Four disciplines behind dependable patient data

A hospital group's patient index is only as good as the registration desks that feed it. In Kerala and Karnataka, the same person may give a house name, an initial or a father's name in a different order on each visit, and many families share one mobile number. I set matching rules with medical-records leads, not only with the system vendor, and I separate certain matches from uncertain ones. Uncertain pairs go to trained reviewers who can see both records' clinical history. I also insist on a documented way to unmerge, because some errors survive any review. I measure the work through wrong-merge reports and duplicate registrations per thousand visits, and I report both to the clinical governance committee. Desk staff receive feedback on their own error rates, which has done more than any instruction from my office.

Insurers and third-party administrators read our records after the patient has gone home, and they read them strictly. Having spent four years on their side, I know which gaps lead to deductions: an extended intensive-care stay with no recorded reason, an implant without its supporting documents, or an ICD-10 diagnosis that does not explain the procedure performed. My team codes discharge summaries and audits a sample every month. I treat coding as the last step, not the remedy. Where the clinical note is incomplete, the correction belongs with the clinician who wrote it, supported by structured fields that take seconds to complete. I do not accept coding that describes more than the record supports. A higher claim value reached that way would be a liability, however it looked in a monthly revenue report.

Under the Digital Personal Data Protection Act, a hospital must be clear about why it processes personal data and what it tells patients, and the Act's obligations are coming into force in stages. The Ayushman Bharat Digital Mission adds another layer: health account numbers, registries of facilities and professionals, and consent-based sharing of records between providers. I explain these frameworks to our teams in general terms and take specific legal questions to counsel. In practice I ask three things of any new use of data: what the patient was told, whether they can refuse without any effect on their care, and who checks that we honour the answer. Our registration notices are printed in Malayalam, Kannada and English. Requests for patient-level data from research partners or vendors go to a review group that includes a clinician, counsel and the privacy officer.

Bed turnaround, theatre utilisation and discharge timing are everyday hospital measures, and they are easy to distort. A discharge recorded at the moment of a doctor's decision looks better than one recorded when the patient leaves the ward. I agree each definition with the department that generates the data and publish it beside the figure. Every measure has a named owner in the hospital, usually a nursing superintendent or a department head, who can correct the source rather than ask my team to adjust the report. We compare hospitals only where definitions match. When our Mangaluru hospital's theatre utilisation appeared ten points higher than Kochi's, the gap came from counting cleaning time differently. Ward leaders trust figures more when they can see how they are made and when a correction does not count against them.