Browse the corpus

Diabetes mellitus and perioperative outcomes: a scoping review of the literature. BACKGROUND: Diabetes mellitus (DM) is frequently encountered in the perioperative period. DM may increase the risk of adverse perioperative outcomes owing to the potential vascular complications of DM. We conducted a scoping review to examine the association between DM and adverse perioperative outcomes. METHODS: A systematic search strategy of the published literature was built and applied in multiple databases. Observational studies examining the association between DM and adverse perioperative outcomes were included. Abstract screening determined full texts suitable for inclusion. Core information was extracted from each of the included studies including study design, definition of DM, type of DM, surgical specialties, and outcomes. Only primary outcomes are reported in this review. RESULTS: The search strategy identified 2363 records. Of those, 61 were included and 28 were excluded with justification. DM was mostly defined by either haemoglobin A1c (HbA1c) or blood glucose values (19 studies each). Other definitions included 'prior diagnosis' or use of medication. In 17 studies the definition was unclear. Type 2 DM was the most frequently studied subtype. Five of seven studies found DM was associated with mortality, 5/13 reported an association with 'complications' (as a composite measure), and 12/17 studies found DM was associated with 'infection'. Overall, 33/61 studies reported that DM was associated with the primary outcome measure. CONCLUSION: Diabetes mellitus is inconsistently defined in the published literature, which limits the potential for pooled analysis. Further research is necessary to determine which cohort of patients with DM are most at risk of adverse postoperative outcomes, and how control influences this association.

This scoping review was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, extended for use with scoping reviews.11 The protocol was developed before commencing the review and is available upon request from the corresponding author. An initial scoping search was performed using PubMed to collate relevant keywords and medical subject headings (MeSH). These were collated into a systematic search strategy combining free text and Boolean logic terms. The search was applied in CINAHL, the Cochrane library, MEDLINE, SCOPUS, and Web of Science. An example search strategy can be found in Appendix 1. The initial search strategy was developed with the assistance of an experienced Information Scientist (NK) in accordance with best practice guidelines.11 Reference lists of included studies and relevant reviews were also searched to supplement the systematic search. Where full texts were not available, we contacted the authors, which was successful in one case. The review question was specifically designed to address the epidemiology of outcomes for patients with DM and therefore studies investigating an intervention, such as RCTs, were excluded. Only manuscripts covering health-related outcomes, including patient-reported outcomes, were included. It was a prerequisite that papers included patients with and without DM undergoing elective, noncardiac surgery. We further limited the search to adult patients as the epidemiology of DM differs significantly between adult and paediatric patients.

th-related outcomes, including patient-reported outcomes, were included. It was a prerequisite that papers included patients with and without DM undergoing elective, noncardiac surgery. We further limited the search to adult patients as the epidemiology of DM differs significantly between adult and paediatric patients. Duplicate references were removed using EndNote (EndNote, Clarivate Analytics). Manuscripts not available in English were also excluded at this stage. After removal of duplicates, studies were uploaded onto Rayyan (Rayyan Systems Inc. Online Software; available from: https://www.rayyan.ai/).12 As part of a consistent and comprehensive screening process two authors (DD and RB) screened all titles and abstracts independently to identify relevant studies for full text review. Rayyan collates a list of disagreements between the two authors, which were then examined by a third author (MA) to determine final inclusion.

rt of a consistent and comprehensive screening process two authors (DD and RB) screened all titles and abstracts independently to identify relevant studies for full text review. Rayyan collates a list of disagreements between the two authors, which were then examined by a third author (MA) to determine final inclusion. The full texts identified for review were re-imported to a reference management software (EndNote, Clarivate Analytics). A screening and data collection tool was created a priori and tested (DD and RB) on the first 10 papers to assess suitability. All full texts were re-assessed against the key inclusion criteria and then relevant data was extracted using Microsoft Excel (Microsoft Corp., Redmond, WA, USA). Three authors were involved in data extraction (DD, RB, and CDF). Consensus was sought between the three extracting authors in cases of uncertainty. We recorded article characteristics such as design, statistical methods, Definition of DM, type of DM, participant counts, surgery types, and outcomes. Many studies looked at multiple outcomes. When it was not clear which was the primary outcome, we selected the outcome most applicable across different surgical specialities (i.e. mortality).

racteristics such as design, statistical methods, Definition of DM, type of DM, participant counts, surgery types, and outcomes. Many studies looked at multiple outcomes. When it was not clear which was the primary outcome, we selected the outcome most applicable across different surgical specialities (i.e. mortality). Microsoft Excel was used to synthesise extracted data. We grouped studies based on the Definition of DM they used and the outcomes they studied, and value cut-offs for descriptors of glycaemic status. Frequencies were produced for the definitions studied, and we visually displayed these using a bubble plot. All included texts were synthesised into a single table to compare the outcomes studied.

