General practitioners’ experiences of a data-driven quality development process

Research highlights that structured data use in general practice can optimise treatment quality and identify patients needing specific treatment [1]. Danish studies have found that electronic feedback based on general practice treatment data can successfully underpin treatment guidelines, making it an effective tool for improving the quality of care in general practices [2, 3]. Nonetheless, clinical decision-making extends beyond treatment guidelines and research evidence. Taking into account clinical variables, patient preferences and their life circumstances is an essential clinical competency, and managing type 2 diabetes (T2D) through a comprehensive, multifactorial approach is crucial to reduce the risk of complications [4]. A population-based approach to data management can enhance clinical practice and minimise unnecessary variations in treatment by targeting specific objective measures [2]. This study was a qualitative evaluation of an intervention in general practices in Northern Zealand (Capital Region), Denmark. The intervention aimed to assess whether increased use of population data in general practices may optimise treatment and strengthen workflows. General practitioners (GPs) can access quality indicators and treatment outcomes related to national T2D guidelines at the population level. Data indicators are displayed as interactive charts and patient-specific lists, strengthening quality development with real-time data.

Methods

In Denmark, healthcare is free of charge, and primary care is provided by self-employed GPs under contractual agreements with the public health system [5]. A GP typically has around 1,600 patients [6]. As gatekeepers of the Danish National Health Service, GPs handle nearly all referrals to specialists and hospitals. The prevalence of T2D in Denmark is 6.8%, and roughly 80% of these patients are treated by a GP [7].

The Danish research initiative – DataSam

This qualitative study is part of the research initiative DataSam, which was designed as a quality development feasibility study [8]. General practice clinics were invited to participate in a newsletter (PraksisNyt), followed by phone calls and oral presentations in general practice medical forums. A total of 19 clinics volunteered to join DataSam (four clinics dropped out during the intervention).

DataSam was a one-year intervention containing a quality development process focusing on data usage in collaboration with an experienced GP and organisational advisor. T2D population data for each clinic were accessed through an internet-based quality programme for GPs in Denmark (treatment pathways, in Danish: forløbsplaner (FLP)) [8]. DataSam aimed to optimise T2D treatment and workflows in general practices. GPs were offered:

- Clinic visits at baseline and at six and 12 months, focusing on data-driven quality development using FLP population data for T2D [8]. Each clinic was offered technical support to ensure proper setup and consistent coding of T2D data, followed by workflow optimisation advice. Each clinic developed a quality development plan based on its T2D population data, data management and treatment outcomes [8]. TKM participated in most of the visits in an observatory role.

- Monthly online dialogue meetings with local hospital specialists, including a short presentation. These 30-minute lunchtime sessions allowed attendees to ask questions and discuss cases from their clinics.

Audio records of clinic visits and participants

Most clinic visits were audio-recorded (all time points: 12 clinics). From recordings (≈ 60 minutes), we assessed clinic-specific data-driven quality development from baseline to conclusion of the intervention interviews. This included changes in the organisation of T2D care, data usage challenges and levels of acceptance of project initiatives. Characteristics are presented in Table 1.

Conclusion of the intervention interviews

At the end of the one-year period, each clinic was invited to participate in a semi-structured evaluation interview. Invitations were extended at the final visit, and participation was voluntary. In total, 14 clinics participated (Table 2). The semi-structured interview guide included open-ended questions on data use, workflow changes and implementation challenges. Participants were encouraged to share other perspectives to address unforeseen issues. Interviews lasted between 13 and 39 minutes (median: 20 minutes), were audio-recorded and conducted individually via Teams, except for two, where a GP and nurse were interviewed together. The interviewer (LH) had no prior DataSam involvement.

Data analysis of visits and interviews

All audio-recordings were transcribed verbatim, and the analysis was inspired by qualitative content analysis [9], involving:

- Conducting line-by-line open coding, identifying statements of potential interest while remaining open to emerging categories.

- Organising recurring codes into themes.

- Performing focused coding after determining themes and searching for patterns developed during coding [10].

The sub-theme “Positive experience of clinic visits” was developed based on codes derived from statements like “It has been useful to share knowledge and tips with another experienced colleague. It is a competence boost …” and “The visits keep you on your toes. You remember to pull data out and pay attention …”. Such expressions reflected how clinic visits were perceived as quality development opportunities. The analysis was conducted cross-sectionally [11], with data analysed at baseline, six and 12 months to identify changes over time. Subsequently, common changes across clinics were identified.

The analysis was iterative, moving back and forth between open and focused coding to allow new themes to emerge. LH and SSEB conducted the primary analysis, while TKM and NB reviewed and discussed the content.

Ethical considerations

As a quality project, the study was registered as “not required approval” by the Regional Ethics Committee of the Capital Region (F-22073139), but it was conducted following the Declaration of Helsinki. Participants were informed orally and in writing that their participation was voluntary and that the interviews would be recorded and used for evaluation on an anonymous, aggregate level. All provided informed consent.

Trial registration: Registered as “not required approval” with the Regional Ethics Committee of the Capital Region (F-22073139).

Results

Initially, GPs were positive and expected that the optimised data structure would simplify data and FLP management, enabling more focused work. They anticipated enhancements in population monitoring, a more structured internal overview and optimised treatment. One clinic raised concerns about the possible effect on core values and emphasised worries about the required time investment.

Shared experiences of data use from baseline to the end of the intervention

Table 3 presents shared experiences across clinics. This primarily included how the T2D care organisation within clinics changed, and the high acceptance of project initiatives. Technical issues and time constraints were encountered.

