Administration and Policy in Mental Health and Mental Health Services Research

, Volume 40, Issue 3, pp 168–178

Implementing an E-Prescribing System in Outpatient Mental Health Programs

Authors

    • Department of Psychiatry, Center for Mental Health Policy and Services Research, Medical CenterUniversity of Pennsylvania
    • School of Social Policy and PracticeUniversity of Pennsylvania
    • Leonard Davis Institute of Health EconomicsUniversity of Pennsylvania
  • Elizabeth Noll
    • Department of Psychiatry, Center for Mental Health Policy and Services Research, Medical CenterUniversity of Pennsylvania
  • Eri Kuno
    • University of Indiana
  • Cynthia Zubritsky
    • Department of Psychiatry, Center for Mental Health Policy and Services Research, Medical CenterUniversity of Pennsylvania
    • Leonard Davis Institute of Health EconomicsUniversity of Pennsylvania
  • Matthew O. Hurford
    • Department of Psychiatry, Center for Mental Health Policy and Services Research, Medical CenterUniversity of Pennsylvania
    • Philadelphia Department of Behavioral Health and Intellectual Disability Services
  • Cordula Holzer
    • Wilmington VA Medical Center
  • Trevor Hadley
    • Department of Psychiatry, Center for Mental Health Policy and Services Research, Medical CenterUniversity of Pennsylvania
Research - Practice Relationships

DOI: 10.1007/s10488-011-0392-6

Cite this article as:
Rothbard, A.B., Noll, E., Kuno, E. et al. Adm Policy Ment Health (2013) 40: 168. doi:10.1007/s10488-011-0392-6

Abstract

This study describes the implementation and evaluation of an electronic prescription ordering system and feedback report in three community-based mental health outpatient agencies and the usefulness of the system in improving psychiatrists’ prescribing behavior. Using the e-prescribing system as a data collection tool, feedback on evidence based prescribing practices for patients diagnosed with schizophrenia spectrum disorder or major affective disorder was provided to agency directors and prescribers via a monthly report. The results of the project were that e-prescribing tools can be installed at a reasonable cost with a short start up period. Although the feedback intervention did not show a significant reduction in questionable prescribing patterns, we should continue to investigate how to best use HIT to improve safety, reduce costs, and enhance the quality of healthcare. A better understanding of what prescribers find useful and the reasons why they are prescribing non-evidenced based medications is needed if interventions of this type are to be effective. Given the availability of administrative claims data and electronic prescribing technology, considerable potential exists to provide useful information for monitoring and clinical decision making in public mental health systems.

Keywords

E-prescribingPublic mental healthEvidence based prescribingClinical decision supportFeedback systems

Background

As the practice of health care becomes increasingly more complex, the incorporation of computer systems into practice settings presents an opportunity to enhance efficiency and effectiveness in all areas of health care. E-prescribing tools are used by clinicians to enhance point-of-care access to medication histories and by organizational entities to monitor prescribing patterns for quality assurance. The use of Health Information Technology (HIT) is of particular relevance in the area of psychiatry where studies reveal substantial divergence from evidence-based prescribing practices by prescribers (Fayek et al. 2003; Lehman et al. 1998; Parks and Surles 2004; Stowell et al. 2009) and high rates of non-adherence to antipsychotic drugs by patients ranging from 20 to 89% (Fenton et al. 1997; Rijcken et al. 2004).

Using health technology to monitor patient and prescriber behavior provides an opportunity to improve quality of care and reduce higher treatment costs due to relapses. Unfortunately, the mental health care field, both private and public, has lagged behind other fields in the area of technology advancement (Hogan 2003). Reported barriers to the use of computer applications include lack of clinician involvement in the planning process, poor implementation planning, and substandard functioning and reliability of the technology (Podichetty and Penn 2004; Miller and Sim 2004). Perceived ease of use, resistance to change, compatibility with workflow, lack of leadership and prior exposure to HIT (Bhattacherjee and Hikmet 2007; Davis et al. 1989; Lapointe and Rivard 2005; Poon et al. 2004), as well as cost considerations, are also seen as a significant obstacle to implementing information technology into practice (Kuno et al. 2007; Spil et al. 2004). Furthermore, despite its face value, surprisingly little empirical evidence exists on the effectiveness of HIT to improve patient care. Consequently, practitioners have been slow to employ HIT in their clinical practice because of limited practical innovations.

