Written descriptions versus photographs in dermatological case reports
Bilge Fettahlioglu Karaman
Department of Dermatology, Faculty of Medicine, Çukurova University, Adana, Turkey
Correspondence Address:
Dr. Bilge Fettahlioglu Karaman
Department of Dermatology, Faculty of Medicine, AÇukurova University, Adana 01330
Turkey
Source of Support: None, Conflict of Interest: None
DOI: 10.4103/ds.ds_24_22
Background: Images are essential for dermatology. However, if a photograph has poor quality, it causes problems in conveying information. Objectives: This study aimed to evaluate the efficiency of photographs in a part of the current literature. Methods: One hundred and eighteen written descriptions and their related photographs were extracted from 89 articles selected from the journal Case Reports in Dermatological Medicine. Frequencies and confidence intervals (CIs) of mentioning 11 features, namely location, side, amount, distribution, dimension, figure, edge, color, touch perception, subjective symptom, and elementary lesion, in the written descriptions were calculated. The photographs were also evaluated to detect features other than touch perception and subjective symptom. Fifty articles randomly selected from five high-impact journals were also reviewed. Results: Touch perception and subjective symptom were mentioned only in 35.6% and 46.6% of the written descriptions, respectively. Distribution and edge were distinguishable in the photographs only in half of the cases, however other features in more than 90%. According to multinomial CIs, the written descriptions were superior only in terms of location, whereas the photographs were superior in terms of side, distribution, dimension, figure, edge, and color. Some features, such as touch perception and subjective symptom, were mentioned less frequently in the reviewed articles of the high-impact journals. Conclusion: The contribution of the photographs to the written descriptions in case reports was clear. However, the low rate for determining sharpness of edges in the photographs was a sign indicating low-quality images. In conclusion, the present study and similar studies will guide more efficient usage of developing image modalities for online publication in dermatology.
Keywords: Case reports, confidence intervals, dermatological photography, online publication, visual aid
Visuality has a priority in dermatology. In other words, visuality has a great importance in dermatological clinical practice. The following cases emphasize this importance. Visually oriented medical students are advised to prefer a career in dermatology.[1] Replaceability of a host of descriptive sentences by good pictures in clinical dermatology has been underlined with a famous English adage: a picture is worth more than a thousand words.[2] Moreover, interest in new imaging modalities has been grown recently. These modalities are confocal microscopy, optical coherence tomography, high-frequency ultrasound, multispectral optoacoustic tomography, Raman spectroscopy, ultraviolet photography, and fluorescence imaging in addition to two-dimensional or three-dimensional total-body digital photography.[3],[4]
On the other hand, the importance of images in dermatological clinical practice has also been exaggerated. For example, a pioneer study in teledermatology was reported with a striking title in 1997: “The substitution of digital images for dermatologic physical examination.”[5] However, there are also opposite facts to the above-mentioned adage. For example, if an image's quality is not reasonable, more than 1000 words may be required to describe the image.[6]
In this study, to make more clear the place of clinical images in dermatology, I evaluated the contribution of photographs to texts of case reports for descriptions of skin lesions.
Materials and MethodsTo evaluate the contribution of photographs in describing skin lesions, all articles of the journal Case Reports in Dermatological Medicine published between 2016 and 2019 were reviewed. There were 117 articles in total. However, case reports without suitable photographs, case reports focused on the skin appendages or extracutaneous organs, case series without individual descriptions, studies, and corrigendum were excluded from the study. Hence, 89 case reports remained. They reported 101 patients and had 118 descriptions suitable for this study.
It was evaluated whether or not eleven features were mentioned in the written descriptions. These features were location, side, amount, distribution, dimension, figure, edge, color, touch perception, subjective symptom, and elementary lesion. Furthermore, photographs were examined for the nine features other than touch perception and subjective symptom.
