Data Quality - You Are What You Eat
The influence of source data on reporting quality
Casey Wike
Data Coordinator
Data is everything, but what we sometimes lose sight of is the importance of quality data. Quality data is the basis for sound decision-making.
As MetricAid’s Data Coordinator, I review the emergency department (ED) data imported from clients and understand first hand the large variance in quality. As such, I’ve put together a list of indicators to watch for when compiling data to ensure that your final output is accurate, fulsome and relevant to your decision making.
Characteristics of Quality Data
Quality data is consistent. Data should be collected using the same units, formats, and layouts and each row should have a uniform configuration to facilitate comparison and categorization. Dates, for example, should be formatted identically in terms of the year, month, day, time, and the use of decimals. Consistent data expedites quality reporting.
Quality data is complete. The scope of your dataset should include all of the data necessary to draw valid conclusions. Make a checklist of all necessary data points; this list will vary depending on the needs of your report. When finished, revisit the checklist to make sure nothing is left incomplete or forgotten. Data can always be removed, but it cannot be created.
Quality data is valid. Establish a consensus among those knowledgeable in the subject matter that the data being considered is factually accurate. Invalid data will fuel poorly drawn conclusions and result in deficient decisions. A report is only as accurate as its source data.
Quality data is timely. Solutions to current problems should be based on current data. Relying on outdated data will lead to decisions that no longer satisfy your department's needs. Actioned old data will solve old problems.
Quality data is relevant. If your ED is making a decision about its scheduling needs, data on the performance of all Canadian emergency departments is unlikely to reflect the needs of your own group. Each dataset should directly relate to the decision being made. Ensure your data is responding to the questions being asked.
Producing Quality Data
When producing data regularly, consider the following tips to improve the quality of your output.
Check with the recipient. When possible, talk with the intended recipient to make sure that what you are about to produce is what they are expecting and/or what they need.
Check your data for outliers. Determining whether or not an observation is an outlier is ultimately a subjective exercise. A good indicator of a possible outlier is a data point that is significantly higher or lower than most of the other values. Sometimes there are legitimate reasons for such values and if you can think of a reason why an outlier is valid you can keep it in the dataset. Otherwise, these outliers should be discarded.
Check for data completion. Check your dataset to make sure that every possible data entry is accounted for. If your dataset represents patient-flow performance metrics, make sure to include data for each patient that is relevant to the report. Failure to include all relevant pieces of data can result in bad reporting, and ineffective decision making.
Check for extraneous data. Always keep in mind the end goal of your data analysis. When collecting data, ask yourself whether or not that data directly relates to the purpose of your reporting. Is it necessary and useful, or will it cause unnecessary confusion and compromise the quality of your data? A dataset needs to be complete, but it also needs to be concise.
Check the format. Deliver your data in the right format. Don’t send a “.csv” file if what’s needed is an Excel file. This will help to expedite the data collection and analysis process by limiting unnecessary back and forth between you and the recipient.
Automate when you can. If you are delivering a standard dataset on a regular basis, automate the process. If your dataset is produced by querying a large database, automate the query so that it runs at regular intervals. A little bit of early effort can save you time and effort in the future.
Follow up. After each delivery of your data follow up with the recipient to make sure that their expectations and needs are met. If they aren’t satisfied, follow through by taking the steps necessary to rectify any problems. Building a strong relationship with colleagues and clients will pay dividends in the long term.
What this Means for MetricAid
MetricAid’s Advanced + EM clients directly benefit from the quality data we collect. We use patient flow and department performance data to build schedules that increase hourly patient assessments, reduce patient wait times, reduce physician burnout and improve overall work/life balance. Our Advanced + EM clients also have access to a robust set of reports based on the data provided to us, such as:
Physician Group productivity
Physician individual productivity
Individual shift productivity
P4R metrics
Customized shift templates
Staffing levels vs staffing needs
Quality data is a core part of our business and it’s imperative that we receive regular and accurate information from our clients to properly create finely-tuned schedules that address everybody’s needs.
If you want to learn more about our Advanced + EM service, contact sales@metricaid.com to start a conversation.
Cheers,
Casey Wike, Data Coordinator
