Impact of Hemoglobin Glycation Index on Complications Following Orthopedic Surgery: A Retrospective Comparative Propensity Score‐Matched Study

ABSTRACT

Objective

Patients with impaired glucose metabolism have an increased incidence of post‐operative complications. The best marker for glycemic control prior to elective orthopedic surgery remains unclear. We aimed to assess the utility of the hemoglobin glycation index (HGI) in predicting early complications following elective orthopedic surgery.

Methods

We retrospectively enrolled 1496 patients who underwent elective orthopedic surgery at Fujian Provincial Hospital in China from Jan 2015 to Jan 2023. Restricted cubic spline (RCS) was used to select the cutoff value of HGI. Propensity score matching (PSM) was performed to reduce confounding bias, and multivariate logistic regression models (with and without adjustment) for complication outcomes were applied to evaluate the odds ratios of HGI.

Results

The U‐shaped curve in RCS analysis suggested dividing HGI into three subgroups: the reference interval (−0.76 to −0.10), the lower group (≤ −0.76), and the higher group (> −0.10). The incidence of early complications significantly increased from the lower (12.5%) and higher (12.2%) subgroups to the reference interval (6.9%). Following PSM, total postoperative complications were more common in patients with lower HGI (OR: 3.272, 95% CI: 1.417–7.556), but patients in the higher HGI subgroup had a higher risk of incision complications (OR: 3.735, 95% CI: 1.295–10.769).

Conclusions

After adjusting for HbA1c levels, higher HGI (> −0.1) was a risk factor for incision complications, but not for other complications. The risk of overall postoperative complications in patients with lower HGIs (≤ −0.76) should not be ignored.

Abbreviations

HGI

Hemoglobin Glycation Index

OR

odds ratio

RCS

restricted cubic spline

1. Introduction

Musculoskeletal disorders affect approximately 1.7 billion people globally, and over 25 million orthopedic surgeries are performed annually worldwide [1]. However, the incidence of early complications associated with orthopedic surgery remains relatively high, and these complications can increase morbidity, lengths, and costs of hospital stays, unplanned readmissions, and mortality rates [2, 3, 4]. Effective and comprehensive assessment of potential risk factors can help orthopedic surgeons remain vigilant with regard to such complications and prevent their occurrence by reducing the probability of treatment failure.

However, these complications can also be affected by the increasing prevalence of diabetes mellitus worldwide [5, 6]. In China, diabetes is the most common comorbidity among inpatients of orthopedic surgical hospitals [7]. In patients undergoing surgeries, the correlation between blood glucose control and perioperative complications has been well documented, particularly in terms of nosocomial infections and wound healing [5, 8]. Previous studies have further attempted to identify the link between surgical outcomes and blood glucose level, which appears to be a key predictor [5, 9, 10, 11, 12]. However, controversy remains regarding exactly which indicators are ideal. Glycated hemoglobin (HbA1c), a highly stable biomarker, can be used to assess glycemic control levels over periods of 3 months preceding surgeries, and represents a measure that is not easily affected by short‐term fluctuations [13]. An exceptionally high HbA1c level before surgery is considered to be a predictor of poor outcomes in joint replacements, particularly in terms of periprosthetic infections and wound complications [14, 15, 16, 17]. However, this finding has been challenged by other studies [16, 18, 19]. HbA1c does not fully represent the functional level of blood glucose metabolism, and is influenced by factors such as race, sex, basal metabolism, and hematological sources [20, 21]. The linear equation formed by aggregate population‐level fasting blood glucose (FBG) and HbA1c data reflects the influence of short‐term blood glucose fluctuations on long‐term glycemic control levels. The hemoglobin glycation index (HGI) was derived from the gap between the predicted HbA1c levels according to this equation and the actual measured levels [22]. As such, it can be used to eliminate the influence of individual HbA1c levels due to short‐term blood glucose fluctuations and further clarify individual glycemic metabolism statuses [23, 24].

Therefore, the aims of this study were as follows: (i) to explore whether the HGI could help to predict the occurrence of early complications following orthopedic surgery; (ii) to compare the differences in the relative levels of effectiveness between HGI and HbA1c levels for predicting such complications in Chinese patients undergoing elective orthopedic surgeries.

