| OCR Text |
Show Original Contribution Section Editors: Clare Fraser, MD Susan Mollan, MD Inappropriate Indexing of Case Reports to the “Papilledema” Subject Heading: A Systematic Review Brendan K. Tao, MD(c), BHSc, Amir R. Vosoughi, MD, MSc(c), Bhadra U. Pandya, MD(c), Junaid Ishaq, MD(c), HBSc, Edward A. Margolin, MD, Jonathan A. Micieli, MD, CM, FRCSC Background: Papilledema must be managed distinctly from other causes of optic disc edema (ODE) due to its basis in raised intracranial pressure (ICP). However, evidence indicates that the term “papilledema” is widely misused across specialties to describe ODE without raised ICP. Sources of this misconception remain undiscerned. Because all physicians consult medical databases, our objective was to evaluate whether nonspecific “papilledema” subject heading definitions misleadingly associate articles on other conditions with papilledema proper. Methods: Systematic review of case reports, prospectively registered on PROSPERO (CRD42022363651). MEDLINE and Embase were searched to July 2022 for any full-length case report indexed to the “papilledema” subject heading. Studies were graded for incorrect indexing, defined as cases lacking evidence for raised ICP. Nonpapilledema diagnoses were assigned to a predefined set of diseases and pathophysiological mechanisms for subsequent comparison. Results: Incorrect indexing occurred in 40.67% of 949 included reports. Embase-derived studies were misindexed significantly less than MEDLINE-derived studies (P , 0.01). There was also significant heterogeneity in incorrect indexing among specific diseases (P = 0.0015) and mechanisms (P = 0.0003). The most commonly misindexed diseases were uveitis (21.24% of errors), optic neuritis (13.47%), and instances with no mention of ODE (13.99%). The most commonly misindexed mechanisms were inflammation Faculty of Medicine (BKT), The University of British Columbia, Vancouver, Canada; Max Rady College of Medicine (ARV), Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, Canada; Temerty Faculty of Medicine (BUP), University of Toronto, Toronto, Canada; Michael G. DeGroote School of Medicine (JI, EAM, JAM), Faculty of Health Sciences, McMaster University, Hamilton, Canada; Department of Ophthalmology and Vision Sciences (EAM, JAM), Toronto Western Hospital, University Health Network, University of Toronto, Toronto, Canada; Division of Neurology (EAM, JAM), Department of Medicine, University of Toronto, Toronto, Canada; Kensington Vision and Research Center (EAM, JAM), Toronto, Canada; St. Michael’s Hospital (EAM, JAM), Toronto, Canada; Toronto Western Hospital (EAM, JAM), Toronto, Canada; and University Health Network (EAM, JAM), Toronto, Canada. The authors report no conflicts of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www. jneuro-ophthalmology.com). B. K. Tao and A. R. Vosoughi contributed equally as co-first authors. Address correspondence to Jonathan A. Micieli, MD, CM, FRCSC, Kensington Vision & Research Centre, 340 College Street, Toronto, ON M5T 3A9, Canada; E-mail: jonathanmicieli@gmail.com Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 (34.97%), other mechanism (e.g., genetic; 25.91%), and ischemia (20.47%). Conclusions: Database subject headings, especially from MEDLINE, do not adequately distinguish between true papilledema and other causes of ODE. Inflammatory diseases were most often incorrectly indexed among other diseases and mechanisms. Current “papilledema” subject headings should be revised to reduce the probability of misinformation. Journal of Neuro-Ophthalmology 2023;43:525–530 doi: 10.1097/WNO.0000000000001895 © 2023 by North American Neuro-Ophthalmology Society P apilledema is defined as optic disc edema (ODE) due to raised intracranial pressure (ICP).1 Previous work indicates that the term “papilledema” is commonly misused across medical specialties during funduscopic examination or to describe nonintracranial pressure–related ODE.2 This is problematic as papilledema is among the top reasons for neuro-ophthalmology referral.3 Indeed, a miscommunicated referral because of an incorrect understanding of the term “papilledema” may disrupt the transfer of care and increase the risk for patient harm.2 Therefore, in all specialty settings, papilledema must be discriminated from other causes of optic disc edema because it implies a unique pathophysiology and management