Sage Journals: Discover world-class research

Abstract

This study sought to determine how the proportion of physicians using electronic prescribing in nine US states was associated with the hospitalization rate for adverse drug events among older adult patients. A discharge-level analysis of the relationship between county electronic prescribing and adverse drug event hospitalization rates was conducted. Data from the 2011 State Inpatient Databases, the Office of the National Coordinator Health IT Dashboard, and the Area Health Resource File were obtained for nine US states. The analysis examined the odds that a discharge for older adults would have been adverse drug event associated, versus other causes, using multivariable logistic regression models. After adjusting for patient, provider, health infrastructure, and community factors, the lowest county electronic prescribing rate quartile was associated with significantly greater odds of an adverse drug event hospitalization (odds ratio: 1.10; 95% confidence interval: 1.02–1.19). Early results indicate greater odds of adverse drug event hospitalizations among older adults living in counties with low electronic prescribing rates when compared to those in high electronic prescribing counties.

Keywords

adverse drug events electronic prescribing health information technology IT healthcare evaluation older adults patient safety

Introduction

Patient safety and quality of care have received increased attention from the healthcare industry since the release of the 1999 Institute of Medicine report, To Err is Human.¹ This heightened attention has continued as tools to identify adverse drug events (ADEs) and reactions are becoming widely available. ADEs, defined as injuries that are the result of medication, are of particular public health importance because of the increasingly high number of Americans taking prescription drugs, particularly older Americans. Across 2007–2010, approximately 40 percent of Americans aged 65 years and over used five or more prescription drugs in the past 30 days.² Occurrences of ADEs are also seen in older adults nearly seven times more often than among adults younger than 65 years, making ADEs an important indicator of patient safety.³

ADEs that occur prior to hospital admissions are of particular interest because they are three times more common than ADEs that originate during a hospital stay.⁴ Of all hospital stays, it is estimated that between 2.4 and 9.0 percent are caused by ADEs that are present at hospital admission.^5–8 Von Laue et al.⁸ determined that over half of ADEs were preventable. With the high risk and cost of ADEs in the older patient population, it is essential to ensure that effective and appropriate ambulatory prescribing habits are followed and ADEs that result in hospitalization are prevented.^9,10

Kaushal et al.¹¹ describe the potential of health information technology (HIT) to improve patient safety in the outpatient setting. Additionally, Bates and Gawande¹² argue that electronic prescribing (e-prescribing) provides a method to reduce errors by preventing errors and adverse events. E-prescribing provides a tool to overcome the common risk factors of ADEs, including the lack of communication among concurrent prescribers.^13–15

In an effort to increase the adoption of e-prescribing, federal legislation and incentive programs have been established. These programs include the Medicare Prescription Drug, Improvement, and Modernization Act of 2003;¹⁶ the e-prescribing incentive program as part of the Medicare Improvements for Patients and Providers Act of 2008;¹⁷ and most recently the electronic health record incentive program through the Health Information Technology for Economic and Clinical Health Act of the American Recovery and Reinvestment Act of 2009.¹⁸ The combination of incentives has resulted in an increase in e-prescribing adoption from approximately 7 percent in December 2008 to 70 percent in April 2014.¹⁹ However, whether the increase in HITs has resulted in meaningful population health improvements is undetermined.

Previous studies examining the impact of HIT on ADEs have been limited to a hospital setting and determining the rate of ADEs occurring within a hospital stay.^20–32 No studies were found that examined the impact of HIT use in ambulatory settings on ADE hospitalizations. This study examines the odds of an older adult having an ADE hospitalization due to prescriptions provided outside a hospital setting, given the rate of e-prescribing in their county of residence.

Methodology

Theoretical model

An ecological modification of the Ancker et al.³³ Triangle Model was used as a theoretical framework for this analysis. While the Triangle Model lends itself to evaluating the impact of HIT within an organization, we utilized the framework at a systems/community level. The four main constructs of the Triangle Model (technology, provider, organization, and patient) were modified to technology, adopter/provider, healthcare infrastructure, community, and patient.

Data sources

Data were drawn from three datasets: 2011 Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases (SID),³⁴ the ONC Health IT Dashboard—Health IT Adoption and Use dataset,³⁵ and the Health Resources and Services Administration (HRSA) Area Health Resource File (AHRF).³⁶ Datasets were merged using state and county Federal Information Processing Standards (FIPS) codes.³⁷ Key data elements for this study from HCUP SID include patient demographics, such as sex, age, race, number of chronic conditions using HCUP’s Chronic Condition Indicator for the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), and length of stay; principal diagnosis; and whether the diagnosis was present on admission (POA).³⁴ The POA indicator, commonly collected to provide documentation to Centers for Medicare and Medicaid Services, allows for the identification of ADEs that occurred in the community and are likely to be associated with outpatient provider prescriptions, rather than during a hospital stay. This study was exempted by the University of South Carolina Institutional Review Board.