Nevertheless, there have been a number of instances where e-prescribing technology has been implemented in psychiatric settings (Siegel et al. 1984; Simon et al. 1993; Finnerty et al. 2002; Essock et al. 2009; Leslie and Rosenheck 2001). Initially, the technology was used in hospital pharmacy settings for detecting adverse drug events as a result of improper drug selections, interactions or incorrect dosages (Chrischilles et al. 2002). An innovative project in the 1980s utilized e-prescribing for identifying evidence-based medication practices in eleven New York State Institutions for Mental Health and Developmental Disabilities. A computerized drug monitoring system was developed to improve hospital prescribing practices by flagging exceptions to a set of user specified guidelines. The results of the study showed that the surveillance techniques improved prescribing practices (Siegel et al. 1984).

In another e-prescribing project, the Veterans Health Administration (VHA) developed an electronic data repository known as “VistA” that incorporates all health information by providers at VHA practice sites for all patient contacts. The “VistA” tool is part of the VHA Quality Enhancement Research Initiative (QUERI), begun in 1998, to facilitate and support implementation of evidence-based practices and improve outcomes and clinical care delivery (Feussner et al. 2000; Rubenstein et al. 2000). Several studies have used the prescription drug records generated from the system to document the quality of pharmacotherapy for individuals being treated in VA facilities (Chen et al. 2000; DeMakis et al. 2000; Dolder et al. 2002; Leslie and Rosenheck 2001; Owen et al. 2004). However, the use of electronic records for decision support at a clinical level is still not widely reported at the VA. This, despite evidence from several studies that suggest feedback on prescribing patterns and patient adherence can change behavior and improve the quality of psychotropic prescribing (Schectman et al. 2004; Stowell et al. 2009; Uttaro et al. 2007).

More recently, a claims monitoring and reporting system known as the Psychiatric Clinical Knowledge Enhancement System (PSYCKES) has been implemented by the New York State Office of Mental Health. PSYCKES uses paid claims prescription data obtained from all outpatient and inpatient providers in publically-funded facilities treating Medicaid clients. Current and historical medication histories as well as best practice reports at the patient, psychiatrist, ward, and facility levels can be accessed by providers via a web based system located in a common data warehouse (Cohen et al. 2004; Essock et al. 2009; Finnerty et al. 2002; White et al. 2004;). A study of psychiatrists who received supervision for specific cases identified by the PSYCKES reports showed reductions in polypharmacy prescribing practices (Uttaro et al. 2007). This monitoring approach does not, however, use an e-prescribing system to document the information on what medication was prescribed during the patient visit. As a result, although gaps in filled prescriptions can be identified retrospectively, patients who never fill their prescription will not be readily identified.

Pharmacy records are increasingly being used in a variety of ways for quality control (QA) purposes to change prescriber behavior around questionable prescribing practices. For example, the Missouri State Office of Mental Health analyzed psychotropic medications using their Medicaid claims data and compared psychotropic medication patterns against a series of nine clinical quality indicators in the form of “questionable practice” (Parks and Surles 2004). Selected prescribers received mailings that included a letter alerting them to possible deviations from best practice, a 90-day pharmacy claims history of any patient in their practice to whom any of the selected indicators applied, a benchmark report comparing their prescribing practice with that of their peers, and a personal medication best practice briefing related to the particular clinical issue. Prescribers were also offered expert consultation in psychopharmacology. Responses to this intervention were reported to be effective in reducing the number of multiple antipsychotic medications prescribed over 6 months (Parks and Surles 2004).

