Patient to Nurse Ratio and Risk of Ventilator-Associated Pneumonia in Critically Ill Patients

EU-VAP/CAP is a prospective, observational survey conducted in 27 ICUs in 9 European countries: Belgium, France, Germany, Greece, Italy, Ireland, Portugal, Spain, and Turkey. Organizational aspects of the study are described elsewhere.²⁷  All patients who were admitted to the ICU for treatment of pneumonia or received invasive mechanical ventilation for more than 48 hours, irrespective of the admission diagnosis, were included in the initial cohort.

VAP was defined as (1) a pulmonary infection that occurred 48 hours or more after endotracheal intubation in patients with no evidence of pneumonia at the time of intubation or (2) the diagnosis of a new pulmonary infection if the initial admission to the ICU was for pneumonia.²⁸  Late-onset VAP was defined as VAP that occurred more than 5 days after intubation. Patient demographics, primary diagnosis, ICU and hospital lengths of stay, duration of mechanical ventilation, and ICU mortality were recorded for all patients. Days at risk for VAP were defined as the number of days of treatment with mechanical ventilation before the onset of VAP or, for patients in whom VAP did not develop, the total number of days of treatment with mechanical ventilation.

Underlying diseases were defined as in the primary article²⁷  on the EU-VAP study.

Severity of underlying disease was assessed by using the McCabe classification of comorbid conditions. In this classification system, patients’ prognoses are roughly estimated and categorized into 3 main groups: likely to survive more than 5 years (nonfatal underlying disease), likely to survive 1 to 5 years (ultimately fatal), or likely to die within 1 year (rapidly fatal).

Severity of acute illness was assessed by using the Simplified Acute Physiology Score (SAPS) II,²⁹  a well-validated scoring system that has been used for years as a standard measure to quantify severity of acute illness. The SAPS II takes into account the type of admission (medical, scheduled or unscheduled surgery), chronic diseases (metastatic carcinoma, hematological malignant tumor), age, score on the Glasgow Coma Scale, hemodynamic and respiratory status, temperature, white blood cell count, urine output, liver tests, and electrolyte balance.

Data were recorded by the investigators at the individual study sites on paper-based case report forms. These forms were sent to the central study site where the data were put into an electronic database and checked for inconsistencies by the principal investigators.

The initial study cohort included 2585 patients. Because the focus of the study reported here was prevention of VAP, data on patients with a clinical diagnosis of community-acquired pneumonia, non–ventilator-associated hospital-acquired pneumonia, or very early VAP (due to aspiration and developing within 48 hours after intubation), were excluded from the analysis. Data on patients from 6 ICUs that did not provide data on nurse staffing levels were also excluded from the analysis. Hence, our study cohort consisted of 1658 patients who had mechanical ventilation for at least 48 hours.

Routine staffing levels for all available ICU beds were considered, irrespective of bed occupancy. Routine staffing level is defined as the patient to nurse ratio that is standard in a particular ICU. As such, unit-based standard nurse staffing levels were used irrespective of acute shortages of staff and number of patients present. Daily bed occupancy levels were not taken into account because this cohort consisted solely of patients who received mechanical ventilation. Hence, actual day-to-day patient to nurse ratios were not available for the analysis. For units with variable staffing levels (eg, 1 to 1 during day shifts and 2 to 1 during night shifts), the highest patient to nurse ratio in a 24-hour period was considered. In European countries, ICU nurses are generally qualified to manage ventilators. The use of respiratory therapists for ventilator management is rare.

The participating centers received ethical approval from the appropriate institutions. Informed consent was waived because the study was observational.

Medians and interquartile ranges were used for continuous variables and numbers and percentages for discrete variables. For comparisons between groups, the Mann-Whitney test and the Fisher exact test or χ² test were used as appropriate. Independent relationships with development of VAP were assessed by using a logistic regression analysis. Variables considered in the logistic regression analysis either showed a moderate relationship (P < .10) in univariate analysis or a logic relationship with the dependent variable. Variables considered were age, SAPS II, underlying diseases, admission diagnosis, and patient to nurse ratio. A stepwise variable elimination was then performed to develop the final model. Variables with P greater than .15 were stepwise removed. The patient to nurse ratio was kept in the model irrespective of the associated P value. According to the reference category in patient to nurse ratio, 4 different logistic regression models were generated: model I with 4 staffing level ratios (1 to 1, 2 to 1, 2.5 to 1, and 3 to 1) and a 1 to 1 ratio as the reference category; model II with 2 staffing level ratios (1 to 1 and >1 to 1) and a 1 to 1 ratio as the reference category; model III with 2 staffing level ratios (2 to 1 and >2 to 1) and a 2 to 1 ratio as the reference category; and model IV with 2 staffing level ratios (<3 to 1 and 3 to 1) and a less than 3 to 1 ratio as the reference category. Results of the regression analysis are reported as odds ratios (OR) and 95% CIs.

