Sage Journals: Discover world-class research

Abstract

Objective

People with disabilities (PWDs) are at greater risk of COVID-19 infection, complications, and death, and experience more difficulty accessing care. We analyzed Twitter tweets to identify important topics and investigate health policies’ effects on PWDs.

Methods

Twitter's application programming interface was used to access its public COVID-19 stream. English-language tweets from January 2020 to January 2022 containing a combination of keywords related to COVID-19, disability, discrimination, and inequity were collected and refined to exclude duplicates, replies, and retweets. The remaining tweets were analyzed for user demographics, content, and long-term availability.

Results

The collection yielded 94,814 tweets from 43,296 accounts. During the observation period, 1068 (2.5%) accounts were suspended and 1088 (2.5%) accounts were deleted. Account suspension and deletion among verified users tweeting about COVID-19 and disability were 0.13% and 0.3%, respectively. Emotions were similar among active, suspended, and deleted users, with general negative and positive emotions most common followed by sadness, trust, anticipation, and anger. The overall average sentiment for the tweets was negative. Ten of the 12 topics identified (96.8%) related to pandemic effects on PWDs; “politics that rejects and leaves the disabled, elderly, and children behind” (48.3%) and “efforts to support PWDs in the COVID crisis” (31.8%) were most common. The sample of tweets by organizations (43.9%) was higher for this topic than for other COVID-19-related topics the authors have investigated.

Conclusions

The primary discussion addressed how pandemic politics and policies disadvantage PWDs, older adults, and children, and secondarily expressed support for these populations. The increased level of Twitter use by organizations suggests a higher level of organization and advocacy within the disability community than in other groups. Twitter may facilitate recognition of increased harm to or discrimination against specific populations such as people living with disability during national health events.

Keywords

COVID-19 social media Twitter sentiment analysis disability health disparity health inequity discrimination crisis standard of care

Introduction

The COVID-19 pandemic disrupted health care systems worldwide. Two years after the start of the global pandemic, the World Health Organization reported more than 452 million cases and 6 million deaths from COVID-19 worldwide on March 11, 2022.¹ At that time, the virus accounted for a reported 78.7 million cases and 957,00 deaths in the USA, 3.34 million cases and 37,000 deaths in Canada, and 19.4 million cases and 162,000 deaths in the UK.¹ Forty-three percent of individuals who acquired COVID-19 experienced or are experiencing postacute sequelae of COVID-19 (PASC) or “long COVID.”² For patients who were or were not hospitalized due to COVID-19, the estimates were 54% and 34%, respectively.³ The COVID-19 pandemic left health care systems overwhelmed and unable to care for the rapidly growing number of patients resulting in capacity measures such as cancelling elective surgeries and nonurgent care. When these measures were insufficient, providers and organizations were forced to choose which patients to treat preferentially or to deny care.⁴

In emergency situations, crisis standards of care (CSC) may be used to determine how to allocate and ration limited resources⁵ by assigning patients CSC scores using acute and chronic severity of illness and life expectancy. Even though scores are supposed to lead to unbiased decisions, CSC priority scores result in fewer scarce resources allocated to Black patients.⁶ Evidence is emerging that CSC may disproportionately affect disadvantaged populations because comorbid conditions and expected lifespan are unequally distributed.⁷ Health service researchers have proposed alternate ways of distributing resources that seek to avoid discrimination against people with severe health conditions.⁸ Section 1557 of the United States Affordable Care Act and Section 504 of the Rehabilitation Act prohibit discrimination based on disability in programs and activities funded by the federal Department of Health and Human Services.⁹ During the COVID-19 pandemic, some health care organizations and providers were accused of failure to comply, with disability advocacy organizations in several states filing complaints with the Office of Civil Rights.^10–14

Existing CSC recommendations support the development of CSC plans that embody fairness and equitable processes, including transparency, consistency, proportionality, and accountability. Furthermore, they call for community and provider engagement, education, and communication, and the rule of law with regard to authority and environment.¹⁵ However, even when CSC are implemented in accordance with antidiscrimination mandates, access to care, the services of caregivers, and other resources may be limited, leading to decline in health and quality of life, disparities, and health inequities.^6,16–19 An analysis of US ventilator allocation guidelines and CSC guidance for ventilator reallocation indicates that inequities may result from the use of such guidelines^18,20 when community attitudes do not support ventilator reallocation.²¹

Adding to the effects of CSC, having any disability negatively affects the likelihood, severity, and outcome of COVID-19 infection. Heightened vulnerability results not only from disease- and disability-related characteristics such as reduced lung function but also from environmental factors such as living in a congregate setting (e.g. long-term care facility) or lack of access to protective equipment.^22–24 Disability often may result in a lower economic status and decreased access to care, medical products, and services (e.g. face masks and home grocery delivery).²⁵ COVID-19 screening and practices are variable and may not be performed in ways that are effective in some people (e.g. those with spinal cord injuries),²⁶ and test kits may not be available to people with disabilities (PWDs).²⁷ In addition, COVID-19 may present differently in PWDs, hampering timely diagnosis.^28,29 Reduced activity due to self-isolation or home lockdown may increase pain, make pain management strategies less effective, or increase the likelihood of health problems associated with lack of exercise.^30,31

Social media³² has been shown to be a valid approach to identifying patient perspective around health conditions, pandemics, preventive measures, and treatments.^33–35 The microblogging platform Twitter facilitates analysis of public perceptions and patient perspectives. More than 145 million daily active users³⁶ post, repost, and comment on “tweets”—short notes of up to 280 characters. Twitter commentary has been used as a reflection of public opinion on health conditions such as diabetes³⁷ and cancer and on legislation (e.g. the Affordable Care Act).³⁸ Twitter has acted as a tool for analysis of public opinion and sentiment on emerging infectious disease outbreaks such as measles,³⁹ influenza,⁴⁰ Zika,⁴¹ Mpox,⁴² and SARS-CoV-2.^34,35 We hypothesized that the public discussion on Twitter would provide insight into the experience of PWDs during the COVID pandemic.

