American Journal of Infection Control 43 (2015) 348-53
Contents lists available at ScienceDirect
American Journal of Infection Control
American Journal of Infection Control
journal homepage: www.ajicjournal.org
Major article
Novel low-resource intervention reduces urinary catheter use and associated urinary tract infections: Role of outcome measure bias? Blair Carl Schwartz MDCM, MHS, FRCPC a, *, Charles Frenette MD, FRCPC b, Todd C. Lee MD, MPH, FRCPC a, c, Laurence Green MD a, Dev Jayaraman MD, MPH, FRCPC a a
Division of General Internal Medicine, Department of Medicine, McGill University, Montreal, QC, Canada Division of Infectious Diseases, Infection Prevention and Control Service, Department of Medicine, McGill University Health Center, Montreal, QC, Canada c Division of Infectious Diseases, Department of Medicine, McGill University, Montreal, QC, Canada b
Key Words: Device-related infection Urinary tract infection Quality improvement Nosocomial
Background: Previous interventions targeting nosocomial urinary tract infections have reduced catheterization and infections, but they require significant resources and may be susceptible to misclassification and surveillance bias. We sought to determine the effectiveness of a novel intervention at reducing catheterization and infections while exploring the potential for bias. Methods: We conducted a prospective study of a brief monthly in-person educational intervention focusing on appropriate urinary catheter use. Results: We studied 1,335 patients (13,753 patient days) on 1 control and 1 intervention ward. After the intervention, the device utilization rate was significantly reduced, with a relative risk of 0.49 (95% confidence interval [CI], 0.32-0.76; P ¼ .001) versus 1.02 (95% CI, 0.58-1.82; P ¼ .93) for controls. Both wards demonstrated a reduction in catheter-associated infections after intervention, with an intervention relative risk of 0.42 (95% CI, 0.16-1.08; P ¼ .07) and 0.51 (95% CI, 0.22-1.20; P ¼ .12) for controls. There was no change in the rate of all nosocomial urine infections, with an intervention relative risk of 0.79 (95% CI, 0.38-1.65; P ¼ .53) and 0.89 (95% CI, 0.48-1.67; P ¼ .72) for controls. Conclusion: Our study demonstrates that our novel educational intervention significantly reduces urinary catheter use in hospitalized patients. The trend towards reduced catheter-associated infections after intervention, coupled with the absence of an improvement in all nosocomial infections supports a potential role of misclassification bias. We suggest that future prospective investigations explore this phenomenon using more robust outcome measures. Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
Urinary tract infections (UTIs) account for nearly 40% of nosocomial infections, and approximately 60%-80% of these are related to the use of a urinary catheter.1,2 After central venous catheters, catheter-associated urinary tract infections (CAUTIs) are the second most common source of device-related blood stream infections and are associated with a 2.8-fold increased odds of death.3 Although a
* Address correspondence to Blair Carl Schwartz, MDCM, MHS, FRCPC, Division of General Internal Medicine, Jewish General Hospital, 3755 Cote Sainte Catherine, G-050, Montreal, QC, H3T 1E2, Canada. E-mail address: [email protected] (B.C. Schwartz). Previous Presentation: The design and preintervention period data from this study were presented in poster format at the International Forum on Quality and Safety in Healthcare, April 16-19, 2013, London, England. Funding/Support: Dr Schwartz was supported by a Health Professionals Training Award (no. 24825) from the Fonds de Recherche Québec - Santé. Conflicts of interest: None to report.
valuable tool in selected cases, some 40%-50% of urinary catheter use in hospitals is without clinical indication.4,5 Consequently, a reduction in the inappropriate use of urinary catheters represents a major strategy in preventing nosocomial UTIs and has been endorsed by the Choosing Wisely Campaign.6 Several interventions have been shown to reduce urinary catheter use.7-13 Although effective, these interventions often require electronic order entry systems,7,9 additional nursing rounds,11,12 or dedicated urinary catheter nurses,8,10 all of which are resources that may not be available in all institutions. No low-resource intervention for the prevention of urinary catheter use in hospital has previously been developed and evaluated. Low-resource approaches to quality improvement have the advantage of ease of implementation regardless of financial resources and with minimal additional workload for already busy health care workers. Such interventions have been successfully
0196-6553/$36.00 - Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajic.2014.12.006
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
used to improve antibiotic prescription behavior in the hospital setting,14 and interventions combining education with a focus on safety culture have been among the most effective and recommended models to date.15,16 Furthermore, prior studies have used the CAUTI rate as their primary effectiveness outcome measure.8-10,17 In these studies, CAUTI is defined as a UTI that occurred in a patient with a catheter in place at the time of infection. However, it is possible that study subjects who had their catheter removed because of the intervention under study, go on to experience a UTI thereafter. Such subjects would be classified as experiencing a UTI, and not a CAUTI, even though the root cause may have been the catheter. As a result, the postintervention CAUTI rate may represent a biased underestimate. The role played by this misclassification bias in studying the effectiveness of quality improvement interventions to prevent nosocomial UTIs has not been previously evaluated. We investigated the effectiveness of a novel, low-resource, educational intervention designed to reduce urinary catheter use and nosocomial UTIs. We developed an educational intervention targeting ward physicians and nurses that focused on accepted urinary catheter indications and alternatives, ward data on the actual burden of nosocomial UTIs, and establishing a safety culture based on avoiding unnecessary catheter use on an internal medicine clinical teaching unit. We performed a prospective controlled trial to evaluate our hypotheses that our novel brief educational intervention leads to a reduction in urinary catheter use and CAUTI rate. We further explored whether the CAUTI rate might be susceptible to misclassification bias. MATERIALS AND METHODS Study design and setting We conducted a prospective, controlled, intervention trial on 1 internal medicine (46 beds) and 1 neurosciences ward (35 beds) at the Montreal General Hospital (Montreal, Canada), a 417-bed tertiary care hospital affiliated with McGill University. The study was conducted from December 5, 2011-August 24, 2012, encompassing 2 months of preintervention, 2 months of intervention, and 5 months of postintervention follow-up. Subjects Our cohort included all patients consecutively admitted to the internal medicine and neurosciences wards at the Montreal General Hospital during the study period. Subjects were excluded if they had a permanent urinary catheter, had a urinary catheter inserted by the urology service during admission, or were admitted to the long-term care service on the internal medicine ward. Subjects were followed until death, hospital discharge, transfer to another service, or study conclusion. This study was granted a waiver of exemption by the McGill University Ethics Review Board. Intervention The study began with 2 months of preintervention data collection on both wards, during which time all clinical personnel were blinded to study objectives. Subsequently, the medical and nursing personnel on the internal medicine ward each received an educational intervention. In contrast, the medical and nursing personnel on the control neurosciences ward received no educational intervention.
