Achieving and Sustaining Ventilator-Associated Pneumonia–Free Time among Intensive Care Units (ICUs): Evidence from the Keystone ICU Quality Improvement Collaborative
Our retrospective analysis of the Michigan Keystone intensive care unit (ICU) collaborative demonstrated that adult ICUs could achieve and sustain a zero rate of ventilator-associated pneumonia (VAP) for a considerable number of ventilator and calendar months. Moreover, the results highlight the importance of adjustment for ventilator-days before comparing VAP-free time among ICUs.
Ventilator-associated pneumonia (VAP) is a common healthcare-associated infection (HAI) that afflicts up to 20% of intensive care unit (ICU) patients who received mechanical ventilation.1 Moreover, VAP is associated with increased mortality rates, hospital lengths of stay, and costs of care.1 Multiple professional societies have issued clinical practice guidelines to prevent VAP that emphasize multifaceted strategies, such as prevention bundles.2,3 When providers have adhered to guideline recommendations, there have been significant reductions in ventilator days, ICU length of stay, VAP rates, and mortality.4,5
Prevention of VAP and other HAIs has become a national public health priority. The Centers for Medicare and Medicaid Services placed VAP on a list of nonreimbursable complications.6 Their optimum goal is to eliminate VAP and to sustain this result over an extended period of time. Nevertheless, many question whether zero is a realistic target, how long ICUs can remain free of VAP, and whether zero cases of VAP can be achieved in diverse ICU settings.6,7 The objective of this study was to evaluate whether diverse ICUs in the Michigan Keystone ICU project achieved and sustained a zero rate of VAP to inform realistic best practice benchmarks.
We conducted a retrospective analysis of VAP data collected from December 2003 to June 2005 as part of the Keystone ICU project. The Keystone project was a cohort collaborative, and the VAP intervention sought to increase caregiver’s use of evidence-based therapies to prevent VAP, coupled with focused efforts to improve safety culture and communication. Details of the project and the initial evaluation of the VAP intervention have been published.8 The VAP rate was calculated as the number of infections per 1,000 ventilator-days. All ICU teams used the Centers for Disease Control and Prevention–National Healthcare Safety Network surveillance criteria for pneumonia applied to patients who received mechanical ventilation throughout the study period when collecting data.9
The primary outcome measure for this analysis was the maximum number of consecutive months of mechanical ventilation without a VAP (VAP-free months) for each ICU. The number of VAP-free months was calculated by totaling the number of ventilator-days without a VAP for each patient in a given ICU during a given period and dividing by 30 days. Therefore, an ICU with 1 patient who received mechanical ventilation for 30 days and an ICU with 10 patients who received mechanical ventilation for 3 days would both be 1 ventilator-month. We chose maximum VAP-free months of mechanical ventilation, rather than calendar months, to account for variations in hospital bed size and ventilator use (exposure to risk) across different ICUs. Our secondary outcome measure was VAP-free calendar months, which was estimated on the basis of the results obtained using ventilator-months and the average number of patients who received mechanical ventilation for each hospital bed size.
An event-time analysis was performed for the VAP-free ventilator months with descriptive statistics, log-rank tests, Cox regressions, and analyses of variance to estimate parameters and adjust for that hospital and ICU-level covariates would predict our primary and secondary outcomes. Hospital-level covariates included region, hospital bed size (less than 200 beds, 200–299 beds, 300–399 beds, and 400 or more beds), and teaching status. The ICU-level covariates were type and patient load, defined as the average number of patients who received mechanical ventilation per day. We used the generalized estimating equation (GEE) regression models with robust variance estimation to account for clustering of ICU data within hospitals. Regression results were reported as hazard ratios with 95% confidence intervals (CIs), and the tests that we used were 2-sided, with statistical significance set at . Analyses were performed using Stata/IC 11.1 software.
For 112 ICUs in 72 hospitals, the mean follow-up time was 1,143 days (38 months), and the mean number of patients who received mechanical ventilation per day was 4 (Table 1). The majority of ICUs were mixed/neurologic (55%).
|Variable||No. (%) of ICUs||Maximum VAP-free ventilator-months, median months (IQR)||Hazard ratioa (95% CI)||P|
|East Central||14 (13)||19.7 (32,6)||1.0 (reference)|
|Mid-Michigan||10 (9)||30.1 (18.5)||1.22 (0.40–3.71)|
|North Central||5 (4)||3.3 (15.8)||3.69 (0.83–16.39)|
|Southeast||55 (49)||26.3 (17.3)||1.35 (0.47–3.84)|
|Southwest||9 (8)||28.0 (25.5)||1.79 (0.54–5.91)|
|Upper Peninsula||1 (1)||N/A||1.36 (0.36–5.10)|
|West Central||12 (11)||16.2 (16.4)||1.33 (0.32–5.49)|
|Out of state||6 (5)||14.9 (28.9)||1.45 (0.51–4.17)|
|Mixed/neurologic||62 (55)||26.3 (25.3)||1.0 (reference)|
|Medical||11 (10)||28.0 (12.9)||1.57 (0.70–3.51)|
|Surgical/trauma/burn||22 (20)||22.8 (14.2)||3.42 (1.87–6.24)|
|Cardiac||17 (15)||19.7 (18.6)||2.76 (1.27–5.99)|
|<200||24 (21)||11.8 (22.7)||1.0 (reference)|
|200–299||25 (22)||21.9 (14.2)||0.43 (0.21–0.91)|
|300–399||21 (19)||34.8 (18.6)||0.16 (0.06–0.39)|
|≥400||42 (38)||26.5 (20.7)||0.20 (0.08–0.50)|
|Nonteachingb||36 (32)||26.3 (29.8)||1.0 (reference)|
|Teaching||76 (68)||26.1 (18.8)||1.27 (0.76–2.11)|
|Overall||112 (100)||26.2 (21.6)|
The overall median maximum number of VAP-free ventilator-months was 26.2 (interquartile range [IQR], 21.6). In the Cox multiple regression model, ICU type was a significant predictor of VAP-free ventilator-months (; Table 1), with the highest hazard ratio in the surgical/trauma/burn ICUs at 3.42 (95% CI, 1.87–6.24). Hospital bed size had a global effect on VAP-free ventilator-months (); ICUs in hospitals with less than 200 beds had a significantly higher hazard of VAP occurrence compared with ICUs in hospitals with 200 or more beds (Figure 1). There was a significant decrease in the risk of VAP for each additional patient who received mechanical ventilation (hazard ratio, 0.92 per additional patient who received mechanical ventilation [95% CI, 0.86–0.97]). Teaching status was not a significant predictor of median maximum VAP-free ventilator-months (hazard ratio, 1.27 [95% CI, 0.76–2.11]).