In 2004, an opportunity presented itself to implement and evaluate the use of an e-prescribing system in improving psychiatrist’s prescribing behavior in mental health community outpatient settings. A four year grant funded from tobacco settlement dollars was awarded to the University research group from the Pennsylvania State Department of Health (ME-02-383, SAP 4100010704; 2004–2007). The grant supported the installation of e-prescribing software, as well as administrator and user training in three outpatient mental health agencies that treated seriously mentally ill clients (Kuno et al. 2007). Support was also provided to investigate the effectiveness of prescriber feedback on enhancing evidence based prescribing practices. Thus, this project afforded a chance to examine the usefulness of an e-prescribing system for individually administered clinics that were not part of a large organizational entity such as the VHA or NYOMH. This paper describes the e-prescribing implementation process, the building of consensus on a set of questionable prescribing practices to monitor and the results of providing feedback reports to medical directors on clients with serious mental illness whose prescribing practices were in question.

Methods

The e-prescribing project was conducted in three phases. The first phase involved the implementation of e-prescribing in the designated agencies and the determination of what constituted questionable prescribing practices for persons with serious mental illness, specifically for diagnoses of schizophrenia and major depressive disorder. The second phase involved the development of feedback reports, followed by the introduction of monthly feedback reporting to the medical directors in two of the three agencies with the third agency acting as a comparison group. An evaluation of qualitative and quantitative changes in provider prescribing patterns for feedback and non-feedback agencies comprised the third phase of the project.

To address patient confidentiality risks related to the evaluation, HIPAA agreements were signed with the agencies participating in the project giving the project team the ability to access e-prescribing records to create feedback reports. Study protocols were approved by the research institution’s Institutional Review Board, Committee on Human Subjects. Because this study relied primarily on the use of HIT, the protection against risk applied to protection of data to insure patient confidentiality and to avoid disclosure.

Phase I: System Implementation

E-prescribing order entry systems were put into operation in three outpatient mental health agencies between 2004 and 2005. These agencies were chosen to participate in the project due to the interest of their administration in installing electronic prescribing and their medical director’s desire to improve the quality of prescribing patterns.

The e-prescribing system was web-based and acted as a data collection tool for the evaluation phase of the project as well as an electronic prescribing system. Patient demographic data was populated in a ‘member’ file via an interface from each mental health agency’s administrative data files to the e-prescribing system which periodically read demographic data, mapped the fields to a specified format, and transmitted the data to the agency database. Each agency had its own information channel to interface with its management information system and an interface to provide data to researchers on patient medication order details, physician and patient demographics, and diagnosis details. Prescriptions were printed out and handed to patients or faxed to their pharmacy. The system also provided printable medication brochures for medications that prescribers would give to their patients.

A series of three all-day training sessions where held by the e-prescribing software vendor for prescribers and administrative staff as well as the project staff. Agencies were reimbursed for psychiatrist and other agency staff time by grant funds. Vendor support continued throughout the implementation stages and involved the Management Information System (MIS) staff at each agency as well as those using the system. All agencies were assessed with respect to their technological needs by project staff. Software and hardware was provided by grant funds, as needed, in order to supply the required technological support (Kuno et al. 2007). Both the CEO and medical director of the clinics participated in meetings focused on implementation of the e-prescribing system.

Phase II: Feedback Reporting

While the e-prescribing system was being implemented, potentially problematic prescribing practices were identified by a workgroup comprised of Medical Directors from each agency, community practitioners, academic psychiatrists from the university and the project team, using a variety of methods. First, evidence based prescribing recommendations based on the Schizophrenia PORT study and from the American Psychiatric Association (American Psychiatric Association 2006) and the Texas Medication Algorithm Project (TMAP) were provided to the workgroup to help identify what were considered questionable prescribing practices (Lehman et al. 1998; Texas Department of State Health Services 2007).

Next, retrospective prescription data from Medicaid paid claims records for agency patients was obtained by the project team and examined by the workgroup in order to identify questionable prescribing in their agencies. The workgroup met over a six month period during the first year of the project to develop a list of potential medication problems for which feedback reports would be used (Rothbard et al. 2003).