strategy. There remains no work to-date that investigates the underlying causes of papilledema term misuse, especially causes that may explain the widespread nature of this error across specialties including ophthalmology.2 MEDLINE and Embase are highly consulted medical databases that use subject headings to index relevant journal articles. Although article indexing has historically proceeded through manual sorting, automatic indexing has been in use since 2011 and 2020 for MEDLINE and Embase, respectively.4 In either case, both manual and automated systems rely on the database-dependent definition of the “papilledema” subject heading.4 An inappropriate definition may therefore result in inappropriate indexing of studies to a particular subject heading. Thorough details on the MEDLINE indexing system are described elsewhere.5 As of November 2022, MEDLINE and Embase both use a “papilledema” subject heading. However, both databases commit the same terminology error with their “papilledema” subject heading definitions because they include studies on ODE that are not exclusively due to raised ICP. 525 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution In MEDLINE, the “papilledema” medical subject heading (MeSH) is defined as, “swelling of the optic disk, usually in association with increasing intracranial pressure,” which incorrectly implies that papilledema is not always accompanied by increased ICP.6 Furthermore, MEDLINE incorrectly indicates that the “papilledema” subject heading be used to index studies that report on cases of retinal edema, optic nerve papillitis, and cases related to decreased intraocular pressure, none of which fundamentally imply raised intracranial pressure.6 For example, a study by Liou et al7 is indexed under this MEDLINE subject heading, where they report on 2 patients with unilateral orbital inflammation and ODE, secondary to myelin oligodendrocyte glycoprotein–associated disease.7 This study reports no evidence of raised ICP and should not be indexed under the “papilledema” subject heading. In Embase, the “papilledema” Emtree subject heading is defined as “edema of the optic disk (papilla), most commonly due to increased intracranial pressure,” which likewise erroneously implies that papilledema is not always secondary to increased ICP.8 However, Embase has superior indications for applying the “papilledema” subject heading compared with MEDLINE given that it has no incorrect indications for its use, such as cases of retinal edema, optic nerve papillitis, and cases related to decreased intraocular pressure.8 For this reason, we hypothesized that Embasederived articles are less often incorrectly indexed. Nonetheless, articles continue to be incorrectly indexed to the Embase “papilledema” subject heading. For example, a study by Burova et al9 is indexed under this Embase subject heading, where they report on a patient with choroidal neovascularization after resolution of multiple evanescent white dot syndrome. This study does not report evidence of raised ICP and likewise should not be indexed to the “papilledema” subject heading. This study’s rationale was to investigate the erroneous application of the “papilledema” subject heading across case reports indexed in Embase and MEDLINE. We hypothesize that such indexing errors are a source of misconception that synonymize “papilledema” with other causes of ODE. The primary objective of this study was to determine the prevalence of inappropriately indexed case reports to the “papilledema” subject heading. Secondary objectives included analyses on whether the extent of inappropriate indexing was related to the origin database (Embase or MEDLINE), the cause of ODE depicted in each case report, or the underlying mechanisms of these causes of ODE. METHODS This systematic review was prospectively registered in PROSPERO (CRD42022363651).10 Under article 2.4 of the Tri-Council Policy Statement, relevant IRB/Ethics Committees waived need for approval because the data were 526 collected from published primary research. This research adheres to the tenets of the Declaration of Helsinki. Where appropriate, this study abides by reporting PRISMA guidelines. Eligibility Criteria for Considering Studies for This Review From Embase and MEDLINE, all English-language case reports indexed under the “papilledema” subject heading were included. We excluded non–full-length