Toward safer prescribing: evaluation of a prospective drug utilization review system on inappropriate prescriptions, prescribing patterns, and adverse drug events and related health expenditure in South Korea

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Available online at www.sciencedirect.com

Public Health journal homepage: www.elsevier.com/puhe

Original Research

Toward safer prescribing: evaluation of a prospective drug utilization review system on inappropriate prescriptions, prescribing patterns, and adverse drug events and related health expenditure in South Korea S.J. Kim a,b, K.-T. Han b,c, H.-G. Kang d, E.-C. Park b,e,* a

Department of Nursing, College of Nursing, Eulji University, Seongnam, Republic of Korea Institute of Health Services Research, Yonsei University, Seoul, Republic of Korea c Department of Research and analysis, National Health Insurance Service Ilsan Hospital, Ilsan, Republic of Korea d Department of Pharmaceutical Management, Health Insurance Review and Assessment Service, Seoul, Republic of Korea e Department of Preventive Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea b

article info

abstract

Article history:

Objectives: This study aimed to evaluate the effect of the prospective drug utilization review

Received 25 January 2018

(DUR) system introduced in Korea in December 2010 as a real-time method to improve

Received in revised form

patient safety, in terms of changes in prescribing practices, adverse drug events (ADEs),

5 April 2018

and ADE-related healthcare expenditure, using non-steroidal anti-inflammatory drugs

Accepted 1 June 2018

(NSAIDs) and their common ADEs as a guide. Study design: We used an interrupted time-series study design using generalized estimating equations to evaluate changes in prescription rate and ADE-related healthcare expendi-

Keywords:

ture. Cox regression analysis was used to evaluate the probability of NSAID-associated

Drug utilization review

ADEs.

Prescription drug misuse

Methods: A total of 154,585 outpatients with musculoskeletal or connective tissue disorders,

Drug-related side-effects and

without pre-existing gastric bleeding or ulcers were included in this study. The primary

adverse reactions

outcome was the level and trend change in prescription rate, drugedrug interactions,

Health expenditure

coprescribed gastro-protective drugs, and defined daily dose (DDD) of NSAIDs. The secondary outcome was the probability of ADEs and changes in ADE-related healthcare expenditure. Results: There was a significant trend change after introducing the DUR system in terms of drugedrug interactions (
3.6%) and coprescribed gastro-protective drugs (þ0.6%). The mean DDD of NSAIDs increased by 0.2. The probability of ADEs decreased overall (
1.7%) and in the high-risk group (age 65 years;
9.6%); however, only the latter was significant. There was no significant trend or level change in ADE-related health expenditure.

* Corresponding author. Department of Preventive Medicine & Institute of Health Service Research, Yonsei University College of Medicine, 50 Yonsei-ro, Seodaemun-gu, Seoul 120-752, Republic of Korea. Tel.: þ82 2 2228 1862; fax: þ82 02 392 8133. E-mail address: [email protected] (E.-C. Park). https://doi.org/10.1016/j.puhe.2018.06.009 0033-3506/© 2018 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

129

Conclusions: The introduction of the DUR system was associated with more efficient prescribing, including a reduction in drugedrug interactions and an increase in the use of gastro-protective drugs. The system had a positive effect on patient outcome but was not associated with reduced ADE-related costs. Further studies are needed to evaluate the long-term effects of the DUR system in Korea. © 2018 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

