Diabetic Foot Disease Research in Aotearoa New Zealand: A Bibliometric Analysis (1970-2020)

The aim of this bibliometric study was to examine trends in the quality and quantity of published diabetic foot disease (DFD) research in Aotearoa/New Zealand (NZ) over the past ve decades. Method In July 2021, the Scopus® database was searched for DFD-related publications (1970-2020) using predetermined search and inclusion criteria. Bibliometric data were extracted from Scopus® and Journal Citation Reports. Retrieved bibliographic indicators were analysed in Biblioshiny, an R Statistical Software interface and reported using descriptive statistics.


Introduction
Diabetic foot disease (DFD) is one of the most devastating, but potentially avoidable complications of diabetes (1,2). DFD is de ned as a foot affected by infection, ulceration or destruction of tissues of the foot of a person with currently or previously diagnosed diabetes mellitus, usually accompanied by neuropathy and/or PAD in the lower extremity (3). Diabetic foot ulcers are the most frequently recognised complication of DFD and a major risk factor for, and nearly always precede, diabetes-related lower-extremity amputation (DRLEA) (4-6). DRLEA is one of the most substantial and debilitating consequences of diabetes (7).
In Aotearoa/New Zealand (NZ), diabetes is common. In 2008/09, the prevalence of diabetes (diagnosed and previously undiagnosed) among those aged ≥15 years was 7.0% overall, and ethnic speci c rates were 9.8%, 15.4% and 6.1% among Māori, Paci c peoples and non-Māori non-Paci c peoples, respectively (8). More than half (58%) of DRLEAs are attributable to diabetes (9) and Indigenous Māori people are more likely to experience DRLEA than their NZ European counterparts (10). Gurney et al. demonstrated that Māori with diabetes were 65% more likely to undergo major DRLEA than NZ European/Other people with diabetes (9).
Understanding DFD research quantity and quality in the NZ context is important, as Indigenous Māori people not only have high rates of DRLEA (10,11), but also fare worse in many other diabetes-related health measures (11,12).
Whilst the understanding of DFD has been advanced over many decades based on international research, the contribution of locally NZ driven DFD-related research appears to be limited (10)(11)(12). Consequently, we do not know if NZ based DFD-related research is strategically targeting areas where research need is the greatest in order to reduce diabetes-related DRLEA and improve health outcomes. The objective of this study was to provide the rst comprehensive bibliometric analysis of DFD-related research generated by NZ based researchers. Speci cally, the study aimed to identify underlying patterns in DFD publications, author-speci c contributions, the volume of scholarly work overtime, the degree of national and international collaborations, and the major topics/areas of research focus.

Data source
This bibliometric analysis of NZ DFD-related publications between 1970 and June 2021 was conducted in July 2021 using data sourced from the Scopus® database (Elsevier, Amsterdam, Netherlands). The Scopus® database was selected as it enables search by document, author or a liation, or use, with the ability to re ne results by author and publication characteristics. It has the largest abstract and citation database of research literature (13). As of January 2020, Scopus® had in excess of 25,100 active titles and over 550 articles in press (14). Additionally, Scopus® includes a more expanded spectrum of journals than PubMed and Web of Science®, and its citation analysis is faster and includes more articles than the citation analysis of Web of Science® (15).

Search strategy
The search strategy was developed through a staged process, involving adaptation of a search strategy used in a previous bibliometric analysis of DFD conducted by some of the authors of this study (16). Initially, Al-Busaidi et al.'s search strategy was run in Scopus® (16), with the 15 most cited articles retrieved. Author keywords, Medical Subject Headings (MeSH) classi cations, and Emtree (Embase subject headings) terms were then downloaded from Scopus® and exported into NVivo Qualitative Data Analysis Software (QSR International Pty Ltd. Version 12, 2018) and analysed by text query analysis to obtain word frequency counts.
Keywords were reviewed and discussed by the authors to develop the nal search strategy displayed in Table  1.

Data processing
The titles and abstracts of all identi ed publications were downloaded from Scopus® database and exported into the online systematic review application Rayyan (http://rayyan.qcri.org) (17). The articles were then independently screened by two authors (MC, ISA) and selected based upon pre-determined inclusion and exclusion criteria agreed by all authors (Box 1). Con icts arising from the screening process were resolved via a consensus meeting. A third author (KC) was available if screening con icts were not resolved at an initial meeting, but this was not required. The reference lists of the nal included articles were also scanned to nd any additional publications. The study retrieval process is displayed in Figure 1.

