Article Text

PDF

Large-scale deployment of the Global Trigger Tool across a large hospital system: refinements for the characterisation of adverse events to support patient safety learning opportunities
  1. V S Good,
  2. M Saldaña,
  3. R Gilder,
  4. D Nicewander,
  5. D A Kennerly
  1. Baylor Health Care System Office of Patient Safety, Dallas, Texas, USA
  1. Correspondence to Donald A. Kennerly, Baylor Health Care System, Office of Patient Safety, 8080 North Central Expressway, Suite 500, Dallas, TX 75206, USA; donaldk{at}baylorhealth.edu

Abstract

Background The Institute for Healthcare Improvement encourages use of the Global Trigger Tool to objectively determine and monitor adverse events (AEs).

Setting Baylor Health Care System (BHCS) is an integrated healthcare delivery system in North Texas. The Global Trigger Tool was applied to BHCS's eight general acute care hospitals, two inpatient cardiovascular hospitals and two rehabilitation/long-term acute care hospitals.

Strategy Data were collected from a monthly random sample of charts for each facility for patients discharged between 1 July 2006 and 30 June 2007 by external professional nurse auditors using an MS Access Tool developed for this initiative. In addition to the data elements recommended by Institute for Healthcare Improvement, BHCS developed fields to permit further characterisation of AEs to identify learning opportunities. A structured narrative description of each identified AE facilitated text mining to further characterise AEs.

Initial findings Based on this sample, AE rates were found to be 68.1 per 1000 patient days, or 50.8 per 100 encounters, and 39.8% of admissions were found to have ≥1 AE. Of all AEs identified, 61.2% were hospital-acquired, 10.1% of which were associated with a National Coordinating Council – Medical Error Reporting and Prevention harm score of “H or I” (near death or death).

Future Direction To enhance learning opportunities and guide quality improvement, BHCS collected data—such as preventability and AE source—to characterise the nature of AEs. Data are provided regularly to hospital teams to direct quality initiatives, moving from a general focus on reducing AEs to more specific programmes based on patterns of harm and preventability.

  • Global Trigger Tool
  • patient safety
  • adverse events

Statistics from Altmetric.com

Concurrent with the launch of the “5 Million Lives Campaign”, the Institute for Healthcare Improvement (IHI) encouraged broad use of the Global Trigger Tool (GTT) for measuring adverse events (AEs) as a means to objectively determine and monitor patient harm that occurs throughout the healthcare system. Although the concept of “trigger tools” to focus medical record reviews and identify AEs was introduced in 1974,1 most institutions continue to rely on voluntary reporting and traditional record reviews to quantify AEs.2 Although these methods provide valuable information regarding some aspects of patient safety, they have substantial inherent limitations: comprehensive manual chart reviews are expensive and time-consuming3; computerised searches of electronic records are quicker and less expensive but, depending on the sophistication of the search strategy, may miss documented events that have been described without using the selected search terms4; and voluntary reporting systems have been found to grossly underestimate the true rate of AEs5 6 and to predominantly capture “near-miss” reports.7

More recently, trigger tools have emerged as a strategy to avoid many limitations inherent to other methods of collecting AE data in healthcare settings.8–16 The original trigger tool described by Classen et al8 9 targeted identifying adverse drug events, was fully automated and “effective but expensive and required customised software linkages to pharmacy databases”.13 It was thus impractical in many healthcare settings. A relatively low-cost and “low-tech” version was developed by a group of experts convened by IHI and Premier in 2000.13 The modified IHI/Premier tool replaced the computerised detection of triggers with manual chart review and expanded the number of triggers from 12 to 24.15 The IHI/Premier trigger tool has been used by more than 200 organisations, and is consistent, reliable and low cost.13 The IHI has also collaborated with healthcare systems such as VHA, Kaiser Permanente, Ascension Health Care and Baylor Health Care System (BHCS) to develop and implement trigger tools specific to intensive care units, ambulatory care settings, neonatal intensive care units, perioperative surgical care and perinatal care.17 18

The GTT is intended to help estimate the prevalence of AEs that occur within a hospital setting by combining high yield triggers in areas applicable to most hospitals: general care, adverse drug event, intensive care unit, surgical care, emergency department and perinatal care.2 Although the GTT provides a good means of tracking the overall harm occurring within a hospital, it is deliberately less helpful with respect to guiding attention to specific areas that are amenable to improvement. For example, the IHI-recommended GTT does not examine the extent to which an AE is preventable or ameliorable,2 or seek to determine the nature of the AE. From an operational perspective, however, there is immense value in greater characterisation of the nature of the AEs as this information can help direct limited quality improvement resources towards AEs that might be more likely to have care processes that could be productivity redesigned.

BHCS has applied the experience gained in working with the IHI and other collaborators on the Outpatient Trigger Tool to develop additional fields to be collected in conjunction with the “standard” GTT data to enhance learning from the identified AEs and improve patient safety. Although comprehensive taxonomies for AEs have been developed and published,19 the purpose of this work was to move in a practical and useful way from a primary focus on measurement to a balanced focus on measurement and learning that could be broadly applied.

