Laterality of lower extremity deep vein thrombosis after colectomy: A retrospective study using the national inpatient sample

Abstract

Background and aims

Deep venous thrombosis (DVT) is known to occur preferentially on the left lower extremity. The renowned surgeon Denis Burkitt advanced the theory that a heavy sigmoid colon would compress the left pelvic veins and predispose to DVT. Our study aimed to evaluate this hypothesis by comparing the laterality distributions with and without a prior colectomy.

Methods

We conducted a retrospective analysis of the 2016 National Inpatient Sample database by stratifying the patients at any age with acute DVT of lower extremity by history of prior colectomy, thereby eliminating local gut mechanical factors in the development of DVT. We compared the laterality distribution (i.e., left, right, bilateral, and unspecified) between the patients with and without a prior colectomy. We also conducted a subgroup analysis by the sex category to examine the difference in laterality distribution for male and female patients. Chi-square test for independence was used. P value ≤0.05 was considered statistically significant.

Results

We found an estimated total of 342,525 cases. Among patients without a prior colectomy, 136,605 (41.6%) were left-sided DVT versus 119,555 (36.4%) right-sided, with 55,555 bilateral and 16,865 unspecified. Among patients with a prior colectomy, 5,750 (41.2%) were left-sided, 5,000 (35.9%) were right-sided, 2,345 were bilateral and 850 were unspecified. The laterality distribution between the two groups was not significantly different (p = .167). The left-side predominance disappeared only in males with a prior colectomy (37.1% for left vs. 38.9% for right, p = .027).

Conclusions

Our findings did not confirm the Burkitt’s hypothesis. The left-side predominance of lower extremity DVT was attenuated only in male patients with a prior colectomy.

Keywords

Deep venous thrombosis pulmonary embolism May-Thurner syndrome colectomy fiber

Introduction

Since Virchow’s time it has been observed that lower extremity and pelvic deep venous thromboses (DVT) happen more often on the left side, and this has been attributed to the crossing of the right common iliac artery over the left iliac vein, impairing blood flow by extrinsic compression. In 1957 May and Thurner, starting from clinical and surgical observations, published a cadaveric case series and concluded that intravascular structures in the left iliac vein (and not external compression by the neighboring iliac artery), which they called spurs, are the best explanation for the observed laterality in DVT.¹ Extrinsic compression cannot be completely dismissed as a component of left-sided predominance in DVT, as exemplified by the fact that 90% of DVT associated with pregnancy happens in large veins of the left side, as opposed to what is observed in the general population, with higher proportion of calf DVT and less laterality predominance.² Moreover, extrinsic pulsatile compression by the iliac artery was hypothesized as the irritating factor that led to the formation of spurs inside left iliac vein.^1,3

While in cases of provoked DVT (such as in post-operative orthopedic cases and trauma with lateralizing lesions) the predominance of DVT ipsilateral to the vessel trauma is obvious,^4,5 among unselected patients, left-side predominance is still easy to demonstrate. For example, 42 vs 39% left vs right among trauma patients without lateralizing injuries, and 57% left-sided DVT in another cohort.^5,6 Contradicting May and Thurner’s findings, Narayan et al. analyzed a cohort of patients who had undergone computed tomographic angiography of the pelvis before developing a DVT, and found that 7.2% of patients had iliac vein compression >70%, and in that group left-sided DVT was more probable with an OR of 3.03.⁷

It has long been observed that DVT is part of a group of diseases known to increase incidence steeply with population transitions to Western lifestyle, the so-called “diseases of civilization.”^8,9 Many factors have been hypothesized to explain this epidemiological transition, including poor data collection, effect of diseases that can now be easily prevented or treated, differences in physical activity levels, genetics, and diet. Decrease in intake of fiber due to modern flour and sugar processing has, along with vitamins and other substances, been proposed as an explanation for the ill health supposedly promoted by changes in diet over the last ∼150 years.^8,10 Clinical studies of fiber intake have shown conflicting results for major health benefits and even for constipation.^11–13 Overall, there is a positive effect of high fiber diet suggested by nutritional epidemiological surveys with supporting basic science mechanisms, with lacking and underwhelming evidence from prospective and clinical trial data.^14,15

The proposed mechanism for benefit is that a high-fiber diet leads to faster gut transit with less constipation, leading to chronically less straining and intrabdominal pressure overload.⁹ To increase fiber in diet would, therefore, decrease immediately the incidence of what Burkitt called “pressure diseases,” including hiatal hernia, appendicitis, diverticulitis, and lower extremity varices and DVT,^8,9 notwithstanding other proposed benefits by modulation of gut flora and increasing stomach fullness with potential decrease in total calorie intake.^16,17 A population with low dietary fiber would have slow colonic transit, constipation, therefore not only the straining (by increasing venous pressure leading to stasis) but also the heavy full sigmoid resting on the left side of pelvis could contribute to DVT, and this latter factor would at least partially explain the predominance of left-sided DVT.

