Using Large-scale Social Media Analytics to Understand Patient Perspectives About Urinary Tract Infections: Thematic Analysis

Background

Current qualitative literature about the experiences of women dealing with urinary tract infections (UTIs) is limited to patients recruited from tertiary centers and medical clinics. However, traditional focus groups and interviews may limit what patients share. Using digital ethnography, we analyzed free-range conversations of an online community.

Objective

This study aimed to investigate and characterize the patient perspectives of women dealing with UTIs using digital ethnography.

Methods

A data-mining service was used to identify online posts. A thematic analysis was conducted on a subset of the identified posts. Additionally, a latent Dirichlet allocation (LDA) probabilistic topic modeling method was applied to review the entire data set using a semiautomatic approach. Each identified topic was generated as a discrete distribution over the words in the collection, which can be thought of as a word cloud. We also performed a thematic analysis of the word cloud topic model results.

Results

A total of 83,589 posts by 53,460 users from 859 websites were identified. Our hand-coding inductive analysis yielded the following 7 themes: quality-of-life impact, knowledge acquisition, support of the online community, health care utilization, risk factors and prevention, antibiotic treatment, and alternative therapies. Using the LDA topic model method, 105 themes were identified and consolidated into 9 categories. Of the LDA-derived themes, 25.7% (27/105) were related to online community support, and 22% (23/105) focused on UTI risk factors and prevention strategies.

Conclusions

Our large-scale social media analysis supports the importance and reproducibility of using online data to comprehend women’s UTI experience. This inductive thematic analysis highlights patient behavior, self-empowerment, and online media utilization by women to address their health concerns in a safe, anonymous way.

Time Trends in Use of Radical Prostatectomy by Tumor Risk and Life Expectancy in a National Veterans Affairs Cohort

IMPORTANCE

Guidelines endorse using tumor risk and life expectancy (LE) to select appropriate candidates for radical prostatectomy (RP), recommending against treatment of most low-risk tumors and men with limited LE.

OBJECTIVE

To investigate time trends in the use of RP by tumor risk and Prostate Cancer Comorbidity Index (PCCI) score in a contemporary, nationally representative Veterans Affairs (VA) cohort.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study of 5736 men treated with RP at 8 VA hospitals from January 1, 2000, to December 31, 2017, used a nationally representative, multicenter sample from the VA SEARCH (Shared Equal Access Regional Cancer Hospital) database. Statistical analysis was performed from June 30, 2018, to August 20, 2020.

MAIN OUTCOMES AND MEASURES

Stratified linear and log-linear Poisson regressions were used to estimate time trends in the proportion of men treated with RP across American Urological Association tumor risk and PCCI (a validated predictor of LE based on age and comorbidities) subgroups.

RESULTS

Among 5736 men (mean [SD] age at surgery, 62 [6] years) treated with RP from 2000 to 2017, the proportion of low-risk tumors treated with RP decreased from 51% to 7% (difference, -44%; 95% CI, -50% to -38%). The proportion of intermediate-risk tumors treated with RP increased from 30% to 59% (difference, 29%; 95% CI, 23%-35%), with unfavorable intermediate-risk tumors increasing from 30% to 41% (difference, 11%; 95% CI, 4%-18%) and favorable intermediate-risk tumors decreasing from 61% to 41% (difference, -20%; 95% CI, -24% to -15%). The proportion of high-risk tumors treated with RP increased from 18% to 33% (difference, 15%; 95% CI, 9%-21%). Among men treated with RP, the proportion with the highest PCCI scores of 10 or more (ie, LE <10 years) increased from 4% to 13% (difference, 9%; 95% CI, 4%-14%). Within each tumor risk subgroup, no significant difference in the rate of tumors treated with RP over time was found across PCCI subgroups.

CONCLUSIONS AND RELEVANCE

In this study, the use of RP shifted from low-risk and favorable intermediate-risk to higher-risk prostate cancer. However, its use among men with limited LE appears unchanged across tumor risk subgroups and increased overall.

Female Lower Urinary Tract Symptom Prevention and Treatment Strategies on Social Media: Mixed Correlation With Evidence

OBJECTIVE

To evaluate the level of evidence behind recommendations on social media for disease prevention in five lower urinary tract symptoms.

MATERIALS AND METHODS

We conducted a digital analysis of anonymous online posts on social media sites collected by a social media data mining service. One thousand posts about pelvic organ prolapse, stress urinary incontinence, overactive bladder, urinary tract infection, and interstitial cystitis/bladder pain syndrome were randomly selected. We analyzed these posts for recommendations regarding the prevention and treatment of these diseases, which were then compared to recommendations in available clinical guidelines and assessed for level of evidence.