It was agreed that for patients with schizophrenia, polypharmacy related to the use of multiple (2 or more) antipsychotics, total use concurrently of four (4) or more psychotropics, and the use of sedative-hypnotics such as benzodiazepines would be monitored. The decision to examine benzodiazepines was due to concerns of the prescribers of the potential long-term effects of the medication (e.g. drug dependence, possible adverse effects on cognitive function).

The questionable prescription patterns agreed on for patients with a diagnosis of major depressive disorder were multiple antidepressants; three (3) or more psychotropic’s; the use of an antipsychotic or a benzodiazepine medication. Patients with other diagnoses were not included in the feedback report evaluation as the evidence-based prescribing patterns were considered to be less clear and many of these patients were in treatment for short time periods.

In addition to monitoring evidence-based prescribing practices, clinicians were interested in whether or not there were gaps in prescriptions for a client previously receiving medication. A report on patients who did not have a follow-up prescription in a timely manner was deemed by the workgroup to be a way to identify patients who had stopped taking their medications.

To insure that feedback reports would reflect all patient prescriptions written by agency physicians, the project team used two procedures. First, random checks were done that involved matching chart records with the e-prescribing system to validate that the medication listed in the chart had been entered electronically. Second, monthly physician usage reports produced by the e-prescribing system were monitored by project team to track the volume of prescriptions and number of patients that psychiatrists were writing prescriptions for. This allowed project staff to identify changes in prescribing practice and led to contacting physicians to ask if they had prescribed manually or given out samples that were not recorded. Missing medication records were then entered into the system at each agency.

Two types of reports were constructed: agency specific and client specific. The agency level report was a summary of monthly practices and identified the proportion of individuals whose prescribing records met the questionable criteria established by the workgroup (Table 1). This information was provided to the Medical Director and the project team in evaluating the extent of changes over time in practice patterns. The patient level report was meant to be used by the Medical Director and prescriber in developing a revised prescribing plan that met evidenced based practice. The patient report identified each prescriber and patient with a questionable practice, and included a three month medication history of the identified patients. Additionally, an algorithm was written that identified patients who had no refill of a prior prescription (i.e., within 30 days of the prescription end date) or no new prescription for ones that had been dropped or changed.
Table 1

Sample monthly feedback agency report

Feedback report for (month, year)

Agency name

Diagnosis of schizophrenia

(N = 269)

N

%

Diagnosis of major depression

(N = 113)

N

%

Benzodiazepine

35

13.0

Antipsychotics

49

43.4

Antidepressant

72

26.8

Benzodiazepine

43

38.1

Number of antipsychotics

  

Number of antidepressants

  

 0

4

1.5

 0

6

5.3

 1

222

82.5

 1

76

67.3

 2

42

15.6

 2

27

23.9

 3

0

0.0

 3

4

3.5

Number of psychotropics

  

Number of psychotropics

  

 1

85

31.6

 1

28

24.8

 2

103

38.3

 2

39

34.5

 3

58

21.6

 3

32

28.3

 4

18

6.7

 4

12

10.6

 5

5

1.9

 5

2

1.8

Questionable prescription patterns

N

%

Questionable prescription patterns

N

%

0 Antipsychotics

4

1.5

0 Antidepressants

6

5.3

Benzodiazepines

35

13.0

Benzodiazepines

43

38.1

2 or more antipsychotics

42

15.6

2 or more antidepressants

31

27.4

4 or more psychotropics

23

8.5

3 or more psychotropics

46

40.7

Beginning in 2005, the medical directors in the two feedback agencies began receiving “monthly” reports on questionable prescribing practices. The reports were e-mailed to the clinic medical director by the project team using an encrypted computer program. The medical directors were asked to follow-up with their prescribers in the way they felt was most appropriate for their clinic. In some instances the medical directors discussed the identified cases with the psychiatrists to determine rationale for the prescribing and to check if reports were accurate and appropriate in identifying questionable practices. In other cases, prescribers were asked to provide written rationale to their medical directors for the identified cases and to explain why they were changing or continuing with their practices. Other procedures involved giving the reports to the prescriber and asking them to review the cases on their own. The rationale for not imposing a specific intervention protocol was to test the extent to which the feedback report itself would change physician practices in a typical community setting where fidelity to a particular intervention was not feasible. Regular workgroup meetings were held during the project with the Medical directors to report on progress and problems and to develop steps to improve administrative processes related to the feedback report.

