Table 1. Summary Table of Included Studies

Table 1. Summary table of included studies

Articles / Population / Intervention / Comparator / Outcomes / Study design / Brief summary of results / Significant indicators / Level of evidence
Alonso-Echanove et al., 2003 / Adult patients (n=4,535) in 8 ICUs in the US admitted with central venous catheter (CVC) / Nurse-patient ratio; ICU nurse or float nurse; patient care assistant-to-patient ratios (number of PCAs per shift per 100 patients) / Proportion of float nurse-days (float nurse defined as nurse who had worked in the unit < 1 year) / CVC-associated bloodstream infections (BSIs) / Prospective cohort study / Risk for CVC-BSIs not associated with nurse-patient ratio or PCA-patient ratio but was associated with proportion of float nurse-days. Risk for CVC-BSI 2.6 times higher in patients cared for by float nurse > 60% of time; rate ratio 2.61 (1.21 to 5.59) / CVC-BSIs / 3e
Amaravadi, et al., 2000 / Adult patients (n=366) who had oesophageal resection in 32 ICUs in 32 non-federal acute care hospitals in Maryland, US / Night time nurse patient ratio of 1 nurse to 1-2 patients / Night time nurse patient ratio of 1 nurse to 3 or more patients / In-hospital mortality, length of stay, pneumonia, reintubation, septicaemia, aspiration, postoperative infection, MI, cardiac arrest, surgical complications, acute renal failure / Cohort study / 39% increase in median LOS, OR 2.4 pneumonia, OR 2.6 reintubation, OR 3.6 septicaemia / Hospital length of stay, pneumonia, reintubation, septicaemia / 3e
Benbenbishty, et al., 2010 / Adult patients physically restrained (n=669) or chemically restrained (n=566) in 34 ICUs in nine European countries / Nurse-patient ratio each shift / Day and night time nurse- patient ratios of 1:1, 1:2, 1:3 or 1:4. / Physical and chemical restraint use (note, results for chemical restraint use not reported in detail) / Cross-sectional / Physical restraint more likely to be used for patients who were in units with lower daytime nurse-patient ratio (χ2 = 17.17, p = 0.001). / Physical restraint use / 3e
Blegen et al., 2011 / Adult patients (approx. 1.1 million) in 54 US hospitals, staffing data from 285 ICUs, 587 general wards, outcomes measured at hospital level / Total hours of care from RN/LPN/NA per patient day and skill mix (RN percent) / Comparing levels between hospitals - no specific comparator / In-hospital CHF mortality, decubitus ulcer, failure to rescue, infection due to medical care, postoperative sepsis, rate of LOS > expected / Cross sectional / Significant association between infection due to medical care and total ICU staffing and between ICU RN skill mix and postoperative sepsis in non-safety-net hospitals. Significant associations between CHF mortality and failure to rescue and ICU RN skill mix in safety-net hospitals / Infection due to medical care, postoperative sepsis, in-hospital CHF mortality, failure to rescue / 4b
Blot et al., 2011 / Adult ICU patients (n=1658) treated for pneumonia or receiving invasive mechanical ventilation ≥ 48 hours in 27 ICUs in 9 European countries. / Patient-to-nurse ratio / Staffing level ratios: Model I (1:1, 2:1, 2.5:2, & 3:1); Model II (1:1 & >1:1); Model III (2:1 & >2:1); Model IV (<3:1 and 3:1). / Ventilator-associated pneumonia (VAP) / Prospective, observational survey / No significant difference in VAP rates in ICUs with patient-to-nurse ratio of 2:1 or > 2:1. A 1:1 patient-to-nurse ratio was associated with lower risk for VAP but not significant after adjustment for confounding covariates. Non-significant association between higher nurse staffing levels and reduced risk for late-onset VAP. / None / 3e
Bolton et al., 2007 / Falls data from 65 step-down units and pressure ulcer and restraint use data from 40 step-down units in Californian hospitals / Hours of nursing care/patient day; nurse/patient ratios and skill mix pre/post introduction of mandated nurse staffing ratios / Staffing changes pre/post introduction of mandated staffing ratios / Falls and falls with injury; hospital-acquired pressure ulcers; restraint use / Longitudinal study pre/post mandated nurse staffing ratios / Significantassociation between % of care by RNs & any falls.