A few years after Burkitt’s papers, in a letter to the British Medical Journal, a surgeon named Conrad Latto reports implementing Burkitt’s advice for fiber supplementation in a surgical service and having experienced a large decrease in DVT and pulmonary embolism for 5 years, although no actual statistics were provided.¹⁸

Except for the findings by May and Thurner,¹ proposed mechanisms to explain lower extremity predominance and left laterality of DVT are largely theoretical. Here we offer to test Burkitt's hypothesis utilizing a large inpatient database, by comparing the laterality of DVT in general inpatients with patients who have undergone colectomy. We hypothesize that patients with a history of colectomy, by eliminating the mechanical factor of a full sigmoid over the pelvic vessels, alongside a decrease in Valsalva straining to pass stools, would present an attenuated lateralization of DVT incidence, when compared with a general inpatient population.

Methods

Design and participants

This was a retrospective observational study to investigate differences in laterality of lower extremity DVT between patients with and without a history of colectomy. We used the National Inpatient Sample for the year 2016 (NIS), that is a publicly available all-payer dataset from all hospitals reporting to Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality. Use of this limited dataset does not require review by institutional review board. We included all inpatients at any age who had a diagnosis of acute DVT of lower extremity during the hospitalization. We identified a total of 68,505 cases with acute DVT of the lower extremity using the code I824 based on the International Classification of Diseases, tenth revision (ICD-10). To classify the laterality of DVT, we used the ICD-10 codes I824*1 (left), I824*2 (right), I824*3 (bilateral), I824*9 (unspecified). History of partial or total colectomy and presence of colostomy were identified by the ICD-10 code Z9049 and Z933, respectively. We also reviewed patients’ age, sex, race, length of stay (LOS), and in-hospital mortality.

Statistical analysis

We performed descriptive analysis for age, sex, race, LOS, mortality, and history of colectomy for each laterality category. One-way ANOVA, Kruskal-Wallis test, or Chi-square test was performed to examine any difference in laterality for each variable, as appropriate. Subgroup analysis by sex was also performed as we were aware that left-side predominance in DVT is prominent in females.⁶ Sex data were missing in 70 cases, leaving 68,435 cases for this subgroup analysis. Chi-square test for independence was performed to compare the proportion of each laterality for female and male patients separately. We did a sensitivity analysis comparing non-pregnant to all female patients. Querying “Z3A*” in the database identified 401 pregnant patients. There was no difference in their laterality distributions between the two groups. No additional benefit was found from multiple regression analysis accounting for age, race, and interaction between sex and colectomy. We therefore present the results from the chi-square tests with the estimated counts representing the US population by weighting the data. P value ≤.05 was considered statistically significant. We used SPSS version 26 (Armonk, NY: IBM Corp.) for all analyses.

Results

Laterality of DVT and patient characteristics

The descriptive analysis showed that the mean age was 65 ± 17 years, 49.4% were female, 70.5% were Caucasian, and 17.5% were African American. Median length of stay was 5 (IQR 3 – 10) days, and overall in-hospital mortality rate was 4.8%.

Of the estimated total of 342,525 patients, 142,355 (41.6%) were left-sided and 124,555 (36.4%) were right-sided. Bilateral and unspecified laterality were 57,900 (16.9%) and 17,715 (5.2%), respectively. History of colectomy was found in 13,945 patients (4.1%). When we compared the laterality stratified by these patient characteristics as shown in Table 1, the left-side predominance remained unchanged except pediatric patients age less than 18 (41.5% for left vs. 43.2% for right). Among the pediatric patients, a history of colectomy was found in only one case. The median age and race were similar between left- and right-sided DVTs of lower extremity. Of the estimated 342,175 patients who had a sex variable on record, left- and right-sided DVT were 43.2% and 34.9% respectively for female as compared to 40.0% and 37.8% for male. This difference in laterality distribution was significantly different (p < .001)). Female patients showed a stronger left-side predominance than male.

Table 1.