Phase III: Evaluation

Although this project was primarily focused on assessing the process of implementing e-prescribing and understanding its acceptance in a public mental health clinical setting, we used the opportunity to evaluate the usefulness of a feedback report on improving the quality of prescribing. The qualitative evaluation involved the use of focus groups. The quantitative analysis used a pre-post trend study to determine changes in questionable prescribing practices as a result of the feedback report intervention.

Focus Groups

Using a focus group strategy we conducted descriptive analyses examining the implementation process, the level of uptake of the new technology with respect to the number of patients and prescriptions written over the study period, and prescriber satisfaction with using the e-prescribing system. Three focus groups were conducted after 12 months of implementing the prescribing system with a total of twelve psychiatrists representing each of the three mental heath centers. A key question explored was whether or not psychiatrists felt either the e-prescribing or the feedback report (in the case of the two intervention agencies) had improved the quality of patient care. We also explored satisfaction with the content and formatting of the report in the feedback agencies.

Pre-Post Study

Since it is seldom feasible to randomly assign patients within publicly funded agencies to test new interventions, we employed a pre post study design to compare changes in two agencies that received feedback reports with changes in another agency that received only the routine reports generated by the prescribing program. The use of a comparison agency was meant to provide a historical control in the event that changes in prescribing were the result of changes in formulary policy or the introduction of different medications into the market.

Monthly prescribing patterns at three selected time points were examined in the intervention agencies and two in the comparison agency. Time 1 was the baseline and represented a typical month during the first year of the project following start-up and prior to feedback reporting in the intervention agencies; Time 2 and Time 3 represented the same month in years two and three following the intervention. The comparison agency had data for two time periods (Time 1 and Time 2) with no feedback intervention. A third data period was not available given they had implemented the electronic prescribing later in the study period. Each time period had several outcome measures related to questionable prescribing practices. It is important to note that the prescribing records in the time periods may or may not be for the same patients and thus represent a cross sectional or point in time sample. Given the fact that we are looking at changes in provider behavior over time, we expected a reduction in questionable prescribing patterns in the two intervention feedback agencies in time periods two and three and no change in the comparison group.

Study Subjects

Study subjects in the three agencies that implemented e-prescribing met the following criteria: (1) had a serious mental disorder (diagnosis of schizophrenia spectrum disorders- ICD-9 295 or major depressive disorder-ICD-9 296.2, 296.3); and (2) received at least one psychotropic prescription from an agency psychiatrist between January 2004 and December 2006 were selected at each time period. The patients were adults aged 19–64, who were predominately Medicaid eligible individuals and had their medication benefits fully reimbursed through a Medicaid managed care behavioral health program. The agencies involved in the project were large community mental health organizations in an urban site that provided services to enrollees in the managed care program. Their clients are predominantly low-income, African-American and females.

Data

The e-prescribing system provided patient-specific information on diagnosis, covered dates of each prescribed medication and whether a prescription was active, discontinued or a dose changed or a similar class of medication substituted. Information on new patients or those who were no longer receiving treatment was continually updated. This information enabled us to select patients at each of the study time periods that met subject criteria and construct algorithms that determined the number, therapeutic class, dosage, whether or not there was a refill of a prior prescription within 30 days of the prescription end date or if there was no new medication prescribed for those that had been discontinued.

For patients with schizophrenia, the patterns we monitored monthly were the number and therapeutic class of antipsychotics used concurrently and the continuity of use of an antipsychotic over time. For patients with a diagnosis of major affective disorder, we tracked the number and therapeutic class of antidepressants used concurrently and the continuity of use of antidepressants over time. The total number of monthly prescriptions for psychotropic’s and benzodiazepines was calculated for patients with a diagnosis of schizophrenia or major affective disorder.

