Quality performance and associated factors in Swiss diabetes care – A cross-sectional study

Introduction Quality indicators and pay-for-performance schemes aim to improve processes and outcomes in clinical practice. However, general practitioner and patient characteristics influence quality indicator performance. In Switzerland, no data on the pay-for-performance approach exists and the use of quality indicators has been marginal. The aim of this study was to describe quality indicator performance in diabetes care in Swiss primary care and to analyze associations of practice, general practitioner and patient covariates with quality indicator performance. Methods For this cross-sectional study, we used medical routine data from an electronic medical record database. Data from 71 general practitioners and all their patients with diabetes were included. Starting in July 2018, we retrieved 12-month retrospective data about practice, general practitioner and patient characteristics, laboratory values, comorbidities and co-medication. Based on this data, we assessed quality indicator performance of process and intermediate outcomes for glycated hemoglobin, blood pressure, cholesterol and associations of practice, general practitioner and patient characteristics with individual and cumulative quality indicator performance. We calculated odds ratios (OR) and 95% confidence intervals (CI) using regression methods. Results We assessed 3,383 patients with diabetes (57% male, mean age 68.3 years). On average, patients fulfilled 3.56 (standard deviation: 1.89) quality indicators, whereas 17.2% of the patients fulfilled all six quality indicators. On practice and general practitioner level, we found no associations with cumulative quality indicator performance. On patient level, gender (ref = male) (OR: 0.83, CI: 0.78–0.88), number of treating general practitioners (OR: 0.94, CI: 0.91–0.97), number of comorbidities (OR: 1.43, CI: 1.38–1.47) and number of consultations (OR: 1.02, CI: 1.02–1.02) were associated with cumulative quality indicator performance. Conclusion The influence of practice, general practitioner and patient characteristics on quality indicator performance was surprisingly small and room for improvement in quality indicator performance of Swiss general practitioners seems to exist in diabetes care.


Introduction
(line 49) typo: "The use of quality indicators (QI) and contingent incentives aim to improve process and outcomes in clinical practice" The following paper may also be useful background: • Huang, J., Yin, S., Lin, Y., Jiang, Q., He, Y. and Du, L. (2013), Impact of pay-for-performance on management of diabetes: a systematic review. JOURNAL OF EVIDENCE-BASED MEDICINE, 6: 173-184. doi:10.1111/jebm.12052 (line 61) You state that: "Currently, a randomized controlled trial (RCT) testing the P4P approach in Swiss primary care using clinical routine data is ongoing (15). The data of this trial offer the opportunity to study the characteristics and QI performance of the P4P study population and to analyze associations of QI performance with practice, GP and patient covariates." While I agree this provides justification for the trial, it does not clearly provide justification for the current paper i.e., while it is important to examine the association of different factors (patient, GP, practice level etc) and quality performance. I would suggest including a line here to justify the current study e.g. did it inform the decision to conduct the trial or will it inform the trial and if so, how? I see that in the Methods you mention that data collection for the current study was conducted as part of the trialit may be worth clarifying this earlier i.e., that the trial was progressing regardless but that it provided an opportunity to explore this association between patient, GP, practice factors and QI.
(line 57) Again, relating to study justification, you state "From literature, we know that not only GP, but also patient characteristics influence QI performance (7-10)". Since the association with practice factors is also mentioned later, should this be introduced earlier?

