Family doctors can take more than a month to record ovarian cancer, once diagnosed by a specialist, in one in 10 cases, indicates research published in the launch issue of the new online journal BMJ Open.
Ovarian cancer was also incorrectly or prematurely classified in 11% of cases, the data show.
The authors base their findings on the "free text" data available in patient records, which are submitted to the General Practice Research Database (GPRD).
The GPRD contains long term anonymised medical data on more than four million patients on the lists of a representative 500 primary care practices across the UK.
The information, which is created during the course of GP consultations or correspondence relating to specialist referrals and diagnostic tests, is widely used for studies on aspects of disease and drug safety.
The authors focused on the Read codes relating to a diagnosis of ovarian cancer in women aged 40 to 80, between 2002 and 2007.
Read codes are used to code symptoms and diagnoses electronically in primary care in the UK. They provide the option to enter free text for each code, such as a letter from the specialist and a date when the 'event' occurred.
The authors also looked at this free text information to establish the time lag between when the specialist notified the GP of the diagnosis and the date it was officially coded as such in the patient's record.
There is a good deal of free text in the database, but it is not always coded by GPs nor accessed by researchers, say the authors.
During the study period, 344 women were given one of three Read codes indicating a diagnosis of ovarian cancer.
Most cases (90%) had free text relating to ovarian disease, and in almost two thirds (64%) this confirmed a diagnosis of ovarian cancer.
But in one in five cases (22%) free text information confirmed this diagnosis before it was officially coded as such in the patient's record. In one in 10 cases more than four weeks had elapsed.
The free text information also showed that mistakes been made in 11% of the records.
Four cases coded for ovarian cancer did not have the disease at all; in another six cases the cancer was a recurrence which was not evident in the coding. And in 35 cases, patients had been coded before a definitive diagnosis had been made
"For diseases which rely on hospital consultants for diagnosis, free text (particularly letters) is invaluable for accurate dating of diagnosis and referrals and also for identifying misclassified cases," conclude the authors.