The effect of diabetes mellitus on chronic rhinosinusitis and sinus surgery outcome

Abstract

Objective

Patients with diabetes mellitus (DM) are known to be liable to infection. However, the association between diabetes and chronic rhinosinusitis (CRS) has not been well studied. We sought to determine the effects of DM on CRS culture results and quality of life (QOL) after functional endoscopic sinus surgery (FESS).

Methods

We conducted a retrospective cohort study. Adult CRS patients undergoing FESS were recruited from 10/1/2007 to 12/31/2011. Patient demographics, comorbidities, medication use, Lund-Mackay CT scores were collected prior to FESS. Intraoperative culture was obtained. Preoperative and 1-, 3- and 6-month postoperative QOL was measured by the 22 item Sinonasal Outcome Test (SNOT-22) scores. A mixed effects model was performed for analysis.

Results

Among the 376 CRS patients included, 19 patients (5.05%) had DM. Compared to non-DM patients, DM patients were significantly more likely to have Pseudomonas aeruginosa (26.32% versus 7.56%, p=0.004) and gram negative rods (26.32% vs. 8.96%, p= 0.013), but there were no significant difference in the prevalence of Staphylococcus aureus; DM patients were also significantly more likely to have nasal polyps and gastroesophageal reflux disease. Additionally, DM patients had significantly less improvement of postoperative SNOT-22 scores from baseline to 6-month follow-up than non-DM patients (adjusted mean=11.14, 95% CI (0.14, 22.15), p=0.047) after adjusting for all the other risk factors for CRS.

Conclusion

DM patients may be prone to gram negative bacterial sinus infections, and have significantly worse short-term postoperative QOL. Special postoperative care may need to be considered in CRS patients with DM

Introduction

Chronic rhinosinusitis (CRS) is defined as a group of disorders characterized by inflammation of the mucosa of the nose and paranasal sinuses of at least 12 weeks duration. It is estimated to affect 14–16% of the population in the United States.¹ CRS is a heterogeneous multifactorial disease². Many risk factors for CRS have been identified, such as age³sex^4,5smoking status⁶nasal polyps^3,7allergic rhinitis⁶asthma^3,7aspirin intolerance⁴infection, biofilms, gastroesophageal reflux (GERD)⁸as well as anatomical abnormalities of upper airways and histological appearance of sinus mucosa.^9–11 However, it is unknown whether diabetes mellitus (DM), the fifth deadliest disease in the United States, affects the pathogenesis and prognosis of CRS.

8.3% of the U.S. population has DM¹². The percentage of people with diagnosed diabetes is increasing over the past decades. Elevated blood sugar levels in patients with DM inhibit the immune system.¹³ In addition, some diabetes-related health issues, such as nerve damage and reduced blood flow also increase the body's vulnerability to infection. Previous study has shown that patients with DM are at increased risk for lower respiratory tract infection, urinary tract infection, and skin and mucous membrane infection.¹⁴ Pseudomonas aeruginosa and Staphylococcus aureus are two of the most common bacterial isolates in patients with CRS.^15,16 It is of considerable interest to assess if CRS patients with DM are more susceptible to certain types of infection and whether DM affects the prognosis of CRS after functional endoscopic sinus surgery (FESS). The specific aims of our study were to determine whether DM was associated with certain types of sinus culture results, especially Staphylococcus aureus or Pseudomonas aeruginosa, as well as the short-term quality of life (QOL) improvement after FESS for CRS.