Methods

Data collection and processing

On 14 January 2022, we used Twitter's application programming interface (API) to access Twitter's public COVID-19 stream.⁴³ We collected all English-language tweets for a 2-year span from 1 January 2020 to 14 January 2022. The tweets contained a combination of keywords related to COVID-19, the disability community, and topics of discrimination and inequity such as “Krip the vote,” “Disability Twitter,” and “Eugenics by CDC” (Appendix). We provided descriptive statistics for tweets, including user profiles and tweet content, and determined long-term tweet availability based on Twitter API status codes (“user suspended” or “no user found with this User ID”). We used Python version 3.9.1 software (Python Software Foundation, Wilmington, DE) for all data processing and analyses.⁴⁴ Institutional review board approval and informed consent were not required because this study used only publicly available data published voluntarily by Twitter users.

Descriptive analysis

We used the python library M3-Inference,⁴⁵ which uses a deep learning model trained on a Twitter data set using profile images, screen names, names, and profile descriptions to predict the Twitter users’ gender, age, and type of account. We used the ethnicolr library⁴⁶ in Python to obtain ethnicity information, which uses a model trained off the 2010 US census data to predict the user's ethnicity.

Sentiment and emotion analysis

Prior to performing a sentiment analysis on the tweets, we preprocessed the tweets into plain text, which required the removal of hyperlinks, Twitter handles, “#” symbols, and replies to tweets. We used Python's SentiStrength Library⁴⁷ to identify and classify these preprocessed tweets’ sentiment (positive or negative). SentiStrength uses a lexicon-based classifier as well as a rule-based algorithm to measure sentiment on a scale of −4 (most negative) to 4 (most positive). Because deleted or suspended accounts may represent individual accounts that ran afoul of the Twitter rules of engagement, we calculated the average monthly sentiment for all the tweets and compared them based on their account status (active, suspended, or deleted). Before performing an emotion analysis on the tweets, we use the Spacy library⁴⁸ to clean the text further. We tokenized and transformed the tokens into their root form through natural language processing techniques such as lemmatizing and removing stop words and nonalphanumeric characters. The removal of stop words eliminates words that provide little semantic meaning, such as “their,” “who,” and “is.” We used the Python library NRCLex⁴⁹ to label the primary emotion for each tweet (fear, anger, anticipation, trust, surprise, sadness, disgust, or joy).

Topic modeling

Topic modeling is an effective methodology for organizing large amounts of unstructured social media data (like a library of tweets) into topics. Conventional methods, such as content analysis, rely on human coders to identify topics, but these can be slow and unreliable. In contrast, topic modeling uses machine learning algorithms, such as latent Dirichlet allocation (LDA), to automatically analyze textual data and identify topics. This approach is especially valuable during fast-moving events, such as pandemics, when timely understanding of public sentiment and timely responses from public health agencies are critical.⁵⁰

To clean the data by removing tweets unrelated to COVID-19 and disabilities, we used the gensim⁵¹ library in Python and applied an unsupervised machine learning algorithm called LDA⁵² to group tweets using a representative set of words into word clusters. We then used the most highly weighted words in each cluster to determine the content of each topic. We removed tweets with topics unrelated to COVID-19 and inequities in the disabled community. With the remaining tweets, we then performed an additional topic modeling and analyzed topics by word clusters to determine the content of each topic. Topic modeling allows the discovery of themes or patterns in a corpus. A pattern describes a set of recurring words that suggest a topic of discussion that reoccurs in multiple tweets.

To optimize the number of topics in our analysis, we trained and evaluated several LDA models with topics ranging from 2 to 50 topics based on their topic coherence score, which summarizes the semantic similarity among high scoring (frequent) words within topics. We ultimately chose a 12-topic LDA model as it produced the highest score. An author without access or insight into the topic modeling labeled the topics using the 20 most frequently used terms, which were ranked by weight. All authors evaluated these topic labels, reached a consensus, and identified example tweets whose content pertained >99% to a specific topic.

Results

Descriptive analysis

Between 1 January 2020 and 14 January 2022, we identified 134,443 English-only tweets related to COVID-19, the disabled community, and topics of discrimination and inequality among the disabled community. After the removal of duplicate tweets, tweets that were replies, and retweets, we retained 94,814 tweets from 43,296 accounts. Of all accounts, 1068 were suspended during our observation period (2.5%) and 1088 (2.5%) were deleted (Table 1) while 41,139 (95%) remained active Twitter accounts. Among verified users tweeting about COVID-19 and disability, suspension (0.13%) or account deletion (0.3%) was significantly less common (p < 0.001).

Table 1.

Characteristics of Twitter users.

	All Users		Active Users		Suspended Users		Deleted Users
	n = 43,296		n = 41,140		n = 1068		N = 1088
User characteristics
Verified Twitter account	3026		3013		4		9
User followers	16,395		17,119		3886		1300
User posts to date	30,967		31,041		43,839		13,001
Demographic characteristics	n	(%)	n	(%)	n	(%)	n	(%)
Age group
≤18	4941	11.4	4567^a,b	11.1	175^c	16.4	199	18.3
19–29	3996	9.3	3702^a,b	9.0	131^c	12.3	163	15.0
30–39	11,854	27.3	11,231^b	27.3	301	28.2	322	29.6
≥40	22,505	52.0	21,640^a,b	52.6	461^c	43.1	404	37.1
Gender
Male	22,879	52.8	21,599^a,b	52.5	670^c	62.7	610	56.1
Female	20,418	47.2	19,542^a,b	47.5	398^c	37.3	478	43.9
Ethnicity
Asian	4786	11.1	4484^d	10.9	166^e	15.6	136	12.5
Hispanic	3668	8.4	3497^a	8.5	79	7.4	92	8.5
Non-Hispanic Black	7414	17.2	7035	17.1	183	17.1	197	18.1
Non-Hispanic White	27,427	63.3	26,124^a,f	63.5	640	59.9	663	60.9
Type of account
Organization	19,027	43.9	18,307^a,b	44.5	374^g	35.0	346	31.8
Personal	24,269	56.1	22,833^a,b	55.5	694^g	65.0	742	68.2

Comparison between the groups is done using the Welch t-test.

Significant difference between active and suspended users (P < 0.001).

Significant difference between active and deleted users (P < 0.001).

Significant difference between suspended and deleted users (P < 0.032).

Significant difference between active and suspended users (P < 0.01).

Significant difference between suspended and deleted users (P < 0.010).

Significant difference between active and deleted users (P < 0.02).

Significant difference between suspended and deleted users (P < 0.017).

