Todd C. Lee, MD, MPH

Urinary catheter use can be associated with urinary tract infections, delirium, trauma, and immobility.[1] Evidence‐based strategies to reduce inappropriate use are available[2]; however, their application across centers is variable.[3] We aimed to characterize the prevalence and indication for catheters among Canadian teaching hospitals with and without catheter reduction programs.

METHODS

Twelve of 17 postgraduate internal medicine training program directors agreed to participate, and 9 Canadian teaching hospitals enrolled in this prevalence study of urinary catheter use among medical inpatients. Data collection used a standardized form and took place over 5 consecutive weekdays during August 2015. Each site anonymously collected the total number of catheters, total number of inpatient‐days, and indications for use from either the bedside nurse or physician. Appropriate clinical indications were based on the 2009 guidelines from the Healthcare Infection Control Practice Advisory Committee.[4] Potentially inappropriate indications included urine output measurement in noncritically ill patients, and other or unknown indications.[4, 5] A catheter reduction program was defined as the presence of a structured system to monitor and reduce use via: nurse‐directed catheter removal, audit‐feedback of use to providers, physician reminders, and/or automatic stop orders.

The primary outcome was the number of catheter days per 100 inpatient‐days. We used generalized estimating equations to adjust the 95% confidence interval (CI) and P value to account for hospital‐level clustering of the responses. The P values are from a 2‐tailed Wald test against the true log scale parameter being equal to zero. The analysis was performed using R version 3.0.2 using the geepack package (Free Software Foundation, Boston, MA).

The McGill University Health Centre Research Ethics Board approved this study with concomitant authorization at participating sites.

RESULTS

The characteristics of participating hospitals are displayed in Table 1. Those with active catheter reduction programs reported established systems for monitoring catheter placement, duration, and catheter‐associated urinary tract infections. More than half of the hospitals lacked a catheter reduction program. Overall, catheters were present on 13.6% of patient‐days (range, 2.3%32.4%). Centers without reduction programs reported higher rates of catheter use both overall and for potentially inappropriate indications. After adjustment for clustering, those with a formal intervention had 8.8 fewer catheter days per 100 patient‐days as compared to those without (9.8 [95% CI: 6.0‐15.6] vs 18.6 [95% CI: 13.0‐26.1], P = 0.03). This meant that the odds of a urinary catheter being present were 2 times (95% CI: 1.0‐3.4) greater in hospitals without reduction programs. Differences in appropriate catheter use did not reach statistical significance.

DISCUSSION

Despite the availability of consensus guidelines for appropriate use and the efforts of movements like Choosing Wisely, many Canadian teaching hospitals have not yet established a urinary catheter reduction program for medical inpatients. Our findings are similar to 2 non‐Canadian studies, which demonstrated that fewer than half of hospitals had implemented control measures.[4, 6] In contrast to those other studies, our study demonstrated that hospitals that employed control measures had reduced rates of catheter use suggesting that systematic, structured efforts are necessary to improve practice.[7, 8]

Ours is the first nation‐wide study in Canada to report urinary catheter rates and the effect of associated reduction programs. Data from the National Healthcare Safety Network suggest our Canadian estimates of urinary catheter rates in medical inpatients are similar to those of the United States (13.6 vs 14.8 catheter days per 100 inpatient‐days, respectively, for general medical inpatients).[9, 10]

Several limitations of this study warrant discussion. First, we sampled only academic institutions at 1 time point, which may not represent annualized rates or rates in community hospitals. However, our findings are similar to those reported in previous studies.[10] Second, our method of consecutive daily audits may have caused individuals to change their behavior knowing that they were being observed, resulting in lower catheter utilization than would have been otherwise present and biasing our estimates of catheter overuse downward. Third, we collected point prevalence data, limiting our ability to make inferences on causality. The key factor(s) contributing to observed differences between hospitals remains unknown. However, pre‐post intervention data available for 3 hospitals suggest that improvements followed active catheter reduction efforts.[7, 8] Fourth, we were unable to obtain outcome data such as catheter‐associated urinary tract infection, delirium, or fall rates. However, catheter reduction is widely recognized as an important first step to reducing preventable harm for hospital patients.

We suggest that the broader uptake of structured models of care that promote early discontinuation of urinary catheters on medical wards is needed to improve their appropriateness. Fortunately, it appears as though a variety of models are effective. Therefore, when it comes to adopting Choosing Wisely's less is more philosophy toward urinary catheter utilization, we suggest that less time be allowed to pass before more proven and structured interventions are universally implemented.

Urinary catheter use can be associated with urinary tract infections, delirium, trauma, and immobility.[1] Evidence‐based strategies to reduce inappropriate use are available[2]; however, their application across centers is variable.[3] We aimed to characterize the prevalence and indication for catheters among Canadian teaching hospitals with and without catheter reduction programs.