Benefits of data use and changes in workflows

Findings from the clinic visits were contextualised during the semi-structured interviews. Three overall themes were identified (Table 4). In Theme 1, focusing on the benefits of data use, all clinics shared how they had adjusted workflows and made data use more systematic. All clinics reported that structured data utilisation improved the use of FLP, and nurses assumed more responsibilities. This allowed many clinics to gain a T2D population overview, optimised treatment, along with increased staff skills, knowledge, job satisfaction and confidence (Table 4). One clinic noted time consumption as a drawback, and another felt minimal impact of the study. Nonetheless, all clinics except one (Table 3) reported increased confidence in a data-driven approach, with some aiming to apply it to other diagnoses.

Implementation challenges in practice

All clinics faced some level of implementation challenges. Although working with data was frequently beneficial, it was not always intuitive, and staff needed time to familiarise themselves with it. The general practices reported spending an average of 4.5 hours (2-14 hours) working with FLP data between visits 1 and 2, which decreased to 2.2 hours (1-4 hours) between visits 2 and 3. Some noted that data use could lead to overtreatment and expressed concerns about potential errors and misuse. Moreover, the knowledge and responsibility obtained remained mainly with the clinic project coordinators. Additionally, some reported that the emphasis on T2D required deprioritising other diagnoses due to time constraints.

Experiences with project activities

The visits were generally positively received, with most clinics noting improvements in data extraction and overview (Table 4). Sharing knowledge with an experienced GP was appreciated, providing feedback and reinforcing adherence to procedures. Only one clinic found the visits stressful, comparing them to an exam and expressing a need for more guidance on treatment due to uncertainty about the adequacy of their actions.

Dialogue meetings were well-conceived and contributed to maintaining engagement, but they were challenging to implement due to timing conflicts and limited resources. Moreover, although meetings were tailored to the project setting and educational in nature, incorporating cases would have been advantageous.

Discussion

GPs were generally positive towards DataSam, reporting population overview, treatment optimisation, workflow improvements and confidence in data-driven decisions. However, technical issues, time constraints, concerns about overtreatment and de-prioritisation of other diagnoses were encountered.

Comparison with previous findings

The interviewees expressed improved workflow through structured data usage. This aligns with a review by Tsang et al. [12], emphasising data-driven optimisation for clinical improvements and highlighting the role of real-time, actionable electronic feedback. Thereby, improving the perceived advantage by enabling goal-setting, problem-solving and ownership, making clinical improvements more likely [12]. Our findings, indicating that clinics that adjusted intra-organisational workflows were more motivated to act on population data, align with Tsang et al.’s emphasis on integrating action plans into quality development efforts rather than focusing solely on measuring performance [12].

General practices in DataSam experienced challenges in a data-driven quality development process, much like shown by previous reports [13]. The GPs noted a heightened awareness of discrepancies between their ideal and actual performance. They expressed concerns about challenges to systematically acting on their data, including a perceived misalignment between population-level quality targets, patient-centred care and competing priorities at individual and clinical levels. These concerns are not unique [12, 13]. However, most clinics were motivated to adopt a systematic approach and achieve population-level quality targets for diabetes (Table 4). This improved the patients’ clinical outcomes related to systolic blood pressure, low-density lipoprotein and ischaemic heart disease [8].

In one interview, a clinic expressed concern that the intervention could lead to increased medication use, prompting reflections on the risk of overtreatment and overdiagnosis. The clinic questioned whether all patients required pharmacological intervention, asking: “Should an 80-year-old be prescribed SGLT2?” While the intervention was seen as supporting a greater focus on treatment targets, it also risked overshadowing other important clinical factors, such as patient age, polypharmacy and economic considerations. The clinic described how data-driven actions can steer clinical attention in a particular direction why the general practice must balance this with professional judgment, emphasising the importance of navigating between competing priorities and evolving guidelines.

Study strengths and limitations

This study provided insights into how increased use of population data can enhance T2D treatment and workflows. The longitudinal qualitative design allowed us to capture changes over time, yielding rich experiences and challenges. However, the study also had limitations. DataSam was a feasibility study that involved voluntary participation from a specific area in the Capital Region. Participants may have had a prior interest in population data and a slightly lower T2D prevalence (4%) than the national average (6%). Our results may reflect a particular interest in data, potentially overlooking clinics with higher frustration levels, implementation challenges or limited outcomes. These factors may affect the generalisability of the results.

One interview revealed a potential barrier associated with the researcher’s observational role and the audio recording. The clinic reported that being observed heightened their sense of pressure, making the visits feel more like an evaluation in a vulnerable position than a routine interaction. This illustrates how the presence of a researcher can unintentionally influence the dynamics of clinical encounters.

Perspectives

The findings of this study should encourage diabetes quality organisations to advance the implementation of a data-driven quality development process. The methods used in this study are likely adaptable to other general practice patient groups with chronic diseases. However, clinics without systems like the FLP might find implementing such changes challenging. Future research should prioritise identifying key challenges to effectively implement the FLP.

Conclusions

The study demonstrated that structured data use in general practices can enhance T2D treatment, workflows and staff confidence in data-driven decision-making. Even though all clinics experienced positive changes, challenges were evident in relation to technical issues, time constraints and concerns about overtreatment and data misuse. The study highlights the potential of using population data to optimise care, although further attention to implementation is needed.