Methods
Since you mention the introduction of P4P in Switzerland and how the current study uses data from a trial of P4P, I think it would be important to provide some more detail here about the P4P programme e.g. what it involves and when it was introduced. When reading this section it is not immediately clear whether the data you use comes from baseline before P4P was introduced or the control group, or whether you are using all data returned to the database from 2017 to 2019 during which P4P was in place. To help the reader it may be useful to provide a figure or timeline, showing PONE-D-19-32658 Reviewer comments data collection points etc. Figure 1 shows the flowchart and I think looking at this that you used data from the retrospective 12-month period and not the intervention group. If the latter is the P4P group then this suggests your data comes from practices and patients who are not within a P4P system -this should be made clearer.
Some acronyms might be unfamiliar to readers (e.g. ICPC, EMR), therefore would suggest spelling these out where possible. Maybe I missed it, but you do not appear to state whether these are patients with type 1 or type 2 diabetes or both types are included in your sample. On this point, did you consider conducting the analysis separately by type or adjusting for this variable in your regression models?
In terms of the statistical analysis, were any further covariates included in the models or just those listed? I would suggest a footnote on the table to clarify these are adjusted OR and what variables they were adjusted for.
You should include an explanation of how cumulative QI performance was calculated -do you mean a composite QI here? Was there a scoring system? I know it is mentioned in the introduction, but in connection with the previous comment I would suggest listing which specific QI were used for the current study in the Methods section. You mention the type of data collected but are not explicit about which data corresponds to the QI to examined. PONE-D-19-32658 Reviewer comments (line 110) You state one objective is "To examine the association of QI performance with practice, GP and patient covariates" (see typo here also), however in the analysis section practice factors are described and listed under 'GP characteristics'. Therefore, this leads me to think that these factors are being explored at the GP level -I would suggest you keep the terminology consistent throughout the paper, so either describe these as three (patient, GP, practice) or two (patient, GP) separate variable categories or levels and be clear about what comes under each.
(line 120) typo: "Number of medications and BMI were not considered as covariates for the regression model as they were used to identify certain comorbidities"

Results
If space allows I think it may be helpful to have a descriptive Table 1 showing the profile of GPs and/or patients. For example, much of the text about the different patient medications and comorbidities could be captured in a table rather than listing in the text.
It is good practice to provide the % and the N -if you have your descriptive table(s) as suggested then you can refer the reader here for the numbers.
(line 132) typo: "These patients were in 57% male and had a mean age of 68.3 years (SD 13.4)." (line 137) typo: "On average, patients had eight consultations ) at the GPs' practice (or 'a GP practice') in the 12 months preceding baseline." In the text it is not always clear whether you mean targets achieved or processes performed. These are different and should be made explicit when reporting the results. For example, in line 161 you state "higher age was significantly associated with a higher HbA1c QI performance but with a lower cholesterol QI performance" . Should 'QI performance' in this case be taken to mean both measurement of HbA1c (process) and achievement of <7.5% (target achieved)?
In Table 1 you show the % patients achieving QI. There are BP and HbA1c targets mentioned (i.e. <140/85mmHg, <7.5%, < 5 mmol/l). These should have been mentioned in the Methods when explaining the QI to be examined. I realise these are common targets in diabetes management, but I PONE-D-19-32658 Reviewer comments feel it would be important to provide a rationale for using them i.e., are they used in national and/or international guidelines. Later in the manuscript you mention the American Diabetes Association guidelines. If the targets you use are based on these guidelines I would suggest citing them.
(line 146) "93.8% of the patients had at least one comorbidity, whereas 37.9% of the patients had three or more comorbidities." -do not begin sentences with a numeral. Please amend this in the rest of the manuscript.
In the section titled 'Associations with QI performance' I would report the OR and 95% CI in the text, particularly the significant findings, perhaps referring the reader to the table to review the nonsignificant findings. I suggest rephrasing line 159 "GP characteristics had almost no significant influence on QI performance" as is this is potentially confusing. It would be better to say: 'Some GP characteristics were associated with QI performance'. I would also advise revising this expression at the start of the discussion.
Similarly, I think saying 'lower' and 'higher' and 'lowered' QI performance (lines 158 -166) is potentially confusing as this implies a continuous rather than binary outcome. You are examining the association of these factors with the odds of either performing a process or achieving a target, so I would reword the text throughout this section to make this clear.
(line 168) 'Cumulative QI' again is mentioned but it is unclear what is meant here -looking at this table it seems the individual QI are again reported rather than a summary or composite.