METHODS

Twelve of 17 postgraduate internal medicine training program directors agreed to participate, and 9 Canadian teaching hospitals enrolled in this prevalence study of urinary catheter use among medical inpatients. Data collection used a standardized form and took place over 5 consecutive weekdays during August 2015. Each site anonymously collected the total number of catheters, total number of inpatient‐days, and indications for use from either the bedside nurse or physician. Appropriate clinical indications were based on the 2009 guidelines from the Healthcare Infection Control Practice Advisory Committee.[4] Potentially inappropriate indications included urine output measurement in noncritically ill patients, and other or unknown indications.[4, 5] A catheter reduction program was defined as the presence of a structured system to monitor and reduce use via: nurse‐directed catheter removal, audit‐feedback of use to providers, physician reminders, and/or automatic stop orders.

The primary outcome was the number of catheter days per 100 inpatient‐days. We used generalized estimating equations to adjust the 95% confidence interval (CI) and P value to account for hospital‐level clustering of the responses. The P values are from a 2‐tailed Wald test against the true log scale parameter being equal to zero. The analysis was performed using R version 3.0.2 using the geepack package (Free Software Foundation, Boston, MA).

The McGill University Health Centre Research Ethics Board approved this study with concomitant authorization at participating sites.

RESULTS

The characteristics of participating hospitals are displayed in Table 1. Those with active catheter reduction programs reported established systems for monitoring catheter placement, duration, and catheter‐associated urinary tract infections. More than half of the hospitals lacked a catheter reduction program. Overall, catheters were present on 13.6% of patient‐days (range, 2.3%32.4%). Centers without reduction programs reported higher rates of catheter use both overall and for potentially inappropriate indications. After adjustment for clustering, those with a formal intervention had 8.8 fewer catheter days per 100 patient‐days as compared to those without (9.8 [95% CI: 6.0‐15.6] vs 18.6 [95% CI: 13.0‐26.1], P = 0.03). This meant that the odds of a urinary catheter being present were 2 times (95% CI: 1.0‐3.4) greater in hospitals without reduction programs. Differences in appropriate catheter use did not reach statistical significance.

DISCUSSION

Despite the availability of consensus guidelines for appropriate use and the efforts of movements like Choosing Wisely, many Canadian teaching hospitals have not yet established a urinary catheter reduction program for medical inpatients. Our findings are similar to 2 non‐Canadian studies, which demonstrated that fewer than half of hospitals had implemented control measures.[4, 6] In contrast to those other studies, our study demonstrated that hospitals that employed control measures had reduced rates of catheter use suggesting that systematic, structured efforts are necessary to improve practice.[7, 8]

Ours is the first nation‐wide study in Canada to report urinary catheter rates and the effect of associated reduction programs. Data from the National Healthcare Safety Network suggest our Canadian estimates of urinary catheter rates in medical inpatients are similar to those of the United States (13.6 vs 14.8 catheter days per 100 inpatient‐days, respectively, for general medical inpatients).[9, 10]

Several limitations of this study warrant discussion. First, we sampled only academic institutions at 1 time point, which may not represent annualized rates or rates in community hospitals. However, our findings are similar to those reported in previous studies.[10] Second, our method of consecutive daily audits may have caused individuals to change their behavior knowing that they were being observed, resulting in lower catheter utilization than would have been otherwise present and biasing our estimates of catheter overuse downward. Third, we collected point prevalence data, limiting our ability to make inferences on causality. The key factor(s) contributing to observed differences between hospitals remains unknown. However, pre‐post intervention data available for 3 hospitals suggest that improvements followed active catheter reduction efforts.[7, 8] Fourth, we were unable to obtain outcome data such as catheter‐associated urinary tract infection, delirium, or fall rates. However, catheter reduction is widely recognized as an important first step to reducing preventable harm for hospital patients.

We suggest that the broader uptake of structured models of care that promote early discontinuation of urinary catheters on medical wards is needed to improve their appropriateness. Fortunately, it appears as though a variety of models are effective. Therefore, when it comes to adopting Choosing Wisely's less is more philosophy toward urinary catheter utilization, we suggest that less time be allowed to pass before more proven and structured interventions are universally implemented.

Urinary catheter use can be associated with urinary tract infections, delirium, trauma, and immobility.[1] Evidence‐based strategies to reduce inappropriate use are available[2]; however, their application across centers is variable.[3] We aimed to characterize the prevalence and indication for catheters among Canadian teaching hospitals with and without catheter reduction programs.

