Introduction
Background
Objectives
Methods
Results
Author
|
Facilitator / enabler for adoption
|
Themes
|
Barrier to adoption
|
Themes
|
Bias or limitation
|
---|---|---|---|---|---|
Bailey, et al. [18] | - Increased utilization of prevention / primary care | Preventative Care | - None identified | None identified | - Limited sample from Oregon which means the results are not generalizable |
- Disease prevention | Preventative Care | ||||
- Utilization can be captured through EHR, even during dramatic upturns | Data Management | ||||
- Improved data quality | Quality | ||||
- Improved workflow | Productivity/ Efficiency | ||||
- Disease surveillance | Surveillance | ||||
- Improved timeliness | Quality | ||||
Houser, et al. [19] | - Interoperability | Interoperability | - Lack of funding | Cost | - Limited sample from Alabama conference |
- Surveillance across all registries and all states | Surveillance | - Lack of medical staff support | Limited staff support | - Response bias | |
- Advancing epidemiologic research | Data Management | - Changing data standards | No standards | - Lack of resources | |
- Quality reporting | Quality | - Lack of full-time commitments | Critical thinking/treatment decisions | ||
- Clinical decision support | Decision support | - Lack of standardized data exchange | No standards | ||
Metroka, et al. [20] | - Improved efficiency | Producivity/ Efficiency | - Records may be missing data | Missing data | - Limited external validity: This study was restricted to immunizations |
- Ease of use | Ease of use | ||||
- Data sharing | Data Management | ||||
Blecker, et al. [21] | - Improved quality | Quality | - Data contains errors | - Limited external validity: Data only collected at one institution. | |
- Ease of data collection | Data Management | Missing data / data error | |||
- Ability to measure intensity of care | Productivity/ Efficiency | ||||
Flood, et al. [22] | - Surveillance | Surveillance | - Missing data | Missing data / data error | - Measurement error can be mitigated with training. |
- Disease prevention | Preventative Care | - Human error in measurement | Missing data / data error | - Interrater reliability between systems needs to be measured and controlled. | |
- EHR samples are convenience samples which may not be representative of the population. | |||||
Martelle, et al. [23] | - Improved accessibility | Productivity/ Efficiency | - Few incentives | Cost | - Small sample leads to low statistical power which reduces the external validity. |
- Improved quality of care | Quality | - Few inmates have email which reduces the demand for a patient portal. | Technology complex | - External validity limited: Study conducted in the correctional setting. | |
- Financial assistance | Financial Assistance | ||||
- Interoperability | Interoperability | ||||
Chambers, et al. [24] | - Improvement to quality | Quality | - None identified | None identified | - Selection bias |
- Surveillance | Surveillance | - Gender bias | |||
- Access to primary care information provides tailored quality improvement initiatives | Productivity/ Efficiency | - External validity limited because the gender/race demographics of the sample are not representative of the U.S. | |||
Moody-Thomas, et al. [25] | - Improved primary care | Quality | - No independent method for determining the quality of data | No standards | - Quasi-experimental |
- Intervention effective in lowering the prevalence of tobacco | Health Outcomes | - Patient-reported behavior on bad behavior can be tempered to avoid uncomfortable discussions. | |||
- Disease prevention | Preventative Care | - No similar group exists for comparison of results. | |||
- Surveillance | Surveillance | ||||
Vogel, et al. [26] | - Sustainability and generalizability | Quality | - Human error | Human error | - The voluntary nature of the Massachusetts League of Community Health Centers can create a fluid status of participating offices, which can also create orphaned data for queries. |
- Health outcomes | Health Outcomes | - Data is typically missing or incomplete | Missing data / data error | ||
- Data management | Data Management | - Data error | Missing data / data error | ||
- Productivity | Productivity/ Efficiency | ||||
Calman, et al. [7] | - Improve surveillance and management of chronic disease | Surveillance | - Cost | Cost | - Not all primary-care entities cooperate and share with public health entities. |
- Efficiency | Productivity/ Efficiency | - No central agency mandating cooperation of public health with primary care entities. | No standards | ||
- Interoperability | Interoperability | ||||
- Decisions about treatment | Decision support | ||||
- Disease prevention | Preventative care | ||||
Duan, et al. [27] | - None identified | None identified | - Electronic system failures | Productivity loss | - Information bias caused by misclassification of errors. |
