Shift length and nurse error rate

[granolapher]

This is a review of an article published in the International Journal of Nursing Studies that examines the effect of shift length on errors

Original review from FOANed Reviews,

study from the International Journal of Nursing Studies.

Jill Clendon and Veronique Gibbons published a systematic review in the International Journal of Nursing Studies that examined the effects of extended shifts on error rates. They reviewed the Cochrane, Joanna Briggs Institute, CINAHL, MEDLINE, Embase, Current Contents, Proquest, and Dissertations International databases for English language research (published or unpublished) conducted prior to August 2014. 5429 studies were identified using: "length of time", "shift work", "error rate", and "nursing" key words and assessed for review. 86 studies were identified as potentially relevant; 60 did not meet inclusion criteria and were discarded, 26 studies met the broad definition and were included for review.

The 26 Studies that met the broad inclusion criteria were quantitative observational study designs, of nurses (any level, excluding students), who work in acute care hospitals, that compared extended (12hrs or more) to non-extended (

Jill Clendon and Veronique Gibbons published a systematic review in the International Journal of Nursing Studies that examined the effects of extended shifts on error rates. They reviewed the Cochrane, Joanna Briggs Institute, CINAHL, MEDLINE, Embase, Current Contents, Proquest, and Dissertations International databases for English language research (published or unpublished) conducted prior to August 2014. 5429 studies were identified using: "length of time", "shift work", "error rate", and "nursing" key words and assessed for review. 86 studies were identified as potentially relevant; 60 did not meet inclusion criteria and were discarded, 26 studies met the broad definition and were included for review.

The 26 Studies that met the broad inclusion criteria were quantitative observational study designs, of nurses (any level, excluding students), who work in acute care hospitals, that compared extended (12hrs or more) to non-extended (

The authors note that the results are too heterogeneous (X = 34.29, p = 0.01) to draw conclusions on the whole population reviewed; but they do discuss some conclusions drawn by studies of similar designs and results. 4 studies found lower error rates associated with extended work days, 6 found higher error rates with extended work days, and 3 found no difference. The 6 studies reporting higher error rates comprised 89% (60,780) of the total review population (n= 67,967), this is what the authors spend the bulk of the paper discussing.

Half (n = 31,627) of the patients in this cohort come from a 2014 study (Griffiths et al) that assessed rates of unfinished work among nurses of varying shift lengths. The study showed higher rates of unfinished care with extended shifts (OR = 1.13; CI 95% 1.09-1.16); but failed to discuss the characteristics of the studied groups, and the nature of the unfinished care. Because the RN's surveyed in this study came from 488 hospitals across 12 European countries, failing to discuss the similarity or differences in patient acuity and hospital census weakens the evidence, as regional characteristics could result in large differences between groups; similarly, not knowing the nature of the unfinished work prevents extrapolation of what the clinical effects of these errors were.

The next largest subset of this group study (n= 25,985) to associate negative outcome with extended shift length was a study that assessed the differences in central line associated infections, urinary tract infections, and pain control among patients cared for by groups of nurses working extended or non-extended shifts. These studies are two separate analyses of results from the same study. The study did find central line associated infections were 2.5 (n = 3710) times more likely to be reported in areas were nurses worked extended shift, that pain was less well controlled (OR = 0.9, p

Of the remaining 3 studies 2 assessed for rates of reported error, or near miss errors. One study (Rogers et al., 2004) used a binary (yes/no) system to code for error during extended work days. They found that extended shifts were associated with an error rate 3 times higher than non-extended shifts (OR = 3.29; p = 0.001). The second study, by Scott et al (2006), used a self reporting log of error, or near errors to assess for risk of error (vigilance) in extended shifts. They found that extended shifts resulted in an "error/risk for error events" rate twice that of regular shifts (OR 1.94, p = 0.03). Taken together these studies observed error rates 2-3 times higher in extended shifts. These are alarming numbers; but the evidence may not be strong enough to make conclusions on the effects of shift length on adverse events. Unfortunately a binary system of coding errors prevents any discussion about the true error incidence rate (actual number of errors), nature of the error, or clinical relevance of the error. Furthermore a tool that assesses for both error and "near error: is likely to be a very sensitive tool, one that doesn't differentiate between error types (true rates of actual error and near error), and is likely to produce a disproportionately large rate of "positives", which limits the clinical relevance of the findings.

