Exploring Resident Care Information Technology Use and Nursing Home Quality

Abstract

A wide array of sophisticated information technology systems are being used in nursing home resident care to improve quality. The purpose of this study was to explore differences in nursing home information technology sophistication, a comprehensive measure of adoption, used in resident care processes based on facility characteristics over four consecutive years and to examine the impact on select long-stay nursing home quality measures. Results indicate information technology systems used in resident care are becoming increasingly sophisticated during these four years. Nursing home bed size, type of ownership, and location were significant predictors of information technology score in areas related to resident care. Results also suggests that as electronic clinical processes and documents increase (e.g. incident reporting, nursing flowsheets, care planning) in resident care, more falls with injury are detected. Continued assessments of NH IT sophistication are important as we continue to evaluate the impact of technology on quality.

Health information technology (IT) systems are designed to assist in the delivery, support, and management of patient care. While implementation of IT systems has grown dramatically in recent years in acute and ambulatory care, other sectors such as nursing homes (NHs) have experienced slower uptake. For example, in 2017 88% of hospitals were able to send electronic patient information to other providers, in contrast, a recent national study found only 46% of NHs with any capability to exchange health information electronically (American Hospital Association, 2019; Powell, Deroche & Alexander, 2020). Although adoption of IT systems was not federally incentivized in NHs, there is increasing pressure to implement such systems. Payment changes including a shift from fee-for-service to prospective, value-based payment and market based pressures are escalating the need for sophisticated IT systems to increase efficiencies, improve quality and safety (Grabowski et al., 2017; Huckfeldt, Weissblum, Escarce, Karaca-Mandic, & Sood, 2018). As NHs take heed of the political and consumer driven demand for improved IT capabilities, it is important to understand trends in adoption and how NH IT impacts quality.

A wide array of IT systems can be used in providing care to NH residents which could impact quality. These systems are used in direct resident care by different members of the interdisciplinary team including nurses, nurse aides, physical therapists, occupational therapists, and others. For example, IT systems can be used to digitize documents such as medication administration records, care plans, nursing flowsheets, and incident reports. While these technologies are being used in some NHs, a systematic assessment of NH IT sophistication is necessary to trend adoption and begin to understand how these systems influence quality of care. Currently there are no reported links between use of IT systems and quality reported on a national level.

Members of our research team has been measuring IT adoption in NHs, also called IT sophistication since 2006 (Alexander, Madsen, Herrick & Russell, 2008). We have developed and tested a survey measuring IT sophistication across three healthcare domains (resident care, clinical support, administrative activities) and three IT dimensions (IT capabilities, extent of use, and degree of integration) in order to develop a complete and systematic assessment of NH IT. We have started to examine the relationship between nationally reported NH quality measures (QMs) and the nine dimensions/domains of IT sophistication but have only begun to look deeper into these dimensions/domains within specific content areas (CAs) and their impact on quality. Each of the nine dimensions and domains are broken down into 27 CAs with associated content items used to describe and trend a full range of IT sophistication measures (Alexander, Deroche, Powell, Mosa, Popejoy & Koopman, 2020).

By examining relationships between specific CAs and QMs, we continue working toward our overall goal of determining which technologies make a difference in NH quality in order to inform practice and policy. Relating specific technologies to quality requires a deeper level of granularity than what has previously been explored. The aims of this study were 1) to explore differences in NH IT sophistication used in resident care processes based on facility characteristics (bed size, location, type of ownership) over four consecutive years and 2) to examine the impact NH IT sophistication in resident care processes have on select QMs. A guiding model of independent variables (NH characteristics, IT sophistication CAs, time in years) and dependent variables (NH QMs) is presented in Figure 1.

Figure 1:

A Conceptual Model to test Nursing Home IT Sophistication and Quality Measures

Methods

This research includes a longitudinal design using surveys of US NHs that were randomly selected from each state using the publicly available Nursing Home Compare dataset. NHs were recruited over a four-year period from January 1, 2014-January 1, 2018. The institutional review board (Protocol #1209004) approved all protocols.

Sample

The research team recruited a randomized sample of NHs from each state in the US. The target goal was 10% of all NHs from each state in the U.S. A total of 15,653 NHs were identified in the final data set resulting in a target sample of 1,565 homes. We oversampled by 15% in each state to account for attrition. Ultimately, 1977 NH administrators agreed to participate in the survey and 815 returned completed surveys in Year 1 (41.5% response rate). Researchers recruited NH administrators from the same 815 facilities in Year 1 to participate in Year 2, Year 3, and Year 4 of the study using the same recruitment strategy. A total of 188/815 NHs (23% response rate) completed all four years of the study. Our sample was significantly different (more non-profit, fewer homes in metropolitan areas but slightly more rural facilities) than remaining homes in the national sample. Differences in facilities based on bed size was nonsignificant (see Table 1). The final sample included NHs from every US state except Delaware, Hawaii, Idaho, Maryland, and The District of Columbia (DC). A detailed summary of the recruitment strategy has been previously published (Alexander et al., 2017).