Methods

Study design and study population

Institutional review board approval was obtained from the University of Pennsylvania. We conducted a retrospective cohort study. Adult CRS patients (≥18 years old) who were scheduled to have FESS at the Department of Otorhinolaryngology–Head and Neck Surgery, the Hospital of the University of Pennsylvania were recruited between 10/1/2007 to 12/31/2011. The inclusion criteria were patients who met the objective and subjective guidelines for the diagnosis of CRS¹⁷. CRS diagnosis is set forth by the Rhinosinusitis Initiative: patients with CRS experienced at least 2 of the typical symptoms, such as headache, nasal obstruction, and nasal drainage, that persisted for 12 weeks or longer and showed edema or erythema of the middle meatus, as identified by means of nasal endoscopy¹⁷. Furthermore, a computed tomographic (CT) scan showed diffuse mucosal thickening, opacification, or swelling of the ethmoidal and maxillary mucosa with bilateral obstruction of the osteomeatal complex.¹⁸ According to the recommendation of guidelines for the studies of CRS,^19,20 patients with the following diseases were excluded: cystic fibrosis, congenital or acquired immunodeficiency, congenital mucociliary, non-invasive fungal balls and invasive fungal disease, systemic vasculitis and granulomatous diseases, cocaine abuse, or neoplasia.

Diabetes and other covariates

Patients’ medical records were reviewed to determine whether the patients had diagnosis of DM. Type I or Type II diabetes diagnosis was listed in the preoperative medical records. Preoperative fasting glucose level, hemoglobin A1c (HbA1c) and diabetic medications were also collected. In addition, basic demographic data and comorbidities including age, sex, race, asthma, nasal polyps, allergic rhinitis, obstructive sleep apnea, smoking status, GERD, prior FESS history, and Samter's triad were collected from patients’ medical records at baseline. The preoperative Lund-Mackay CT scores were assessed by a research nurse and a medical assistant. The use of oral antibiotics, oral steroids, nasal steroid spray, and nasal irrigations in the month before surgery was also recorded.

Outcomes

There were two outcomes of our study. One was to identify any bacterial isolates from sinus culture at the time of FESS, especially S. aureus or P. aeruginosa. The other was to evaluate the improvement of QOL from the latest preoperative visit to 1-, 3- and 6-month postoperative follow up visits. Each patient had endoscopically guided sinus culture taken from one of the diseased sinuses during FESS using standard method^21–23. The samples were immediately sent to the hospital microbiology laboratory for standard microbiological speciation. QOL was measured by the 22 item Sinonasal Outcome Test (SNOT-22) scores. The SNOT-22 contains 22 symptoms related items on sinonasal symptoms, sleep, emotional and psychosocial functioning, where each item is scored on a range of 0 to 5 depending on severity, thus giving a total score of 0 to 110.²⁴ Higher scores represent worse QOL. Clinically significant change of SNOT-22 score was defined as difference of at least ½ SD of the baseline SNOT-22 score in the reference group.^25,26

Statistical Analysis

First, we summarized all the patients’ characteristics and tested whether the continuous variables were normally distributed. If the distribution was not normal, the variable was log transformed. In the univariate analyses, χ2 tests were used for categorical variables and t tests were used for continues variables. We compared the isolation rates of different bacteria between patients with and without DM using χ2 tests. To determine whether DM was independently associated with the improvement of SNOT-22 scores from baseline to 1-, 3- and 6-month postoperative follow up visits, we conducted the analysis using the linear mixed-effects regression model, with fixed effects for time modeled as a categorical variable, diagnosis of diabetes, and the interaction between the two in order to allow the effects of diabetes on postoperative QOL to vary over time. The model also included random effects for patient to account for the correlation between repeated measures per patient. If a SNOT-22 score was missing at a certain time point, the rest of the scores from that same patient were still included in the mixed effects model. Potential confounding variables listed above were considered, and these with p<0.2 in the univariate analyses were entered into the final multiple regression model. The adjusted β coefficients and 95% confidence intervals (95% CI) were calculated from the linear mixed effects regression model to assess the magnitudes of the associations. The adjusted β coefficient can be interpreted as the adjusted mean difference of postoperative improvement of SNOT-22 scores in CRS patients with and without the condition while holding the other variables in the model constant. Significance was based on α<0.05 and all hypothesis tests were 2-sided. The analyses were performed using Stata version 11.0 (Stata Corp, College Station, Texas).