There were four categories for mentioning a feature in the written descriptions and detecting it in the photographs: (1) neither mentioned in the written descriptions nor detected in the photographs; (2) mentioned in the written descriptions, but not detected in the photographs; (3) not mentioned in the written descriptions, but detected in the photographs; and (4) both mentioned in the written descriptions and detected in the photographs. These categories were labeled with “nonverbal nonvisual,” “verbal nonvisual,” “nonverbal visual,” and “verbal visual,” respectively.
The “verbal visual” category was further divided into five categories: (1) the written descriptions and the photographs were equal in information; (2) the written descriptions had both common and different information, whereas the photographs had only common information; (3) the written descriptions had only common information, whereas the photographs had both common and different information; (4) both the written descriptions and the photographs had both common and different information; and (5) the written descriptions and the photographs had no common information. In other words, they were completely different, even conflicting, in terms of information offered. These categories were labeled with “equal,” “verbal superior,” “visual superior,” “complementary,” and “conflicting,” respectively. Hence, the total number of the categories increased to eight. Examples of them for the feature “elementary lesion” are given in [Table 1].
Table 1: Examples of eight categories for the feature “elementary lesion”Mentioning a feature in the written descriptions and detecting a feature in the photographs were defined as “success.” R[7] and its library “DescTools”[8] were used to calculate 95% exact binomial confidence intervals (CIs) according to the Clopper–Pearson method and multinomial CIs according to the Sison–Glaz method for the success rates. If these intervals did not intersect for a given feature, the success rate was accepted to be significantly different between the written descriptions and the photographs. In the evaluation of multinomial CIs, categories “nonverbal nonvisual,” “equal,” “complementary,” and “conflicting” were not taken into consideration, since they were not in favor of neither betterness of the written descriptions nor that of the photographs. Among the rest categories, “verbal nonvisual” and “verbal superior” were accepted in favor of betterness of the written descriptions, only when at least one category's CI was above those of both opposite categories, namely “nonverbal visual” and “visual superior.” Similarly, “nonverbal visual” and “visual superior” were accepted in favor of betterness of the photographs.
The website of Scimago Journal and Country Rank was reviewed for dermatological journals with high-impact factors. Among these journals, the following five were chosen, since the library of the university where I work has access to their full texts: The British Journal of Dermatology, the Journal of the American Academy of Dermatology, the Journal of Investigative Dermatology, the Journal of the European Academy of Dermatology and Venereology, and the Journal of Dermatological Science. For each of these five journals, the list of case report articles published between 2016 and 2019 was extracted from PubMed. Each list was randomly ordered in R.[7] Beginning from the top of each randomly ordered list, articles were reviewed, and some of them were excluded according to the above criteria. Eventually, ten case reports were selected from each journal. A total of 50 case reports published in journals with high-impact factors were evaluated in the same manner, in order to verify the findings obtained from the evaluation of the Case Reports in Dermatological Medicine articles. They reported 54 patients and had 73 descriptions suitable for this study.
ResultsEvaluation of articles published in the journal Case Reports in Dermatological Medicine
The success rates and their CIs of the written descriptions and the photographs are given in [Table 2]. The most commonly mentioned features were location, amount, elementary lesion, and color in the written descriptions. The two features which could not be detected in the photographs, namely touch perception and subjective symptom, were mentioned in 35.6% and 46.6% of the written descriptions, respectively. While 34 (61.8%) of the 55 written descriptions mentioning subjective symptom stated one or more symptoms, the rest 21 descriptions (38.2%) did note that there was no symptom. Seven features other than distribution and edge were detected in the photographs with a rate of over 90%. These two exceptions were detected in half of the cases. Among the nine features for which a comparison was possible between the written descriptions and the photographs, only one feature, namely location, was more common in the written descriptions, and others, in the photographs.
Table 2: The success rates and their confidence intervals of mentioning features in the written descriptions and detecting them in the photographs for 118 descriptions extracted from 89 case reports published in the journal Case Reports in Dermatological MedicineThe CIs for location, amount, and elementary lesion intersected. However, those for side, distribution, dimension, figure, edge, and color did not intersect. The success rates for these six features were higher in the photographs [Figure 1].