Prescription records entered into the Infoscriber system were provided to the project staff monthly. Computer programs were run against the prescription records to identify patients with questionable practices, create summary statistics for monitoring trends and to create three months of prescription information to be included in feedback reports to the intervention agencies.

Analysis

A within study group statistical comparison of pre-post changes was employed. We used a two proportion z-score test with pooled sample proportion standard error [z = (P1 − P2)/SE] to determine whether the difference observed between proportions of patients whose prescription regimens were flagged for not meeting selected evidence-based practices was statistically significant from pre to post baseline. Comparisons were made separately for patients diagnosed with Schizophrenia and Major Depressive Disorder in the feedback and the no feedback groups.

Results

Prescriber Satisfaction

The majority of the psychiatrists participating in the focus groups felt that the e-prescribing system had a positive impact on their prescribing practices because it saved time, made it easier for them to read previous medications than hand-written chart notes, and encouraged them to review and update their pharmacologic knowledge. Many felt that communication with their patients increased as a result of the psycho-educational modules of the e-prescribing tool that provided printable medication brochures that prescribers considered more informative than pharmaceutical handouts. Communication with large pharmacies that had reliable fax machines was perceived to be much easier as a result of using an electronic prescribing system because of the quick turnaround time between the prescribing fax and the pharmacies filling of the order and the reduction in phone calls from the pharmacies related to reading the handwriting of the prescriber. Prescribers expressed that communication involving helpful information also increased between the Medical Director and the psychiatrists through the monthly review of prescribing reports that were provided to the study sites that had feedback interventions.

Concerns of the psychiatrists early in the implementation stage involved what they described as the “nature of using the computer during a treatment session”. Psychiatrists felt that dividing their attention between the computer and the patient would be a barrier to maintaining a bond with the patient. This idea was dismissed in the later focus groups as this practice has become more widespread in all areas of clinical practice. The only psychiatrists who disliked using the e-prescribing system were those who had not used computers prior to this study.

Focus group participants from the two feedback groups suggested some minor formatting revisions to the reports such as summarizing the patient-level detail portion of the report to make it shorter, and providing detail only for patients with a continuing history of problem prescribing rather than for all patients with problems in a particular month. Prescribers and medical directors agreed that monthly reporting was adequate and manageable.

E-Prescribing Utilization

Three months following implementation of the e-prescribing system, psychiatrists were using the new technology to write prescriptions for the majority of patients. The annual number of patients receiving prescriptions electronically in Agency I went from 322 in the first year to 675 individuals with a diagnosis of schizophrenia in the third year; for those with major depression it went from 137 to 269 individuals. The increase after year 1 was the result of using the e-prescribing system for patients at that agency’s new outpatient clinic site. In Agency II, the annual number of patients receiving prescriptions electronically remained fairly constant after year 1 (387 in the first year and 322 patients in the second year for schizophrenia and 331 to 309 for major depression). In Agency III, the comparison site, the number of patients with a diagnosis of schizophrenia dropped between year 1 and 2 from 499 to 343 and increased from 1,131 to 1,498 for patients with depression. Twenty-six (26) psychiatrists from the three agencies were involved in the prescribing process during the study period.

Changes in Prescribing Patterns

Figure 1 depicts trends in selected questionable prescribing patterns for the feedback and no feedback groups by diagnosis and prescribing practice. For the feedback group, the measures of polypharmacy for patients with a diagnosis of schizophrenia remained stable or increased less than 2%. We observed somewhat greater increases in the no-feedback group, most notably in psychotropic polypharmacy which increased by 7.1%. Prescriptions for benzodiazepines increased in both groups over the post period but less so for the feedback group (2.2–2.5%) than for the no feedback group (6.8%).
https://static-content.springer.com/image/art%3A10.1007%2Fs10488-011-0392-6/MediaObjects/10488_2011_392_Fig1_HTML.gif
Fig. 1