Study sample

Due to budgetary constraints and per-state costs for HCUP SID files, this analysis used data from nine states: Arizona, Arkansas, Florida, Maryland, Michigan, Mississippi, New Jersey, Vermont, and Washington. States were chosen due to the availability of patient residence and POA indicator data elements while also allowing for the representation of at least one state from each of the four major US Census regions.

To ensure that our analysis did not inaccurately estimate the association between e-prescribing and ADE hospitalizations, several criteria were used to exclude discharges, hospitals, and counties from the county- and discharge-level analyses. Counties with missing e-prescribing rates were excluded (n = 16). Mirroring the analysis conducted by Weiss et al.,⁴ hospitals that reported all diagnoses as POA and hospitals with over 20 percent of discharges with missing POA indicators for non-missing diagnoses were excluded. The number of hospitals represented was reduced from 1099 to 899 after hospital exclusion criteria were applied. We also excluded discharges with missing age or age less than 65 years, missing POA information for non-missing diagnoses, and missing county of residence. Finally, discharges for out-of-state patients were excluded. The final study population consisted of 2,725,378 discharges across 404 counties.

Dependent variable

Following the algorithm utilized by Encinosa and Bae,³⁸ we modified the Lucado et al.³ method to flag ADEs. To only identify ADEs that occurred within the community and outside of a hospitalization, our modification defines ADE hospitalizations as hospital discharges that had (1) a POA indicator and (2) an ICD-9-CM diagnosis code for an ADE. Both criteria were used to ensure that ADEs were acquired prior to hospitalization. The Lucado et al.³ list of ICD-9 codes and external cause of injury codes were used to determine ICD-9 codes for ADEs. Discharges with ICD-9 codes that pertain to accidental poisonings, self-inflicted poisonings, and/or assault were not flagged as ADEs. The analysis utilized a dichotomous indicator of having an ADE hospitalization.

Independent variable

The ONC Health IT Dashboard—Health IT Adoption and Use dataset was utilized to determine the independent variable, the percentage of physicians e-prescribing in each county. The ONC Health IT Dashboard is an open government project that provides percentage of e-prescribers on the country’s largest e-prescribing network, Surescripts, by county.^35,39 The adoption rate of counties was categorized as either high/low when compared to the median rate (35%).

Control variables

Multivariable analyses controlled for potential confounders at two levels: patient and county. Patient demographics controlled for patient sex, age group (65–74, 75–84, and 85+ years), race/ethnicity (White, African American, Hispanic, and Other), number of chronic conditions (0–20), and length of hospital stay in days. The HRSA AHRF was the source for control variables. Adopters/providers were determined through the AHRF by including the number of primary care physicians per 1000 population in the county (expressed in quartiles). Healthcare infrastructure was represented by including the number of hospital beds per 1000 population in the county (expressed in quartiles), whether the county had a hospital, and whether a county was designated as a Health Professional Shortage Area (HPSA; whole, partial, and none). Last, the community variables utilized included whether a county was rural (defined by the 2003 Urban Influence Code⁴⁰ of greater than 3), the percentage of the population living in poverty (expressed in quartiles), and the percentage of the population with less than a high school education (expressed in quartiles).

Analytic approach

A discharge-level analysis was used to determine whether the odds of a hospitalization being attributable to ADE were associated with county e-prescribing rates. Discharges were assigned to counties based on the patient county of residence. Differences across county e-prescribing adoption rates were assessed using Wald chi-square tests (α = 0.05). Multivariable logistic regression was utilized to identify patient and county characteristics that were associated with having an ADE hospitalization. Five logistic regression models were conducted to determine the impact of each major construct of the theoretical framework: technology, patient, adopters/providers, healthcare infrastructure, and community.

Results

Table 1 displays the characteristics of hospital discharges in the study population, by the county e-prescribing rate. Discharges for patients who resided in the highest e-prescribing counties were more likely female, between the ages of 65 and 74 years, White, and had a shorter length of hospital stay. The counties in which these patients reside have fewer primary care physicians, fewer hospital beds per population, are more likely to be whole county HPSA, are more likely to be rural, and have close to median levels of poverty and high school education levels.

Table 1.

Characteristics of hospital discharges among older adults, by electronic prescribing (e-prescribing) rates, 2011.