Introduction Non-steroidal anti-inflammatory drugs (NSAIDs) are among the most widely used medications in many countries. They are considered effective for the treatment of chronic conditions of pain and inflammation.1e3 However, NSAIDs are associated with many adverse drug events (ADEs), including gastrointestinal (GI) bleeding and the risk of cardiovascular disease.4e6 Approximately 18% of patients using NSAIDs experience ADEs, either due to overutilization or inappropriate use.7,8 Prescribing drugs involves weighing up the benefits and risks; careful decision-making is needed to improve their safe use. Thus, many countries have focused on preventing potential ADEs associated with medical errors and have adopted systems that monitor the prescription of the drugs (drug utilization review [DUR] systems) to reduce ADEs and improve patient safety. DUR is defined as an ongoing systematic approach to review drug prescription and utilization.9e11 The DUR system operates prospectively (detecting inappropriate prescription) and retrospectively (providing feedback to the provider, including preventable drug-related morbidity and specific drug utilization). Many countries, including the US, have adopted this system to improve patient outcomes.12 In Korea, the need for a DUR system was raised in 2003. In response, in 2004, the Ministry of Food and Drug Administration announced a list of age-related drug contraindications and drugedrug interactions.13 The list of registered drugs has increased annually to reduce inappropriate prescriptions, but there has been no integrated approach to improving patient safety because the list was adopted separately through a different claim submission system. Hence, the government decided to introduce a DUR system in Korea. Starting in April 2008, a DUR system was piloted in one city. This provided useful information, including information on drugedrug interactions and the duplication of ingredients.13,14 After this pilot test, the DUR system was rolled out nationwide in December 2010. To improve patient safety and reduce pharmaceutical expenditure, this system provides information on drugs that were previously prescribed or prescribed by different hospitals and checks whether the prescriptions are inappropriate in real time. According to previous studies, DUR systems have reduced inappropriate drug prescription and drugedrug interactions.15e18 In addition, actionable drug information of DUR systems has reduced healthcare expenditure and controlled drug utilization, particularly of frequently associated with the risk of ADEs.19,20 Some researchers have suggested that a DUR

system alone is insufficient to change prescribing patterns and that such change depends on physicians' drugprescribing choices.12,21 In Korea, many researches have suggested that the DUR system has improved prescribing and reduced pharmaceutical expenditure.13,14,22,23 These studies focused on actual change in prescribing as a result of the DUR and its operating system and evaluated overall prescribing changes. In addition, most of these studies focused on the changes in utilization of antibiotic agents. There was a lack of evidence of changes in the prescription of NSAIDs, the most commonly used prescription drugs in Korea. We investigated whether the introduction of a DUR system would result in a change in prescription patterns and in quality of care. First, we hypothesized that drug information provided by the DUR system would change prescription patterns and drug utilization. As the introduction of the DUR system could impact drug choices, we studied whether there was any change in drugedrug interactions involving NSAIDs. In addition, as the information provided could also influence physicians' use and choice of gastro-protective drugs to reduce NSAID-related ADEs, we evaluated changes in the prescription of proton-pump inhibitors (PPIs) and H2 receptor antagonists (H2RA). Furthermore, it could influence the utilization of NSAIDs; hence, we evaluated changes in defined daily dose (DDD) prescribed. Second, a change in prescription patterns may impact patient outcome; we examined NSAIDrelated ADEs (GI ulcer or bleeding) after prescription of NSAIDs. Finally, ADE-related health expenditure may also decrease after the introduction of a DUR system; thus, we evaluated changes in healthcare expenditure due to ADEs.

Methods Database and data collection We used National Health Insurance Services (NHIS) sampling cohort data from 2002 to 2013. Using NHIS sampling cohort data, we selected outpatients who had musculoskeletal or connective tissue disorders based on the International Classification of Disease (ICD)-10 codes (M00-M99) during the period 2008e2013. First, we considered conditions that may mimic NSAID-related ADEs (GI bleeding or ulcer) as these may have created noise in our outcome variable. Thus, we excluded patients who attended hospital services with GI disease as a major diagnosis (ICD-10 codes: K226, K228, K25eK29) during the period 2002e2007. Second, we excluded patients who were not prescribed NSAIDs based on

130

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Anatomical Therapeutic Chemical (ATC) codes (M01A). Third, we excluded patients diagnosed with GI disease before being diagnosed with musculoskeletal or connective tissue disorders. Next, we excluded patients based on age (<20 or 80 years) and residence in healthcare facilities (long-term care facilities or community health centers). We also excluded patients who could not calculate a DDD. Finally, we excluded patients who lived in Goyang-si as this was the city where the pilot test for the DUR system was conducted. Ultimately, 154,585 patients were included in our study.