Box 1 Inclusion and exclusion criteria
Inclusion criteria Publications were included if: 1. They were original articles, or systematic reviews with meta-analysis; and 2. The research was conducted within an NZ institution; and 3. Data were reported that was conducted on an NZ population; and 4. The article had at least one author with an a liation to an NZ research institution; and 5. They were published in English, and 6. The eld of research was related to DFD (including screening, prevention, diagnosis, management, complications, and workforce) and relevant conditions (peripheral neuropathy, neuroarthropathy, peripheral artery disease, infections, deformity, ulceration, and amputation); and 7. They were published between 1970 to the current date (date of search)

Exclusion criteria
The following studies were excluded: non-original research publications, non-systematic reviews, case reports, commentaries, letters, and editorials DFD: diabetic foot disease; NZ: Aotearoa New Zealand.
The following data and bibliometric indicators were extracted from each publication: title, year of publication, journal name, journal impact factor (IF), citation count, author names, total authors per manuscript, institutional a liation, collaboration network and funding source. Collaborative networks were classi ed into four categories: (1) "international collaborative" articles involving collaboration with international authors, (2) "bi-national link" articles originating from authors a liated to only two NZ institutions, (3) "multi-national link" articles authored by researcher from three or more NZ institutions, and (4) "no collaboration" articles representing publications where all authors were a liated to the same institution (16, 18). Funding sources were classi ed into two categories: (1) industry sponsored funding, and (2) academic/public funding (funding derived from universities, hospitals, or government bodies). Using the 2019 IWGDF guidelines as guidance, articles were classi ed into the following predetermined categories; (1) screening and prevention of DFD, (2) management of DFD related conditions, (3) epidemiology, and (4) other/miscellaneous (publications that did not t into one of the above groups) (19). Articles were also characterised by type of study (basic/clinical research articles, systematic review with meta-analyses, and randomised controlled trials (RCT)) (20).
As a measure of research quality, the journal IF attained in the year prior to publication was obtained using the  (21). Biblioshiny was used to extract the following data: general characteristics of the included articles, annual scienti c production, average citations per year, and most relevant authors, and collaboration world map. The nonparametric Mann-Kendall test was applied to data to detect statistically signi cant trends in publication numbers and collaborative networks.

Results
The characteristics of the included studies are displayed in Table 2. A total of 140,489 publications were initially identi ed of which 444 articles satis ed the inclusion criteria ( Figure 1). After application of the exclusion criteria 47 articles were included in the nal analysis. Volume of scienti c production

Funding sources
Fifteen articles (32%) reported receiving funding support. Eleven articles (23%) reported receiving NZ based funding. Of these, 10 (21%) were funded through university/public sources, with one industry funded. Four (9%) international multicentre studies reported funding from international sources.

Author and authorship network
Authorship networks for articles published between 1988 and 2020 are displayed in Figure 3. No studies were single authored. Articles had a median of ve co-authors per article (range: 2 -18). Seventeen (36%) of the included articles had an international authorship link with a "bi-national link" in ten articles (21%), a "multinational link" in six articles (13%), and "no collaborative link" outside of one institution in 14 articles (30%).
There  (Table 3). Five most cited articles The top ve source journals are displayed in Table 4. The most cited article was published by Rajamani et al (22) in The Lancet in 2009 (209 citations). This article represented 22% of the total citations for all included articles. The ve most cited articles represented 46% of the total citations. Four of the most cited articles were international multicentre RCTs. Only one of the ve most cited studies had a NZ researcher as rst author (23).
With the exception of Behrendt et al's observational study (24), all studies were supported by funding.  Table 5 details the most cited articles where an author from NZ was the rst author. With the exception of Dobson et al. (23), all were observational studies based in NZ (12,23,25). Three studies declared funding from an academic/public source withing NZ while two studies were not supported nancially (26, 27).