Significant positive association between higher total hours of care per patient day & higher prevalence of stage II+ pressure ulcers -
No significant results or trends in relationship between staffing variables & restraint prevalence. / Falls, falls with injuries, hospital-acquired pressure ulcers / 3e
Checkley et al., 2014 / 69 ICUs in 42 hospitals centres in the US. May include some paediatric ICUs. No information on patient numbers / No intervention, Bed-to-nurse ratio of 1.5:1 / Bed-to-nurse ratio of 2:1 / ICU mortality / Prospective observational study / Significantly lower mortality, 1.8% lower when the bed-to-nurse ratio decreased from 2:1 to 1.5:1 / ICU mortality / 3e
Chittawatanarat et al., 2014 / 155 ICUs in 87 hospitals in Thailand(90% of sample were adult ICUs but some paediatric). 104,046 patient admissions used for mortality calculations / Average nurse-to-patient ratio (NPR) each shift / Higher/lower NPR / Mortality (monthly crude mortality rate), average ventilator days, ICU length of stay (ICU-LOS) / Retrospective observational study / Increased NPR significantly associated with decreased number of ventilator days (-2.08 [-5.377 to -0.166], p = 0.037). NPR not significantly associated with crude mortality or ICU-LOS. / Ventilator days / 3e
Cho et al., 2008 / Adult patients (n=27,372) in ICUs in 236 Korean hospitals / No specific intervention / No specific comparator / In-hospital mortality / Cross sectional / Significant relationship between nurse patient ratio and mortality in secondary hospitals (OR 1.43) / In-hospital mortality / 4b
Cho & Yun, 2009 / Adult patients (n=6,957) with acute stroke admitted to ICUs in 185 Korean hospitals
(No of ICUs not given as for hospitals with more than 1 ICU results were aggregated) / Bed-to-nurse (RN) ratio / Bed-to-nurse ratios categorised into 9 staffing grades (Grades 1-9). Grade 1 (best) number of beds per nurse < 0.50, Grade 9 (worst) ratio is ≥ 2.0. (Note, for the analysis, staffing grades collapsed into four groups: 1-3, 4 & 5, 6 & 7, and 8 & 9.) / In-hospital mortality & 30-day mortality, / Cross sectional / Worsening ICU staffing associated with stepwise decrease in the proportion of low 30-day mortality group. Hospitals with Grades 4 & 5 ICU staffing less likely to have high in-hospital mortality than reference group (OR, 0.26; 95% CI, 0.09-0.80). Better ICU staffing grades (1-3 and 4 & 5) had lower probability of classification into high 30-day mortality group. / In-hospital mortality and 30-day mortality / 4b
Dancer et al., 2006 / Adult patients (n=174) in one ICU in Scotland / No intervention - measured the relationship between workload and number of nurses on shift and described understaffing as weeks staffed below the line of best fit / None / MRSA / Retrospective observational study / No significant relationship between understaffing and MRSA acquisition / None / 4c
Dang et al., 2002 / Adult patients (n=2,606) undergoing abdominal aortic surgery admitted to ICU in 38 hospitals in Maryland / No specific intervention, used nurse-to-patient ratios (NPR) Compared high-intensity nurse staffing (NPR ≤ 1:2 on both day & night shifts) to two other groups / Low-intensity nurse staffing (NPR ≥ 1:3 on both day & night shifts); medium-intensity nurse staffing (NPR ≥ 1:3 on day or night shift but not both); / Cardiac complications (acute myocardial infarction, cardiac arrest, cardiac complications post-procedure); respiratory complications (aspiration, pneumonia, pulmonary insufficiency post-procedure, mechanical ventilation > 96 h, tracheal reintubation); Other complications (acute renal failure, septicaemia, platelet transfusion) / Retrospective, longitudinal study / Medium-intensity nurse staffing: significant association with cardiac complications OR 1.78, (1.16-2.72); other complications OR 1.74, (1.15-2.63); cardiac complications after a procedure OR 2.10, (1.26-3.50); Low-intensity nurse staffing: significant association with respiratory complications OR 2.33, (1.50-3.60); pulmonary insufficiency OR 5.11, (2.89-0.04);mechanical ventilation after 96 h OR 2.39, (1.55-3.69); reintubation OR 2.09, (1.47-3.03). / Cardiac complications, Respiratory complications, Other complications / 3e