Variables Primary outcome variables in this study were prescription rate (drugedrug interactions, gastro-protective drugs) and drug utilization (DDD per day). First, we used drug information data and evaluated whether coprescription of drugs that potentially interact with NSAIDs existed (see Appendix). If patients were prescribed drugs that may interact with NSAIDs, then we classified this as 1, and if not, this was classified as 0. Second, we considered PPIs (ATC code: A02BA) and H2RA (ATC code: A02BC) as gastro-protective drugs.24 If patients were coprescribed gastro-protective drugs, we classified the gastro-protective drug as ‘yes’. Third, we calculated DDD per day based on ATC code (M01AA~M01AX). The secondary outcome was change in NSAID-related ADEs and related health expenditure. We defined ADEs as patients who were treated for GI disease within 8 weeks after prescription of NSAIDs (ICD-10 codes: K226, K228, K25-K29).1 The followup period was considered to start on the first date of prescription of NSAIDs and to end on the date of treatment for GI disease. For participants who were not treated for GI disease, the end date was calculated using end date of NSAIDs plus 8 weeks (see Appendix). Next, we calculated health expenditure of ADEs including outpatient care, pharmaceutical costs, and admission costs. Finally, all outcome variables were aggregated by month. The percentage was calculated based on the number of events of outcome variables divided by the total number of patients per month. The primary variable of interest in this study was the level and overall trend change in outcome variables after introduction of a DUR system in Korea. The baseline trend represents the rate of the monthly change in outcome variables based on data before the intervention period. Level change indicates the change at the moment of intervention. Trend change is the rate of change of an outcome variable, defined as an increase or decrease in the slope of the segment after the intervention compared with the segment preceding the intervention (see Appendix). The index date of introduction of the DUR system was in December 2010. To investigate changes in outcome variables, we used a dummy variable based on the index date; the time before the introduction of the DUR system was defined as 0, and the time after the introduction of the DUR system was defined as 1. In addition, our analysis of trends examined linear changes after introduction of the DUR system. The overall trends were stratified by month and included data from January 2008 to December 2013. We adjusted for patient and medical institution characteristics when analyzing changes in outcome variables after

the introduction of the DUR system. Patient characteristics such as sex, age (20e39, 40e59, or 60e79 years), Charlson comorbidity index, insurance type (medical aid, self-employed, or employee), income status (bottom, middle-bottom, middle-top, or top), type of disease (arthropathies, systemic connective tissue disorders, spinal diseases, soft tissue disorders, osteopathies and chondropathies, or other diseases), and year of healthcare utilization were included in our analysis. The characteristics of a medical facility included the type of institution (clinic, hospital, or general hospital), location (metropolitan or non-metropolitan), number of beds (low, moderate, high), and number of doctors.

Statistical analysis The distribution of each categorical variable was examined by an analysis of frequencies and percentages, and c2 tests were performed. Analysis of variance was also performed to compare average values and standard deviations for the continuous variables. To evaluate changes in outcome variables after introduction of the DUR system, we used an interrupted time-series study design using generalized estimating equations.25e28 We performed Poisson regression analysis with a log link function to evaluate changes in the pattern of prescriptions and health expenditure. In addition, we used Cox regression models to determine the association between onsets of ADEs after prescription of NSAIDs. Hazard ratios (HRs) for the onset of ADEs were estimated, starting at the first date of prescription of NSAIDs and accounting for the duration until the onset of ADEs or 8 weeks after the end of NSAIDs. Because age older than 65 years was considered as a high risk for GI adverse effects, we evaluated further analysis based on age groups (<65, 65 years). All statistical analyses were performed using SAS, version 9.4, (SAS Institute, Inc.; Cary, NC, USA). P-values <0.05 were considered statistically significant.

Ethical consideration This study was approved by the Institutional Review Board, Yonsei University Graduate School of Public Health (IRB number: Y-2017-0057).

