BACKGROUND: Pediatric hospital-acquired (HA) venous thromboembolism (VTE) is a vexing problem with improvement efforts hampered by lack of robust surveillance methods to establish accurate rates of HA-VTE.
METHODS: At a freestanding children’s hospital, a multidisciplinary team worked to develop a comprehensive surveillance strategy for HA-VTE. Starting with diagnosis codes, we implemented complementary detection methods, including clinical and radiology data, to develop a robust surveillance system. HA-VTE events were tracked by using descriptive statistics and a statistical process control chart. Detection methods were evaluated via retrospective application of each method to every identified HA-VTE. Initial detection method was tracked.
RESULTS: A total of 68 HA-VTE events were identified and the median number of events per 1000 patient days increased from 0.18 to 0.34. No single detection method would have identified all events. Each detection method initially identified HA-VTE events.
CONCLUSIONS: Implementation of multiple detection methods has optimized timely detection of HA-VTE. This allows the establishment of a reliable baseline rate, enabling quality improvement efforts to address HA-VTE.
Hospital-acquired (HA) venous thromboembolism (VTE), including deep venous thrombosis (DVT) and pulmonary embolism, has increasingly become a focus for the prevention of harm.1,2 Although rare,3 VTE events may result in loss of vascular access, pulmonary hypertension, postthrombotic syndrome, and death.4–14 Notably, patients who develop HA-VTE also experience increased lengths of stay and subsequently higher costs.15 In his “Call to Action to Prevent Deep Vein Thrombosis and Pulmonary Embolism” in 2008, the US surgeon general called for the development of effective surveillance for VTE.2
Guidelines exist to assess risk for development of pediatric HA-VTE and to suggest strategies for prevention, including mechanical and pharmacologic prophylaxis16–22; however, surveillance methods to detect HA-VTE are not well established. The most commonly used method to detect HA-VTE uses the International Classification of Diseases, Ninth Revision (ICD-9) diagnostic codes,3,8,12,23,24 but evidence suggests this method has low specificity (23% for HA-VTE)24 and only fair sensitivity (77% in 1 study).25 The Solutions for Patient Safety network, a network of children’s hospitals working collaboratively to reduce HA conditions including VTE, has recommended that diagnostic codes should not be the sole method of VTE detection; however, the network had no clear specifications on surveillance strategies for HA-VTEs.26 Several successful detection strategies in adults have used both passive and active surveillance methods, such as ICD-9 codes (passive) and radiology reports (active).27
At our center, guidelines exist for VTE risk assessment and prevention,16,20,28 but no system existed to reliably identify HA-VTEs, which is an essential first step to ultimately reduce the incidence. Hence, we developed a surveillance system using multiple detection methods to improve the reliable and timely identification of HA-VTEs. We hypothesized that a combination of methods, including both active and passive surveillance, would optimize event identification.
Design and Setting
We conducted a cross-sectional study of identified cases of HA-VTE between December 2012 and November 2014. This study was conducted at Cincinnati Children’s Hospital Medical Center, which is a tertiary care children’s hospital. In keeping with the Solutions for Patient Safety network’s definition of HA-VTE during the study period, an event was defined as a DVT and/or pulmonary embolism that occurred >48 hours after hospital admission in patients >6 months of age on any unit or service within the hospital. Arterial and superficial venous thromboses were excluded.
The development of the surveillance system occurred in 3 phases. For each phase, an additional detection modality was added to those already in use (Table 1). During Phase 1, from December 2012 to January 2013, 22 ICD-9 codes were used to identify HA-VTE events by using a monthly report. During Phase 2, from February 2013 to October 2013, surveillance data from our local vascular access team (VAT) was added to data from ICD-9 codes. At our institution, the VAT consists of specialized nurses who perform the monitoring of all inpatient central venous catheters (CVC), which include peripherally inserted central catheters, tunneled CVCs, and nontunneled CVCs. Their expertise includes early recognition and management of CVC complications, including VTEs. A weekly report was generated from their database to identify instances of CVC occlusion or thrombosis. In Phase 3, from November 2013 to November 2014, radiologic data were added to ICD-9 codes and VAT data. The radiology reports included information from any ultrasound ordered to assess for VTE within our hospital system (eg, an ultrasound with Doppler ordered to assess for DVT in a single extremity). These reports were generated by using an automated electronic reporting system based on specific ultrasound order types.
