PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of phimLink to Publisher's site
 
Perspect Health Inf Manag. 2008; 5: 12.
Published online 2008 September 2.
PMCID: PMC2533117

A Descriptive Analysis of a Nursing Home Clinical Information System with Decision Support

Gregory L. Alexander, PhD, RN, assistant professor

Abstract

Clinical information systems are absent in most nursing homes. Therefore, the vast numbers of elderly people who reside in the facilities are not reaping the benefits that these systems are thought to have, including better management of chronic conditions, greater efficiencies, and improved access to information. It is important for early adopters of nursing home information systems to share experiences to foster implementation of other systems and improve design and evaluation of these systems. The purpose of this study is to describe the activity of clinical decision support alerts and associated triggers during an analysis of an information system implemented in three nursing homes. The proportion of triggers used for each active alert in each alert category is described across resident diagnoses within the nursing home decision support system during six months of data collection.

Key Words: Nursing home, technology, clinical decision support, quality of care

Introduction

Leading healthcare organizations have stressed the importance of integrating information technology (IT) into healthcare systems to improve provider practices, increase the quality of patient care, and reduce medical errors.1,2 One missing link plaguing the deployment of technology into healthcare is how to incorporate practical, usable clinical information systems into the work environment of providers.3 This missing link is most evident in the organizational structure of nursing homes.4

Clinical IT systems are absent in most nursing homes; however, a few homes are implementing highly sophisticated technology into their workflow to support resident management, enhance clinical support services, and improve administrative oversight.5,6 Describing the experiences of early adopters of clinical IT provides detail about successes and failures for those who adopt these systems later. Sharing this knowledge results in convergence of IT design and practice, better management and rollout of information systems in these settings, more effective IT leadership, and better-defined strategies for technology evaluation, system use, and outcome measurement.7

Improvements in nursing home care quality cannot be expected until IT is implemented that provides accurate measures of the actual care provided to residents.8 Clinical data stored in an information system assists us to define the type of care documented and perhaps delivered to residents of the nursing home. Analyses of alert and trigger frequencies in decision support systems have important implications. For example, triggers or alerts with higher frequencies would be an important focus during implementation if limited resources were available to address the alerts; additionally, alerts or triggers associated with more positive resident outcomes would also be an important focus for quality and risk management oversight. The purpose of this study was to describe the activity of clinical decision support alerts and associated triggers using data gathered during an analysis of an information system called OneTouch Technology (now Optimus EMR, Inc.) that was implemented in three nursing homes. The proportion of triggers for each active alert in each alert category is described across resident diagnoses within the nursing home decision support system during six months of data collection.

Within this study, an alert is defined as a clinical decision support mechanism that automatically sends an automated message to an end user; the trigger is the mechanism, (including logic) that generates the alert sent to the end user as notification of a potential patient problem. Figure Figure11 illustrates the relationships between the alerts (automated messages) and triggers (set of rules that negotiate clinical assessments) that generate clinical alerts if a rule is satisfied. Alert categories included the following:

  • dehydration
  • constipation
  • skin integrity
  • weight loss
  • weight gain
  • decline in condition
  • improvement in condition
Figure 1
Data Flow Diagram of Nursing Home Clinical Information System with Decision Support

A trigger is a predetermined criterion within the resident assessment data that is used to make a clinical alert active. Table Table1 1 describes the clinical alerts and associated triggers within the resident assessment data in the OneTouch system at the time of this evaluation. Triggers are used by clinicians to identify new resident conditions or changes in existing conditions. In response, alert frequencies may rise if triggers are repetitively used or fall if triggers are not selected by the staff. Rising alert frequencies can lead to alert fatigue, resulting in decreased user responsiveness to the alerts.

Table 1
Alert Calculations

In this study, alerts and triggers can indicate either positive (improvement in condition) or negative (decline in condition) resident changes. For example, in this decision support system, a dehydration alert is sent to a provider if an episode of bowel incontinence, fever, insufficient fluid intake, or emesis is reported within a 24-hour period (Table (Table1).1). Additionally, a dehydration alert is issued if 25 percent or more of food is left uneaten at meals. In contrast, some alerts in this system are based on significant changes documented by providers on successive clinical assessments rather than episodic events. A decline in condition alert is sent to an end user if the ability to make decisions declines over two successive minimum data set (MDS) assessments (see Table Table1).1). MDS assessments are specialized assessments required to be performed at admission, quarterly, and in the event of a significant change in condition for any nursing home resident.