349
On the intervention ward, the medical and nursing personnel received separate education sessions. The medical personnel (attending physicians, residents, medical students) received, on the first day of each 1-month clinical rotation, a 15-minute education session given by a project leader. The presentation included information on epidemiology and morbidity of nosocomial UTIs; accepted indications for, and alternatives to, urinary catheter use with emphasis on reducing inappropriate catheter use; and urinary catheter use and UTI incidence data for their ward from the preintervention phase. The nursing personnel received, during an education day prior to the postintervention phase, a single 45-minute education session given by the same project leader. The presentation included all the elements of the medical personnel session, with additional emphasis on the use of condom catheters, straight in-out catheterization, enhanced mobilization, and good bowel hygiene as a means to avoid urinary catheter use. Discussion was held on anticipated nursing barriers to implementation of a catheter reduction strategy, and solutions were identified in a multidisciplinary fashion. Both medical and nursing interventions included strong messages of a ward-wide commitment to a safety-based culture of reduced urinary catheter usage, but neither focused on techniques surrounding insertion or maintenance of these devices. All sessions were attended by representatives of ward medical and nursing leadership, who emphasized their support for this intervention and made themselves available to all members of the staff during the intervention period. During the postintervention period, control ward personnel remained blinded to study objectives. Postimplementation blinding of personnel on the intervention ward was precluded by the nature of the intervention; however, study personnel made no attempts to influence clinical practice on that ward. Outcomes The coprimary outcomes were the device utilization rate (DUR) (catheter days/100 patient days) and CAUTI rate (number of CAUTIs/1,000 patient days). Secondary outcomes included rates per 1,000 patient days of noncatheter-associated UTIs, all urinary infections, positive urine cultures, and urine sampling (number of urine cultures/ 1,000 patient days). The proportion of patients who ever had a urinary catheter and duration of catheterization were also recorded. As balancing patient safety measures we tracked the catheter reinsertion rate (catheter reinsertions per 1,000 patient days) and incidence of urogenital trauma during reinsertion, with the latter requiring immediate reporting to the principal investigator. A catheter day was defined as the presence of a urinary catheter at 10:00 AM each day, and a positive urine culture was defined as the growth of >105 colony forming units/mL of a single uropathogenic species. A UTI was defined as the presence of a positive culture plus at least one of the following: fever >37.9 C, peripheral white blood cell count >10.5 109/L, urinary tract symptoms, mental status change, or new onset hypotension, all in the absence of another evident clinical explanation. A UTI was classified as catheter associated if the culture was sent with a catheter in place or within 7 days of catheter removal. Catheter reinsertion was defined as reinsertion of a previously removed urinary catheter within 7 days of initial removal. Data were collected Monday-Friday by trained study medical personnel using a specifically designed data extraction sheet; weekend data were abstracted the following Monday.
350
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
Table 1 Cohort demographic and clinical history by ward and intervention period Control ward Total (n ¼ 1,335)
Preintervention (n ¼ 126)
Postintervention (n ¼ 348)
Median age (IQR) Median LOS (IQR) (d) Male sex (%) Diabetes (%) Admission ABX (%) Admission UTI (%)
69 (52-80) 9 (4-17) 61.9 28.6 34.9 10.3
67 (49-81) 8 (4-17) 62.4 20.4 34.2 12.9
Intervention ward P value* .76 .64 .93 .06 .88 .44
Preintervention (n ¼ 239)
Postintervention (n ¼ 622)
69 (57-80) 7 (4-11) 56.5 29.3 66.1 9.2
70 (56-81) 6 (3-11) 52.3 33.4 59.9 8.7
P value* .74 .02 .28 .24 .10 .81
ABX, antibiotics; IQR, interquartile range; LOS, length of stay; UTI, urinary tract infection. *P values from Pearson c2 test for categorical variables and Wilcoxon rank-sum test for continuous variables.
Demographic data (age and sex), comorbidity (diabetes status), and baseline clinical data (UTI or antibiotic history in hospital prior to current ward admission) were collected from the ward admission note and laboratory information system. Urinary catheter presence and reinsertion status were determined from the nursing records, with bedside confirmation in the case of discrepant or unclear recording. Information regarding urinary culture sampling, positive culture results, and leukocytosis were determined from the computerized laboratory and microbiology system. The complete medical record of all patients with positive urine cultures was reviewed to determine classification as a UTI. In the case of uncertainty, the data collector discussed the outcome with a senior internal medicine physician (D.J.) or infectious disease physician (C.F.) to obtain consensus.
Table 2 Preintervention outcome rates and 95% CI by ward Control ward Outcome DUR (per 100 patient days) CAUTI rate N-CAUTI rate NUTI rate Positive culture rate
Intervention ward
Preintervention*
95% CI
Preintervention*
95% CI
17.0
15.1-19.1
15.7
14.1-17.4
5.4 3.0 8.4 14.4
2.5-10.2 0.9-7.0 4.6-14.1 9.2-21.5
3.4 0.8 4.3 7.7
1.5-6.7 0.1-3.1 2.1-7.9 4.6-12.2
CAUTI, catheter-associated urinary tract infection; CI, confidence interval; DUR, device utilization rate; N-CAUTI, noncatheter-associated urinary tract infection; NUTI, nosocomial urinary tract infection. *Preintervention rate per 1,000 patient days unless otherwise indicated.