In each phase, the captured events were initially reviewed by 2 separate team members who were nurses specializing in clinical data review. After the inclusion of radiology reports, this process included reviewing the radiologist interpretation of ordered ultrasounds to determine which were read as positive for VTE (eg, the use of the words “thrombus,” “thrombi,” “DVT,” or “VTE”). The list of events then underwent final review by a single physician observer (R.S.C) to determine if they met the definition of HA-VTE, including final determination of clinical relevance.
For each month and phase, the median number of VTE events per 1000 patient days was calculated by using the process in place at the time and a statistical process control chart was used to track the rate over time. We also described the initial method of VTE event identification during each phase and retrospectively applied each detection method to every HA-VTE event.
During the study period, we identified 68 HA-VTE events, with an average of 2.8 events per month and a range of 0 to 6 events per month. The median number of VTE events per 1000 patient days for each month is depicted in Fig 1. From Phases 1 and 2 to Phase 3, the median number of detected VTE events per 1000 patient days increased from 0.18 to 0.34. Table 2 depicts the initial detection method for each event and demonstrates that the majority of the events in Phase 3 were initially identified via radiology reports. Finally, when retrospectively applying each detection method to every event, 97.1% would have been identified by radiology reports, 88.2% by ICD-9 codes, and 38.2% by VAT reports (Table 3). Of the events that could have been identified via ICD-9 codes, 6.7% were not identified until billing during outpatient follow-up visits, which would delay detection via this method by at least 1 month. Of the 7 events (10.3%) which would have been detected by only 1 detection modality, 1 was detected by ICD-9 codes and the other 6 were detected by radiology reports. The events not detected by radiology reports reflect the detection of HA-VTE by other imaging modalities: 1 via echocardiography and 1 via a contrast study by using fluoroscopy to assess a CVC. Of these 2 events, both would ultimately have been detected by ICD-9 codes, but 1 was identified earlier by using VAT reports.
Our report demonstrates increased detection of HA-VTE events at our center through the implementation of a trimodal surveillance strategy. The addition of radiology reports allowed for identification of events that would not have been captured by using ICD-9 data alone, in addition to the more rapid detection of events. Although the addition of VAT team data did not capture any additional events, this modality allows for earlier event identification than ICD-9 codes.
There are no previous studies describing the effectiveness of detection strategies for pediatric HA-VTE. A pediatric study by Branchford et al25 demonstrated the poor sensitivity and specificity of using diagnosis codes alone to identify VTE events. Hence, the authors suggest pediatric HA-VTE surveillance employ additional methods to identify events. Two recent studies of adults support the concept of complementing administrative data with other methods. Wendelboe et al27 sought to create a surveillance system to detect VTE using passive (ICD-9) and active (review of imaging tests such as chest computed tomography or compression ultrasound) methods across inpatient and outpatient health care systems. The authors noted that ICD-9 codes were a reasonable initial screen but required the addition of active methods to confirm cases. A study in surgical patients seeking to optimize a surveillance system for postoperative VTE explored the use of diagnosis codes (ICD-9) combined with pharmacy records of anticoagulation prescriptions to identify postoperative VTE occurring in the inpatient setting.29 The authors did not comment on how the addition of pharmacy records impacted the rates of VTE detection.
