[ad_1]
DISCUSSION
Data management activities were implemented in the primary care organisation we studied. This implementation was associated with significantly greater increases in coding for chronic conditions studied compared to other Canadian practices.
There is limited evidence on which interventions are most effective in improving data quality.56,57 Repeated assessments, feedback and training may be effective.57 However, this represents a significant time investment for practitioners; the extra work could compete with the already extensive requirements associated with providing clinical care,16 which may limit acceptance. Using an automated EMR alert based on clinical criteria and prompting the clinician to add a condition to the problem list if it is missing has been found to be effective.58 However, programming this in commercial EMRs routinely used may be challenging. There are about 20 EMR applications used across Canada and each one would require individual programming and support for this process. Regulations could be used to mandate EMR-based data improvement activities.
During the pilot for this project, we found that a simple approach minimizing physician workload was not costly and was acceptable to the physicians involved.52 We used their expertise only to verify cases, with no additional training; this did not interfere with clinical encounters. Most of the work was delegated to other members of the primary care team or to data clerks as appropriate. Acceptance in this project was high, with 83%–90% of physicians returning their lists of patients.
The organisation we studied was interested in data quality and had already made efforts to implement some coding for chronic conditions prior to the project.52 Data clerks had entered the diagnostic code for diabetes in the previous year. This explains the high rate of baseline coding for that condition (81.7% in NYFHT) compared to other CPCSSN physicians (63.6%). Coding for diabetes increased less than for the other conditions possibly because of ceiling effects due to prior efforts; however, the increase was still greater than for the comparator group. No consistent efforts had been made for the other three conditions; baseline coding prevalence was similar to other physicians across Canada.
The lack of adoption of terminology standards and the prevalence of uncoded, ‘local’ or idiosyncratically coded data presents challenges in terms of electronic communication and interoperability.59,60 We used the most commonly entered codes in a national database for each condition studied as an initial step towards more consistent terminology. We demonstrated that coding conforming to an external norm could be implemented in a complex and distributed primary care organisation in Canada.
Maintaining data quality will require ongoing efforts. In order to improve sustainability, the data clerks documented the processes used for this project. A handbook is provided at http://drgreiver.com/NYFHTSummerStudentProgramHandbook.pdf. NYFHT has also developed a manual for standardized data entry, which is available to all members of the team and has been shared with other teams. Data quality activities using the same approach are ongoing at NYFHT and have been expanded to include additional conditions. Scalability should also be considered; small primary care teams may not have the resources to implement this approach. However, in Ontario, primary care analysts (QIDSS) have been embedded in FHTs. The DPT and associated processes have already been provided to additional analysts, each supporting multiple FHTs in Ontario, as well as to primary care networks in Alberta. To assist with governance and processes, we have provided templates and tools developed as part of this project for privacy, and data entry and analytics to the Association of Family Health Teams of Ontario.61 The Association has been tasked with assisting the provincial implementation of analytics in FHTs through the QIDSS program.47 Additional resources for support, continuing development and broader implementation of DPT in primary care are being actively pursued by CPCSSN.
Similar approaches could be used elsewhere. In the U.S., significant funding has been devoted to improving data in EMRs. For example, a problem list needs to be used for 80% or more of patients in order to meet meaningful use goals.62 Adaptations of these processes could be used to rapidly improve the completeness and coding of data in problem lists.
Limitations
This study was a convenience sample for both cohorts. However, physicians participating in CPCSSN were reasonably similar to others in Canada.63,64 An observational cohort study was used; this is subject to both measured and unmeasured confounders. We used statistical adjustments for factors we measured. We could not measure factors possibly affecting coding, such as dictation; however, we compared the change in coding over time rather than providing a cross-sectional comparison. Data reflects only patients seen for care over time; it is possible that more frequent visits could lead to improved recognition of a chronic condition and increases in associated coding. However, physicians at NYFHT had less frequent patient visits than the national cohort. The specificity of CPCSSN case definitions varied, with some false positive cases. It is also possible that clinicians may not recognize that a condition was present for some patients on their verification list.
In conclusion, data management activities were implemented by the primary care organisation; this was associated with an increase in standardized coding for four chronic conditions. A similar environment currently exists in other primary care organisations in Ontario. Planning for resources and activities that would allow the adoption of data management within primary care organisations in order to support data and quality improvement may be worthwhile.
Where this study fits in
EMR data quality in primary care including standardized coding for chronic conditions is currently limited.
We implemented data management activities in a large primary care organisation in Ontario, Canada. This included the return of merged, cleaned EMR data and reporting software to the organisation’s data manager.
The team used these resources to code and standardise designations for chronic conditions; coding improved to a greater degree than in comparable practices across Canada.
Improving data quality in primary care using this approach appears to be feasible.
[ad_2]
Source link