- Inaccurate data (data errors) | Missing data / data error | ||||
- Complexity | Technology complex | ||||
Kawamoto, et al. [28] | - Disease prevention | Preventative care | - None identified | - Quasi experimental | |
- Improved productivity | Productivity/ efficiency | None identified | - Control group comparison data were created using a model. | ||
- Improved efficiency | Productivity/ efficiency | ||||
Behrens, et al. [29] | - Surveillance | Surveillance | - None identified | None identified | - Binning, as is common in Monte Carlo simulations, can cause bias in data. |
- Machine logic was used for best fit. | |||||
Cross, et al. [30] | - Support care coordination | Communication | - Interoperability | No standards | - Sample restricted to the state of Michigan. |
- Increased productivity | Productivity/ efficiency | - Cost | Cost | ||
- Data management | Data management | - Total adoption is a barrier because some physicians don’t want to adopt unless referrals will have the technology. | Resistance to change | ||
- Technology is up to date | Current technology | - EHRs can often obscure relevant information. | Missing data / data error | ||
Tanner, et al. [31] | - Patient safety for medications | Quality | - Fear of unintended consequences from EHRs. | Missing data / data error | - Selection bias: Only pre-meaningful use era adopters were queried. |
- Interoperability | Interoperability | - Does not address causation | |||
- Improved productivity | Productivity/ efficiency | ||||
Emani, et al. [32] | - Decrease medical errors | Quality | - Resistance to change | Resistance to change | - Study was limited to two academic medical centers in one region. |
- Physician satisfaction | Satisfaction | - Did not include factors such as practice size. | |||
- Self-efficiency | Productivity/ efficiency | ||||
Benkert, et al. [33] | - Overall positive impact overtime | Satisfaction | - Data failures/ challenges | Productivity loss | - Factors beyond the EHR that can affect poor outcomes were not measured. |
- Improved productivity | Productivity/ efficiency | - Data quality bias with level of user experience with the EHR. | |||
- Improved data collection | Data management | - Neither time lags nor staggered time points were measured or controlled. | |||
Merrill, et al. [34] | - Improved efficiency | Productivity/ efficiency | - Structural limitation | Limited staff support | - The registry database limited comparison of EHR-submitted vs non-EHR submitted data. |
- Improved productivity | Productivity/ efficiency | - Missing data | Missing data / data error | ||
- Improved compliance | Decision support | ||||
- Disease prevention | Preventative care | ||||
Glicksberg, et al. [35] | - Disease prevention | Preventative care | - None identified | None identified | - External validity: One study group was not representative of the population. |
McAlearney, et al. [36] | - Consistent communication | Communication | - Productivity loss during implementation | Productivity loss | - Small sample size greatly reduces statistical power and external validy. |
- Careful planning | Productivity/ efficiency | - Resistance to change | Resistance to change | ||
- System failure | Productivity loss | ||||
- Poor computer skills | Limited staff support | ||||
- Slow queries | Productivity loss | ||||
Polling, et al. [37] | - Data collection | Data management | - Missing data | Missing data / data error | - Not all data in the set could be matched with a record due to anonymity requirements. This limited the ability to compare data between records, and therefore limited the number of data points that were analyzed. These data points could have been dramatically different than those in the comparison. |
- Disease prevention | Preventative care | ||||
Zhao, et al. [38] | - Data collection | Data management | - Interoperability | No standards | - Limited validity and reliability |
Roth, et al. [39] | - Data collection | Data management | - Interoperability | No standards | - Data error was controlled by removing records that contained implausible values. This may have skewed the data because, while implausible, the data could have described an unusually sick population. |
- Surveillance | Surveillance | - Prone to data-entry error | Missing data / data error | - Free-text fields are inherently difficult to include in analysis. The data contained within free-text fields may have skewed the results differently. | |
- Missing data | Missing data / data error | - Without a time-series or longitudinal study, it is difficult to generalize the results. | |||
Barnett, et al. [40] | - None identified | None identified | - none identified | None identified | - The small sample of 17 hospitals reduces statistical power which may limit the generalizability of the results. |