2. Patients and Methods

2.1. Database

This study was approved by the ethics committee of our hospital (No. K2021‐07‐038), and the researchers received authorization to use the hospital's patient medical information database. The Yidu‐Cloud database was searched for course materials, surgical records, basic patient demographics, and laboratory data between January 2015 and January 2023. Follow‐up records from within 30 days following surgeries comprised outpatient or telephone follow‐up records.

2.2. Sample Selection

This retrospective study analyzed data from patients who underwent elective orthopedic surgeries between January 2015 and January 2023. The inclusion criteria were as follows: (a) orthopedic patients with joint or spinal diseases who underwent elective surgical treatments, including primary shoulder, elbow, hip, and knee replacements, as well as spinal fusions; (b) patients undergoing primary surgeries for the first time; and (c) participants who underwent measurements of HbA1c levels prior to surgeries. The exclusion criteria were as follows: (a) patients who received immunosuppressive therapies prior to their surgeries, to treat malignant tumors or immune system diseases; (b) patients who had severe organ failure or infectious diseases before their surgeries that could not be completely corrected; (c) patients who underwent revision surgeries; and (d) patients for whom complete 30‐day postoperative follow‐up data were not available. (Figure 1).

FIGURE 1.

Flow chart of our study population's selected protocol.

2.3. Outcomes

We focused on complications that occurred within 30 days following surgery, including organ system and surgical site complications. These medical records were corroborated by complementary examinations. Early postoperative complications included pneumonia, urinary tract infection, myocardial infarction, cerebral infarction, renal failure, liver failure, venous thrombosis, pulmonary embolism, and incisional complications [2]. Poor healing and infection were the two subtypes of wound‐related complications identified. All incision infections needed to be accompanied by microbial culture evidence from secretions or drainage fluids to be confirmed. Other incision‐related issues such as redness, swelling, exudation, dehiscence, and bleeding were attributed to poor healing [25].

2.4. Covariates

The demographics and other baseline patient information analyzed in this study included age, sex, body mass index, smoking, and alcohol status, perioperative biochemical examination data, and surgical records. Perioperative markers from laboratory tests were applied to individually assess each patient's physiological status. Albumin and total cholesterol were the two nutritional indicators we studied. Alanine aminotransferase and aspartate levels were used to indicate the liver function. Serum creatinine and urea nitrogen levels were used as indicators of renal function. Surgical records were reviewed for surgical procedures, surgical duration, and blood transfusion rate. HbA1c, measured using high‐performance liquid chromatography, and FBG levels were the main markers of glucose metabolism. FBG was quantified using the fasting plasma hexokinase method on the first postoperative day. The linear relationship between HbA1c and FBG was estimated from the linear regression analysis of the study subjects' data (HbA1c = 3.703 + 0.024 × FBG (mg/dL), n = 1496, Figure 1). A predicted HbA1c level was then calculated from this equation using each subject's FBG value. HGI was defined as the difference between the actual HbA1C levels and the predicted levels using linear equations(HGI = measured HbA1c—predicted HbA1c). In accordance with The China Diabetes Society (CDS) criteria [26], diabetes was defined in participants with self‐reported diabetes diagnosed by a health professional or with a fasting plasma glucose level of 126 mg/dL or greater, a 2‐h plasma glucose level of 200 mg/dL or greater after a 75‐g oral glucose challenge, or HbA1c level of 6.5% or greater.

2.5. Statistical Analysis

All data were analyzed using SPSS software version 23.0 (SPSS Inc., Chicago, IL, USA), SAS 9.2 software (SAS Institute Inc., Cary, NC, USA), and R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). All examinations were two‐sided, and statistical significance was set at p < 0.05. Odds ratios (ORs) were calculated, and results are reported as ORs and 95% confidence intervals (CIs).

