other two studies was 1.37 (95% CI 1.10–1.71) (Figure 2c).

3.2.3. Mosquito-Coil Burning During Adulthood

Six previous case–control studies reported associations between mosquito-coil burning during adulthood and NPC risk. A meta-analysis of the four studies with NOS scores of at least 6 generated a pooled OR of 1.31 (95% CI 0.99–1.74) in a random-effects model, with high heterogeneity across studies (p = 0.002, I² = 80%) (Figure 3). The pooled OR did not substantially change in the sensitivity analysis removing one study at each time (Supplementary Figure S3).

3.2.4. Mosquito-Coil Burning During Childhood

The three case–control studies from the research group of Yu et al. examined the associations between mosquito-coil burning at early ages and NPC risk, and none of them reported statistically significant findings. Another recent study reported an OR of 0.84 (95% CI 0.75-0.95) for NPC risk associated with mosquito-coil burning in summer at the age of 10 years [27].

4. Discussion

This updated systematic review of existing epidemiological studies investigates the associations of incense and mosquito-coil burning with NPC risk. The results reveal an increased risk of NPC associated with incense burning during both adulthood and childhood periods and mosquito-coil burning during adulthood, but evidence regarding mosquito-coil burning during childhood remains limited.

The carcinogenic potentials of emissions from incense and mosquito-coil burning have been intensively documented in the previous literature on environmental chemistry and toxicology. Incense smoke is an important source of residential indoor PMs, including those less than 2.5 μm in the diameter (PM_2.5), which pose the largest health risks in all types of PMs [11,36,37,38,39,40]. It has been reported that incense burning produced PM over 45 mg/g burned, which was dramatically higher than that burned for cigarettes (10 mg/g) [41]. Incense burning also generates various types of VOCs, including benzene, PAHs, and formaldehyde, which are confirmed or probable carcinogens to humans [42,43,44,45,46,47,48,49,50]. Burning of mosquito coils is also a powerful producer of carcinogenic agents, such as PAHs and formaldehyde [16,17]. The emission of PM_2.5 mass from burning one mosquito coil was as high as that from burning 75–137 cigarettes. Burning one mosquito coil would release the same amount of formaldehyde as from burning 51 cigarettes [16]. Toxicological studies have shown that PM extracts from incense smoke is mutagenic in the Ames Salmonella test with TA98 and activation, and the genotoxicity of certain incense smoke condensates in mammalian cells may be even higher than that of tobacco smoke condensates [11,14,15]. Given the fact that inhalation is the primary route of exposure to smokes from incense and mosquito-coil burning, burning incense and mosquito coils may be an important risk factor for malignancies in the respiratory tract, including NPC.

The pooled results of meta-analysis in the present study showed an increased risk of NPC associated with incense burning. The inconsistency between findings from previous studies and the pooled estimates may be explained by differences in study design, population characteristics, exposure assessment, control of potential confounding, and chance. Notably, the findings should be interpreted with caution, given the considerable methodological limitations in previous studies. However, restricting our meta-analysis to studies with NOS scores of at least 6 diminished this influence to some extent. No associations between mosquito-coil burning during childhood and NPC risk were indicated in previous studies. It is possible to explained this by non-differential exposure misclassification regarding exposure in early life, which would have biased the results towards null, as well as the fact that burning mosquito coils is only an occasional practice in selective seasons, unlike incense burning, which is often a long-term daily practice.

Epidemiological studies on the health effects of environmental pollutants need accurate exposure assessments. An inaccurate exposure assessment can cause exposure misclassification, which may result in biased results and/or loss of study power, in addition to a failure in documenting a clear exposure–response relationship. Because incense and mosquito-coil burning was not the primary risk factor of interest in previous case–control studies, the exposure assessment methods concerning burning incense and mosquito coils were quite crude, e.g., only dichotomized to ever/never-exposed categories or just the frequency of burning. Only one of previous case–control studies considered other important aspects of exposure [21]. In addition, these studies only used the usual (recent) burning practice for exposure assessment, although several studies also assessed exposure at early ages. Changes of behavior over time or cumulative exposures were not considered. Even in the cohort study of reasonable study quality, the exposure assessment was based on incense burning practice at baseline only and did not adequately reflect lifetime exposure. Lack of accurate exposure assessment, particularly if non-differential, in previous studies may have diluted the observed associations between domestic incense and mosquito-coil burning and NPC risk.