Research Question

Clinical IT in nursing homes is expected to facilitate management of chronic conditions, support the delivery of effective patient care, and improve efficiency by avoiding waste of equipment, supplies, and energy.9 Potential uses of IT in nursing homes include the use of clinical decision support mechanisms at the point of care to make the most appropriate treatment options available for providers, increase accessibility of evidence-based protocols specific for nursing home residents, enhance data collection, and allow more rigorous analysis of outcomes.10 In this study, investigators sought to determine the frequency and types of triggers in active alerts in a clinical decision support system for residents with specified diagnoses during six months of data collection. The research question was as follows:

What is the proportion of triggers for each active alert in each alert category within and across resident diagnoses?

Background

A resident in a nursing home receives care from a multitude of providers including physicians, nurses, nurse's aides, medical technicians, physical therapists, dieticians, and others. All of these professionals provide necessary treatments, education, and assessments that are recorded in medical records. The motivation for developing clinical information systems in these settings is to allow multiple stakeholders to have access to patient information from anywhere in order to make the best possible decisions about resident care. The goal of the clinical information system with decision support is to facilitate multiple, coordinated, asynchronous interactions between providers while enhancing decision making through better artificial intelligence; to coordinate resident data from all the sources listed; and to organize the data to facilitate explicit decision support for computer-mediated evidence-based care.11

The data flow diagram in Figure Figure11 shows the relationships between the entities (persons, places, things, and events) in a nursing home clinical information system with decision support. A data flow diagram groups data into entities, or objects that exist and are distinguishable from other objects, lists the relationships between those entities, and illustrates how data is maintained and stored.

The rectangles in Figure Figure11 represent names of entities. The data flow, or the movement of data collected by each entity, is illustrated by the solid line between two entities; the name of the data flow is given above the line. In this study, the frequency of documentation of clinical assessment data, triggers, and alerts is evaluated. The staff provides important clinical data about residents that populate the electronic fields in the resident assessment. When the data is entered into the clinical information system, the internal decision support system negotiates the rules table, comparing the clinical data to evidence-based protocols that are part of the rules table. When the clinical data and evidence-based protocols don't match, an alert with a time and date stamp notifies the staff of the change in condition. In theory, these alerts should then prompt the staff to take action, which should be documented in care plans, daily progress notes, task lists generated by the staff to direct care delivery, interventions performed by staff members, and progress made toward resident goals.

Method

Selection and description of participants

The three nursing homes in this study were participants in a more comprehensive study to investigate the use of a point-of-care clinical information system (provided by the OneTouch Technology) to demonstrate how IT can improve quality of care.12 All procedures were approved by the university's Institutional Review Board before the study began. Data obtained during this study included all patient care data recorded by the nursing home staff in each of the three facilities, starting six months after implementation and ending 12 months after implementation. Facility and resident characteristics are shown in Table Table22.

Table 2
Nursing Home Quality Measures (QM)

The evaluation study used a stratified purposive approach to recruit facilities. Facility size was taken into account during the recruitment phase. Additionally, nursing home facilities representing profit, not-for-profit, and governmental ownership structures were represented. Resident characteristics of the facilities around the time of the study are described using Nursing Home Compare data downloaded from the Centers for Medicare and Medicaid Services national database during the first quarters of 2004 and 2005, as shown in Table Table22.

Nursing home technology

The OneTouch system provided point-of-care technology that enabled healthcare providers to access and enter resident information outside of confined nursing stations. Few nursing homes have incorporated integrated clinical information systems into their clinical practices to support clinical practice, enhance clinical support between different providers (laboratory, physical therapy, social services, etc.), and provide better administrative oversight.13 The point-of-care modules in the OneTouch system integrated specialized technology that facilitated electronic tracking of resident care, provided personal digital assistants for data entry, and used wireless remote technology to enable wider access to patient data and decision making. Benefits and limitations of this technology have been described previously, including the ability to track resident care back to providers and the ability to view resident care from many places in the facility.14 At the time of this study, integrated systems used by the nursing home staff included clinical alerts, provider-to-provider messages, nurse assistant task lists, care plan items, treatments, electronic medication administration, and more.

