How to Improve Programmers’ Workflow Process
Efficiency is everything in a statistical programming environment. “The first rule of any technology used is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.” — Bill Gates
Complex, lengthy work processes and high volumes of data will bring challenge to any programming environment. A typical band aid solution is hiring additional staff to address new workload requirements before doing any analysis to ensure efficiency and quality.
Well engineered processes are absolutely essential for a lean, effective statistical programming environment. Only by conducting a proper business analysis can you be sure your programming environment will deliver timely and quality product with the proper staffing, effective processes and appropriate technology. The goal is a compliant statistical computing environment.
Here is a suggested strategy:
- Identify a mission for your efficiency project. For example, “To build quality processes and reduce the time needed to produce clinical study reports”.
- Break the project up into phases. A first phase can be to conduct the analysis of the current situation.
- Interview staff, review current work processes and existing programmers addressing various work processes.
- Produce a report explaining the current environment along with recommendations comparing old processes to new suggested processes and the benefits.
- Identify current costs and where savings without compromising quality can be achieved.
- Obtain feedback from staff; have discussions with management.
- Obtain approval to correct problems.
- Implement new proven processes that will bring quality and savings.
Some corrective actions we have suggested in conducting assessments of excellence in statistical computing environments have been to automate repetitive manual processes, eliminate unnecessary and repetitive steps and modify and implement standard operating procedures (SOPs). We standardize processes for best practices and apply software solutions appropriately.
Analyzing a programmer’s work environment and converting it to a compliant statistical computing environment can be a complex job. An example of one such project identified five areas recommended for improvement that was estimated to take nine months to complete. Some of our recommendations are included below:
- Change departmental processes so work flow is optimized. Examples of improvement opportunity was to deliver a statistical analysis plan before data was collected, combine reviews with plans, include programming validation and implement a single review of tables, listings and graphs.
2. Create an automated process for extracting data from the clinical data management system to gain access to clinical data.
3. Design a comprehensive statistical programming environment with a standard project directory structure to apply to all projects and protocols. Train programmers and statisticians.
4. Implement Standard Analysis Dataset Structures.
5. Expand macro-driven reporting and bring it under more structured control. Implement a Report Macro Library.
The new statistical programming recommendations were implemented in nine months. The recommendations reduced the data processing time to a third of the original time. The number of programmers required was reduced by 50 percent. Our client improved their statistical computing environment significantly and showed a $2M savings from these improvements.