An initial scoping search was performed using PubMed to collate relevant keywords and medical subject headings (MeSH). These were collated into a systematic search strategy combining free text and Boolean logic terms. The search was applied in CINAHL, the Cochrane library, MEDLINE, SCOPUS, and Web of Science. An example search strategy can be found in Appendix 1. The initial search strategy was developed with the assistance of an experienced Information Scientist (NK) in accordance with best practice guidelines.11 Reference lists of included studies and relevant reviews were also searched to supplement the systematic search. Where full texts were not available, we contacted the authors, which was successful in one case.

The review question was specifically designed to address the epidemiology of outcomes for patients with DM and therefore studies investigating an intervention, such as RCTs, were excluded. Only manuscripts covering health-related outcomes, including patient-reported outcomes, were included. It was a prerequisite that papers included patients with and without DM undergoing elective, noncardiac surgery. We further limited the search to adult patients as the epidemiology of DM differs significantly between adult and paediatric patients.

Duplicate references were removed using EndNote (EndNote, Clarivate Analytics). Manuscripts not available in English were also excluded at this stage. After removal of duplicates, studies were uploaded onto Rayyan (Rayyan Systems Inc. Online Software; available from: https://www.rayyan.ai/).12 As part of a consistent and comprehensive screening process two authors (DD and RB) screened all titles and abstracts independently to identify relevant studies for full text review. Rayyan collates a list of disagreements between the two authors, which were then examined by a third author (MA) to determine final inclusion.

The full texts identified for review were re-imported to a reference management software (EndNote, Clarivate Analytics). A screening and data collection tool was created a priori and tested (DD and RB) on the first 10 papers to assess suitability. All full texts were re-assessed against the key inclusion criteria and then relevant data was extracted using Microsoft Excel (Microsoft Corp., Redmond, WA, USA). Three authors were involved in data extraction (DD, RB, and CDF). Consensus was sought between the three extracting authors in cases of uncertainty. We recorded article characteristics such as design, statistical methods, Definition of DM, type of DM, participant counts, surgery types, and outcomes. Many studies looked at multiple outcomes. When it was not clear which was the primary outcome, we selected the outcome most applicable across different surgical specialities (i.e. mortality).

Microsoft Excel was used to synthesise extracted data. We grouped studies based on the Definition of DM they used and the outcomes they studied, and value cut-offs for descriptors of glycaemic status. Frequencies were produced for the definitions studied, and we visually displayed these using a bubble plot. All included texts were synthesised into a single table to compare the outcomes studied.

The systematic search produced 2363 records. After duplications had been removed, 1714 title and abstracts were screened and 1625 excluded. The full text of 89 articles was assessed for eligibility. A total of 61 papers were included13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73 based on the predetermined criteria (described above). The remaining 28 articles were excluded with the reasons summarised in Fig 1. Table 1 summarises the included papers. Publication date ranged from 1984 to 2020. All studies used observational designs; most were retrospective designs with 12 studies utilising prospective methodologies. The range of patients with DM studied was from 8 to 423 050. The range of control patients was 18–2 145 944. The surgical specialties represented included: dental, spinal, vascular, ophthalmic, orthopaedic, urology, gynaecological, general, head and neck, and maxillo-facial.Fig 1PRISMA flowchart demonstrating the full scoping review process from initial search to abstract screening and full text assessment. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.Fig 1Table 1Summary of all papers included in the review. Data are grouped according to their primary outcome measure. ∗Not reported.

monstrating the full scoping review process from initial search to abstract screening and full text assessment. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.Fig 1Table 1Summary of all papers included in the review. Data are grouped according to their primary outcome measure. ∗Not reported. AVf, arterio-venous fistula; CI, confidence interval; DM, diabetes mellitus; HD, haemodialysis; HR, hazard ratio; IGT, impaired glucose tolerance; IOP, intraocular pressure; IPTW, inverse probability of treatment weighting; JOA, Japanese Orthopaedic Association (Changes in motor, sensory and bladder function); LOS, length of stay; OR, odds ratio; PJI, prosthetic joint infection; QuickDASH, Quick Disabilities of the Arm, Shoulder and Hand (patient reported outcome measure assessing disability); RR, relative risk; sd, standard deviation; SSI, surgical site infection; UTI, urinary tract infection; WOMAC, Western Ontario and McMaster Universities Arthritis Index.Table 1YearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (mortality)Reported differences between groupsDM associated with outcome2014Guzman64Retrospective cohort423 0502 145 944SpinalMortalityOR=1.44; 95% CI, 1.19–1.74; P=0.0001Yes2014Guzman63Retrospective cohort223 9081 378 237SpinalMortalityOR=2.08; 95% CI, 1.72–2.50; P<0.0001Yes2009Marchant62Retrospective cohort109 458920 555OrthopaedicMortalityControlled DM vs no diabetes: OR=0.855; 95% CI, 0.679–1.076; P=0.182Uncontrolled diabetes vs no diabetes OR=2.700; 95% CI, 1.647–4.426; P<0.001Yes2015Lee54Retrospective case-control4192656UrologyAll-cause mortalityOR=1.825; P=0.001Yes2019Zarrouk66Retrospective cohort3971709VascularMortalityIPTW adjusted Cox regression (RR=0.98; CI, 0.75–1.29; P=0.91)No2019Long44Retrospective cohort261790Vascular30-day mortalityDM: 2.5%Control (glucose >180 mg dl−1): 8.5% (P=0.02)No2016Hjellestad17Prospective cohort study858VascularAll-cause mortalityMultivariate Cox regressionHR death=6.35; 95% CI, 1.49–27.1; P=0.01YesYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (Composite measure of morbidity and mortality)Reported differences between groupsDM associated with outcome2019Wysocki51Retrospective cohort3431375GeneralOverall morbidity rateDM: 7.27%Control: 5.58%Pre-diabetes: 6.64%; P=0.571No2019Guetta45Retrospective cohort143841GeneralMild complication (Clavien–Dindo classification <3a)OR=2.32; 95% CI, 1.16–4.6; P=0.017Yes