Clinical alerts and triggers

Automated clinical alerts in the OneTouch system assist in identifying when a resident might be experiencing constipation, dehydration, a skin integrity change, weight loss, weight gain, or other changes in condition. Each alert mechanism has a specific alert calculation. Alert calculations incorporate triggers selected by the staff during a resident assessment that defines current clinical conditions (see Table Table1). 1). Triggers are identified and alerts become active when data collected by the nursing home staff at the point of care are combined with detailed data elements from resident assessment data into a relational database. Immediate access to this database through electronic information displays and system reports can be used to manage resident care activities more quickly and efficiently.

Data collection procedures

Resident data were collected during the six-month period noted in Table Table22 at each of the nursing homes. The data from each of the nursing homes were received on a preformatted hard disk and loaded into an Access (v. 2003) database for analysis. Before data were provided for analysis, all resident and facility identifying information was removed; fictitious unique facility and resident identifiers were assigned in the dataset to replace nursing home and resident names.

Statistics

The OneTouch dataset was queried daily for six months to identify the types of alerts activated and triggers selected for each facility. While controlling for specific categories of alerts and the primary resident diagnosis, descriptive statistics were used to identify trends in trigger frequencies for active alerts in each alert category during the six-month period. Each alert category and patient diagnosis was given a unique dummy-coded variable to allow them to be manipulated in the dataset. The frequencies and types of triggers associated with active alerts for each diagnosis type were determined and will be reported here. All statistical analysis was performed using SPSS version 14.0 and Excel.

Results

Alert activity at the facility level

Initial daily queries of active alert data were evaluated to observe patterns at the facility level. The investigator assumed that there would be significant variability among daily alert frequencies based on the diversity of care delivered to residents, changes in chronic conditions, and complexity of care.15 However, contrarily, little to no variation in alert frequencies was noted during the analysis period in all three facilities.

In Facility A, during the first 10 days no active alerts or triggers were documented for dehydration, decline in condition, weight loss, or weight gain; conversely, on a consistent daily basis there were 136 constipation alerts, three skin integrity alerts, and 40 improvement in condition alerts. Similarly from day 21 of the fourth month to the last day of the sixth month, all alert counts were the same. From day 11 of the first month to day 20 of the fourth month, there was variable activity in all alert categories.

There was frequent variability in Facility B beginning day one of the first month and continuing to day 25 of the second month, when suddenly all daily alert totals for each category became consistently the same. In Facility C, total active daily alerts remained consistent during the entire six-month period of data collection. In facility C no active alerts occurred for the entire period for skin integrity, dehydration, or decline in condition. Consistently, daily active alert totals were 83 and 18 throughout the six months.

As a result, only active alerts and associated triggers from day 11 of the first month to day 20 of the fourth month in Facility A and day 1 of the first month to day 24 of the second month in Facility B were included in this analysis, for a total of 155 days of alert activity. Since no variability in the alerts was noticed in Facility C, data from that facility were not used. For more results and discussion of alert frequencies and potential reasons for these findings, see other studies by this investigator.1618 The remainder of this section discusses the trigger frequencies associated with these active alerts and the types of resident diagnoses they were associated with.

Trigger frequencies by resident diagnosis

A total of 172 residents with eight primary diagnoses were included in this analysis (see tables tables3310). The largest category (52 percent) included 89 residents who had ventilation pneumonitis as their primary diagnosis. These residents had nearly 17,000 triggers selected by the staff during the 155 clinical days considered in this evaluation (see Table Table3).3). The most frequent trigger within this category was related to the amount of food left uneaten, which resulted in an active dehydration alert. The proportion of food left uneaten was vigilantly documented by the staff; this trigger, which was associated with the dehydration alert, was in the top two of all triggers for every resident diagnosis. Its use ranged from 39.73 in residents with ventilation pneumonitis (Table (Table3)3) to 21.92 percent in residents with osteoarthritis (Table (Table99).

Table 3
Residents with Ventilation Pneumonitis (89)
Table 9
Residents with Osteoarthritis (7)
Table 10
Residents with Pneumonia (7)

A large number of residents in the sample had neurological issues. Residents with Alzheimer's disease (Table (Table4)4) represented 7.5 percent of cases. Dietary intake, continence, positioning, and emotional level were some of the most frequent triggers selected for this group of residents. Other important triggers, although selected less frequently in this class of residents, were the changing level of locomotion on and off the unit and the ability to walk in corridors.