Dimick et al., 2001 / Adult patients (n=569) who had hepatic resection in 33 non-federal acute care hospitals in Maryland, US / 1 nurse for 1 or 2 patients / 1 nurse for 3 or 4 patients / In-hospital mortality, postoperative complications including aspiration, pulmonary insufficiency, pneumonia, reintubation, septicaemia, postoperative infection, cardiac complications, cardiac arrest, acute MI, acute renal failure, LOS / Cohort study / No significant difference between the two groups on in-hospital mortality, length of stay. Significant association between fewer ICU nurses at night and reintubation, no significant associations with other postoperative complications. / Reintubation / 3e
Dodek et al., 2014 / Patients (n=42,381) in 37 ICUs in British Columbia and 20 ICUs in Ontario (n=14,165). (Sample may include paediatric patients) / No specific intervention. Compared nurse-patient ratio / Nurse-patient ratio 1:1 vs 1:2 / Hospital mortality, ICU length of stay (ICULOS), hospital length of stay (LOS) / Cross-sectional / OR 2.08 (1.45-2.97) nurse patient ratio 1:1 vs 1:2 in BC and 1.37 (0.81-2.33) in Ontario for mortality; RR 1.4 (1.09-1.79) in BC and 0.99 (0.80-1.23) in Ontario for ICU LOS; RR 1.38 (1.12-1.71) in BC and 1.27 (1.04-1.56) in Ontario for Hospital LOS / In-hospital mortality, ICU LOS, Hospital LOS / 3e
Dorsey et al., 2000 / Adult patients (n=52) in a surgical ICU in a general hospital in San Francisco / No intervention - used the mean staffing quotient for each month of the study / No specific comparator / Presence of Enterobacter cloacae or Serratia marcescens in the blood or CSF / Retrospective cohort study / Temporal relationship between infection outbreaks and lower staffing. During outbreak months, on average one less staffed empty bed each shift compared to non-outbreak months / Enterobacter and Serratia presence / 3e
Falk & Wallin, 2016 / Patients over 15 years (n=33,032) in ICUs in 7 hospitals in Sweden / Nurse to patient ratio; nurse assistant to patient ratio / Compared hospitals which had 1:1, 0.6:1 and 0.5:1 ratios / Length of ICU stay, ICU mortality, 30-day mortality, estimated mortality risk, unplanned extubation, time on ventilation / Longitudinal / Nurse to patient ratio of 0.5-0.6 per patient - patients spent longer on ventilation, higher nurse-patient ratios significantly longer length of stay. Higher mortality in hospitals with lowest ratios of registered staff. Other results not reported. / Time on ventilation, length of stay, mortality / 3e
Glance et al., 2012 / Adult patients (n=70,142) admitted with trauma from 77 Level I and II hospitals in the United States (sample may include paediatric patients) / Hours per Patient Day (HPPD) for registered nurses (RN), licensed practical nurses (LPN), and nurses' aides (NA) / (i) Bottom & top quartiles of LPN HPPD;(ii) staffing ratio of LPN staffing to total nurse staffing / In-hospital mortality, healthcare associated infections, failure-to-rescue (FTR) / Cross sectional / 15 min increase in LPN HPPD associated with death OR 1.15 (1.05, 1.25) and sepsis OR 1.27 (1.11, 1.45). 1% increase in ratio of LPN to total nursing time associated with mortality OR 1.04 (1.02-1.06) and sepsis OR 1.06 (1.03-1.10). 15 min increase in NA HPPD associated with pneumonia OR 0.92 (0.87, 0.98). Hospitals in highest quartile of LPN staffing had 3 excess deaths per 1000 patients and 5 more episodes of sepsis than hospitals in lowest quartile LPN staffing. / In-hospital mortality; sepsis; pneumonia / 4b