ty)Reported differences between groupsDM associated with outcome2019Wysocki51Retrospective cohort3431375GeneralOverall morbidity rateDM: 7.27%Control: 5.58%Pre-diabetes: 6.64%; P=0.571No2019Guetta45Retrospective cohort143841GeneralMild complication (Clavien–Dindo classification <3a)OR=2.32; 95% CI, 1.16–4.6; P=0.017Yes 2015Reategui43Retrospective case series130703OrthopaedicMedical, infectious, mechanical, and surgical complications–∗No2015Goodenough72Prospective cohort129888GeneralMajor complication using (Clavien–Dindo classification system) within 30 days of surgeryOR=1.17; 95% CI, 0.57–2.41; P=0.66No2019Law23Retrospective cohort104104OrthopaedicComplication rateDM: 5.8%Control:4.8%No2015Kallio69Retrospective cohort study103100OrthopaedicComplication rateDM (A1c <10%) + referral: 0.78 (1.01)DM no referral: 1.27 (1.18)No DM: 0.36 (0.63) (P=0.124)DM (A1c <8%) + referral: 0.50 (0.89) (P=1)Yes2016Swirska21Retrospective cohort9191GynaecologicalNumber of perioperative complications (e.g. UTI, impaired wound healing)OR=1.83; 95% CI, 0.68–4.96; P=0.24No2006Hofmann42Prospective cohort80544VascularPeriprocedural complications (fatal and non-fatal stroke, non-fatal myocardial infarction)Inadequate control (HbA1c >7%) OR=3.7; 95% CI, 1.5–9.1; P=0.005Yes2012Myers68Retrospective cohort7474OrthopaedicAny complication (infection, non-infection [e.g.

5% CI, 0.68–4.96; P=0.24No2006Hofmann42Prospective cohort80544VascularPeriprocedural complications (fatal and non-fatal stroke, non-fatal myocardial infarction)Inadequate control (HbA1c >7%) OR=3.7; 95% CI, 1.5–9.1; P=0.005Yes2012Myers68Retrospective cohort7474OrthopaedicAny complication (infection, non-infection [e.g. non-union])OR=2.9; 95% CI, 1.42–5.96; P<0.005Yes2018Kamarajah27Prospective cohort study49132GeneralOverall complications (Clavien–Dindo)Multivariate logistic regression: OR=2.08; 95% CI, 1.04–3.99; P=0.031Yes2020Law67Retrospective cohort4080OrthopaedicOverall complication rate (infection, reoperation, non-union)DM: 17.5%Control: 23.8% (P=0.489)No2019Rudolph24Retrospective cohort39112GeneralMajor complicationsDM: 53% (P=0.514)No2016Bianchini28Retrospective cohort31137Head and neckPostoperative complicationsMultivariate logistic regression: OR=1.042; 95% CI, 0.416–2.607; P=0.930NoYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (Infections)Reported differences between groupsDM associated with outcome2018Cancienne61Retrospective cohort13 470103 586OrthopaedicDeep infection within 6 months requiring debridementDM: 0.33%Control: 0.19% (P=0.001)Yes2019Lipsky58Retrospective cohort447810 491UrologyInflatable penile prosthesis infectionHR=1.32; 95% CI, 1.05–1.66; P=0.016Yes2013Kwon56Retrospective cohort40987532GeneralComposite infectionsNon-insulin DM: OR=0.51; 95% CI, 0.37–0.69Insulin DM: OR=0.52; 95% CI, 0.35–0.76No2015Maradit Kremers59Retrospective cohort350716 664OrthopaedicPJIHR=1.23; 95% CI, 0.87–1.74No2017Hoelzer53Retrospective cohort4522285PainInfection rateDM: 1.99%Control: 2.54% (P=0.49)No2014Wukich52Prospective cohort3231737OrthopaedicSSIOR=3.99 (95% CI, 2.39–6.68)Yes2011Wukich49Prospective cohort2211241OrthopaedicSSI (within 30 days)DM: 9.5%Control: 2.4% (P<0.00)Yes2017Rahimi-Nedjat50Retrospective cohort1201254Maxillo-facialInfectionsDM: 15.0%Control: 12.1% (P=0.383)No2014Fisichella33Retrospective case-control111176OrthopaedicSSIOR=8.7Yes1984Vannini48Retrospective cohort471180OrthopaedicDeep phlagosis (infection)DM: 11%Control: 2% (P<0.001)Yes2020Keavy39Retrospective cohort study43321GynaecologicalInfectionDM: 12.4%Control: 7.5% (P<0.05)Yes2006Liao37Retrospective cohort39298SpinalInfectionDM: 10.3%Control: 0.7% (P=0.003)Yes2014Hikata38Retrospective case-control36309SpinalSSIDM: 16.7%Control: 3.2% (P=0.0005)Yes2016Iavazzo36Prospective cohort34266GynaecologicalInfective complicationsDM: 32.4% (P=0.048)Yes2