Characteristics of interest	County e-prescribing adoption rate
Characteristics of interest	All counties	High adoption (⩾35%)	Low adoption (⩽34%)	p-value, high vs low
All discharges	100.0	53.3	46.7
Number of discharges	2,725,378	1,451,544	1,273,834
Patient
Male^a	56.0	55.3	56.9	<0.0001
Age group (years)
65–74	38.5	39.3	37.5	<0.0001
75–84	37.5	37.6	37.5	0.1903
85+	24.0	23.1	25.0	<0.0001
Race^b
White	75.6	82.9	67.4	<0.0001
African American	10.7	7.2	14.8	<0.0001
Hispanic	7.1	3.8	10.9	<0.0001
Other	2.6	2.2	2.9	<0.0001
Number of chronic conditions, mean	6.6	6.6	6.6	<0.0001
Length of stay (days), mean	5.3	5.2	5.5	<0.0001
Adopters/providers
Primary care physicians (per 1000 county residents)
0.00–0.57	28.2	26.9	29.7	<0.0001
0.58–0.71	24.0	31.8	15.0	<0.0001
0.72–0.84	23.0	17.5	29.3	<0.0001
0.85–1.85	24.8	23.8	26.0	<0.0001
Healthcare infrastructure
Hospital beds (per 1000 county residents)
0.00–2.21	25.3	25.8	24.7	<0.0001
2.22–2.89	24.0	33.3	13.5	<0.0001
2.90–3.43	24.8	20.0	30.2	<0.0001
3.44–21.4	26.0	31.6	21.0	<0.0001
No hospital in county	1.3	1.8	0.8	<0.0001
Health Professional Shortage Area (HPSA)
Whole	30.7	33.3	27.6	<0.0001
Partial	63.0	61.3	64.9	<0.0001
None	6.4	5.4	7.5	<0.0001
Community
Rural	13.1	13.3	12.9	<0.0001
Population living in poverty
0.0–11.8	23.5	20.1	27.4	<0.0001
11.9–15.4	24.2	28.4	19.4	<0.0001
15.5–17.8	24.1	26.1	21.9	<0.0001
17.9–36.0	28.2	25.5	31.3	<0.0001
Population with less than a high school education
0.0–7.1	23.9	23.1	24.9	<0.0001
7.2–8.7	23.5	24.8	22.0	<0.0001
8.8–10.2	24.0	31.2	15.9	<0.0001
10.3–24.7	28.6	21.0	37.2	<0.0001
State
AR	5.0	7.3	2.4	<0.0001
AZ	9.0	15.7	1.4	<0.0001
FL	36.2	37.1	35.1	<0.0001
MD	8.2	7.6	8.8	<0.0001
MI	17.1	12.5	22.3	<0.0001
MS	3.4	2.1	4.8	<0.0001
NJ	13.9	7.4	21.3	<0.0001
VT	0.5	0.7	0.2	<0.0001
WA	6.9	9.7	3.8	<0.0001

All differences were assessed using Wald chi-square tests.

A total of 22 discharges were missing sex due to missing information.

A total of 108,902 discharges were missing race due to missing information.

A total of 6400 of 2,725,378 (0.23%) discharges were due to ADEs that occurred in the community. The unadjusted proportion of all hospitalizations among persons 65 and older that included a community ADE was higher for counties with high e-prescribing adoption rates (0.25 versus 0.22; see Table 2). Other factors that were associated with proportionately more ADE hospitalizations included being female, between the ages of 65 and 74 years, White, living in a county with one or more hospitals, living in a HPSA or urban county, and residing in Washington, Arizona, or Florida.

Table 2.

Percent of older adult discharges that included a community ADE, by electronic prescribing (e-prescribing) rates, 2011.