RESULTS Statistical analysis
Subjects
Statistical analysis was performed using Stata 12 (StataCorp, College Station, TX), with a prespecified significance level of P < .05. Pre- and postintervention subject demographic and clinical data were compared with the Wilcoxon rank-sum test for continuous variables, and the Pearson c2 test was used for categorical data. Primary catheter and UTI outcome measures were analyzed by determining the incidence rate ratio (IRR) comparing the postintervention rate with the preintervention rate for each individual ward. The crude DUR IRR was determined by negative binomial regression after a Poisson goodness-of-fit test, data exploration, and likelihood-ratio test of the a coefficient indicated significant overdispersion in the DUR data. An adjusted IRR was determined by fitting a multivariate negative binomial regression model with the prespecified covariates of sex and history of UTI on admission. Crude and adjusted infectious outcome measure IRRs were determined by Poisson regression models fitted with and without the prespecified covariates of age, sex, diabetes history, and history of UTI or antibiotic use prior to ward admission. Model adequacy was tested with the use of the Poisson goodness-of-fit test. The IRRs for urine culture sampling and catheter reinsertion were determined in an unadjusted manner with the use of negative binomial regression because of overdispersion of data. Proportions of positive cultures and patients with urinary catheters in each time period were compared by Pearson c2 analysis. We performed 2 post hoc sensitivity analyses of the DUR outcome by excluding subjects with catheter duration >14 and >7 days to ensure that findings were not influenced by the presence of prolonged catheterization and conducted 1 post hoc subgroup analysis evaluating the effect of subject sex on the intervention to reduce the DUR.
We evaluated 1,364 patients who were consecutively admitted to the 2 study wards during the 2-month preintervention and 5-month postintervention periods. Of these patients, 26 (1.9%) were excluded from analysis because of a permanent (n ¼ 19) or urology-inserted (n ¼ 7) urinary catheter. A further 3 patients (0.2%) had missing data. Given that this data was the result of clerical error, we classified the data as missing completely at random, and given the low percentage, we excluded these 3 patients. Therefore, 1,335 patients, representing 13,753 patient days, were included in our final analysis. Within ward, demographic and clinical factors were evenly distributed when comparing the pre- and postintervention periods, apart from a 1-day shorter median length of stay on the intervention ward (Table 1).
Outcomes The preintervention period rates of all catheter and infectious outcome measures for both wards are listed in Table 2. Urinary catheter utilization After the intervention there was a highly significant reduction in the DUR among intervention ward patients, with an unadjusted IRR of 0.52 (95% confidence interval [CI], 0.34-0.81; P ¼ .003), whereas there was no significant change in catheter use among control ward patients, with an unadjusted IRR of 1.1 (95% CI, 0.62-1.97; P ¼ .74). (Table 3). On adjusting for the potential confounders of patient sex and admission history of UTI, the findings on intervention effectiveness were unchanged. The proportion of patients with a urinary catheter in the intervention period dropped significantly on the intervention ward (26.78% to 16.4%, P ¼ .0005), with no change on
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
351
Table 3 Crude and adjusted device utilization rate outcome IRRs by ward and intervention period Crude (unadjusted) Variable Intervention ward Postintervention Preintervention Control ward Postintervention Preintervention Sex Female Male UTI at admission Yes No
IRR
95% CI
0.52 1.0 (reference)
0.34-0.81 NA
1.1 1.0 (reference)
Adjusted* P value
IRR
95% CI
P value
.003 NA
0.49 1.0 (reference)
0.32-0.76 NA
.001 NA
0.62-1.97 NA
.74 NA
1.02 1.0 (reference)
0.58-1.82 NA
.93 NA
0.83 1.0 (reference)
0.61-1.15 NA
.27 NA
0.78 1.0 (reference)
0.56-1.07 NA
.12 NA
1.76 1.0 (reference)
1.06-2.91 NA
.028 NA
1.94 1.0 (reference)
1.17-3.24 NA
.011 NA
CI, confidence interval; IRR, incidence rate ratio; NA, not applicable; UTI, urinary tract infection. *Model adjusted for sex and history of UTI on admission.
the control ward (30.17% to 28.17%, P ¼ .67). There was no significant change in median duration of catheterization postintervention on either ward (data not shown). A history of UTI prior to ward admission was associated with a significantly increased DUR compared with patients with no UTI on admission (IRR, 1.94; 95% CI, 1.17-3.24; P ¼ .011). There was no significant association between patient sex and DUR in either univariate or multivariable regression (adjusted IRR, 0.78; 95% CI, 0.56-1.07; P ¼ .12), and there was no evidence for sex effect modification (P ¼ .42 for interaction term). Sensitivity analyses of the primary outcome, after excluding subjects with catheter duration of >14 and >7 days, respectively, did not show differences in point estimates or inference for intervention effectiveness (data not shown). UTI outcomes Among intervention ward patients, the postintervention period was associated with a reduced unadjusted CAUTI rate (IRR, 0.39; 95% CI, 0.15-1.02; P ¼ .054) and a reduced age, sex, diabetes and history of UTI or antibiotic use adjusted CAUTI rate (IRR, 0.42; 95% CI, 0.16-1.08; P ¼ .072), both of which however failed to reach statistical significance (Table 4). Among control patients, the intervention period was associated with a nonsignificant reduction in the adjusted CAUTI rate (IRR, 0.51; 95% CI, 0.22-1.20; P ¼ .12). The intervention period was associated with nonsignificant increases in the adjusted incidence rates of noncatheter-associated UTIs in both groups of patients; the increase was 2.40-fold (95% CI, 0.53-10.7; P ¼ .25) among intervention patients and 1.57-fold (95% CI, 0.59-4.21; P ¼ .37) among control patients. There were no significant changes in the rate of all nosocomial UTIs comparing the postintervention to the preintervention period on either the intervention ward (IRR, 0.79; 95% CI, 0.38-1.65; P ¼ .53), or the control ward (IRR, 0.89; 95% CI, 0.48-1.67; P ¼ .72). Similarly, there were no within-ward differences in the rate of positive urine cultures (intervention ward: IRR, 0.82; 95% CI, 0.471.46; P ¼ .50; control ward: IRR, 0.89; 95% CI, 0.55-1.43; P ¼ .63). Urine culture sampling Compared with the preintervention period, the rate of urine culture sampling on the control ward decreased significantly, by 25% (P ¼ .04), in the postintervention period. In contrast, on the intervention ward, there was no significant change in rate of urine culture sampling. There was no significant change in the proportion of urine samples returning as positive cultures between the pre- and postintervention periods on either ward (Table 5).