ologicalInfectionDM: 12.4%Control: 7.5% (P<0.05)Yes2006Liao37Retrospective cohort39298SpinalInfectionDM: 10.3%Control: 0.7% (P=0.003)Yes2014Hikata38Retrospective case-control36309SpinalSSIDM: 16.7%Control: 3.2% (P=0.0005)Yes2016Iavazzo36Prospective cohort34266GynaecologicalInfective complicationsDM: 32.4% (P=0.048)Yes2 008Olsen70Retrospective nested case-control29199OrthopaedicSSIOR=3.5 (95% CI, 1.2–10.0)Yes2013Motta13Prospective case-control2818DentalClinical complications (surgical site infection, systemic infection)Controlled DM: 7.7%Uncontrolled DM: 13.3%No DM: 5.6%, Fisher's exact test (P=0.81)No2010Ata65Retrospective medical record review–∗–∗General/vascularPostoperative infectionVascular – adjusted OR=1.84 (95% CI, 1.20–2.82)General – adjusted OR=1.80 (95% CI, 1.12–2.90)YesYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (specialty Specific)Reported differences between groupsDM associated with outcome2013Adams60Retrospective cohort756732 924OrthopaedicRevision arthroplastyHbA1c <7%, OR=1.32 (95% CI, 0.99–1.76)HbA1c >7%, OR=1.03 (95% CI, 0.68–1.54)No2013Takahashi26Retrospective cohort41124SpinalJOA scoreDM: 22.7 (sd 5.6)Control: 24.4 (sd 4.2) (P=0.137)No2000Kawaguchi14Retrospective case-control1834SpinalJOA scoreDM: 12.6 (2.0)Control: 13.3 (2.1) (P=0.25)No2012Dokai19Retrospective case series1365SpinalJOA scoreDM: 12.1 (7–16.5)Control: 12.4 (6.5–17) (P=0.578)No2017Brock22Retrospective matched cohort100100OrthopaedicWOMAC scores (pain, stiffness, and physical function)–∗Yes2018Moazzeni16Prospective case-control4848SpinalRate of Fusion at 1 yrDM: 58%Control: 79% (P=0.02)Yes2018Sun29Retrospective case-control11141OrthopaedicNew onset or exacerbation of nerve symptomsDM: 27%Normal glucose tolerance: 9%Impaired regulation: 19% (P=0.112)No2019Singh30Prospective cohort150150OphthalmicEye Complications (transient corneal oedema)–∗No1993Kodama34Retrospective cohort36184OphthalmicOphthalmic Complications (Macular Oedema and Transient elevation of intraocular pressure)IOP; DM: 13%, Control 4%Oedema; DM: 18%, Control 2% (P<0.05)Yes2013Law18Retrospective case-control2964OphthalmicRate of qualified surgical success (IOP <15 and >5 mm Hg, without complications)DM: 61%Control: 64.2% (P=0.881)NoYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (Other including cardiovascular, renal, and LOS)Reported differences between groupsDM associated with outcome2014Underwood46Retrospective

IOP <15 and >5 mm Hg, without complications)DM: 61%Control: 64.2% (P=0.881)NoYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (Other including cardiovascular, renal, and LOS)Reported differences between groupsDM associated with outcome2014Underwood46Retrospective cohort449888General/vascularLOSDM: 6.