Characteristics of interest	County e-prescribing adoption rate
Characteristics of interest	All counties	High adoption (⩾35%)	Low adoption (⩽34%)	p-value, high vs low
All discharges (n = 2,725,378)	0.23	0.25	0.22	<0.0001
Patient
Sex^a				0.5907
Male	0.20	0.21	0.20
Female	0.26	0.27	0.24
Age group^b (years)
65–74	0.31	0.33	0.30	0.0473
75–84	0.21	0.21	0.20	0.2015
85+	0.15	0.16	0.14	0.2813
Race
White	0.24	0.25	0.23	<0.0001
African American	0.22	0.22	0.22	<0.0001
Hispanic	0.19	0.24	0.17	<0.0001
Other	0.23	0.22	0.24	0.0023
Adopters/providers
Primary care physicians (per 1000 county residents)
0.00–0.57	0.22	0.23	0.21	0.0020
0.58–0.71	0.25	0.25	0.24	<0.0001
0.72–0.84	0.25	0.27	0.23	<0.0001
0.85–1.85	0.23	0.24	0.21	0.2354
Healthcare infrastructure
Hospital beds (per 1000 county residents)
0.00–2.21	0.23	0.24	0.21	0.1470
2.22–2.89	0.24	0.24	0.24	<0.0001
2.90–3.43	0.24	0.26	0.22	<0.0001
3.44–21.4	0.23	0.25	0.23	<0.0001
Number of hospitals in county				0.0006
0	0.20	0.21	0.18
1 or more	0.24	0.25	0.22
Health Professional Shortage Area (HPSA)
Whole	0.25	0.25	0.25	0.0311
Partial	0.23	0.25	0.21	0.3368
None	0.21	0.23	0.19	0.0182
Community
Rurality				0.5914
Urban	0.24	0.25	0.23
Rural	0.20	0.21	0.20
Population living in poverty
0.0–11.8	0.21	0.22	0.21	<0.0001
11.9–15.4	0.25	0.25	0.25	<0.0001
15.5–17.8	0.25	0.27	0.23	<0.0001
17.9–36.0	0.22	0.24	0.21	<0.0001
Population with less than a high school education
0.0–7.1	0.24	0.25	0.22	0.4990
7.2–8.7	0.24	0.25	0.23	0.0193
8.8–10.2	0.26	0.26	0.26	<0.0001
10.3–24.7	0.20	0.21	0.20	<0.0001
State
AR	0.16	0.15	0.19	<0.0001
AZ	0.28	0.28	0.28	<0.0001
FL	0.25	0.26	0.24	0.8254
MD	0.24	0.22	0.26	<0.0001
MI	0.22	0.24	0.21	<0.0001
MS	0.22	0.25	0.21	<0.0001
NJ	0.18	0.19	0.17	<0.0001
VT	0.19	0.18	0.23	0.0767
WA	0.29	0.29	0.29	<0.0001

ADE: adverse drug event.

All differences were assessed using Wald chi-square tests, α = 0.05.

A total of 22 discharges were missing sex due to missing information.

A total of 108,902 discharges were missing race due to missing information.

The multivariate logistic model (see Table 3) estimated the odds of having an ADE hospitalization versus other diagnoses. Because this approach is based on all hospitalizations rather than all persons, our model controls for potential differences in the likelihood of hospitalization associated with regional variation in practice patterns. The first model included only the county e-prescribing rate. It found that residents of a county with a 30- to 34-percent adoption rate had lower odds of having an ADE hospitalization when compared to residents living in counties with an adoption rate of over 48 percent. When patient demographic information was added into the model (model 2), the odds of an ADE hospitalization for residents of a county with a 30- to 34-percent adoption rate were no longer significant. Other factors that were associated with having higher odds of an ADE hospitalization, when taking patient demographics into account, included being female and having additional chronic conditions. When adding the number of primary care physicians per 1000 population into the model (model 3), being female, having additional chronic conditions, and living in a county that have 0.72–0.84 primary care physicians per 1000 population were significantly associated with increased odds of ADE hospitalizations. Model 4 took healthcare infrastructure into account and there were no changes in significant factors. The final model (model 5) included all the previous variables and community factors. Model 5 indicates that the county e-prescribing rates that were in the lowest quartile (0%–29%) resulted in greater odds of an older adult having an ADE hospitalization when compared to counties in the highest quartiles of e-prescribing rates (⩾42%; odds ratio (OR): 1.10 (1.02–1.19)). Additionally, the model indicates that the odds of an ADE hospitalization are higher for females, individuals having more chronic conditions, individuals living in counties with 8.8–10.2 percent of the population having less than a high school education, and individuals living in Florida or Washington.

Table 3.

Adjusted odds of having an adverse drug event hospitalization among older adults, 2011.

