Applications of Data Collected in Clinical Management Systems

Your clinical practice system stores a lot of useful data and offers you the opportunity for applying clinical epidemiological methods for auditing the quality of care you provide, and in enclosed populations examining population epidemiology within a town. The capacity for population studies depends on the coverage of the population by the practice – the ideal being a single practice covering the entire town. If there is a small group of practices consider pooling your data but beware of duplications.

Data quality

Even in the best managed systems there will be data errors these should be checked for on a regular basis and eliminated as found. Typical errors are:

  • Duplication of person due to name misspelling
  • Duplication of person due to birthdate error
  • Sex reversal or absence of classification
  • Blending of persons due to same name (due to failure of checking before recording information)
  • Use of free text for recording
    • reason for prescribing
    • reason for encounter
    • disease classification
  • multiple medication names
  • failure to delete medications no longer used
  • failure to record chronic illness
  • failure to record family history.

The best way to eliminate medication errors is to prescribe generically unless you mark every prescription that brand substitution is not to be undertaken. Beware however, for pharmacists will tell patients that they cannot obtain the particular brand for ..... weeks and ask them if they can substitute without informing you. That this leads to medication errors is obvious but that is government policy as determined by the pharmacy lobby. The ideal practice system would allow you to prescribe whatever brand you like but record the generic name as well for auditing purposes but as far as I am aware no system offers this at present.

Overcoming data quality problems

Most of you will have well established clinical systems with a large collection of data and will find the prospect of fixing your data errors a bit overwhelming. There is always the problem of the currency of medications. Your system may record that a person has not had a medication prescribed for years, but they may be obtaining it from another source, or it may be an OTC medication hence simply eliminating all old prescriptions is not a suitable policy – the only solution is a medication review for each an every person who is chronically taking three or more medications every time they are seen.

Missing data can be filled on an ad hoc basis but this will not provide the assurance that the overall quality of data is improving. A planned review of the data and planned collection of missing information is the only solution.

1. Look at the population first

  1. Marked deceased persons as such – generating a recall for the deceased is profoundly embarrassing.
  2. Ensure the sex is correct and recorded – I once recalled a young man for a PAP smear – fortunately he had a good sense of humour and declined the offer on the grounds that it would be too painful.
  3. Ensure the date of birth is correctly recorded – again recalling a 100 year old person for their measles immunisation is a bit inappropriate.
  4. Inactivate people who no longer attend the practice – develop your own policy but a person with a chronic illness who formerly attended monthly and has not been seen for six or more months is unlikely to reattend.

The best way to conduct this is print out a list of all current patients and cross them off as they are reviewed – to save paper this could be as a spreadsheet on a laptop or palm/ pocket PC.

2. Next look at the presence of clinical flag recording

  1. Have allergies/ adverse drug reactions been recorded or marked as none?
  2. Has the family history been recorded?
  3. Have previous surgical procedures been recorded (this should be recorded separately from the general medical history but I have not seen a system which does this)
  4. Have at risk persons been identified as such (Aboriginal and Torres Strait Islanders, gestational diabetes, etc)
  5. Have all immunisations been recorded? (your receptionist staff can ensure that the “blue book” accompanies all children when they visit.
  6. Have all mothers had an obstetric history recorded? (ideally it should be possible to cross link children to their mother using the internal database reference link – as yet I have not seen a system which offers this)
  7. Non use of the clinical encounter system provided – there are various systems provided docle, ICPC etc – none have I found to be sufficiently comprehensive as most of my work has been in rural areas of Australia where I have encountered problems not thought of by the urban, and /or non Australian creators of these systems – but the classification provided is indexed, and any free text you add is not hence you will need to become familiar with the idiosyncrasies of the system you are using and try to eliminate any free text classification.

3. Next look for incongruities.

  1. Medication with no indication – when I first audited one of the practices in which I worked I found 10% of people on insulin were not recorded as having diabetes. (Medical Director will allow you to readily identify these with the database search function – use find all patients on Insulin NOT diabetes).
  2. Apparent mismedication – diagnosis of asthma with beta blocker therapy, use of beta agonists with beta blocker therapy etc.
  3. Medications with no indications – a valid indication is “Prescribed by someone else for as yet unknown reason” to cover the hospital discharges and the specialists who have not yet learned to write a letter.
  4. Investigations with no indications – Beta HCG with no record of pregnancy concern, HbA1c with no diabetes, urine albumin/Creatinine ratio with no diabetes for renal disease etc.

The in initial review of the clinical data quality may take some time, do not expect to fix everything in a few weeks, and do not start until the demographic data are correct. Depending on the age of the practice there will be between 3000 and 6000 patient records per practitioner and you will need to become familiar with the capabilities of the system you are using. As I mentioned above Medical Director version 2 offered a good database search function, but I found it easier to export the data to another database for analysis. Most of you will not have the software to do this so discuss with your software provider how you may conduct searches to look for the data errors I have listed above.

Privacy Issues.

I have been unsuccessful in obtaining a rational answer on the issue of data privacy from various offices of privacy commissioners. The attitude taken by some University ethics committees is that the data cannot be used for self directed audit even if the purpose is to improve the quality of care offered, or to identify errors and omissions affecting the quality of treatment offered to the individual.

The question revolves around who owns the information. I think there is little doubt that the medical practitioner owns the medical record, but the patient has the right to dispute the content. Where there is a medical practice company the company may own the record de facto if the practitioner has entered into a master servant relationship with that company and the practitioner agrees to use the clinical record system provided (this applies to both paper and electronic records). Neither a company, nor an individual has the right to release the information in the record system that identifies any individual to a party external to the existing arrangement to provide clinical services. Thus the individual practitioners may come and go from a company but the records will be made available to the new practitioners as this is part of the agreed system of provision of care. The company or individual practitioner may not release the information to an auditor, the government, marketers, or staff of the division of general practice without the express (written) permission of the owner of the data or the patient (or their legal guardian).

What this means is that effective audit can only be conducted “in house” as the sampling bias introduced by seeking written permission, and the necessarily small sample make any other form of audit useless – would you change a policy on the basis of a 0.1% sample of anything?

The opportunity to improve the quality of care you provide is not just in your hands – it is only you who can provide the necessary audit of your services given the current Constraints in Australia.

Posted in Australian eHealth

You need to log in to post comments. If you don't have a Pulse+IT website account, click here to subscribe.

Sign up for Pulse+IT eNewsletters

Sign up for Pulse+IT website access

For more information, click here.

Copyright © 2017 Pulse+IT Magazine
No content published on this website can be reproduced by any person for any reason without the prior written permission of the publisher.