Results The data used in our study were from 154,585 patients and 17,469 medical institutions including clinics, hospitals, and general hospitals. The average prescribed dose of NSAIDs was 1.11 DDD before the introduction of the DUR system. It decreased after adoption of the DUR system to 1.07 DDD. The mean ADE-related health expenditure was 50,567 Korean Won (KRW) before and 46,233 KRW after the introduction of the DUR system. However, these changes were not statistically significant. Individuals aged 35e49 years comprised the largest age group being prescribed NSAIDs (n ¼ 30,706 [32.3%]) before the DUR was introduced; this shifted to 50- to 64-yearolds (n ¼ 34,260 [31.6%]) after introduction of the DUR system. Most patients had spinal diseases before (35.5%) and after (36.1%) introduction of the DUR system (Table 1).

131

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Table 1 e Baseline characteristics of patients and hospitals by introduction of the drug utilization review (DUR) system. Characteristics

DUR system

P-value

Before Outcome variable (mean ± SD) DDD/day 1.11 Costs for ADEs (KRW) 50,567 Patient characteristics (n ¼ 154,585) Sex (n [%]) Male 43,999 Female 51,044 Age in years (n [%]) 25e34 17,399 35-49 30,706 50-64 27,896 65-79 19,042 CCI 1.65 Insurance type (n [%]) Medical aid 1816 Self-employed 35,080 Employees 58,147 Income (n [%]) Bottom 16,061 Middle-bottom 24,586 Middle-top 30,219 Top 24,177 Type of disease (n [%]) Arthropathies 29,106 Spinal disease 33,726 Soft tissue disorders 30,853 Osteopathies and chondropathies 994 Other disease 364 Year of healthcare utilization (n [%]) 2008 45,673 2009 27,873 2010 21,497 2011 2012 2013 Medical institution characteristics (n ¼ 17,469) Type of facilities (n [%]) Clinic 12,407 Hospital 1027 General hospital 339 Location (n [%]) Metropolitan 6338 Non-metropolitan 7435 No. of beds (n [%]) Low 6748 moderate 5248 high 1777 4.11 No. of doctorsa

After ±0.12 ±18,288

1.07 46,233

±0.03 ±26,011

0.0327 0.4186

(46.29) (53.71)

51,527 57,066

(47.45) (52.55)

<0.0001

(18.31) (32.31) (29.35) (20.04) 1.39

18,608 32,642 34,260 23,083 1.77

(17.14) (30.06) (31.55) (21.26) 1.42

<0.0001

(1.91) (36.91) (61.18)

6559 34,749 67,285

(6.04) (32.00) (61.96)

<0.0001

(16.90) (25.87) (31.80) (25.44)

21,982 25,995 33,009 27,607

(20.24) (23.94) (30.40) (25.42)

<0.0001

(30.62) (35.48) (32.46) (1.05) (0.38)

32,409 39,247 35,346 1144 447

(29.84) (36.14) (32.55) (1.05) (0.41)

0.0016

7764 46,016 30,168 24,645

(7.15) (42.37) (27.78) (22.69)

(90.08) (7.46) (2.46)

12,409 1111 343

(89.51) (8.01) (2.47)

0.2197

(46.02) (53.98)

6400 7463

(46.17) (53.83)

0.8044

(48.99) (38.10) (12.90) 31.22

6852 5198 1813 4.34

(49.43) (37.50) (13.08) 33.08

0.5762

(48.06) (29.33) (22.62)

<0.0001

0.5570

a

Mean units and standard deviation. ADEs, adverse drug events; CCI, Charlson Comorbidity Index; DDD, defined daily dose; SD, standard deviation; KRW, Korean Won.

The average monthly prescription rate of cyclooxygenase-2 (COX-2) selective inhibitors and coprescribed gastroprotective drugs increased. The number of patients who were prescribed drugs with potential drugedrug interactions was small and seemed to decrease after the DUR system was introduced. In terms of ADEs, the number of ADEs fluctuated over (calendar) time and tended to decrease from the end of 2012 (Fig. 1). The interrupted analysis showed that monthly prescription rates changed after the introduction of the DUR system. A decreasing trend was observed in the coprescription of drugs

with drugedrug interactions; this rate decreased by an average of 3.6% per month after the DUR system was introduced, although this change was not statistically significant. In addition, a significant increasing trend was observed in coprescription of gastro-protective drugs; the prescription rate increased by 0.6% per month compared with previous periods. A significant trend was observed in the mean dosage of NSAIDs after introduction of the DUR system; the DDD increased by 0.2 (Fig. 2). ADEs were evaluated based on the date of prescription of NSAIDs; a potential event was considered an ADE if it occurred