Table 4 Crude and adjusted infectious outcome IRRs by ward and intervention period Unadjusted Outcome CAUTI Control Intervention N-CAUTI Control Intervention NUTI Control Intervention Positive cultures Control Intervention
Adjusted*
IRRy
95% CI
P value
IRRy
95% CI
P value
0.54 0.39
0.23-1.26 0.15-1.02
.15 .054
0.51 0.42
0.22-1.20 0.16-1.08
.12 .072
1.49 2.65
0.56-3.97 0.59-11.8
.42 .20
1.57 2.40
0.59-4.21 0.53-10.7
.37 .25
0.88 0.80
0.47-1.64 0.39-1.68
.69 .56
0.89 0.79
0.48-1.67 0.38-1.65
.72 .53
0.89 0.86
0.55-1.43 0.49-1.52
.62 .60
0.89 0.82
0.55-1.43 0.47-1.46
.63 .50
NOTE. Poisson goodness-of-fit for all models, P ¼ 1.00. CAUTI, catheter-associated urinary tract infection; CI, confidence interval; DUR, device utilization rate; IRR, incidence rate ratio; N-CAUTI, noncatheter-associated urinary tract infection; NUTI, nosocomial urinary tract infection. *Adjusted for age, sex, diabetes status, history of UTI on admission, and history of antibiotics on admission. y Reference group for all IRRs is the preintervention period for each ward.
Safety outcomes There were no postintervention differences in the rate of catheter reinsertion on either ward (intervention ward: IRR, 0.97; 95% CI, 0.44-2.30; P ¼ .92; control ward: IRR, 0.79; 95% CI, 0.32-2.12; P ¼ .58). There were no cases of reinsertion-related urogenital trauma reported during the study (Table 5).
DISCUSSION Reduction in catheter use Our low-resource intervention led to a 50% reduction in the DUR. This rate of reduction exceeded that achieved in 2 other investigations of urinary catheter interventions. Fakih et al11 achieved a 20% reduction in catheter use by incorporating a dedicated urinary catheter nurse onto daily ward multidisciplinary rounds. Knoll et al7 also achieved a 20% reduction in the DUR, with the use of a dedicated nurse to prompt catheter removal, and a 10% reduction in the DUR during a phase of intense education and audit and feedback. In both phases of Knoll’s study, however, device use returned to baseline rates after the intense intervention was stopped. In contrast with the present and aforementioned studies, a greater reduction in the DUR (81%) on a medical unit was reported by Topal
352
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
Table 5 Urine culture and catheter reinsertion outcomes by ward and intervention period Control ward Variable Urine sampling rate* % positive cultures Catheter reinsertion rate*
Intervention ward
Preintervention
Postintervention
P value
Preintervention
Postintervention
P value
45.7 31.6 4.8
34.3 37.3 3.8
.04 .40y .58
44.5 17.3 4.3
39.1 16.4 4.2
.29 .84y .92
*Rates per 1,000 patient days, and P values for incidence utilization rate. Pearson c2 test.
y
et al,9 with the use of a computerized order entry-based reminder system. Therefore, our reported DUR reduction from a lowerresource intervention compares favorably with previous reports of resource intense interventions. That the proportion of patients ever with a catheter dropped significantly postintervention, with no change in median catheter duration, suggests that the main effect of our intervention was in prompting removal on ward admission of unnecessary catheters, a key component of our educational intervention, which highlighted admission as a cardinal time to make providers aware their patient had a catheter and to evaluate continued need. Interestingly, we did find that a history of UTI on admission was strongly associated with an increase in the DUR, which may reflect the longstanding notion that these infections must be completely treated prior to catheter removal. This finding may represent a target for further education about appropriate catheter utilization. Reduction in CAUTI Our educational intervention was associated with a reduced rate, albeit not statistically significant, of CAUTI in the postintervention period. Given the relative rarity of the outcome, it is possible that we did not accrue enough person time to reach statistical significance during the time of our study. Although also failing to reach significance, we did unexpectedly find a marginal reduction in the CAUTI rate on the control ward during the postintervention period, despite no change in catheter use. Although this may equally represent chance, this has not previously been evaluated or reported. This observation highlights the importance of control arms in evaluations of quality improvement interventions in comparison with the commonly used before and after design. Such control arm designs permit a more robust assessment of the true improvement effect of the studied intervention and allows for the detection of secular trends and other confounders. Possible surveillance bias Our findings suggest that there may have been surveillance bias. On our control ward, which received no educational intervention, there was a significantly reduced rate of urine culture sampling during the postintervention period compared with preintervention, despite a similar proportion of positive cultures. The reduced rate of sampling could be attributed to a reduction in unnecessary cultures or a true decrease in the incidence of UTIs on the control floor. Although we masked the nature of our study on the control floor and no unit personnel crossed over between units, personnel may have become aware of our study objectives through spillover. The nonstandardized approach to urine sampling, seen in many health care settings, implies that the UTI incidence rate may be closely related to the frequency with which they are sought. Recently, Leis et al18 demonstrated that inpatient urine cultures are associated with a confirmed infection in
Contents lists available at ScienceDirect
American Journal of Infection Control
American Journal of Infection Control
journal homepage: www.ajicjournal.org
Major article
Novel low-resource intervention reduces urinary catheter use and associated urinary tract infections: Role of outcome measure bias? Blair Carl Schwartz MDCM, MHS, FRCPC a, *, Charles Frenette MD, FRCPC b, Todd C. Lee MD, MPH, FRCPC a, c, Laurence Green MD a, Dev Jayaraman MD, MPH, FRCPC a a
Division of General Internal Medicine, Department of Medicine, McGill University, Montreal, QC, Canada Division of Infectious Diseases, Infection Prevention and Control Service, Department of Medicine, McGill University Health Center, Montreal, QC, Canada c Division of Infectious Diseases, Department of Medicine, McGill University, Montreal, QC, Canada b
Key Words: Device-related infection Urinary tract infection Quality improvement Nosocomial