Questionable prescribing patterns for feedback versus non-feedback agencies

Polypharmacy prescribing patterns fluctuated for patients with major depression in the feedback agencies, increasing from Time 1 to Time 2 and then falling below Time 1 levels at Time 3. Prescribing patterns for benzodiazepines showed a consistent downward trend decreasing between 1.6 and 3.4% in the feedback group over the two year post period versus an increase of 4.9% in the no-feedback group over the 1 year post period. The changes in prescribing patterns were, however, not statistically significant across time in either the feedback agencies or the no feedback agency (Table 2).
Table 2

Questionable prescribing patterns for feedback versus no feedback by diagnosis

Feedback agencies

t1

(N = 453)

t2

(N = 518)

t3

(N = 479)

P value

t2–t1

P value

t3–t1

Schizophrenia spectrum disorders

     

Two or more antipsychotics

19.2

19.7

21.1

0.848

0.475

Four or more psychotropics

13.2

13.3

13.6

0.972

0.885

Any benzodiazepines

9.7

12.2

11.9

0.224

0.283

Major depressive disorders

     

Two or more antidepressants

31.9

35.7

29.6

0.476

0.653

Three or more psychotropics

34.7

38.5

31.4

0.487

0.544

Any benzodiazepines

31.2

29.7

27.8

0.758

0.495

No feedback agencies

t1

(N = 215)

t2

(N = 333)

 

P value

t2–t1

 

Schizophrenia spectrum disorders

     

Two or more antipsychotics

16.7

17.4

 

0.838

 

Four or more psychotropics

20.5

27.6

 

0.058

 

Any benzodiazepines

18.1

24.9

 

0.062

 

Major depressive disorders

     

Two or more antidepressants

26.6

27.6

 

0.695

 

Three or more psychotropics

51.0

55.4

 

0.133

 

Any benzodiazepines

41.4

46.3

 

0.090

 

Tests: two proportion z-score test with pooled sample proportions [z = (P1 − P2)/SE]

Discussion

Notwithstanding the widespread interest in HIT, the implementation of electronic technology throughout the healthcare industry has been slow, primarily limited to inpatient facilities and organizational settings involving a single payer. Currently e-prescribing technology is underutilized in the public sector with the applications mostly confined to improving patient safety and billing procedures rather than on improving best practice guidelines (Ash et al. 2004; Burt and Hing 2005).

Our findings suggest, however, that e-prescribing tools can be installed efficiently and can provide information for monitoring and clinical decision making in dispersed systems as well as in a single large system such as the VHA or New York’s Department of Mental Health, which have realized considerable success in electronic implementation.

The successful implementation of electronic prescribing in the project agencies was due in part to reducing the barriers noted in the literature. Issues such as the lack of clinician involvement in the planning process, poor implementation planning, and substandard functioning and reliability of the technology were addressed from the onset. Technical support from the vendor was also readily available to users. In addition, the question of cost, identified as a significant obstacle to implementing information technology, particularly in not-for-profit agencies, was also dealt with. Grant funds were used to enhance the MIS systems of the agencies and pay for training and program costs during the four years of the project. We also did a cost analysis of implementing the e-prescribing system for future reference by other agencies (Kuno et al. 2007). We found the start up and operating expenditures for an outpatient agency with 10 FTE psychiatrists in 2005 were $27,549 in the first year. This included pre-implementation consultation, technology and system integration and training costs. The annual ongoing maintenance cost, at the time of the study, was estimated at $14,677 for that size agency, which is a modest expenditure.

Further implementation of e-prescribing may occur through government sponsored financial incentives. The intent of the HITECH Act, which is part of the American Recovery Act (2009), is to provide financial incentives to physicians and medical facilities to implement electronic health records which includes e-prescribing. Unfortunately, the legislation does not currently cover behavioral health clinics, although it does allow individual psychiatrists to apply for funding (American Recovery and Reinvestment Act of 2009, Title XIII—Health Information Technology, Subtitle B—Incentives for the Use of Health Information Technology). This issue is currently being addressed in Congress and should be resolved in a way to provide behavioral healthcare clinics similar financial incentives to purchase e-prescribing software in the near future. Reports, such as the one produced for this study, could easily be constructed and would enable behavioral health providers to meet criterion for “meaningful use” of technology, which is a required condition for receiving financial reimbursement.