Table 1 shows the demographics of Twitter users, including age, gender, and ethnicity. The largest group of users included users 40 years or older (52%). Compared to other topics our group had explored regarding COVID-19, this percentage was higher (Plandemic/Scamdemic 45% and Breastfeeding (manuscript in preparation) 33%).⁵³ Males and non-Hispanic Whites represented the largest groups of all user groups (Table 1). Male users under the age of 40 were overrepresented significantly among accounts that had been suspended or deleted. Asian users were significantly overrepresented in suspended accounts compared to active accounts (15.6% vs. 10.9%, p < 0.001), whereas Hispanic users were significantly underrepresented among suspended accounts (7.4% vs. 8.5%, p < 0.001). The vast majority of suspended and deleted users tweeted from personal accounts, 65.0% and 68.2%, respectively. Suspended users, on average, posted more than active users and had fewer followers than active users.

Of all 94,814 tweets, 90,500 (95.5%) tweets were published by 41,140 (95.0%) active users (2.2 tweets per user), 2585 (2.7%) tweets were published by 1068 (2.5%) subsequently suspended users (2.4 tweets per user), and 1729 (1.8%) of tweets were published by 1088 (2.5%) user accounts subsequently deleted (1.6 tweets per user) (Table 2).

Table 2.

Characteristics of tweets.

	All Tweets		Active		Suspended Tweets			Deleted Tweets
Tweets	n = 94,814		n = 90,500		n = 2585			n = 1729
Users	n = 43,296		n = 41,140		n = 1068			n = 1088
Characteristics	N	(%)	N	(%)	N	(%)	N		(%)
Has reply	10,782	11.4	10,439^a,b	11.5	209	8.1	134		7.8
Has like	25,611	27.0	24,741^a,b	27.3	559^c	21.6	311		18.0
Has retweet	28,861	30.4	27,623^a,b	30.5	884^d	34.2	354		20.5
Has been quoted	8673	9.1	8429^a,b	9.3	158	6.1	86		5.0
Twitter source
Twitter Web App	33,682	35.5	32,341^a	35.7	757^d	29.3	584		33.8
Twitter for iPhone	22,780	24.0	21,226^a,b	23.5	1014^e	39.1	540		31.2
Twitter for Android	18,058	19.0	16,967^a,b	18.7	632	24.4	459		26.5
Hootsuite Inc.	3811	4.0	3792^a,b	4.2	4^d	0.01	15		0.1
TweetDeck	2553	2.7	2528^a,b	2.8	18	0.7	7		0.4

Comparison between the groups is done using chi-square test.

Significant difference between active and suspended users (P < 0.001).

Significant difference between active and deleted users (P < 0.001).

Significant difference between suspended and deleted users (P < 0.003).

Significant difference between suspended and deleted users (P < 0.001).

Significant difference between suspended and deleted users (P < 0.002).

Users who were suspended were statistically more likely to have their tweets retweeted (p < 0.001) compared to active users. Twitter Web App was the most used platform by active users (35.5%). Users who were suspended were statistically more likely to be iPhone users (39.1%) compared to active (23.5%) and deleted users (31.2%). Tweets posted by active users were statistically more likely to contain replies, likes, and be quoted by other Twitter users (p < 0.001).

Our analysis explores whether accounts belong to individuals versus organizations. With 43.9%, the percentage of organizational accounts was higher for the disability topic than any other COVID-19-related topic that we reviewed in the past.

Emotion analysis

Emotions were similarly distributed among active, suspended, and deleted users. Generalized negative and positive emotions were most common followed by sadness, trust, anticipation, and anger for all tweet groups (Figure 1). Tweets from suspended accounts were more likely to express fear than tweets from active and deleted accounts.

Figure 1.

Emotion analysis for active, suspended, and deleted accounts.

Sentiment analysis

The overall average sentiment for the tweets on a scale from −4 (most negative) to 4 (most positive) was negative, as shown in Figure 2. The degree of negative sentiment varied among the groups, with tweets from active users being less negative than those posted by suspended and deleted users. Throughout the study period, suspended tweets contained the most negative sentiment out of all the groups.

Figure 2.

Sentiment analysis (scales −4 to 4). Suspended accounts generated tweets that were more negative.

Topic models

The LDA model identified 12 topics expressed in our sample of tweets. The topics were labeled subjectively based on their respective keywords (Table 3). Two topics (representing only 3.2% of all tweets) were not per se related to PWD. One topic (1.4%) discussed the disinformation relating to how the vaccine “can disable” a person and the other (1.8%) focused on the disability resulting from long COVID. The remaining 10 topics (96.8%) were focused on the effects of the pandemic on PWD. Nearly half of the tweets belonged to the topic “Politics that neglect and leave behind PWD, older adults, and children” (48.3%). The next most popular topic was “Efforts to support PWD in the COVID-19 crisis” (31.8%). The third most common topic, “Disabled people need better access to vaccines and other support,” accounted for only 6.3% of tweets. Analysis of sentiment strength, mean retweets, and mean likes (Table 4) indicated that the most negative sentiments (calls for political action to protect PWD from COVID-19 and politics that neglect and leave behind PWD, older adults, and children) did not necessarily receive the largest numbers of retweets and likes.

Table 3.

Table of topic models.