VAP developed in 393 of the 1658 patients (23.7%) during their ICU stay; 220 of the patients with VAP had late-onset VAP (13.3%). The patient to nurse ratio was constantly 1 to 1 in only 1 ICU. In 10 ICUs, the highest patient to nurse ratio was 2 to 1; 4 units had a 2.5 to 1 ratio, and 6 had a 3 to 1 ratio. VAP rates in units with patient to nurse ratios of 1 to 1, 2 to 1, 2.5 to 1, and 3 to 1 were 9.3%, 25.7%, 18.7%, and 24.2%, respectively (P = .003; Table 1). Rates were significantly lower (P = .002) in units with a ratio of 1 to 1 (9.3%) than in units with a ratio of more than 1 patient to 1 nurse (24.4%).

However, important differences in patients’ characteristics must also be considered (Table 1). Compared with other units, units with a ratio of more than 1 patient to 1 nurse had more patients admitted because of a medical condition or trauma and fewer patients admitted after elective surgery. Furthermore, duration of mechanical ventilation was significantly longer in the units with the ratio of more than 1 patient to 1 nurse (median, 8 vs 3 days; P < .001), as were the time at risk for VAP (median, 7 vs 3 days; P < .001), and the ICU stay (median, 12 vs 5 days; P < .001). On the other hand, severity of disease at the time of admission as indicated by the SAPS II was significantly higher among patients cared for in the unit with a 1 to 1 patient to nurse ratio than in the unit with more than 1 patient to 1 nurse (median, 53 vs 45; P = .002). After adjustments for such confounding covariates, a ratio of more than 1 patient to 1 nurse was no longer associated with increased risk for VAP (Table 2, models I and II).

Univariate analysis indicated no significant difference in VAP rates between patients cared for in ICUs with a patient to nurse ratio of 2 to 1 and those cared for in units with a ratio of more than 2 patients to 1 nurse (24.6% vs 22.1%; P = .25; Table 1). Also in multivariate analysis (Table 2), higher patient to nurse ratios, either more than 2 patients to 1 nurse or 3 patients to 1 nurse, were not associated with a higher risk for VAP (Table 2, models III and IV). In all logistic regression models, the following 3 variables were identified as independent risk factors for the acquisition of VAP: the number of days at risk, admission because of trauma, and higher SAPS II. Results of the logistic regression models did not change when late-onset VAP was used as the dependent variable. With a patient to nurse ratio of 1 to 1 as the reference category, no other ratio was associated with risk for late-onset VAP (patient to nurse ratio 2 to 1: OR, 2.05; 95% CI, 0.62–6.69; patient to nurse ratio 2.5 to 1: OR, 1.68; 95% CI, 0.48–5.85; patient to nurse ratio 3 to 1: OR, 2.19; 95% CI, 0.66–7.32). In this regression model, the variable trauma (OR, 1.93; 95% CI, 1.35–2.73) and an increasing number of days at risk (OR, 1.06; 95% CI, 1.05–1.07) were the predominate risk factors for late-onset VAP.

In this study, based on a large cohort of patients from 21 European ICUs who were treated with mechanical ventilation, we found no association between high staffing levels (patient to nurse ratio <2 to 1) and reduced risk for VAP. Although a patient to nurse ratio of 1 to 1 was associated with a lower risk of VAP in univariate analysis, after adjustment for covariates, this observation was no longer significant. Factors such as admission because of trauma, number of days at risk, and disease severity seem to be stronger determinants of VAP.

