[version 1; peer review: 1 approved, 3 approved with reservations]
No competing interests were disclosed.
Tuberculosis (TB) is a major health challenge in Indonesia, which ranks second globally in 2024. As the 2030 elimination target approaches, gaps in early detection and public education persist. The public’s tendency to seek health information online before consulting professionals presents an opportunity to leverage infodemiology for public health surveillance. Therefore, this study aimed to assess the relationship between multi-term Google search trends and annual TB report data in Indonesia to disseminate the potential use of digital search data as a complementary indicator for epidemiological surveillance.
A cross-sectional design was adopted to analyze the relationship between search volumes for 53 TB-related terms on Google Trends and official Indonesia Health Profile data from 2019 to 2023 across 34 provinces. Case data were normalized (0–100) to reflect the Relative Search Volume (RSV). Statistical analysis was performed using the Spearman correlation test to assess the relationship between digital searches and actual cases.
A consistently strong and positive correlation between TB search terms and case numbers across 34 Indonesian provinces (p < 0.001). Key correlations included “Characteristics of Pulmonary TBC (
Google Trends data correlated strongly with official TB epidemiological data in Indonesia. These results suggest digital search trends can serve as complementary indicators to conventional surveillance and early warning systems.
TB is a chronic infection threat in the world with an estimated 10.7 million sufferers in 2024. The burden is concentrated in Southeast Asia (34%) and the Western Pacific (27%) regions.
To determine whether search trends in digital or online spaces are in line with reality, official Indonesian government data is needed. In this case, the Indonesian Health Profile serves as a reference to present verified and validated case numbers at various levels, from regional to national. In contrast to daily data, which tends to fluctuate, the Health Profile data are compiled comprehensively to describe TB cases across the 34 provinces each year.
Global health information search behavior shows that the internet, specifically through Google, has become the primary source for discovery about symptoms, diagnosis, and treatment options before consulting with a health professional.
Health information activity on Google produces a digital footprint in the form of Relative Search Volume (RSV), which describes fluctuations in public interest, concerns, and pre-consultation behavior in the community.
Several studies have specifically described the use of Google Trends in health studies. In a report, Google Trends data was used to analyze search volumes for infectious disease symptoms before comparing with official surveillance data to assess correlations and potential predictions of case trends.
Previous studies showed a strong correlation with the term TB (r = 0.97–1.00) and national reporting data. However, a small number of medical terms were used without comprehensively exploring the diversity of TB terminology.
The method used was a quantitative approach with an observational cross-sectional design. This study was adapted from previous reports related to Google Trends correlation analysis.
Data were sourced from the official annual Indonesian Health Profile reports published by the Ministry of Health for 2019, 2020, 2021, 2022, and 2023, respectively (
This official Ministry of Health report served as a standard value for validating web search information from Google Trends (
The dependent variable for this study was the total number of confirmed pulmonary TB cases of all types, age groups, and gender in 34 provinces across Indonesia in 2019, 2020, 2021, 2022, and 2023. The independent variable was the volume of TB-related searches on Google Trends, or the frequency of each search term. A total of 53 search terms within the Google Trends search volume, including mentions of TB, characteristics, symptoms, treatment, prevention, transmission, and TB in children and infants in 34 provinces across Indonesia in 2019, 2020, 2021, 2022, and 2023, were examined. Based on terms frequently used by Indonesians and the availability of data on Google Trends in 2019–2023, a total of 53 terms were selected. The details of the 53 terms are presented in
| Category | Term |
|---|---|
| General terms used to refer to tuberculosis in Indonesian society |
|
| Characteristic |
|
| Symptom |
|
| Drug |
|
| Prevention |
|
| Transmission |
|
| Child |
|
| Baby |
|
TB = Tuberculosis, TBC = Tuberculosis, Lung spot = Lung spot is a term used by several Indonesian people to refer to tuberculosis.