Our work demonstrates that utilizing diagnostic codes alone would have resulted in us missing events at our center and the addition of VAT data did not substantially increase our event capture rate. The vast majority of our events are captured by the radiology reporting system, with a small number of events added by our additional detection modalities. One clear benefit of the radiology and VAT reporting system over the use of ICD-9 codes is that these reports are available more quickly than discharge codes, particularly if a VTE event is not captured by billing codes until a follow-up visit. Additionally, we have standardized the radiology reports to ease data abstraction and improve reliability of the surveillance system.
A limitation of our radiology process is its sole reliance on ultrasounds ordered to detect VTE, which means our radiologic data may miss events detected on other imaging modalities, such as computed tomography, echocardiography, or contrast studies of CVCs by using fluoroscopy. The expansion of our radiology system to include other imaging modalities may be a reasonable next step as we continue to refine our local surveillance system.
Additionally, since the end of our study period, the ICD-9 system has transitioned to the International Classification of Diseases, 10th Revision, and we have not reexamined the reliability of the ICD system since that change, although we have transitioned to equivalent International Classification of Diseases, 10th Revision codes.
We also considered collecting pharmacy data, specifically enoxaparin use, but given the challenges of interpreting prophylaxis versus treatment given pediatric weight-based dosing, we did not use it. However, pharmacy data may be useful for other centers seeking to implement a VTE detection system and could serve as a future addition to our process.
HA-VTE remains a safety concern for hospitalized children. Efforts to decrease rates have been hampered by a lack of reliable detection methods. In this study, we report the development of a reliable HA-VTE detection system that resulted in enhanced and more rapid detection of events. This work is foundational to all subsequent quality improvement efforts to address HA-VTE. Hence, we believe our learning may benefit other centers seeking to implement a detection system for HA-VTE in children.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- ↵Children’s Hospitals’ Solutions for Patient Safety. SPS Recommended Bundles: Hospital Acquired Condition: VTE. 2013. Available at: www.solutionsforpatientsafety.org/wp-content/uploads/SPS-Recommended-Bundles.pdf. Accessed December 22, 2014
- ↵Office of the Surgeon General (US) National Heart Lung, and Blood Institute (US). The Surgeon General’s Call to Action to Prevent Deep Vein Thrombosis and Pulmonary Embolism. Rockville, MD: Office of the Surgeon General (US); 2008. Available at: www.ncbi.nlm.nih.gov/books/NBK44178/. Accessed November 20, 2014
- Andrew M,
- David M,
- Adams M,
- et al
- Raffini L,
- Huang YS,
- Witmer C,
- Feudtner C
- Goldenberg NA,
- Donadini MP,
- Kahn SR,
- et al
- Goudie A,
- Dynan L,
- Brady PW,
- Fieldston E,
- Brilli RJ,
- Walsh KE
- Meier KA,
- Clark E,
- Tarango C,
- Chima RS,
- Shaughnessy E
- Monagle P,
- Chan AK,
- Goldenberg NA,
- et al
- ↵Multidisciplinary VTE Prophylaxis BESt TeamCincinnati Children’s Hospital Medical Center. Best evidence statement (BESt) Venous Thromboembolism (VTE) prophylaxis in children and adolescents. 2014. Available at: https://www.cincinnatichildrens.org/-/media/cincinnati%20childrens/home/service/j/anderson-center/evidence-based-care/recommendations/default/venous%20thromboembolism%20best%20181.pdf?la=en. Accessed January 27, 2014
- Raffini L,
- Trimarchi T,
- Beliveau J,
- Davis D
- Branchford BR,
- Mourani P,
- Bajaj L,
- Manco-Johnson M,
- Wang M,
- Goldenberg NA
- ↵Children’s Hospitals’ Solutions for Patient Safety. Operating Definitions October 2013. 2013. Available at: www.solutionsforpatientsafety.org/wp-content/uploads/sps-operating-definitions.pdf. Accessed November 20, 2014
- Wendelboe AM,
- Campbell J,
- McCumber M,
- et al
- Nelson RE,
- Grosse SD,
- Waitzman NJ,
- et al
- Copyright © 2017 by the American Academy of Pediatrics