132

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Fig. 1 e Overall trends in prescription rates, including drugedrug interactions, coprescribed gastro-protective drug, selective non-steroidal anti-inflammatory drugs, and adverse drug events. DUR, drug utilization review.

within 8 weeks of NSAID prescription. We measured ADEs overall and in high-risk (age 65 years) and low-risk (age <65 years) groups. The probability of ADEs decreased after introducing the DUR system; however, this was not statistically significant overall (HR: 0.98, p-value ¼ 0.4189). In the high-risk group, the probability of GI ulcer or bleeding decreased after introduction of the DUR system (HR: 0.90, p-value; 0.0299). A non-significant increase in the probability of an ADE was observed in the low-risk group. Although there was a change in ADE-related healthcare expenditure, the trend was nonsignificant overall and in the high- and low-risk groups (Table 2).

Discussion In this study, we evaluated the effects of a DUR system on patient care in terms of prescription and patient outcome. First, we found that introduction of a DUR system had partially positive effects on the pattern of NSAID prescription. There was a decreased trend in coprescription of drugs that potentially interact with NSAIDs. In addition, the prescription of gastro-protective drugs increased. These findings are similar to those of other studies evaluating the effect of a DUR system on prescription; DUR is associated with reduced number of prescribing errors.17,19,29,30 There was a 24% change in more appropriate use of therapeutic agents with the DUR system. Outpatient prescription was also influenced by alerting prescribers of any potentially harmful prescriptions.31 Although this changed the inappropriate use of drugs, another study suggested that the DUR did not improve prescription quality.32 Different effects of the DUR system on

prescription have been suggested; the present study suggests that drug monitoring systems have important implications for improving patient safety. The alarm notification may have changed physicians' prescribing decisions, as drug information was provided explaining why the prescription was considered inappropriate. Hence, this may have led to the reduction in coprescription of drugs with drugedrug interactions and the potential for ADEs. It may also have impacted the decision to prescribe other drugs to reduce the risk of ADEs. To reduce GI bleeding or ulcers, both of which frequently occur in patients taking NSAIDs, physicians either changed the prescription or coprescribed gastro-protective drugs. Although the rate of inappropriate prescription changed after introduction of the DUR system, a slight increase in DDD was observed in our study; an increase of approximately 0.2% was observed, compared with previous trends. A plausible explanation is related to the registered list of NSAIDs in the DUR system: in 2013, a total of 28 drugs were included with a caution for dosage, and in 2014, dosage contraindications for NSAIDs (such as diclofenac and ibuprofen) were introduced. Before 2013, there was no information about dosage contraindications with NSAIDs; therefore, NSAID utilization may have differed after updating the drug list. However, there might have been little or no change in dosage because physicians frequently choose NSAIDs as primary drugs for treatment in many diseases. In addition, gastroprotective drugs would be added rather than changing the NSAIDs dosage. Second, we found that the probability of ADEs decreased after introduction of the DUR. The probability of GI ulcer or bleeding decreased in the high-risk groups but was

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

133

Fig. 2 e The results of predicted outcome variables after introduction of the drug utilization review (DUR) system. DDD, defined daily dose. *Statistically significant. The baseline trend shows the rate of monthly change before the intervention period. The level change indicates the change at the moment of intervention. A change in trend was defined as an increase or decrease in the slope of the segment after the intervention, compared with the segment preceding the intervention.

unchanged overall. This may be associated with physicians' prescribing decisions for elderly individuals who are considered to be more vulnerable to ADEs.33 In general, patients aged 65 years have chronic diseases, for which many drugs are

often prescribed. Therefore, physicians would have carefully considered two aspects: potential risk of ADEs with prescribed non-selective NSAIDs and drugedrug interactions between NSAIDs and drugs prescribed for other diseases. The DUR