The annual aggregate of TB case count data from the official report of the Indonesian Ministry of Health from the Health Profile was converted to the same interval as the Google Trends (RSV) data, namely on a scale of 0 to 100. A scale of 0 signified no case, while 100 represented the highest number of cases in the five years from 2019 to 2023. The official report was calculated and normalized based on the annual RSV to correlate the annual aggregate TB case count report with SPSS. The following formula was used for the data normalization process
Keywords were collected from Google Trends, frequently used by internet users in Indonesia. These words were grouped into eight categories, namely mentions of TB, characteristics, symptoms, treatment, prevention, transmission, tuberculosis in children, and infants.
Data analysis used the Spearman correlation test to assess the relationship between two variables when the data were not normally distributed. Interpretation followed the criteria presented in
| Absolute magnitude of observed correlation coefficient | Interpretation |
|---|---|
| 0,00–0,10 | Negligible correlation |
| 0,10–0,39 | Weak correlation |
| 0,40–0,69 | Moderate correlation |
| 0,70–0,89 | Strong correlation |
| 0,90–1,00 | Very strong correlation |
The Spearman correlation test was used in this study to examine the relationship between search terms related to TB, characteristics, symptoms, medications, prevention, transmission, TB in children and infants, as well as the annual Indonesian Health Profile report. Furthermore, statistical analysis was performed using SPSS version 25. This study did not use human participants in the methodology, hence, ethical approval was not required.
Table 3 (
In 2020, a strong and positive correlation between TB-related search terms and the number of TB cases in 34 Indonesian provinces was observed for “Characteristics of Pulmonary TB (
In general, the analysis results in Table 3 (
Among all the keywords, the most consistently significant terms over the five years included (1) characteristics of TB, lung spot, pulmonary TB (
In 2020–2021, due to the COVID-19 pandemic, respiratory-related searches generally increased, but the public’s focus was on the virus, leading to a decrease in TB-related terms. However, “lung spots (
Approaching the 2022–2023 period, the correlation pattern strengthened again and showed greater consistency compared to the pandemic period. In 2022, most terms related to the characteristics, symptoms, treatment, and transmission of TB continued to show significant positive correlations with moderate to strong strength across various provinces. This was evident in keywords such as “characteristics of lung spot (
In 2023, the strengthening trend in correlations became even more evident, particularly for clinical and specific terms. Several keywords, such as “characteristics of pulmonary TB (
A correlation analysis of Google Trends for TB cases in Indonesia shows that keywords based on symptoms, clinical characteristics, transmission, and treatment have a more stable relationship than general terms. As a result, the keywords are potentially being used as digital indicators to show spatio-temporal disease dynamics. Several infodemiology-based studies have shown that TB-related search volume on search engines significantly reduces cases and can reflect epidemiological trends. A study in South Africa showed a significant positive correlation between Google Trends search volume and reported incidence rates. Eight search terms showed moderate to strong associations, including “tuberculosis” and “TB,” as well as searches related to symptoms and diagnostic tests such as the “Mantoux test.” Furthermore, a strong correlation pattern was observed for terms related to comorbidities, particularly diabetes and HIV, reflecting the established pattern of HIV-TB infection in the region and increasing public awareness of the risk of diabetes to TB.
Another study used Google Trends to correlate measles clinical cases using Pearson correlation in 30 European countries and Japan. The results showed a very strong correlation at the regional level in developed countries, as observed in Okinawa Prefecture, Japan, during the 2017–2019 period. This is much higher than the correlation at the national level, proving that digital data searches are much more accurate in capturing signals of specific outbreaks in specific locations than on a broader scale. Search behavior is more sensitive to acute outbreaks that appear suddenly with many cases in a short period. Conversely, prolonged outbreaks with few cases per week often fail to be captured by Google Trends because individuals tend to no longer actively search for information when the disease is considered normal or not widespread.
Search for health information is often triggered by the development of disease symptoms that then trigger anxiety. When certain symptoms are experienced, search engines might be used to conduct initial identification before seeking professional medical help.
