The 100 most influential publications in paracetamol poisoning treatment: a bibliometric analysis of human studies

Background Analysis of the most influential publications within paracetamol poisoning treatment can be helpful in recognizing main and novel treatment issues within the field of toxicology. The current study was performed to recognize and describe the most highly cited articles related to paracetamol poisoning treatment. Methods The 100 most highly cited articles in paracetamol poisoning treatment were identified from the Scopus database in November 2015. All eligible articles were read for basic information, including total number of citations, average citations per year, authors’ names, journal name, impact factors, document types and countries of authors of publications. Results The median number of citations was 75 (interquartile range 56–137). These publications were published between 1974 and 2013. The average number of years since publication was 17.6 years, and 45 of the publications were from the 2000s. A significant, modest positive correlation was found between years since publication and the number of citations among the top 100 cited articles (r = 0.316; p = 0.001). A total of 55 journals published these 100 most cited articles. Nine documents were published in Clinical Toxicology, whereas eight documents were published in Annals of Emergency Medicine. Citations per year since publication for the top 100 most-cited articles ranged from 1.5 to 42.6 and had a mean of 8.5 citations per year and a median of 5.9 with an interquartile range of 3.75–10.35. In relation to the origin of the research publications, they were from 8 countries. The USA had the largest number of articles, 47, followed by the UK and Australia with 38 and nine articles respectively. Conclusions This study is the first bibliometric assessment of the top 100 cited articles in toxicology literature. Interest in paracetamol poisoning as a serious clinical problem continues to grow. Research published in high-impact journals and from high income countries is most likely to be cited in published paracetamol research.


Background
N-acetylcysteine (NAC) is a well-established antidote for a paracetamol (acetaminophen) overdose and is highly effective in minimising liver injury if administered promptly. However, NAC is associated with a number of adverse effects, including nausea, vomiting and non-immunoglobulin E anaphylactic (anaphylactoid) reactions (Zyoud et al. 2010c). A number of methods have been applied in clinical practice for risk stratification, such that NAC is administered to patients thought to be at significant risk of paracetamol-induced liver injury. Most prominently, the extent of paracetamol exposure may be estimated from a measured serum concentration at a known interval after ingestion. However, there are difficulties in relying on patient reporting, and errors may occur concerning the reported drug name, dose and timing of ingestion (Zyoud et al. 2012;Hewett et al. 2013;Rutter et al. 2013). A history of acute or chronic alcohol ingestion and biochemical tests of malnutrition have been explored as means of detecting an increased susceptibility to paracetamol liver toxicity; however, none of these are sufficiently reliable for routine clinical application (Waring et al. 2008a, b, d;Zyoud et al. 2011).
The original NAC infusion regimen described by Prescott in the late 1970s gives rise to very high initial blood concentrations, which are associated with the development of adverse effects; a number of alternative regimens have been described, which incorporate a slower initial rate of infusion (Prescott et al. 1977;Zyoud et al. 2010a, b;Waring 2012). Novel NAC infusion regimens are associated with a lower rate of occurrence of adverse effects, but too few data are currently available for a comparison of efficacy in preventing paracetamolinduced liver injury (Chiew et al. 2016). Nonetheless, the regulatory authorities in the UK made substantial changes to the NAC Marketing Authorisation in September 2012, incorporating a slower delivery of the initial loading dose of 150 mg per kg, extended from 15 to 60 min. At the same time, amendments were made to the criteria for NAC so that treatment should be considered for any patient that ingests more paracetamol than 75 mg per kg of body weight or has a measured paracetamol concentration higher than the 100-line nomogram that was formerly used for "high risk" patients only. Other countries have followed suit and incorporated these amendments into local policy. At present, few data exist concerning the impact on rates of occurrence of liver injury or adverse acetylcysteine effects. Early reports from the UK indicate that, since September 2012, there have been increased numbers of patients receiving acetylcysteine, increased hospitalisation and a substantial financial burden (Thompson et al. 2013;Bateman et al. 2014a).
Other novel therapeutic approaches have been examined in patients receiving an intravenous NAC antidote after paracetamol poisoning. For example, co-administration of cimetidine has been found to have no significant effect on patient outcome (Ebrahimi et al. 2015). Co-administration of ondansetron is highly effective in reducing the adverse effects associated with NAC administration, and this might be considered for routine administration in patients at high risk of anaphylactoid reactions, for example those with a history of asthma or previous adverse effects after NAC (Bateman et al. 2014b).
While numerous bibliometric reports have been performed to investigate factors related to research output in the toxicology field (Zyoud et al. 2010b(Zyoud et al. , 2014a(Zyoud et al. , 2015a, to our knowledge, no study has attempted to evaluate the most influential publications within a particular subspecialty in toxicology. The main aim of our study was to identify the 100 most frequently cited articles related to paracetamol poisoning treatment. Therefore, a bibliometric evaluation of scientific literature in a particular field may be used to recognize the impact of influential scholarly work, authors, subjects, countries, etc. The total number of citations that a published article has achieved indicates the importance that published article has on that area of practice. Analysis of the most influential publications within paracetamol poisoning treatment can be helpful in recognizing main and novel treatment issues within the field of toxicology.