METHODS

Twelve of 17 postgraduate internal medicine training program directors agreed to participate, and 9 Canadian teaching hospitals enrolled in this prevalence study of urinary catheter use among medical inpatients. Data collection used a standardized form and took place over 5 consecutive weekdays during August 2015. Each site anonymously collected the total number of catheters, total number of inpatient‐days, and indications for use from either the bedside nurse or physician. Appropriate clinical indications were based on the 2009 guidelines from the Healthcare Infection Control Practice Advisory Committee.[4] Potentially inappropriate indications included urine output measurement in noncritically ill patients, and other or unknown indications.[4, 5] A catheter reduction program was defined as the presence of a structured system to monitor and reduce use via: nurse‐directed catheter removal, audit‐feedback of use to providers, physician reminders, and/or automatic stop orders.

The primary outcome was the number of catheter days per 100 inpatient‐days. We used generalized estimating equations to adjust the 95% confidence interval (CI) and P value to account for hospital‐level clustering of the responses. The P values are from a 2‐tailed Wald test against the true log scale parameter being equal to zero. The analysis was performed using R version 3.0.2 using the geepack package (Free Software Foundation, Boston, MA).

The McGill University Health Centre Research Ethics Board approved this study with concomitant authorization at participating sites.

RESULTS

The characteristics of participating hospitals are displayed in Table 1. Those with active catheter reduction programs reported established systems for monitoring catheter placement, duration, and catheter‐associated urinary tract infections. More than half of the hospitals lacked a catheter reduction program. Overall, catheters were present on 13.6% of patient‐days (range, 2.3%32.4%). Centers without reduction programs reported higher rates of catheter use both overall and for potentially inappropriate indications. After adjustment for clustering, those with a formal intervention had 8.8 fewer catheter days per 100 patient‐days as compared to those without (9.8 [95% CI: 6.0‐15.6] vs 18.6 [95% CI: 13.0‐26.1], P = 0.03). This meant that the odds of a urinary catheter being present were 2 times (95% CI: 1.0‐3.4) greater in hospitals without reduction programs. Differences in appropriate catheter use did not reach statistical significance.

DISCUSSION

Despite the availability of consensus guidelines for appropriate use and the efforts of movements like Choosing Wisely, many Canadian teaching hospitals have not yet established a urinary catheter reduction program for medical inpatients. Our findings are similar to 2 non‐Canadian studies, which demonstrated that fewer than half of hospitals had implemented control measures.[4, 6] In contrast to those other studies, our study demonstrated that hospitals that employed control measures had reduced rates of catheter use suggesting that systematic, structured efforts are necessary to improve practice.[7, 8]

Ours is the first nation‐wide study in Canada to report urinary catheter rates and the effect of associated reduction programs. Data from the National Healthcare Safety Network suggest our Canadian estimates of urinary catheter rates in medical inpatients are similar to those of the United States (13.6 vs 14.8 catheter days per 100 inpatient‐days, respectively, for general medical inpatients).[9, 10]

Several limitations of this study warrant discussion. First, we sampled only academic institutions at 1 time point, which may not represent annualized rates or rates in community hospitals. However, our findings are similar to those reported in previous studies.[10] Second, our method of consecutive daily audits may have caused individuals to change their behavior knowing that they were being observed, resulting in lower catheter utilization than would have been otherwise present and biasing our estimates of catheter overuse downward. Third, we collected point prevalence data, limiting our ability to make inferences on causality. The key factor(s) contributing to observed differences between hospitals remains unknown. However, pre‐post intervention data available for 3 hospitals suggest that improvements followed active catheter reduction efforts.[7, 8] Fourth, we were unable to obtain outcome data such as catheter‐associated urinary tract infection, delirium, or fall rates. However, catheter reduction is widely recognized as an important first step to reducing preventable harm for hospital patients.

We suggest that the broader uptake of structured models of care that promote early discontinuation of urinary catheters on medical wards is needed to improve their appropriateness. Fortunately, it appears as though a variety of models are effective. Therefore, when it comes to adopting Choosing Wisely's less is more philosophy toward urinary catheter utilization, we suggest that less time be allowed to pass before more proven and structured interventions are universally implemented.

In 2013, there were more than 15 million Americans receiving proton pump inhibitors (PPIs),[1] with an associated drug cost of nearly $79 billion between 2007 and 2011.² PPI use is reaching epidemic proportions, likely due to the medicalization of gastrointestinal symptoms coupled with pervasive marketing and academic detailing being performed by the pharmaceutical industry.

