Skip to main content
Dear Dr Behan,
Thank you for taking the time to appraise our study.
You've highlighted many of the limitations when relying solely on data from EMRs (i.e. you have no sight of those not turning up in the data).
Hence I understand what you mean by the term "attendance rate" ideally being reserved for calculations where you can factor in the population level visitation behaviour, which we of course could not. Therefore, a limitation of the study might read "As this study only examined attendance behaviour of patients presenting to GP services, it is not possible to comment on overall population attendance rates".
Thus, perhaps use of the term "rate" in the Abstract and Results might distract from the thrust of the article. However, it is this thrust of the article ( i.e. more patients turning up, and more frequently) that is covered in detail in the discussion.
Also, Table 1 does point out we are discussing "visits per patient", which has no connotations of gauging the behaviour of the population.
I also agree that perhaps Table 1 might be improved if we removed the "Total" rows as some patients jumped between cohorts (which is particularly apparent in the "Year After Policy Change"). As you say the average visits per patient should not be greater than the subgroups.
However, we were careful to include the explainer point "b" in this Table to clearly identify this issue.
Regarding the use of different figures as denominators for the various calculations, I don't feel there is any optimal way to work with imperfect data, where there is limited visibility on individual patients moving between patient cohorts, services and those who may feature in one year of data collected and not in the other. With this is mind, and given the transient nature of patients in terms of the services they attend (especially "out of pocket" patients, and particularly in the "out of hours setting"), we felt it was a reasonable decision to group patients together first by year and then by the other parameters outlined (as this {1 yr of visits -> policy change -> 1 yr of visits} approach captures the sudden change on 1st July 2015 where patients could avail of free GP services).
Since publication of the article thankfully in Ireland we have seen the kind of large scale population surveys that are needed to address (chiefly the Healthy Ireland surveys https://www.gov.ie/en/collection/231c02-healthy-ireland-survey-wave/) some of the issues highlighted herein (i.e. what the various age cohort population attendance rates are).
Your modified table, using data from other sources, is logical and could indeed be combined with such Healthy Ireland data in future work.
However, I do feel the article's main message is still valid and valuable as it clearly shows that free visits leads to increased workload for GPs.
As to whether the issues outlined in your response (and in this reply) represent "errors" or practical decisions around data analysis and presentation is a matter for the reader to consider, based on the evidence presented in the article and this discussion.
Regards,
Dr Michael O'Callaghan