134

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Table 2 e Change in adverse drug events (ADEs) and ADE-related health expenditure after introduction of the drug utilization review (DUR) system. HRa/RRb ADEs Overall (20  age <80 years) High risk (age 65 years) Nonehigh risk (age <65 years) ADE-related health expenditure Overall (20  age <80 years) Baseline trend Level change Trend change High risk (age 65 years) Baseline trend Level change Trend change Nonehigh risk (age <65 years) Baseline trend Level change Trend change

95% CI

P-value

0.983 0.904 1.006

0.942 0.826 0.959

1.025 0.990 1.056

0.4189 0.0299 0.8038

1.062 1.289 0.974

0.978 0.739 0.924

1.153 2.249 1.027

0.1529 0.3717 0.3327

1.202 0.718 0.990

0.994 0.325 0.887

1.453 1.583 1.104

0.0571 0.4113 0.8514

1.013 1.568 0.969

0.929 0.840 0.910

1.105 2.926 1.032

0.7679 0.1578 0.3261

CI, confidence interval. Adjusted for sex, age, Charlson Comorbidity Index, type of disease, insurance type, income, gastro-protective drugs, type of facilities, no. of beds, location, and no. of doctors. The baseline trend shows the rate of monthly change before the intervention period. The level change indicates the change at the moment of intervention. A change in trend was defined as an increase or decrease in the slope of the segment after the intervention, compared with the segment preceding the intervention. a HR (hazard ratio): HR indicates the probability of ADEs. b RR (rate ratio): RR indicates health expenditure for ADEs. It shows results of exponentiated estimates and is interpretable as percentage changes in health expenditure.

system provides information about ADEs related to nonselective NSAIDs; this may have changed physicians' prescribing decisions, increasing the use of selective COX inhibitors rather than non-selective NSAIDs. Monthly prescription rates were higher in those 65 years old, and this seemed to increase after the introduction of the DUR system (see Appendix). Furthermore, reimbursement restrictions on selective COX inhibitors also possibly had an effect on the increase in utilization in this age group; high-risk groups (such as 65 years old, existing GI ulcer or bleeding, or those on a steroidal agent) can be insured under the National Health Insurance. In addition, the increased use of gastro-protective drugs may have also impacted the probability of ADEs in the high-risk group. Eventually, these changes in prescribing practices may have a positive effect on patient outcomes, possibly leading to reduction in the probability of GI ulcer or bleeding in patients aged 65 years. However, no significant change in ADE-related healthcare expenditure was observed. Thus, further studies considering health expenditure with other ADEs are needed to evaluate the effects of the DUR system. To the best of our knowledge, our study is the first to evaluate the effect of a DUR system on prescribing patterns, ADEs, and ADE-related healthcare expenditure in Korea. Previous investigations have considered drugedrug interactions and pharmaceutical expenditure, but no prior study evaluated the ADEs of NSAIDs and ADE-related healthcare expenditure. Second, the interrupted time-series analysis design is a powerful method for evaluating the change. Our results provide valuable evidence to policy makers regarding the effects

of the DUR system in Korea. Third, we used the NHIS sampling cohort data, including a large, representative sample of individuals in Korea; therefore, our results should be of significance to policy makers. Finally, our results are also meaningful to other countries adopting a DUR system to improve patient care and reduce ADEs. Despite these strengths, our study does have some limitations. First, we used claims data; therefore, we were unable to assess the clinical conditions that affected prescribing decisions; this information could have impacted the changes in prescription of NSAIDs and the prescription of gastroprotective drugs. Second, unmeasured patient factors, such as dietary habits, can affect GI bleeding or ulcers. However, we excluded patients with pre-existing or previous GI ulcers or bleeding and limited the period during which these events would be considered ADEs, to reduce noise in our study. Third, we only considered changes in the prescription of NSAIDs; different results may be found if the prescribing patterns of other drugs are assessed. Thus, further studies considering other drugs are needed to evaluate the effects of the DUR system. Fourth, because it was updated, the registered lists of drugs and the information therein did not remain constant. Evaluation over a longer period of time is needed to further improve the DUR system. Fifth, we only considered GI bleeding and ulcers; no other ADEs were evaluated. Furthermore, we only considered insured benefit of ADEs as we could not evaluate uninsured costs that were not submitted to the NHIS claim system. Finally, we were unable to evaluate the use of over-the-counter NSAIDs; this may have affected the measurement of NSAID-related ADEs.