Results

376 CRS patients were included in the analysis. Table 1 shows the patients’ characteristics in our study. The mean ± SD age was 48 ± 13 years. 42.55% of the patients were female (n=160) and 88.8% were white (n=333). 5.05% patients (n=19) had DM. Among them, 90% type II diabetes and 10% were type I; the range of fasting glucose level was 65–367 mg/dL, the mean was 162.5 mg/dL; the range of HbA1c was 5.9–8.8%, the mean was 7.12%; for diabetic medications, 25% had metformin only, and 40% had insulin from the preoperative records. However, approximately 10% of the patients only have the diagnosis of diabetes in the medical records without any of the detailed information. Compared to non-DM patients, CRS patients with DM were significantly more likely to have nasal polyps (89.47% versus 58.54%, p=0.007) and GERD (47.37% versus 24.37%, p=0.025). 27.66% patients (n=104) had isolate of S. aureus and 8.51% (n=32) had P. aeruginosa. The culture results are listed in Table 2. Compared to CRS patients without DM, CRS patients with DM were significantly more likely to have P. aeruginosa (26.32% versus 7.56%, p=0.004) and other gram-negative rods (26.32% versus 8.96%, p=0.013), but not S. aureus (21.05% versus 28.01%, p=0.509), S. pneumonia (0 versus 5.30%, p=0.302), Coagulase-negative staphylococcus (73.68% versus 56.86%, p=0.148) and anaerobes (45.45% versus 55.40%, p=0.518).

Table 1.

Patients’ characteristics by diabetes

	Total	Diabetes, n (%)		p-value
		No	Yes
Age (mean, SD)	48 ± 13	50 ± 14	48 ± 13	0.671
Sex (female)	160(42.55)	153(42.86)	7(36.84)	0.605
Race
Black	31(8.27)	30(8.43)	1(5.26)	0.131
Other	11(2.93)	10(2.81)	1(5.26)
Preoperative antibiotics	67(17.82)	63(17.65)	4(21.05)	0.705
Nasal rinse				0.341
Saline	123(32.98)	114(32.20)	9(47.37)
With Medications	8(2.15)	8(2.26)	0(0)
Steroid use				0.523
Nasal	65(17.29)	63(17.65)	2(10.53)
Oral	119(31.65)	112(31.37)	7(36.84)
Both	31(8.25)	28(7.84)	3(15.79)
Prior FESS	229(60.9)	214(59.94)	15(78.95)	0.098
Asthma	205(54.52)	193(54.06)	12(63.16)	0.438
Nasal polyps	226(60.11)	209(58.54)	17(89.47)	0.007
Allergic rhinitis	248(65.96)	234(65.55)	14(73.68)	0.466
Sleep apnea	17(4.52)	17(4.76)	0(0)	0.330
Smoking				0.923
Former	75(19.95)	71(19.89)	4(21.05)
Current	51(13.56)	49(13.73)	2(10.53)
Gastroesophageal reflux disease	96(25.53)	87(24.37)	9(47.37)	0.025
Samter’s triad	30(7.98)	29(8.12)	1(5.26)	0.654
Preoperative Lund-Mackay CT scores (mean ± SD)	13 ± 6	12.63	13.19	0.732
Preoperative SNOT-22 scores (mean ± SD)	40 ± 22	40.35	40.84	0.926

Table 2.

Culture results by diabetes

Positive Culture	Total	Diabetes, n (%)		p-value
	N=376	Yes n=19	No n=357
S. aureus	104	4 (21.05)	100 (28.01)	0.509
P. aeruginosa	32	5 (26.32)	27 (7.56)	0.004
Other gram-negative rods	37	5 (26.32)	32 (8.96)	0.013
S. pneumonia	17	0	17 (5.30)	0.302
Coagulase-negative staphylococcus	217	14 (73.68)	203 (56.86)	0.148
Anaerobes (n=224)*	123	5 (45.45)	118 (55.40)	0.518

^*224 patients had anaerobic culture results.