A linear regression equation was applied to predict the HbA1c and FBG levels of all the enrolled participants (n = 1496). HGI was calculated as the difference between the actual and predicted HbA1c values. The HGIs from the participants were ranked in descending order and trisected. A restricted cubic spline (RCS) was applied to evaluate relationships between HGIs and patient outcomes from early complications. Three knots are located at the 10th, 50th, and 90th positions within the HGI distribution. The turning point at which the effect of HGI on prognosis changed was identified as a log OR of 0 in the RCS plot. Using these values, the patients were divided into three subgroups according to two HGI cutoff values: the reference interval (−0.76 to −0.10), the lower group (≤ −0.76), and the higher group (> −0.10).

We further analyzed differences in demographics, baseline data, and the incidences of various early postoperative complications. Nonparametric analyses were performed using the Kruskal–Wallis test (for continuous variables) and Pearson Chi‐squares analyses (for categorical variables). Propensity score matching (PSM) was applied to reduce selection and confounding biases. Variables that differed between the HGI groups and were found to be independent risk factors for complications were matched. In our PSM, each matched variable was used in the same form as in the one in which it was presented within the baseline characteristics. The caliper value of the PSM was 0.02. To investigate the impact of the HGI subgroups on early postoperative outcomes, we applied binary logistic regression analysis models to test for associations, by gradually adjusting for adjusting factors. Model 1 adjusted for sociodemographic characteristics (age and sex). Model 2 further adjusted for body mass index, surgical site, lifestyle (smoking, drinking), liver function (alanine transaminase), renal function (creatinine), hemoglobin, and nutritional condition total (cholesterol and albumin). In model 3, HbA1c and FBG were further added to Model 2. There is no multicollinearity between all variables.

3. Results

3.1. Baseline Characteristics

Of the total 1496 patients included in this study, 671 (44.9%) were women and the median age of the cohort was 66 years. Of the surgical procedures, 686 (44.7%) were joint replacements in the extremities and 828 (32.3%) were spinal fusions. The median operative time was 150 min, and 42 patients (2.8%) received blood transfusions during or after surgery (Table 1). The incidence of early complications within 30 days following surgery was 9.89% (148/1496), with incision complications (33.76%) accounting for the largest proportion (Figure 2).

TABLE 1.

Baseline characteristics and perioperative data.

Variables	All participants (n = 1496)	No complications (1348)	Early complications (n = 148)	p b
Age, years	66 (56–73)	65 (55–73)	71 (64–77)	0.000 a
Sex				0.266
Male	825 (55.1%)	737 (54.7%)	88 (81.6%)
Female	671 (44.9%)	611 (45.3%)	60 (66.4%)
Body‐mass index, kg/m2	24.1 (22.0–26.6)	24.1 (22.0–26.6)	24.2 (22.0–26.7)	0.733
Albumin	44.3 (41.6–46.2)	44.3 (41.7–46.3)	43.8 (40.5–46.0)	0.049 a
Total cholesterol	4.87 (4.19–5.65)	4.87 (4.17–5.69)	4.80 (4.25–5.38)	0.268
Triglyceride	1.43 (1.02–1.95)	1.44 (1.03–1.96)	1.31 (0.98–1.90)	0.101
CK‐MB	13.2 (10.8–16.5)	13.2 (10.8–16.5)	12.3 (10.6–15.8)	0.158
Alanine transaminase	19.0 (14.0–28.0)	19.0 (14.0–28.0)	18.7 (13.0–27.0)	0.311
Aspartate transaminase	19.0 (16.0–24.0)	19.0 (16.0–24.0)	18.5 (15.9–25.0)	0.984
Blood urea nitrogen	5.46 (4.48–6.60)	5.46 (4.47–6.60)	5.49 (4.56–6.71)	0.418
Creatinine	65 (55–77)	65 (55–77)	65 (54–77)	0.883
HGI	−0.11 (−0.39–0.20)	−0.12 (−0.40–0.17)	0.03 (−0.29–0.48)	0.000 a
Hemoglobin	138 (130–146)	138 (130–146)	136 (125–146)	0.017 a
Preoperation‐FBG, mg/dL	99 (91–114)	98 (91–113)	102 (93–126)	0.010 a
Postoperation‐FBG, mg/dL	116 (100–136)	116 (100–136)	118 (100–141)	0.605
Hemoglobin A1c, %	6.0 (5.6–6.5)	6.0 (5.6–6.5)	6.2 (5.8–7.3)	0.000 a
Operative site				0.000 a
Extremities	656 (43.9%)	570 (42.3%)	86 (58.1%)
Spine	840 (56.1%)	778 (57.7%)	62 (41.9%)
Operation time, min	150 (115–190)	150 (115–190)	150 (116–184)	0.972
Blood transfusion	42 (2.8%)	36 (2.7%)	6 (4.1%)	0.481
Smoker	233 (15.6%)	216 (16.0%)	17 (11.5%)	0.148
Alcohol	71 (4.7%)	65 (4.8%)	6 (4.1%)	0.677
DM	483 (32.3%)	423 (31.4%)	60 (40.5%)	0.024 a