- Researchers unable to explore the associataion of EHR implementation with inpatient outcomes stratified by implementation context, hospital, or EHR characteristics. | |||||
Drawz, et al. [41] | - Improved performance | Productivity/ efficiency | - Limited functionality | Technology complex | - The lack of nationwide data eliminates comparisons to a national benchmark. |
- Interoperability | Interoperability | ||||
- Improve measuring data (data collection) | Data management | ||||
Thirukumaran, et al. [42] | - None identified | None identified | - Temporary decrease in quality | Productivity loss | - Limited generalizability |
Adler-Milstein J, Everson J, Lee SY. [43] | - Increased quality | Quality | - None identified | None identified | - Adherence to process measurers was high across hospitals which reduces the opportunity to observe EHR-driven improvements. |
- Increased efficiency for hospital care | Productivity/ efficiency | - This study only analyzed the Medicare arm of the Meaningful Use program. | |||
- Patient satisfaction | Satisfaction | ||||
- positive relationship between EHR adoption and performance | Interoperability | ||||
Ananthakrishnan, et al. [44] | - Health outcomes | Health outcomes | - Misclassification (data error) | Missing data / data error | - The cohort studied represents a small population: Therefore, the external validity of results are limited. |
- Quality in documentation | Quality | - Interoperability | No standards | - All provider notes may not have been captured if a participant saw a physician through a private setting. | |
- Lends generalizability to findings | Productivity/ efficiency | ||||
Carayon, et al. [45] | - Increased productivity | Productivity/ efficiency | - Increased amount of time spent on documentation and clinical review | Productivity loss | - Not generalizable nationwide because data were collected at only one location. |
- Efficiency gains | Productivity/ efficiency | - Decreased direct patient care (quality) | Decreased quality | - Physicians were not identified, and therefore their contributions may have occurred both pre and post treatment. Having this information would have made analysis easier. This can introduce observer bias that has not been controlled for. | |
Redd, et al. | - None identified | None identified | - Negative impact on productivity and efficiency | Productivity loss | - Face validity: Clinical volume is not an exact match for provider productivity, but other studies have used this measure. |
[46] | - Time consuming | Technology complex | - Construct validity: Due to the lack of baseline data available, it is difficult to discern that the intended measure is accurate. | ||
- Missing data | Missing data / data error | ||||
Jones & Wittie [47] | - Widespread adoption | Current technology | - Lacked functionality | Accessibility/ utilization | - Limited external validity due to the uncertainty that Beacon communities across the country are homogeneous. |
- Improved quality | Quality | - Complexity | Technology complex | - Self-report data can be questionable, but sufficient research has been conducted using similar data, researchers felt comfortable. | |
- Care coordination (communication with data exchange) | Interoperability | ||||
- Layering of financial incentives | Financial assistance | ||||
- Technical assistance | Communication | ||||
Hammermeister, et al. [48] | - Data collection | Data management | - Missing data | Missing data / data error | - Limited clinic-level data precludes comparison characteristics between hih and low outlier clinics. |
- Inexpensive data collection (cost) | Financial assistance | - External validity is limited because the sample is not representative of the national population. | |||
Benson, et al. [49] | - Interoperability between EHR and primary care systems | Interoperability | - Potential missing data | Missing data / data error | - Infrequency of visits creates missing data. |
- Efficient comparison of patients | Productivity/ efficiency | - Some privacy concerns | Privacy concerns | - Standardized measures for risk factors do not exist. | |
- Inability to conduct certain logistic functions (lack of functionality) | Productivity loss | - Self-report data can be unreliable. | |||
Soulakis, et al. [50] | - Communication between patients and providers | Communication | - Complex analysis | Technology complex | - The measure of interrater reliability is confounded because some providers served on many teams. |
- Preventative care | Preventative care | ||||
Burke, et al. [51] | - Improved over quality of outpatient clinical notes | Quality | - Standards across EHRs | - Not generalizable to all EHR systems because only one was studied. | |
- Accessibility | Ease of use | No standards | |||