Data acquisition
In November 2015, we used the Scopus database to retrieve the most frequently cited articles on paracetamol poisoning treatment. Scopus is the largest electronic scientific database. It is larger than either MEDLINE or Web of Science and is more accurate than Google Scholar. Furthermore, Scopus is accessible and allows researchers to obtain information and do analysis that might not be readily achievable in MEDLINE or Google Scholar.
The search terms used to retrieve articles related to paracetamol poisoning treatment were elected from previous bibliometric studies related to paracetamol (Robert et al. 2009;Zyoud et al. 2015a, d). The articles entitled with the terms "acetaminophen" or "paracetamol" or "acetamidophenol" or "hydroxyacetanilide" or "tylenol" or "n-acetyl-p-aminophenol" or "panadol" or "APAP" or "acephen" were included in the research. Then, the following keywords were entered as terms in the article title or abstract: "poison*" or "overdose*" or "*toxic*". All the previous terms were followed by "treat*" or "antidot*" or "detoxification" or "intervention" or "manag*". To include all possible publications related to paracetamol treatment, we included the following strategy in addition to the previous one. The following keywords, "NAC" or "acetylcysteine" or "N-acetyl-l-cysteine", were entered as terms in the article title or abstract and followed by the following terms: "acetaminophen" or "paracetamol" or "acetamidophenol" or "hydroxyacetanilide" or "tylenol" or "n-acetyl-p-aminophenol" or "panadol" or "APAP" or "acephen". The Scopus search was conducted by applying the previous strategies of search for all previous years. No limits were placed on the time period for this search. Furthermore, all documents related to paracetamol, with no language restriction, were searched in Scopus database. Wildcard characters were used to include variations of a word by using an asterisk (*) to make our search strategy simpler. For example, in the Scopus search engine, when we entered "*toxic*", it offered a wildcard character capability, and we got results for toxic, toxicity, toxicant, toxicities, intoxication-briefly, any possible word that might include the five letters (i.e. toxic).

Data analysis
We sorted the extracted results from the largest number of citations to the lowest. The results were then evaluated by two independent researchers to extract only the top 100 cited articles related to paracetamol poisoning treatment in human research subjects. Total numbers of citations, average citations per year, authors' names, journal name, impact factors (IF), document types, countries of publication, and number of authors for only the top 100 cited articles were noted. The impact factor was obtained from 2014 Journal Citation Reports ® (Thomson Reuters, New York, NY, USA) ( Thomson Reuters 2015) for all journals that published the top 100 cited articles. This is useful in clarifying the significance of absolute citation frequencies. This search used the electronic version of this database by ranking the top 100 cited publications using the standard competition ranking (SCR). It is possible that some articles were cited more frequently than others because of the long time elapsed since their publication. Therefore, a citation index was calculated for each article to avoid the bias created by the time elapsed since publication. Citation index is calculated as the average number of citations divided by number of years elapsed since the article was initially published. A Pearson's correlation coefficient test was applied to assess the correlation between the impact factor of the journal or years since publication or citation index and the number of citations for the top 100 cited articles included in the study. This was carried out using Statistical Package for Social Sciences (SPSS) Version 15.0. Descriptive statistics were presented as frequencies, medians and interquartile ranges or averages. p values <0.05 were considered statistically significant for all the comparisons.

Average number of citations per year
Average number of citations per year for the top 100 cited articles ranged from 1.5 to 42.6 with a mean of 8.5 citations per year and a median of 5.9 (interquartile range 3.75-10.35). Table 2 ranks the top 10 publications based on the highest average number of citations per year. Nine of the articles in the list were published after year 2000. The top three articles based on average number of citations per year were the followings: "Acetaminopheninduced hepatotoxicity" with 42.6 average number of citations per year, "Mechanistic biomarkers provide early and sensitive detection of acetaminophen-induced acute liver injury at first presentation to hospital" with 42.0 average number of citations per year and "The chemistry and biological activities of N-acetylcysteine" with 38.6 average number of citations per year. Interestingly, the total number of citations was significantly correlated with citation index (r = 0.485, p < 0.001).

Authorship
The total number of authors for the top 100 cited articles was 419, for an average of 4.15 authors per paper. Authors per paper ranged from 1 to 23, and 20 articles were written by one author. Table 3 ranks the top 10 most prolific authors, who have published at least four publications among the 100 most cited publications. Twenty-eight documents were published in the ten journals with an IF >10. Only three journals for the top 100 cited articles were without IF. A significant, modest positive correlation was found between the journal impact factor and the number of citations among the top 100 cited articles (r = 0.426; p < 0.001).