Search volume tends to increase sharply in areas with a high disease burden or during outbreaks. In Indonesia, studies on dengue fever and COVID-19 have shown a strong correlation between the number of actual cases and search trends for keywords such as “symptoms” or “transmission”.
Pandemics, such as COVID-19, can create an overshadow effect that disrupts the seasonal patterns of other infectious diseases. During the peak of a pandemic, public attention and medical resources are highly focused on the novel virus. This leads to a decrease or change in search behavior toward other diseases, such as RSV or influenza. After pandemic mitigation measures are relaxed, search patterns and outbreaks of other diseases often reappear at irregular times and with greater intensity. This emphasizes the importance of using digital monitoring tools such as Google Trends to stay aware of multiple health risks amid the dominance of a single large outbreak.
The present study has several limitations, including its reliance on internet access, which is not evenly distributed across Indonesia. This implies that Google Trends data may be more representative of the behavior of residents in urban areas than in rural areas. Furthermore, the intent behind keyword searches cannot be fully ascertained, whether being performed by TB patients for treatment or by healthy individuals simply sourcing general information. The study also did not include user demographic factors such as age and education level, which may influence how search terms are formulated in search engines.
In conclusion, digital search data through Google Trends had a significant positive correlation with the burden of TB cases in Indonesia, offering significant potential as a complementary epidemiological surveillance (infodemiology) instrument. Key results showed that the public was more prone to use non-medical or popular terms such as “lung spots” and specific keywords related to “characteristics,” “symptoms,” and “treatment” than formal technical terms. The consistent, strong correlation across children and infants also confirmed high demand for digital information among vulnerable groups. These search patterns reflected real-world public health behavior, which, when integrated with official reporting systems, could strengthen early warning systems and support more targeted health communication strategies aimed at achieving the 2030 TB elimination target.
The results of this study provide an empirical basis for policymakers to integrate Google Trends data as an early warning system to complement conventional surveillance systems. Public health communication strategies should be optimized by using popular terms in the community, such as “lung spot (
Figshare: Public Search Behavior and Tuberculosis Cases in Indonesia 2019–2023: An Infodemiology Method Using Google Trends,
The project contains the following underlying data:
Data.xlsx. (Tuberculosis Case Data and Google Trends Search Data)
Data are available under the terms of the
Figshare: Public Search Behavior and Tuberculosis Cases in Indonesia 2019–2023: An Infodemiology Method Using Google Trends,
This project sontaints the following extended data:
Table 3. The relationship between the 53 TB search terms on Google Trends and number of TB cases in Indonesia 2019–2023
Data are available under the terms of the
The authors are grateful to the Postgraduate School, Universitas Negeri Semarang, for funding this study and all the teams involved in this work.
Rahayu SR
However, I have a few comments that should be addressed prior to Indexing:
Brief introductory note for Google Trends and its use for public health surveillance would strengthen the manuscript.
TB transmission is often associated with shared and crowded environments, where marginalized populations are disproportionately represented. These groups may have limited access to smartphones or internet connectivity. Could the authors elaborate on whether Google Trends data might have underrepresented such populations?
While rural settings are briefly discussed, a more detailed consideration of access disparities and their impact on data interpretation would strengthen the manuscript. TB shares several symptoms with other respiratory diseases. How do the authors account for potential confounding from searches related to other conditions with overlapping symptoms?
Specifically, how is signal specificity ensured, and what strategies are used to minimize noise arising from non-TB-related searches?
Please clarify whether the current analysis supports an early warning system, or whether it should be described more cautiously as exploratory infodemiological evidence.
References 1 and 2 appear to be highly similar and could be consolidated to avoid redundancy. References 6-9 seem to direct to a single web page containing information for multiple years.
These could potentially be combined into a single reference for clarity and conciseness.
Although the authors state that age was not included as a factor in the analysis, exploring age and sex differences among TB patients could be a valuable avenue for future research.
Is the work clearly and accurately presented and does it cite the current literature?
Yes
If applicable, is the statistical analysis and its interpretation appropriate?
I cannot comment. A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?
Partly
Is the study design appropriate and is the work technically sound?