Figure 1: The CIs of the success rates (%) of mentioning features in the written descriptions (white rectangles) and detecting them in the photographs (black rectangles). CI: Confidence intervals.The percentages of eight categories, namely nonverbal nonvisual, verbal nonvisual, nonverbal visual, equal, verbal superior, visual superior, complementary, and conflicting, for nine features, are given in [Table 3]. For location, the most common categories were equal and verbal superior. For side, the most common categories were equal and nonverbal visual. For amount, it seemed as if that there was only a single category, namely equal, since percentages of other categories were very low. For distribution and edge, the most common categories were nonverbal nonvisual and nonverbal visual. For dimension and figure, nonverbal visual was by far the most common category. For color, the most common categories were equal, nonverbal visual, and visual superior. For elementary lesion, the most common categories were complementary, visual superior, equal, and verbal superior.
Table 3: The percentages of eight categories for nine features mentioned in the 118 written descriptions and/or detected in their related photographs extracted from 89 case reports published in the journal Case Reports in Dermatological MedicineCIs of proportions of the eight categories were also calculated for the nine features. Thus, 144 numbers were obtained. It was very difficult to interpret these numbers. Hence, these intervals were visualized separately for each feature [Figure 2]. In the evaluation of multinomial CIs, the written descriptions were better in location, whereas the photographs were better in the side, distribution, dimension, figure, edge, and color [Table 4].
Figure 2: The CIs of the percentages of eight categories for nine features. CI: Confidence intervals.Table 4: Comparison of the photographs to the written descriptions according to frequencies, binomial confidence intervals, and multinomial confidence intervalsEvaluation of articles published in the journals with high-impact factors
In [Table 5], the success rates in the written descriptions and the photographs obtained from the evaluation of articles published in the journals with high-impact factors are given together with those obtained from the evaluation of the Case Reports in Dermatological Medicine articles. Although some exceptions, the rates obtained from these two evaluations were very similar. Articles of Case Reports in Dermatological Medicine had somewhat higher success rates for touch perception, subjective symptom, and elementary lesion in written descriptions and also for location and side in photographs.
Table 5: Comparison of the evaluation of articles published in the journals with high-impact factors and the evaluation of the Case Reports in Dermatological Medicine articles in terms of the success rates in the written descriptions and the photographs DiscussionAt first glance, it seems impossible to detect some features of skin lesions, such as touch perception and subjective symptom, from photographs. Therefore, at the present time, written description still keeps its indispensable place in conveying information about a skin lesion to others. On the other hand, it is possible to transmit tactile sensations from a remote object by developments in the field of haptics.[9] Moreover, tactile rendering systems have been developed generating a virtual haptic surface even from a single two-dimensional skin image, so providing an opportunity to examine surface roughness.[10],[11] However, until the use of such systems becomes widespread, mentioning touch perception in written descriptions preserves its validity and importance.
In this study, it was found that touch perception and subjective symptom were mentioned only in approximately one-third and one-half of the written descriptions, respectively. In my opinion, these rates are low. Palpation of skin lesions gives information about a lot of subfeatures, such as consistency, tenderness, and signs such as Nikolsky sign, deepness, mobility, dryness, tension, adherence, fragility, warmth, roughness, and tightness. Therefore, almost all skin lesions have some changes detectable only by palpation. As for subjective symptom, there are a lot of subjective symptoms for skin lesions, such as itching, burning, stinging, tingling, and pain. It is not a correct approach to talk about subjective symptom only when there are some symptoms. Absence of symptoms should also be underlined as in 21 of 118 written descriptions examined in this study. One word, not a thousand, is enough for this underlining: “asymptomatic.”