- Improved efficiency | Productivity/ efficiency | ||||
Keck, et al. [52] | - Surveillance | Surveillance | - Limited design, deployment and function (complexity) | - Construct validity is questionable due to lack of baseline data. | |
- Increased time availability (productivity) | Productivity/ efficiency | Technology complex | - Generalizability limited because only the Indian Health System was studied. | ||
- Improved data validity and reliability | Quality | ||||
Roth, et al. [53] | - Surveillance | Surveillance | - Fail to capture important discrete necessary data (missing data) | Missing data / data error | - Selection bias due to a convenience sample. |
- Reduce health disparities (health outcomes) | Health outcomes | - Lack of workflow integration paradigms(productivity) | Productivity loss | - Smoking data is inherently underreported, so the effects of this study are understated. | |
De Moor, et al. [54] | - Reduce duplication and errors | Quality | - Regional diversity in languages. | No standards | - None identified |
- Data collection | Data management | - Interoperability | No standards | ||
- Improved efficiency | Productivity/ efficiency | - Inconsistent documentation | Missing data / data error | ||
- Data quality | Decreased quality | ||||
Chang, et al. [55] | - None identified | None identified | - Missing data | Missing data / data error | - External validity limited: While the computed algorithm satisfactorily predicted one behavior, it is uncertain if such models can be developed for all. |
Reed, et al. [56] | - Increased/positive impact on critical thinking skills | Decision support | - None identified | None identified | - Response bias decreased the number of participants. |
Inokuchi, et al. [57] | - Productivity (reduced time) | Productivity/ efficiency | - No patient outcomes | - Need larger sample size | |
- Increased physician satisfaction | Satisfaction | Decreased quality | - The Hawthorne effect may have increased bias toward the new EMR. | ||
- Increased use of information | Decision support | - External validity may be questionable because only one EMR was studied. | |||
- Organizational impact | Productivity/ efficiency | ||||
Silfen, et al. [58] | - Prompt healthcare providers to screen for chronic health issues (preventative care) | Productivity/ efficiency | - No return on investment | - External validity limited because data were not available for all organizations and anything outside of New York City. | |
- Facilitate provider referrals | Communication | Cost | - Data were not complete | ||
- Supplies rapid feedback to providers | Decision support | ||||
- Track patient outcomes | Health outcomes | ||||
- Monetary/financial incentive | Financial assistance | ||||
Zera, et al. [59] | - None identified | - No effect on the rates of diabetes screening | Disease management | - Data bias may have skewed results toward the null result. | |
None identified | - No access to screening responses (structural limitation) | Missing data / data error | - External validity limited: Small numbers in the control group reduces the statistical power. | ||
- No patient outcomes | Decreased quality | ||||
Baus, et al. [60] | - Surveillance | Surveillance | - The EHR is designed for patient care, not for research. | Accessibility/ utilization | - Not generalizable, sample bias: Clinics were chosen through purposive sampling. |
- Preventative care | Preventative care | - Human error in recording data in the EHR. | Human error | - Unable to combine data for extrinsic information (structural limitation). | |
- Data quality for population health management | Quality | ||||
Baus, et al. [61] | - Preventative care | Preventative care | - Difficulty of extracting necessary data (technical challenges). | Technology complex | - Limited variability in participants limits external validity. |
- Surveillance | Surveillance | - Cost | Cost | ||
- Improved efficiency | Productivity/ efficiency | - Interroperability | No standards | ||
- Improve decision support | Decision support | ||||
- Increase the application of patient data to care | Data management | ||||
- Improve health outcomes | Health outcomes | ||||
Haskew, et al. [62] | - Real time access (efficiency) | Productivity/ efficiency | - Cost | Cost | - Limited external validity due to short time studied and difficulty of implementation model. |
- Sharing data (communication) | Communication | - Limited staff | Limited staff support | ||
Puttkammer, et al., [63] | - Preventative care | Preventative care | - Missing data | - Self-report data is questionable and subject to ability to recall or social desirability. | |
- Data/information accessibility | Ease of use | Missing data / data error | - Missing data that could have skewed the results. | ||