Although PPIs are generally considered safe, they are not as innocuous as many physicians believe. In 2011 and 2012, the US Food and Drug Administration and Health Canada, respectively, issued safety advisories regarding the use of these medications related to Clostridium difficile, fracture risk, and electrolyte derangement.[3, 4, 5, 6] There have also been numerous other harmful associations reported,[7, 8, 9, 10] suggesting it would be prudent to follow Health Canada's advice that: PPIs should be prescribed at the lowest dose and shortest duration of therapy appropriate to the condition being treated.[4] In many cases this implies stopping the PPI after an appropriate duration of therapy or attempting nonpharmacological or H2‐blocker therapy instead.

Nevertheless, despite numerous cautionary publications, PPI use for nonevidence‐based indications remains common. Because they are generally thought of as outpatient medications, PPIs are frequently continued in hospitalized patients, and inappropriate outpatient therapy is rarely addressed.[11, 12, 13] Likewise, inappropriate de novo use can also be observed during hospitalization and may continue on discharge.[13, 14, 15] Hospitalization may consequently present an opportunity to employ meaningful interventions targeting outpatient medication use.[16] We developed an opportune inpatient intervention targeting inappropriate PPI therapy.

Our study had 2 aims: first, to determine the magnitude of the problem in a contemporary Canadian medical inpatient population, and second, we sought to leverage the inpatient admission as an opportunity to promote change when the patient returned to the community through the application of an educational and web‐based quality‐improvement (QI) intervention.

METHODS

RESULTS

During the preintervention period, 464 patients were admitted, of whom 209 (45%) were taking a PPI prior to admission. During their hospitalization, an additional 53 patients (21% of nonusers) were newly prescribed a PPI that was continued at discharge. During the intervention period, a total of 640 patients were admitted, of whom 281 (44%) were taking a PPI prior to admission. During their hospitalization, 60 patients (17% of nonusers) were newly prescribed a PPI that was continued at discharge. Neither the monthly proportions admitted on PPIs from prior to admission (level P=0.59, slope P=0.46) or those newly initiated on a PPI (level P=0.36, slope P=0.18) were significantly different before compared to after the intervention. However, there was both a clinically and statistically significant difference in the proportion of preadmission PPIs that were discontinued at hospital discharge from a monthly mean of 7.7% (or 16/209) before intervention to 18.5% (or 52/281) afterward (Figure 1; level P=0.03, slope P=0.48).

Figure 1

During the intervention period, our teams prospectively captured PPI indications and patient comorbidities for 54% (152/281) of the patients admitted on a PPI using the online assessment tool. The baseline characteristics of the population in whom the online tool was applied are shown in Table 2. These patients had a mean age of 69.6 years, and 49% were male. Thirty‐two percent had diabetes, 20% had chronic renal insufficiency, and 13% had experienced a gastrointestinal hemorrhage within the 3 months prior to admission. It was frequent for PPI users to receive systemic antibiotics (44%) or to have diagnoses potentially associated with PPI use such as community‐acquired pneumonia (25%) or C difficile (11%).

Fifty‐four percent (82/152) of patients in whom the online tool was applied had an evidence‐based indication (Table 3). The most common indication for PPI prescription was the receipt of antiplatelet/anticoagulant or nonsteroidal anti‐inflammatory drug with 2 other known risk factors for upper gastrointestinal bleeding (20%). In the remaining 46% (70/152) of patients, the prescription of a PPI was deemed nonevidence based. Of these, 34 (49%) had their PPI discontinued. When patients were approached to discontinue therapy, the rate of success was high, with only 2 refusals.

DISCUSSION

In this prospective intervention, 44% of patients admitted to an acute‐care medical ward were prescribed a PPI prior to their admission. In the subgroup of patients for whom the indication for PPI use was recorded through our online tool, less than half had an evidence‐based indication for ongoing therapy. Our intervention was successful in increasing the proportion of patients in whom preadmission PPI prescriptions were stopped at discharge from an average of 7.7% in the preintervention phase to 18.5% during the intervention. This intervention is novel in that we were able to reduce active community prescriptions for PPIs in patients without obvious indication by nearly 50%.

Our population's rates of PPI prescription were consistent with previous reports,[11, 12, 13, 15, 25, 26, 27, 28] and it is clear that many hospitalized patients continue their PPIs at discharge without clear indications. We propose that hospitalization can serve as an opportunity to reassess the necessity of continuous PPI use. Previous systematic attempts to reduce inappropriate PPI prescriptions in hospital have met with varied success. Several of these studies were unable to achieve a demonstrable effect.[23, 29, 30, 31] In contrast, Hamzat et al.[30] described a successful educational intervention targeting inpatients on a geriatric ward. A 4‐week educational strategy was employed, and they were able to discontinue PPIs in 10 of 60 (17%) patients without indication during a limited period of study. Another successful intervention by Gupta et al.[32] involved a before‐and‐after study combining a half‐hour physician education session with the introduction of a medication reconciliation tool. They showed a decrease in inappropriate discharge prescriptions of 50%. Not only did our study demonstrate an equally sizable reduction in inappropriate discharge prescriptions, but we also employed a more methodologically sound time‐series analysis to control for unmeasured contemporaneous factors such as rotating staff practices or monthly differences in patient composition. We demonstrated an immediate and sustained improvement in performance that lasted over 6 months. Furthermore, in contrast to other interventions, which addressed inappropriate inpatient use persisting on discharge, our intervention also addressed the appropriateness of PPI use that antedated hospitalization.