Table 4
Residents with Alzheimer's Disease (13)

Eighteen residents in this study had experienced cerebral vascular accidents (CVAs) (Table (Table5).5). Amount eaten at meals (28.79 percent), urinary continence (21.58 percent), and hydration status (5.87 percent) were the most frequently selected triggers for residents with CVAs. Twenty residents with a dementia diagnosis (Table (Table6)6) had 9263 triggers selected during the study period; 29 percent were for bladder incontinence, which related to the skin integrity alert. Bladder incontinence was also the most frequently selected trigger (26.89 percent) in residents with osteoarthritis (Table (Table9).9). Bladder incontinence is another assessment parameter that seems to be monitored closely. Bladder incontinence was the second most frequent trigger selected in residents with Alzheimer's disease, hypertension, cerebral vascular accidents, pneumonia, and depressive disorders.

Table 5
Residents with Cerebral Vascular Accident (18)
Table 6
Residents with Dementia (20)

A smaller number of residents (seven) in this study were experiencing a depressive disorder (Table (Table7).7). Again, dietary intake at meals and bladder incontinence were observed most frequently by the staff for residents with this diagnosis. The staff also documented other individualized assessments in this group of residents, including the ability to make decisions, performance of personal hygiene measures, and increased or decreased dressing, which all related to the clinical alert of improvement or decline in condition.

Table 7
Residents with Depressive Disorder (7)

A wider range of assessments were selected by the staff for the 11 residents who had a primary diagnosis of hypertension (Table (Table8).8). Although the majority of the triggers indicated that residents left their meals uneaten and were incontinent, other important information documented for these residents included locomotion, walking and transferring activities, changes in behavioral symptoms, amount of toilet use, and ability to make decisions.

Table 8
Residents with Hypertension (11)

The most frequent triggers selected in the residents with osteoarthritis were related to bladder and bowel incontinence, the maintenance of turning and repositioning programs, and mobility (Table (Table9).9). Emotional levels, personal hygiene, and ability to make decisions were also part of these residents' individualized assessments.

For the residents with pneumonia (Table (Table10),10), dietary intake, continence, mobility, and maintenance of proper positioning accounted for the majority of computerized triggers selected. Less frequently, assessments regarding changes in their overall condition and self-care abilities were documented.

The percentages of weight gain and weight loss were documented less frequently for residents under each diagnosis. Weight loss and gain was documented in the system as a percentage of total weight. Weight loss of 3.5 percent of the resident's total body weight over 30 days or 7 percent over 180 days resulted in an active alert. Weight gain of greater than 3.5 percent of the resident's total body weight over 30 days or 7 percent over 180 days resulted in an active alert.

Discussion

Anecdotal evidence during the evaluation and implementation of the decision support system indicated that the nursing home staff accepted the technology, although some staff members did not.19Important factors shaping use of the system included staff perception of how the system functioned, resources available to assist with trouble-shooting of system problems, equipment availability, and preparation for the process change. In regard to system functioning, staff members indicated they were not always aware of where alert messages existed in the system and how to address them once they found them. For example, the staff members were not able to tell which triggers were responsible for activating an alert; therefore, they did not know how to respond appropriately to an active alert. Resources and equipment availability were important for maintaining integrity of the data. Staff members indicated if there was not enough equipment, they didn't document care given. Finally, the staff was frustrated and confused by the lack of standardization of terminologies during the conversion process, which in some facilities resulted in necessary care being omitted. In decision-support systems, accuracy of alerts is dependent on data integrity, responsiveness to resident needs, and consistent documentation about resident assessments and care delivered. Therefore, it is vital to conduct ongoing evaluation of these systems to determine the effect of usage on resident care outcomes.

By far, the most frequently documented assessment parameters were related to dietary intake, continence, hydration status, and maintenance of positioning programs. Given the percentages of residents and their conditions reported in the Nursing Home Compare dataset for 2004 and 2005 (Table (Table2), 2), finding greater documentation of these types of assessments is not really that surprising. During the time this study was conducted, a significant number of incontinent residents were present, including low-risk residents who lost control of their bowel or bladder (30–80 percent), while very few residents (1–10 percent) had a catheter inserted and/or left in their bladder. Another contributing factor was that large numbers of residents experienced mobility problems and required more help performing their activities of daily living (see Table Table2).2). Residents with these types of needs require more help getting to and from meals that provide adequate nourishment and prevention of dehydration.

