Factors associated with COVID-19 vaccine uptake in people with kidney disease: an OpenSAFELY cohort study

Objective To characterise factors associated with COVID-19 vaccine uptake among people with kidney disease in England. Design Retrospective cohort study using the OpenSAFELY-TPP platform, performed with the approval of NHS England. Setting Individual-level routine clinical data from 24 million people across GPs in England using TPP software. Primary care data were linked directly with COVID-19 vaccine records up to 31 August 2022 and with renal replacement therapy (RRT) status via the UK Renal Registry (UKRR). Participants A cohort of adults with stage 3–5 chronic kidney disease (CKD) or receiving RRT at the start of the COVID-19 vaccine roll-out was identified based on evidence of reduced estimated glomerular filtration rate (eGFR) or inclusion in the UKRR. Main outcome measures Dose-specific vaccine coverage over time was determined from 1 December 2020 to 31 August 2022. Individual-level factors associated with receipt of a 3-dose or 4-dose vaccine series were explored via Cox proportional hazards models. Results 992 205 people with stage 3–5 CKD or receiving RRT were included. Cumulative vaccine coverage as of 31 August 2022 was 97.5%, 97.0% and 93.9% for doses 1, 2 and 3, respectively, and 81.9% for dose 4 among individuals with one or more indications for eligibility. Delayed 3-dose vaccine uptake was associated with younger age, minority ethnicity, social deprivation and severe mental illness—associations that were consistent across CKD severity subgroups, dialysis patients and kidney transplant recipients. Similar associations were observed for 4-dose uptake. Conclusion Although high primary vaccine and booster dose coverage has been achieved among people with kidney disease in England, key disparities in vaccine uptake remain across clinical and demographic groups and 4-dose coverage is suboptimal. Targeted interventions are needed to identify barriers to vaccine uptake among under-vaccinated subgroups identified in the present study.

and tested. Overall, the work is sound in design, statistical analysis, and interpretation, and my comments are therefore minor, and largely relate to presentation and interpretation.
-The description of people of non white race/ethnicity as "nonwhite" ie in relation to their lack of whiteness. This never sits well with me, and I am sure there must be some editorial guidance on more acceptable descriptors? (This not me making a point about the analysis of white versus other in results, which is largely driven distribution and low numbers of the other groups, which is acceptable) -Intersectional disadvantageno attempt to measure or estimate? no interaction terms in the model? The analysis tackles standard markers of disadvantage, in a standard way, but doesn't take the opportunity of framing these in a way that might make the way forward clearerwhat are the implications for practice? Can you identify the most likely to be unvaccinated? Then intervention points might be clearer. What about a "non-white" female 70 year old from low SES background relative to a white affluent male 30 year old? I think giving a prediction of the absolute differences might be more impactful.
-Measurement/classification of CKD stage. If I have understood correctly this was inferred from one measurement of creatinine onlycould the authors reflect on the potential for measurement error here? This should at least be discussed as a limitation -Solutions -I feel the results are not surprising, so in discussion, this could be more of a call for reasonable action with some suggestions. There is data overwhelm here, and I feel the authors could do more to draw the reader to the most pertinent findings, and suggest lines of enquiry/solution-testing strategy in the discussion. The abstract conclusion :identifying how is… is a priority" is a little lame in the context of the massive data they have to draw on.
- Tables -T1 needs a  For the background: Just curious how much of the general population does this group of people with kidney disease make up?
Author response: Thank you for the suggestion. We have added the following to the Background section to address this: "Approximately 2.6 million individuals over 16 years of age are estimated to be affected by stage 3-5 CKD in England (prevalence of ~6%) 6 ." Line 183: commas needed around therefore to make sentence not feel awkward (",therefore,") Author response: We have reviewed the sentence in question and are happy with the current phrasing. We are happy to defer to the editor for additional guidance.
Line 264: reference for JCVI recommendations Author response: We now cite relevant references (7 and 8) at the end of this sentence.
Line 277: reference needed Author response: As suggested, we now cite reference 17 (chapter 14 the UK HSA green book), which outlines the JCVI priority groups for primary vaccination in Table 2.
