Ensuring Robust Data Quality for Quality Improvement Projects

Why this topic matters

Imagine implementing a new pathway based on data showing a high incidence of a particular complication, only to discover later that the data was inaccurately reported. Or perhaps you're trying to demonstrate the impact of a ward-based intervention, but inconsistencies in data collection for your chosen metrics make it impossible to draw clear conclusions. These scenarios highlight why data quality is not just a technical concern but a fundamental enabler of effective QI.

Poor data quality can lead to:

• Misguided interventions: Acting on incorrect information can waste valuable resources and potentially worsen outcomes.
• Inaccurate measurement of impact: Without reliable baseline and post-intervention data, it's impossible to know if an intervention has truly made a difference.
• Loss of credibility: Projects built on shaky data may lose buy-in from staff and leadership.
• Increased workload: Time spent correcting errors or trying to interpret unclear data diverts resources from actual improvement work.
• Governance and reporting risks: Inaccurate data can hinder compliance with reporting requirements and impact organisational assurance.

What constitutes 'good' data quality?

Good data quality is often described using several key dimensions:

• Accuracy: Data is correct and reflects the real-world situation.
• Completeness: All required data points are present; there are no significant gaps.
• Consistency: Data is recorded uniformly across different sources and over time.
• Timeliness: Data is available when needed for decision-making.
• Validity: Data measures what it is intended to measure.
• Reliability: Data collection methods consistently produce the same results under similar conditions.
• Uniqueness: No duplicate records exist for the same entity.

Practical explanation

Data quality is not an accident; it's the result of deliberate planning and ongoing vigilance throughout the lifecycle of a QI project. It starts at the very beginning when you define your project aims and metrics, and it extends through data collection, analysis, and reporting.

Sources of poor data quality

Understanding common sources of data quality issues can help in proactively mitigating them:

• Human error: Mistakes during data entry, transcription, or interpretation.
• Lack of clear definitions: Ambiguous definitions of metrics or data fields can lead to inconsistent recording.
• Poorly designed collection tools: Forms, spreadsheets, or electronic systems that are difficult to use, or lack validation rules.
• System integration issues: Problems transferring data between different IT systems.
• Inadequate training: Staff collecting data may not understand its importance or how to do it correctly.
• Workload pressures: Staff may rush data entry, leading to errors or omissions.
• Lack of feedback: Data collectors may not be aware of issues with the data they are submitting.

Step-by-step approach to ensuring data quality in QI

This structured approach can help embed data quality into your QI project from conception to completion.

Step 1: Define your measures and data requirements clearly

Before you collect any data, you must be absolutely clear about what you are measuring and why. This is foundational to data validity and accuracy.

• Develop clear, unambiguous definitions: For each metric, define exactly what it represents, how it will be measured, and what units will be used. Ensure all terms are uniformly understood across the team.
• Identify data sources: Determine whether you will use existing routinely collected data (e.g., from electronic patient records, national databases) or if you need to collect new primary data.
• Specify data collection points: Pinpoint exactly when and where the data will be gathered.
• Establish inclusion and exclusion criteria: Define the population or events that will be included or excluded from your data set.

Step 2: Design and pilot your data collection methods

Thoughtful design of data collection tools and processes can prevent many data quality problems.

• Standardise collection tools: Whether it's a paper form, a survey, or an electronic template, ensure everyone uses the same tool. Incorporate clear instructions and fields.
• Build in validation: For electronic systems, implement input validation rules (e.g., only numbers for age, specific date formats). For paper forms, design fields to minimise ambiguity (e.g., tick boxes instead of free text where possible).
• Pilot test: Before full-scale data collection, conduct a small pilot. Test the collection tool with a few cases or team members to identify any ambiguities, difficulties, or missing fields. Refine based on feedback.
• Consider automated data extraction: Where possible, leverage existing digital systems for automated data extraction to reduce manual entry errors. However, validate the extracted data against original sources.

Step 3: Train and support data collectors

The people collecting the data are your first line of defence against poor quality.