IOP <15 and >5 mm Hg, without complications)DM: 61%Control: 64.2% (P=0.881)NoYearFirst authorStudy design# Patients (DM)# Patients (control)Surgical specialtyPrimary outcome measure (Other including cardiovascular, renal, and LOS)Reported differences between groupsDM associated with outcome2014Underwood46Retrospective cohort449888General/vascularLOSDM: 6. (6.6)Control: 5.2 (5.3); P<0.0001Yes2018Lenguerrand40Prospective cohort64523OrthopaedicLOSDM: 5 daysControl: 4 days (P=0.3)No2019Villamiel25Retrospective cross-sectional44113GeneralLOST2DM: 5.8 (sd 3.8)Control: 6.4 (sd 5.1); P=0.476No2013Bakker47Retrospective cohort3291133Vascular30-day cardiovascular complicationsOR=1.80; 95% CI, 1.24–2.61; P<0.01Yes2011Biteker71Retrospective cohort204344MixedPerioperative cardiovascular events (PCEs)DM: 24.5%IGT: 9.8%Control: 5% (P<0.001)Yes2017Shin73Retrospective cohort603448 811SpinalAcute renal failureOR Controlled DM 1.863; 95% CI, 1.35–2.58; P<0.05OR Uncontrolled DM 4.84; 95% CI, 1.75–13.39; P<0.05Yes2008Feringa35Retrospective cohort69220VascularIschaemic EventsOR=2.6 (95% CI, 1.4–4.9)Yes2012Afsar32Retrospective cohort73160VascularFailure of AVF before first HD sessionDM HbA1c >7%: 52.8%DM HbA1c <7%: 29.7%Control: 27.5% (P=0.013)Yes2020Reinstatler20Retrospective cohort9281Urology30-day postoperative visits for pain (ED or clinic)–∗No2011Hwang31Retrospective cohort92159UrologyRecurrence free survival in monthsKaplan–Meier: HR=2.11; 95% CI, 1.4–3.2; P=0.001Yes2020Chung41Retrospective cohort67538UrologyPost-void residual volume at 3 months (ml)DM: 30.6 (41.3)Control: 47.6 (89.4); P=0.306No2018Schroer55Retrospective cohort2376107OrthopaedicMean 90-day charges$5074 increase in costYes2019Zimmerman57Retrospective cohort15039139PlasticQuickDASH scoreDM: 25Control: 27 (P=0.263)No2008Tawil15Prospective case-control4545DentalImplant survival–∗No

ume at 3 months (ml)DM: 30.6 (41.3)Control: 47.6 (89.4); P=0.306No2018Schroer55Retrospective cohort2376107OrthopaedicMean 90-day charges$5074 increase in costYes2019Zimmerman57Retrospective cohort15039139PlasticQuickDASH scoreDM: 25Control: 27 (P=0.263)No2008Tawil15Prospective case-control4545DentalImplant survival–∗No PRISMA flowchart demonstrating the full scoping review process from initial search to abstract screening and full text assessment. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Summary of all papers included in the review. Data are grouped according to their primary outcome measure. ∗Not reported. AVf, arterio-venous fistula; CI, confidence interval; DM, diabetes mellitus; HD, haemodialysis; HR, hazard ratio; IGT, impaired glucose tolerance; IOP, intraocular pressure; IPTW, inverse probability of treatment weighting; JOA, Japanese Orthopaedic Association (Changes in motor, sensory and bladder function); LOS, length of stay; OR, odds ratio; PJI, prosthetic joint infection; QuickDASH, Quick Disabilities of the Arm, Shoulder and Hand (patient reported outcome measure assessing disability); RR, relative risk; sd, standard deviation; SSI, surgical site infection; UTI, urinary tract infection; WOMAC, Western Ontario and McMaster Universities Arthritis Index.

; PJI, prosthetic joint infection; QuickDASH, Quick Disabilities of the Arm, Shoulder and Hand (patient reported outcome measure assessing disability); RR, relative risk; sd, standard deviation; SSI, surgical site infection; UTI, urinary tract infection; WOMAC, Western Ontario and McMaster Universities Arthritis Index. The definition of DM used varied substantially as illustrated in Fig 2. HbA1c was used in 19 studies. A HbA1c of >6.5% was the most common cut-off applied in five studies. Notably, in one study HbA1c values within the preceding 1–2 yr before study were accepted as diagnostic.45 In the 19 studies reporting blood glucose, a range of diagnostic methods were reported including random, fasting, and glucose tolerance tests. In one study, diabetes mellitus was self-reported as part of the functional co-morbidity index, which was later corroborated against participants' medications. They found all patients had correctly reported their status but were unable to distinguish between type 1 and type 2 DM.40 In 17 studies the definition was unclear, or not reported in the manuscript. In six cases, DM was defined by use of hypoglycaemic agents. Three studies specifically referenced international guidelines (WHO and American Diabetes Association).Fig 2Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.Fig 2

HO and American Diabetes Association).Fig 2Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.Fig 2 Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.

HO and American Diabetes Association).Fig 2Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.Fig 2 Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c. It was not always clear which type of DM was being studied. Forty studies specified DM, but not which type. Five studies used ‘Type 1 and Type 2’ to classify DM. In nine studies patients with type 1 DM were excluded, and only type 2 DM was studied. In one study the registry from which patients were identified contained 98% patients with type 2 DM but the analysis was done using presence vs absence of DM.66 Fourteen studies subclassified DM by control.13, 15, 32, 39, 40, 42, 43, 52, 55, 62, 63, 64, 72, 73 Of these, 11 used HbA1c to define control. Cut-offs included 6.5%, 7%, 8%, 7–9%, 8–9%, and 47 mmol mol−1. The remaining three used International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes which are linked with complications (such as ophthalmic manifestations).62, 63, 64 Other ways of subcategorising patients included by management. For example, insulin-vs non-insulin-dependent DM was found in four studies. Overall, there was no consistency in the way DM or control of DM was defined.