RESULTS

A total of 158 of 1000 posts contained 239 prevention strategies. For pelvic organ prolapse, there were 41 strategies identified, 25 (61%) of which had no evidence. For urinary tract infection 14 of 58 (29%) had no evidence, including recommendations for dietary modifications and urinary alkalization. For overactive bladder 8 of 28 (29%) had level 4 or no evidence. For stress urinary incontinence, 12 of 34 (36%) of prevention strategies had no evidence, such as laser rejuvenation and bladder training. Interstitial cystitis had the highest number of prevention strategies, and most were low or nonevidence based (70/79, 89%).

CONCLUSION

Prevention and treatment strategies are common in online discussions of pelvic floor disorders, but at least one third of these recommendations have no evidential support. There is a role for further online education and social media engagement by health care specialists to promote evidence-based practices.

Social media analytics of overactive bladder posts: what do patients know and want to know?

OBJECTIVE

To assess women's knowledge, patient experience, and treatment decision making regarding overactive bladder (OAB) using digital ethnography.

METHODS

Online posts were identified using a data mining service. Two hundred randomized posts were reviewed and coded using grounded theory. We then applied a latent Dirichlet allocation (LDA) probabilistic topic modeling process to review the entire collection of identified posts.

RESULTS

A total of 2618 posts by 1867 unique users from 203 different websites were identified. Our analysis yielded six themes: the impact of OAB on quality of life, patient-physician interactions, online engagement, symptom management, patient knowledge acquisition, and alternative therapies.

CONCLUSION

Overall, online communities are a source of support for women to self-manage the OAB symptom complex and help overcome treatment pathway challenges. Digital ethnography provides insight into patient knowledge and barriers to patient-centered care, which are important to improve patient outreach. Additionally, we identify similar findings to prior work, indicating the reliability of studying social media.

Using Digital Ethnography to Understand the Experience of Women With Pelvic Organ Prolapse

OBJECTIVE

To characterize the decision-making process and illness experience of women with pelvic organ prolapse (POP) using large-scale social media analysis.

METHODS

Digital ethnographic analysis of online posts identified through data mining was performed. Grounded theory methodology was applied to 200 posts via traditional hand coding. To supplement our qualitative approach, we applied a Latent Dirichlet Allocation probabilistic topic modeling process to review the entire data set of identified posts to ensure thematic saturation.

RESULTS

There were 3451 posts by 2088 unique users from 117 websites worldwide that were identified via social media data mining. We found that the anonymity of online forums allowed for information and support exchange among women with POP. Our analysis revealed that the exchange of online information aids in the decision-making process and, in some instances, appears to be the primary source of information. There was confusion about the anatomical and surgical complexities of prolapse. Our study also identified misconceptions, perceived risk factors, prevention methods, and management recommendations that were discussed online.

CONCLUSIONS

This large-scale online community-based analysis demonstrated the utility of social media to better understand women's experiences with POP. Thematic findings highlighted essential concerns and challenges involved in the surgical decision-making process and the understating of the anatomical complexity of sector defects, specifically to cystocele, rectocele, State specific defects.

Women's Experience with Stress Urinary Incontinence: Insights from Social Media Analytics

PURPOSE

We conducted a large-scale digital ethnographic analysis of anonymous online posts to capture the complete patient experience, knowledge and perceptions among women with stress urinary incontinence.

MATERIALS AND METHODS

Online posts were identified through data mining. Overall, 200 randomized posts were analyzed using grounded theory qualitative methods. To ensure full thematic discovery we also applied a Latent Dirichlet Allocation probabilistic topic modeling approach to the entire data set of identified posts. Latent Dirichlet Allocation topics are represented as a distribution of words, similar to a word cloud, which were manually reviewed to identify themes.

RESULTS

A total of 985 online posts by 762 unique users were extracted from 98 websites. There was significant overlap between the grounded theory and Latent Dirichlet Allocation identified themes. Our analysis suggests that these online communities help women manage the quality of life impact of their stress urinary incontinence, navigate specialty care and reach a decision regarding surgical vs nonsurgical management. Additionally, we identified risk factors, prevention strategies and treatment recommendations discussed online.

CONCLUSIONS

Findings demonstrated patient values that may influence decision making when seeking care for stress urinary incontinence and choosing a treatment. Social media interactions provide insight into patient behaviors that are important in order to improve patient centered care and decision making.