- External validity limited because only two organizations studied. | |||||
Zheng, et al. 66] | - Surveillance | Surveillance | - Difficulty combining information from EHR with structured data | Technology complex | - External validity limited |
- Data collection | Data management | ||||
Wu, et al. [64] | - Data collection | Data management | - Missing data on smoking status | - Citation bias | |
- Smoking surveillance | Surveillance | Missing data / data error | - Self-report data on smoking is limited due to social desirability, therefore the results of this study may be understated. | ||
- Facilitating care identification | Communication | ||||
Nguyen & Yehia [65] | - Data collection | Data management | - Different documentation rates at Different healthcare systems | No standards | - External validity limited because only one health system in one region of the U.S. was studied. |
- Preventative care | Preventative care | ||||
Tomayko, et al. [66] | - Data collection | Data management | - None identified | None identified | - External validity limited because the demographics do not match that of the U.S. |
- Preventative care | Preventative care | - Self-report data can be questionable and subject to bias due to recall and social desirability. | |||
- Disease management/monitoring (child obesity) | Surveillance | ||||
- Quality improvement | Quality | ||||
- Greater surveillance of a population | Surveillance | ||||
- Cost effective | Financial assistance | ||||
Romo, et al. [67] | - Surveillance | Surveillance | - Data is often skewed toward those who seek care. | Missing data / data error | - Data bias: Missing values were filled with estimates which may skew the results. |
- Generalizability | Productivity/ efficiency | - Self-report data is questionable and subject to bias due to recall and social desirability. | |||
- External validity limited to U.S. only. | |||||
Chambers, et al. [68] | - Data collection | Data management | - None identified | None identified | - Quasi experimental. |
- Selection bias. | |||||
Wang, et al. [69] | - Improved quality | Quality | - None identified | None identified | - External validity limited: Only 151 organizations studied, therefore generalizing outside those practices may be limited. |
- Work flow variability (productivity) | Productivity/ efficiency | - Structural limitation | |||
- Selection bias: Early adopters were selected for the study. | |||||
Chiang, et al. [70] | - Increased faculty providers | Satisfaction | - Initial decrease in clinical volume | Productivity loss | - Interrater reliability was controlled by using a stable group of providers. |
- Longer notes | Communication | - Increased time expenditure and documentation times | Technology complex | - Baseline data was established during a three-month period (Nov-Jan). | |
- More automatically generated texts (efficiency) | Productivity/ efficiency | - Increased reliance on textual descriptions and interpretations (human error) | Human error | - Construct validity limited because clinical volume may not be an equal measure of productivity. | |
- Little to no increase in clinical volume | Productivity loss | - External validity limited: The only EHR studied was at a large academic medical center which may not be representative of all organizations in the U.S. |
Additional analysis
Facilitators | Reference | Occurrences | Reference | Barriers | |
---|---|---|---|---|---|
Productivity/ efficiency | 19,21,22,24,25,27,28,30*,32,33–36*,38,43,45–47*,51,53-56,59*,60, 63,64,70,72,73 | 33 | 24 | 21,22,23*,27*,29,32,33,36,39,41*,46,48,50,51,55,56,57,61,65,67,70 | Missing data / data error |
Quality | 19*,20,22,24-27,33,34,45,46,49, 53,54,56,62,69,72 | 19 | 13 | 20*,26,28,32,40,41,46,53,56*,63,68 | No standards |
Data management | 19–22,27,28,32,35,39, 40,41,43,50,56,66-69,71 | 19 | 12 | 29,35,38*,44,47,48,51,55,73* | Productivity loss |
Surveillance | 19,20,23,25,26,28,31,41,54,55, 62,63,66,67,69*,70, | 17 | 10 | 24,29,43,48,49,52,54,63,66,73 | Technology complex |
Preventative care | 19*,23,26,28,30,36,37,39,52,62, 63,65,68,69 | 15 | 7 | 20,24,28,31,60,63,64 | Cost |
Communication | 32,38,49,52,60,64,67,73 | 8 | 4 | 47,56,59,61 | Decreased quality |
Interoperability | 20,24,28,33,43,45,49,51 | 8 | 4 | 20,36,38,64 | Limited staff support |
Decision support | 20,28,36,58-60,63 | 7 | 3 | 32,34,38 | Resistance to change |
Health outcomes | 26,27,46,55,60,63 | 6 | 3 | 27,62,73 | Human error |
Satisfaction | 34,35,45,59,73 | 5 | 2 | 49,62 | Accessibility/ utilization |
Financial assistance | 24,49,50,60,69 | 5 | 1 | 61 | Disease management |
Ease of use | 21,53,65 | 3 | 1 | 20 | Critical thinking/treatment decisions |
Current technology | 32,49 | 2 | 1 | 51 | Privacy concerns |
* more than one occurrence | 147 | 85 | |||
None identified | 29,42,44,48,57,61 | 19,25,30,31,37,42,45,58,69,71,72 |