Table 1.

Study sample vs. national sample according to organizational variables.

	National Sample		Study Sample
	N	%	N	%	p-value
Ownership
For-profit	12002	76.88	113	60.11	< 0.0001
Non-profit	3610	23.12	75	39.89	< 0.0001
Region
Metro (>50000)	10034	64.41	105	55.85	0.0142
Micro (10000–49999)	2274	14.60	24	12.77	0.4773
Small town (2500–9999)	1875	12.04	33	17.55	0.0203
Rural (<2500)	1396	8.96	26	13.83	0.0194
Bed size
>120	4474	28.66	44	23.40	0.1107
60–120	8162	52.28	105	55.85	0.3278
<60	2977	19.07	39	20.74	0.5600

Survey Measures

The survey measured trends in NH IT sophistication each year. Excellent reliability estimates have been reported previously (Alexander et al., 2017). To answer our research questions, we focused on six CAs in the domain of resident care across the three dimensions of IT sophistication to explore impacts on QMs measured at the same time.

NH Quality Measures

We included six of sixteen long-stay QMs reported in the Nursing Home Compare dataset. These QMs were selected because they were found to be significantly correlated with total IT sophistication in a previous study (Alexander et al., 2019). The six QMs used in this study are presented in Figure 1.

Statistical Approach

To account for the dependency of observations (i.e. measures on the same NH over time), a General Estimating Equations (GEE) population-average model with an exchangeable working correlation structure was fit using 752 observations (188 NH times 4 years). GEE is similar to regression but has multiple advantages such as using all available pairs of data, providing more reliable parameter estimates, and producing more robust findings (i.e. accurate standard errors) (Hardin & Hilbe, 2012). To address our first aim, we fit a GEE model predicting each resident care CA score (CA 1–6) by year, controlling for the NH characteristics of bed size (<60, 60–120, >120), location (metropolitan, micropolitan, rural, small town), and type of ownership (for profit, non-profit). To address aim 2, we fit a GEE model predicting each QM score by year and each resident care CA score, controlling for the NH characteristics of bed size, location, and ownership. The significance level was set at α = .05. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

Results

Content Areas and NH Characteristics

Time in years was a significant predictor of the IT sophistication score for all six CAs while controlling for bed size, location, and ownership. (see Table 2 in supplemental files). There was a steady trend (increasing) in IT sophistication scores in all six CAs over the four-year study period. IT scores in CAs 1, 3, and 5 were not significantly impacted by NH bed size, location, or type of ownership. IT scores in CA 2 were significantly impacted by bed size and type of ownership. NHs with smaller bed size (<60) were estimated to have a 1.05 point lower (95% CI: 3.22, 5.47) IT sophistication score in CA 2 compared to NHs with >120 beds. This indicates NHs with smaller bed sizes have lower IT sophistication in resident care technologies including computerized clinical documents and processes. For profit NHs were estimated to have a 0.59 point higher score (95% CI: 0.01,1.17) in CA 2 compared to non-profit NHs. CA 4 and CA 6 were significantly impacted by bed size as well. NHs with <60 beds were estimated to have a 4.12 point lower score (95% CI: −7.07, −1.16) in CA 4 compared to NHs with >120 beds. Smaller NHs were also estimated to score lower (95% CI: −1.32, −0.03) in CA 6 compared to NHs with >120 beds.

Content Areas and Quality Measures

Time in Years was a significant predictor of QM score while controlling for bed size, location, and ownership in five of the six QMs examined in this study (see Table 3 in supplemental files). Year was not a significant predictor of the QM score for 410 (falls with injury). CA 1 was significantly associated with QM 410 (falls with injury). For every one point increase in CA 1, the QM 410 score is expected to increase by 0.11 points (95% CI: 0.02, 0.18). This suggests that as electronic clinical processes and documents increase (e.g. incident reporting, nursing flowsheets, care planning) in resident care more falls with injury are detected. QM 401 (ADL increase), QM 402 (pain), and QM 403 (pressure ulcers) were not significantly associated with the six CAs, bed size, location, or type of ownership. QM 408 (depressive symptoms) was associated with both NHs with <60 beds and located in rural areas. NHs with <60 beds were estimated to have a 5.73 point higher score in QM 408 compared to NHs with >120 beds (95% CI: 0.04, 2.68). Rural NHs were estimated to have a 3.42 point lower score in QM 408 (depressive symptoms) compared to NHs located in metropolitan areas (95% CI: −5.70, −1.13). Lastly, QM 409 (restraints) was found to be significantly associated with both bed size and location. NHs with 60–120 beds were estimated to have a 0.43 point lower score in QM 409 compared to NHs with >120 beds (95% CI: −0.83, −0.03). NHs located in rural areas were estimated to have a 0.53 point lower score in QM 409 compared to NHs located in metropolitan areas (95% CI: 0.02, 1.03).