Topic	Tweets/Topic, n (%)	Keywords	Representative Tweet
1. Anger over antivaxxers	1186 (1.3)	covid, disability, student, vaccine, new, year, home_community, disability_barrier_aodaalliance, onpoli_crpd_accessiblecanada, year_uphill_battle_unprecedented, end_report_accessibility_aoda, people, coronavirus, disabled, impaired, study, impairment, shortage_carehome_disabled, sars, metropolitanpolice_riotpolice	You’re congratulating a man who was complicit in the deaths of over 80 Samoans, mostly kids, and the sickness/disability of 6000 more, because he persuaded them to avoid vaccination against measles.He only escaped jail on a technicality.#AntiVaxKills#AntiVax is #ChildAbuse
2. Discussing how vaccines can disable you (vaccine complications)	1374 (1.4)	covid, vaccine, cognitive_impairment, disability, symptom, patient, people, new, vaccination, disabled, city, brain_fog, impairment, information, day_christmas_pls_sign, petition_help_carers_shortage, carehome_disabled_home_community, activity, learn, long_covid	@OllyOllyInFree #StopTheEXPERIMENT no one is liable if you are disabled or killed using the experimental transfection agent disguised as corona ðŸ’‰,they skipped animal trials so you’re the lab rat that may experience fatal immune enhancement reactionhttps://t.co/wrB6J1cpVN
3. Masks protect disabled kids, antimask rules disproportionately affect them	871 (0.9)	covid, disability, people, vaccine, school_mask, sign_support, life_threaten, mask_school_protect, prophesy_coronavirus_december_ravaging, locusts_plague_live_jesus, law_attorney_lawyer, law_news_disabilitylaw_news, lord_acripplehaswalked, instagram, long_covid, disabled, coronavirus, kid, chronlaw_disability, death_permanent	Federal judge rules Texas school mask ban violates Americans with Disabilities Act.#ChicagoPd #AEWDynamite #MAFS #CMAawards #MarriedAtFirstSight #SistasOnBET #COVID19 https://t.co/ZkdpZmRlsI
4. PWDs are affected more through worse outcomes and discrimination	3119 (3.3)	people, disability, covid, disabled, coronavirus, learn, pandemic, england, patient_learn, family, intellectual_disability, support, life_save, time_likely_die, need, lockdown, facebook, fury_resuscitate_notice_give, vaccine, onpoli_crpd	@kidneyontario @ottawadonors @UHNTransplant @CanadasLifeline News release: The Danger of flagrant disability discrimination in access to life-saving critical care grows as overloaded Ontario hospitals near breaking Point https://t.co/nEFf75lstc #accessibility #AODAfail #COVID19 #triage #onpoli #CRPD
5. CSC—crisis standards of care—and the fear that they will be used to withhold/deny care to PWD.	935 (1.0)	covid, disability, disabled, people, woman, vaccine, healthcare_oakgar_oakgarrecruitment_carehome, healthandsocialcare_socialcare, learningdisabilitie_elderly_dementia_supportedlive, mentalhealth, highriskct, age_base, aoda_accessibility, disabilities_remain_live_worry, vaccination, year_secret_ontario_critical, hospitals_threat_discrimination_patient, care_triage_protocol_send, danger, home_community_care_pls	@POTUS @GovNedLamont @CDCgov @NatCounDis @DrNunezSmith Connecticut has no procedure for disabled residents with CDC-listed high-risk conditions to request a reasonable exception or modification to their age-based-only COVID vaccine distribution policy. We are last in line. Please help us. #HighRiskCT
6. PWDs need better access to vaccines and other support	5981 (6.3)	vaccine, disability, people, covid, auspol, vaccination, vaccinate, learning_disability, lose_job, eligible, worker, australia, resident, read, hate_ask_money_know, vaccine_rollout, aged_care, mutual_aid_request_long, term_covid_stephanie, dose	@karamargin, @AZDanielRuiz CDC: States should prioritize people with disabilities as they broaden vaccine access. Listen up AZ! #CovidVaccine #HighRiskAZ, #Disabilityhasnoagerequirement, #ADAaccess, #ADAcompliance
7. Efforts to support PWD in the COVID-19 crisis	30,140 (31.8)	disability, people, covid, pandemic, support, person, impact, learn, need, community, work, help, disabled, new, access, inclusion, today, world, include, live	Last week we organized a consultation with over 15 OPD reps in Indonesia to help @FCDOGovUK refresh its #disability inclusion strategy. Grantees spoke of the challenges persons with disabilities have faced during #COVID19 & how donors can support ongoing efforts led by OPDs. https://t.co/IAV289rt2I
8. Donation and aid to PWD	1236 (1.3)	help, disability, people, covid, vaccine, learn, disabled, easy_read, christmas, support, donate, app, link, vaccination, instagramdisable, need, video, pandemic, check, gofundme	Donated Ration and groceries supplies to the visually impaired families of Gomathipuram, Thiruninravur 5/06/21#geoindiafoundation #covid19relief #covidindia #ration #charity #communityhelp #chennai #ngo #visuallyimpaired https://t.co/F0iozlkw1e
9. Politics that neglect and leave behind PWD, older adults, and children	45,803 (48.3)	covid, disabled, people, disability, long_covid, vaccine, die, death, need, child, disable, know, get, like, life, go, care, year, time, live	$100 for a rapid #covid test in America over the counter during a pandemic thanks to capitalism & #JenPsaki mocks the idea of sending Americans free covid tests while sick, poor & disabled people wait in lines outside in the cold for hours to get testedAmerica is a failed state
10. Long COVID and the resulting disability	1699 (1.8)	help, long_covid_survivor_financially, need, long_covid, disability, disabled, unable_work_struggle_desperately, disable_yr, fault_help_senatedems_nytime, aoc_ewarren_sensanders_repcori, fellow_hurt_human, tesla_billgates_microsoft_help, one_pandemic_amazon_elonmusk, people, work_forget, income, neisvoid_disability_twitter, covid, cfs, covidisnotover	@MSNBC Please help #LongCovid survivors financially. We are unable to work and struggling desperately. Iâ€™ve been disabled by #COVID19 for 1 yr through no fault of my own. HELP!! ðŸ†˜â€¼ï¸@HouseDemocrats @SenateDems @nytimes @AOC @ewarren @SenSanders @RepCori @PressSec @RepPressley @elonmusk
11. Lack of access and the need for technology and/or other solutions that would create access	1131 (1.2)	covid, disability, vaccine, people, home_community_based_services, discrimination_ontario_criticalcare_triageprotocol, publichealth_publicservicefoodsecuritysanitation_disinfection_services, prioritization_coronavirus, vaccinations_elderlythe, disable_homelesspublicservants_critical_mission, blast_disability, new_torontostar_column_davidlepofsky, ford_government_fordnation_secrecy, aoda_onpoli_onhealth_crpd, surround_triage_plan_accessibility, ethiopia, disabled, addisababa, key_include_ppl, disability_community_exacerbate	@SenDuckworth Home & Community Based Services are key to including ppl w/ #disabilities in the community, but #COVID19 exacerbated a long-standing workforce crisis in this field. Pls support $400B proposed by @POTUS for #Medicaid #HCBS in budget bill!
12. Calls for political action to protect PWD from COVID-19	1339 (1.4)	disability_need_real, action_plan_delay_create, hardship_covid_crisis_aoda, people, doug_ford_fordnation_day, accessibility, call_strong_new_action, year_get_hondavidonley_report, covid, action_accessibility_people, report_call_strong_new, disability, help_desperately, financially_disabled_unable_work, forget_one_pandemic_humanitarian, help_covid_longhaulers, month_get_hondavidonley, gaza_strip, zionists_besiege, disability_freepalestine	@TheDuhalde @triofrancos My alma mater, too! #WeSupport students & workers in their calls for protection.#COVID19 is a humanitarian crisis & mass disabling event. @WhiteHouse & @POTUS must do more.Take action with us THURSDAY: https://t.co/lnpR8T3rbX

Table 4.