Characteristics of interest	Model 1OR (95% CI)	Model 2OR (95% CI)	Model 3OR (95% CI)	Model 4OR (95% CI)	Model 5OR (95% CI)
Technology
Electronic prescribing rate
0%–29%	0.97 (0.91–1.04)	1.00 (0.93–1.07)	1.00 (0.93–1.07)	1.00 (0.93–1.07)	1.10 (1.02–1.19)
30%–34%	0.90 (0.84–0.97)	0.98 (0.91–1.06)	0.94 (0.87–1.01)	0.94 (0.87–1.02)	0.95 (0.87–−1.04)
35%–41%	1.09 (1.02–1.17)	1.09 (1.02-1.16)	1.05 (0.98–1.12)	1.06 (0.98–1.14)	1.07 (0.99–1.17)
⩾42%	Ref.	Ref.	Ref.	Ref.	Ref.
Patient
Sex
Male		Ref.	Ref.	Ref.	Ref.
Female		1.35 (1.28–1.42)	1.35 (1.28–1.42)	1.35 (1.28–1.42)	1.35 (1.28–1.42)
Age group (years)
65–74		Ref.	Ref.	Ref.	Ref.
75–84		0.65 (0.61–0.69)	0.65 (0.61–0.68)	0.65 (0.61–0.68)	0.65 (0.61–0.69)
85+		0.46 (0.43–0.50)	0.46 (0.43–0.49)	0.46 (0.43–0.49)	0.46 (0.43–0.50)
Race
White		Ref.	Ref.	Ref.	Ref.
African American		0.87 (0.80–0.94)	0.85 (0.78–0.93)	0.84 (0.77–0.92)	0.86 (0.78–0.93)
Hispanic		0.78 (0.70–0.87)	0.74 (0.66–0.82)	0.74 (0.66–0.82)	0.75 (0.67–0.84)
Other		0.93 (0.80–1.09)	0.93 (0.79–1.09)	0.93 (0.79–1.09)	0.92 (0.79–1.08)
Each additional chronic condition		1.04 (1.03–1.05)	1.04 (1.03–1.05)	1.04 (1.03–1.05)	1.04 (1.03–1.05)
Each additional day in length of stay		0.92 (0.92–0.93)	0.92 (0.92–0.93)	0.92 (0.92–0.93)	0.92 (0.92–0.93)
Adopters/providers
Primary care physicians (per 1000 county residents)
0.00–0.57			0.92 (0.86–1.00)	0.95 (0.88–1.04)	0.92 (0.84–1.02)
0.58–0.71			1.04 (0.97–1.12)	1.03 (0.95–1.12)	0.90 (0.82–1.00)
0.72–0.84			1.17 (1.09–1.26)	1.16 (1.06–1.26)	0.93 (0.84–1.03)
0.85–1.85			Ref.	Ref.	Ref.
Healthcare infrastructure
Hospital beds (per 1000 county residents)
0.00–2.21				0.94 (0.86–1.03)	0.92 (0.83–1.01)
2.22–2.89				0.96 (0.89–1.05)	0.91 (0.83–1.00)
2.90–3.43				0.99 (0.91–1.07)	0.98 (0.90–1.07)
3.44–21.4				Ref.	Ref.
Number of hospitals
0				0.89 (0.70–1.13)	1.08 (0.84–1.38)
1 or more				Ref.	Ref.
Health Professional Shortage Area (HPSA)
Whole				1.12 (1.00–1.27)	0.91 (0.79–1.04)
Partial				1.08 (0.96–1.21)	0.86 (0.76–0.98)
None				Ref.	Ref.
Community
Rurality
Urban					Ref.
Rural					0.85 (0.77–0.94)
Population living in poverty
0.0–11.8					1.03 (0.93–1.15)
11.9–15.4					1.10 (0.98–1.24)
15.5–17.8					1.07 (0.94–1.21)
17.9–36.0					Ref.
Population with less than a high school education
0.0–7.1					1.05 (0.93–1.18)
7.2–8.7					1.12 (1.01–1.23)
8.8–10.2					1.09 (0.99–1.20)
10.3–24.7					Ref.
State
AR					0.74 (0.61–0.88)
AZ					1.16 (0.99–1.37)
FL					1.21 (1.07–1.37)
MD					Ref.
MI					1.01 (0.89–1.15)
MS					1.12 (0.93–1.35)
NJ					0.84 (0.74–0.95)
VT					0.98 (0.66–1.47)
WA					1.34 (1.15–1.58)

OR: odds ratio; CI: confidence interval.

Bold indicates significantly different from the reference group (α = 0.05).

A total of 108,977 observations were excluded from the model due to missing race variable and/or age variable.

Discussion

The analysis found e-prescribing rates at the county-level to be significantly associated with community-occurring ADE hospitalizations, but not in a simple manner. In unadjusted analysis, higher county-level e-prescribing was associated with a higher proportion of ADE admissions. However, in analysis adjusted for patient characteristics, the odds that an admission would be attributable to a community-occurring ADE were higher in counties with the lowest e-prescribing rate than among those in the highest category. Previous research in inpatient settings, including pediatric hospitals and intensive care units, have found no direct effect on ADEs.^27,31,32,41 Research in other settings, such as a primary care clinic and long-term care facility, have demonstrated improved adherence to best-practice guidelines, but have been unable to demonstrate a reduction in ADE rates.^26,30

Patient characteristics associated with increased odds of ADE hospitalization included being female and having a greater number of chronic conditions, while being older than 75 years, living in partial HPSA or rural areas, being African American or Hispanic, and having longer hospital stays were associated with reduced odds of ADE hospitalizations.

