Analyzing Google Trends to Understand Seasonality and Interest in Spinal Cord Stimulators

Introduction: Spinal cord stimulators are implanted into over 30,000 patients’ spines annually. Google Trends (GT) can be utilized to determine public interest in topics like spinal cord stimulators within a certain filter. With approximately 35% of patients looking up their diagnoses1, GT can be analyzed for external events and seasonality. The purpose of this study is to determine any correlation between search interest and placement of spinal cord stimulators while also examining seasonality and state-level trends. We hypothesize a strong positive correlation between searches for spinal cord stimulators and the implantation of spinal cord stimulators.

Methods: With IRB exemption, GT analysis for “Spinal cord stimulator” was performed from 2004-2021 under the “Health” category in the United States and worldwide. The Healthcare Cost and Utilization Project Database (HCUP) was used for 2016-2018 CNS012 CNS Neurostimulator procedures because it was the most recent continuous timespan with no incomplete clinical data. Statistical analysis was performed in SPSS.

Results: Recent “spinal cord stimulator” searches are strongly correlated with spinal cord stimulator procedures from 2016-2018 (R2=0.9808). One-way ANOVA from before COVID-19 (2004-2019) and including COVID-19 (2004-2021) detected a statistically significant repeating monthly pattern (or seasonality) in GT scores for August in the United States (p=0.0253,p=0.0246) as well as worldwide (p=0.0128,p=0.0092). There was no statistically significant change in GT scores due to COVID-19, based on two-tailed, two-sample z-tests. State-level analysis indicated increased interest in spinal cord stimulators in South Dakota (p=0.0125) and decreased interest in Rhode Island (p=0.0379).

Conclusion: The recent, strong positive correlation between spinal cord stimulator public search interest (2016-2018) and neurostimulator procedures indicates how receiving the procedure may affect search interest. This also suggests that GT data can be used to reflect future clinical data. Increased interest in August should be noted by public health initiatives to guide information and education efforts. Increased interest in South Dakota and decreased interest in Rhode Island can optimize public information efforts. GT provides data on search interest but does not provide contextual information on the user performing the search, representing a limitation of the study. Because GT analysis can only indicate correlation, there may be other unexplored factors driving search interest; however, given how robust the positive correlation was, we assume this to be unlikely. Future studies can elucidate the underlying factors explaining temporal and state-level differences in search interest.