Note: Categorical variables presented as n (%); continuous variables described as median (interquartile range).

Abbreviations: CK‐MB, Creatine kinase‐myocardial band; DM, diabetes mellitus; FBG, fasting blood glucose; HGI, hemoglobin glycation index.

^a Statistically significant at alpha = 0.05.

^b Overall differences between outcome subgroups using Kruskal‐Wallis tests, or χ ² tests.

FIGURE 2.

Categories and proportions of early postoperative complications. The frequencies do not add up to the total number of complications because each patient may have more than one postoperative complication. Lower extremity DVT (deep vein thrombosis) and PE (pulmonary embolism) must be confirmed by vascular ultrasound and pulmonary CTA. Major cardiovascular events (MCVE) were defined as myocardial infarction, pulmonary edema, and nonfatal severe arrhythmia. The diagnosis of infection should be confirmed by microbiology test.

The characteristics of the patients with and without early postoperative complications are shown in Table 1. Compared to the control group without complications, the patients with early postoperative complications were older (p < 0.001) and had lower albumin levels (p = 0.049). It is worth noting that, in terms of glucose metabolism‐related indicators, the patients with complications had higher preoperative FBG (p = 0.010) and HbA1c (p < 0.001) levels, as well as a higher prevalence of diabetes mellitus (p = 0.024). However, there was no significant difference in terms of postoperative FBG levels between the groups (p = 0.605).

3.2. Demographic and Biochemical Data According to the Hemoglobin Glycation Index Cutoff Point

A linear regression model was applied to calculate the predicted HbA1c using baseline FBG and HbA1c data, which was then used to calculate HGI. The predicted HbA1c (%) was calculated as HbA1c (%) = 3.703 + 0.024 × FBG (mg/dL); R² = 0.505 (Figure 3).

FIGURE 3.

Linear regression equation for all participants (n = 1496). Predicted HbA1c (%) = 3.703 + 0.024 FBG (mg/dL), R² = 0.505. FBG, fasting blood glucose; HbA1c, hemoglobin A1c.

HGI was assessed as a risk predictor of early postoperative complications, and significant differences were observed between the groups (Table 1; p < 0.001). Our RCS plot revealed a reduction in the risk of complications within the range for HGIs, which reached a lower level of risk near −0.76 to −0.10 and increased thereafter (Figure 4). Therefore, we selected between −0.76 and −0.10 as the HGI reference interval and divided the cohort into a lower and a higher group.

FIGURE 4.

RCS plots revealing the relationship between HGI values and early complication risk. The turning point at which the effect of the HGI on prognosis changes was identified at an OR of 1.

Participants in the higher HGI group were older (p < 0.001), had lower albumin levels (p = 0.002), and had higher HbA1c levels (p < 0.001) than those in the other groups (Table 2). Therefore, it was logical that the subjects in the higher HGI group also had a higher proportion of patients with diabetes mellitus. No significant differences were observed among the three subgroups in terms of surgical procedures, operative duration, or transfusion rates. However, the incidence of early complications significantly increased from the lower and higher subgroups to the reference interval (12.5%, 12.2%, and 6.9% in the lower, higher, and middle ranges, respectively). Myocardial Infarction, acute renal failure, and pulmonary embolism occurred only in patients of the lowest HGI group. The high HGI group had the highest percentage of incision complications (5.3%) compared to the low (2.9%) and moderate HGI groups (1.8%; Table 2).