The final study was a longitudinal survey (n = 2273) that examined the rate of needlestick injury for nurses working differing length shifts. It found that nurses who worked 12 hours or more were more likely (OR = 1.68, p = 0.001) to have a needlestick injury in the previous year compared to nurses working less than 12 hrs. This, much like the study examining pain and infection rate would be need to be reviewed to assess if the statistically significant increase was in any way clinically significant.

The authors suggest that there may be an increased risk for errors associate with extended work days. However the generalizability of this review is limited by the heterogeneity of the studies, the conflicting results, and the broad definition of error. It does provide an overview of some of the literature examining the effect of shift length on error rates; but it falls far short of generating usable recommendations. I don't think the quality of evidence presented here is strong enough to make generalizations about the safety of extended shifts in general. From a nursing perspective I would suggest reviewing the primary research included in this review in isolation if you're attempting to examine the effects of extended work days on nurse or patient clinical outcomes. I would suggest considering your own quality of life measures if you're attempting to determine if extended work days are a better fit for your individual nursing practice.

Clendon, J., & Gibbons, V. (2015). 12h shifts and rates of error among nurses: A systematic review.International journal of nursing studies.

12 h shifts and rates of error among nurses: A systematic review. - PubMed - NCBI

[MunoRN, RN]

That's a really long way of saying they are just repeating the findings of the Rogers (2004) and Scott (2006) studies, which represent a failure to understand the basic principles of statistical analysis so severe and moronic, we should all hang our heads in shame.

The often repeated "statistic", that nurses working 12 hours or more are 2-3 times more likely to have an error or near error, comes from the Rogers study. The error rate in that study was based on errors per shift. Per shift. Per shift. At this point anyone who passed a statistics 101 class, and probably even some who failed it, should be able to recognize the flaw in their analysis.

When looking at potential for risk, the potential is calculated as the potential for risk given a the same opportunity for risk between the two groups you are comparing. "Per shift" is not an equal opportunity for error when you are talking about two different shift lengths. Lets say for a given patient load there 100 opportunities for error in a 24 hour period. That means that for 8 hour shifts, the per shift opportunity for error is 33, while the opportunity for error in a 12 hour shift is 50. So to correctly calculate risk for error in a way where you could actually compare the two, you would have to correct for this differing opportunity for error, for instance you could calculate errors per hour.

If you take the Rogers et al data, the risk is actually about the same, but the 12 hour data is skewed because it includes 12 hour shifts as well as all shifts longer than 12 hours, while the 8 hour data did not include any shifts longer than 8 hours (it excluded excess of shift which is one if the best predictors of error rate). The Scott et al study did actually include error rates for 12 hour shifts and 8 hour shifts equally, and per hour worked the error rate was actually much lower in 12 hour shifts.

[granolapher]

That's a great point, and one that I failed to cosider on top of all the other shortcomings of the research. Thanks for bringing it up.

[ambr46]

All I know is that I am tired after 12 hours and don't have the same mental energy and clarity that I have earlier in the day. What is really scary though is the nurse who clocks out after 12 to head to their other job. Yikes!!!

[ambr46]

I did work with a nurse who was so burned out from working so much that she was making errors left and right. She finally got fired after one of her patients nearly chocked on their pulls. According to the CNA who was present at the time, the nurse came in, dumped pills in the sleeping patients mouth, and proceeded to walk out. I don't know what happened to the patient but I do know that she was terminated shortly after due to accumulating errors.