Background: Previous interventions targeting nosocomial urinary tract infections have reduced catheterization and infections, but they require significant resources and may be susceptible to misclassification and surveillance bias. We sought to determine the effectiveness of a novel intervention at reducing catheterization and infections while exploring the potential for bias. Methods: We conducted a prospective study of a brief monthly in-person educational intervention focusing on appropriate urinary catheter use. Results: We studied 1,335 patients (13,753 patient days) on 1 control and 1 intervention ward. After the intervention, the device utilization rate was significantly reduced, with a relative risk of 0.49 (95% confidence interval [CI], 0.32-0.76; P ¼ .001) versus 1.02 (95% CI, 0.58-1.82; P ¼ .93) for controls. Both wards demonstrated a reduction in catheter-associated infections after intervention, with an intervention relative risk of 0.42 (95% CI, 0.16-1.08; P ¼ .07) and 0.51 (95% CI, 0.22-1.20; P ¼ .12) for controls. There was no change in the rate of all nosocomial urine infections, with an intervention relative risk of 0.79 (95% CI, 0.38-1.65; P ¼ .53) and 0.89 (95% CI, 0.48-1.67; P ¼ .72) for controls. Conclusion: Our study demonstrates that our novel educational intervention significantly reduces urinary catheter use in hospitalized patients. The trend towards reduced catheter-associated infections after intervention, coupled with the absence of an improvement in all nosocomial infections supports a potential role of misclassification bias. We suggest that future prospective investigations explore this phenomenon using more robust outcome measures. Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
Urinary tract infections (UTIs) account for nearly 40% of nosocomial infections, and approximately 60%-80% of these are related to the use of a urinary catheter.1,2 After central venous catheters, catheter-associated urinary tract infections (CAUTIs) are the second most common source of device-related blood stream infections and are associated with a 2.8-fold increased odds of death.3 Although a
* Address correspondence to Blair Carl Schwartz, MDCM, MHS, FRCPC, Division of General Internal Medicine, Jewish General Hospital, 3755 Cote Sainte Catherine, G-050, Montreal, QC, H3T 1E2, Canada. E-mail address: [email protected] (B.C. Schwartz). Previous Presentation: The design and preintervention period data from this study were presented in poster format at the International Forum on Quality and Safety in Healthcare, April 16-19, 2013, London, England. Funding/Support: Dr Schwartz was supported by a Health Professionals Training Award (no. 24825) from the Fonds de Recherche Québec - Santé. Conflicts of interest: None to report.
valuable tool in selected cases, some 40%-50% of urinary catheter use in hospitals is without clinical indication.4,5 Consequently, a reduction in the inappropriate use of urinary catheters represents a major strategy in preventing nosocomial UTIs and has been endorsed by the Choosing Wisely Campaign.6 Several interventions have been shown to reduce urinary catheter use.7-13 Although effective, these interventions often require electronic order entry systems,7,9 additional nursing rounds,11,12 or dedicated urinary catheter nurses,8,10 all of which are resources that may not be available in all institutions. No low-resource intervention for the prevention of urinary catheter use in hospital has previously been developed and evaluated. Low-resource approaches to quality improvement have the advantage of ease of implementation regardless of financial resources and with minimal additional workload for already busy health care workers. Such interventions have been successfully
0196-6553/$36.00 - Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajic.2014.12.006
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
used to improve antibiotic prescription behavior in the hospital setting,14 and interventions combining education with a focus on safety culture have been among the most effective and recommended models to date.15,16 Furthermore, prior studies have used the CAUTI rate as their primary effectiveness outcome measure.8-10,17 In these studies, CAUTI is defined as a UTI that occurred in a patient with a catheter in place at the time of infection. However, it is possible that study subjects who had their catheter removed because of the intervention under study, go on to experience a UTI thereafter. Such subjects would be classified as experiencing a UTI, and not a CAUTI, even though the root cause may have been the catheter. As a result, the postintervention CAUTI rate may represent a biased underestimate. The role played by this misclassification bias in studying the effectiveness of quality improvement interventions to prevent nosocomial UTIs has not been previously evaluated. We investigated the effectiveness of a novel, low-resource, educational intervention designed to reduce urinary catheter use and nosocomial UTIs. We developed an educational intervention targeting ward physicians and nurses that focused on accepted urinary catheter indications and alternatives, ward data on the actual burden of nosocomial UTIs, and establishing a safety culture based on avoiding unnecessary catheter use on an internal medicine clinical teaching unit. We performed a prospective controlled trial to evaluate our hypotheses that our novel brief educational intervention leads to a reduction in urinary catheter use and CAUTI rate. We further explored whether the CAUTI rate might be susceptible to misclassification bias. MATERIALS AND METHODS Study design and setting We conducted a prospective, controlled, intervention trial on 1 internal medicine (46 beds) and 1 neurosciences ward (35 beds) at the Montreal General Hospital (Montreal, Canada), a 417-bed tertiary care hospital affiliated with McGill University. The study was conducted from December 5, 2011-August 24, 2012, encompassing 2 months of preintervention, 2 months of intervention, and 5 months of postintervention follow-up. Subjects Our cohort included all patients consecutively admitted to the internal medicine and neurosciences wards at the Montreal General Hospital during the study period. Subjects were excluded if they had a permanent urinary catheter, had a urinary catheter inserted by the urology service during admission, or were admitted to the long-term care service on the internal medicine ward. Subjects were followed until death, hospital discharge, transfer to another service, or study conclusion. This study was granted a waiver of exemption by the McGill University Ethics Review Board. Intervention The study began with 2 months of preintervention data collection on both wards, during which time all clinical personnel were blinded to study objectives. Subsequently, the medical and nursing personnel on the internal medicine ward each received an educational intervention. In contrast, the medical and nursing personnel on the control neurosciences ward received no educational intervention.