Table of topics with sentiment strength, mean retweets, and mean likes.

Topic	Sentiment Strength	Mean Retweets	Mean Likes
1. Anger over antivaxxers	−0.75	1.13	1.99
2. Discussing how vaccines can disable you (vaccine complications)	−0.79	2.94	5.88
3. Masks protect disabled kids, antimask rules disproportionately affect them	−0.48	3.79	6.08
4. PWDs are affected more through worse outcomes and discrimination	−0.83	2.75	3.63
5. CSC (crisis standards of care) and the fear that they will be used to withhold/deny care to PWD	−0.35	1.94	3.81
6. PWDs need better access to vaccines and other support	−0.58	3.26	8.08
7. Efforts to support PWD in the COVID-19 crisis	−0.11	3.25	7.50
8. Donation and aid to PWD	−0.05	3.20	3.90
9. Politics that neglect and leave behind PWD, older adults, and children	−0.95	9.49	29.14
10. Long COVID and the resulting disability	−0.66	1.12	1.93
11. Lack of access and the need for technology and/or other solutions that would create access	−0.66	1.36	1.96
12. Calls for political action to protect PWD from COVID-19	−1.11	0.85	1.26

Discussion

Our research on early COVID-related Twitter data, which predates even the coining of the term “COVID,” uncovered important themes that had not yet gained widespread public attention. As early as January 2020 (2 months prior to the stock market crash), Twitter users expressed concerns about the pandemic's economic impact.^33,34 Moreover, conversations at that time prioritized respiratory protection measures such as wearing masks and practicing social distancing, even before it was established that COVID-19 primarily spreads through respiratory droplets. Harnessing social media data could enable us to leverage the collective knowledge and insights of the masses to identify emerging issues and potential interventions.

PWDs experience worse outcomes following COVID-19 infection than those without disabilities.^23,54 Any disability is associated with increased risk for in-hospital mortality.⁵⁵ People with intellectual and developmental disabilities (IDDs) are at a higher risk for infectious disease-related emergency room visits, hospitalization, and death in hospitals.^56,57 They were admitted to hospital with more severe symptoms (e.g. seizures) than non-IDD individuals, deteriorated more quickly, were younger at death, and experienced more difficulty accessing care teams than those without IDDs.^58,59 PWDs had longer hospital stays, more frequent readmissions within 30 days, and a greater likelihood of severe outcomes including death.^60–62 As the pandemic progressed and health care systems became strained, some PWDs found it more difficult to receive their usual level of care⁶³ or needed medical supplies.⁶⁴

Our analysis of tweets between 1 January 2020 and 14 January 2022 demonstrated that the most important topic (48.3% of tweets) was a discussion on how politics and policies leave disadvantaged populations like PWD, older adults, and children behind. This immediately linked to a discussion of how to help, aid, and protect these populations (31.8%). This focus on the need for support of PWD reinforces research findings that poorer mental well-being, greater occupational stress, difficulty using digital health tools for support, and having their needs go unacknowledged and unaddressed within the community contribute to COVID-related illness and death.⁶⁵

Sadness and fear were among the highest ranked emotions, and overall, the tweets had a negative sentiment throughout. This finding is not surprising given that PWDs were more likely to experience symptoms of psychological stress (e.g. depression, suicide ideation, and anxiety) and report loneliness than people without a disability.^66–68 These sources of stress included pandemic-related concerns such as inadequate access to health care, as well as nonpandemic issues such as emotional or physical abuse and inadequate food.⁶⁷ The poorer health outcomes experienced within this community, noted previously, have been found to have resulted in part from COVID-19 control measures that failed to adequately address PWD human rights and socioeconomic well-being.⁶⁹ Heightened visibility of medical ableism^70,71 and the shortcomings of “one size fits all” approaches to public health planning have renewed decades-long calls for rethinking and restructuring of health care for PWD.^72,73

In addition, our samples of tweets were more likely to be tweeted by organizations than other topics we have explored in the past. This may reflect a high level of organization and advocacy among the PWD community, as well as a growing movement for disability justice focusing on health care-related institutions and policies. Recent work in this area has advocated for the involvement of PWD in disaster planning to address the needs of PWD better,⁷⁴ more accurate data collection about PWD to quantify COVID-19-related disparities better and disparity reduction strategies,⁷⁵ engagement with the disability community on the development of informatics-based solutions for health disparity reduction,⁷⁶ explicit efforts to identify and eliminate structural ableism within health care including medical rationing,⁷⁷ and rethinking of the field of bioethics’ reflexive disregard for PWD.⁷⁸

Although fewer accounts were suspended in the analysis of this disability and COVID-19 theme than in a prior study on the hashtags #scamdemic and #plandemic, in which more than 20% of accounts were suspended, still in this work, 2.5% of accounts were suspended and 2.5% deleted after a year, which is a fairly high rate. Further, we found that tweets generally had negative sentiments. These findings suggest that this topic generated passionate discussion or outrage, which is more likely to result in reporting of tweets by other users. Users whose accounts were suspended had the most retweets and the highest negative sentiments throughout the observation period, suggesting inflammatory statements more likely generating interests by other Twitter users. In January 2023, iPhones and Android phones accounted for 55% and 45% of the market, respectively.⁷⁹ We noticed a relative increase in suspended users who used iPhones, which may have resulted from greater usability of iPhones for PWDs.⁸⁰

Limitations

Our study has a number of limitations. First, we used existing tools to analyze sentiments and emotion of tweets that are not specific to health care topics, which could have skewed our analysis. Second, because we targeted only tweets in English and are unable to determine precise geographic locations for users, we are limited in making conclusions about specific countries or countries where English is the not the predominant language. Third, our keywords selected 10 topics that were indeed focused on PWDs and COVID-19, but 2 topics were not related to this topic and accounted for 3.2% of all tweets.