Partly
Are the conclusions drawn adequately supported by the results?
Partly
Are sufficient details of methods and analysis provided to allow replication by others?
Partly
Reviewer Expertise:
Tuberculosis, Immunology, Aging
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
Review Report :
The title is informative; however, the phrase
or
The sentence:
“Tuberculosis (TB) is a major health challenge in Indonesia, which ranks second globally in 2024.”
should be revised for clarity because it implies Indonesia itself is “the challenge.” Suggested revision:
“Tuberculosis (TB) remains a major public health challenge in Indonesia, which ranked second globally in TB burden in 2024.”
The statement:
“Case data were normalized (0–100) to reflect the Relative Search Volume (RSV).”
requires additional explanation. The abstract should briefly clarify
The results are excessively detailed for an abstract. Listing many correlation coefficients makes the abstract difficult to read. Please summarize only the major findings and report the strongest correlations more concisely.
For example:
“Several symptom- and treatment-related search terms showed consistently strong positive correlations with TB case reports across provinces.”
The sentence:
“However, control remains at 12%, a value far from the target of 50% by 2025.”
is unclear because “control” is not specifically defined. Please clarify whether this refers to case detection, treatment success, or another TB control indicator.
The statement:
“This gap shows that sufferers and the public tend to express their health awareness regarding TB through searching for information in digital spaces…”
is speculative and unsupported by evidence. The authors should provide a supporting reference or rephrase more cautiously.
Suggested revision:
“This gap may indicate increased public reliance on digital platforms to seek TB-related health information.”
The research objectives are not explicitly stated. Please clearly state:
The primary objective. Secondary objectives. The novelty of study unclear
A dedicated final paragraph for study objectives is strongly recommended.
The manuscript describes the study as:
“cross-sectional”
However, the analysis uses multi-year provincial data (2019–2023), which is more consistent with an ecological longitudinal or repeated cross-sectional design.
Please revise and justify the study design terminology appropriately.
The authors should clarify:
Whether Google Trends data were extracted using:
“search term” or “topic” “health category” or “all categories” “web search” or other search types Whether the searches were conducted in Indonesian language only.
These details are essential for reproducibility.
The explanation of the 53 keywords is insufficient.
Please explain:
How the keywords were generated. Whether experts validated them. Whether pilot testing was conducted. Why certain colloquial terms such as “flek paru” were included.
The normalization formula is incomplete and poorly formatted:
“Normalization Formula ¼ Highest scale in this case 100…”
This section requires major revision because the mathematical explanation is unclear.
Please:
Rewrite the formula correctly. Explain each variable. Explain why normalization was necessary before Spearman analysis.
The manuscript only reports Spearman correlation coefficients and p-values.
Please additionally report:
Confidence intervals. Interpretation criteria with supporting references. Whether assumptions for Spearman correlation were checked.
Since 53 terms × 5 years were analyzed, there is a high risk of Type I error.
The authors should discuss:
Bonferroni correction, False Discovery Rate (FDR), or Justification for not adjusting multiple comparisons.
This is a major methodological concern.
The Results section is overly narrative and repetitive.
For example, the repeated sentence structure:
“a strong and positive correlation was observed…”
appears excessively throughout Pages 4–6.
Please summarize findings more efficiently using:
tables, heatmaps, trend figures, or clustered correlation plots.
There are inconsistencies in terminology:
“Lung spot” “Lungs spot”
Please standardize terminology throughout the manuscript.
Decimal notation uses commas instead of periods:
“0,70–0,89”
Please standardize according to journal style:
“0.70–0.89”
The sentence:
“technical terms are less sensitive in describing the dynamics of the caseload”
needs stronger scientific explanation. Why were non-medical terms more strongly correlated? Consider discussing:
health literacy, public language preferences, cultural terminology, digital search behavior theory.
The discussion regarding COVID-19 overshadowing TB searches is interesting but lacks supporting Indonesian references. Please strengthen this section with local or regional evidence.
Several paragraphs merely restate the Results section without deeper interpretation.