ion (ICD-9-CM) codes which are linked with complications (such as ophthalmic manifestations).62, 63, 64 Other ways of subcategorising patients included by management. For example, insulin-vs non-insulin-dependent DM was found in four studies. Overall, there was no consistency in the way DM or control of DM was defined. Table 1 outlines the primary outcomes studied. Some studies used outcomes that would be applicable to all surgical specialties (i.e. mortality and length of stay), whereas others used outcomes specific to a surgical speciality (e.g. Japanese Orthopaedic Association [JOA] scores and need for revision arthroplasty). Seven studies analysed mortality and found that DM was associated in five cases. Mortality was analysed within different timescales including: inpatient, 30-day, and longer term (up to 8 yr).44, 54, 62, 63, 64, 66 In one study mortality was included as part of a composite measure (‘adverse postoperative outcomes’).73 In 12 studies perioperative complications were analysed as a group, as a composite measure. Patients with DM experienced higher rates of complications in five of these studies. Infection was analysed in 17 studies. Like mortality, infective outcomes were not consistently defined. Definitions of infection included surgical site infection, unrelated infection (such as urinary tract, or pulmonary), and operation specific infection (prothesis infection). Infection was frequently included in composite outcomes. Where it was the primary outcome, DM was associated with postoperative infection in 12 cases.

nitions of infection included surgical site infection, unrelated infection (such as urinary tract, or pulmonary), and operation specific infection (prothesis infection). Infection was frequently included in composite outcomes. Where it was the primary outcome, DM was associated with postoperative infection in 12 cases. Of the 60 papers included in this scoping review, 33 reported that DM was associated with their primary outcome measure. Often multiple outcome measures were studied; this meant that DM may have been associated with one outcome, but not another in the same study.62,63

The definition of DM used varied substantially as illustrated in Fig 2. HbA1c was used in 19 studies. A HbA1c of >6.5% was the most common cut-off applied in five studies. Notably, in one study HbA1c values within the preceding 1–2 yr before study were accepted as diagnostic.45 In the 19 studies reporting blood glucose, a range of diagnostic methods were reported including random, fasting, and glucose tolerance tests. In one study, diabetes mellitus was self-reported as part of the functional co-morbidity index, which was later corroborated against participants' medications. They found all patients had correctly reported their status but were unable to distinguish between type 1 and type 2 DM.40 In 17 studies the definition was unclear, or not reported in the manuscript. In six cases, DM was defined by use of hypoglycaemic agents. Three studies specifically referenced international guidelines (WHO and American Diabetes Association).Fig 2Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.Fig 2 Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.

The definition of DM used varied substantially as illustrated in Fig 2. HbA1c was used in 19 studies. A HbA1c of >6.5% was the most common cut-off applied in five studies. Notably, in one study HbA1c values within the preceding 1–2 yr before study were accepted as diagnostic.45 In the 19 studies reporting blood glucose, a range of diagnostic methods were reported including random, fasting, and glucose tolerance tests. In one study, diabetes mellitus was self-reported as part of the functional co-morbidity index, which was later corroborated against participants' medications. They found all patients had correctly reported their status but were unable to distinguish between type 1 and type 2 DM.40 In 17 studies the definition was unclear, or not reported in the manuscript. In six cases, DM was defined by use of hypoglycaemic agents. Three studies specifically referenced international guidelines (WHO and American Diabetes Association).Fig 2Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c.Fig 2 Bubble plot of diabetes mellitus definitions. Bubble plot depicting the range of definitions used in the published literature studying the association between diabetes mellitus and adverse postoperative outcomes. ∗OGTT, oral glucose tolerance test. HbA1c, haemoglobin A1c. It was not always clear which type of DM was being studied. Forty studies specified DM, but not which type. Five studies used ‘Type 1 and Type 2’ to classify DM. In nine studies patients with type 1 DM were excluded, and only type 2 DM was studied. In one study the registry from which patients were identified contained 98% patients with type 2 DM but the analysis was done using presence vs absence of DM.66 Fourteen studies subclassified DM by control.13, 15, 32, 39, 40, 42, 43, 52, 55, 62, 63, 64, 72, 73 Of these, 11 used HbA1c to define control. Cut-offs included 6.5%, 7%, 8%, 7–9%, 8–9%, and 47 mmol mol−1. The remaining three used International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes which are linked with complications (such as ophthalmic manifestations).62, 63, 64 Other ways of subcategorising patients included by management. For example, insulin-vs non-insulin-dependent DM was found in four studies. Overall, there was no consistency in the way DM or control of DM was defined.