We further evaluated whether the preoperative severity of CRS was different between patients with and without DM. The preoperative Lund-Mackay CT scores were not significantly different between patients with and without DM (13 ± 7 versus 13 ± 6, p=0.732), and neither were the preoperative SNOT-22 scores (41 ± 18 versus 40 ± 23, p=0.926). However, compared to non-DM patients, CRS patients with DM had less improvement of QOL from baseline to 1-, 3- and 6-month postoperative visits as shown in Figure 1.

Figure 1.

Pre- and post-FESS SNOT-22 scores by diabetes

After adjusting for all the covariates in the linear mixed-effects regression model, compared to patients without DM, the improvement of SNOT-22 scores in CRS patients with DM were marginally significantly less from baseline to 1-month postoperative follow-up (adjusted β-coefficient=9.74, 95% CI (-0.18, 19.66), p=0.054), and significantly less from baseline to 6-month postoperative follow-up (adjusted β-coefficient=11.14, 95% CI (0.14, 22.15), p=0.047). The results are shown in Table 3. This adjusted mean SNOT-22 score difference from baseline to 6 month postoperative follow-up, 11.14, was greater than the ½ SD of baseline SNOT-22 scores of CRS patients without DM, 11, and thus, CRS patients with DM had clinically significantly less QOL improvement from baseline to 6-month postoperative visits.

Table 3.

Adjusted QOL improvement after FESS at 1-, 3- and 6-month follow up

	Unadjusted means	Adjusted Coef.	95% Conf. Interval	p-value
Difference of SNOT-22 scores between patients with and without diabetes
From baseline to 1-month follow-up	10	9.74	(−0.18, 19.66)	0.054
From baseline to 3-month follow-up	3	7.34	(−3.66, 18.33)	0.191
From baseline to 6-month follow-up	11	11.14	(0.14, 22.15)	0.047
Difference of SNOT-22 scores between patients with and without variable below*
Nasal polyps	NA	−0.16	(−3.87, 3.55)	0.932
Gastroesophageal reflux disease	NA	0.72	(−3,30, 4.74)	0.726
P. aeruginosa	NA	2.47	(−3.76, 8.70)	0.438
Other gram-negative rods	NA	−2.89	(−8.84, 3.05)	0.340
Prior FESS	NA	3.24	(−0.53, 7.00)	0.092

^*Their effects on QOL improvement were averaged across all postoperative follow-up time points.

In our sub-analysis, we further explored whether patients with Type I versus Type II diabetes had different culture results and postoperative QOL improvements, but did not find any statistically significant results.

Discussion

To our knowledge, this appears to be the first study to assess the impact of DM on CRS in a large clinical cohort. DM increases patients’ susceptibility to various types of infections due to the metabolic derangements.^14,27,28 The most common affected sites in DM are the skin and urinary tract,^29,30 such as staphylococcal follicular skin infections, fungal infections, cellulitis, erysipelas, lower urinary tract infections and acute pyelonephritis. However, it is unclear how DM would affect bacteria isolated from CRS and FESS outcome. Our study showed that CRS patients with DM were significant more likely to have P. aeruginosa and other gram negative rods isolated from the infected sinus, as well as clinically significantly less improvement of postoperative SNOT-22 scores from baseline to 6-month postoperative follow up. Interestingly, CRS patients with DM were also significantly more likely to have nasal polyps and GERD as compared to patients without DM.

Our study found that CRS patients with DM had 3.5 times higher isolation rate of P. aeruginosa and other gram negative rods than CRS patients without DM, however, the isolation rate of S. aureus, S. pneumonia, Coagulase-negative staphylococcus and anaerobes were similar between patients with and without DM. DM infection is usually polymicrobial in nature³¹. S. aureus is common in skin and soft tissue infections in patients with diabetes, but P. aeruginosa and other gram negative rods have been associated with worse outcomes in hospitalized patients.³² Additionally, P. aeruginosa has been associated with malignant external otitis in diabetic patients decades ago³³. Our study appears to be the first one to demonstrate a high prevalence of P. aeruginosa and other gram negative rods from sinus culture in CRS patients with DM. P. aeruginosa and other gram negative rods are important causes of infection, especially in patients with compromised host defense mechanisms. The common bacterial isolates from CRS patients include P. aeruginosa, other gram negative rods and S. aureus^16,34–36. The bacterial isolation rates in our study consist with previous studies^34,37. Further study is needed to better understand the biological mechanism regarding how DM makes the sinus mucosa susceptible for P. aeruginosa and other gram negative rods in CRS.