Grundmann et al., 2002 / Adult patients (n=782) in one ICU in a tertiary care hospital in the UK / No intervention - used the average number of nurses per shift over a 24 hour period / No specific comparator / Presence of MRSA / Observational study / Relationship between nursing staff deficits (greater than 1:1 nurse/patient ratio) and development of MRSA - Risk ratio 1.05 (1.02, 1.08) / MRSA presence / 3e
Halwani, et al., 2006 / Adult ICU patients (n=430) with ICU stay > 48 h in one hospital in Nottingham, UK / Nurse-patient ratio / Understaffing defined as less than optimal staffing level (1:1 during 24-h period) / Cross-contamination with nosocomial infections / Prospective cohort study / Relationship between cross-contamination and patients experiencing understaffing for whole ICU stay compared with patients who were not exposed tounderstaffing OR 3.07 (1.41-6.66) / Cross-contamination with nosocomial pathogens / 3e
He et al., 2012 / Fall rates among adult patients in 8915 nursing units in 1171 US acute care hospitals (included 1994 critical care units & 1328 step-down units) / No intervention - used annual average nursing hours per patient day and percent of total hours of care provided by RNs / No specific comparator / Inpatient falls / Retrospective observational study / Higher HPPD and RN skill mix were significantly associated with lower fall rates (IRR: 0.95 for HPPD and 0.99 for RN skill mix) for all units together, associations consistent at unit level, results not reported / Fall rate / 3e
Hugonnet, Chevrolet, & Pittet, 2007 / Adult ICU patients (n=1,883) with ICU stay > 48 h in Geneva, Switzerland / Nurse workload measured by 24-h nurse-to-patient ratio. / Low staffing level / Nosocomial infection / Prospective cohort study / Significant association between nurse – patient ratio and nosocomial infections (RR 0.69 (0.50 – 0.95) / Nosocomial infection / 3e
Hugonnet, Uçkay, & Pittet, 2007 / Adult patients (n=2470) admitted to a medical ICU in Switzerland / No intervention - calculated a 24 hour nurse/patient ratio and a mean ratio per shift / No specific comparator / Ventilator-associated pneumonia / Prospective cohort study / Significant association between nurse/patient ratio and late-onset VAP HR: 0.42 (0.18, 0.99) but not for early-onset VAP HR: 0.78 (0.42, 1.45) / Late-onset ventilator associated pneumonia / 3e
Hugonnet, Villaveces & Pittet, 2007 / Adult patients (n=366) admitted to a medical ICU in Geneva, Switzerland / Nurse/patient ratio / Lower nurse staffing (< 1.9) / Nosocomial infection / Case-crossover and cohort designs / Relationship between nurse-patient ratio and infection. For cohort design OR 1.47 (1.03, 2.11); for case-crossover designlogistic analysis OR 1.89 (1.16, 3.07) and GEE analysis OR 1.58 (1.08, 2.33) / Nosocomial infection / 3c & 3d
Kelly, 2012 / Adult patients (n=56,826) who were mechanically ventilated in ICUs in 303 hospitals in the US / No intervention - used mean patients per nurse / No specific comparator / 30-day mortality, ICU and hospital LOS, ventilator associated pneumonia, central line associated bloodstream infection / Retrospective cross sectional / No significant relationship between staffing and 30-day mortality, hospital LOS, central line infection or ventilator associated pneumonia - Significant association between staffing and ICU LOS / ICU LOS / 3e