case reports (e.g., short photographic cases). Case series were included if they described individual cases in sufficient detail per patient. Search Methods for Identifying Studies Ovid MEDLINE (from 2011) and Embase (from 2022) were searched by BT to July 1, 2022. Using the search criteria (See Supplemental Digital Content, Appendix, http://links.lww.com/WNO/A697), we collected any English-written case report indexed under the “papilledema” subject heading. The search years were adjusted to target an a priori sample size of approximately 1,000 initial case reports. As in our previous report, this target sample was chosen to yield increased study power, while balancing against practical considerations for manually evaluating studies.2 As one study objective was to compare the indexing patterns between Embase and MEDLINE, it was preferable that both databases contribute a near-equal number of reports to the total. Exploratory searches using the documented search strategy revealed a lower limit search year of 2011 for MEDLINE and 2020 for Embase, which yielded just more than 500 titles from each database and collectively totaled just more than 1,000 articles. Data Collection and Risk of Bias Assessment In Covidence (Melbourne, Australia), paired reviewers (A.V., B.T., J.I., and B.P.) conducted study title and abstract screening and subsequent full-text evaluation for the final inclusion. Exclusion criteria included non–full-text studies (e.g., conference abstracts and photographic cases), noncase report studies, and non–English-language articles. Data extraction proceeded in duplicate (J.I., B.P., B.T., and A.V.) to ascertain the final case diagnosis, whether it be papilledema proper or an inappropriately indexed condition. A set of final case diagnoses was determined a priori and included papilledema, optic disc drusen, anomalousappearing optic nerves, myopic tilted nerves, nonarteritic anterior ischemic optic neuropathy (NAION), arteritic anterior ischemic optic neuropathy (AAION), central retinal artery occlusion (CRAO), branch retinal artery occlusion (BRAO), associated with central retinal vein occlusion (CRVO), optic neuritis, uveitis, other optic neuropathies and/or diagnoses (e.g., related to toxins or neoplasia), or no mention at all of any optic disc edema. These conditions were aggregated into their respective pathophysiological Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution mechanisms, including i) papilledema; ii) pseudopapilledema; iii) ischemic; iv) inflammatory or v) infiltrative and/or neoplastic; and vi) other mechanism. Conflicts at all stages of the review were adjudicated by the senior review author (J.M.), an experienced neuro-ophthalmologist. Finally, because case report indexing is entirely an independent process of the database sorting algorithm, case report risk of bias was not assessed as it was beyond the scope of the study research question. Data Synthesis and Analysis For the primary outcome, the crude prevalence of incorrect study indexing was calculated across all studies. Secondarily, we calculate this prevalence separately according to database of origin. Chi-square analysis and phi correlation analyzed the relationship between database of origin and the prevalence of misindexing. Studies that were indexed by both databases were accounted for in both database totals. As well, we calculate the prevalence of study misindexing according to each predefined condition and disease mechanism. Subsequent Kruskal–Wallis analysis with post hoc pairwise testing (using false detection rate correction) evaluated whether various diagnoses or disease mechanisms were more likely than others to be incorrectly indexed. Owing to the nature of this study, there were no missing data. Analysis was completed by BT in RStudio 2022.02.0+443. RESULTS We identified 949 unique records for final inclusion, composed of 471 studies uniquely indexed to Embase, 403 reports from MEDLINE, and 75 studies indexed in both databases. Figure 1 shows the study identification process.11 The inter-rater reliability was strong for distinguishing reports with papilledema proper vs untrue papilledema (kappa = 0.81%, 95% confidence interval [0.78–0.85], corresponding to “almost perfect agreement”, 91.7% crude agreement) and for classifying the type of nonpapilledema diagnosis (81.7% crude agreement).12 Conflicts were predominately