PTU-076 Using google search trend data to assess public interest in upper GI cancer symptoms

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction

Search engine data has been used to predict disease outbreaks based on search volumes for symptoms. Trend data can also assess information-seeking behaviour for cancer symptoms.

Introduction

‘Be Clear on Cancer’ (BCOC) for upper GI cancers ran 26/1/15–28/2/15 and resulted in increased two week wait (2ww) OGD referrals. We use Google search data to identify information-seeking behaviour for symptoms of upper GI cancers and corresponding local 2ww referral patterns in response to BCOC.

Methods

Search data from 10/2/13–3/2/18 for: ‘heartburn’, ‘indigestion’, ‘reflux’ and ‘heartburn AND cancer’ was extracted from ‘Google Trends’ and ‘Google Adwords’. Google Trends normalised weekly search traffic to a ‘relative search volume’ (RSV) index between 0 (<1% of the peak weekly search volume) and 100 (equivalent to the highest weekly search volume during the study period).

Methods

Data was compared to monthly 2ww referrals for upper GI cancers at our centre over the study period. Lower GI cancer referrals were used as a control.

Results

A mean annual increase in weekly RSV (compared to the annual mean RSV) for the terms ‘heartburn’ and ‘indigestion’ in the last week of December annually (from 2013 to 2017) of 67% and 60% respectively (figure 1). In the last week of January 2015, there was also a peak in RSV for ‘heartburn’ of 73%, corresponding to BCOC. This peak was higher than the Christmas 2014 increase (56%) and not replicated in the study period.

Results

Searches for ‘heartburn AND cancer’ reached their peak RSV (100) in the last week of January 2015, correlating with BCOC (figure 2). Mean RSV over the entire five year reference period was 7.48, with a 1336% increase in searches during the first week of BCOC, dropping to 722% in week 2% and 173% by week 5 - mean 673% overall.

Results

Data from our centre showed a mean 19.2% increase in 2ww referrals for suspected upper GI cancers in January of every year (compared to the mean number of monthly 2ww referrals for the preceding 12 months), except January 2016 where there was an 11% reduction in the number of referrals. There was an 29.9% increase in 2ww referrals from January to February 2015 which corresponds to BCOC. Suspected colorectal cancer referrals observed a mean 3% increase each January compared to the preceding 12 month average.

Conclusions

Peaks in information-seeking behaviours were observed each Christmas for upper GI symptoms resulting in a significant increase in 2ww referrals. BCOC generated a larger increase in interest with additional cancer association, but impact (as measured by information seeking) fell sharply over the course of the campaign.

Conclusions

Future BCOC programmes may benefit from 2 separate campaigns of reduced duration. Service pressures during such campaigns may be eased by such campaigns avoiding January.

Related Topics

    loading  Loading Related Articles