There are common themes to the successful programs. The more time spent educating and reminding the prescribing physicians, the more successful the intervention. Nonetheless, we believe our intervention is not onerous or overly time consuming. We performed a short presentation each month to educate rotating physicians, and the tool took less than 1 minute to complete once the information on PPI indication was available. Frequent education sessions may be initially necessary given the comfort that many physicians have developed in prescribing PPIs. A further prerequisite for success may be a familiarity with PPI indications and potential adverse effects. Without this, the intervention may not show a demonstrable effect, as was seen in a study of pulmonologists.[23] We hypothesize that some subspecialist physicians may not have the same appreciation of the adverse effects of PPIs nor the confidence to stop them when not indicated, as compared to general internists or hospitalists.

The proportion of patients with newly initiated PPIs at discharge decreased after the intervention, but this did not reach statistical significance. Our study's power may have precluded this. However, we had also previously put in place unpublished interventions to diminish inappropriate gastric prophylaxis in the hospital, which may have diminished the effect of this intervention.

Unfortunately, although we demonstrated a clinically significant effect on PPI exit prescription rates, we still found that nearly half of the patients who were evaluated using the online tool were discharged on PPIs despite our physicians' acknowledgment that they had no identifiable indication. It is possible that clinicians do not feel comfortable stopping these medications, owing to a fear that there is an indication that they are not aware of. In certain cases, it is possible that a reappraisal of the benefits, risks, and costs might reassure the clinician that they could safely stop the drug; however, therapeutic inertia is often hard to overcome.

CONCLUSION

We have demonstrated that in medical inpatients, both PPI use and misuse remain common; however, with a combined educational and Web‐based QI intervention, we could successfully decrease inappropriate exit prescriptions. Hospitalization, particularly at academic centers, should serve as an important point of contact for residents in training and expert faculty physicians to reconsider and rationalize patient medications. We should take the opportunity to engender a culture of responsibility for all of the medications that we represcribe at discharge, including an appraisal of the relevant harms and benefits, particularly when a medication is potentially unnecessary. We ought to then communicate the rationale for any changes to our community partners to maintain continuity of care. In this way, hospitalists can help treat the prescription indigestion that has become a common affliction in modern medicine.

In 2013, there were more than 15 million Americans receiving proton pump inhibitors (PPIs),[1] with an associated drug cost of nearly $79 billion between 2007 and 2011.² PPI use is reaching epidemic proportions, likely due to the medicalization of gastrointestinal symptoms coupled with pervasive marketing and academic detailing being performed by the pharmaceutical industry.

Although PPIs are generally considered safe, they are not as innocuous as many physicians believe. In 2011 and 2012, the US Food and Drug Administration and Health Canada, respectively, issued safety advisories regarding the use of these medications related to Clostridium difficile, fracture risk, and electrolyte derangement.[3, 4, 5, 6] There have also been numerous other harmful associations reported,[7, 8, 9, 10] suggesting it would be prudent to follow Health Canada's advice that: PPIs should be prescribed at the lowest dose and shortest duration of therapy appropriate to the condition being treated.[4] In many cases this implies stopping the PPI after an appropriate duration of therapy or attempting nonpharmacological or H2‐blocker therapy instead.

Nevertheless, despite numerous cautionary publications, PPI use for nonevidence‐based indications remains common. Because they are generally thought of as outpatient medications, PPIs are frequently continued in hospitalized patients, and inappropriate outpatient therapy is rarely addressed.[11, 12, 13] Likewise, inappropriate de novo use can also be observed during hospitalization and may continue on discharge.[13, 14, 15] Hospitalization may consequently present an opportunity to employ meaningful interventions targeting outpatient medication use.[16] We developed an opportune inpatient intervention targeting inappropriate PPI therapy.

Our study had 2 aims: first, to determine the magnitude of the problem in a contemporary Canadian medical inpatient population, and second, we sought to leverage the inpatient admission as an opportunity to promote change when the patient returned to the community through the application of an educational and web‐based quality‐improvement (QI) intervention.