Discussion
Bibliometric analyses are useful for providing an overview of published literature, identifying knowledge gaps, uncovering emerging trends, and deriving novel ideas for future investigation (28). The present study analysed the research quantity (i.e., publication output, areas of research focus and trends over time) and quality (i.e., the level of evidence, degree of collaboration, impact factor and citation rates) of NZ DFD-related research.
Whilst the number of articles is relatively small, between 1988 and 2020, the number of NZ produced DFD publications has steadily increased. Notably from 2006 onwards there was an increase in published articles with a peak of six new articles in 2018, and an increase in research growth and national collaboration. The underlining reasons for the increase in DFD research are not attributable to any one factor but may be resultant from numerous diabetes quality of care strategies, policies and initiatives implemented in the early 2000s in NZ. Additional File 1 presents a timeline of policies/strategies/initiatives that may have potentially in uenced NZ DFD research. Notably during this period, the NZ Government strategy elevated the importance of diabetes with the release of The NZ Health Strategy (2001) (29). Reducing the incidence and impact of diabetes was one of the 13 health objectives chosen for implementation in the short to medium term.
Whilst the signi cance of diabetes in context of NZ came to the fore with changes in health policy and strategy, research development prior to 2000 was hindered by limited research funding. In 2000, health research accounted for only 1% of the national health budget (30). It was not until 2005 that the Health Research Council (HRC) of NZ became a Crown agent, charged with putting into effect government policy in relation to health research. The opportunities for NZ researchers to obtain funding has improved in the past 15 years, however our results indicate DFD research by NZ researchers is still poorly funded with only 32% of the included articles declaring research funding support. Most of these studies were funded by national organisations (DHBs, universities, HRC; n = 11, 23%) and represent studies that were largely observational in nature. Consequently, as these types of studies are of a lower level of evidence, they are often insu cient to change clinical practice in comparison to large international multicentre and randomised studies funded through multiple international sources. Without signi cant investment, resourcing to implement high level of evidence locally relevant studies such as RCTs will remain limited. Despite the increase in number of publications and increased national and international collaborations, the majority of the identi ed publications (87%) represent studies classi ed as basic/clinical research, which represents a lower level of evidence as per the Oxford Centre for Evidence-based Medicine (level 3 or 4 evidence) (31). Based upon the Oxford levels of evidence rating, and the relatively low citations rates, the majority of the identi ed NZ DFD-related publications were categorised as of poor quality. Of the top ve cited articles (contributing 46% of total citations), four were RCTs (one rst-authored by NZ researcher) and four were multicentre studies rst-authored and led by international researchers. The most cited articles are either multicentre national or international RCTs representing high quality of evidence.
The median journal IF (4.31, IQR: 2.75-6.81) was relatively low for included publications. Only ve articles were published in journals with an IF of greater than 10. Of note, most publications were published in the New Zealand Medical Journal (n = 10, 21%), which does not currently hold an IF. IF is the most common metric for evaluating bibliometric impact of published research, however the value of the research is not necessarily re ected by the IF (16). This nding is interesting as researchers often seek publication in so called higher impact/prestigious journals with the intention of improving their personal citation rate, and h-indices. Furthermore, academic staff promotions at universities often depends upon the publication of a certain number of articles in scienti c journals (32). However, authors sometimes chose to publish in journals based on the intended audience/readership where the article may have the most context and/or clinical impact. This may be more common when authors come from a clinical rather than purely academic background, where their driver maybe to improve outcomes rather than produce high ranked research outputs. Alternatively, the decision to publish in a particular journal may be dictated by the availability of funding to support the fees associated with publication. A combination of the above factors may be likely reasons for many NZ authored DFD articles being published in the New Zealand Medical Journal.
Categorisation of research by type found that there was a relatively even spread number of articles categorised as screening/prevention, management of diabetes-related foot complications, and epidemiological studies. However, of the studies categorised as screening/prevention none were interventional studies. With few studies aimed at improved care or prevention of diabetic foot ulceration/amputation, coupled with recent international calls to reduce foot ulcer incidence by at least 75% and local NZ health priorities to reduce health inequities for Māori, a shift in DFD research priorities is essential (6). Therefore, the rst steps towards this goal are to evaluate the performance of diabetic foot services aimed at prevention and early detection of DFD, and the ability of services to reduce inequities in access to services and health outcomes. This is a priority in NZ given the regional variation in DRLEA and signi cantly higher amputation rates for Māori (7,9,10).
The results of this study have several limitations that must be considered. All metrics were extracted based upon our pre-de ned search terms, and data only from the Scopus® database, which may not include all publications that meet our inclusion criteria. Some peer-reviewed journals are not indexed in Scopus®.
However, we also checked for additional publications by screening reference lists of identi ed articles from the initial search. Due to the nature of a bibliometric analysis, no ethical considerations were required.

Consent for publication
Not applicable.

Availability of data and material
All data used in this article can be found on the Scopus database using the search strategy outlined in the Methods section. A complete list of all included papers in available upon reasonable request from the corresponding author.

Author contributions
All authors designed the study. MC and ISA collected, screened, and analysed all data. MC and ISA drafted the manuscript. All authors reviewed subsequent manuscript draft versions. All authors read and approved the nal manuscript prior to submission.

Competing interests
Matthew Carroll an Editorial Board member of the Journal of Foot and Ankle Research. All other co-authors declare no competing interests.

Funding
This research received no speci c grant from any funding agency.

Figure 1
Flow chart for the search and retrieval process  Visual representation of international collaborative authorship networks

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AdditionalFile1.pdf