Foot Complications in a Representative Australian Inpatient Population

We investigated the prevalence and factors independently associated with foot complications in a representative inpatient population (adults admitted for any reason with and without diabetes). We analysed data from the Foot disease in inpatients study, a sample of 733 representative inpatients. Previous amputation, previous foot ulceration, peripheral arterial disease (PAD), peripheral neuropathy (PN), and foot deformity were the foot complications assessed. Sociodemographic, medical, and foot treatment history were collected. Overall, 46.0% had a foot complication with 23.9% having multiple; those with diabetes had higher prevalence of foot complications than those without diabetes (p < 0.01). Previous amputation (4.1%) was independently associated with previous foot ulceration, foot deformity, cerebrovascular accident, and past surgeon treatment (p < 0.01). Previous foot ulceration (9.8%) was associated with PN, PAD, past podiatry, and past nurse treatment (p < 0.02). PAD (21.0%) was associated with older age, males, indigenous people, cancer, PN, and past surgeon treatment (p < 0.02). PN (22.0%) was associated with older age, diabetes, mobility impairment, and PAD (p < 0.05). Foot deformity (22.4%) was associated with older age, mobility impairment, past podiatry treatment, and PN (p < 0.01). Nearly half of all inpatients had a foot complication. Those with foot complications were older, male, indigenous, had diabetes, cerebrovascular accident, mobility impairment, and other foot complications or past foot treatment.


Introduction
Active foot disease (ulcers, infection, or ischaemia) is commonly precipitated by the foot complications of previous amputations, previous foot ulcers, peripheral arterial disease (PAD), peripheral neuropathy (PN), and foot deformity in both diabetes and nondiabetes populations [1][2][3][4]. These foot complications not only increase the risk of developing active foot disease in the community but also have been found to increase the risk of developing active foot disease, falls, and pressure injuries in the inpatient setting [3,[5][6][7][8]. Thus, it seems important for clinicians, researchers, and policy makers to understand how often these foot complications present and what other factors may precipitate them in inpatient populations.
Studies investigating foot complications in inpatient populations have predominantly focused within diabetes and geriatric inpatient populations [4][5][6][7][8][9]. These studies suggest that up to 80% of diabetes inpatients, and up to 50% of geriatric inpatients, have at least one of these foot complications [4][5][6][7][8][9]. A recent systematic review of this field found previous studies have investigated either a single foot complication (such as PAD) in a representative adult inpatient population (defined as a typical hospital's inpatient population inclusive of patients admitted for any reason, with or without diabetes and of any age) or multiple foot complications in a specific inpatient population (such as all foot complications in diabetes inpatients only) [3]. This review concluded that no previous study had investigated the prevalence or factors associated with multiple foot complications in a representative inpatient population [3]. Thus, the primary aim of this study was to investigate the point prevalence and factors independently associated with foot complications in a representative adult inpatient population. A secondary aim was to investigate if there were any differences in the prevalence of foot complications between diabetes and nondiabetes inpatients.

Materials and Methods
This study was a secondary analysis of data collected from the Foot disease in inpatients study, a large multisite observational point prevalence study with the aim of investigating active foot disease and other foot-related conditions in a representative inpatient population [1,2]. Previous papers from the Foot disease in inpatients study have investigated and reported on the outcomes of (i) primary and secondary admissions for any foot-related conditions [1] and (ii) the presence of active foot disease (ulcers, infection, and ischaemia) [2]. Thus, the design and methodology of the Foot disease in inpatients study that form the basis of this paper have been described in detail elsewhere [1,2]. This paper now aims to investigate and report on the outcomes of the presence of different foot complications (previous amputations, previous foot ulcers, PAD, PN, and foot deformity).
In brief, the Foot disease in inpatients study investigated all adult inpatients present (in hospital at the time of the study for any medical reason) in five representative public hospitals in Queensland (considered representative of the five different categories of Australian hospitals) on one designated day and they were all invited to participate on that same day, excluding those with a cognitive deficit and those in a maternity or psychiatric ward [1]. Of a total of 1146 inpatients present on those days, 883 were eligible for the study and 733 consented to participate [1]. Those eligible participants who consented reported no age or sex differences to those who did not consent [1]. This sample of 733 inpatients has been reported to be highly reflective of the demographic, social determinant, medical history, and reason for admission characteristics reported elsewhere for representative inpatient populations present in Australia and in other developed nations [1,2,10]. Highly trained and tested data collectors collected all data on the 733 participants by surveying each participant to determine their selfreported medical history and physically assessing their feet to clinically diagnose any foot-related conditions [1,11]. All data were captured on a validated data collection instrument (the Queensland Foot Disease Form) [1,2].