TABLE 2.

Demographic and biochemical data of the participants according to hemoglobin glycation index.

Variables	Reference interval (−0.76 −0.10, n = 654)	Lower (< −0.76, n = 108)	Higher (≥ −0.10, n = 734)	p b
Age, years	65 (53–73)	64 (52–71)	67 (59–74)	0.020 a
Sex				0.034 a
Male	362 (55.2%)	45 (43.3%)	418 (56.8%)
Female	294 (44.8%)	59 (56.7%)	318 (43.2%)
Body‐mass index, kg/m2	24.03 (21.75–26.45)	23.65 (21.26–26.03)	24.34 (22.28–26.67)	0.166
Albumin	44.7 (41.9–46.6)	43.9 (40.9–45.7)	44.0 (41.2–46.0)	0.002 a
Total cholesterol	4.87 (4.18–5.72)	4.73 (4.12–5.24)	4.88 (4.21–5.55)	0.117
Triglyceride	1.45 (0.96–1.96)	1.30 (1.05–1.75)	1.45 (1.05–1.99)	0.088
CK‐MB	13.1 (10.6–16.5)	13.6 (11.0–17.3)	13.1 (10.9–16.48)	0.587
Alanine transaminase	19 (14–28)	20 (15–29)	20 (14–28)	0.679
Aspartate transaminase	19 (16–24)	21 (16–27)	19 (16–24)	0.186
Blood urea nitrogen	5.38 (4.48–6.49)	5.50 (4.32–7.22)	5.51 (4.48–6.66)	0.672
Creatinine	65 (55–77)	66 (55–80)	66 (55–78)	0.740
Hemoglobin	138 (131–146)	138 (129–149)	138 (129–146)	0.798
Preoperation‐FBG, mg/dl	98 (91–108)	120 (104–156)	98 (89–115)	0.000 a
Postoperation‐FBG, mg/dl	116 (101–135)	109 (98–130)	117 (100–138)	0.169
Hemoglobin A1c, %	5.7 (5.5–5.9)	5.5 (5.2–6.1)	6.3 (6.0–7.0)	0.000 a
Operative site				0.654
Extremities	296 (45.1%)	43 (41.3%)	317 (43.1%)
Spine	360 (54.9%)	61 (58.7%)	419 (56.9%)
Operation time, min	148 (115–190)	150 (110–189)	150 (120–190)	0.360
Blood transfusion	19 (2.9%)	4 (3.8%)	19 (2.6%)	0.753
Smoker	98 (14.9%)	18 (17.3%)	117 (15.9%)	0.780
Alcohol	32 (4.9%)	14 (13.5%)	25 (3.4%)	0.000 a
DM	94 (14.3%)	49 (47.1%)	340 (46.2%)	0.000 a
Early complications	45 (6.9%)	13 (12.5%)	90 (12.2%)	0.002 a
Incision complications	12 (1.8%)	3 (2.9%)	39 (5.3%)	0.002 a
Other complications	33 (5.0%)	10 (9.6%)	53 (7.2%)	0.099

Note: Categorical variables presented as n (%); continuous variables described as median (interquartile range).

Abbreviations: CK‐MB, Creatine kinase‐myocardial band; DM diabetes mellitus; FBG fasting blood glucose; HGI, hemoglobin glycation index.

^a Statistically significant at alpha = 0.05.

^b Overall differences between HGI subgroups using Kruskal‐Wallis tests, or χ ² tests.

3.3. Association Between Hemoglobin Glycation Index and the Incidence of Early Complications

To investigate any independent association between HGI and early postoperative complications, PSM was applied to match comparable baseline characteristics. A subsequent analysis revealed that patients with early complications had higher HGIs (p = 0.023) and HbA1c levels (p = 0.010; Table 3). However, neither preoperative nor postoperative FBG levels showed any significant difference between the groups.

TABLE 3.

Baseline characteristics and perioperative data of the participants with PSM.

