Introduction

Deep vein thrombosis (DVT) after TKA is a common and relatively benign treatable condition, while pulmonary embolism (PE) is frequently life-threatening and sometimes fatal. The literature suggests a substantial variation exists in the rates of DVT and PE in different countries after TKA [23]. Several recent Asian reports [13, 24, 30, 34] suggest an incidence of PE of 0% to 1.3% and an incidence of DVT of 26.6% to 60.8% after TKA without chemoprophylaxis. A number of studies [7, 8, 12, 15, 18, 24, 26, 27, 29] show the incidence of DVT and PE is relatively lower in Asian countries than in Western countries. However, several recent studies [5, 13, 42] suggest the incidence after TKA in Asian patients is comparable to that in Western populations. One meta-analysis focusing on the rate of DVT and PE after major hip and knee surgery from Asia [23] reported lower rates of proximal DVT (8.7% after TKA and 9.6% after hip surgery) in Asian countries than in Western countries (> 20% after major hip surgery) [12, 16]. Furthermore, the authors questioned the true advantage of routine chemoprophylaxis in the Asian population since they found no reported deaths due to PE regardless of a possible trend of increasing incidence of proximal DVT. That study, however, included not only hip and knee arthroplasties but also hip fracture surgery. Moreover, it was unclear whether the incidence of PE and DVT without chemoprophylaxis after TKA is similarly low among different Asian countries.

We therefore determined (1) the overall incidence of symptomatic PE and DVT after TKA without chemoprophylaxis in Asian populations, (2) whether the rates of symptomatic PE and proximal DVT have increased over time in Asian patients, and (3) the differences in the incidence of symptomatic PE and DVT among different Asian countries through a meta-analysis.

Search Strategy and Criteria

We performed a systematic review and meta-analysis under the Meta-analysis of Observational Studies in Epidemiology guidelines [38] and the Cochrane Reviewers’ Handbook [17]. We systematically searched the PubMed, Embase, Cochrane Library, Web of Science, and Google Scholar websites for publications published between January 1996 and December 2011. The database search was performed using the terms shown in Appendix 1. We applied the search to English language titles and abstracts. The reference lists of identified studies and key review articles were also searched. Using this search strategy, we identified 5432 articles (Fig. 1).

Fig. 1
figure 1

A flowchart shows the selection process of studies.

Full size image

Of these, we included only randomized clinical trials and prospective cohort studies reporting the incidence of DVT and PE after TKA in the Asian population. DVT had to be confirmed by venography, ultrasonography, or CT venography (Table 1). Imaging studies for DVT had to be the routine practice after TKA. For symptomatic PE, clinical symptoms had to be determined by perfusion or ventilation scintigraphy, pulmonary angiography, or spiral CT. Studies were excluded if patients received anticoagulants prophylaxis or mechanical prophylaxis (eg, foot pumps, intermittent pneumatic compression devices, etc) during or after surgery. Three authors (HSK, HJL, SSS) independently screened the 5432 titles and abstracts and reviewed the published articles to assess suitability for inclusion. Disagreements were resolved by consensus. After removing nonscientific or basic research articles not related to the topic, we identified 246 studies. There were 154 studies after removing duplicates. Forty of 154 studies satisfied the inclusion criteria. Two of us (WSL, SSS) then reviewed the full text of 40 articles. We excluded 22 articles based on overlapping cohorts, review article, inadequate data, mechanical prophylaxis, and inadequate imaging studies (Fig. 1). These 22 exclusions left 18 studies (three randomized controlled trials and 15 prospective observation studies) involving 1947 patients [2–5, 10, 13, 14, 21, 24, 28, 29, 32, 35, 39–42, 45]. Of the 18 studies, six were conducted in Taiwan [3, 39–42, 45], three in Korea [2, 24, 28], two in Japan [13, 14], two in Thailand [5, 35], two in Singapore [4, 32], and one each in Hong Kong [29], Malaysia [10], and India [21] (Table 1). Two authors (WSL, KIK) assessed study quality using the Newcastle-Ottawa Scale [43], which evaluates studies on selection, comparability, and outcome in cohort studies. Inconsistencies were resolved by consensus between the two authors. No difference in quality was observed between the studies.

Table 1 Summary of all studies included in the meta-analysis
Full size table

For data extraction, we used the following information from each study: midyear of surgery, number of patients, patient characteristics, country, imaging methods for DVT and PE, and thromboembolic events (Table 1). To assess DVT and PE events, we recorded cases of overall DVT, proximal DVT, distal DVT, symptomatic DVT, symptomatic PE, fatal PE, and death after surgery. Overall DVT was defined as DVT confirmed by routine radiographic studies regardless of DVT symptoms. Proximal DVT was defined when thrombosis developed in the popliteal vein and above. If thrombi were detected in both proximal and distal veins, DVTs were classified as proximal DVT. Two reviewers (WSL, KIK) extracted all data independently, according to the selection criteria. Inconsistencies in the extracted data were settled by consensus.

