Once topics and subtopics were identified, it is now possible to examine trends to identify any patterns that may emerge within the data.

Overall trend of MDR-TB research

Increase in research publications from around 2005/2006

Plotting the number of publications per year over time reveals an interesting insight; prior to 2005/2006 there was a steady increase in the volume of publications, but after 2006 the number of publications increased rapidly.

trends_tb
Figure 1: Graph depicting number of publications per year and articles with the highest citations

Why might this be so? Researching this further, I noticed that 2006 was a turning point in the fight against drug-resistant tuberculosis, when data regarding XDR-TB (extremely drug-resistant TB) was described for the first time by the CDC and WHO. Given that XDR-TB is the most severe level of drug-resistance at the time, the emergence of it may have spurred the rapid increase in publication numbers on MDR-TB in the following years as efforts to prevent the rise of XDR-TB was intensified. This also corresponded with results from a separate citation analysis which revealed the publication of an influential paper on XDR-TB in 2006 by N Gandhi in the Lancet.

Highly cited articles in MDR-TB

The purple circles denote the 30 most highly cited publications that were found containing the search term ‘MDR-TB’ over the time period 1992-2018 (right axis).

The publication by Boehme et. al stands out with the number of citations that far exceeded all other publications. This article describes the results of assessment of the performance of Xpert MTB/RIF, a new rapid diagnostic tool for the detection of drug-resistant strains to Rifampicin (Boehme et al., 2010).

Specific trends within topics and subtopics

Largest increase in publication numbers seen for ‘Treatment optimization’

Publication output for the four largest topics can be seen to increase over the years. The output in terms of publications per year can be seen to be the highest for Topic 5 (Treatment optimization), followed by the topic ‘Drug-related research’.

trend_topics
Figure 2: Growth of publication numbers for the specific topics over time

The topics of ‘Molecular typing’ and ‘Immunology’ do not appear to be growing

In contrast for topics 0 and 6 which are the smallest topics identified by the algorithm, the output appears fairly constant over time with no clear trend twards a surge in publications in recent years, suggesting that these fields are not growing.

trend_topics_2
Figure 3: Growth of publication numbers for the two smallest topics in the study

Clustering of highly cited articles when research output increases

For all topics, there appear to be clustering of highly cited articles that correspond with an increase in the amount of research output within the field. This can be clearly seen in the case of topic 1 where highly cited articles began to appear around the time period when there was rapid growth in the number of research articles, beginning around and after 2004, which was initially preceded by a slower growth phase in terms of article numbers. On the other hand, this correlation appears less obvious for topic 5 as the increase in research output already began from the beginning of the data available (around 1996), thus all highly cited publications are evident from the start of the graph.

We can therefore observe that topic 5 began growing around 1996, followed by topic 1 around 2003/2004 and lastly topics 2 and 3 around 2006.

Insights from examination of subtopics

For topic 5, the most highly productive subtopic can be seen to be within subtopic 2 (Treatment regimens) and followed by subtopic 3 (Risk factors). For subtopic 2, a surge in publications appears to centre around 2012. For subtopic 3, publication output began to increase around 2008 after a period of slow growth.

trend_subtopics_5
Figure 4: Growth of publication numbers for subtopics 2 and 3 of Topic 5.

The biggest subtopic within topic 1 is subtopic 6 (New compounds) which appears to have rapidly grown between 2005 and 2009 where the number of publications per year peaked. This was followed by a decrease in output following 2009. On the other hand, subtopic 0 (Resistance genes) has generally been increasing from around 2007 up to the present moment.

trend_subtopics_1
Figure 1: Growth of publication numbers for subtopics 6 and 0 of Topic 1.

Although the number of articles are perhaps too low to derive any definitive conclusions for T1_S6, the trend suggests that the search for new compounds may have peaked around 2009/2010. It would be interesting to know if this is indeed the case and if so, why that may be. Also interesting is the concentration of highly cited articles with first authors from Indian institutions involved in this subtopic.

Conclusion

  • These visualizations allow us to view the relationship between publication numbers within the specific topics and subtopics and time.
  • By including the highest cited articles, we can have a sense of when a topic of research is booming.
  • Any patterns that are identified could be used as a starting point for deeper investigations.