Estimates of VAP-free calendar-months varied by hospital bed size (; data not shown). The overall median number of VAP-free calendar-months was 6.1 (IQR, 5.6), varying from 5.1 (IQR, 5.1) calendar months in ICUs with 400 or more hospital beds to 9.1 (IQR, 11.2) calendar months in ICUs with less than 200 beds.
We found that diverse ICUs in the Keystone project can achieve and sustain a zero rate of VAPs among patients who receive mechanical ventilation. Half of the participating ICUs achieved 26.2 consecutive VAP-free ventilator-months. Moreover, half of the ICUs went over 6 calendar-months and three-quarters of ICUs went nearly 1 calendar-year without a case of VAP. Our finding that VAP risk was reduced as patient load increased confirms previous evidence that a single patient’s risk of VAP is cumulative with increasing time spent receiving mechanical ventilation.10
Our study advances the existing literature by using device exposure time (ventilator-months) rather than calendar time as the primary outcome variable. Although calendar time is a simple and easily understood measure, it does not account for the varying exposures to risk across different organizations. For example, a 10,000-employee factory might have a much higher risk for an industrial accident than a 10-employee factory. Thus, the finding of 100 days since the last accident would have different meanings for these 2 factories. In our study, ICUs with less than 200 hospital beds have the lowest number of VAP-free ventilator-months and the highest number of VAP-free calendar-months. Studying device exposure time (ventilator-months) accounts for variations in bed size and ventilator use across ICUs and likely provides a more accurate benchmark for evaluating performance.
Our study has several limitations. First, the parent study had potential limitations that may influence our results, including voluntary participation in the project, no concurrent control group, lack of VAP outcome validation, and missing data. Second, truncating follow-up at the end of the observation period could underestimate the outcome if ICUs eventually had their longest VAP-free period after the end of follow up. Third, information about the date of occurrence of a VAP incident is not available. Thus, our estimates of maximum VAP-free ventilator time may be overestimated or underestimated by up to 20 ventilator-days, on average.
In conclusion, our study suggests that ICUs in diverse hospitals can eliminate VAP and can remain VAP-free for nearly a year. This has important public health implications and informs realistic best-practice benchmarks. Furthermore, the use of ventilator-months to account for variations in device exposure across ICUs likely provides a more accurate benchmark for evaluating performance.
We thank Christine G. Holzmueller, BLA, for her thoughtful review and assistance in editing and organizing the manuscript content and Annette Levering, BS, for her administrative support.
Financial support. For the period October 2003–September 2005, the Keystone ICU project was supported by the Agency for Healthcare Research and Quality grant 1UC1HS14246. Additional support for this project was provided by the Michigan Health and Hospital Association for the biannual statewide meetings.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article. All authors submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and the conflicts that the editors consider relevant to this article are disclosed here.
- 1. Safdar N, Dezfulian C, Collard HR, Saint S. Clinical and economic consequences of ventilator-associated pneumonia: a systematic review. Crit Care Med 2005;33:2184–2193.
- 2. Klompas M. Prevention of ventilator-associated pneumonia. Expert Rev Anti Infect Ther 2010;8:791–800.
- 3. Pogorzelska M, Stone PW, Furuya EY, et al. Impact of the ventilator bundle on ventilator-associated pneumonia in intensive care unit. Int J Qual Health Care 2011;23:538–544.
- 4. Wip C, Napolitano L. Bundles to prevent ventilator-associated pneumonia: how valuable are they? Curr Opin Infect Dis 2009;22:159–166.
- 5. O’Keefe-McCarthy S, Santiago C, Lau G. Ventilator-associated pneumonia bundled strategies: an evidence-based practice. Worldviews Evid Based Nurs 2008;5:193–204.
- 6. Klompas M. Ventilator-associated pneumonia: is zero possible? Clin Infect Dis 2010;51:1123–1126.
- 7. Marra AR, Cal RG, Silva CV, et al. Successful prevention of ventilator-associated pneumonia in an intensive care setting. Am J Infect Control 2009;37:619–625.
- 8. Berenholtz SM, Pham JC, Thompson DA, et al. Collaborative cohort study of an intervention to reduce ventilator-associated pneumonia in the intensive care unit. Infect Control Hosp Epidemiol 2011;32:305–314.
- 9. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008;36:309–332.
- 10. Ashraf M, Ostrosky-Zeichner L. Ventilator-associated pneumonia: a review. Hosp Pract (Minneap) 2012;40:93–105.