349
On the intervention ward, the medical and nursing personnel received separate education sessions. The medical personnel (attending physicians, residents, medical students) received, on the first day of each 1-month clinical rotation, a 15-minute education session given by a project leader. The presentation included information on epidemiology and morbidity of nosocomial UTIs; accepted indications for, and alternatives to, urinary catheter use with emphasis on reducing inappropriate catheter use; and urinary catheter use and UTI incidence data for their ward from the preintervention phase. The nursing personnel received, during an education day prior to the postintervention phase, a single 45-minute education session given by the same project leader. The presentation included all the elements of the medical personnel session, with additional emphasis on the use of condom catheters, straight in-out catheterization, enhanced mobilization, and good bowel hygiene as a means to avoid urinary catheter use. Discussion was held on anticipated nursing barriers to implementation of a catheter reduction strategy, and solutions were identified in a multidisciplinary fashion. Both medical and nursing interventions included strong messages of a ward-wide commitment to a safety-based culture of reduced urinary catheter usage, but neither focused on techniques surrounding insertion or maintenance of these devices. All sessions were attended by representatives of ward medical and nursing leadership, who emphasized their support for this intervention and made themselves available to all members of the staff during the intervention period. During the postintervention period, control ward personnel remained blinded to study objectives. Postimplementation blinding of personnel on the intervention ward was precluded by the nature of the intervention; however, study personnel made no attempts to influence clinical practice on that ward. Outcomes The coprimary outcomes were the device utilization rate (DUR) (catheter days/100 patient days) and CAUTI rate (number of CAUTIs/1,000 patient days). Secondary outcomes included rates per 1,000 patient days of noncatheter-associated UTIs, all urinary infections, positive urine cultures, and urine sampling (number of urine cultures/ 1,000 patient days). The proportion of patients who ever had a urinary catheter and duration of catheterization were also recorded. As balancing patient safety measures we tracked the catheter reinsertion rate (catheter reinsertions per 1,000 patient days) and incidence of urogenital trauma during reinsertion, with the latter requiring immediate reporting to the principal investigator. A catheter day was defined as the presence of a urinary catheter at 10:00 AM each day, and a positive urine culture was defined as the growth of >105 colony forming units/mL of a single uropathogenic species. A UTI was defined as the presence of a positive culture plus at least one of the following: fever >37.9 C, peripheral white blood cell count >10.5 109/L, urinary tract symptoms, mental status change, or new onset hypotension, all in the absence of another evident clinical explanation. A UTI was classified as catheter associated if the culture was sent with a catheter in place or within 7 days of catheter removal. Catheter reinsertion was defined as reinsertion of a previously removed urinary catheter within 7 days of initial removal. Data were collected Monday-Friday by trained study medical personnel using a specifically designed data extraction sheet; weekend data were abstracted the following Monday.
350
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
Table 1 Cohort demographic and clinical history by ward and intervention period Control ward Total (n ¼ 1,335)
Preintervention (n ¼ 126)
Postintervention (n ¼ 348)
Median age (IQR) Median LOS (IQR) (d) Male sex (%) Diabetes (%) Admission ABX (%) Admission UTI (%)
69 (52-80) 9 (4-17) 61.9 28.6 34.9 10.3
67 (49-81) 8 (4-17) 62.4 20.4 34.2 12.9
Intervention ward P value* .76 .64 .93 .06 .88 .44
Preintervention (n ¼ 239)
Postintervention (n ¼ 622)
69 (57-80) 7 (4-11) 56.5 29.3 66.1 9.2
70 (56-81) 6 (3-11) 52.3 33.4 59.9 8.7
P value* .74 .02 .28 .24 .10 .81
ABX, antibiotics; IQR, interquartile range; LOS, length of stay; UTI, urinary tract infection. *P values from Pearson c2 test for categorical variables and Wilcoxon rank-sum test for continuous variables.
Demographic data (age and sex), comorbidity (diabetes status), and baseline clinical data (UTI or antibiotic history in hospital prior to current ward admission) were collected from the ward admission note and laboratory information system. Urinary catheter presence and reinsertion status were determined from the nursing records, with bedside confirmation in the case of discrepant or unclear recording. Information regarding urinary culture sampling, positive culture results, and leukocytosis were determined from the computerized laboratory and microbiology system. The complete medical record of all patients with positive urine cultures was reviewed to determine classification as a UTI. In the case of uncertainty, the data collector discussed the outcome with a senior internal medicine physician (D.J.) or infectious disease physician (C.F.) to obtain consensus.
Table 2 Preintervention outcome rates and 95% CI by ward Control ward Outcome DUR (per 100 patient days) CAUTI rate N-CAUTI rate NUTI rate Positive culture rate
Intervention ward
Preintervention*
95% CI
Preintervention*
95% CI
17.0
15.1-19.1
15.7
14.1-17.4
5.4 3.0 8.4 14.4
2.5-10.2 0.9-7.0 4.6-14.1 9.2-21.5
3.4 0.8 4.3 7.7
1.5-6.7 0.1-3.1 2.1-7.9 4.6-12.2
CAUTI, catheter-associated urinary tract infection; CI, confidence interval; DUR, device utilization rate; N-CAUTI, noncatheter-associated urinary tract infection; NUTI, nosocomial urinary tract infection. *Preintervention rate per 1,000 patient days unless otherwise indicated.