Line 338-9: reference needed Author response: We have amended the phrasing (replacing "The principal aim…" with "A principal aim…") to better reflect that this reflects a subjective statement rather than a citation of an external source. The final sentence now reads as follows: "A principal aim of the COVID-19 vaccine roll-out is to protect those at greatest risk of severe disease." Line 346-7: What does it mean for it to be lower? Why could it be lower? Line 349: What does it mean if it's lower? Author response: Thank you for these comments. In each case, the conclusions are based on the reported hazard ratios (HRs) in Figure 2 and the Supplementary Tables. The reviewer is correct that the term 'lower' is imprecisefailing to capture the fact that these inferred from time-to-event analyses. We have therefore replaced 'higher' with 'faster' and 'lower' with 'slower' in this paragraph, reflecting the more precise language used throughout the Results section. We also highlight that the final coverage was lower in ethnic minority groups and in areas of higher social deprivation as follows: "Minority ethnicity and social deprivation were associated with delayed vaccine administration and lower final coverage…" Line 390: reference needed Author response: We now cite the relevant reference (7) at the end of this sentence. Comments to the Author: This is a descriptive analysis of vaccine uptake in people with kidney disease in England, with a focus on equity of vaccination. Somewhat unsurprisingly those disadvantaged in society were also disadvantaged in covid-19 vaccination. Nevertheless it is important to demonstrate, such that solutions may be designed and tested. Overall, the work is sound in design, statistical analysis, and interpretation, and my comments are therefore minor, and largely relate to presentation and interpretation.
-The description of people of non white race/ethnicity as "non-white" ie in relation to their lack of whiteness. This never sits well with me, and I am sure there must be some editorial guidance on more acceptable descriptors? (This not me making a point about the analysis of white versus other in results, which is largely driven distribution and low numbers of the other groups, which is acceptable) Author response: Thank you for this valuable comment. As indicated by the reviewer, the grouping of non-White individuals in our interpretation was a statistical decision given their low combined prevalence (<10%) in the study population. We have now replaced 'non-White' with 'ethnic minority groups' throughout the manuscript and appendix, as recently recommended in the BMJ (https://doi.org/10.1136/bmj.m4493), but are happy to change the phrasing if an alternative expression is preferred.
We have also added the following sentence to the Methods section describing the new intersectional subgroup analysis (see below) to clarify that our decision to combine ethnic minority groups is driven by statistical considerations: "While recognising their heterogeneity, ethnic minority groups are combined in this subgroup analysis due to their low combined prevalence (<10%) within the study population." -Intersectional disadvantageno attempt to measure or estimate? no interaction terms in the model?
The analysis tackles standard markers of disadvantage, in a standard way, but doesn't take the opportunity of framing these in a way that might make the way forward clearerwhat are the implications for practice? Can you identify the most likely to be unvaccinated? Then intervention points might be clearer. What about a "non-white" female 70 year old from low SES background relative to a white affluent male 30 year old? I think giving a prediction of the absolute differences might be more impactful.
Author response: Thank youthis is very valuable feedback. We agree that a more granular analysis of intersectional disadvantage would be a valuable addition. Our view is that an analysis of cumulative coverage in key subgroups is easier to interpret than the inclusion of interaction terms in the statistical models, and draws attention to key under-vaccinated populations that should be the focus of future interventions.
We have therefore added an analysis of 3-and 4-dose coverage in subgroups stratified by ethnicity, IMD quintile, age, and kidney disease severity. This additional analysis is described as follows: Methods: "Finally, we calculated cumulative 3-and 4-dose coverage in subgroups defined by ethnicity, IMD quintile, age, and kidney disease severity to identify populations at particular risk of under-vaccination at the end of follow-up. While recognising their heterogeneity, ethnic minority groups are combined in this subgroup analysis due to their low combined prevalence (<10%) within the study population." Results (3-dose coverage): "Cumulative coverage estimates stratified by ethnicity and IMD quintile revealed population subgroups with particularly low vaccination rates ( Figure  3 and Supplementary Table 4). Among ethnic minority groups, 3-dose coverage ranged from 67% in the lowest IMD quintile to 88% in the highest, and was <90% across age and kidney disease subgroups (Supplementary Table 4). Among White individuals, 3-dose coverage fell below 90% among individuals <70 years of age and RRT recipients in the most deprived IMD quintiles." Results (4-dose coverage): "Once again, stratified coverage estimates revealed several under-vaccinated subgroups (Figure 3). Among ethnic minority groups, 4-dose coverage ranged from 35% in the most deprived IMD quintile to 71% in the least deprived, with particularly low coverage (<50%) among individuals aged 16-64 years in all IMD quintiles (Supplementary Table 8). In White individuals, 4-dose coverage was below 75% among individuals <70 years of age in all IMD quintiles, and among RRT recipients in the most deprived IMD quintiles." Discussion: "Notably, under-vaccinated populations in the present study included: ethnic minority groups of any age or disease severity in areas of both low and high social deprivation; and White individuals <75 years of age in areas of both medium and high social deprivation. Furter research such as qualitative surveys within these under-vaccinated populations could offer a valuable opportunity to identify key barriers to vaccine uptake." The raw data underlying Figure 3 are presented in Supplementary Table 4 (3-dose estimates) and 8 (4-dose estimates). We thank the reviewer for raising this issue, and feel that the changes above have substantially strengthened the final manuscript.