The consistent maintenance of a positioning program to ensure that skin remained intact was also frequently not documented by the staff, resulting in more frequent skin integrity alerts. One consistent observation across the facilities is the decline of the percentage of high-risk residents with pressure ulcers from year one to year two (Table (Table2).2). The purpose of this study was not to empirically test the effect of clinical information systems on resident outcomes; however, it is certainly worth noting that increased reminders to all staff members at the point of care regarding positioning draws attention to these important programs, resulting in better maintenance and decreased pressure sore development. This would be consistent with findings from other related studies.20

An additional factor that might have contributed to the documentation of positioning was the significant numbers of residents that were being restrained in these facilities, as high as 12 percent (Table (Table2). 2). Restraining residents limits their ability to move about and reposition themselves independently; it also prevents them from getting up and going to the bathroom. Therefore, in facilities that have high numbers of residents that are physically restrained, incontinence, positioning, and hydration will be more frequently assessed and a greater number of triggers will be documented by the staff for these conditions.

The opportunity to develop other types of clinical alerts exists for this clinical decision support system, including the development of an alert and triggers associated with resident pain levels. At the time of this study, no alerts or triggers were being used in the OneTouch system to identify when a resident was experiencing increased pain or relief from preexisting pain. In these facilities 20 to 40 percent of the short-stay residents were experiencing moderate to severe pain; less frequently, the longer term residents were experiencing painful episodes (3–14 percent; see Table Table2).2). The ability to document and improve awareness of resident pain through better decision support would lead to earlier intervention and a better quality of life for nursing home residents.

Limitations

This study has limitations. The small number of nursing homes included in the study, all of which are located in one state, affects generalizability. However, as stated, very few nursing homes have implemented such systems; therefore, the significance of this project lies in describing how early adopters of clinical information systems with decision support are using these systems and how the data can be used to describe resident care.

Another limitation lies in the assumptions made about the clinical information in the decision support system captured in this study. Because of the lack of variability, an assumption was made by the investigators that something was happening in the facilities during those periods affecting the quality of the decision support data, and therefore some periods were not included. Based on anecdotal evidence, causes of this lack of variability included poor clinical documentation resulting in loss of data integrity, lack of existing in-house clinical IT support to educate users about the function of triggers and location of alert messages, and alert fatigue or poor response to alerts caused by constant electronic end-user messaging to the staff. Anecdotal evidence from observations and interviews of staff members at the time of this study indicate that there was concern about these causal factors and their effect on the system.

Conversely, another assumption was made that when there was variable data, the quality of the decision support system was intact. Researchers assumed that the variability of assessments increased as documentation changed between assessments and as effective clinical responses occurred to the procedures performed by care providers. For example, the most frequent trigger chosen (29 percent) was bladder incontinence for residents with dementia in Facility B, while in comparison, the Facility A staff used this trigger much less frequently (6 percent). These assumptions cannot be proven since the investigators were not physically present when all the data were collected. One remedy for this situation was to describe the makeup of the resident population physically present in the facilities using the Nursing Home Compare dataset as discussed. The frequency of documented triggers appears to align with potential problems likely to be experienced by high percentages of residents according to the nursing home quality measures.

Conclusion

The analysis of nursing homes that are early adopters of clinical information systems with decision support provides significant details on how these tools are being used for resident care. These assessments inform developers how providers use the current system and identify potential areas of future development, which leads to greater convergence between system design and practice. Linking the designers and the users of nursing home information systems is a critical step in creating information structures that are practical and usable in these settings.