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

Conclusion In conclusion, our findings suggest that use of a DUR system is associated with improved prescribing practices, including a reduction in drugedrug interactions and an increase in the use of gastro-protective drugs, and with reduced ADEs among the elderly. Further studies that consider other drug information and have longer periods of follow-up are needed. Nonetheless, these findings provide valuable evidence to improve prescription quality and prevent ADEs.

Author statements Acknowledgments This study used the National Health Insurance Services sampling cohort made by the National Health Insurance Services (Research Number: REQ0000017678).

Ethical approval This study was approved by the Institutional Review Board, Yonsei University Graduate School of Public Health (IRB number: Y-2017-0057).

Funding This research was funded by the Health Fellowship Foundation. However, the funding sources did not have input into the study, such as study design or data interpretation.

Competing interests None declared.

Contributors S.J.K. conceived the idea for the study, carried out the statistical analysis, interpreted the data, and drafted the manuscript. K.-T.H. substantially contributed to the interpretation of the data and analysis. H.-G.K. substantially contributed to drafting of the article. E.-C.P. acted as an advisor to the study design and substantially contributed to the conception and drafting of the article.

references

1. Dreischulte T, Donnan P, Grant A, Hapca A, McCowan C, Guthrie B. Safer prescribingda trial of education, informatics, and financial incentives. N Engl J Med 2016;374:1053e64. 2. Day RO, Graham GG. Republished research: non-steroidal anti-inflammatory drugs (NSAIDs). Br J Sports Med 2013;47. 1127-. 3. Bjarnason I, So A, Levi AJ, Peters T, Williams P, Zanelli G, et al. Intestinal permeability and inflammation in rheumatoid arthritis: effects of non-steroidal anti-inflammatory drugs. Lancet 1984;324:1171e4. 4. Laine L. GI risk and risk factors of NSAIDs. J Cardiovasc Pharmacol 2006;47:S60e6.