RESULTS Statistical analysis
Subjects
Statistical analysis was performed using Stata 12 (StataCorp, College Station, TX), with a prespecified significance level of P < .05. Pre- and postintervention subject demographic and clinical data were compared with the Wilcoxon rank-sum test for continuous variables, and the Pearson c2 test was used for categorical data. Primary catheter and UTI outcome measures were analyzed by determining the incidence rate ratio (IRR) comparing the postintervention rate with the preintervention rate for each individual ward. The crude DUR IRR was determined by negative binomial regression after a Poisson goodness-of-fit test, data exploration, and likelihood-ratio test of the a coefficient indicated significant overdispersion in the DUR data. An adjusted IRR was determined by fitting a multivariate negative binomial regression model with the prespecified covariates of sex and history of UTI on admission. Crude and adjusted infectious outcome measure IRRs were determined by Poisson regression models fitted with and without the prespecified covariates of age, sex, diabetes history, and history of UTI or antibiotic use prior to ward admission. Model adequacy was tested with the use of the Poisson goodness-of-fit test. The IRRs for urine culture sampling and catheter reinsertion were determined in an unadjusted manner with the use of negative binomial regression because of overdispersion of data. Proportions of positive cultures and patients with urinary catheters in each time period were compared by Pearson c2 analysis. We performed 2 post hoc sensitivity analyses of the DUR outcome by excluding subjects with catheter duration >14 and >7 days to ensure that findings were not influenced by the presence of prolonged catheterization and conducted 1 post hoc subgroup analysis evaluating the effect of subject sex on the intervention to reduce the DUR.
We evaluated 1,364 patients who were consecutively admitted to the 2 study wards during the 2-month preintervention and 5-month postintervention periods. Of these patients, 26 (1.9%) were excluded from analysis because of a permanent (n ¼ 19) or urology-inserted (n ¼ 7) urinary catheter. A further 3 patients (0.2%) had missing data. Given that this data was the result of clerical error, we classified the data as missing completely at random, and given the low percentage, we excluded these 3 patients. Therefore, 1,335 patients, representing 13,753 patient days, were included in our final analysis. Within ward, demographic and clinical factors were evenly distributed when comparing the pre- and postintervention periods, apart from a 1-day shorter median length of stay on the intervention ward (Table 1).
Outcomes The preintervention period rates of all catheter and infectious outcome measures for both wards are listed in Table 2. Urinary catheter utilization After the intervention there was a highly significant reduction in the DUR among intervention ward patients, with an unadjusted IRR of 0.52 (95% confidence interval [CI], 0.34-0.81; P ¼ .003), whereas there was no significant change in catheter use among control ward patients, with an unadjusted IRR of 1.1 (95% CI, 0.62-1.97; P ¼ .74). (Table 3). On adjusting for the potential confounders of patient sex and admission history of UTI, the findings on intervention effectiveness were unchanged. The proportion of patients with a urinary catheter in the intervention period dropped significantly on the intervention ward (26.78% to 16.4%, P ¼ .0005), with no change on
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
351
Table 3 Crude and adjusted device utilization rate outcome IRRs by ward and intervention period Crude (unadjusted) Variable Intervention ward Postintervention Preintervention Control ward Postintervention Preintervention Sex Female Male UTI at admission Yes No
IRR
95% CI
0.52 1.0 (reference)
0.34-0.81 NA
1.1 1.0 (reference)
Adjusted* P value
IRR
95% CI
P value
.003 NA
0.49 1.0 (reference)
0.32-0.76 NA
.001 NA
0.62-1.97 NA
.74 NA
1.02 1.0 (reference)
0.58-1.82 NA
.93 NA
0.83 1.0 (reference)
0.61-1.15 NA
.27 NA
0.78 1.0 (reference)
0.56-1.07 NA
.12 NA
1.76 1.0 (reference)
1.06-2.91 NA
.028 NA
1.94 1.0 (reference)
1.17-3.24 NA
.011 NA
CI, confidence interval; IRR, incidence rate ratio; NA, not applicable; UTI, urinary tract infection. *Model adjusted for sex and history of UTI on admission.
the control ward (30.17% to 28.17%, P ¼ .67). There was no significant change in median duration of catheterization postintervention on either ward (data not shown). A history of UTI prior to ward admission was associated with a significantly increased DUR compared with patients with no UTI on admission (IRR, 1.94; 95% CI, 1.17-3.24; P ¼ .011). There was no significant association between patient sex and DUR in either univariate or multivariable regression (adjusted IRR, 0.78; 95% CI, 0.56-1.07; P ¼ .12), and there was no evidence for sex effect modification (P ¼ .42 for interaction term). Sensitivity analyses of the primary outcome, after excluding subjects with catheter duration of >14 and >7 days, respectively, did not show differences in point estimates or inference for intervention effectiveness (data not shown). UTI outcomes Among intervention ward patients, the postintervention period was associated with a reduced unadjusted CAUTI rate (IRR, 0.39; 95% CI, 0.15-1.02; P ¼ .054) and a reduced age, sex, diabetes and history of UTI or antibiotic use adjusted CAUTI rate (IRR, 0.42; 95% CI, 0.16-1.08; P ¼ .072), both of which however failed to reach statistical significance (Table 4). Among control patients, the intervention period was associated with a nonsignificant reduction in the adjusted CAUTI rate (IRR, 0.51; 95% CI, 0.22-1.20; P ¼ .12). The intervention period was associated with nonsignificant increases in the adjusted incidence rates of noncatheter-associated UTIs in both groups of patients; the increase was 2.40-fold (95% CI, 0.53-10.7; P ¼ .25) among intervention patients and 1.57-fold (95% CI, 0.59-4.21; P ¼ .37) among control patients. There were no significant changes in the rate of all nosocomial UTIs comparing the postintervention to the preintervention period on either the intervention ward (IRR, 0.79; 95% CI, 0.38-1.65; P ¼ .53), or the control ward (IRR, 0.89; 95% CI, 0.48-1.67; P ¼ .72). Similarly, there were no within-ward differences in the rate of positive urine cultures (intervention ward: IRR, 0.82; 95% CI, 0.471.46; P ¼ .50; control ward: IRR, 0.89; 95% CI, 0.55-1.43; P ¼ .63). Urine culture sampling Compared with the preintervention period, the rate of urine culture sampling on the control ward decreased significantly, by 25% (P ¼ .04), in the postintervention period. In contrast, on the intervention ward, there was no significant change in rate of urine culture sampling. There was no significant change in the proportion of urine samples returning as positive cultures between the pre- and postintervention periods on either ward (Table 5).