Since NPC is a relatively rare disease, the cohort study design does not seem to be an efficient method of investigation for the risk factors of NPC. The reasonably large cohort study among a Singapore Chinese population only observed 175 NPC cases during follow-up and might not have adequate statistical power. Particularly, the limited length of follow-up might not have allowed for all NPC cases to occur. Case–control, especially in high-risk regions, may continue to be the major study design for investigating the etiology of NPC due to its feasibility and efficiency. The potential risk of selection bias in this systematic review needs to be discussed. A comprehensive searching strategy has been adopted to identify all of the published studies reporting the associations of interest. Because only a limited number of studies were specifically designed to investigate the associations between incense burning and NPC risk, and because the majority of these studies were published in earlier periods, several included studies were not identified using electronic searching, suggesting that electronic databases may not be sufficient for identifying studies dating very far back. Although the original authors were also contacted for detailed information when the published results were not sufficient for meta-analysis, no additional information was obtained, mainly because the data records were destroyed after the completion of their studies. The results of the meta-analysis are probably prone to publication bias because many studies with lower NOS scores but that found no associations were not included in the meta-analysis.

The validity of the results of the meta-analysis could be affected by those from primary studies. Although nine of the included case–control studies reported considerably high response rates, the remaining three did not report response rates, posing a certain risk of selection bias. Three studies did not mention whether recruited cases were incident or prevalent. If incense or mosquito-coil burning is associated with poor survival in NPC patients, recruiting prevalent cases might have resulted in underestimating the effects of incense or mosquito-coil burning because those with poor survival who were more likely to burn incense or mosquito coils might have been missed. Seven studies used hospital-based controls, among which, one even used cancer patients as controls [22]. If control participants were diagnosed with diseases associated with incense or mosquito-coil burning, the results might have been biased, for which the direction away from the true estimates depends on the specific diseases which the control participants were diagnosed with. If incense or mosquito-coil burning is associated with an increased risk for the diseases which control participants had, the effects of incense or mosquito-coil burning might have been underestimated. However, if control participants were affected by some specific diseases (e.g., asthma), they were likely to avoid exposures to smoke, which might have resulted in overestimating the effects of incense or mosquito-coil burning on NPC risk. Community-based controls were used in six of the included studies. Two studies also used friend or neighborhood controls, and it was possible that this caused overmatching and moved the estimates towards the null. In addition, even when community-based controls were used, the risk of misclassification of outcomes would be minor, given that NPC is rare in the general population. Although the same methods of exposure assessment were used for cases and controls in previous studies, there remains certain risk of misclassification in previous case–control studies because of the retrospective nature of such designs. However, it was probably non-differential, since incense burning and mosquito-coil burning are not well-established risk factors for NPC; hence, the resulting bias, if any, would have been towards the null. It is notable that limited efforts have been made to control for potential confounders in most of previous studies. Since incense burning is a traditional practice in some Asian populations, it is probably correlated with some other types of traditional lifestyle. Failure to control for potential confounding effects from other risk factors associated with traditional lifestyle may have compromised the validity of the results, leading to over- or under-estimated effects. Overall, the results of the meta-analysis in this systematic review were possibly subject to bias as a result of the inaccurate magnitude of associations reported in the original studies that could have arisen from various sources of bias in those studies. Notably, the limited number of eligible studies, most of which were conducted several decades ago, reflects the current scarcity of epidemiological research on this topic. In addition, the evidence included in this systematic review and meta-analysis was primarily derived from studies conducted in East Asian and Southeast Asian populations, which may not be generalized to other populations. These facts highlight a substantial research gap and the need for updated large-scale studies using contemporary study designs and exposure assessment methods.

5. Conclusions

In conclusion, although the synthesized results of previous studies suggest an increased risk of NPC associated with burning incense and mosquito coils, existing evidence from epidemiological studies remains far from enough to accurately assess such associations. We highlight the need for refined exposure assessment methods, increased efforts to control for confounders, and increased sample sizes for adequate study power in future investigations.