Patient characteristics per laterality of acute lower extremity DVT.

Variables		Laterality of acute lower extremity DVT								P
Variables		Left		Right		Bilateral		Unspecified		P
Age (Mean±SD)		64 ± 17		64 ± 17		67 ± 16		64 ± 17		<0.001
	AGE < 18	1235	(41.5%)	1285	(43.2%)	330	(11.1%)	125	(4.2%)	0.001
	AGE >=18	141,120	(41.6%)	123,270	(36.3%)	57,570	(17.0%)	17,590	(5.2%)
RACE	White	96,560	(41.7%)	84,465	(36.4%)	39,650	(17.1%)	11,110	(4.8%)	<0.001
	Black	23,315	(40.7%)	20,995	(36.6%)	9,755	(17.0%)	3,295	(5.7%)
	Hispanic	10,850	(42.9%)	8,965	(35.5%)	3,705	(14.7%)	1,770	(7.0%)
	Other	5,805	(41.1%)	5,005	(35.4%)	2,450	(17.3%)	865	(6.1%)
Length of stay in days (median)		5 (IQR 3–9)		5 (IQR 3–9)		7 (IQR 4–12)		5 (IQR 3–9)		<0.001
In-hospital mortality		5,525	(3.9%)	5,340	(4.2%)	4,310	(7.5%)	1,290	(7.3%)	<0.001

DVT: deep venous thrombosis; SD: standard deviation; IQR: interquartile range.

Laterality of DVT with and without a prior colectomy

For those who did not have a prior colectomy (N = 328,579), left- and right-sided DVTs were 136,605 (41.6%) and 124,555 (36.4%), respectively. Of the estimated 13,945 patients who had a prior colectomy, 5,750 (41.2%) were left-sided, 5,000 (35.9%) were right-sided. These proportions of laterality did not differ between the groups with and without a prior colectomy (p = .167). Similarly, the subgroup analysis stratified by the sex variable showed statistically insignificant results for the female group. Of 8,240 female patients who had a prior colectomy, left- and right-sided DVTs were observed in 3,640 (44.2%) and 2,780 (33.7%) cases, respectively (p = .722). On the other hand, male patients with a prior colectomy (N = 5,675), 2,105 (37.1%) were left-sided and 2,205 (38.9%) were right-sided. The left-side predominance was not present in this particular group. A statistically significant difference was noted in the laterality distribution between male patients with and without a prior colectomy (p = 0.027).

Discussion

We did not confirm our hypothesis that a history of colectomy would attenuate the laterality preference of incident DVT of the lower extremities. Our findings support prior research suggesting that either external compression by other vascular structures or intravascular anatomical abnormalities explain the left-sided predominance of lower extremity DVT.

Since the original May & Thurner report,¹ several studies detailed the anatomic characteristics of the left iliac vein in both diseased and asymptomatic populations. Cockett et al., in 1967, published a case series of patients with post-thrombotic syndrome. The overwhelming majority had left-sided symptoms, and 65% of patients had evidence of stenosis of the iliac vein at the point of crossing of the iliac artery.¹⁹ Hassell et al. published a case report of chronic left leg edema due to external compression by a tortuous left iliac artery.²⁰ Kibbe et al. published an important study in 2004, evaluating a cohort of 50 consecutive patients who underwent computed tomographic imaging for abdominal symptoms, without suspicion for DVT, and found that 33 patients had more than 25% compression of the left iliac vein, 12 had over 50% compression, and confirmed that the most common structure causing compression was the right common iliac artery (84%).³

These studies evaluating intravascular anatomical features and extravascular compression, while explaining the lateral predominance of DVT, do not address the original cause of thrombosis, therefore cannot explain the observed apparent secular trend of increased DVT incidence.²¹ Virchow’s classic triad of vessel wall, blood flow and coagulability still holds true, even with a deeper understanding of the coagulation cascade in more recent times.²² It is possible that, while anatomic relationships of the left iliac vein predispose to DVT, increased number of surgical interventions, increased risk factors including heart failure and obesity, and likely improved diagnosis with wider availability of computed tomography and ultrasound all contribute to an apparent progressive increase in population incidence of DVT.^21,23

Interestingly, our study found that males with a history of colectomy had a slight right-side predominance in DVT (39.7 vs 36.0% right vs left). Shin’s study mentioned above found 57% left-sided predominance of DVT, but the finding seemed to be restricted to females who had left-predominant DVT in 74.2%.⁶ The case-control study by Narayan et al. found that female sex had odds ratio 6.41 for compression of the left iliac vein, providing an underlying cause for the observed laterality associated with female sex.⁷ It has been proposed that the accentuated lordosis of the lower spine in females pushes the left iliac vein forward against the right iliac artery (briefly reviewed by Harbin et al.)²⁴ We did not perform adjustment for multiple comparisons because all analyses were driven by specific hypotheses.²⁵ Future studies may address whether the left-side predominance is restricted to female sex and what is the extent of its clinical significance.