To identify the incidences of overall DVT, proximal DVT, distal DVT, symptomatic DVT, and symptomatic PE, we pooled DVT and PE events among the studies using random-effects or fixed-effects models, depending on the heterogeneity of the included studies. The random-effects model was utilized by applying the method of DerSimonian and Laird [9], and the fixed-effects model was utilized by applying the method of Mantel and Haenszel [31]. We tested for heterogeneity by calculating the I2 tests and interpreting I2 values of greater than 50% as indicating heterogeneity [17]. We found heterogeneity among the studies for overall DVT (I2 = 98.9%), proximal DVT (I2 = 87.2%), distal DVT (I2 = 98.6%), and symptomatic DVT (I2 = 95.9%). Therefore, we used the random-effects model to pool the DVT events. There was no heterogeneity among the studies for symptomatic PE (I2 = 0.0%). We used the fixed-effects model to pool the symptomatic PE events. Subgroup analyses were conducted based on midyear of research and country in each study. To determine whether the incidences of symptomatic PE and proximal DVT in Asian populations changed over time, we used the year of surgery or research rather than the year of publication in each study. Because publishing a paper takes a long time, sometimes many years, time periods during which surgeries were performed were believed to be a more valid surrogate for time. Therefore, to assess temporal trends in the incidence of symptomatic PE and proximal DVT, we enrolled 17 studies and stratified the midyear of surgery of each study (range, 1993–2008) into three time periods: 1993 to 1998 (414 patients) [10, 14, 32, 40], 1999 to 2002 (917 patients) [3, 13, 21, 28, 29, 35, 39, 41, 42], and 2003 to 2008 (564 patients) [2, 4, 5, 24]. The incidence ratios (IRs) of proximal DVT and symptomatic PE at the last two time periods were compared with the first time period as a reference. For the analysis by country, there had to be at least two studies from a country and the pooled number of patients had to be more than 100. The incidences of overall DVT, proximal DVT, and symptomatic PE in other Asian countries with available data were compared with those of Taiwan, from which the greatest number of studies were reported.

To assess the publication bias, we conducted funnel plots and visually evaluated their symmetry. Rates of overall DVT were broadly symmetrical and rates of PE were narrowly symmetrical. Statistical analyses were conducted using Stata® 10 software (StataCorp LP, College Station, TX, USA).

Results

Symptomatic PE was rare and the incidence for proximal DVT was low. The incidence of symptomatic PE was 0.01% (95% CI, 0.00–0.07) in 16 studies [3–5, 10, 13, 14, 21, 24, 28, 29, 32, 35, 39–42], which included 1838 patients (Fig. 2). No fatal PE was reported in any of these studies. The incidence of overall DVT (asymptomatic and symptomatic) after TKA was 40.4% (95% CI, 28.5–52.3). The incidences varied from 0% to 76% depending on the study (Fig. 3). Sixteen studies [2–5, 13, 14, 21, 24, 28, 29, 32, 35, 39–42], which included 1861 patients, reported incidences of proximal DVT and distal DVT, which were 5.8% (95% CI, 3.7–7.9) (Fig. 4) and 26.6% (95% CI, 23.6–29.6) (Fig. 5), respectively. Seven studies [3, 24, 28, 35, 39–41], which included 917 patients, provided data on incidence of symptomatic DVT: 1.9% (95% CI, 0.6–3.1) (Fig. 6).

Fig. 2
figure 2

The forest plot shows rates of symptomatic PE with a table of summarized data in Asian countries. The incidence of symptomatic PE was extremely low (0.01%; 95% CI, 0.00–0.07).

Full size image
Fig. 3
figure 3

The forest plot shows rates of overall DVT with a table of summarized data in Asian countries. The incidence of overall DVT after TKA was 40.4% (95% CI, 28.5–52.3).

Full size image
Fig. 4
figure 4

The forest plot shows rates of proximal DVT with a table of summarized data in Asian countries. The incidence of proximal DVT after TKA was 5.8% (95% CI, 3.7–7.9).

Full size image
Fig. 5
figure 5

The forest plot shows rates of distal DVT with a table of summarized data in Asian countries. The incidence of distal DVT after TKA was 26.6% (95% CI, 23.9–29.6).

Full size image
Fig. 6
figure 6

The forest plot shows rates of symptomatic DVT with a table of summarized data in Asian countries. The incidence of symptomatic DVT after TKA was 1.9% (95% CI, 0.6–3.1).