cases that were near consensus but with an extraneous diagnosis that was overlooked by one reviewer. The multiplicity of various nonpapilledema diagnoses precluded kappa calculation. Supplemental Digital Content (see Table S1, http://links.lww.com/WNO/A698 tabulates all included case reports from this review). For the primary outcome, 386 of 949 studies (40.67%) were incorrectly indexed under the “papilledema” subject heading. For secondary analysis, misindexing occurred in 215 of 478 MEDLINE-originated reports (44.98%) and 199 of 546 Embase-derived studies (36.45%). Embase committed significantly fewer errors than MEDLINE (x2 = 7.35; P ,0.01; u = 0.0867). Among all included studies, we cataloged 65 unique diagnoses which are summarized in Supplemental Digital Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 Content (see Table 1, http://links.lww.com/WNO/A698). Table 1 summarized the prevalence of case reports that described diagnoses that were defined a priori, but without accompanying papilledema. There was a significant relationship between certain predefined diagnoses and higher rates of misindexing (P = 0.0015). Subsequent pairwise testing revealed significant differences between uveitis and BRAO (P = 0.04), uveitis and optic disc drusen (P = 0.006), optic neuritis and optic disc drusen (P = 0.041), NAION and optic disc drusen (P = 0.041), and no mention of ODE with optic disc drusen (P = 0.019), and the former of each pair was more often indexing incorrectly. The distribution of disease mechanisms of all included studies is summarized in Supplemental Digital Content (see Table 1, http://links.lww.com/WNO/A698). Table 2 summarizes the prevalence of case reports that described underlying disease mechanisms that were defined a priori, but without accompanying papilledema. There was a significant relationship between certain predefined disease mechanisms and higher rates of misindexing (P = 0.0003). Subsequent pairwise testing revealed significant differences between mechanisms of ischemia and infiltration/neoplasia (P = 0.004), inflammation and infiltration/neoplasia (P = 0.015), no mechanism (no evidence of ODE) and infiltration/neoplasia (P = 0.015), pseudopapilledema and inflammation neoplasia (P = 0.008), ischemia and other mechanism (e.g., genetic) (P = 0.031), ischemia and pseudopapilledema (P = 0.004), and ischemia and pseudopapilledema (P = 0.004), no mechanism (no evidence of ODE) and pseudopapilledema (P = 0.015), and the former of each pair was more often incorrectly indexed. DISCUSSION A significant number of studies were misindexed to the “papilledema” subject heading, especially from MEDLINE. There was also a significant heterogeneity in the rates of incorrect indexing among prespecified diseases and mechanisms, with inflammatory conditions and mechanisms constituting most misindexed cases. Although a seemingly innocuous phenomenon, this error may contribute to broad misinformation on the term “papilledema” and its implication of raised ICP. Indeed, for all clinicians, medical databases are the mainstay source of continuing medical education. Disorganization of these databases, such as through misleading subject headings, may subtly conflate studies on papilledema proper with other causes of ODE. As centers for knowledge dissemination, databases must use accurate definitions if they are to be resources for continuing medical education. Medical databases are used as research tools for conducting systematic reviews. Incorrect study indexing may therefore impede or even mislead such research endeavors. Fundamentally, this indexing error results in the mixing of 527 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution FIG. 1. Preferred reporting items for systematic reviews flow diagram. studies on ODE from various etiologies, each of which implicate different diagnostic workups and treatment regimens. For instance, this error may increase the risk of articles on papilledema proper being considered in reviews of other optic disc edemas (such as on optic neuritis) and vice versa. Although systematic reviews permit manual exclusion of irrelevant reports, the ongoing and proven conflation of “papilledema” with other ODE causes across clinicians and disciplines increases the risk of failing to make this distinction and subsequently including inappropriate studies into article synthesis.2 Hence, the results of such