METHODS

RESULTS

During the preintervention period, 464 patients were admitted, of whom 209 (45%) were taking a PPI prior to admission. During their hospitalization, an additional 53 patients (21% of nonusers) were newly prescribed a PPI that was continued at discharge. During the intervention period, a total of 640 patients were admitted, of whom 281 (44%) were taking a PPI prior to admission. During their hospitalization, 60 patients (17% of nonusers) were newly prescribed a PPI that was continued at discharge. Neither the monthly proportions admitted on PPIs from prior to admission (level P=0.59, slope P=0.46) or those newly initiated on a PPI (level P=0.36, slope P=0.18) were significantly different before compared to after the intervention. However, there was both a clinically and statistically significant difference in the proportion of preadmission PPIs that were discontinued at hospital discharge from a monthly mean of 7.7% (or 16/209) before intervention to 18.5% (or 52/281) afterward (Figure 1; level P=0.03, slope P=0.48).

Figure 1

During the intervention period, our teams prospectively captured PPI indications and patient comorbidities for 54% (152/281) of the patients admitted on a PPI using the online assessment tool. The baseline characteristics of the population in whom the online tool was applied are shown in Table 2. These patients had a mean age of 69.6 years, and 49% were male. Thirty‐two percent had diabetes, 20% had chronic renal insufficiency, and 13% had experienced a gastrointestinal hemorrhage within the 3 months prior to admission. It was frequent for PPI users to receive systemic antibiotics (44%) or to have diagnoses potentially associated with PPI use such as community‐acquired pneumonia (25%) or C difficile (11%).

Fifty‐four percent (82/152) of patients in whom the online tool was applied had an evidence‐based indication (Table 3). The most common indication for PPI prescription was the receipt of antiplatelet/anticoagulant or nonsteroidal anti‐inflammatory drug with 2 other known risk factors for upper gastrointestinal bleeding (20%). In the remaining 46% (70/152) of patients, the prescription of a PPI was deemed nonevidence based. Of these, 34 (49%) had their PPI discontinued. When patients were approached to discontinue therapy, the rate of success was high, with only 2 refusals.

DISCUSSION

In this prospective intervention, 44% of patients admitted to an acute‐care medical ward were prescribed a PPI prior to their admission. In the subgroup of patients for whom the indication for PPI use was recorded through our online tool, less than half had an evidence‐based indication for ongoing therapy. Our intervention was successful in increasing the proportion of patients in whom preadmission PPI prescriptions were stopped at discharge from an average of 7.7% in the preintervention phase to 18.5% during the intervention. This intervention is novel in that we were able to reduce active community prescriptions for PPIs in patients without obvious indication by nearly 50%.

Our population's rates of PPI prescription were consistent with previous reports,[11, 12, 13, 15, 25, 26, 27, 28] and it is clear that many hospitalized patients continue their PPIs at discharge without clear indications. We propose that hospitalization can serve as an opportunity to reassess the necessity of continuous PPI use. Previous systematic attempts to reduce inappropriate PPI prescriptions in hospital have met with varied success. Several of these studies were unable to achieve a demonstrable effect.[23, 29, 30, 31] In contrast, Hamzat et al.[30] described a successful educational intervention targeting inpatients on a geriatric ward. A 4‐week educational strategy was employed, and they were able to discontinue PPIs in 10 of 60 (17%) patients without indication during a limited period of study. Another successful intervention by Gupta et al.[32] involved a before‐and‐after study combining a half‐hour physician education session with the introduction of a medication reconciliation tool. They showed a decrease in inappropriate discharge prescriptions of 50%. Not only did our study demonstrate an equally sizable reduction in inappropriate discharge prescriptions, but we also employed a more methodologically sound time‐series analysis to control for unmeasured contemporaneous factors such as rotating staff practices or monthly differences in patient composition. We demonstrated an immediate and sustained improvement in performance that lasted over 6 months. Furthermore, in contrast to other interventions, which addressed inappropriate inpatient use persisting on discharge, our intervention also addressed the appropriateness of PPI use that antedated hospitalization.

There are common themes to the successful programs. The more time spent educating and reminding the prescribing physicians, the more successful the intervention. Nonetheless, we believe our intervention is not onerous or overly time consuming. We performed a short presentation each month to educate rotating physicians, and the tool took less than 1 minute to complete once the information on PPI indication was available. Frequent education sessions may be initially necessary given the comfort that many physicians have developed in prescribing PPIs. A further prerequisite for success may be a familiarity with PPI indications and potential adverse effects. Without this, the intervention may not show a demonstrable effect, as was seen in a study of pulmonologists.[23] We hypothesize that some subspecialist physicians may not have the same appreciation of the adverse effects of PPIs nor the confidence to stop them when not indicated, as compared to general internists or hospitalists.

