Factors affecting drug reviews on WebMD

Michael Chin
Posted on Feb 18, 2018

 

For this web scraping project, the user reviews of three drugs were scraped from the website WebMD and analyzed to gain further insights into public opinions expressed on these class of anti-depressant drugs. Aggregating data directly from patient reviews produces insights into patient satisfaction not obtainable through clinical trials -where the focus is determining the safety and efficacy of the drug - and provides a few facet of data that can be used ensure that the patient is prescribed with the medication that would be the most effective for them. Medical professionals could use this information when making decisions about which medications to prescribe and to whom. Pharmaceutical companies may also use this information to better tune their marketing strategies, focusing their efforts on demographics that respond best to any particular treatment. In parallel, this data may help Pharmaceutical R&D centers identify where their current portfolio of products are lacking and plan for future products accordingly. To highlight the potential of these methods, 'Customer Behavior change' and 'Growth of consumerism' in health care are among the top five in perceived opportunities amongst Pharma company directors surveyed in a 2017 Healthcare Marketers Trend report.

 

Methods:

The reviews of three SNRI (Serotonin-Norepinephrine Reuptake Inhibitors) were scraped for this project: Cymbalta (Duloxetine by Eli Lilly), Effexor (Venlafaxine by Pfizer), and Pristique (Desvenlafaxine also by Pfizer). SNRI medications work by blocking the reuptake of neurotransmitters and are effective for treating depression and are also sometimes used to treat other conditions such as anxiety disorders and long-term pain. These three medications were selected because of the relatively high number of reviews on WebMD, and since all SNRI's work in a similar way, side effects are more comparable.

Scrapy, a Python-based open-source, web scraping engine, was used to obtain the review information:

Each review yields the following information

  • medical condition being treated
  • review date
  • reviewer age
  • gender
  • treatment length
  • user ratings on perceived effectiveness, ease of use, and satisfaction.
  • comments

 

Results and Discussion:

After collecting and cleaning the data, the initial exploratory analysis indicated that the mode demographics of each drug were the same. Females, between the age of 45 and 54, being treated for depression were the most common demographic for all drugs. In regards to satisfaction rating, all three drug satisfaction ratings were wide-ranging, with mean satisfaction rating at 2.9, 3.1 and 3.1 for Cymbalta, Effexor, and Pristiq respectively. Despite these apparent similarities, there is a potential lag in satisfaction rating comparing Cymbalta to it's competition. For the first iteration of the investigation, I took a closer look at these reviews to determine if A) This difference in mean review score is significant and B) If the factors contributing to this difference can be found from the data scraped. The distribution of user satisfaction is very non-normal. With a significant portion of the population leaving 1-star reviews.   

With the distribution as it is, treating the ratings as discrete-ordinal values would be inappropriate for T-testing. Instead, the reviews were binned into two categories. 1 and 2-star ratings were considered 'poor' or 'unsatisfactory' group while reviewers leaving a 4 or 5-star rating for satisfaction were defined as 'good' or 'satisfactory' group. Comparisons of these binary categories reveal a  stark difference between Cymbalta and the other two drugs.

But what could cause such a major uptick in negative reviews for Cymbalta? To answer this I also binned to various the drugs are used to treat into two major categories: Pain or Mood Altering, and comparison of the conditions distribution is made for each drug:

Cymbalta is the only medication of the three where it is significantly being prescribed for pain relief. Sure enough, the overall ratings for these SNRI's as a pain relief medication are generally poor, driving overall ratings of Cymbalta lower.

 

 

While these differences are statistically significant, further modeling will be necessary to determine the weights each factor has in determining patient satisfaction.

Conclusions and next steps:

With this examination into WebMD reviews of three SNRI class of drugs, we can begin to understand the how public opinion is formed on different drugs, despite similarities in function and target demographics. Cymbalta has significantly more poor reviews for their product because of its secondary application as a pain medication. With this information, medical practitioners may consider prescribing alternatives for chronic pain, and Eli Lilly may pivot marketing and research resources to addressing the concerns patients are having with their product as a pain medication. Future iterations of this work may include creating a logistical regression model to aid in predicting satisfaction and drug effective ratings for a given patient.

 

 

About Author

Michael Chin

Michael Chin

Michael is a Data Scientist with a strong background in data exploration and data analysis in physical science fields. 10+ years of experience in all aspects of research, from hypothesis formation to data collection, preparation, analysis, and communication.
View all posts by Michael Chin >

Related Articles

Leave a Comment

No comments found.

View Posts by Categories


Our Recent Popular Posts


View Posts by Tags

#python #trainwithnycdsa 2019 airbnb Alex Baransky alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus API Application artist aws beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep Bundles California Cancer Research capstone Career Career Day citibike clustering Coding Course Demo Course Report D3.js data Data Analyst data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization Deep Learning Demo Day Discount dplyr employer networking feature engineering Finance Financial Data Science Flask gbm Get Hired ggplot2 googleVis Hadoop higgs boson Hiring hiring partner events Hiring Partners Industry Experts Instructor Blog Instructor Interview Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter lasso regression Lead Data Scienctist Lead Data Scientist leaflet linear regression Logistic Regression machine learning Maps matplotlib Medical Research Meet the team meetup Networking neural network Neural networks New Courses nlp NYC NYC Data Science nyc data science academy NYC Open Data NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time Portfolio Development prediction Prework Programming PwC python python machine learning python scrapy python web scraping python webscraping Python Workshop R R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn Selenium sentiment analysis Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau team TensorFlow Testimonial tf-idf Top Data Science Bootcamp twitter visualization web scraping Weekend Course What to expect word cloud word2vec XGBoost yelp