systematic reviews may be at a higher risk of bias, and as the highest form of evidence, may mislead care delivery in clinical practice. The broader impact of erroneous “papilledema” term usage has been detailed elsewhere.2 Our results support our hypothesis of Embase committing significantly fewer indexing errors than MED528 LINE. Although both databases had incorrect papilledema subject headings, we noted that MEDLINE’s definition had a greater degree of error, as it stated the specific cases of retinal edema, optic nerve papillitis, and cases related to decreased intraocular pressure as prototype conditions that would fall under the papilledema subject heading. However, none of these supposedly prototypical diagnoses necessarily imply raised intracranial pressure.6 By comparison, Embase offers no such prototypical examples. Presumably, the more specific yet misleading subject heading definition of MEDLINE may have contributed to its significantly higher rate of study misindexing. This result concords with previous evidence that implies a greater degree of robustness with Embase over MEDLINE. For instance, one study reported that Embase demonstrated superior improvement in its indexing accuracy with implementation of the Standards for Reporting of Diagnostic Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 1. Represented diagnoses from incorrectly indexed studies Diagnosis Optic neuritis Nonarteritic anterior ischemic optic neuropathy Optic disc drusen Associated with central retinal vein occlusion Uveitis No mention of any optic disc edema Branch retinal artery occlusion Arteritic anterior ischemic optic neuropathy Central retinal artery occlusion n (Prevalence Among All Reports; N = 949) n (Prevalence Among Indexing Errored Reports; N = 386) 52 (5.8%) 32 (3.48%) 14 (2.11%) 23 (2.42%) 82 (8.64%) 54 (5.69%) 5 (0.53%) 6 (0.63%) 18 (1.9%) 13.47% 8.29% 3.63% 5.96% 21.24% 13.99% 1.30% 1.55% 4.66% This table depicts the prevalence of case reports that described diagnoses that were defined a priori, but without accompanying papilledema. Prevalence is calculated among all included reports and all reports that were incorrectly indexed. For the former, the remaining prevalence is accounted for by diagnoses that were not predefined. Accuracy.13 An Elsevier Whitepaper comparing its Emtree system with the MeSH system of MEDLINE also elucidates potential reasons for better indexing in Embase. Relevant reasons include using more preferred and up-to-date terminology than MeSH headings; its adoption of more than 300,000 synonyms and 71,000 preferred terms (compared with 220,000 synonyms and 27,000 preferred terms in MEDLINE), which may improve the likelihood of hosting subject headings for conditions that cause ODE but are unrelated to raised ICP; its reliance on authors’ use of terms to define their headings (as opposed to rigid definitions set in MEDLINE), which may reduce errored subject heading definitions as at least some proportion of reports will use terms correctly in the literature; and triyearly terminology updates (as compared with annual updates in MEDLINE).14 However, despite these advantages, neither database was effective at distinguishing studies with papilledema proper vs other causes of ODE. Their further optimization is recommended. Our results imply that most indexing errors corresponded to studies on diseases or mechanisms involving inflammation. Indeed, the most commonly misindexed diseases were uveitis (21.24% of errors), optic neuritis (13.47%), and instances with no mention of ODE (13.99%). As well, the most commonly misindexed mechanism was inflammation (34.97%). Although this study is unable to directly investigate the indexing decision-making process, whether manual or automatic due to a lack of hands-on access to the sorting algorithm, our results may reflect portions of the MEDLINE definition for the “papilledema” subject heading. For instance, MEDLINE reports conditions such as retinal edema and optic nerve papillitis as prototypical examples of conditions that belong to its “papilledema” subject heading.6 However, although neither condition necessarily implicates raised ICP, both have underlying bases in inflammation, which corresponds to our findings of inflammatory-related conditions as the leading examples of misindexed studies. Our disease- and mechanism-specific results may inform ways to reiterate