Full size image

The incidences of symptomatic PE and proximal DVT in more recent two time periods were similar to those of the first time period, and no increasing temporal patterns were observed between the earlier and more recent time periods. The incidence of symptomatic PE was 0.017% from period 2003 to 2008 [4, 5, 24], 0.012% in 1999 to 2002 [3, 13, 21, 28, 29, 35, 39, 41, 42], and 0.017% in 1993 to 1998 [10, 14, 32, 40] (Fig. 7). With the period 1993 to 1998 as a reference, we identified similar incidences of symptomatic PE between the periods 2003 to 2008 (IR = 1.00; 95% CI, 0–∞) and 1999 to 2002 (IR = 0.71; 95% CI, 0–∞). The incidences of proximal DVT were 5.92% from 2003 to 2008 [2, 4, 5, 24], 5.47% from 1999 to 2002 [3, 13, 21, 28, 29, 35, 39, 41, 42], and 6.69% from 1993 to 1998 [14, 32, 40] (Fig. 7). With the period 1993 to 1998 as a reference, there were no differences in incidences of proximal DVT between patients observed from 1999 to 2002 (IR = 0.82; 95% CI, 0.52–1.28) and 2003 to 2008 (IR = 0.89; 95% CI, 0.55–1.43).

Fig. 7
figure 7

A graph shows the incidences of symptomatic PE and proximal DVT stratified by midyear of study. The incidences of symptomatic PE and proximal DVT did not increase over time in Asian patients.

Full size image

Among the different Asian countries, the incidences of symptomatic PE were similar, but the incidences of overall and proximal DVT were dissimilar. Compared with Taiwan as a reference, there was no difference in incidences of overall DVT in Japan (IR = 1.02; 95% CI, 0.88–1.18), but there were lower incidences of overall DVT in Thailand (IR = 0.78; 95% CI, 0.66–0.97), Korea (IR = 0.51; 95% CI, 0.44–0.59), and Singapore (IR = 0.24; 95% CI, 0.17–0.34) (Fig. 8). We identified higher incidences of proximal DVT in Japan (11.04%) (IR = 3.11; 95% CI, 1.73–5.57) and Thailand (13.05%) (IR = 3.67; 95% CI, 2.03–6.64) but similar incidences in Korea (4.76%) (IR = 1.34; 95% CI, 0.76–2.36) and Singapore (3.46%) (IR = 0.97; 95% CI, 0.44–2.15), compared with Taiwan (3.55%) (Fig. 8).

Fig. 8
figure 8

A graph shows the reported incidences of symptomatic PE, proximal DVT, and overall DVT per country. The incidences of overall and proximal DVT were different among Asian countries, but the incidences of symptomatic PE were similarly low.

Full size image

Discussion

In contrast to the Western literature, which has typically reported the incidence of PE and DVT after TKA with chemoprophylaxis, most Asian literature has reported the incidence without chemoprophylaxis [7, 8, 12, 18, 24, 26, 27, 29]. This may reflect a low incidence of DVT and PE after TKA in Asian populations even without chemoprophylaxis. Only one meta-analysis study of the incidence of PE and DVT after major orthopaedic surgery has been reported from Asia so far and it suggested rates of proximal DVT and symptomatic PE lower than those in Western reports [23]. However, that study included not only TKAs but also THAs and hip fractures. Since the incidence and pattern of PE and DVT after TKA, THA, and hip fracture are not same [11, 12, 16, 44], that study could not give a clear indication of the incidence of PE and DVT after TKA. It has been also unclear whether the incidence of PE and DVT after TKA is similarly low among Asian countries. Furthermore, several recent studies have reported the incidence after TKA in Asian patients is increasing [5, 13, 42]. We therefore determined (1) the overall incidence of symptomatic PE and DVT after TKA without chemoprophylaxis in the Asian population, (2) whether the rates of symptomatic PE and proximal DVT have increased over time in Asian patients, and (3) the incidences of symptomatic PE and DVT among Asian countries through a meta-analysis.

In our systematic survey, we identified a number of limitations in the literature, and our survey methods were also associated with some limitations. First, methods for detecting DVT were not the same in the 18 studies enrolled. Venography was used in 12 studies, ultrasonography in five studies, and CT venography in one study (Table 1). Although invasive venography was considered the classical standard in the diagnosis of DVT, the noninvasive ultrasonography now is widely used and has been proven to have almost equal sensitivity and specificity [1, 46]. Furthermore, the American College of Radiology Appropriateness Criteria® recently recommended the use of ultrasonography for initial evaluation of DVT rather than venography [19]. Second, a limited number of studies have been published in each country. Although we did our best to find more publications from more countries, we used only eight countries with 18 articles to assess the differences in incidences among the countries. We found no additional literature satisfying our inclusion criteria, such as no prophylaxis and routine radiographic evaluation for DVT regardless of any symptoms. Third, our study does not convey information on the effects of potential risk factors for DVT or PE, such as age, sex, obesity, types of anesthesia, or operation time, because relevant data were not available in most included studies. Therefore, we cannot suggest risk factors for DVT or PE. Finally, we lacked control data from Western countries to compare differences in the incidence of PE and DVT between Asian and Western populations. Although we considered inclusion of control groups in Western patients, we found no comparable study without prophylaxis after TKA in Western patients. Hence, we only briefly mentioned the difference compared with the usual Western literature for the reference.