Our study suggests that the mechanism hypothesized by Burkitt misses some important factors. The strengths of our study include large sample size and an innovative clinical question that prevents a systematic bias in the data (i.e., it is unlikely that clinicians would have been more or less aware of laterality of DVT in a patient with a history of colectomy, as an intact sigmoid is not a commonly suspected risk factor for left-sided DVT). Moreover, colectomy often will address conditions that feature higher risk of DVT such as colon cancer, which probably enriched our colectomy group to some degree.

A major limitation of our study is the lack of diagnosis ascertainment, as data were selected based on diagnostic codes. There were many patients with unspecified and bilateral codes for DVT (approximately 22% of the sample, see Table 2). In future studies that access more detailed data, cases of bilateral DVT can still be analyzed based on the predominant side, by applying the thrombus scoring system proposed by Shin et al.⁶ Systematic errors that could have been introduced by our data extraction methodology would be expected to decrease our power to detect a meaningful effect. It is also unclear how different DVT patients who were diagnosed and treated as outpatient are from hospitalized patients since our study examined only hospitalized patients. Still, with our very large database analysis, we conclude that any risk factor that could be reversed by colectomy, if present, would be clinically insignificant. Finally, our diagnostic codes may have included several patients with diverting colostomy but an intact sigmoid colon. To exclude such patients would not be expected to change our findings, as a diverting colostomy would make accumulation of heavy stools in the sigmoid unlikely. Finally, because ICD-10 codes do not differentiate provoked from unprovoked DVT, that is potentially worth investigating in future research.

Table 2.

Laterality of acute lower extremity DVT with and without a prior colectomy stratified by sex.

		Laterality of acute lower extremity DVT								P
	Total (N)	Left		Right		Bilateral		Unspecified		P
All cases	342,525	142,355	(41.6%)	124,555	(36.4%)	57,900	(16.9%)	17,715	(5.2%)	0.167
No colectomy	328,579	136,605	(41.6%)	119,555	(36.4%)	55,555	(16.9%)	16,865	(5.1%)
Colectomy	13,945	5,750	(41.2%)	5,000	(35.9%)	2,345	(16.8%)	850	(6.1%)
Female	169,155	73,065	(43.2%)	58,975	(34.9%)	27,985	(16.5%)	9,130	(5.4%)	0.722
No colectomy	160,915	69,425	(43.1%)	56,195	(34.9%)	26,635	(16.6%)	8,660	(5.4%)
Colectomy	8,240	3,640	(44.2%)	2,780	(33.7%)	1,350	(16.4%)	470	(5.7%)
Male	173,020	69,145	(40.0%)	65,435	(37.8%)	29,870	(17.3%)	8,570	(5.0%)	0.027
No colectomy	167,345	67,040	(40.1%)	63,230	(37.8%)	28,880	(17.3%)	8,195	(4.9%)
Colectomy	5,675	2,105	(37.1%)	2,205	(38.9%)	990	(17.4%)	375	(6.6%)

DVT: deep venous thrombosis.

In conclusion, we did not observe a significant difference in laterality distribution of DVTs with and without prior colectomy. We interpret findings to suggest that the anatomical relationship between the distal gut and the pelvic vessels is not a sole cause of the left-sided predominance that is commonly observed with lower extremity DVT. We further interpret that Burkitt’s prediction of the importance of colonic transit time and dietary fiber intake in DVT is not confirmed in the present cohort. Further studies are needed to investigate why the left-side predominance is more prominent in females and why it is attenuated in males.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical approval

Not applicable. This study is based on a publicly available database.

Guarantor

Not applicable.

Contributorship

Ivan E Saraiva: idea, background/introduction, final manuscript preparation. Hirotaka Kato: database analysis, statistical analyses, final manuscript preparation.

Acknowledgements

None.

ORCID iD

Ivan E Saraiva

References

May

Thurner

The cause of the predominantly sinistral occurrence of thrombosis of the pelvic veins. Angiology 1957; 8: 419–427.

