Estimating the Incidence of Symptomatic Postoperative Venous Thromboembolism: Title and subTitle BreakThe Importance of Perspective

Venous thromboembolism (VTE) is a major health care problem, with an incidence of 1 per 1000 person-years.¹ Sudden death is the initial clinical presentation for nearly one-quarter of patients with pulmonary embolism. VTE frequently recurs and is associated with long-term complications, including postthrombotic syndrome and pulmonary hypertension. Despite extensive efforts to identify VTE risk factors and to develop and implement effective and safe VTE prophylaxis, the incidence of VTE has not changed significantly.¹ The large number of potentially preventable US deaths attributable to VTE prompted the 2008 Surgeon General's Call to Action to Prevent Deep Vein Thrombosis and Pulmonary Embolism.² Recent hospitalization for major surgery accounts for almost one-quarter of all incident VTE events, and total hip replacement (THR) and total knee replacement (TKR) surgery are 2 of the surgeries posing the highest risk for VTE.¹

The Institute of Medicine considers failure to provide appropriate VTE prophylaxis to hospitalized, at-risk patients a medical error. The Agency for Healthcare Research and Quality (AHRQ) ranks VTE prophylaxis as among the most important interventions to improve patient safety.³ In 2002, AHRQ defined a set of Patient Safety Indicators for identifying potential episodes of compromised patient safety, including prevention of VTE, based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) hospital discharge diagnosis codes.⁴ In 2006, the National Quality Forum/The Joint Commission published performance measures for in-hospital VTE prophylaxis,⁵ and The Joint Commission, AHRQ, and the Centers for Medicare & Medicaid Services have established quality measurement, quality improvement, and pay-for-performance initiatives for in-hospital VTE based on specific ICD-9-CM hospital discharge diagnosis codes.⁶ Thus, establishing rates of symptomatic VTE before hospital discharge can be a useful benchmark for quality improvement.

In this issue of JAMA, Januel and colleagues⁷ report findings from a systematic review and meta-analyses that estimate the in-hospital incidence of acute symptomatic VTE among patients receiving appropriate prophylaxis after total or partial hip arthroplasty (TPHR) and total or partial knee arthroplasty (TPKR). The authors propose these estimates as benchmark values for quality measurements and for counseling patients contemplating elective hip or knee replacement surgery. The meta-analysis included clinical trials and observational studies of adults undergoing elective TPHR or TPKR who received low-molecular-weight heparin or a direct or indirect thrombin or factor Xa inhibitor as VTE prophylaxis. Rates of symptomatic postoperative VTE before hospital discharge were 0.53% and 1.09% after TPHR and TPKR, respectively.

However, rates of symptomatic VTE before hospital discharge after THR or TKR may be suboptimal safety indicators, because the period of VTE risk extends beyond the length of hospitalization for surgery. In a large California administrative data set (1991-1993), the 91-day cumulative incidence rates for symptomatic VTE among 19 586 patients undergoing THR and 24 059 patients undergoing TKR were 2.8% and 2.1%, respectively.⁸ Median times to symptomatic VTE after THR and TKR were 17 days and 7 days, respectively. Mean lengths of hospital stay were about 7 days, and 76% of THR-associated VTE events and 47% of TKR-associated VTE events occurred after hospital discharge. Using Medicare claims data from beneficiaries who underwent THR (n = 12 956) during 1995-1996, the 1- and 6-week incidence rates of pulmonary embolism were 0.3% to 0.4% and 0.7% to 0.8%, respectively.⁹ Similarly, among Medicare beneficiaries undergoing primary (n = 124 986) or revision (n = 11 726) TKR during the year 2000, the 90-day cumulative incidence rates of pulmonary embolism were 0.8% and 0.5%, respectively.¹⁰

Although these estimates may be somewhat outdated because of changes in practice over time, and their accuracy is potentially limited by uncertainty regarding ICD-9-CM hospital discharge coding practices and the type and duration of VTE prophylaxis rendered, contemporary cohort studies that used currently recommended VTE prophylaxis regimens provide similar estimates.¹¹

Based on the studies conducted to date,^{8 - 11} the postoperative period of risk for VTE after THR and TKR is about 10 to 12 and 4 to 6 weeks, respectively, and substantially beyond the period of initial hospitalization for surgery. The 3-month cumulative incidence of symptomatic VTE and pulmonary embolism after elective THR is 2.5% to 3.4% and 1.1%, respectively, and the rate of fatal pulmonary embolism is 0.22%. Corresponding rates of symptomatic VTE and pulmonary embolism after elective TKR are 1.8% to 2.4% and 0.8%, respectively, and the rate of fatal pulmonary embolism is 0.15%. From the perspective of the patient contemplating elective THR or TKR, these cumulative rates are likely to be most important for decision making. Similarly, these 3-month VTE incidence rates should be most important for the purposes of quality improvement and pay-for-performance initiatives. However, no validated, simple, and convenient method is available for collecting individual-level outcomes data after hospital discharge.

Accepting this limitation, how accurate are the estimates of in-hospital VTE rates after TPHR and TPKR in the study by Januel et al?⁷ Of the patients included in their meta-analysis, more than 80% were enrolled in randomized clinical trials. The generalizability of the estimated in-hospital VTE rates to all patients undergoing TPHR and TPKR is uncertain. Because the authors did not have individual patient–level data on the dates of surgery, postoperative VTE events, and death or other loss to follow-up, their VTE rates are not adjusted for differing periods of observation. Since clinical trials typically mandate some form of leg vein imaging between postoperative days 7 and 10 and patients with identified asymptomatic DVT were usually treated, the study by Januel et al likely underestimated the true rate of symptomatic VTE for the reported mean durations of follow-up (13 days). However, because the current duration of hospitalization for TPHR and TPKR is only 3 to 4 days, the current rates of symptomatic VTE prior to hospital discharge likely are lower than those reported by Januel et al.

Given these limitations and the poor operating characteristics of AHRQ Patient Safety Indicators in correctly identifying postoperative in-hospital VTE events,¹² the appropriateness of in-hospital VTE after THR or TKR as a patient safety indicator for quality improvement and pay-for-performance initiatives is uncertain. As pointed out by Januel et al,⁷ future efforts should be directed toward enlarging the period of observation to accurately collect VTE outcome events that reflect the entire period of VTE risk after THR and TKR. A longitudinal electronic health record may allow better surveillance for medical errors, complications, and events that occur both during and after hospital discharge.