Sources of Support

This project was supported by grant number K08HS016862 from the Agency for Healthcare Research and Quality (Alexander, PI). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

Notes

1. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academies Press; 2001.
2. Ozbolt J. G., Bulechek G., Graves J. R. Report of the Priority Expert Panel on Nursing Informatics. Bethesda, MD: National Center for Nursing Research, National Institutes of Health; 1993.
3. Sensmeier, J., and C. Delaney. “Nursing Informatics Collaboration Task Force Letter.” 2004. Available at http://www.nitrd.gov/pitac/reports/20040401_amia_himss.pdf.
4. American Health Information Management Association (AHIMA). “AHIMA Responses to LTC Health Information Technology Summit.” 2005. Available at http://www.ahima.org/infocenter/whitepapers/AHIMAResponse_LTCHITSummit-Final.pdf.
5. Alexander, G. L., D. Madsen, S. Herrick, and B. Russell. “Measuring IT Sophistication in Nursing Homes.” Advances in Patient Safety. Agency for Healthcare Research and Quality, 2008 (in press).
6. Alexander, G. L. “IT Sophistication in Nursing Homes.” Long Term Care Interface (2008, in press).
7. Lorenzi N. M., Novak L. L., Weiss J. B., Gadd C. S., Unertl K. M. “Crossing the Implementation Chasm: A Proposal for Bold Action.” Journal of the American Medical Informatics Association. 2008;15(no. 3):290–296. [PMC free article] [PubMed]
8. Schnelle J. F., Bates-Jensen B. M., Chu L., S. F. Simmons. “Accuracy of Nursing Home Medical Record Information about Care-Process Delivery: Implications for Staff Management and Improvement.” Journal of the American Geriatrics Society. 2004;52:1378–1383. [PubMed]
9. Institute of Medicine. Committee on Data Standards for Patient Safety. Key Capabilities of an Electronic Health Record System. Washington, DC: National Academies Press; 2003.
10. McAlearney A. S., Hoshaw S. A. Long Term Care Interface. 2005, April. “The Potential for Handheld Computers in LTC Settings.” pp. 36–40.
11. Nebeker J. R., Hurdle J. F., Bair B. D. “Medical Informatics in Geriatrics.” Journal of Gerontology: Medical Sciences. 2003;58A(no. 9):820–825. [PubMed]
12. Rantz, M. J., J. Scott-Cawiezell, G. L. Alexander, V. Conn, R. W. Madsen, G. F. Petroski, L. Hicks, M. Zwygart-Stauffacher, and RTI International. “Evaluation of the Use of Bedside Technology to Improve Quality of Care in Nursing Facilities.” 2006. Available at http://www.cms.hhs.gov/reports/downloads/Rantz.pdf.
13. 6. Alexander, G. L. “IT Sophistication in Nursing Homes.” Long Term Care Interface(2008, in press).
14. Alexander G. L., Rantz M. J., Flesner M., Diekemper M., C. Siem. “Clinical Information Systems in Nursing Homes: An Evaluation of Initial Implementation Strategies.” Computers, Informatics, Nursing. 2007;25(no. 4):189–197. [PubMed]
15. Rantz M. J., Hicks L. L., Petroski G. F., Madsen R. W., Mehr D. R., Conn V., Zwygart-Staffacher M., Maas M. “Stability and Sensitivity of Nursing Home Quality Indicators.” Journal of Gerontology. 2004;59A(no. 1):79–82. [PubMed]
16. Alexander G. L. “Issues of Trust and Ethics in Computerized Clinical Decision Support Systems.” Nursing Administration Quarterly. 2005;30(no. 1):21–29. [PubMed]
17. Alexander G.L. “An Analysis of an Integrated Clinical Decision Support System in Nursing Home Clinical Information Systems.” Journal of Gerontological Nursing. 2008;34(no. 2):15–20. [PubMed]
18. Alexander G. L. Proceedings of the American Medical Informatics Association. Madison, WI: Omnipress; 2007. “An Analysis of Skin Integrity Alerts Used to Monitor Nursing Home Residents.” pp. 11–15. [PubMed]
19. Alexander G. L., Rantz M. J., Flesner M., Diekemper M., Siem C. CIN: Computers, Informatics, Nursing. no. 4. Vol. 25. 2007. “Clinical Information Systems in Nursing Homes: An Evaluation of Initial Implementation Strategies.” pp. 189–197. [PubMed]
20. Rantz, M. J., J. Scott-Cawiezell, G. L. Alexander, V. Conn, R. W. Madsen, G. F. Petroski, L. Hicks, M. Zwygart-Stauffacher, and RTI International. “Evaluation of the Use of Bedside Technology to Improve Quality of Care in Nursing Facilities.” 2006. Available at http://www.cms.hhs.gov/reports/downloads/Rantz.pdf.

Articles from Perspectives in Health Information Management are provided here courtesy of American Health Information Management Association