Origins and publication type
In relation to the origin of the research publications for the highly cited articles, they were from eight countries. The USA had the largest number of articles with 47 articles. The UK and Australia published 38 and nine articles respectively, whereas Denmark, Switzerland, Canada, New Zealand and Israel published four, four, three, two and two articles respectively. In the terms of "collaboration with other countries", we found that eight articles were published and co-authored by researchers from multiple countries. Sixty-six articles were original articles, 32 articles were review articles and three were other types of publications, including conference papers, editorials and notes.

Discussion
The present study was designed to rank and characterize the top 100 cited publications in the field of paracetamol poisoning treatment. The most obvious finding to emerge from this study is that results of this study may explain how developments in this area of clinical toxicology have progressed over time. It becomes obvious which key publications and authors have made exceptional contributions that have played an integral role in shaping the guidelines related to the treatment of paracetamol. Furthermore, the results of this study enhance our understanding about the leading publications that have contributed to the development of this field of toxicology. Common treatment guidelines for paracetamol poisoning that are currently in use include Australian and New Zealand guideline by Daly et al. (2008), USA guideline by Dart et al. (2006) and UK guideline by Wallace et al. (2002). These guidelines ranked 33rd, 55th and 77th respectively. Furthermore, these common guidelines were based on several articles in the top 100 cited publications such as Smilkstein et al. (1988), Prescott et al. (1979), Keays et al. (1991), Prescott et al. (1977), Rumack (2002), and Smilkstein et al. (1991).
The top 100 cited studies in our study were published from 1974 to 2013 with a citation range of 44-553 times since publication. Compared to citations in other medical  (Chen et al. 2015), compared to 582-7248 for hypertension (Oh and Galis 2014). The difference in the total number of citations in each area reveals the number of researchers working in each area (Chen et al. 2015). It is also well known that editors for the main journals with high IFs consider articles with a high citation rate to sustain the IF progress of their journals (Zyoud et al. 2014a(Zyoud et al. , b, 2015a. Furthermore, since the toxicology field is considered as a very narrow field with a very small readership and number of researchers, it should not be surprising for toxicology topics or toxicology journals to have a small numbers of citations (Bird 2008;Zyoud et al. 2014b). Our results confirm that there was a modest relationship between the number of years elapsed since the time of publication and the number of citations. These results support previous findings which reported that older articles attained more citations because their citable period was longer (Loonen et al. 2008;Lefaivre et al. 2011;Aminian et al. 2014;Joyce et al. 2014;Lee et al. 2015). The annual total number of citations for any article fluctuates with time and for some articles the total annual number of citations might decrease with time while for other articles, the total annual number of citations might increase or remain steady with time. In our study, the low correlation between the total citations and citation index seems to be related to papers that have many citations when they are first published but then drop off in later years because researchers might be tended to preferentially cite the most recent studies (Azer and Azer 2016). Our study showed that the average number of citations per year for articles published after year 2000 was higher than those published before 2000. This finding may be due to the tendency of authors to cite recent papers which is a common practice among authors (Van Noorden et al. 2014). The most influential article in paracetamol poisoning treatment was conducted by Smilkstein and experts from the USA (Smilkstein et al. 1988) and was published in 1988 in New England Journal of Medicine. This article described the final outcomes of 2540 patients with paracetamol overdose treated with a loading dose of 140 mg per kg of oral NAC followed by 70 mg per kg given every 4 h for an additional 17 doses. It was concluded that NAC treatment should be started within 8 h of an paracetamol overdose (Smilkstein et al. 1988). In contrast to earlier findings, recently, in the last decade, described novel NAC infusion regimens offer different rates of intravenous NAC administration in both the loading and maintenance doses, and this is associated with a lower rate of the occurrence of adverse effects (Pakravan et al. 2008;Waring et al. 2008c). These studies were in the list of influential papers in the top 100 cited articles.
The first paper for paracetamol poisoning treatment was cited only 25 times and was published by Maclean et al. (1968) in 1968 in The Lancet. This article recommended immediate gastric lavage, intravenous hydrocortisone, forced diuresis and antihistamine for paracetamol poisoning treatment. This article achieved a low rate of citation to be listed in the top 100 cited publications in the field of paracetamol poisoning treatment because it was based on treatment regimens, which limited recommendations for the treatment of such cases. The two pioneering publications in the field of general paracetamol poisoning were conducted by Mitchell et al. (1973a, b). They explained the mechanism of paracetamol hepatotoxicity, were published in Journal of Pharmacology and Experimental Therapeutics and are considered as remarkable papers in paracetamol poisoning (Rumack and Bateman 2012). These two