-Measurement/classification of CKD stage. If I have understood correctly this was inferred from one measurement of creatinine onlycould the authors reflect on the potential for measurement error here? This should at least be discussed as a limitation Author response: We have added the following limitation to address this issue: "Third, individuals with CKD were identified based on a single serum creatinine measurement. This approach is highly sensitive and minimises delay in identification of affected individuals, but lacks confirmation of chronicity 24 ." -Solutions -I feel the results are not surprising, so in discussion, this could be more of a call for reasonable action with some suggestions. There is data overwhelm here, and I feel the authors could do more to draw the reader to the most pertinent findings, and suggest lines of enquiry/solutiontesting strategy in the discussion. The abstract conclusion :identifying how is… is a priority" is a little lame in the context of the massive data they have to draw on.
Author response: As detailed in the responses above, we have added additional subgroup analyses and a new Figure 3 to highlight key under-vaccinated subgroups within our study population. These are perhaps the most pertinent findings from a policy perspective, and we have therefore augmented the Discussion to draw more direct attention to these groups (and the need for specific solution-testing) as follows: "Notably, under-vaccinated populations in the present study included: ethnic minority groups of any age or disease severity in areas of both low and high social deprivation; and White individuals <75 years of age in areas of both medium and high social deprivation. Further research such as qualitative surveys within these under-vaccinated populations could offer a valuable opportunity to identify key barriers to vaccine uptake." We have amended the final sentence of the Abstract to reflect a similar sentiment, as follows: "Targeted interventions are needed to identify barriers to vaccine uptake among under-vaccinated subgroups identified in the present study." This comment on data overwhelm is well taken. To address this, we have moved Figures 3  and 4 to the supplementary materials given that they consolidate trends reported elsewhere.
We have instead added a new Figure 3 that presents cumulative coverage across key subgroups, as described above.
- Tables -T1 needs a  In Supplementary Tables 2 and 5, we agree that the cross-referencing symbols are unclear. We have therefore removed these and added the following text to the figure legends to interpret: "Minimally adjusted models included age, care home residence, and health and social care worker status given their use in vaccine prioritisation criteria. Partially adjusted models additionally included housebound status, receipt of end-of-life care, setting, sex, ethnicity, IMD quintile, prior SARS-CoV-2 infection, immunosuppression, and haematologic cancer. Fully adjusted models included all covariates" -We have removed the 'N (n events)' columns of Supplementary Table 3 (given their overlap with data presented in Table 1) and increased the font size of the remaining columns to improve readability. We have done the same for Supplementary Table 7 Table 6 (formerly Supplementary Table 5). We agree that the visual format is more informative. The supplementary table is intended as a supportive adjunct that also provides event counts alongside outputs from minimally and fully adjusted models.
-Figures -F1can you use monochromatic scheme to make the plot lines easier to distinguish? (dots and dashes?)? Also add some x-ticks for months. F3 is tiny and impossible to read Author response: We have amended Figure 1 as suggested. Figure 3 has been moved to the supplement (as Supplementary Figure 2), and its width expanded to improve readability. Note that underlying data is also presented in Supplementary Table 3. Editor(s)' Comments to Author: Please delete the sections about what is known and what the study adds as they are not part of the BMJ Open format.