Methods

Setting

BHCS is an integrated healthcare delivery system in the Dallas-Fort Worth metroplex, comprising 14 owned, leased and affiliated hospitals and >100 primary, speciality and senior healthcare ambulatory centres. AE data were collected for the eight general acute care hospitals, two inpatient speciality cardiovascular hospitals and two rehabilitation/long-term acute care hospitals. The paediatric rehabilitation facility was excluded from this analysis because the GTT tool was not designed for use in paediatric populations.

Patient population

The following patient inclusion criteria were used: the record was closed and completed (discharge summary and all coding completed), the patient was formally admitted to the hospital and had a length of stay ≥3 days and the patient was ≥18 years of age. Patients admitted for psychiatric or addictive diseases were excluded because of the low numbers of such admissions to BHCS hospitals and the heightened confidentiality standard applied to such records. A random sample of 10–35 charts per month (based on facility discharge volume) was selected using the SAS V.9.1 Survey Select Procedure for patients discharged from each of 12 BHCS facilities from 1 July 2006 to 30 June 2007. These numbers of charts represent the floor and ceiling caps on the 2% random sample chosen from each facility: where 2% was <10 charts per month or >35 charts per month, these numbers were substituted. Stated alternatively, 10 charts per month were selected for hospitals with annual admissions of <6000, 35 charts per month were selected for hospitals with annual admissions of >21 000 and a 2% sample was selected for hospitals with intermediate annual admissions. The lower limit to the number of charts reviewer was applied to ensure data were not too sparse for meaningful analysis, and the upper limit in acknowledgement that chart review is a costly and time-consuming endeavour and that this initiative was subject to resource limitations. In comparison, the IHI recommends reviewing 10 charts per hospital twice a month, irrespective of hospital volume, and finds little additional benefit in reviewing >40 charts per month.2

Study measures

We examined the three AE rate measures defined by the GTT, recognising that different representations of the data are better suited to certain purposes: AEs/1000 patient days (the traditional measure), AEs/100 encounters (a measure that was used in preliminary trigger tool studies and which is more readily understood by leadership without extensive background in clinical quality and safety research) and the percentage of admissions during which the patient experienced at least one AE (a measure that can easily be interpreted by lay leadership and the general public and defined as the number of inpatient encounters with at least one event/total records reviewed×100).2 20

Data collection

A Microsoft Access tool was developed for data collection and entry and pilot tested by the lead nurse reviewer for the project and the BHCS GTT initiative project manager by reviewing the same 10 charts from a single BHCS facility. Figure 1 shows the main data entry screen after revision based on user feedback.

Figure 1

Screenshot of the data entry screen for the Microsoft Access data collection tool developed as part of the Baylor Health Care System refinement of the Institute for Healthcare Improvement's Global Trigger Tool.

Audit process

Randomly selected patient records meeting the inclusion criteria were reviewed by external nurse auditors trained on the use of the GTT and the Microsoft Access tool. Initial training included face-to-face didactic sessions, data entry and review of standardised training cases, and discussions of case evaluations, lasting approximately 12 h. This training programme was developed by a team consisting of a physician, a nurse and the BHCS GTT project manager, and was delivered by the project manager (who has 10 years of patient safety experience within BHCS, including a substantial component relating to chart review). Additional training, in the form of face-to-face sessions and conference calls, has been provided as needed after revisions to the data collection tool and audit or data management processes. Reviewers applied the GTT to the medical record for a maximum of 20 min per record (including documentation). Data were submitted to the BHCS Office of Patient Safety monthly and reviewed by the Office of Patient Safety team, at which point final consensus regarding the occurrence, severity, type and source of each AE was reached. At the end of the first 6 months of data collection, preliminary results were reviewed and a round-table discussion with reviewers regarding their experience with the GTT and data collection tool was held. Based on these events, the review process was modestly refined and clarified resulting in the flow chart presented in figure 2. Specifically, revisions were made to ensure consistency for incomplete medical records and for records with more positive triggers than AEs (resulting in a situation where individual AEs were being recorded multiple times under different triggers). In addition, the “recommended review order” for the medical record was revised (from the original IHI recommendations) based on reviewer experience.

Figure 2

Review process used by chart auditors for the Baylor Health Care System's large-scale deployment of the refined Global Trigger Tool.

Results

A total of 2369 admissions were reviewed for discharges between 1 July 2006 and 30 June 2007. AE rates, as estimated by each of the three measures recommended by the IHI, are shown in table 1 both for all AEs and for only those AEs that occurred during hospitalisation.