ADE admission rates varied across states, from 0.16 percent in Arkansas to 0.29 percent in Washington. Some differences persisted after adjusting for patient characteristics, with residents of Arkansas and New Jersey being less likely than residents of the referent state (Maryland) to experience an ADE admission, while Florida and Washington states remained elevated. The state variation in odds of an ADE hospitalization cannot be explained by variations in e-prescribing rates or POA reporting experience. In 2011, the states studied varied in e-prescribing rates from lowest to highest in the following order: New Jersey (31%), Mississippi (32%), Maryland (33%), Michigan (36%), Florida (37%), Arizona (42%), Washington (44%), Arkansas (45%), and Vermont (51%).³⁵ Additionally in 2011, Vermont had just begun reporting POA data to HCUP, Mississippi had reported for 2 years, Arizona, Arkansas, Michigan, New Jersey, and Washington had been reporting for 3 years, and Florida and Maryland had been reporting HCUP POA data for 4 years.

Although our findings detected a significant association between e-prescribing and community-occurring ADEs, the results should be interpreted with caution. Rogers’ Diffusion of Innovation theory suggests that the rate of innovation adoption gains momentum after 20 percent and continues to increase at an accelerated rate.⁴² The national rate of e-prescribing increased from 7 percent in December 2008 to 70 percent in April 2014.¹⁹ The mean e-prescribing rate for studied counties in 2011 was 37.9 percent, similar to the national mean of 41.4 percent. Rogers’ theory categorizes adopters during this stage as the “early majority.” Thus, our study potentially included counties comprising only early adopters. In addition, it is possible that e-prescribing rates need to be markedly higher than those observed to have consistent and sizeable effects on ADE occurrence.

Westphal and Shortell⁴³ found that early adopters of total quality management practices were more likely to adapt technologies to improve efficiencies within their organizations, while late adopters were more likely to conform to norms by adopting well-established programs. Studies also demonstrate that the true value of technology is how it is utilized by individuals and organizations.^44,45 These studies demonstrate the need to address organizational and human factor changes to avoid unintended consequences.^46–49 Further studies are necessary to determine whether gains in patient safety indicators continue to be seen with increasing adoption rates in a short- and long-term setting.

Limitations

This analysis may underestimate ADE hospitalization due to the underreporting of ADEs in administrative datasets.^50,51 However, studies have noted the utility of using administrative data in detecting adverse events.^51–54 This analysis was also restricted to nine states, which limits the generalizability of its findings.

We limited this analysis to ADE hospitalizations because they provide insight into the safety of ambulatory prescribing.¹⁰ However, the definition utilized in this study was broad. Future analysis should indicate whether ADE hospitalizations are considered preventable, potential, or non-preventable. This distinction will allow for a better understanding of the performance of technologies such as e-prescribing. Previous studies have found decreases in preventable and potential ADEs.^20,55 In addition, we only examined one HIT, e-prescribing; other technologies have also shown potential in decreasing ADEs and increasing patient safety. Given that e-prescribing can be viewed as an entry-level system when compared to more comprehensive HITs, this study provides a stepping stone to determine the impact HIT adoption is having throughout our nation.⁵⁶ We were also unable to determine the number of medications taken by an individual patient, patient medication adherence rates, and the underlying conditions of the patient in this analysis. However, data indicate that 89.7 percent of older adults have taken one or more prescription medications in the past 30 days.⁵⁷ Finally, we used an ecological measure for e-prescribing rates. This limited our ability to determine whether a patient with an ADE hospitalization was seen at a practice with e-prescribing capabilities.

Conclusion

Our results suggest that residents of low e-prescribing counties have greater odds of experiencing an ADE hospitalization, when compared to residents of high e-prescribing counties. Further research and monitoring that utilizes clinical, prescription, and other administrative datasets is necessary to determine the potential impact of e-prescribing and other health information technologies on patient safety indicators.

Footnotes

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article:

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was partially supported by a Graduate Research Grant from the Office of the Vice President for Research at the University of South Carolina.

References

Institute of Medicine. To err is human: building a safer health system. Washington, DC: Institute of Medicine, 1999.

U.S. Food and Drug Administration. Reports received and reports entered into FAERS by year, http://www.fda.gov/drugs/guidancecomplianceregulatoryinformation/surveillance/adversedrugeffects/ucm070434.htm (2015, accessed 3 October 2016).

Lucado

Paez

Elixhauser

. Medication-related adverse outcomes in U.S. hospitals and emergency departments. HCUP statistical brief #109, 2011. Rockville, MD: Agency for Healthcare Research and Quality, https://www.ncbi.nlm.nih.gov/books/NBK54566/

Weiss

Elixhauser

Bae

, et al. Origin of adverse drug events in U.S. hospitals, 2011. HCUP statistical brief #158, 2013. Rockville: MD, Agency for Healthcare Research and Quality, https://www.ncbi.nlm.nih.gov/books/NBK169247/

Bates

Cullen

Laird

. Incidence of adverse drug events and potential adverse drug events. JAMA 1995; 274(1): 29–34.