Table 1 outlines the primary outcomes studied. Some studies used outcomes that would be applicable to all surgical specialties (i.e. mortality and length of stay), whereas others used outcomes specific to a surgical speciality (e.g. Japanese Orthopaedic Association [JOA] scores and need for revision arthroplasty). Seven studies analysed mortality and found that DM was associated in five cases. Mortality was analysed within different timescales including: inpatient, 30-day, and longer term (up to 8 yr).44, 54, 62, 63, 64, 66 In one study mortality was included as part of a composite measure (‘adverse postoperative outcomes’).73 In 12 studies perioperative complications were analysed as a group, as a composite measure. Patients with DM experienced higher rates of complications in five of these studies. Infection was analysed in 17 studies. Like mortality, infective outcomes were not consistently defined. Definitions of infection included surgical site infection, unrelated infection (such as urinary tract, or pulmonary), and operation specific infection (prothesis infection). Infection was frequently included in composite outcomes. Where it was the primary outcome, DM was associated with postoperative infection in 12 cases. Of the 60 papers included in this scoping review, 33 reported that DM was associated with their primary outcome measure. Often multiple outcome measures were studied; this meant that DM may have been associated with one outcome, but not another in the same study.62,63

This scoping review is the first of its type to examine the existing literature studying the relationship between DM and adverse perioperative outcomes in the noncardiac surgery literature. Understanding this relationship is important for guiding future research in this area and identifying where targeted interventions will benefit patients most. This is increasingly important as the burden of DM increases among surgical patients.74,75 Through the systematic search of the available literature, we found that DM is defined in multiple ways: using HbA1c values, blood glucose investigations, patient records, and prescribed medications. Furthermore, the cut-offs applied for the Definition of DM, and glycaemic control varied substantially. Such inconsistencies have the potential to undermine the value of pooled analyses.76 It may be possible to perform a systematic review and meta-analysis with the existing literature, but it would require asking a well-formulated question, focusing on just one of the outcome measures such as mortality or infection, which both have studies addressing them with thousands of patients included. However, to take mortality as an example, caution would be necessary as all seven studies from this scoping review used a different definition of DM.

ulated question, focusing on just one of the outcome measures such as mortality or infection, which both have studies addressing them with thousands of patients included. However, to take mortality as an example, caution would be necessary as all seven studies from this scoping review used a different definition of DM. In addition, variable definitions and cut-offs may cause confusion when discussing glycaemic control. This is an important consideration for the perioperative multi-disciplinary team who need to communicate with diabetes specialists, support preoperative optimisation, and decide whether it is appropriate to proceed with an operation.2 The American Diabetes Association have published consensus guidelines advising the use of hyperglycaemia, hypoglycaemia, time in range, and diabetic ketoacidosis as clinically meaningful outcomes measures in type 1 DM, which could be modified for use in type 2 DM.77

ther it is appropriate to proceed with an operation.2 The American Diabetes Association have published consensus guidelines advising the use of hyperglycaemia, hypoglycaemia, time in range, and diabetic ketoacidosis as clinically meaningful outcomes measures in type 1 DM, which could be modified for use in type 2 DM.77 Existing guidelines focus on HbA1c as a marker of glycaemic control.78 HbA1c gives an average of control over 2–3 months and is widely used because of its low costs and reproducibility of measurement. However, literature studying type 1 DM highlights its limitations, specifically its inability to assess short-term glycaemic variability (GV) and its inability to quantify hypoglycaemic burden. Moreover, HbA1c is inaccurate in patients with anaemia or abnormalities in renal function, both of which are common in surgical patients, limiting the value of HbA1c in this setting.77,79 Short-term GV, as measured with continuous glucose monitors, was not studied in any of the papers found in this scoping review. Incorporating short-term measures of GV are an important consideration for future perioperative research with a potential role in aiding preoperative optimisation but also improved care during a hospital stay by enabling patient autonomy and closer monitoring for complications such as hypoglycaemia.79,80 Their utility is currently being explored in the cardiac surgery literature.81 Numerous alternative novel biomarkers for DM diagnosis, control, and complications are currently being explored. Micro-RNAs are currently being studied as potential biomarkers for the early detection of DM and its associated complications. Although scientific and methodological barriers remain before they can be implemented in clinical practice, they show promise and may be relevant to perioperative practice in the next decade.82,83

ro-RNAs are currently being studied as potential biomarkers for the early detection of DM and its associated complications. Although scientific and methodological barriers remain before they can be implemented in clinical practice, they show promise and may be relevant to perioperative practice in the next decade.82,83 On the question of whether DM is a risk factor for adverse postoperative outcome, the answer is patient, operation, and perioperative outcome specific. This scoping review reports primary outcomes, but most papers studied multiple outcomes and many reported associations with some – but not all – outcomes. Interestingly, the four largest studies (>10 000 patients with DM) found an association between DM and their primary outcome measure. For the remaining studies, no trends were seen between study size and likelihood of a detected association. No further trends were noted between factors such as methodology or DM Definition and outcomes.

e four largest studies (>10 000 patients with DM) found an association between DM and their primary outcome measure. For the remaining studies, no trends were seen between study size and likelihood of a detected association. No further trends were noted between factors such as methodology or DM Definition and outcomes. The outcomes found in this scoping review can be classified as either (1) generalisable to the whole surgical population or (2) specific to certain surgical specialties. In terms of outcomes relevant to all surgical patients, such as mortality, the literature reports an association in most cases, but not all. Similar mixed findings were reported for length of stay (LOS), infection, and composite complication measures. For specialty-specific measures, no differences were seen between groups in studies using the JOA score, but significant differences were found for ophthalmic and vascular complications, which is unsurprising given the micro- and macro-vascular complications associated with DM. The need for consistent definitions for seemingly dichotomous variables such as mortality has been discussed elsewhere.76 This review corroborates those observations with studies using various mortality endpoints (in-patient vs 30-day mortality). We support the call for standardised endpoints in observational studies.