Kendall-Gallagher & Blegen, 2009 / Adult patients from 48 ICUs in 29 hospitals in the US / No specific intervention: Used total hours of nursing care per day and RN skill mix / No specific comparator / Medication administration errors, falls, skin breakdown, urinary tract infection, blood stream infection, & central catheter infection / Cross sectional / Total hours of nursing care per patient day was positively associated with medication administration errors (0.39; p=0.006). No other significant associations. No significant relationships with skill mix. / Medication administration errors / 4b
Kim et al., 2012 / Adult patients (n=251) in 28 adult ICUs in Korea / No intervention - used nurse/patient ratios / No specific comparator / Mortality from sepsis / Prospective observational study / Hospitals with a 1:2 ratio had significantly lower hospital, ICU and 28-day mortality than those with a 1:3 or 1:4 or more / In-hospital mortality from sepsis / 3e
Lake et al., 2010 / 1293 ICUs in a sample of 5,388 nursing units in 636 hospitals in the US / No intervention - used nursing care hours per patient day / No specific comparator / Inpatient falls / Retrospective cross-sectional study / An additional RN hour per patient day was associated with a 3% lower fall rate in ICUs / Fall rate / 3e
Merrill, 2011 / Nine hospitals in Ottawa (including ICU and step down) / Total productive nursing hours per patient day (HPPD); number productive registered nurses (RN) HPPD, Licensed Practical Nurse (LPN) HPPD, unlicensed assisted personnel (UAP) HPPD; skill mix (percent of RNs) / Standard staffing ratios of 1:1 for ICU and 1:2 or 1:3 for step down departments / Falls, pressure ulcers, catheter-associated urinary tract infection, medication errors / Observational study / Relationship between patient falls and RN HPPD (r = -.764, p = .002) and total HPPD (r = -.753, p = .003). No other significant associations. / Falls / 3e
Mitchell Scott et al., 2014 / Adult and paediatric patients (n=125) in an ED in one hospital in Victoria / No intervention - recorded if there were unfilled nursing deficits but didn't define this / No specific comparator / Medication errors / Prospective descriptive survey / Unfilled nursing deficits resulted in a 19.1% increase in errors of omission. Nursing skill mix not associated with errors of omission / Medication errors / 4b
Neuraz et al., 2015 / Adult patients (n=5,718) in 8 ICUs in 4 hospitals in Lyon, France / Patient-nurse ratio by shift / Categories of patient-nurse ratio: 1:1; > 1:1 to less than or equal to 1.5:1; >1.5:1 to less than or equal to 2:1; > 2:1 to less than or equal to 2.5:1; and > 2.5:1. / ICU mortality / Longitudinal study / Increased risk of mortality for the highest patient-nurse ratios. ICU risk of death increased by factor of 3.5 (1.3-9.1) when ratio > 2.5 patients per nurse. / ICU mortality / 3e
Park, 2011 / 38 acute care hospitals in the US - 60 surgical ICUs / No intervention - staffing calculated at hospital level as nursing care hours per patient day / No specific comparator / Failure to rescue, postoperative sepsis, haemorrhage or haematoma, physiologic and metabolic derangements, respiratory failure, pulmonary embolism, DVT / Retrospective observational study / RN HPPD was associated with higher observed/expected ratios of post-operative haemorrhage and haematoma, no other relationships found. Higher non-RN HPPD related to lower O/E ratios of PE and DVT / Post-op haemorrhage and haematoma; PE; DVT / 3e