Discussion
(line 182) "In this study, we explored associations of QI performance with patient, GP and practice characteristics." -I would rephrase this line, as earlier in the manuscript you lead with the exposure (i.e. GP, practice, patient factors) and finish with the outcome (i.e. QI performance) and this is the convention. In general, I would check this phrasing throughout the manuscript to ensure consistency -I believe the order may be reversed in the results also. PONE-D-19-32658 Reviewer comments I suggest rephrasing this sentence (line 188) "Number of years since diabetes diagnosis was much shorter in our study, which is explicable through limited data availability before GPs participated in the FIRE project." as it is unclear what you mean -is it the case that people may have been diagnosed earlier but you do not have access to data on these people? I would not use the phrase "well above 90%" (line 196). Either provide a range or simply state they were over 90%. In general, it is better to give readers a sense of the scale i.e. %. This also applies to the next line where you comment that the other Swiss study found "slightly higher" rates.
(line 200) You state that "Reasons for such poor performances might be clinical inertia to intensify treatment, poor patient adherence or misjudgment by our QI scheme (25,26)." -I would be interested to know whether you think there is something about the patient or GP population studied which might explain why issues such as inertia would be more common, than, for example, in the Djalali et al. 2014 study which you say is the same EMR database. Does this imply these are similar patients and GPs?
On this point, you mention in the strengths and limitations, using the "full spectrum" (line 236) of patients and that "could show high generalizability of our data due to given patient characteristics." -I feel the descriptive Table 1 in the Results showing the patient profile would be helpful. In the text you only describe patient in terms of age, sex and diabetes duration, therefore the reader may not understand how this compares to the general population with diabetes. In the strengths and limits you mention 'potential missing data' (line 244). It would be helpful report in the Results whether any of the variables you used had missing data. As you only report % in the text it is difficult for the reader to determine the level of missingness.
The meaning of this sentence (line 250) is unclear and needs to be rewritten "This bias is probably of larger issue with lower performance rates as in QI for cholesterol." (line 252) You mention "We tried to minimize the amount of missing data by setting a minimum standard of data availability for each GP to be contacted for recruitment." . This is important and should be mentioned earlier in the Methods section when explaining participants. From the way you describe it in results "125 GPs met the data availability criteria of a) delivering data on a regular PONE-D-19-32658 Reviewer comments basis since 01.01.2017" it was not clear to me that there was a threshold, or what is meant by 'regular basis' -do you mean they had to deliver a complete dataset annually or at some other interval?
I think there is potential to expand on your overall implications and conclusions and make these stronger and better situate them in the broader literature. For example, given there were a number of covariates which you could not explore in your analysis: 1) do you feel there is scope for future work (and what would this look like?), 2) is there scope to expand on (e.g. to include further variables) or improve (e.g. your comment about missingness) the type of data collected in the FIRE database, 3) do you think this database should or can be used for monitoring quality of care and providing feedback to GPs/practices in the future?
As I commented earlier there needs to be a more explicit justification for the current study in the Introduction. Once this is made clearer it will be easier for you to link your conclusions and implications back to the reason for conducting the study. For example, considering your conclusion regarding the 'room for improvement', was your initial aim to determine whether the quality of Swiss diabetes management needs to be improved and how to do so and/or identify factors to inform case-mix adjustments? How do your findings inform the P4P trial? Earlier in the Introduction you mention the lack of understanding around QI performance within P4P systems, particularly in Switzerland. However, in the Discussion there is no mention of P4P and how your findings relate to research from other countries which have introduced P4P -for example, the Quality and Outcomes Framework in the UK. I understand this omission if you used data from the patients and GPs who were not in the P4P intervention group. However, even if you looked at QI performance in a group who are not within P4P system, still I think you need to be clearer about the purpose and relevance of this current study, how it relates to the trial and how it fits in with the bigger picture.