Our observations differ from the results of Hugonnet et al,³⁰  who explored the relationship between staffing level and development of VAP in a single-center cohort of 2470 medical ICU patients in which the average patient to nurse ratio was 1.9 to 1. Multivariate Cox regression analysis indicated that higher staffing levels reduced the risk for late-onset VAP, although the reduction was borderline significant (adjusted hazard ratio, 0.42; 95% CI, 0.18 to 0.99) and was not evident in early-onset VAP (hazard ratio, 0.78; 95% CI, 0.42 to 1.45) or when all VAP cases were considered (hazard ratio, 0.66; 95% CI, 0.40 to 1.10). We focused on late-onset VAP and did not find a significant association between higher staffing levels and a reduced risk for late-onset VAP.

Although data on the relationship between nurse staffing levels and VAP are scarce, more reports are available on the relationship between staffing levels and hospital-acquired pneumonia. In a systematic review, Lang et al³¹  examined the relationship between the hospital-wide risk for pneumonia and nurse staffing levels. Their analysis revealed mixed results. First, some investigators^32,33  reported a deleterious effect of lower staffing levels on pneumonia rates in both medical and surgical wards but not among patients who underwent invasive vascular procedures.³³  Second, Lichtig et al³⁴  found an inverse relationship between staffing level and the occurrence of pneumonia in hospitals in California but not in hospitals in New York. In a cohort of patients after esophagectomy, Amaravadi et al³⁵  found that the risk for several postoperative pulmonary and infectious complications and the use of resources increased when 1 ICU nurse provided care for more than 2 patients at night. Finally, in some studies,^36–38  the researchers found no link between staffing levels and risk for pneumonia. On the basis of their systematic review, Lang et al³¹  consequently concluded that the existing evidence neither confirms nor rules out an inverse relationship between nurse staffing and pneumonia rates. A certain relationship probably exists, but the effect is rather discrete and is easily diminished when a patient has other, more powerful, risk factors, such as admission because of trauma and (severity of) underlying conditions.

Despite the lack of an obvious relationship with risk for infection, higher nurse staffing levels in ICUs have been associated with better survival rates. Cho et al³⁹  investigated the deleterious effect of lower nurse staffing levels in a large Korean study with 27 372 patients from 42 tertiary and 194 secondary hospitals. Compared with a ratio of 2 patients to 1 nurse, each 1-patient increase in the ratio was associated with a 9% increase in the odds of death (OR, 1.09; 95% CI, 1.04 to 1.14). Because Cho et al did not have data for a patient to nurse ratio of 1 to 1, they could not evaluate the eventual beneficial effect of this high (1:1) staffing. These data indicate that more favorable staffing levels are associated with better quality of care and hence better patient outcomes.

Several limitations of our study must be addressed. The study was a secondary analysis and thus was not specifically designed for the research question. Consequently, the study was hampered by several confounders. First, among the ICUs, a patient to nurse ratio of 1 to 1 was the standard in only a single unit. Therefore, the external validity can be questioned. In addition, important differences in patient characteristics existed between this particular ICU and units with a lower staffing level (>1 patient to 1 nurse). The ICU with the 1 to 1 ratio had more patients who had had elective surgery and fewer trauma patients than the other units did. In our study, as well as in previous reports,^2,40  trauma is well recognized as a major and independent risk factor for VAP.

Furthermore, patients cared for in ICUs with a patient to nurse ratio of 1 to 1 experienced fewer days on the ventilator and thus fewer days at risk for VAP. A possible explanation is that the shorter period of ventilation dependency is a direct consequence of the 1 to 1 staffing level with a more proactive weaning strategy,⁴¹  a situation that may reduce the risk of exposure to time-dependent complications such as nosocomial infections. However, we think that the shorter period of dependency is due to the preponderance of elective surgery patients. Compared with medical or trauma patients, patients who have elective surgery most likely have a less troublesome weaning. Anyhow, after adjustment for the most important risk factors—ICU admission because of trauma, days at risk, and disease severity—higher staffing levels were not associated with a reduced risk for VAP in our cohort of patients.