Conclusions

Twitter, the microblogging social media platform, hosted a robust and in part passionate (as indicated by suspended and closed accounts) discussion on the topic of disability and COVID-19. The predominant discussion focused on how politics and policies leave disadvantaged populations like PWD, the elderly, and children behind followed by the topic focused on helping these populations. Twitter may be a medium that allows public health officials to detect an increase in discrimination of specific populations like people living with disability in the context of a national crisis.

Footnotes

Contributorship

CP, CUL, RJM, and MID conceived the study. MID performed the analysis. CP and CUL led the manuscript writing, with critical inputs from RJM and MID. All authors interpreted the findings, reviewed, contributed to, and approved the final manuscript. The article processing charges for this article are supported by the Clinical Informatics Center at UT Southwestern Medical Center.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RJM has received grant funding from the Centers for Disease Control and Prevention and research funding from Verily Life Sciences, the Sergey Brin Family Foundation, and the Texas Health Resources Clinical Scholar program.

Ethical approval

Not applicable, because this article does not contain any studies with human or animal subjects.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Guarantor

CP.

ORCID iD

Carolyn Petersen

Trial registration

Not applicable, because this article does not contain any clinical trials.

Appendix

Keywords contained in Twitter's public tweet stream:

Crip the Vote

CripTheVote

CSC

Disability

DisabilityTwitter

Disabled

Disabled and access

Disabled and equity

EugenicsByCDC

Handicap

Impaired

Incapable and disabled

Incapacitated

Krip the Vote

KripTheVote

Paralyzed

#NoBodyIsDisposable

References

World Health Organization. WHO Coronavirus (COVID-19) Dashboard, https://covid19.who.int/ (2022, accessed 30 July 2022).

Al-Aly

Xie

Bowe

. High-dimensional characterization of post-acute sequelae of COVID-19. Nature 2021; 594: 259–264.

Chen

Haupert

Zimmermann

, et al. Global prevalence of post COVID-19 condition or long COVID-19: a meta-analysis and systematic review. J Infect Dis 2022; 226(9): 1593–1607.

Idaho Department of Health and Welfare. Idaho expands crisis standards of care statewide due to surge in COVID-19 patients requiring hospitalization, https://healthandwelfare.idaho.gov/news/idaho-expands-crisis-standards-care-statewide-due-surge-covid-19-patients-requiring-0 (2021, accessed 30 July 2022).

Brous

. Crisis standards of care. Am J Nurs 2021; 121: 51–54.

Ashana

Anesi

Liu

, et al. Equitably allocating resources during crises: racial differences in mortality prediction models. Am J Respir Crit Care Med 2021; 204: 178–186.

Cleveland Manchanda

Sanky

Appel

. Crisis standards of care in the USA: a systematic review and implications for equity amidst COVID-19. J Racial Ethn Health Disparities 2021; 8: 824–836.

Basu

Carlson

Veenstra

. Health years in total: a new health objective function for cost-effectiveness analysis. Value Health 2020; 23: 96–103.

Office of the Assistant Secretary for Administration & Management. Section 504, Rehabilitation Act of 1973, https://www.dol.gov/agencies/oasam/centers-offices/civil-rights-center/statutes/section-504-rehabilitation-act-of-1973 (1973, accessed 30 July 2022).

10.

Alabama Disabilities Advocacy Program. Letter re: Complaint of Alabama Disabilities Advocacy Program and The Arc of the United States, https://adap.ua.edu/uploads/5/7/8/9/57892141/al-ocr-complaint_3.24.20.pdf (2020, accessed 30 July 2022).

11.

Disability Rights Oregon. Letter re: Complaint Regarding Oregon’s Crisis Care Guidance, https://static1.squarespace.com/static/5d645da3cf8e4c000158e55a/t/60417948e3b52c4f1ba22576/1614903632382/2020.05.08-Letter-to-HHS-OCR-Regarding-Crisis-Care-Guidance-in-Oregon.pdf (2020, accessed 30 July 2022).

12.

Disability Rights Pennsylvania. Letter re: Complaint of Disability Rights Pennsylvania concerning Pennsylvania’s Interim Crisis Standards of Care for Pandemic Guidelines, https://www.disabilityrightspa.org/wp-content/uploads/2020/04/DRP-9.pdf (2020, accessed 30 July 2022).

13.

Disability Rights Tennessee. Letter re: Complaint of Erin Brady Worsham, Jean Marie Lawrence, Toni Corbin, John and Pam Bryan, Jennifer Aprea, Disability Rights Tennessee, the Tennessee Disability Coalition, The Arc Tennessee, The Arc of the United States, Civil Rights Education and Enforcement Center, Disability Rights Education and Defense Fund, Autistic Self Advocacy Network, Epilepsy Foundation of Middle & West Tennessee, National Kidney Foundation, and National Multiple Sclerosis Society, https://www.disabilityrightstn.org/documents/tn_orc_complaint-re-healthcare-rationing-guideline.pdf (2020, accessed 30 July 2022).

14.

Disability Rights Washington. Letter re: Complaint of Disability Rights Washington, Self Advocates in Leadership, The Arc of the United States, and Ivanova Smith Against the Washington State Department of Health (WA DOH), the Northwest Healthcare Response Network (NHRN) and the University of Washington Medical Center (UWMC), https://www.centerforpublicrep.org/wp-content/uploads/2020/03/OCR-Complaint_3-23-20-final.pdf (2020, accessed 30 July 2022).

15.

Institute of Medicine (US) Committee on Guidance for Establishing Standards of Care for Use in Disaster Situations. Guidance for establishing crisis standards of care for use in disaster situations: a letter report. Washington, DC: National Academies Press (US), 2009.

16.

Huang

Siddiqui

Touchett

, et al. Protecting the most vulnerable among us: access to care and resources for persons with disability from spinal cord injury during the COVID-19 pandemic. PM R 2021; 13: 632–636.

17.

Gershengorn

. Inequitable resource allocation amidst a pandemic—a crisis within a crisis. JAMA Netw Open 2022; 5: e221751.

18.

Wunsch

Hill

Bosch

, et al. Comparison of 2 triage scoring guidelines for allocation of mechanical ventilators. JAMA Netw Open 2020; 3: e2029250.

19.

Riviello

Dechen

O’Donoghue

, et al. Assessment of a crisis standards of care scoring system for resource prioritization and estimated excess mortality by race, ethnicity, and socially vulnerable area during a regional surge in COVID-19. JAMA Netw Open 2022; 5: e221744.

20.