The explanatory variables for this study were grouped into the domains of participant demographics (age and sex), social determinants (socioeconomic status, geographical remoteness, education levels, country of birth, and indigenous status), medical condition history (diabetes, hypertension, dyslipidaemia, myocardial infarct, cerebrovascular accident, chronic kidney disease, smoking, cancer, arthritis, depression, and acute foot trauma), self-care ability (mobility impairment, vision impairment, and main footwear worn inside and outside the home), and past foot treatment in the year prior to hospitalisation (by podiatrist, general practitioner, specialist physician, surgeon, nurse, orthotist, and other) [1,2]. All variables have been defined in detail elsewhere [1,2,11].
2.1. Statistical Analysis. All data were analysed using SPSS 22.0 for Windows (SPSS Inc., Chicago, IL, USA) or Graph-Pad Software. Descriptive statistics were used to display all variables. Prevalence with 95% confidence intervals (95% CI) was evaluated for all foot complication outcome variables. Pearson's chi-square tests and Student's t-tests were used to test for differences in categorical variables (proportions) and continuous variables (mean (standard deviation)), respectively. Mann-Whitney U tests were also used to test for differences in continuous variables (median (interquartile ranges)) if they were identified not to be normally distributed using Kolmogorov-Smirnov tests. Univariate logistic regression was used to test for crude associations between all explanatory variables and each foot complication outcome (p < 0 05). All explanatory variables achieving a statistical significance of p < 0 2, except those deemed illogical to be potentially on a causal pathway, were included in backwards stepwise multivariate logistic regression analysis for that foot complication outcome until only variables reaching statistical significance remained (p < 0 05) (unadjusted model) [1,18,19]. All omitted variables from the unadjusted models were reentered and retained in the models as confounders if the beta effect estimates of any unadjusted independent explanatory variable changed by >20% (adjusted model) [1,18,19]. Missing data were treated by excluding cases with any missing data in all models as the proportion of missing data cases was minimal (<5% in most cases) [1,18,19]. Table 1 displays the prevalence proportions (95% CI) for all foot complications in all participants (n = 733), including diabetes participants (n = 172) and nondiabetes participants (n = 561). Overall, 336 participants (46.0% (95% CI: 42.4-49.7%)) had at least one foot complication, including 175 (23.9% (20.9-27.1%)) with multiple foot complications and 81 (11.1% (9.0-13.6%)) with a history of previous foot disease (previous amputation or foot ulceration). Diabetes participants had fewer numbers with a foot complication (n = 112) than nondiabetes participants (n = 224). Yet diabetes participants had much higher proportions of foot complications (65.5%) than nondiabetes participants (40.1%), including those with multiple foot complications (38.6% diabetes, 19.5% nondiabetes) and previous foot disease (22.1% diabetes, 7.7% nondiabetes) (all p < 0 001).

Discussion
This appears to be the first study to investigate a representative inpatient population for foot complications. Our findings indicate nearly half (46%) of all inpatients had at least one foot complication that places them at risk of developing active foot disease, including nearly a quarter (24%) at higher risk with multiple foot complications and a tenth (11%) at very high risk of developing active foot disease with a history of previous foot disease. Inpatients with diabetes had significantly higher proportions of all foot complications than those without diabetes; however, interestingly, there were more patients with foot complications that did not have diabetes than did have diabetes due to the greater overall proportion of inpatients without diabetes. Foot complications in inpatients were associated with older age, males, indigenous peoples, diabetes, cerebrovascular accident (CVA) history, mobility impairment, other foot complications, and past foot treatment. Overall, these findings suggest that foot complications are relatively common in inpatient populations and also have common factors independently associated with them in both diabetes and nondiabetes inpatients.