In this study, comparison of the photographs to the written descriptions was done in three steps: (1) frequencies, (2) binomial CIs, and (3) multinomial CIs [Table 4]. In the first step, the verbal descriptions were better only in location, whereas the photographs were better in the rest of the features, namely side, amount, distribution, dimension, figure, edge, color, and elementary lesion. According to binomial CIs, some of these advantages were lost, so, only betternesses of the photographs in side, distribution, dimension, figure, edge, and color kept their significance. According to multinomial CIs, the betterness of the written descriptions in location regained its significance. Briefly, the followings were concluded: (1) the written descriptions gave more contribution in location; (2) the photographs gave more contribution in side, distribution, dimension, figure, edge, and color; (3) the written descriptions and the photographs were interchangeable for amount; and (4) the written descriptions and the photographs were complementary for elementary lesions.
In a recent study, it was found that CIs were reported for the primary outcome only in 22% of 97 studies selected from six dermatology journals.[12] On this low reporting rate, it was advised that, if applicable, authors should report CIs, editors should review manuscripts for the necessity of CIs, and reader should rely on CIs instead of P values.[13] The present study also emphasizes the importance of CIs, since conclusions according to frequencies may change after analyses based on either binomial or multinomial CIs.
At the beginning, I decided to review only one case report journal, namely the journal Case Reports in Dermatological Medicine, in order to easily reach many case descriptions as possible, since such journals are composed almost exclusively of case report articles. However, it is not the best way to extract significant conclusions from a single journal in a study design like mine. Hence, randomly selected case report articles from journals with high-impact factors were also evaluated to verify the results. Success rates of both kinds of journals in the written descriptions and the photographs were very similar. However, higher success rates for touch perception, subjective symptom, and elementary lesion in written descriptions were observed in the evaluation of the journal Case Reports in Dermatological Medicine. Moreover, photographs of this journal also gave higher success rates for location and side. This might be due to that articles of journals with high-impact factors are usually focused on advanced test results such as those of genetic analyses, so they usually describe clinical characteristics with only a few sentences in their texts and usually use clinical images as small and cropped photographs.
Even at the end of the twentieth century, the usual practice in taking a medical image was film photography. Quality of an image taken in this way could be evaluated only after shooting the entire film and developing it, in other words, only after at least several days. Digital photography became widespread two decades ago, and using smartphones, one decade ago. Digital photography has offered many advantages, such as cost reduction, storage in very little space and without loss of quality, quick retrieval, and easy replication.[14] Quality of an image taken by a digital camera or a smartphone can be evaluated just after shooting, in other words within seconds. If necessary, it can be erased and retaken immediately.
It is not certain whether or not these advantages of digital photography have enhanced quality of images taken in dermatological practice. More attention was used to pay in film photography, since compensation of errors was usually impossible. However, it seems that a less attention is paid in digital photography. The following result of the present study could support this idea. The feature “edge” was distinguishable only in half of the cases, since it was not possible to decide whether or not the edge was well defined in some photographs. This means that the sharpness of these photographs was not enough. As compared to dedicated digital cameras, smartphones may produce less sharp images with spherical distortion.[15] In my opinion, other problems of hastily taken digital images are closing up without using an object as size reference and excluding parts of the body useful in location determination.
Teledermatology has gained more importance under COVID-19 pandemic. At the very beginning of the pandemic, it was reported that teledermatology may be a useful tool in the diagnosis of irritant contact dermatitis due to increased hand hygiene.[16] Recently, a systematic review has been performed on articles dealing with teledermatology under the pandemic.[17] In this analysis, it has been concluded that teledermatology is a useful tool in the management of ambulatory dermatoses. However, it has also been stated that the success of teledermatologic consultations depends on high-quality clinical photographs. This statement supports the fact that a picture gives a contribution to words in conveying clinical dermatological information, only when it is high in quality.
Scientific literature is becoming increasingly online. By rapid development in digital technology, online publication is already able or will be able in near future to put advanced imaging modalities such as three-dimensional photography, total body photography, and high-resolution images into service. Hence, online publication will provide opportunities to examine a lesion from all angles, to look at not only a specific area but also all parts of the body requested by the user, and to see a lesion very closely.
ConclusionI think studies investigating contributions and defects of images in case reports, as the present study has done, will be useful in guiding the future of online publication.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References
留言 (0)