Epidemiology of diabetic foot disease and diabetes-related lower-extremity amputation in Australia: a systematic review protocol

The following databases will be searched: PubMed and Excerpta Medica Database (EMBASE) via Ovid SP and will include publications in any language and without restrictions in date. Before performing the search, a validation set of approximately 10 publications was created and tested. Each publication in the set needed to be identified in the search. The validation set was created by including key publications known to the authors that fit the scope of this systematic review. The final search string is shown in Table 1.

Preliminary searches resulted in 155 records and 20 potentially eligible publications. Records and data can therefore be managed using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA). All records will be imported from PubMed and EMBASE into an Excel file. This file will be used for study selection. Keeping the master copy and all management of this file is the responsibility of the first author. The Statistical Package for the Social Sciences (IBM Corporation, Armonk, NY, USA) will be used for data analysis, with data imported by the first author.

Title and abstracts will be screened by two independent investigators (JJvN and MB) for inclusion in the full-text assessment. Criterion for inclusion for full-text assessment is any mention of epidemiological data on diabetic foot disease or lower-extremity amputation in an Australian setting. A publication identified for inclusion by either one of the investigators will be included for full-text assessment. Cohen’s kappa will be calculated for screening agreement by the two investigators [16].

Included full-text publications will be assessed by two independent investigators (JJvN and MB) for eligibility using the inclusion and exclusion criteria. Inclusion criteria will be based on publications reporting Australian population-based incidence or prevalence of diabetic foot disease or diabetes-related lower-extremity amputation. Exclusion criteria will be based on publications reporting data from a single centre or selected centres (unless general- or diabetes-population information for the selected centres is provided or if data is adjusted or compared to the overall Australian population demographics); publications reporting lower-extremity amputation incidence or prevalence without differentiating between people with and without diabetes; and publications not reporting original data. No restrictions are placed on study duration, study period, or date of publication. Disagreements between investigators will be resolved with discussion until consensus is reached, with a third investigator to be involved if needed.

Diabetic foot disease will be defined as “infection, ulceration or destruction of tissues of the foot associated with neuropathy and/or peripheral artery disease in the lower extremity of people with diabetes” [17]. Diabetes-related lower-extremity amputation will be defined as “resection of a segment of the lower-extremity through a bone or joint” in a person with diabetes [17]. This will be further differentiated as major amputation (any resection proximal of the ankle) and minor amputation (any resection through or distal of the articulation of the ankle) [17]. However, we are mindful that different definitions may have been used in different studies and we will also include (and report) other definitions if used.

The denominator population will be defined as the general or diabetes population reported for a specified geographical area, with key characteristics (age and gender) to be reported for this population in the publication. Publications reporting in-hospital epidemiology of diabetic foot disease or diabetes-related lower-extremity amputation will be included, when data from the complete hospital population is given, and provided more than one hospital is included in the publication. Common risk factors for diabetic foot disease may also be used to further specify the “at-risk population” in the denominator (when available in a publication), and these include peripheral neuropathy, peripheral artery disease, and foot deformities [17, 18].

After completion of selection of the final included publications, a forward literature search will be performed via Google Scholar (https://​scholar.​google.​com.​au). This may help to identify early online publications that have not been indexed (since Google Scholar does not rely on indexation before publications are included in their database), and it may capture some grey literature as well (such as dissertations and conference proceedings). One author will assess the publications found in the forward literature search. All publications identified for inclusion by this author will be screened for eligibility as per the above study selection procedure of full-text publications.

As the authors are aware of government publications on the primary outcomes of this systematic review, a search to identify those publications and other grey literature meeting the inclusion and exclusion criteria will be undertaken. This search will consist of the following steps, performed by one author:

As both EMBASE and Google Scholar include conference presentations and dissertations, the searches in those databases described elsewhere in this manuscript can also be regarded as covering part of the grey literature search.

All publications identified for inclusion in the grey literature search will be screened for eligibility as per the above study selection procedure of full-text publications.

Data from each publication will be extracted by one investigator, after which a second investigator will check the extracted data. Investigators will not extract data from an article of which they were (co-)author to prevent potential bias or conflict of interest. The primary outcome is the incidence or prevalence of diabetic foot disease and diabetes-related lower-extremity amputation. Data extracted will include setting, period, study design, use of prevalence or incidence, number of general or diabetes population, key characteristics (age, sex, foot disease risk factors, other) within the population, definition of diabetic foot disease or amputation, classification of diabetic foot disease or amputation level, other characteristics of diabetic foot disease or amputation, and if the diabetic foot disease or amputation is the first ever incidence or a recurrence. Authors will be contacted if data to be extracted are unclear from the original publication.

As diabetic foot disease and amputation pose specific requirements on study reporting, no specific quality assessment instrument is suited to the data and research aim of this review. Therefore, a customised approach will be followed. In line with conclusions from a recent systematic review on tools for assessing the quality and susceptibility to bias in observational studies, a transparent checklist approach has been chosen, concentrating on the few—principal—potential sources of bias identified from this systematic review [19]. The customised checklist developed to assess selection and information bias by the investigators is based on checklists used by other systematic reviews on population-based studies investigating mortality after amputation [20] and diabetic foot disease [21], which in turn were based broadly on the Meta-analysis Of Observational Studies in Epidemiology guidelines [22] and the Newcastle-Ottawa scale [23]. Our checklist aims to assess selection and information bias.

Five criteria will be assessed for each included publication by two independent investigators (Table 2). Item C is outcome specific and is scored twice when both primary outcomes are included. Each criterion is scored with a “+”, “?”, or “−”. The criteria chosen are based primarily on the quality of the reporting of the geographical area, diabetic foot disease or amputation numerator, population denominator, information flow, and study design. These criteria are all known to be important aspects of bias in diabetic foot disease research [21, 24]. Disagreements will be resolved by discussion until consensus is reached, with a third investigator to be involved if consensus is unable to be reached. Investigators will not assess the quality of an article of which they were (co-)author to prevent any conflict of interest.

As grey literature is not peer-reviewed, an additional quality assessment of the grey literature will be performed, using the AACODS (Authority, Accuracy, Coverage, Objectivity, Date, Significance) checklist [25].

The summary statistic used for each study will be an incidence (rate per population) or prevalence proportion (%) of diabetic foot disease or diabetes-related lower-extremity amputation. The standard error for each proportion will be calculated.

A meta-analysis will be performed if three or more publications, with data collected within 10 years of each other, reporting on the same outcome and in a similar geographical area or with similar denominator population are found, providing that quality assessment items A, B, and C are scored with a “+”. Meta-analysis will be used to calculate pooled incidence or prevalence estimates using a random-effects model. Random effects will be used to give an average estimate across heterogeneous studies weighted on total sample size. The I ² test will be used to examine heterogeneity across studies included in meta-analysis.

No meta-bias or confidence in the cumulative evidence will be calculated or assessed. Availability of prospectively registered protocols of epidemiological studies is unlikely, and as such unpublished results to calculate publication bias cannot be investigated or estimated. Further, this systematic review does not concern intervention trials, rendering the Grading of Recommendations Assessment, Development and Evaluation system not applicable. The confidence in the cumulative evidence will be a descriptive analysis based on the separate quality assessment outcomes of the included studies.