Pronovost et al., 2001 / Adult patients (n=2,606) post-abdominal aortic surgery in ICUs in 38 Maryland hospitals / Nurse-to-patient ratio during day and evening / Nurse-to-patient ratio categorised as: 1:1 or 1:2 = "More ICU nurses"; ratios of 1:3 or 1:4 = "fewer ICU nurses" / Pulmonary insufficiency, tracheal reintubation, cardiac complications after a procedure, acute renal failure, septicaemia, acute myocardial infarction, cardiac arrest, surgical complications after a procedure, puncture, perforation or haemorrhage during surgery, reoperation for bleeding / Observational study / Nurse patient ratios of 1:3 or 1:4 were associated with increased risk for any complication; RR 1.7 (1.3 to 2.4), any medical complication; RR 4.5 (1.5 to 2.9), pulmonary insufficiency after procedure; RR 4.5 (2.9-6.9,) and reintubation; RR 1.6 (1.1 to 2.5). All other outcomes non-significant. / Any complication, any medical complication, pulmonary insufficiency after procedure, and reintubation / 3e
Robert et al., 2000 / Adult patients in a surgical ICU in one hospital in the US - 28 patients with BSI and 99 controls / No intervention - looked at nurse-patient ratio in 3 days prior to index date as the exposure variable / No specific comparator / Nosocomial bloodstream infections / Case-control study / Significant relationship between BSI and lower nurse to patient ratios and higher pool nurse to patient ratios / BSI / 3d
Sakr et al., 2015 / Adult ICU patients (n=13,796) in 1,265 ICUs in 75 countries (Western and Eastern Europe, Central and South America, Asia, North America, Oceania, and Africa) / Nurse-patient ratio at two points: 10:00-11:00 (day) and 20:00-23:00 (night) / Nurse-patient ratio < 1:2; 1:1.5-1:1.99; 1:1-1:1.49; >1:1 / In-hospital mortality / Cross sectional study / A nurse-patient ratio >1:1.5 was associated with lower risk of in-hospital mortality. Nurse-patient ratios 1:1.5-1:1.99; OR, 0.84 (0.70-1.01); nurse-patient ratios 1:1-1:1.49; OR, 0.71 (0.57-0.87); nurse-patient ratios > 1:1; OR, 0.69 (0.53-0.90) / In-hospital mortality / 4b
Sales et al., 2008 / Adult patients (n=33,020) in 171 ICUs in 123 acute care hospitals in the US / No intervention - measured nursing hours per patient day and RN skill mix / No specific comparator / Mortality / Retrospective cross sectional study / No significant relationship between staffing and mortality in the ICU / None / 4b
Schwab et al., 2012 / Patients (n=159,400) in 182 ICUs in Germany in 2007 (sample may include paediatric patients) / Nurse-patient ratio per shift and nurse-ventilated patient ratio per shift / Nurse-patient ratio (per shift) and nurse-ventilated patient ratio (per shift) categorised into 4 percentile groups: / Ventilator orcentral venous catheter (CVC) associatedpneumonia and bloodstream infection (BSI) / Observational study / Higher nurse-ventilated patient ratio was significantly associated with a lower incidence of BSI and pneumonia: >p25-p50; IRR 0.77 (0.67-0.88); >p50-p75; IRR 0.64 (0.54-0.75); >p75; IRR 0.42 (0.32-0.55). The nurse-patient ratio was not significantly associated with the incidence of BSI or pneumonia / Ventilator-associated pneumonia; CVC-associated BSI / 3e
Seynaeve et al., 2011 / Adult patients (n=79) admitted to one ICU in Belgium / No specific intervention, counted number of nurses on morning, afternoon and night shifts but didn't include this as a nurse/patient ratio / No specific comparator, used the three shift times / Adverse drug events / Cross-sectional study / Hypoglycaemic episodes significantly higher during night shifts. No significant differences for hypokalaemia or prolonged activated partial thromboplastin time / Hypoglycaemic episodes / 4b
Shuldham et al., 2009 / Adult (n=23,192) and paediatric (n=2,315) cardiac, cardiothoracic, and respiratory patients admitted to critical care areas (intensive care and high dependency units) of two NHS Trust hospitals in England / Nursing hours per patient day (NHPPD) for permanent and temporary nursing staff (hospital staff bank or agency nurses) at ward/unit level / Higher/lower NHPPD / Pressure sores, patient falls, upper gastrointestinal (GI) bleed, pneumonia, sepsis, shock, deep vein thrombosis (DVT) / Retrospective observational study / Significant association between sepsis and the ratio of permanent nursing staff hours. IRR 0.985 (0.972 - 0.998) and IRR 0.983 (0.972 - 0.995). No other significant associations. / Sepsis / 3e