Budnitz

Pollock

Mendelsohn

, et al. Emergency department visits for outpatient adverse drug events: demonstration for a national surveillance system. Ann Emerg Med 2005; 45(2): 197–206.

Leape

Cullen

Clapp

, et al. Pharmacist participation on physician rounds and adverse drug events in the intensive care unit. JAMA 1999; 282(3): 267–270.

Von Laue

Schwappach

Koeck

. The epidemiology of preventable adverse drug events: a review of the literature. Wien Klin Wochenschr 2003; 115(12): 407–415.

Willams

. Using medications appropriately in older adults. Am Fam Physician 2002; 66(10): 1917–1924.

10.

Thürmann

. Detection of drug-related adverse events in hospitals. Expert Opin Drug Saf 2003; 2(5): 447–449.

11.

Kaushal

Barker

Bates

. How can information technology improve patient safety and reduce medication errors in children’s health care? Arch Pediatr Adolesc Med 2001; 155(9): 1002–1007.

12.

Bates

Gawande

. Improving safety with information technology. N Engl J Med 2003; 348(25): 2526–2534.

13.

Green

Hawley

Rask

. Is the number of prescribing physicians an independent risk factor for adverse drug events in an elderly outpatient population? Am J Geriatr Pharmacother 2007; 5(1): 31–39.

14.

Boockvar

Liu

Goldstein

, et al. Prescribing discrepancies likely to cause adverse drug events after patient transfer. Qual Saf Health Care 2009; 18(1): 32–36.

15.

Jena

Goldman

Weaver

, et al. Opioid prescribing by multiple providers in medicare: retrospective observational study of insurance claims. BMJ 2014; 348: 1–12.

16.

Medicare Prescription Drug. Improvement, and Modernization Act of 2003 (Public Law 108-173), http://www.gpo.gov/fdsys/pkg/PLAW-108publ173/pdf/PLAW-108publ173.pdf (2003, accessed 3 October 2016).

17.

Medicare Improvements for Patients and Providers Act of 2008 (Public Law 110-275), http://www.gpo.gov/fdsys/pkg/PLAW-110publ275/pdf/PLAW-110publ275.pdf (2008, accessed 3 October 2016).

18.

Title XIII—Health Information Technology for Economic and Clinical Health Act as part of the American Recovery and Reinvestment Act of 2009 (Public Law 111-5), https://www.hhs.gov/sites/default/files/ocr/privacy/hipaa/understanding/coveredentities/hitechact.pdf (2009, accessed 3 October 2016).

19.

Gabriel

Swain

. E-prescribing trends in the United States. ONC data brief no 18, 2008, https://www.healthit.gov/sites/default/files/oncdatabriefe-prescribingincreases2014.pdf

20.

Bates

Teich

Lee

, et al. The impact of computerized physician order entry on medication error prevention. J Am Med Inform Assoc 1999; 6(4): 313–321.

21.

Devine

Wilson-Norton

Lawless

, et al. The impact of an ambulatory CPOE system on medication errors. J Am Med Inform Assoc 2010; 17: 78–84.

22.

Gandhi

Weingart

Seger

, et al. Outpatient prescribing errors and the impact of computerized prescribing. J Gen Intern Med 2005; 20(9): 837–841.

23.

Colpaert

Claus

Somers

, et al. Impact of computerized physician order entry on medication prescription errors in the intensive care unit: a controlled cross-sectional trial. Crit Care 2006; 10: R21.

24.

Evans

Pestotnik

Classen

, et al. Preventing adverse drug events in hospitalized patients. Ann Pharmacother 1994; 28(4): 523–527.

25.

Evans

Pestotnik

Classen

, et al. A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 1998; 338(4): 232–238.

26.

Gurwitz

Field

Rochon

, et al. Effect of computerized provider order entry with clinical decision support on adverse drug events in the long-term care setting. J Am Geriatr Soc 2008; 56(12): 2225–2233.

27.

King

Paice

Rangrej

, et al. The effect of computerized physician order entry on medication errors and adverse drug events in pediatric inpatients. Pediatrics 2003; 112(3 Pt 1): 506–509.

28.

Mullett

Evans

Christenson

, et al. Development and impact of a computerized pediatric antiinfective decision support program. Pediatrics 2001; 108(4): E75.

29.

Pestotnik

Classen

Evans

, et al. Implementing antibiotic practice guidelines through computer-assisted decision support: clinical and financial outcomes. Ann Intern Med 1996; 124(10): 884–890.

30.

Steele

Eisert

Witter

, et al. The effect of automated alerts on provider ordering behavior in an outpatient setting. PLoS Med 2005; 2(9): e255.

31.

Upperman

Staley

Friend

, et al. The impact of hospitalwide computerized physician order entry on medical errors in a pediatric hospital. J Pediatr Surg 2005; 40(1): 57–59.