The proportion of patients with newly initiated PPIs at discharge decreased after the intervention, but this did not reach statistical significance. Our study's power may have precluded this. However, we had also previously put in place unpublished interventions to diminish inappropriate gastric prophylaxis in the hospital, which may have diminished the effect of this intervention.

Unfortunately, although we demonstrated a clinically significant effect on PPI exit prescription rates, we still found that nearly half of the patients who were evaluated using the online tool were discharged on PPIs despite our physicians' acknowledgment that they had no identifiable indication. It is possible that clinicians do not feel comfortable stopping these medications, owing to a fear that there is an indication that they are not aware of. In certain cases, it is possible that a reappraisal of the benefits, risks, and costs might reassure the clinician that they could safely stop the drug; however, therapeutic inertia is often hard to overcome.

CONCLUSION

We have demonstrated that in medical inpatients, both PPI use and misuse remain common; however, with a combined educational and Web‐based QI intervention, we could successfully decrease inappropriate exit prescriptions. Hospitalization, particularly at academic centers, should serve as an important point of contact for residents in training and expert faculty physicians to reconsider and rationalize patient medications. We should take the opportunity to engender a culture of responsibility for all of the medications that we represcribe at discharge, including an appraisal of the relevant harms and benefits, particularly when a medication is potentially unnecessary. We ought to then communicate the rationale for any changes to our community partners to maintain continuity of care. In this way, hospitalists can help treat the prescription indigestion that has become a common affliction in modern medicine.

In 2013, there were more than 15 million Americans receiving proton pump inhibitors (PPIs),[1] with an associated drug cost of nearly $79 billion between 2007 and 2011.² PPI use is reaching epidemic proportions, likely due to the medicalization of gastrointestinal symptoms coupled with pervasive marketing and academic detailing being performed by the pharmaceutical industry.

Although PPIs are generally considered safe, they are not as innocuous as many physicians believe. In 2011 and 2012, the US Food and Drug Administration and Health Canada, respectively, issued safety advisories regarding the use of these medications related to Clostridium difficile, fracture risk, and electrolyte derangement.[3, 4, 5, 6] There have also been numerous other harmful associations reported,[7, 8, 9, 10] suggesting it would be prudent to follow Health Canada's advice that: PPIs should be prescribed at the lowest dose and shortest duration of therapy appropriate to the condition being treated.[4] In many cases this implies stopping the PPI after an appropriate duration of therapy or attempting nonpharmacological or H2‐blocker therapy instead.

Nevertheless, despite numerous cautionary publications, PPI use for nonevidence‐based indications remains common. Because they are generally thought of as outpatient medications, PPIs are frequently continued in hospitalized patients, and inappropriate outpatient therapy is rarely addressed.[11, 12, 13] Likewise, inappropriate de novo use can also be observed during hospitalization and may continue on discharge.[13, 14, 15] Hospitalization may consequently present an opportunity to employ meaningful interventions targeting outpatient medication use.[16] We developed an opportune inpatient intervention targeting inappropriate PPI therapy.

Our study had 2 aims: first, to determine the magnitude of the problem in a contemporary Canadian medical inpatient population, and second, we sought to leverage the inpatient admission as an opportunity to promote change when the patient returned to the community through the application of an educational and web‐based quality‐improvement (QI) intervention.

METHODS

RESULTS

During the preintervention period, 464 patients were admitted, of whom 209 (45%) were taking a PPI prior to admission. During their hospitalization, an additional 53 patients (21% of nonusers) were newly prescribed a PPI that was continued at discharge. During the intervention period, a total of 640 patients were admitted, of whom 281 (44%) were taking a PPI prior to admission. During their hospitalization, 60 patients (17% of nonusers) were newly prescribed a PPI that was continued at discharge. Neither the monthly proportions admitted on PPIs from prior to admission (level P=0.59, slope P=0.46) or those newly initiated on a PPI (level P=0.36, slope P=0.18) were significantly different before compared to after the intervention. However, there was both a clinically and statistically significant difference in the proportion of preadmission PPIs that were discontinued at hospital discharge from a monthly mean of 7.7% (or 16/209) before intervention to 18.5% (or 52/281) afterward (Figure 1; level P=0.03, slope P=0.48).

Figure 1

During the intervention period, our teams prospectively captured PPI indications and patient comorbidities for 54% (152/281) of the patients admitted on a PPI using the online assessment tool. The baseline characteristics of the population in whom the online tool was applied are shown in Table 2. These patients had a mean age of 69.6 years, and 49% were male. Thirty‐two percent had diabetes, 20% had chronic renal insufficiency, and 13% had experienced a gastrointestinal hemorrhage within the 3 months prior to admission. It was frequent for PPI users to receive systemic antibiotics (44%) or to have diagnoses potentially associated with PPI use such as community‐acquired pneumonia (25%) or C difficile (11%).

