Management of Diabetes Patients across the PeriOperative Pathway: A Systematic Review

Peri-operative environments are a hazardous setting for diabetes patients. A systematic review of literature regarding the management of diabetes patients across the peri-operative pathway has been undertaken to assess if the management of patients within this pathway is suitable and effective for patients. A database search of Google Scholar, CINAHAL, Embase, OVID, Cochrane Library, Joanna Briggs institute and PUBMED was undertaken from 15 th of March 2019 to 30 th of March 2019. A total of 57 papers were found and reduced down to 11 final papers that answered the review question and met the inclusion and exclusion criteria. Inclusion criteria were: Full text, English language, human subjects, adult patients only and studies that focused on diabetes care in a section of the peri-operative pathway. Exclusion criteria: children or adults and children, studies that looked a one particular intervention or type of surgery. No date limit was set. PICO tool was used to frame the study question. 1. Poor patient outcomes; Longer length of stay (LOS); Lack of adherence to guidance and or protocols and glycaemic control. Elective patients had advantageous outcomes compared to emergency surgical patients. Hyperglycaemia still remained a problem with an increase in other medical complications for diabetes patients. LOS in hospital was found to have increased due to medical complications. Adherence to protocols and guidance was found to be beneficial in monitoring and managing hyperglycaemia. However, this review found that best practice guidance and hospital protocol is not always adhered to. A liberal approach to glycaemic control is beneficial. Conclusion This systematic review investigated the management of diabetes patients across the peri-operative pathway. Three main themes emerged from the literature: poor patient outcomes; length of stay; and lack of adherence to guidance and or protocols and glycaemic control. We concluded the perioperative environment is a hazardous setting for a diabetes patients. Elective patients had slightly more advantageous outcomes than emergency patients. Hyperglycaemia still remains a problem which leads to poor patient outcomes and longer LOS. Adherence to protocols and guidance was found to be beneficial in monitoring and managing hyperglycaemia.


Introduction
Poor patient outcomes 8 out of 11 studies reported on the outcomes of patients with diabetes. Studies 2, 3,5,6,8,9,10 and 11 discussed surgical outcomes directly related to diabetes management. McCavert, Monem and Dooher, et al [8] found that best practice of glycaemic control, inline with hospital protocols, saw a 25.4% reduction of peri-operative complications. Overall complications being 29% (out of 69 patients). Elective patients with T2DM were more prone to complications. 5 out of 17 (29.4%) of T2DM elective patients experienced complications; in contrast, only 4 out of 21 (19.0%) of elective patients with T1DM developed a complication such as wound infection or peritonitis. For emergency patients, the rate of complications was slightly higher for those with T1DM (5 out of 14; 35.7%) versus 6 out of 17 patients (35.3%) with T2DM. Complications such as; Wound dehiscence, septicaemia, wound infection, wound infection, confusion, deep vein thrombosis and lower respiratory tract infection were reported as a complication. Frisch, Chandra, Smiley, et al [9] similarly analysed outcomes of mobility contrasting both diabetes and non-diabetes patients. Outcomes such as pneumonia (12.1 vs 5.4%; p=0.001), wound and skin infections (5 vs 2.3%; p<0.001), systematic blood infection (3.6 vs 1.1%; p<0.001), urinary tract infections (4.5vs 1.4%, p<0.001) acute myocardial infarction (2.6 vs 1.2 %; p< 0.001) were reported. Patients who experienced complications had a strong affiliation with high blood glucose levels pre and post-operatively.
Wang, Chen, Li, et al (2019) found that patients over 65-years old, male, high mean post-operative blood glucose (BG), diabetes complications, abnormal kidney function and have underwent general surgery were the highest risk category for poor patient outcomes. The study compared surgery type and patient outcomes. Of the 301 (19.8%) of all patients with diabetes complications, 295, (98.0%) had major vascular complications, 8 (27. %) had diabetes nephropathy, 3 (0.7%) had diabetes retinopathy, 5 (1.7%) had diabetes foot post-operatively. Post-operative adverse events occurred in 118 (7.7%) including 43 (36.4%) delayed extubation caused by surgery-related respiratory failure or muscle weakness. 15 (12.7%) patients had circulatory disorders, 23 (19.5%) had respiratory and circulatory abnormalities. 11 (9.3%) had non-healing of the incision. 15 (12.7%) had infections at other sites. 8 (6.8%) patients with other complications. 3 (2.5%) patients died due to pulmonary embolism and two cases of septic shock. Kotgal, Symods, Hirsch, lrl, et al [12] did not correlate BG management with patient outcomes, but results showed that patients had a greater chance of poorer outcomes with any level of hyperglycaemia versus those who had better diabetes control.
In contrast, Sathya, Davis, Taveria, et al [13] found that stroke, atrial fibrillation and wound infection were the most significant complications from pooled results of 6 studies. Mixed results were noted; 2 pooled results found that the incidence of post-operative stroke was reduced by liberal glycaemic regimes, but pooled results from a further 3 studies suggested that there was no significant difference between the effect of moderate vs strict control on stroke outcomes (odds ratio, 18.5, 95% CI 0.72-4.74, p=0.020). Sathya et al [13] also examined the relationship between atrial fibrillation as a patient outcome and diabetes control. Again, pooled estimates from 2 pooled studies found that moderate versus liberal control had no direct effect on atrial fibrillation as an outcome (Odds ratio 0.54, 95% CI 0.17-1.76, p =0.31). In addition, pooled results from 3 other studies found that there was no significant difference between strict versus moderate control in relation to atrial fibrillation (odds ratio: 0.71, 95% CI0. 39-1.30, p=0.27). Wound infection was also not found to have a significant link to the effects of moderate versus glycaemic control from the results of 2 pooled studies.