Variables	All participants (n = 630)		p b
	Non‐complications (n = 504)	Early Complications (n = 126)
Age, years	69 (63–75)	70 (62–75)	0.879
Sex			0.545
Male	215 (42.7%)	50 (39.7%)
Female	289 (57.3%)	76 (60.3%)
Body‐mass Index, kg/m2	24.22 (22.03–26.71)	24.44 (22.01–27.00)	0.936
Albumin	44.6 (41.7–46.5)	43.9 (40.5–46.0)	0.697
Total cholesterol	4.88 (4.14–5.76)	4.79 (4.22–5.41)	0.233
Triglyceride	1.40 (1.01–1.95)	1.31 (0.98–1.92)	0.427
CK‐MB	13.1 (10.6–16.3)	12.2 (10.6–15.7)	0.359
Alanine transaminase	19.0 (14.0–28.0)	18.0 (13.0–27.0)	0.554
Aspartate transaminase	19.0 (15.9–24.0)	18.0 (15.0–25.0)	0.948
Blood urea nitrogen	5.57 (4.54–6.80)	5.56 (4.55–6.78)	0.840
Creatinine	65 (54–78)	65 (55–76)	0.837
HGI	−0.95 (−0.36–1.93)	0.03 (−0.29–0.37)	0.023 a
Higher HGI	196 (38.9%)	64 (50.8%)	0.015 a
Hemoglobin	138 (129–146)	135 (124–145)	0.036
Preoperation‐FBG, mg/dl	100 (92–115)	100 (93–126)	0.160
Postoperation‐FBG, mg/dl	114 (99–133)	117 (99–141)	0.359
Hemoglobin A1c, %	6.0 (5.7–6.5)	6.2 (5.8–7.0)	0.010 a
Operative site			0.811
Extremities	266 (52.8%)	68 (54.0%)
Spine	238 (47.2%)	58 (46.0%)
Operation time, min	150 (120–194)	140 (110–180)	0.331
Blood transfusion	19 (3.8%)	5 (4.0%)	0.917
Smoker	68 (13.5%)	16 (12.7%)	0.815
Alcohol	27 (5.4%)	6 (4.8%)	0.789
DM	179 (35.5%)	49 (38.9%)	0.481

Note: Categorical variables presented as n (%); continuous variables described as median (interquartile range).

Abbreviations: CK‐MB, Creatine kinase‐myocardial band; DM, diabetes mellitus; FBG, fasting blood glucose; HGI, hemoglobin glycation index.

^a Statistically significant at alpha = 0.05.

^b Overall differences between HGI subgroups using Kruskal‐Wallis tests, or χ ² tests.

We subsequently performed a multivariate analysis of the HGI subgroups for both the adjusted and unadjusted models (Table 4). In the unadjusted model, the OR (95% CI) for incision complications in the higher HGI group, compared to the reference group, was 3.548 (1.718–7.329). This relationship did not change significantly in the adjusted models. In Model 3, those in the higher HGI group had a 3.272‐fold greater risk for incision complications than those in the reference group (p = 0.006). For total complications risk, covariates HbA1C and FBG significantly affect the results. In the crude model and Models 1–2, no elevated risk for total complications was observed. However, while further adjusted for HbA1c and FBG in Model 3, those in the lower HGI group had a significant risk for total complications compared to those in the reference group (OR(95% CI):1.295–10.769; p = 0.015). Notably, HbA1c, rather than HGI, remained a significant risk factor for other complications both with and without adjustment.

TABLE 4.

Association between the HGI and early postoperative complications in the PSM group.