and improve on current “papilledema” subject headings. Our results are limited to the repertoire of case reports and an inability to analyze all possible diagnoses and mechanisms outside those prespecified (e.g., ocular hypotony and genetics). As well, whether reports were indexed to TABLE 2. Represented disease mechanisms from incorrectly indexed studies Mechanism n (Prevalence Among All Reports; N = 949) Raised intracranial pressure Pseudopapilledema Ischemia Inflammatory Infiltrative/neoplastic Other None 563 (59.33%) 20 (2.11%) 79 (8.32%) 135 (14.23%) 4 (0.42%) 100 (10.54%) 54 (5.69%) n (Prevalence Among Indexing Errored Reports; N = 386) N/A 5.18% 20.47% 34.97% 1.04% 25.91% 13.99% This table depicts the prevalence of case reports that described disease mechanisms that were defined a priori, but without accompanying evidence of raised intracranial pressure. Prevalence is calculated among all included reports and all reports that were incorrectly indexed. For the former, the remaining prevalence is accounted for by diagnoses that were not predefined. Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 529 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution other relevant subject headings to ODE was not conducted. Although comprehensive, we believed that such an analysis would detract from the central focus on inappropriate papilledema-related misindexing. Embase and MEDLINE do not adequately differentiate papilledema from other causes of ODE, most often conflating true papilledema with inflammatory conditions. For this reason, clinicians should continue to critically appraise published cases of ODE because distinction of papilledema is required for proper management. We recommend more specific subject heading definitions to avoid conflating distinct causes of ODE. Future work should investigate other contributors to the “papilledema” term misunderstanding such as gaps in medical education. REFERENCES 1. Rigi M, Almarzouqi S, Morgan M, Lee A. Papilledema: epidemiology, etiology, and clinical management. Eye Brain. 2015;7:47–52. 2. Tao B, Vosoughi A, Margolin E, Micieli JA. Inappropriate use of the term “papilledema” in the medical literature: a systematic review of case reports across specialties. Ophthalmology. 2023;130:129–136. 3. Stunkel L, Sharma RA, Mackay DD, et al. Patient harm due to diagnostic error of neuro-ophthalmologic conditions. Ophthalmology. 2021;128:1356–1362. 4. Frequently Asked Questions About Indexing for MEDLINE [Internet]. Bethesda, MD: National Library of Medicine, 2022. Available at: https://www.nlm.nih.gov/bsd/indexfaq.html. Accessed November 22, 2022. 530 5. Mork JG, Jimeno-Yepes A, Aronson A. The NLM Medical Text Indexer System for Indexing Biomedical Literature. Bethesda, MD: National Library of Medicine, 2013. 6. Papilledema [Internet]. Bethesda, MD: National Library of Medicine; 1999. Available at: https://meshb.nlm.nih.gov/ record/ui?ui=D010211. Accessed November 22, 2022. 7. Liou VD, Yoon MK, Maher M, Chwalisz BK. Orbital inflammation in myelin oligodendrocyte glycoprotein antibody–associated disease: a case report and review of the literature. J Neuroophthalmol. 2022;42:e56–e62. 8. Scope Note for: Papilledema [Internet]. Amsterdam, NL: Elsevier; 1974. 9. Burova M, Stepanov A, Almesmary B, Jiraskova N. Choroidal neovascularization in a patient after resolution of multiple evanescent white dot syndrome: a case report. Clin Case Rep. 2022;10:e05802. 10. Tao B, Vosoughi A, Micieli J. Inappropriate Indexing of Case Reports to the “Papilledema” Subject Heading: A Systematic Review Across Specialties [Internet]. York: Centre for Reviews and Dissemination, 2022. Available at: https://www.crd.york. ac.uk/prospero/display_record.php?ID=CRD42022363651. Accessed November 14, 2022. 11. Haddaway NR, Page MJ, Pritchard CC, McGuinness LA. PRISMA2020: an R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and Open Synthesis. Campbell Syst Rev. 2022;18:e1230. 12. McHugh M. Interrater reliability: the kappa statistic. Biochem Med (Zagreb). 2012;22:276–282. 13. Wilczynski N, Haynes R. Indexing of diagnosis accuracy studies in MEDLINE and EMBASE. AMIA Annu Symp Proc. 2007;2007:801–805. 14. Whitepaper A. Comparison of Emtree and MeSH. Amsterdam, NL: Elsevier R&D Solutions. Pharma & Life Sciences, 2015. Tao et al: J Neuro-Ophthalmol 2023; 43: 525-530 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. |