Table 1

Adverse event (AE) rates for Baylor Health Care System (BHCS) acute care hospitals (1 July 2006–30 June 2007) using the Institute for Healthcare Improvement's Global Trigger Tool2

Of the AEs identified in this sample, 38.8% of the AEs were present on admission and 61.2% of the AEs occurred during hospitalisation. Of those encounters with at least one recognised AE, 352 (37.3%) had ≥1 AE present on admission and none that were hospital acquired; 518 (54.9%) had ≥1 AE that was hospital acquired and none that were present on admission; and 74 (7.8%) ≥1 AE that was present on admission and ≥1 AE that was hospital acquired. Table 2 shows the distribution of hospital acquired AEs by National Coordinating Council – Medical Error Reporting and Prevention harm score.21

Table 2

Distribution of identified adverse events by National Coordinating Council–Medical Error Reporting and Prevention harm score21

Discussion

Based on anecdotal discussions, concerns regarding the audit time and cost associated with GTT implementation, the non-specificity of the results and the limited value of the resultant data have led some facilities to stop or delay its use. Therefore, BHCS's key goal, beyond the establishment of baseline AE rates, has been the identification of learning opportunities from the GTT audit process that could drive productive improvements to care. The GTT/audit process has, as envisioned by the IHI, resulted in important baseline performance data both for organisations as a whole and for each hospital. Although these data have prompted important leadership discussions, the collection of additional data about the nature of each AE is felt to be at least as valuable.

Because resources to apply to process improvement are limited, focus is critical. The scope of data collected via the GTT audit process was expanded to achieve this focus related to BHCS AE data. Physicians, nurses, and leaders within our organisation were concerned by the fact that the original GTT tool did not distinguish between AEs that were present on admission versus those that occurred during hospitalisation. Although professionals and institutions carry broad responsibility for care and need to be aware of the prevalence of all AEs, focus on hospital-acquired AEs was an organisational priority. Therefore, once an AE was identified, the reviewer determined whether the AE was present on admission or was acquired during the hospitalisation. That 38.8% of AEs were present on admission is notable and is being separately reviewed by clinical groups that refer patients to BHCS hospitals.

This focus has intensified interest in examining AEs identified by the GTT audit process in greater detail to identify opportunities for improvement. BHCS therefore simultaneously collects data relating to preventability (preventable, probably preventable, possibly preventable, not preventable, unable to determine) and potential contributing factors (patient assessment, work environment, patient-related, communication, equipment/device, delayed treatment, delayed diagnosis, other, unknown) for the AEs identified through the GTT audit, as well as requiring the nurse auditors to provide a narrative description of the AE following the “Situation-Background-Assessment” format. The detailed characterisation of AEs based on these data is beyond the scope of this article and will be presented separately. Our current model for organisational learning from AEs is shown in figure 2.

An important facet of using the GTT as a learning tool to drive quality improvement is characterising AEs and documenting them using the Microsoft Access tool developed for this project. Although the GTT can be successfully applied in a paper-based context, the electronic data collection tool offers substantial advantages. It creates and facilitates a structured review process by prompting reviewers to complete all relevant fields. Electronic data entered by the nurse auditors can be directly transmitted to and integrated with a centralised database, eliminating the need for a separate data entry step, reducing the opportunities for data entry errors. The electronic format also facilitates comparison of AE rates between hospitals by trigger module or AE severity—for example, which can provide useful information regarding opportunities to improve quality. A very important refinement made to the GTT is felt to be the mandatory structured description of AEs identified by chart auditors. This narrative electronic data allow text mining analysis that in turn provides additional detail regarding the nature of AEs for quality improvement and/or to revise the audit tool or process to capture important information regarding AEs. In addition, electronic data can help ensure data quality. For example, during review of the data collection process conducted 6 months after the GTT was first implemented, a problem was identified relating to charts with multiple positive triggers related to a single AE. Under this scenario, some auditors were recording the single AE multiple times. The decision was made to record only the most relevant trigger to avoid inaccurate estimates of the prevalence of AEs. Using this Microsoft Access tool, patients with multiple positive triggers associated with an AE were easily identified and any instances where a single AE was recorded multiple times were corrected.

Our initial data have resulted in a perceived value for consistently categorising AEs to inform prioritisation of hospital-level and system-level improvement directed by specific AE review teams. Text mining of the initial comments is being undertaken to determine the types of AEs present on admission and those that occur during hospitalisation. Once the key categories have been developed, assigning the AE to a category will be added to the chart review/data collection process. Reporting AEs by category will further enhance prioritisation of improvement work and help direct quality initiatives undertaken by both outpatient and inpatient practitioners. Although this method has been initially applied in a US environment, its basic principles—like the IHI's GTT that is being applied and/or has corollaries in the United Kingdom, Canada, and Australia—are broadly applicable. Although minor adaptations to the local data sources may be needed for other settings within the USA and abroad, the BHCS adaptations and additions to the GTT data collection process and use provide a patient safety organisational learning model for healthcare providers in general.

Acknowledgments

The authors thank KDJ Consultants for their work in collecting the data related to this project, Robert Page, MPA, and Flor De Maria Valle for the development of the Microsoft Access data collection tool; Brian Adams for data management; S Quay Mercer, MT(ASCP), for sharing her expertise related to the Outpatient Trigger Tool and assistance in developing the auditor training program; and Briget da Graca, MS, ELS, for background research and writing assistance in preparing this article.

References

View Abstract

Footnotes

  • Funding This work was paid for through the Baylor Health Care System operational funds.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.