Outcomes for postoperation		Unadjusted			Model 1			Model 2			Model 3
		OR	95% CI	p	OR	95% CI	p	OR	95% CI	p	OR	95% CI	p
Total complications	Lower HGI	1.779	0.782–4.048	0.169	1.810	0.788–4.156	0.162	2.119	0.865–5.192	0.101	3.735	1.295–10.769	0.015
	Higher HGI	2.018	1.293–3.149	0.002	2.019	1.293–3.152	0.002	2.227	1.379–3.598	0.001	1.290	0.715–2.326	0.398
	HbA1c	1.564	1.278–1.914	0.000							2.690	1.650–4.386	0.000
Insicion complication	Lower HGI	1.744	0.451–6.751	0.420	1.824	0.466–7.141	0.388	1.853	0.433–7.921	0.405	2.224	0.478–10.349	0.308
	Higher HGI	3.548	1.718–7.329	0.001	3.564	1.724–7.369	0.001	3.890	1.806–8.382	0.001	3.272	1.417–7.556	0.006
	HbA1c	1.346	1.080–1.676	0.008							1.340	0.851–2.110	0.206
Other complication	Lower HGI	1.595	0.650–3.917	0.308	1.600	0.644–3.973	0.311	1.833	0.679–4.947	0.232	2.797	0.906–8.638	0.074
	Higher HGI	1.213	0.731–2.013	0.455	1.209	0.728–2.006	0.464	1.309	0.757–2.262	0.335	0.806	0.419–1.552	0.519
	HbA1c	1.359	1.116–1.655	0.002							2.188	1.419–3.374	0.000

Note: Model 1: Adjusted for Age and Sex; Model 2: Body Mass Index, Surgical site, Smoking, Drinking, Alanine Transaminase, Creatinine, Hemoglobin, Total Cholesterol and Albumin added to Model 1; Model 3: HbA1c and FBG added to Model 2.

Abbreviations: CI, confidence interval; HbA1c, hemoglobin A1c; HGI, hemoglobin glycation index; OR, odd ratios.

4. Discussion

4.1. Main Findings of This Study

Our study analyzed 1496 surgical patients and found a 9.89% incidence of multiple kinds of early complications following orthopedic surgery, with incision issues being the most common. The U‐shaped relationship in RCS plots was used for group sorting of HGI. Using a linear regression model, patients in the higher HGI group (≥ −0.10) had a 3.27‐fold higher risk of incision complications, while those in the lower HGI group (< −0.76) had an unexpected increase in severe postoperative complications. Overall, HGI and HbA1c are indicators that are closely related, but also have their own biological characteristics, especially in perioperative risk prediction models.

4.2. Effect of HbA1c on HGI Prediction Models

Previous studies have suggested that some individuals have higher or lower HbA1c levels than those predicted based on population data, which has resulted in some controversy regarding postoperative outcomes [13, 15, 27, 28, 29]. The evidence for HGI as an independent predictor of perioperative complications is less clear [30, 31]. This retrospective analysis of medical data from southeastern China demonstrated that a high absolute HGI value (< −0.76 or ≥ −0.10) was strongly associated with poor outcomes in patients who underwent elective orthopedic surgeries. A higher HGI (≥ −0.10) was found to be a significant influencing factor for poor wound healing and incision infection within the first 30 days following spinal and articular surgeries. However, it had no impact on non‐incisional complications, which were significantly affected by HbA1c levels.

Diabetes mellitus with impaired glucose metabolism has been shown to be an independent risk factor for postoperative complications [5, 6]. In our study, patients with higher HGIs also had higher HbA1c levels and lower preoperative fasting blood levels. These results are consistent with the characteristics assigned to the HGI by our mathematical transformation. Poor glycemic control and hyperglycemia result in a disturbed inflammatory response and decreased angiogenesis, which ultimately delay wound healing [32, 33]. Intuitively, hyperglycemia is the mechanism by which HGI increases the incidence of incision‐related complications. However, the risk of a higher HGI for incisional complications did not disappear following correction for HbA1c levels. Advanced glycosylation end products (AGEs) contribute to endothelial damage through various mechanisms, including oxidative stress and inflammation, and are thought to be important with regard to the involvement of HGI in tissue repair [34]. AGEs accumulate continuously in the body, particularly in long‐lived macromolecules such as collagen [33, 35]. This also showed that HGI has a better predictive effect on wound complications than HbA1c under certain conditions because it takes into account a wider range of glucose metabolism characteristics. This requires a prospective study with a larger sample size to confirm. Note that our results remind that patients with long‐term stable HbA1c but drastic changes in blood sugar levels during the perioperative period should not be ignored in the clinic. They are the characteristics of the lower HGI population.