Another novel finding of our study is that we found CRS patients with DM were significantly more likely to have nasal polyps compared to non-DM patients. As patients with asthma, frequent sinus infections and allergies are more likely to develop nasal polyps,^38–40 further studies are needed to assess whether diabetes affects the inflammatory status of the nasal cavity and sinuses to facilitate the development of nasal polyps. In addition, our finding agrees with previous studies that patients with DM had higher prevalence of GERD.^41,42 Recent studies have shown that GERD was more common in diabetic patients with neuropathy or nerve damage, which are common complications of diabetes.^42,43

After adjusting for the confounding effects of nasal polyps and GERD on postoperative QOL improvement in the linear mixed effects regression model, only DM was a significant independent risk factor for less postoperative improvement of SNOT-22 scores from baseline to 6 month follow up visit; nasal polyps, GERD, P. aeruginosa, other gram-negative rods and prior FESS were not significantly associated with postoperative QOL improvement. Previous studies have shown that prior FESS was independently associated with poor postoperative QOL outcomes.^4,44 However, prior FESS was not significant in our final model. This may be because we included diabetes, P. aeruginosa and other gram negative rods in the final regression model, which were not adjusted in previous studies. Thus, our findings added evidence to the literature that DM was a clinically significant independent risk factor for short-time QOL improvement after FESS for CRS. Future studies with long-term follow up can help to better evaluate the effect of DM on long-term QOL outcome after FESS for CRS.

There are several limitations in our study. First, we did not have objective measures of the postoperative outcomes, such as nasal endoscopy score. However, the use of patient-reported outcome measures is rapidly growing in studies of clinical effectiveness and quality of care. Second, not every CRS patient in our institute completed the SNOT-22 scores. We compared patients with and without SNOT-22 scores and there were no statistically significant differences between them. Third, due to the fact that many patients followed up with specialist outside of Penn for diabetes control, we did not have objective measures of diabetes across different time points to determine whether diabetes control and compliance of diabetes treatment affect the outcome of CRS. Fourth, the prevalence of DM was 5% in our population, which is lower than the reported 8.3% in the US population. However, no prior studies have shown whether the prevalence of diabetes in the CRS population should be higher or lower than the general US population. Our results may be due to the fact that CRS patients tend to be younger in age than those with many chronic medical conditions with a lower number of patients having yet to develop type II DM. In addition, we cannot rule out the possibility that some diabetic patients were not captured in this retrospective cohort study, but if undiagnosed diabetic patients were misclassified as non-diabetic, the effects would be biased towards the null, so the real effects would be strengthened. Finally, there may be other unknown or uncollected factors associated with postoperative QOL outcome, which were not evaluated in our study.

Conclusion

We have demonstrated in a large clinical cohort study that DM appears to have a significant impact on CRS. Compared to CRS patients without DM, patients with DM were significantly more likely to have nasal polyps, GERD, positive P. aeruginosa and other gram negative rods isolated from sinus cultures, and clinically significantly less improvement of postoperative SNOT-22 scores from baseline to 6 month follow up visits. Thus, CRS patients with DM may require special postoperative care, such as using oral steroids with caution, and clinicians treating CRS in diabetic patients may consider antibiotics with improved pseudomonal coverage. Future studies with long-term postoperative follow up and objective measures of postoperative outcomes and diabetes controls, such as nasal endoscopy scores and HbA1c, are needed to further evaluate the impact of DM on long-term CRS prognosis after FESS, as well as to determine whether DM control improves the prognosis of CRS after FESS.