135

5. Motsko SP, Rascati KL, Busti AJ, Wilson JP, Barner JC, Lawson KA, et al. Temporal relationship between use of NSAIDs, including selective COX-2 inhibitors, and cardiovascular risk. Drug Saf 2006;29:621e32. 6. Winkelmayer WC, Waikar SS, Mogun H, Solomon DH. Nonselective and cyclooxygenase-2-selective NSAIDs and acute kidney injury. Am J Med 2008;121:1092e8. 7. Guthrie B, McCowan C, Davey P, Simpson CR, Dreischulte T, Barnett K. High risk prescribing in primary care patients particularly vulnerable to adverse drug events: cross sectional population database analysis in Scottish general practice. Bmj 2011;342:d3514. 8. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. Bmj 2004;329:15e9. 9. Fulda TR, Lyles A, Pugh MC, Christensen DB. Current status of prospective drug utilization review. J Manag Care Pharm 2004;10:433e41. 10. Moore WJ. Medicaid drug utilization review: a critical appraisal. Med Care Res Rev 1994;51:3e37. 11. Chui M. Evaluation of online prospective DUR programs in community pharmacy practice. J Manag Care Pharm 2000;6:27e32. 12. Hennessy S, Bilker WB, Zhou L, Weber AL, Brensinger C, Wang Y, et al. Retrospective drug utilization review, prescribing errors, and clinical outcomes. JAMA 2003;290:1494e9. 13. Yang J-H, Kim M, Park Y-T, Lee E-K, Jung CY, Kim S. The effect of the introduction of a nationwide DUR system where local DUR systems are operatingdthe Korean experience. Int J Med Inf 2015;84:912e9. 14. Heo JH, Suh DC, Kim S, Lee E-K. Evaluation of the pilot program on the real-time drug utilization review system in South Korea. Int J Med Inf 2013;82:987e95. 15. Peng CC, Glassman PA, Marks IR, Fowler C, Castiglione B, Good CB. Retrospective drug utilization review: incidence of clinically relevant potential drug-drug interactions in a large ambulatory population. J Manag Care Pharm 2003;9:513e22. 16. Armstrong EP, Wang SM, Hines LE, Patel BV, Leslie RS, Malone DC. Prescriber perceptions of a near real-time fax alert program for potential drug-drug interactions. J Manag Care Pharm 2014;20:494e500. 17. Monane M, Matthias DM, Nagle BA, Kelly MA. Improving prescribing patterns for the elderly through an online drug utilization review intervention: a system linking the physician, pharmacist, and computer. J Am Med Assoc 1998;280:1249e52. 18. Hanlon JT, Fillenbaum GG, Kuchibhatla M, Artz MB, Boult C, Gross CR, et al. Impact of inappropriate drug use on mortality and functional status in representative community dwelling elders. Med Care 2002;40:166e76. 19. Daubresse M, Gleason PP, Peng Y, Shah ND, Ritter ST, Alexander GC. Impact of a drug utilization review program on high-risk use of prescription controlled substances. Pharmacoepidemiol Drug Saf 2014;23:419e27. 20. Trygstad TK, Christensen D, Garmise J, Sullivan R, Wegner SE. Pharmacist response to alerts generated from Medicaid pharmacy claims in a long-term care setting: results from the North Carolina polypharmacy initiative. J Manag Care Pharm 2005;11:575e83. 21. Piecoro LT, Browning SR, Prince TS, Ranz TT, Scutchfield FD. A database analysis of potentially inappropriate drug use in an elderly Medicaid population. Pharmacotherapy 2000;20:221e8. 22. Shin JY, Kim MH, Shin SM, Lee SH, Park BJ. Dramatic decrease in fluoroquinolones in the pediatric population in Korea. Pharmacoepidemiol Drug Saf 2014;23:1320e4.

136

p u b l i c h e a l t h 1 6 3 ( 2 0 1 8 ) 1 2 8 e1 3 6

23. Kim D-S, Huh S, Lee S. Potentially inappropriate medication use at ambulatory care visits by elderly patients covered by National Health Insurance in Korea. Int J Clin Pharm Ther 2015;53:819e29. 24. Chan EW, Lau WC, Leung WK, Mok MT, He Y, Tong TS, et al. Prevention of dabigatran-related gastrointestinal bleeding with gastroprotective agents: a population-based study. Gastroenterology 2015;149. 586e595. e3. 25. Zeger SL, Liang K-Y, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics 1988:1049e60. 26. Hanley JA, Negassa A, Forrester JE. Statistical analysis of correlated data using generalized estimating equations: an orientation. Am J Epidemiol 2003;157:364e75. 27. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Therapeut 2002;27:299e309. 28. Zhang F, Wagner AK, Soumerai SB, Ross-Degnan D. Methods for estimating confidence intervals in interrupted time series analyses of health interventions. J Clin Epidemiol 2009;62:143e8. 29. Moore WJ, Newman RJ. Drug formulary restrictions as a costcontainment policy in Medicaid programs. J Law Econ 1993;36:71e97.

30. Starner CI, Norman SA, Reynolds RG, Gleason PP. Effect of a retrospective drug utilization review on potentially inappropriate prescribing in the elderly. Am J Geriatr Pharmacother 2009;7:11e9. 31. Biru TT, Defersha AD, Gelaw BK, Tegegne GT. Drug utilization review of ciprofloxacin in the outpatient department of Boru Meda Hospital, South Wollo Zone, Amhara Region, Ethiopia. 2014. goire J-P, Moisan J, Potvin L, Chabot I, Verreault R, Milot A. 32. Gre Effect of drug utilization reviews on the quality of in-hospital prescribing: a quasi-experimental study. BMC Health Serv Res 2006;6:1. 33. Laine L. Approaches to nonsteroidal anti-inflammatory drug use in the high-risk patient. Gastroenterology 2001;120:594e606.

Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.puhe.2018.06.009.