The need for consistent definitions for seemingly dichotomous variables such as mortality has been discussed elsewhere.76 This review corroborates those observations with studies using various mortality endpoints (in-patient vs 30-day mortality). We support the call for standardised endpoints in observational studies. It is important to distinguish between definitions of DM control. Many of the studies included in this review specifically referred to glycaemic control (HbA1c), but well-controlled DM refers to more than just a glycaemic marker such as HbA1c. It may include factors such as blood pressure, weight, or lipid status. Three studies had used ICD-9-CM codes to stratify their groups by control, which include reference to microvascular complications.62, 63, 64 A closer examination of the association between presence of DM complications and perioperative outcomes would be of value in future research. This has been explored by our group in colorectal cancer, suggesting that presence of complications is associated with both postoperative mortality (90-day) and death during the surgical episode.84

Through the systematic search of the available literature, we found that DM is defined in multiple ways: using HbA1c values, blood glucose investigations, patient records, and prescribed medications. Furthermore, the cut-offs applied for the Definition of DM, and glycaemic control varied substantially. Such inconsistencies have the potential to undermine the value of pooled analyses.76 It may be possible to perform a systematic review and meta-analysis with the existing literature, but it would require asking a well-formulated question, focusing on just one of the outcome measures such as mortality or infection, which both have studies addressing them with thousands of patients included. However, to take mortality as an example, caution would be necessary as all seven studies from this scoping review used a different definition of DM. In addition, variable definitions and cut-offs may cause confusion when discussing glycaemic control. This is an important consideration for the perioperative multi-disciplinary team who need to communicate with diabetes specialists, support preoperative optimisation, and decide whether it is appropriate to proceed with an operation.2 The American Diabetes Association have published consensus guidelines advising the use of hyperglycaemia, hypoglycaemia, time in range, and diabetic ketoacidosis as clinically meaningful outcomes measures in type 1 DM, which could be modified for use in type 2 DM.77

Existing guidelines focus on HbA1c as a marker of glycaemic control.78 HbA1c gives an average of control over 2–3 months and is widely used because of its low costs and reproducibility of measurement. However, literature studying type 1 DM highlights its limitations, specifically its inability to assess short-term glycaemic variability (GV) and its inability to quantify hypoglycaemic burden. Moreover, HbA1c is inaccurate in patients with anaemia or abnormalities in renal function, both of which are common in surgical patients, limiting the value of HbA1c in this setting.77,79 Short-term GV, as measured with continuous glucose monitors, was not studied in any of the papers found in this scoping review. Incorporating short-term measures of GV are an important consideration for future perioperative research with a potential role in aiding preoperative optimisation but also improved care during a hospital stay by enabling patient autonomy and closer monitoring for complications such as hypoglycaemia.79,80 Their utility is currently being explored in the cardiac surgery literature.81 Numerous alternative novel biomarkers for DM diagnosis, control, and complications are currently being explored. Micro-RNAs are currently being studied as potential biomarkers for the early detection of DM and its associated complications. Although scientific and methodological barriers remain before they can be implemented in clinical practice, they show promise and may be relevant to perioperative practice in the next decade.82,83

On the question of whether DM is a risk factor for adverse postoperative outcome, the answer is patient, operation, and perioperative outcome specific. This scoping review reports primary outcomes, but most papers studied multiple outcomes and many reported associations with some – but not all – outcomes. Interestingly, the four largest studies (>10 000 patients with DM) found an association between DM and their primary outcome measure. For the remaining studies, no trends were seen between study size and likelihood of a detected association. No further trends were noted between factors such as methodology or DM Definition and outcomes.

It is important to distinguish between definitions of DM control. Many of the studies included in this review specifically referred to glycaemic control (HbA1c), but well-controlled DM refers to more than just a glycaemic marker such as HbA1c. It may include factors such as blood pressure, weight, or lipid status. Three studies had used ICD-9-CM codes to stratify their groups by control, which include reference to microvascular complications.62, 63, 64 A closer examination of the association between presence of DM complications and perioperative outcomes would be of value in future research. This has been explored by our group in colorectal cancer, suggesting that presence of complications is associated with both postoperative mortality (90-day) and death during the surgical episode.84

In conclusion, robust observational studies are warranted to further expand our understanding of the relationship of DM to adverse postoperative outcomes. This will be aided by consistent definitions and considering a wider perspective on DM control, including GV and complication status. Defining cohorts of patients with DM who are most at risk will allow implementation of targeted intervention to improve outcomes.

Conceptualisation: DJD, RJB, SH, RA. Methodology: DJD, RJB, SH, RA. Formal analysis: DJD, RJB, CDSF, MA. Investigation: DJD, RJB, CDSF, MA. Writing of original draft: DJD, RJB. Writing, review, and editing: DJD, RJB, CDSF, MA, SH, RA. Visualisation: DJD, RJB. Supervision: SH, RA.