6 Data Entry and fieldwork quality control
6.1 Once the survey was issued, paper copies of questionnaires received were scanned on a daily basis by staff at Ciconi Ltd. A verification process was then carried out for each batch scanned and a number of checks were undertaken to ensure that the scanning process had worked correctly.
6.2 To ensure clarity of information and ultimately accuracy, the following were set up and tested before the data entry began:
- Data structure
- Data entry spreadsheet
- Data capture instructions
- Capture questionnaire
6.3 The people entering the data were required to enter data into a test environment prior to commencing work. The test data files were individually checked, matching each questionnaire to each record, to ensure accuracy. Once the test files had been verified and approved, the person was allowed to commence work. A number of formal procedures were used to help increase accuracy including name of the person who entered the data recorded against each entry and the unique reference numbers that link a survey to a patient were entered twice.
Verification & upload process
6.4 The data entry system ensured that only valid answer codes for each question could be entered and that the correct data appeared in each field. Other checks included ensuring that numeric data was the correct format and that fields were not truncated in error.
6.5 Once confirmation had been received that the scanned batch of completed surveys had been verified; trays containing the hardcopy forms were tipped into a locked cage and stored for collection. Periodically the cages were collected and sent for secure shredding. Batches that were destroyed were logged and a certificate of destruction was received.
6.6 The survey asked respondents if there was anything else that they would like to tell us about their experiences of their local GP or other local healthcare services, or care and support services.
6.7 Just over 90,000 comments were left with approximately 50,000 relating to the GP practice, roughly 28,000 relating to other local NHS services, and roughly 14,000 relating to care and support services.
6.8 Disclosive details that could be used to identify patients were suppressed when the comments were entered. These details included names, addresses, ages, dates, medications and medical conditions. Staff names were also suppressed.
6.9 Checks were undertaken on 10% of records to ensure that the free text comments were linked to the correct record and that the instructions for suppressing disclosive details were followed.
Email: Andrew Paterson