To the best of our knowledge, the only foot complication to have been previously investigated in a representative inpatient population was PAD [3,4]. Our 21% prevalence for PAD seemed low compared to the 29% and 36% reported in the two previous similar studies [3,20,21]; however, an interrogation of these studies suggests closer alignment. The previous studies used an ankle-brachial index to diagnose PAD in inpatients over 40 years, whereas our study used toe systolic pressures to diagnose PAD in inpatients over 18 years [20,21]. Our equivalent PAD prevalence for our subgroup of inpatients over 40 years of age was 24% (149/623), and toe pressures have been found to decrease false positive PAD identification compared with the ankle brachial indices [13,15,16]. These methodological differences appear to explain the differences in our PAD prevalence findings compared to previous findings and suggest our findings are generalisable. We also found that diabetes patients (35%) had much higher proportions of PAD than nondiabetes patients (17%) which is also consistent with previous literature [20,21]. Although previous amputation, previous foot ulceration, PN, and foot deformity have not been previously investigated in representative inpatient populations, our findings are generally consistent to those reported for diabetes and geriatric inpatient populations [3,4]. Previous amputation prevalence reported by other studies were 1-8% within diabetes inpatients [4,22,23] and 0-7% within over 60 years old [4,7,24], which were similar to our findings of 9% and 6% (24/433), respectively. Previous foot ulcer prevalence reported by other studies was 12-20% within diabetes inpatients [4,[25][26][27] and 1-15% within over 60 years old [4,6,24] which again were similar to our findings of 20% and 11% (48/433), respectively. PN prevalence reported by other studies was 12-81% within diabetes inpatients [4,9,28] and 26% within over 60 years old [5,6], which were again similar to our findings of 43% and 29% (124/433), respectively. Although the 43-50% foot deformity prevalence has only been previously reported in geriatric inpatients [4,7,29], and was much higher than our geriatric finding of 31% (133/433), this was likely explained by the different foot deformity definitions used between our study and the previous studies. Our study required at least three clinical characteristics of a foot deformity to be present to be defined as a foot deformity [13,17], whereas previous studies required a much lower threshold of diagnosis with just one-foot deformity characteristic being required [7,29]. The overall interpretation of our diabetes and geriatric specific inpatient findings reassures us that our foot complication prevalence findings in representative inpatient populations are plausible and generalisable.
There has also been a general lack of literature investigating independent factors associated with foot complications in representative populations (inpatient or outpatients with and without diabetes). Yet interestingly, our representative inpatient findings for factors associated with foot complications were very similar to previously reported diabetes outpatient findings, even after we adjusted for diabetes. We found previous amputation was most strongly associated with previous foot ulcers which have been consistently identified in the diabetes literature to be the major precipitating risk factor for amputation [12-14, 16, 17, 30]. Other factors identified in our study were foot deformity, CVA history, and past foot treatment by a surgeon. The association with foot deformity is most likely explained by a minor amputation procedure often producing a foot deformity in the remaining partial foot and a major amputation often precipitating a compensatory foot deformity in the remaining contralateral foot [13,17]. A CVA history has also been identified in other recent studies [30,31] and may be explained by the similar macrovascular pathophysiology that occurs in both CVA and PAD which can subsequently result in amputation [30,31]. Lastly, past foot treatment by a surgeon in the previous year is perhaps not surprising considering amputation procedures can only be performed by surgeons;    however, we only captured past foot treatment for the year prior to hospitalisation and did not record the duration since the previous amputation was performed. Thus, further research would be required to determine how long people with a previous amputation maintain ongoing foot treatment with their surgeon after their procedure.
The independent factors we identified to be associated with previous foot ulceration were also consistent with those reported in the diabetes outpatient literature [12-14, 17, 32].
Our findings indicate that PAD, PN, and past foot treatment factors are important factors associated with foot ulceration, regardless of diabetes status [13,17,32]. Furthermore, a recent study also identified that PAD and PN were independently associated with active foot ulcers, regardless of diabetes; however, that study identified an association with past foot treatment by a surgeon rather than a podiatrist or nurse [2]. This adds weight to previous recommendations that best practice guidelines must emphasise the need for a podiatrist and nurse, in conjunction with a surgeon, to be part of the recommended outpatient multidisciplinary foot team to prevent foot ulcer inpatient admissions in diabetes and nondiabetes patients, rather than wait until the foot ulcer has healed to seek podiatry and nursing foot treatment as our findings suggest happens [13,33,34]. Nevertheless, these findings suggest, in both diabetes and nondiabetes populations, that these foot complications significantly increase  the risk of developing future active foot disease (ulcers, infection, or ischaemia) which in turn increases the risk of potential hospitalisation and amputation. PAD in our study was independently associated with older age, male gender, and PN, which is consistent with previous outpatient literature [20,21,35]. Furthermore, our study identified Australian indigenous peoples as an independent factor associated with PAD which has also been identified by two previous diabetes-related Australian studies [36,37]. The independent associated factor of past surgical treatment for PAD is a welcome finding and suggests patients with PAD are being regularly assessed by vascular surgeons as recommended by best practice guidelines [13,15,16], whereas our finding that a cancer history decreased the likelihood of having PAD is potentially a novel finding. It is perhaps most likely explained by cancer sufferers being hospitalised at younger ages [10], but more likely because our definition of cancer history was broad and included any cancers in the participant's history. Thus, it is recommended that any further research into this association captures data on different cancer types, severity, durations, and treatments to determine if the association is with particular types of cancers or treatments.