Discussion

Caring for older adults residing in NHs involves a dynamic process of ongoing assessment, transitions, and shifting care plans. Strategic use of IT systems to lessen the burden and improve outcomes is possible, however, a comprehensive and systematic assessment of the capabilities, extent of use, and level of integration of different systems is a prerequisite. We used an instrument designed to systematically measure NH IT adoption and focused specifically on technological components contained in CAs related to resident care. Results from this study offer some insight into how NHs are using IT in resident care and the impact on QMs.

In this study, we examined six CAs representing IT systems used in resident care across the IT domains: capabilities, extent of use, and degree of integration. Time in Years was a significant predictor for all six CAs. The biggest change that occurred across the four years of the study was in CA 3: the extent of use of technologies in resident care and CA 5: degree to which resident care systems are integrated with other systems in the NH. This finding demonstrates that while IT capabilities are growing in NHs, more substantial growth is occurring in the extent to which these systems are being used and degree to which they are integrated with other systems. For example, CA 3 includes specific technologies such as telemedicine, sensors, and clinical decision support. This finding is consistent with recent literature reporting more extensive use of technologies like telemedicine in NHs for specialty consultations (Edirippulige, Martin-Khan, Beattie, Smith, & Gray, 2013) and use of sensors for early detection of behavioral changes and risk for falls (Cameron et al., 2018; Goerss et al., 2019).

Our study revealed several important differences in CAs that included items specific to physical and occupational therapy (PT/OT). First, NHs with <60 beds scored lower in these CAs compared to NHs with larger bed size. Second, the only significant difference in CAs according to profit status was found in CA 2: PT/OT processes that are computerized. For-profit NHs were estimated to have a higher score in CA 2 compared to non-profit NHs. These findings are especially interesting given the recent changes in the Medicare payment model for therapy in skilled nursing facilities. As of October 1, 2019, the Patient Driven Payment Model (PDPM) replaced the Prospective Payment System (PPS) to improve appropriateness of payments by focusing on patient diagnoses and characteristics rather than the volume of services provided (CMS, 2020). Since this policy took effect after our study took place, we are unable to determine what effect if any these changes had on adoption of IT systems specifically intended to support PT/OT. We could speculate that for-profit NHs took a proactive approach to implementing IT systems in PT/OT in anticipation of these changes. Reassessment of the impact of IT related to PT/OT should be done as findings could have important implications for policy evaluation.

Our examination of the relationship between resident care CAs and QMs yielded several remarkable findings. First, the only CA found to be a significant predictor of any QM was CA 1: clinical processes or documents that are computerized and QM 410 (falls with injury). We speculate that more sophisticated computerized processes does not actually result in more resident falls with injury but perhaps aids in accurate reporting of such events, possible through electronically enhanced incident reporting systems. Second, QM 408 (depressive symptoms) and 409 (restraints) were significantly different based on NH location. NHs located in rural areas had a lower percentage of long-stay residents having symptoms of depression and a higher percentage of residents who were physically restrained compared to NHs located in metropolitan areas. This finding is consistent with other urban-rural disparities reported in the literature. For example, NHs located in rural areas have been found to have lower staffing, poorer quality end of life care, and are less likely to achieve high start quality ratings, a measure based on performance among QMs (Bowblis, Meng & Hyer, 2013; Lutfiyya, Gessert & Lipsky, 2013; Temkin-Greener, Zheng & Mukamel, 2012). Further research is needed to understand how enhanced IT systems can be used to meet the unique needs of rural NH residents.

Limitations

In this longitudinal study, we relied on survey data collected from NH administrators over four years. There may have been bias in the responses to the NH IT sophistication survey. Presumably, some NHs may not participate because they have no technology which could result in a reported overall higher levels of IT capabilities, extent of use, and degree of integration. We limited our analyses to three organizational variables (NH bed size, location, and ownership). These variables were selected because they demonstrated significance in prior work, however, future studies should include a broader set of organizational variables for example staffing, bed occupancy, and chain affiliation. Lastly, we focused on six of the sixteen QMs reported in the Nursing Home Compare dataset. While we selected these QMs based on prior work and expertise of the research team, repeating the analysis with all sixteen QMs could provide more insight into the relationship between technology and quality.

Conclusion

The purpose of this study was to explore the relationship between NH IT systems used in resident care processes and select long-stay QMs. We took a more granular approach by looking at six specific CAs related to resident care among the IT dimensions of capabilities, extent of use, and degree of integration. Findings revealed IT systems are increasing in sophistication over time, especially in areas directly related to resident care. We also discovered important differences according to NH bedside, location, and ownership. Continued assessments of NH IT sophistication are important as we continue to evaluate the impact of technology on quality.