Piscitello

Kapania

Miller

, et al. Variation in ventilator allocation guidelines by US state during the coronavirus disease 2019 pandemic: a systematic review. JAMA Netw Open 2020; 3: e2012606.

21.

Biddison

ELD

Gwon

Schoch-Spana

, et al. Scarce resource allocation during disasters: a mixed-method community engagement study. Chest 2018; 153: 187–195.

22.

Landes

Turk

Formica

, et al. COVID-19 outcomes among people with intellectual and developmental disability living in residential group homes in New York State. Disabil Health J 2020; 13: 100969.

23.

Shakespeare

Ndagire

Seketi

. Triple jeopardy: disabled people and the COVID-19 pandemic. Lancet 2021; 397: 1331–1333.

24.

Kuper

Banks

Bright

, et al. Disability-inclusive COVID-19 response: what it is, why it is important and what we can learn from the United Kingdom’s response. Well Open Res 2020; 5: 79.

25.

Turk

McDermott

. The COVID-19 pandemic and people with disability. Disabil Health J 2020; 13: 100944.

26.

Korupolu

Stampas

Gibbons

, et al. COVID-19: screening and triage challenges in people with disability due to spinal cord injury. Spinal Cord Ser Cases 2020; 6: 35.

27.

Stillman

Capron

Alexander

, et al. COVID-19 and spinal cord injury and disease: results of an international survey. Spinal Cord Ser Cases 2020; 6: 21.

28.

Dicks

Clements

Gibbons

, et al. Atypical presentation of COVID-19 in persons with spinal cord injury. Spinal Cord Ser Cases 2020; 6: 38.

29.

Rodriguez-Cola

Jimenez-Velasco

Gutierrez-Henares

, et al. Clinical features of coronavirus disease 2019 (COVID-19) in a cohort of patients with disability due to spinal cord injury. Spinal Cord Ser Cases 2020; 6: 39.

30.

Matsuoka

Sumida

. The effect of the COVID-19 pandemic on the health-related quality of life in home-based patients with spinal cord injuries in Japan. J Spinal Cord Med 2021; 45(5): 1–5.

31.

Marco-Ahullo

Montesinos-Magraner

Gonzalez

, et al. Impact of COVID-19 on the self-reported physical activity of people with complete thoracic spinal cord injury full-time manual wheelchair users. J Spinal Cord Med 2021; 45(5): 755–759.

32.

Mamidi

Miller

Banerjee

, et al. Identifying key topics bearing negative sentiment on Twitter: insights concerning the 2015-2016 Zika epidemic. JMIR Public Health Surveill 2019; 5: e11036.

33.

Venuturupalli

Kumar

Bunyan

, et al. Using patient-reported health data from social media to identify diverse lupus patients and assess their symptom and medication expressions: a feasibility study. Arthritis Care Res (Hoboken) 2022. DOI: 10.1002/acr.24868. Online ahead of print.

34.

Medford

Saleh

Sumarsono

, et al. An “infodemic": leveraging high-volume Twitter data to understand early public sentiment for the coronavirus disease 2019 outbreak. Open Forum Infect Dis 2020; 7: ofaa258..

35.

Saleh

Lehmann

McDonald

, et al. Understanding public perception of coronavirus disease 2019 (COVID-19) social distancing on Twitter. Infect Control Hosp Epidemiol 2021; 42: 131–138.. Epub 2020 Aug 6.

36.

Twitter. Q1 2019 Earnings Report, https://investor.twitterinc.com/financial-information/quarterly-results/default.aspx (2019, accessed 30 July 2022).

37.

Gabarron

Dorronzoro

Rivera-Romero

, et al. Diabetes on Twitter: a sentiment analysis. J Diabetes Sci Technol 2019; 13: 439–444.

38.

Davis

Zheng

Liu

, et al. Public response to Obamacare on Twitter. J Med Internet Res 2017; 19: e167.

39.

Mollema

Harmsen

Broekhuizen

, et al. Disease detection or public opinion reflection? Content analysis of tweets, other social media, and online newspapers during the measles outbreak in the Netherlands in 2013. J Med Internet Res 2015; 17: e128.

40.

Chew

Eysenbach

. Pandemics in the age of Twitter: content analysis of Tweets during the 2009 H1N1 outbreak. Sampson M, editor. PLoS ONE 2010; 5: e14118.

41.

Chen

Safarnejad

. Effects of information veracity and message frames on information dissemination: a case study of 2016 Zika epidemic discussion on Twitter. Health Commun 2021; 36: 1560–1570.

42.

Cooper

Radunsky

Hanna

, et al. Analyzing an emerging pandemic on Twitter: monkeypox. Open Forum Infect Dis 2023; 10: ofad142..

43.

Twitter. Twitter API | products | twitter developer platform, https://developer.twitter.com/en/products/twitter-api (2022, accessed 30 July 2022).

44.

Van Rossum

Drake

FL.

Python 3 Reference Manual, https://dl.acm.org/doi/book/10.5555/1593511 (2009, accessed 30 July 2022).

45.

Wang

Hale

Adelani

, et al. Demographic inference and representative population estimates from multilingual social media data. In: Proceedings of the 2019 World Wide Web Conference, 13–17 May 2019; San Francisco, CA, USA.

46.

Laohaprapanon

Sood

Naji

Appeler/ETHNICOLR: predict race and ethnicity based on the sequence of characters in a name, https://github.com/appeler/ethnicolr (2021, accessed 30 July 2022).

47.

Thelwall

Buckley

Paltoglou

. Sentiment strength detection for the social web. J Am Soc Inf Sci Technol 2012; 63: 163–173.

48.

Honnibal

Montani

spaCy 2: natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing, https://sentometrics-research.com/publication/72/ (2017, accessed 30 July 2022).

49.

Bailey

. NRCLEX. PyPI, https://pypi.org/project/NRCLex/ (2019, accessed 30 July 2022).

50.

Wang

Shi

Zhang

. What public health campaigns can learn from people’s Twitter reactions on mask-wearing and COVID-19 vaccines: a topic modeling approach. Cogent Social Sciences 2021; 7: 1959728.

51.

Rehurek

Sojka

. Parallelized latent Dirichlet allocation. genism, https://radimrehurek.com/gensim/models/ldamulticore.html (accessed 30 July 2022).

52.

Blei

Jordan

. Latent dirichlet allocation. J Mach Learn Res 2003; 3: 993–1022.

53.