Another limitation of our study is that the patient to nurse ratios we used were not the actual ratios as calculated on a daily basis (number of patients/number of nurses present per day). We used unit-based standard nurse staffing levels irrespective of acute staff shortages (eg, due to sick leave) and number of patients present. Hence, actual day-to-day patient to nurse ratios were not available for the analysis. A further limitation is that quality of care is not fully reflected by degree of staffing. Quality of care at the level of individual patients strongly depends on nurses’ competencies. Recently, we discovered substantial shortages in the average knowledge level of European ICU nurses about evidence- based guidelines for VAP prevention,^42,43  and in addition, implementation of such recommendations is problematic.⁴⁴  Obviously, the database we used did not allow us to adjust for either knowledge and/or implementation level of best-practice recommendations. Finally, we have no knowledge of the compliance of distinct ICUs with evidence-based guidelines for the prevention of VAP, such as drainage of subglottic secretions, semi-recumbent positioning, and chlorhexidine-based oral care.

In this cohort of patients treated with mechanical ventilation, a patient to nurse ratio of 1 to 1 appeared to be associated with a lower risk for VAP. After adjustment for confounding covariates, however, the difference was no longer significant. Although higher staffing levels may be beneficial for other outcomes, the effect of trauma, general disease severity, and duration of mechanical ventilation are more important risk factors for VAP. Our data indicate that efforts to reduce the number of days at risk should be a priority in the prevention of VAP. Thus, our results underscore the value of a proactive extubation policy with a “sedation vacation” as recommended in current guidelines.¹⁵  Further research is necessary to evaluate the relationship between higher staffing levels (patient to nurse ratio <2 to 1) and compliance rates with distinct evidence-based strategies to prevent VAP. In our study, the actual patient to nurse ratio should be taken into account (actual number of patients per nurse each day).

This study was endorsed by the European Critical Care Research Network of the European Society of Intensive Care Medicine.

Members of the EU-VAP/CAP study group were as follows: Djilali Annane (Raymond Poincaré University Hospital, Garches, France), Rosario Amaya-Villar (Virgen del Rocio University Hospital, Seville, Spain), Apostolos Armaganidis (Attikon University Hospital, Athens, Greece), Stijn Blot (Ghent University Hospital, Ghent, Belgium), Christian Brun-Buisson (Henri-Mondor University Hospital, Paris, France), Antonio Carneiro (Santo Antonio Hospital, Porto, Portugal), Maria Deja (Charite University Hospital, Berlin, Germany), Jan De Waele (Ghent University Hospital, Ghent, Belgium), Emili Diaz (Joan XIII University Hospital, Tarragona, Spain), George Dimopoulos (Attikon University Hospital and Sotiria Hospital, Athens, Greece), Silvano Cardellino (Cardinal Massaia Hospital, Asti, Italy), Jose Garnacho-Montero (Virgen del Rocio University Hospital, Seville, Spain), Muhammet Guven (Erciyes University Hospital, Kayseri, Turkey), Apostolos Komnos (Larisa General Hospital, Larisa, Greece), Despoina Koulenti (Attikon University Hospital, Athens, Greece, and Rovira i Virgili University, Tarragona, Spain), Wolfgang Krueger (Tübingen University Hospital, Tübingen, Germany, and Constance Hospital, Constance, Germany), Thiago Lisboa (Joan XIII University Hospital and CIBER Enfermedades Respiratorias), Antonio Macor (Amedeo di Savoia Hospital, Torino, Italy), Emilpaolo Manno (Maria Vittoria Hospital, Torino, Italy), Rafael Mañez (Bellvitge University Hospital, Barcelona, Spain), Brian Marsh (Mater Misericordiae University Hospital, Dublin, Ireland), Claude Martin (Nord University Hospital, Marseille, France), Ignacio Martin-Loeches (Mater Misericordiae University Hospital), Pavlos Myrianthefs (KAT Hospital, Athens, Greece), Marc Nauwynck (St. Jan Hospital, Bruges, Belgium), Laurent Papazian (Sainte Marguerite University Hospital, Marseille, France), Christian Putensen (Bonn University Hospital, Bonn, Germany), Bernard Regnier (Bichat-Claude-Bernard University Hospital, Paris, France), Jordi Rello (Joan XIII University Hospital), Jordi Sole-Violan (Dr Negrin University Hospital, Gran Canaria, Spain), Giuseppe Spina (Mauriziano Umberto I Hospital, Torino, Italy), Arzu Topeli (Hacettepe University Hospital, Ankara, Turkey), and Hermann Wrigge (Bonn University Hospital, Bonn, Germany.