Even if implementation and cost issues could be easily resolved, insuring the effective use of the electronic prescribing system has its own set of challenges, particularly in the public sector environment. The community mental health system is characterized by high provider and patient turnover rates. Agencies must ensure ongoing training for new prescribers. New patients and their demographic information must be continually updated so that all prescriptions are entered into the system. Conversely, patients who are no longer in treatment need to be removed so they do not generate alerts for questionable prescribing practice such as gaps in prescriptions. This requires that management information systems be kept current and relevant data items need to be linked to the e-prescribing tool. Also, decisions need to be made on how to enter sample medications to insure a comprehensive medication history. Most e-prescribing tools allow medications prescribed by outside providers (e.g., the primary care provider or medical subspecialist) to be entered into the system. This feature enables the tool to check for potential drug–drug interactions and helps agencies comply with regulatory requirements for maintaining a complete and reconciled medication list for each patient. This upkeep requires agency effort and is more difficult for independent clinics than large delivery systems like the VA which likely have better support and infrastructure. Without the proper maintenance, reports will become cumbersome and unreliable and will lose their credibility with prescribers as a useful decision support tool.

The lack of a significant impact in reducing questionable prescribing patterns in our feedback agencies could be due to certain study limitations. First, there may have been selection bias in that the agencies that volunteered to participate in the feedback study were those with a higher quality of care with fewer questionable cases to change. Another limitation that requires further examination is that of incomplete medication history due to patients receiving care in another setting that is not linked to the e-prescribing software. Coordinating of electronic health information across payer networks is needed to remedy this issue (Grossman et al. 2007). This undertaking requires the participation of e-prescribing technology vendors, hospitals, and physician practices and should improve in the future as electronic records increase in usage.

Finally, a more rigorous intervention design using a larger sample of agencies and a more structured feedback process may be required to fully test the hypothesis regarding the effectiveness of the approach. The most promising strategies for changing provider prescribing behavior, according to Soumerai (1998), involve educational outreach, continuous reminders and audit feedback strategies. Academic detailing, involving a trained person known as an opinion leader who meets with providers in their practice setting and provides specific feedback designed to change provider behavior, has been effective as well in increasing the use of guideline-based practices (Smith 2000). Also, constructing interventions that involve payer incentives to change practices may provide a better method of changing provider behavior using the feedback information available from e-prescribing. Furthermore, an analytic approach using a between group analysis for future studies would be more robust in determining the differences between groups. Unfortunately, it was not feasible in this project due to data limitations (e.g. lack of records on all patients who met criteria prevented us from performing inferential statistics on differences over time between study groups).

Conclusion

Our experience with implementing electronic prescribing provided evidence that prescribers in outpatient public sector mental health settings were able to quickly use the technology regularly in their clinical prescribing practices. The agencies participating in our study are typical of large urban public mental health clinics with sizeable caseloads and limited psychiatric time suggesting the feasibility of broadening the application of e-prescribing technology in the public sector. Given the availability of administrative claims data and electronic prescribing technology, interventions using feedback mechanisms have considerable potential for clinical support. Several studies using Medicaid claims data for monitoring and reporting purposes (PSYCKES) have in fact shown improvements in reducing polypharmacy practices. They also provide alerts to prescribers regarding co-morbid conditions of patients (e.g., diabetes and hypertension) in order to improve the type of antipsychotics prescribed for individuals with schizophrenia (Kealy 2009; Finnerty 2009).

It is important that we continue to investigate how to best use HIT to improve safety, reduce costs, and enhance the quality of healthcare. However, a better understanding of what prescribers find useful and the reasons why they are prescribing non-evidenced based medications is needed if interventions of this type are to be effective.

Copyright information

© Springer Science+Business Media, LLC 2012