publications achieved higher citations in the field of general paracetamol poisoning than in the field of paracetamol poisoning treatment, 1052 and 904 citations each respectively (data not shown), but did not appear in our list because they involved animal research. What is surprising is that this group was in our list and got an advanced position (i.e. the seventh most frequently cited paper) by providing a rationale for therapy in humans by indicating that cysteamine could prevent hepatotoxicity (Mitchell et al. 1974). Prescott et al. (1974) reported the successful treatment of patients with severe paracetamol overdose with cysteamine, but this article was not among the top 100 cited articles. This might be due to the adverse effects in patients making it a less than ideal antidote and of low interest to researchers. Surprisingly, Prescott et al. (1979) were found to be in the list of the top 100 cited articles for a different article, which achieved the third position in the list. They reported that intravenous NAC was more effective in the treatment of paracetamol poisoning than cysteamine and methionine and was markedly free from adverse effects (Prescott et al. 1979). Furthermore, it is somewhat surprising that Prescott and his colleagues in Edinburgh published a series of influential papers in the top 100 cited articles (Prescott et al. 1976(Prescott et al. , 1977(Prescott et al. , 1989Prescott 1983Prescott , 2000Prescott , 2005Prescott and Critchley 1983).
One of the pioneering article in the field of general paracetamol poisoning was written by Rumack and Matthew (1975). This article is considered as one of the highly cited articles in the field of general paracetamol poisoning, rather than in the field of paracetamol poisoning treatment. It achieved 303 citations (data not shown) but did not appear in our list. Contrary to expectations, several studies (Schiodt et al. 1997;Larson et al. 2005) were dropped from the list of the top 100 cited articles in the field of paracetamol poisoning treatment because these studies did not mention paracetamol poisoning related treatment terms in their titles or abstract. The most obvious finding to emerge from our study is that Rumack published a series of influential papers in the top 100 cited articles (Rumack and Peterson 1978;Rumack et al. 1981;Rumack 1984Rumack , 1986Rumack , 2002. In the present study, 28 documents were published in the ten journals with an IF >10, including New England Journal of Medicine, The Lancet, Annual Review of Pharmacology and Toxicology, Annals of Internal Medicine, British Medical Journal, Archives of Internal Medicine, Gastroenterology, Gut, Annual Review of Medicine and Hepatology. Our results confirm the modes relationship between the journal impact factor and the number of citations among the top 100 cited articles. These results seem to be consistent with other research, which found that the most cited articles in the field of tuberculosis are often published in journals that top the impact factor list (Chen et al. 2015). The results of our study show that more than half of the publications originated from the USA, followed by the UK. These results match those observed in earlier studies (Loonen et al. 2008;Joyce et al. 2014;Oh and Galis 2014;Chen et al. 2015;Dolan et al. 2015). Research activity in these countries is most likely due to their economic strength (Li et al. 2013;Yun et al. 2015), or numerous large poison centers (Forrester 2016), or the number of researchers or general research activity in this scientific field (Zyoud et al. 2015a, d), or number of poisoning incidents in these countries (Bateman 2014;Mowry et al. 2015).
There are several limitations of this study. First, our analysis to choose publications with a primary focus on paracetamol poisoning treatment research likely excluded articles that had otherwise notably influenced thinking in the field, including some of the highly cited articles that refer to investigations of paracetamol poisoning in animals or in vitro. Second, it was based on the Scopus database alone; Scopus does not index all journals, and we may have missed journals that figure in other databases such as Google Scholar. Third, articles published in recent years had less of a chance to be among the top 100 cited articles because less time has elapsed since the date of publication to allow citation. Another limitation is that some publications did not mention paracetamol poisoning related treatment terms in their titles or abstract, so it is possible that not all publications about paracetamol were considered. The present study design means that we are unable to include some clinicians that have influenced the treatment of paracetamol poisoning through other means, including unpublished research work, presentations at scientific meetings, and pioneering clinical practice. For example, many influential toxicologists, such as Matthew H., Jaeschke H., Piperno E. and Khairallah E.A. (Rumack 2002;Proudfoot and Prescott 2009;Rumack and Bateman 2012), were not identified in our study.
In conclusion, we carried out a bibliometric analysis of the most cited publications focused on paracetamol poisoning research, revealing a number of characteristics related to these influential publications, including the country of origin, type of study, journal and authorship. This study is the first bibliometric assessment of the top 100 cited articles in toxicology literature. The most influential report in paracetamol poisoning treatment research appears to be conducted by Smilkstein et al. from the USA and was published by in 1988 in New England Journal of Medicine. Interest in paracetamol poisoning as a serious clinical problem continues to grow. Research published in high-impact journals and from high income countries is most likely to be cited in published paracetamol research.