PN was found to be independently associated with older age, diabetes, and mobility impairment in our study, again all of which have been reported in the diabetes outpatient literature [32,38]. In contrast to best practice guideline recommendations that people with PN should receive regular foot monitoring by a podiatrist to prevent active foot disease, falls, or pressure injuries [13,33], our study found no past foot treatment variables were independently associated with PN. PN has been reported to be the most important foot complication that precipitates the development of active foot disease, falls, and pressure injuries [3,[5][6][7][8]. Thus, our findings that one in every five inpatients, including nearly one in every two diabetes inpatients, has PN and is unlikely to have received any past foot treatment are concerning and highlight that further strategies are necessary to identify and monitor these patients in both the inpatient and outpatient settings [13,33,34]. The independent factors for foot deformity of older age, PN, mobility impairment, and past podiatry treatment identified in our inpatient study are also consistent with similar outpatient literature [17,[38][39][40]. The association between foot deformity and PN has been consistently identified in the diabetes literature, and our findings suggest this link may be of similar importance in nondiabetes patients and should be investigated further in the future [17,[38][39][40].
Our overall findings suggest that foot complications that have been commonly reported to precipitate the development of active foot disease in the community are also present frequently in the inpatient population and have common factors independently associated with them, regardless of diabetes. Further research is recommended to more precisely determine the causal relationships for foot complications in nondiabetes populations in particular. It is recommended that policy makers and clinicians adopt simple hospital triage procedures that identify inpatients with these foot complications early to ensure that they do not develop into future active foot disease, falls, or pressure injuries whilst in hospital [1,2,33,34]. These procedures could be as simple as questioning all inpatients on admission, particularly those with diabetes or over 60 years of age, as to their previous foot disease history or using simple screening tools to identify PAD, PN, and foot deformity [1,2,13,34]. Nevertheless, clinicians and policy makers should continue to recommend inpatients identified with foot complications be assessed and managed whilst in an hospital and discharged to an outpatient multidisciplinary foot team for ongoing management in order to prevent potential hospitalisation from active foot disease, falls, and pressure injuries in the future [1,2,33,34].
Additionally, although the independent associations between past foot treatment and most foot complications appear encouraging in our findings, this was not the case for PN. It could be argued that PN is the most critical foot complication that leads to active foot disease, falls, and pressure injuries, and thus, best practice guidelines need to better highlight that patients with PN require ongoing monitoring to ensure they can identify problems early to prevent possible future hospitalisation [13,17,[32][33][34]. Lastly, it is recommended that policy makers incorporate these foot complications into their existing inpatient bedside audit programs alongside general diabetes, falls, and pressure injury bedside audit programs [2,10,26]. This should enable efficient monitoring of the influence of these foot complications on inpatient populations in the future and particularly their impact on inpatient adverse events [1,2].

Strengths and Limitations
. This study has a number of strengths and limitations which have been discussed elsewhere [1,2]. In brief, the strengths were this study investigated a highly representative Australian inpatient population [1,2,10]; data collectors were highly experienced, trained, and reliable in collecting validated and internationally agreed definitions of clinically diagnosed foot complications [11,14]; and multivariate logistic regression models were used to adjust for confounding variables [1,18,19]. The limitations were this study was a cross-sectional study and was unable to test for causal relationships; excluded a large number of older cognitively impaired patients which may have led to a more conservative prevalence estimate of foot complications; had to aggregate minor and major previous amputations which are arguably the result of different causal pathways due to small numbers of both; did not exclude patients with active foot disease; and was a secondary analysis of a large dataset [1,2] which increases the likelihood of type 1 statistical error [18,19].

Conclusions
This study was the first to investigate multiple foot complications in a representative inpatient population. It identified that half of all inpatients had at least one-foot complication, with a quarter having multiple foot complications, which have been reported to be risk factors for the development active foot disease, pressure injuries, or falls whilst in hospital. The findings of this study suggest that regardless of having diabetes or not, common factors precipitate these foot complications. It is recommended that all inpatients are screened for these common foot complications on admission, particularly those with diabetes, and are managed accordingly to potentially prevent the large burden that foot disease already imposes on inpatient and outpatient populations.