Fifty‐four percent (82/152) of patients in whom the online tool was applied had an evidence‐based indication (Table 3). The most common indication for PPI prescription was the receipt of antiplatelet/anticoagulant or nonsteroidal anti‐inflammatory drug with 2 other known risk factors for upper gastrointestinal bleeding (20%). In the remaining 46% (70/152) of patients, the prescription of a PPI was deemed nonevidence based. Of these, 34 (49%) had their PPI discontinued. When patients were approached to discontinue therapy, the rate of success was high, with only 2 refusals.

DISCUSSION

In this prospective intervention, 44% of patients admitted to an acute‐care medical ward were prescribed a PPI prior to their admission. In the subgroup of patients for whom the indication for PPI use was recorded through our online tool, less than half had an evidence‐based indication for ongoing therapy. Our intervention was successful in increasing the proportion of patients in whom preadmission PPI prescriptions were stopped at discharge from an average of 7.7% in the preintervention phase to 18.5% during the intervention. This intervention is novel in that we were able to reduce active community prescriptions for PPIs in patients without obvious indication by nearly 50%.

Our population's rates of PPI prescription were consistent with previous reports,[11, 12, 13, 15, 25, 26, 27, 28] and it is clear that many hospitalized patients continue their PPIs at discharge without clear indications. We propose that hospitalization can serve as an opportunity to reassess the necessity of continuous PPI use. Previous systematic attempts to reduce inappropriate PPI prescriptions in hospital have met with varied success. Several of these studies were unable to achieve a demonstrable effect.[23, 29, 30, 31] In contrast, Hamzat et al.[30] described a successful educational intervention targeting inpatients on a geriatric ward. A 4‐week educational strategy was employed, and they were able to discontinue PPIs in 10 of 60 (17%) patients without indication during a limited period of study. Another successful intervention by Gupta et al.[32] involved a before‐and‐after study combining a half‐hour physician education session with the introduction of a medication reconciliation tool. They showed a decrease in inappropriate discharge prescriptions of 50%. Not only did our study demonstrate an equally sizable reduction in inappropriate discharge prescriptions, but we also employed a more methodologically sound time‐series analysis to control for unmeasured contemporaneous factors such as rotating staff practices or monthly differences in patient composition. We demonstrated an immediate and sustained improvement in performance that lasted over 6 months. Furthermore, in contrast to other interventions, which addressed inappropriate inpatient use persisting on discharge, our intervention also addressed the appropriateness of PPI use that antedated hospitalization.

There are common themes to the successful programs. The more time spent educating and reminding the prescribing physicians, the more successful the intervention. Nonetheless, we believe our intervention is not onerous or overly time consuming. We performed a short presentation each month to educate rotating physicians, and the tool took less than 1 minute to complete once the information on PPI indication was available. Frequent education sessions may be initially necessary given the comfort that many physicians have developed in prescribing PPIs. A further prerequisite for success may be a familiarity with PPI indications and potential adverse effects. Without this, the intervention may not show a demonstrable effect, as was seen in a study of pulmonologists.[23] We hypothesize that some subspecialist physicians may not have the same appreciation of the adverse effects of PPIs nor the confidence to stop them when not indicated, as compared to general internists or hospitalists.

The proportion of patients with newly initiated PPIs at discharge decreased after the intervention, but this did not reach statistical significance. Our study's power may have precluded this. However, we had also previously put in place unpublished interventions to diminish inappropriate gastric prophylaxis in the hospital, which may have diminished the effect of this intervention.

Unfortunately, although we demonstrated a clinically significant effect on PPI exit prescription rates, we still found that nearly half of the patients who were evaluated using the online tool were discharged on PPIs despite our physicians' acknowledgment that they had no identifiable indication. It is possible that clinicians do not feel comfortable stopping these medications, owing to a fear that there is an indication that they are not aware of. In certain cases, it is possible that a reappraisal of the benefits, risks, and costs might reassure the clinician that they could safely stop the drug; however, therapeutic inertia is often hard to overcome.

CONCLUSION

We have demonstrated that in medical inpatients, both PPI use and misuse remain common; however, with a combined educational and Web‐based QI intervention, we could successfully decrease inappropriate exit prescriptions. Hospitalization, particularly at academic centers, should serve as an important point of contact for residents in training and expert faculty physicians to reconsider and rationalize patient medications. We should take the opportunity to engender a culture of responsibility for all of the medications that we represcribe at discharge, including an appraisal of the relevant harms and benefits, particularly when a medication is potentially unnecessary. We ought to then communicate the rationale for any changes to our community partners to maintain continuity of care. In this way, hospitalists can help treat the prescription indigestion that has become a common affliction in modern medicine.