Stone et al., 2007 / Adult patients (n=15,902) in 51 ICUs in 31 hospitals in the US. / No specific intervention - measured RN hours per patient day from monthly data and divided these into quartiles / Compared the 2nd, 3rd and 4th quartiles to the 1st quartile / Central line associated bloodstream infections, ventilator-associated pneumonia, catheter-associated urinary tract infections, 30-day mortality, pressure ulcer / Observational study / Units with higher staffing had lower incidence of CLBSI (3rd quartile OR 0.32), VAP (4th quartile OR 0.21), 30-day mortality (3rd quartile OR 0.81) and PU (3rd quartile OR 0.69). Other quartiles for these outcomes and all quartiles for CAUTI not significant / CLBSI, VAP, 30-day mortality and PU / 4b
Tarnow-Mordi et al., 2000 / Adult medical and surgical patients (n=1,025) admitted to an ICU in a hospital in Scotland / ICU workload measured by occupancy, ratio of occupied to appropriately staffed beds, and nursing requirement per shift / Overall ICU workload during stay categorised as moderate, intermediate, and high (composite measure of average nursing requirement per occupied bed per shift and peak occupancy in any shift in the stay). / In-hospital mortality / Retrospective analysis from prospective cohort study / Adjusted mortality was related to (i) ratio of occupied to appropriately staffed beds per shift and (ii) peak occupancy. / In-hospital mortality / 3e
van den Heede et al., 2009 / Adult patients (n=9054) admitted to 28 Belgian acute care hospitals for a CABG or heart valve procedure. / No specific intervention - calculated NHPPD for each nursing unit / Modelled the outcome to consider the result if all units increased their staffing to the 75th percentile / in-hospital mortality / Observational study / No significant association between ICU staffing levels and mortality / None / 3e
van der Kooi et al. 2015 / Adult patients (n=379) admitted to 25 ICUs in 25 hospitals in the Netherlands / Nurse to patient ratio ≤ 1 to 1 / Nurse to patient ratio >1 to 1 / Restraint use / Cross-sectional / No significant relationship between restraint use and nurse to patient ratio / None / 4b
West et al., 2014 / Adult patients (n=38,168) admitted to 65 ICUs in England / Number of FTE permanent direct care registered nurses (RNs) and health care assistants / Lower number of FTE permanent direct care RNs / Mortality (in-hospital and in-ICU) / Observational study / Significant relationship between number of direct care nurses per bed and ICU mortality; OR 0.90 (0.84, 0.97) and hospital mortality; OR 0.92 (0.87, 0.98). / ICU mortality and in-hospital mortality / 3e
Whitman et al., 2002 / NSO rates in 95 patient care units across 10 adult acute care hospitals in the US. Included 15 cardiac ICUs and 7 non-cardiac ICUs / No specific intervention - calculated NHPPD / No specific comparator / Central line blood-associated infections, pressure ulcers, falls, medication errors, restraint application duration / Secondary analysis of prospective observational data / No significant relationships between staffing and any outcome for non-cardiac ICUs. Significant relationship between staffing and medication error in cardiac ICUs (r = -.55), all other relationships in cardiac ICUs non-significant / Medication errors / 4b

Levels of evidence: Joanna Briggs Institute. (2013). New JBI levels of evidence. Adelaide, SA, Joanna Briggs Institute. Retrieved from