32.

Van Doormaal

van den Bemt

Zaal

, et al. The influence that electronic prescribing has on medication errors and preventable adverse drug events: an interrupted time-series study. J Am Med Inform Assoc 2009; 16(6): 816–825.

33.

Ancker

Kern

Abramson

, et al. The Triangle Model for evaluating the effect of health information technology on healthcare quality and safety. J Am Med Inform Assoc 2012; 19(1): 61–65.

34.

HCUP Databases. Healthcare Cost and Utilization Project (HCUP)—overview of the State Inpatient Databases (SID). Agency for Healthcare Research and Quality, http://www.hcup-us.ahrq.gov/sidoverview.jsp (2014, accessed 3 October 2016).

35.

Office of the National Coordinator for Health Information Technology. Datasets and documentation. HealthIT.gov, http://dashboard.healthit.gov/data/ (2016, accessed 3 October 2016).

36.

Health Resources and Services Administration. Area Health Resource File, http://ahrf.hrsa.gov/ (accessed 3 October 2016).

37.

Federal Information Processing Standards (FIPS) Codes, https://www.census.gov/geo/reference/codes/cou.html (accessed 3 October 2016).

38.

Encinosa

Bae

. Will meaningful use electronic medical records reduce hospital costs? Am J Manag Care 2013; 19(10 Spec No): eSP19–eSP25.

39.

Surescripts, http://surescripts.com/ (accessed 3 October 2016).

40.

United States Department of Agriculture Economic Research Service. Urban Influence Codes, http://www.ers.usda.gov/data-products/urban-influence-codes.aspx (2003, accessed 3 October 2016).

41.

Evans

Classen

Pestotnik

, et al. A decision support tool for antibiotic therapy. Proc Annu Symp Comput Appl Med Care 1995; 651–655.

42.

Rogers

. Diffusion of innovations. 4th ed. New York: Free Press, 1995.

43.

Westphal

Shortell

. Customization or conformity? Institutional and network perspective on the content and consequences of TQM adoption. Adm Sci Q 2014; 42(2): 366–394.

44.

Prince

Herrin

. The role of information technology in healthcare communications, efficiency, and patient safety: application and results. J Nurs Adm 2007; 37(4): 184–187.

45.

Karsh

B-T

Weinger

Abbott

, et al. Health information technology: fallacies and sober realities. J Am Med Inform Assoc 2010; 17(6): 617–623.

46.

Menachemi

Burke

Brooks

. Adoption factors associated with patient safety-related information technology. J Healthc Qual 2004; 26(6): 39–44.

47.

Damberg

Ridgely

Shaw

, et al. Adopting information technology to drive improvements in patient safety: lessons from the Agency for Healthcare Research and Quality health information technology grantees. Health Serv Res 2009; 44(2 Pt 2): 684–700.

48.

Nanji

Rothschild

Salzberg

, et al. Errors associated with outpatient computerized prescribing systems. J Am Med Inform Assoc 2011; 18(6): 767–773.

49.

Edmondson

Bohmer

Pisano

. Disrupted routines: team learning and new technology implementation in hospitals. Adm Sci Q 2001; 46: 685–716.

50.

Hohl

Kuramoto

, et al. Evaluating adverse drug event reporting in administrative data from emergency departments: a validation study. BMC Health Serv Res 2013; 13: 473.

51.

Cano

Rozenfeld

. Adverse drug events in hospitals: a systematic review. Cad Saude Publica 2009; 25(Suppl. 3): S360–S372.

52.

Bates

Evans

Murff

, et al. Detecting adverse events using information technology. J Am Med Inform Assoc 2003; 10(2): 115–128.

53.

Hougland

Pickard

, et al. Performance of International Classification of Diseases, 9th Revision, Clinical Modification codes as an adverse drug event surveillance system. Med Care 2006; 44(7): 629–636.

54.

Weingart

Iezzoni

Davis

, et al. Use of administrative data to find substandard care: validation of the complications screening program. Med Care 2000; 38(8): 796–806.

55.

Bates

Leape

Cullen

, et al. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA 1998; 280(15): 1311–1316.

56.

Bell

Friedman

. E-prescribing and the medicare modernization act of 2003. Health Aff 2005; 24(5): 1159–1169.

57.

National Center for Health Statistics. Health, United States, 2012: with special feature on emergency care. Hyattsville, MD: National Center for Health Statistics, 2013.

Community-level electronic prescribing and adverse drug event hospitalizations among older adults

Abstract

Keywords

Introduction

Methodology

Theoretical model

Data sources

Study sample

Dependent variable

Independent variable

Control variables

Analytic approach

Results

Discussion

Limitations

Conclusion

Footnotes

Declaration of conflicting interests

Funding

References