Our meta-analysis showed the incidence of overall DVT was 40.4%, but the incidences of proximal and symptomatic DVT were 5.8% and 1.9%, respectively. Since distal DVT is typically stable and usually does not progress to PE [22, 27, 33, 36], most surgeons focus on proximal DVT. Moreover, the incidence of symptomatic PE was 0.01% and there were no deaths related to DVT or PE. Our findings were similar to those reported in one meta-analysis from Asia [23], although that study included not only TKAs but also THAs and hip fracture surgeries. While we found no control (without chemoprophylaxis) studies of Western patients, our findings showed incidences of symptomatic PE and proximal DVT after TKA that seem low compared to those in Western studies [15] reporting an incidence of DVT of 41% to 85% without prophylaxis and an incidence of proximal DVT of 5% to 22%. Furthermore, we found no death related to PE and DVT in the Asian literature. The lower incidence of PE and DVT has been attributed to the low prevalence of obesity, the low frequency of hyperlipidemia, and the absence of some genetic factors in the Asian population [25, 26, 28]. Differences in the dietary patterns (vegetarian versus nonvegetarian diets) have also been suggested as one of the causes of the low risk of DVT [30]. Kim and Kim [25] suggested the absence of thrombophilic polymorphisms with low clinical prothrombotic risk factors make the occurrence of PE after THA rare in the Asian population. Similar low rates are seen in Asians living in the West [37]. However, the reasons why the incidence of symptomatic DVT or PE is low in Asian patients after TKA even without chemoprophylaxis are still not clear.

Recent epidemiologic studies [6, 30, 34 ] have reported relatively high incidences of DVT and PE in the Asian population after TKA. Moreover, a previous meta-analysis from Asia [23] suggested a possible trend toward increasing incidence of proximal DVT. The Scottish Arthroplasty Project showed PE and DVT rates that have remained unchanged between 1992 and 2001 despite improved prophylaxis usage during this period [20]. In contrast, one Western systematic review article [44] reported a decreasing trend of development of proximal DVT after TKA with time. Finally, we found no evidence of a recent increasing tendency of PE and DVT in Asia, which differed from the results of the previously mentioned Asian meta-analysis [23]. That study enrolled 14 studies, with midyear of patient recruitment ranging from 1979 to 2003 for assessment; among these 14 studies, only seven studies related to TKA, with midyear ranging from 1993 to 2003. Moreover, that study enrolled not only TKAs but also THAs and hip fractures for assessing the temporal trend. We performed our assessment by focusing purely on TKAs with longer observation periods and collected from the most recent literature. These might be reasons why the trend of incidence of PE and DVT over time was different between these two meta-analyses from Asia.

The comparisons among Asian countries suggest, despite the overall low incidence of symptomatic DVT or PE in the region, considerable variances may exist among Asian countries. Although Malaysia had the highest incidence of DVT and India had the lowest incidence (Table 1), we identified only one study in each country that matched our criteria, and each sample was small (< 100 patients). Excluding these countries plus Hong Kong with a small sample size as well, there was a difference in the incidence of DVT among countries, such as Singapore (12.7%), Korea (26.9%), Thailand (42.5%), and Japan (54.2%), compared to Taiwan (53.2%). We identified no clear reason why the various countries had different reported incidences. Two studies [2, 35] had a much higher incidence of PE than any other Asian countries because the studies reported the incidence of PE with asymptomatic PE included while other studies [3–5, 10, 13, 14, 21, 24, 28, 29, 32, 39–42] reported only the incidence of symptomatic PE. Therefore, if we exclude asymptomatic PE in the previously mentioned studies [2, 35], the incidence of symptomatic PE might be similar in each country; specifically, it was extremely low. Considering the differences among Asian countries, data acquired from each country might be considered in developing optimal DVT prevention guidelines for a country.

In conclusion, our study demonstrates the reported incidences of symptomatic PE and proximal DVT after TKA were low even without chemoprophylaxis in Asian patients. No increasing temporal patterns were identified for the incidence of symptomatic PE or proximal DVT, regardless of the Westernizing lifestyles and an increasingly aging populace. Hence, routine chemoprophylaxis following Western protocols after TKA is still debatable in Asian populations. Further investigation with large randomized studies is needed to confirm our results and identify the risk factors that can lead to PE and DVT.