Length of stay
LOS was a significant finding in studies 2, 3, 6 and 8. Although not a complication in itself, LOS was linked to or reported alongside poor patient outcomes.
McCavert et al [8] found that Emergency patients had a significantly longer LOS in hospital than the elective groups. Frisch et al (2010) [9] also reports that diabetes patients had a higher rate of complications than non-diabetes counterparts (p=0.105). Patients with diabetes were found to have a greater LOS (and LOS in ICU) than non-diabetes patients. It was also noted that African American patients were not at an increased risk of mortality than other races. No other study compared likelihood of surgical outcomes and race.
Patients with diabetes were also more likely to have greater complications including LOS. Underwood et al, 2014 [11] however, reported that patients with A1C levels >6.5-8% had a similar LOS to the control group. Patients with higher A1C ≤6.5 up to greater than 10% had a significantly longer LOS compared to control subjects. This was the most significant difference of the various A1C groups compared in the study. Higher A1C level was more significant than any other variable such as a diabetes patient's race, gender or type of surgery in relation to LOS. Longer LOS in the hospital was found by Hommel et al [14] to be associated with higher dissatisfaction of patients regarding patient centred-ness in their assessment of results.

Lack of adherence to guidance and or protocols and glycaemic control
The third key theme that emerged from the literature was adherence to guidance, such as hospital protocols and national guidelines and glycaemic control. This theme was disused in studies 1,2,5,7 and 10.
McCavert et al [8] studied both elective and emergency surgical patients. 60% of elective patients with T1DM were not treated according to hospital protocol. Elective patients who were treated according to protocol had a complication rate of 6.3 %. For emergency surgical patients, 7.3% of T1DM patients who were treated as per protocol developed a complication. 12.3% of scheduled blood glucose measurement were not completed. 11.1% of T1DM elective patients did not have their blood glucose checked, and 6.8% of emergency T1DM patients. For T2DM, blood glucose was not checked in 17.4% of elective patients and 12.7% in emergency cases.
Similarly, Coan, Schlinkert, Brandon et al [15] note that capillary BG was taken in 89% of cases in the pre-operative area, and only52% of patients had a HBA1C. Intra-operatively, 33% of patients had a BG check, and the post-operative figure was 87%. 90% of preoperative BG was point of care (POC), and 4% was venous sampling. Intraoperatively, 10% of patients had POC BG values, 16% had POC blood gas sampling. In the PACU, 86% of BG were obtained by POC and 1% was venous. Similarly, Jackson, Patvardhan et al (2015) reported that only 71% of patients had a HBA1C recorded preoperatively and 56% intra-operatively via CBG. 73% of patients had a CBG performed in recovery (PACU) contrary to national guidance. Hommel, Van Gurp, Tack et al's [16] quality indicators suggest that best-practice involved measuring BG 4 hours pre-operatively, every 2 hours intra-operatively, and 1 hour post-operatively. Hommel et al [14] reported that in relation to patient satisfaction and person centeredness, 20% of 362 patients were not informed about intraoperative BG level and its effect. 15% were also not informed that insulin was administered during surgery. This correlated to overall low score from patients' involvement in the survey. Sathya et al [13] report that patients undergoing a liberal target for glycaemic control had significantly better post-operative outcomes (less or no complications) than other groups. No difference with wound infection or atrial fibrillation were found. Bibble (1983) commented from the 3 case studies that protocols for glycaemic control were directed towards managing 'average' diabetes patients rather than complex ones, making guidance non-beneficial.
Future recommendations would be to undertake extensive quantitative and qualitative research across the peri-operative pathway with staff who have direct responsibility for diabetes patients undergoing surgery. The views and attitudes of staff members regarding diabetes management may shed light on the barriers as to why this is still a problem despite being highlighted by several studies seen in this review since 1983. Any further research conducted needs to be influential on practice in order to drive change.