Please reduce repetition and focus on:
implications, mechanisms, comparison with previous studies, theoretical explanations.
The limitation section is currently too brief.
Please additionally discuss:
Ecological fallacy risk. Google Trends algorithm instability. Influence of media coverage. Unequal internet penetration across provinces. Potential bias from urban populations dominating online searches.
This section needs stronger critical reflection.
The conclusion is too long and partially repetitive.
Please shorten and focus on:
the most important findings, practical implications, future surveillance potential.
Avoid re-discussing detailed results.
Is the work clearly and accurately presented and does it cite the current literature?
Partly
If applicable, is the statistical analysis and its interpretation appropriate?
Partly
Are all the source data underlying the results available to ensure full reproducibility?
Yes
Is the study design appropriate and is the work technically sound?
Yes
Are the conclusions drawn adequately supported by the results?
Yes
Are sufficient details of methods and analysis provided to allow replication by others?
Partly
Reviewer Expertise:
Tuberculosis, Family Support, medication compliance, TB promotion
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
The study design is appropriate, and the work is technically sound. The manuscript also demonstrates novelty, particularly in its attempt to link Google Trends terminology with TB burden at the provincial level. In addition, the methods and analytical procedures are described in sufficient detail to allow replication by other researchers. The underlying source data appear to be adequately available to support reproducibility, and the conclusions are generally supported by the results. Nevertheless, minor refinements in the discussion would further strengthen the manuscript. Several points should be highlighted. First, the discussion requires stronger support from current and relevant literature, particularly in paragraphs 2–5, to ensure that the interpretation is evidence-based and does not appear speculative. Second, although the statistical analysis is generally appropriate, it could be further strengthened by extending the analysis to a multivariate approach, considering the use of Google Trends terminology and its significant association with TB burden across more than two provinces.
Is the work clearly and accurately presented and does it cite the current literature?
Partly
If applicable, is the statistical analysis and its interpretation appropriate?
Partly
Are all the source data underlying the results available to ensure full reproducibility?
Yes
Is the study design appropriate and is the work technically sound?
Yes
Are the conclusions drawn adequately supported by the results?
Yes
Are sufficient details of methods and analysis provided to allow replication by others?
Yes
Reviewer Expertise:
Tuberculosis, Neuroscience, Neurondocrine, Physiology
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Please strengthen the Introduction by explaining more clearly why Google Trends is expected to reflect tuberculosis burden in the Indonesian setting. The Discussion section contains some repetitive interpretation of symptom-related and “lung spot” search terms. Consider tightening these paragraphs for better readability. Please include a few more international studies on TB-related infodemiology/digital surveillance to provide broader global context beyond Indonesia and South Africa. Some conclusions in the manuscript appear stronger than the presented findings. Consider using slightly more cautious language throughout the Discussion and Conclusion sections.
Please specify the exact Google Trends extraction settings used, including:
geographic filters, language settings, search category, search type. Clarify whether RSV values were extracted separately for each province and keyword before analysis. The normalization formula should be rewritten more clearly to improve reproducibility. Consider adding the complete extraction strategy as supplementary material.
The availability of both underlying and extended datasets through Figshare is a major strength of the manuscript and supports transparency and reproducibility.
The conclusions are broadly supported by the observed correlations. However, statements suggesting Google Trends as an “early warning system” should be phrased more cautiously, as the current analysis demonstrates association rather than predictive validation. Consider framing Google Trends as a complementary exploratory surveillance tool rather than a definitive epidemiological monitoring system.
Is the work clearly and accurately presented and does it cite the current literature?
Partly
If applicable, is the statistical analysis and its interpretation appropriate?
I cannot comment. A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?
Yes
Is the study design appropriate and is the work technically sound?
Partly
Are the conclusions drawn adequately supported by the results?
Partly
Are sufficient details of methods and analysis provided to allow replication by others?
Partly
Reviewer Expertise:
Tuberculosis, Public Health, Infectious Diseases, Epidemiology
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.