Lanier

Diaz

Saleh

, et al. Analyzing COVID-19 disinformation on Twitter using the hashtags #scamdemic and #plandemic: retrospective study. PLoS One 2022; 17: e0268409.

54.

Choi

Han

Lee

, et al. Risk of COVID-19 and major adverse clinical outcomes among people with disabilities in South Korea. Disabil Health J 2021; 14: 101127.

55.

Harvey

Hermez

Schanz

, et al. Healthcare disparities correlated with in-hospital mortality in COVID-19 patients. Int J Gen Med 2021; 14: 5593–5596.

56.

Zandam

Mitra

Akobirshoev

, et al. Infectious-diseases-related emergency department visits among non-elderly adults with intellectual and developmental disabilities in the United States: results from the National Emergency Department Sample, 2016. Popul Health Manag 2021. DOI: 10.1089/pop.2021.0218. Online ahead of print.

57.

De Cauwer

Spaepen

. Are patients with down syndrome vulnerable to life-threatening COVID-19? Acta Neurol Belg 2021; 121: 685–687.

58.

Baksh

Pape

Smith

, et al. Understanding inequalities in COVID-19 outcomes following hospital admission for people with intellectual disability compared to the general population: a matched cohort study in the UK. BMJ Open 2021; 11: e052482.

59.

Heslop

Byrne

Calkin

, et al. Deaths of people with intellectual disabilities: analysis of deaths in England from COVID-19 and other causes. J Appl Res Intellect Disabil 2021; 34: 1630–1640.

60.

Koyama

Koumans

Sircar

, et al. Severe outcomes, readmission, and length of stay among COVID-19 patients with intellectual and developmental disabilities. Int J Infect Dis 2022; 116: 328–330.

61.

Turk

Landes

Formica

, et al. Intellectual and developmental disability and COVID-19 case-fatality trends: triNetX analysis. Disabil Health J 2020; 13: 100942.

62.

Czeisler

Board

Thierry

JAM

, et al. Mental health and substance use among adults with disabilities during the COVID-19 pandemic—United States, February-March 2021. MMWR Morb Mortal Wkly Rep 2021; 70: 1142–1149.

63.

Gustafson

Stillman

Capron

, et al. COVID-19 and spinal cord injury and disease: results of in international survey as the pandemic progresses. Spinal Cord Ser Cases 2021; 7: 13.

64.

Monden

Andrews

Pilarski

, et al. COVID-19 and the spinal cord injury community: concerns about medical rationing and social isolation. Rehabil Psychol 2021; 66: 373–379.

65.

Chemerynska

Marczak

Kucharska

. What are the experiences of clinical psychologists working with people with intellectual disabilities during the COVID-19 pandemic? J Appl Res Intellect Disabil 2022; 35: 587–595.

66.

Kavanagh

Hatton

Stancliffe

, et al. Health and healthcare for people with disabilities in the UK during the COVID-19 pandemic. Disabil Health J 2022; 15: 101171.

67.

Okoro

Strine

McKnight-Eily

, et al. Indicators of poor mental health and stressors during the COVID-19 pandemic, by disability status: a cross-sectional analysis. Disabil Health J 2021; 14: 101110.

68.

Hearn

Rohn

Monden

. Isolated and anxious: a qualitative exploration of the impact of the COVID-19 pandemic on individuals living with spinal cord injury in the UK. J Spinal Cord Med 2022; 45: 691–699.

69.

Wickenden

Hart

TGB

Thompson

, et al. How did South Africans with disabilities experience COVID-19? Results of an online survey. Afr J Disabil 2023; 12: 1119.

70.

Iezzoni

Rao

Ressalam

, et al. Physicians’ perceptions of people with disability and their health care. Health Aff (Millwood) 2021; 40: 297–306.

71.

Lagu

Haywood

Reimold

, et al. ‘I am not the doctor for you’: physicians’ attitudes about caring for people with disabilities. Health Aff (Millwood) 2022; 41: 1387–1395.

72.

Goggin

Ellis

. Disability, communication, and life itself in the COVID-19 pandemic. Health Sociol Rev 2020; 29: 168–176.

73.

Guidry-Grimes

Savin

Stramondo

, et al. Disability rights as a necessary framework for crisis standards of care and the future of health care. Hastings Cent Rep 2020; 50: 28–32.

74.

Engelman

Craig

Iles

. Global disability justice in climate disasters: mobilizing people with disabilities as change agents. Health Aff (Millwood) 2022; 41: 1496–1504.

75.

Sabatello

Landes

McDonald

. People with disabilities in COVID-19: fixing our priorities. Am J Bioeth 2020; 20: 187–190.

76.

Valdez

Lyon

Wellbeloved-Stone

, et al. Engaging the disability community in informatics research: rationales and practical steps. J Am Med Inform Assoc 2022; 29: 1989–1995.

77.

Andrews

Ayers

Brown

, et al. No body is expendable: medical rationing and disability justice during the COVID-19 pandemic. Am Psychol 2021; 76: 451–461.

78.

Mukherjee

Tarsney

Kirschner

. If not now, then when? Taking disability seriously in bioethics. Hastings Cent Rep 2022; 52: 37–48.

79.

Statista. Market share of mobile operating systems in North America from January 2018 to January 2023, https://www.statista.com/statistics/1045192/share-of-mobile-operating-systems-in-north-america-by-month/#:∼:text=Share%20of%20mobile%20operating%20systems,America%202018%2D2023%2C%20by%20month&text=Google’s%20Android%20and%20Apple’s%20iOS,accounted%20for%20around%2055%20percent (2023, accessed 6 March 2023).

80.

Web Usability. Android vs iOS: how do smartphone platforms compare on accessibility?, https://info.webusability.co.uk/blog/android-vs-ios-how-do-smartphone-platforms-compare-on-accessibility (2021, accessed 6 March 2023).

The lived experience of people with disabilities during the COVID-19 pandemic on Twitter: Content analysis

Abstract

Objective

Methods

Results

Conclusions

Keywords

Introduction

Methods

Data collection and processing

Descriptive analysis

Sentiment and emotion analysis

Topic modeling

Results

Descriptive analysis

Emotion analysis

Sentiment analysis

Topic models

Discussion

Limitations

Conclusions

Footnotes

Contributorship

Declaration of conflicting interests

Ethical approval

Funding

Guarantor

ORCID iD

Trial registration

Appendix

References