Conclusion
This systematic review examined the management of diabetes patients across the peri-operative pathway. Three main themes emerged: poor patient outcomes; longer length of stay; and lack of adherence to guidance and or protocols and glycaemic control. We concluded the peri-operative environment can be a hazardous setting for diabetes patients. Elective patients had slightly more advantageous outcomes than emergency patients. Hyperglycaemia still remains a problem which leads to poor patient outcomes and longer LOS. Adherence to protocols and guidance was found to be beneficial in monitoring and managing hyperglycaemia. Likert rating scale ranging from 1-9, not relevant -relevant. An answer of cannot assess was made available. An indicator was considered unmeasurable if > 25% of patents individual l indicators score could not be computed because of missing data. Quality indicators were selected according to their relevance , median scores of 7,8 and 9 if there was no disagreement 30%or more ratings in both the 1-3 and 7-9 tertiles.
1100 patients with diabetes who underwent major surgery were identified. Practice test was under taken with the finial 36 quality indicators on a sample of 389 patients.
Clinometric properties for the outcome's indicators Please see table 1 and 2 in study.
Measurability ->25% of scores were missing at he time of data extraction.

Applicability
As hypoglycaemia rarely occurred, the indicator intravenous glucose ordered for pasting patients was inapplicable.
Reliability The indicators blood glucose measurement every 4-6 ours ordered for fasting patients had a kappa of 5.1, which indicted moderate interobserver reliability. In other indicators kappa scores of >0.6 considered.
Improvement potential the potential for improvement as low for four process indicators and one outcome indicator.
Case-mix stability was assess for 12 indicators that had good climetric properties (1,4,8,9,11,15,16,18,(21)(22)(23)26) . All these needed correction for case-mix. The indicators blood glucose every 2 hours during surgery was influenced by gender. Complete data was 298 of 690 participants. 362 returned a completed questionnaire. 24 did not completed an informed consent form and were therefore excluded. 17 medical records showed that criteria were not met and were excluded. 23 were excluded because full data was not available.

Study No
65% of participants had reported care as described for each item.
The scores form the dimensions 'access to care' varied from 52% to 92%. 52% of participants had the opportunity to contact his/her internist.
Overall scores for 'information' were low-13 of 16 items had scores of below 65%. Preoperatively few patients received information 31% and information about target BG. Post operatively few patients were told about their intraoperative BG and insulin administered 20% and 15%.
Patient involvement was also low. 7 of 11 scores were less than 65%.
Scores for communication and education were higher than 85%.
Physical comfort had scores under and over 65%.
Overall scores for co-ordination and integration of care were low. 5 out of 8 items had scores below 65%. 13% did not know who their caregiver was in charge of their diabetes treatment. Or who to contact during their stay 17%.
Transition and continuity varied from 42% to 92%. 49% indicated their G.P knew about their diabetes treatment when they were discharged.
Variation between hospitals was not significant a range of less than 20% was noted.
Insulin treatment prior to hospital admission was associated with higher mean scores for the dimensions of patient's involvement (p<0.001) and co-ordination (p<0.005). Older participants had lower mean scores for the dimensions of information (p=0.002) and co-ordination (p=0.012). these patients who were also treated with oral hypoglycaemic agents or diet prior to hospital admission had the lowest scores for information (p=0.023).
Longer hospital stay was associated with higher scores for coordination (p=0.003).