Ginsberg

JS.

Peripheral venous disease. In: Goldman

Shafer

(eds) Goldman-Cecil medicine. 25th ed. Philadelphia, PA: Elsevier , 2015, p.512.

Kibbe

Ujiki

Goodwin

, et al. Iliac vein compression in an asymptomatic patient population. J Vasc Surg 2004; 39: 937–943.

Davidson

Mazzu

Gage

, et al. Screening for deep venous thrombosis in asymptomatic postoperative orthopedic patients using color Doppler sonography: analysis of prevalence and risk factors. AJR Am J Roentgenol 1996; 166: 659–662.

Weiss

Hayanga

Efron

, et al. Laterality of deep venous thrombosis among trauma patients: are we screening our patients adequately? J Surg Res 2007; 141: 68–71.

Shin

Park

JK.

The laterality of deep vein thrombosis in the pelvic and lower extremity veins. Vasc Specialist Int 2014; 30: 56–61.

Narayan

Eng

Carmi

, et al. Iliac vein compression as risk factor for left- versus right-sided deep venous thrombosis: case-control study. Radiology 2012; 265: 949–957.

Burkitt

DP.

Varicose veins, deep vein thrombosis, and haemorrhoids: epidemiology and suggested aetiology. Br Med J 1972; 2: 556–561.

Burkitt

DP.

Dietary fibre and “pressure diseases”. J R Coll Physicians Lond 1975; 9: 138–146.

10.

Taubes

Good calories, bad calories: fats, carbs, and the controversial science of diet and health. New York, NY: Anchor Books. 2008.

11.

Veronese

Solmi

Caruso

, et al. Dietary fiber and health outcomes: an umbrella review of systematic reviews and Meta-analyses. Am J Clin Nutr 2018; 107: 436–444.

12.

Tan

CYM

Daud

MAM

, et al. Stopping or reducing dietary fiber intake reduces constipation and its associated symptoms. World J Gastroenterol 2012; 18: 4593–4596.

13.

McRorie

Fahey

Gibb

, et al. Laxative effects of wheat bran and psyllium: resolving enduring misconceptions about fiber in treatment guidelines for chronic idiopathic constipation. J Am Assoc Nurse Pract 2020; 32: 15–23.

14.

O’Keefe

SJ.

The association between dietary fibre deficiency and high-income lifestyle-associated diseases: Burkitt’s hypothesis revisited. Lancet Gastroenterol Hepatol 2019; 4: 984–996.

15.

Hartley

May

Loveman

, et al. Dietary fibre for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev 2016; 2016: CD011472.

16.

Holscher

HD.

Dietary fiber and prebiotics and the gastrointestinal microbiota. Gut Microbes 2017; 8: 172–184.

17.

Hassanzadeh-Rostami

Faghih

Effect of dietary fiber on serum leptin level: a systematic review and meta-analysis of randomized controlled trials. Exp Clin Endocrinol Diabetes. Epub ahead of print 11 December 2019.DOI: 10.1055/a-0998-3883.

18.

Latto

Postoperative deep vein thrombosis in Nigerians on high-fibre diets. Br Med J 1979; 1: 199–199.

19.

Cockett

Thomas

Negus

Iliac vein compression.–its relation to iliofemoral thrombosis and the post-thrombotic syndrome. Br Med J 1967; 2: 14–19.

20.

Hassell

Reifsteck

Harshfield

, et al. Unilateral left leg edema: a variation of the May-Thurner syndrome. Cardiovasc Intervent Radiol 1987; 10: 89–91.

21.

Huang

Goldberg

Anderson

, et al. Secular trends in occurrence of acute venous thromboembolism: the Worcester VTE study (1985–2009). Am J Med 2014; 127: 829–839.e5.

22.

Chung

Lip

GYH.

Virchow’s triad revisited: blood constituents. Pathophysiol Haemost Thromb 2003-2004; 33: 449–454. DOI: 10.1159/000083844.

23.

Goldhaber

SZ.

Venous thromboembolism: epidemiology and magnitude of the problem. Best Pract Res Clin Haematol 2012; 25: 235–242.

24.

Harbin

Lutsey

PL.

May‐Thurner syndrome: history of understanding and need for defining population prevalence. J Thromb Haemost 2019; 18: 534–542.

25.

Perneger

TV.

What’s wrong with Bonferroni adjustments. BMJ 1998; 316: 1236–1238.