• Provide comprehensive training: Ensure all staff involved in data collection understand the definitions, the collection process, the importance of accuracy, and the overall purpose of the QI project.
• Offer ongoing support: Be available to answer questions and address issues as they arise during data collection.
• Emphasise the 'why': Help staff understand how their accurate data contribution directly impacts patient care and the success of the project.

Step 4: Implement data quality checks and monitoring

Data quality is not a one-off task; it requires ongoing monitoring.

• Regular spot checks: Periodically review a sample of collected data against original sources (e.g., patient notes) to verify accuracy and completeness.
• Range and consistency checks: Use simple analytics to look for outliers, impossible values (e.g., age over 150), or inconsistencies (e.g., discharge date before admission date).
• Missing data analysis: Track the proportion of missing data for key fields. If a field consistently has missing values, investigate why and address the root cause.
• Data review meetings: Schedule regular sessions where the QI team reviews data performance, discusses quality issues, and agrees on corrective actions.

Step 5: Document and communicate

Clear documentation and open communication are vital for sustaining data quality.

• Create a Data Dictionary: This document should contain definitions for all data fields, units of measurement, acceptable values, and collection instructions. Share it widely.
• Log data quality issues and resolutions: Keep a record of any data quality problems identified and the steps taken to resolve them. This helps prevent recurrence and informs future projects.
• Provide feedback to data collectors: Share the results of quality checks and thank staff for their diligent work. If issues are found, use them as learning opportunities rather than punitive measures.

Example in clinical practice

A multidisciplinary team on a surgical ward identifies a high rate of post-operative urinary tract infections (UTIs) in patients undergoing elective hip replacements. They decide to initiate a QI project to reduce this rate.

Data Quality in Action:

1. Defining Measures: The team clearly defines 'post-operative UTI' using specific diagnostic criteria (e.g., positive urine culture with specific bacterial counts, symptoms requiring antibiotic treatment). They define 'time to diagnosis', 'antibiotic prescribed', and 'length of stay post-op'.
2. Collection Tool Design: They create a standardised collection form, integrated into their electronic patient record (EPR) system where possible, with mandatory fields for key UTI diagnostic results and treatment details. For new entries, a drop-down menu is used for common antibiotic choices. They pilot this form with a few recent patient records.
3. Training: All surgical nurses and junior doctors involved in patient care receive training on the precise definition of a post-op UTI and accurate documentation within the EPR/form. Emphasis is placed on the importance of timely and complete data entry for the QI project.
4. Monitoring: The QI lead runs weekly reports from the EPR system to check for completeness of UTI-related fields and unusual entries. For instance, they check that 'antibiotic start date' is not before 'UTI diagnosis date'. Any anomalies are flagged and reviewed against the original patient notes by a designated nurse.
5. Documentation: A 'UTI Data Dictionary' is published on the ward's shared drive, outlining all definitions and collection guidance. Learning points from data quality issues (e.g., initial confusion between 'colonisation' and 'infection') are shared at ward meetings, reinforcing best practice.

By meticulously managing data quality, the team ensures that the observed reduction in UTI rates (or lack thereof) accurately reflects their intervention's impact, allowing them to make informed decisions about continuing, modifying, or scaling their improvement efforts.

How Lazomis can help

Lazomis offers tools that can significantly support data quality within your QI projects:

• QI Project Setup: Our structured project planner guides you through defining clear metrics, data sources, and collection plans from the outset. This ensures clarity and reduces ambiguity.
• Data Collection Forms: Lazomis enables the creation of customisable digital data collection forms with built-in validation rules, helping to ensure accuracy and completeness at the point of capture.
• Dashboards & Visualisations: Real-time dashboards allow you to monitor data completeness and consistency, helping you spot anomalies quickly and address them before they impact your analysis. Trend charts can highlight unexpected deviations, prompting further investigation into data quality.
• Reporting Tools: Integrated reporting features facilitate the documentation of data definitions (data dictionary) and can highlight areas where data quality might be suboptimal, guiding focused improvement efforts.

This resource supports, but does not replace, clinical judgement. Local policy, formulary and specialist advice should be followed.

Key takeaways