Table 4 Crude and adjusted infectious outcome IRRs by ward and intervention period Unadjusted Outcome CAUTI Control Intervention N-CAUTI Control Intervention NUTI Control Intervention Positive cultures Control Intervention
Adjusted*
IRRy
95% CI
P value
IRRy
95% CI
P value
0.54 0.39
0.23-1.26 0.15-1.02
.15 .054
0.51 0.42
0.22-1.20 0.16-1.08
.12 .072
1.49 2.65
0.56-3.97 0.59-11.8
.42 .20
1.57 2.40
0.59-4.21 0.53-10.7
.37 .25
0.88 0.80
0.47-1.64 0.39-1.68
.69 .56
0.89 0.79
0.48-1.67 0.38-1.65
.72 .53
0.89 0.86
0.55-1.43 0.49-1.52
.62 .60
0.89 0.82
0.55-1.43 0.47-1.46
.63 .50
NOTE. Poisson goodness-of-fit for all models, P ¼ 1.00. CAUTI, catheter-associated urinary tract infection; CI, confidence interval; DUR, device utilization rate; IRR, incidence rate ratio; N-CAUTI, noncatheter-associated urinary tract infection; NUTI, nosocomial urinary tract infection. *Adjusted for age, sex, diabetes status, history of UTI on admission, and history of antibiotics on admission. y Reference group for all IRRs is the preintervention period for each ward.
Safety outcomes There were no postintervention differences in the rate of catheter reinsertion on either ward (intervention ward: IRR, 0.97; 95% CI, 0.44-2.30; P ¼ .92; control ward: IRR, 0.79; 95% CI, 0.32-2.12; P ¼ .58). There were no cases of reinsertion-related urogenital trauma reported during the study (Table 5).
DISCUSSION Reduction in catheter use Our low-resource intervention led to a 50% reduction in the DUR. This rate of reduction exceeded that achieved in 2 other investigations of urinary catheter interventions. Fakih et al11 achieved a 20% reduction in catheter use by incorporating a dedicated urinary catheter nurse onto daily ward multidisciplinary rounds. Knoll et al7 also achieved a 20% reduction in the DUR, with the use of a dedicated nurse to prompt catheter removal, and a 10% reduction in the DUR during a phase of intense education and audit and feedback. In both phases of Knoll’s study, however, device use returned to baseline rates after the intense intervention was stopped. In contrast with the present and aforementioned studies, a greater reduction in the DUR (81%) on a medical unit was reported by Topal
352
B.C. Schwartz et al. / American Journal of Infection Control 43 (2015) 348-53
Table 5 Urine culture and catheter reinsertion outcomes by ward and intervention period Control ward Variable Urine sampling rate* % positive cultures Catheter reinsertion rate*
Intervention ward
Preintervention
Postintervention
P value
Preintervention
Postintervention
P value
45.7 31.6 4.8
34.3 37.3 3.8
.04 .40y .58
44.5 17.3 4.3
39.1 16.4 4.2
.29 .84y .92
*Rates per 1,000 patient days, and P values for incidence utilization rate. Pearson c2 test.
y
et al,9 with the use of a computerized order entry-based reminder system. Therefore, our reported DUR reduction from a lowerresource intervention compares favorably with previous reports of resource intense interventions. That the proportion of patients ever with a catheter dropped significantly postintervention, with no change in median catheter duration, suggests that the main effect of our intervention was in prompting removal on ward admission of unnecessary catheters, a key component of our educational intervention, which highlighted admission as a cardinal time to make providers aware their patient had a catheter and to evaluate continued need. Interestingly, we did find that a history of UTI on admission was strongly associated with an increase in the DUR, which may reflect the longstanding notion that these infections must be completely treated prior to catheter removal. This finding may represent a target for further education about appropriate catheter utilization. Reduction in CAUTI Our educational intervention was associated with a reduced rate, albeit not statistically significant, of CAUTI in the postintervention period. Given the relative rarity of the outcome, it is possible that we did not accrue enough person time to reach statistical significance during the time of our study. Although also failing to reach significance, we did unexpectedly find a marginal reduction in the CAUTI rate on the control ward during the postintervention period, despite no change in catheter use. Although this may equally represent chance, this has not previously been evaluated or reported. This observation highlights the importance of control arms in evaluations of quality improvement interventions in comparison with the commonly used before and after design. Such control arm designs permit a more robust assessment of the true improvement effect of the studied intervention and allows for the detection of secular trends and other confounders. Possible surveillance bias Our findings suggest that there may have been surveillance bias. On our control ward, which received no educational intervention, there was a significantly reduced rate of urine culture sampling during the postintervention period compared with preintervention, despite a similar proportion of positive cultures. The reduced rate of sampling could be attributed to a reduction in unnecessary cultures or a true decrease in the incidence of UTIs on the control floor. Although we masked the nature of our study on the control floor and no unit personnel crossed over between units, personnel may have become aware of our study objectives through spillover. The nonstandardized approach to urine sampling, seen in many health care settings, implies that the UTI incidence rate may be closely related to the frequency with which they are sought. Recently, Leis et al18 demonstrated that inpatient urine cultures are associated with a confirmed infection in