4.3. The U‐Shaped Relationship Between HGI and Postoperative Outcomes

The roughly U‐shaped curve relationship between postoperative outcomes following orthopedic surgeries and HGI values confirmed the previously established adverse effects of both short‐ and long‐term glycometabolic disorders [36, 37, 38]. Increasing evidence has shown that severe glucose fluctuations and hyperglycemia before surgery are closely associated with diabetic complications [39]. In the complication subgroup analysis, we found that these patients with low HGI were not at low risk for wound complications. Contrary to common belief, both the higher and the lower HGI are risk factors for wound healing in this study. Our results show that the lower HGI subgroup presented with relatively low HbA1c but high preoperative FBG levels, which have been shown to be a risk factor for impaired β‐cell function in childhood diabetes [22, 40]. Acute hyperglycemia promotes oxidative stress and notable mitochondrial damage, resulting in renal and cardiovascular injuries [41, 42]. Thus, different glycemic control goals and strategies should be adopted for patients in different baseline HGI groups. Similar procedures can be applied to determine the timing of surgery; however, prospective studies are warranted to verify this.

4.4. Prospect and Limitations

The study analyzed 1496 patients over 8 years, using RCS, PSM, and multivariate models to ensure robust statistical validity. We identified a U‐shaped relationship between HGI and postoperative complications, showing that both high and low HGI values increase risk. These new findings provide new ideas for postoperative assessment and choice of the ideal glycemic biomarker, which emphasized the need for personalized glycemic control strategies in orthopedic surgery.

This study has several key limitations which should be noted. First, this is a retrospective study, so a certain degree of unintentional selection bias cannot be ruled out. In order to obtain complete data, we had to select patients for whom both HbA1c and fasting glucose data were available, which likely led to a class I bias. We cannot explain why the remaining unenrolled patients who underwent elective orthopedic surgeries did not undergo HbA1c testing. One reason may be that the doctors believed that their risks of diabetes were so low that they did not need to be tested for anything other than fasting blood sugar levels. Second, FBG replaced MBG (Mean blood glucose) as a factor in the HbA1c predictive value equation [43]. We referred to the previous literature and believe that FBG, to a certain extent, can reflect immediate blood glucose levels in the short term [22, 44, 45]. However, it is undeniable that single data points suffer from fluctuation errors. Third, in our multi‐factor regression model, we did not include all risk factors and targeted only the most influential ones, such as age, sex, and body mass index. The most direct objective of our study was to determine the effects of HGI and HbA1c levels on short‐term outcomes. Finally, we were also limited by a smaller sample size for our subgroup analyses. The overall sample size was insufficient to assess whether the low incidence of early complications was associated with the glycation indicators. A poor record of the complications experienced by the different subgroups led to a lack of detailed analyses. Despite these limitations, this study represents a novel attempt to use HGI in the analysis of surgical complications.

5. Conclusion

Overall, in the present study, we analyzed the relationship between glycosylation indicators and early complications in patients who underwent elective orthopedic surgeries. The results showed that higher HGI (> −0.1) was a risk factor for incision‐related complications, but not for other complications after adjusting for HbA1c levels. The risk of overall postoperative complications in patients with lower HGIs (≤ −0.76) should not be ignored. Our results suggest that further research is warranted to analyze the association between glycosylation indicators and specific mechanisms.

Author Contributions

Yiyang Xu data curation, funding acquisition, investigation, methodology, software, validation, writing – original draft, writing – review and editing. Bochen Sun: methodology, writing – original draft, writing – review and editing. Guoyu Yu: supervision, validation. Fenqi Luo: data curation, resources. Long Chen: methodology, resources, software, validation. Jie Xu: supervision. Jun Luo: data curation, funding acquisition, software, visualization. Ting Xue data curation, funding acquisition, investigation, methodology, software, validation, visualization, writing – original draft, writing – review and editing. Yuhua Xiao: data curation, funding acquisition, supervision, writing – review and editing.

Ethics Statement

Ethics Statement for this study was obtained from Fujian Provincial Hospital, Fuzhou, China (No. K2021‐07‐038). This work was performed at Fujian Provincial Hospital, Fuzhou, China.

Conflicts of Interest

The authors declare no conflicts of interest.