On 2025-11-30 16:49:17, user Cyril Burke wrote:
[Note: This is the first of several reviews of an earlier version of our combined manuscript that aims to reduce ‘racial’ disparity in kidney disease. The comments were kindly offered by nephrologists, through a medical journal, and we remain grateful to them for the time and care they gave to improve our manuscript.
We removed identifying features and will include our responses in a subsequent comment. The changing title and line numbers refer to versions prior to our medRxiv preprints.]
April 1, 2022<br />
Screening for early kidney disease and population health using longitudinal serum creatinine
Dear Dr. Burke III,
REDACTED.
Reviewer #1: Burke et al submit a somewhat unusual paper, devoted to a topic of potential major clinical relevance, and as yet understudied.
General comments
-
The thesis of the authors, that using the baseline serum creatinine of a given patient would potentially improve the earlier diagnosis of kidney disease, even in the normal range, is in line with the experience of this reviewer, who always retrieves , whatever the difficulty of reaching that goal, past results of blood tests, and uses them as a way to date the onset of kidney disease, sometimes with important prognostic implications.
-
Yet, the authors do not provide data strongly supporting their thesis. For instance, when looking at case 2, should the last point (the most recent one) be omitted, there would be very little evidence supporting progressive early kidney disease.
-
The claim that the statistics fit the data better when all points are used (page 9,11) should not come as a surprise. Using thresholds instead of the full range of values has long been known to be more powerful for statistical analysis. But fitting the data does not equal to a high positive predictive value!
-
A key question is whether in a real world context, the earlier diagnosis of kidney disease would be possible, without too much background noise from intercurrent illness (functional), drugs (NSAIDS, etc..). In other words, would the specificity (or PPV) of the suspicion of early kidney disease be reasonable enough to catch the attention of clinicians
-
Even though there has been improvement in the standardization of measurement of serum creatinine (IDMS), the comparability of results measured by different labs remains suboptimal, at least in the experience of this reviewer, and medical shopping is not uncommon, making the availability of all previous results in the same graph a logistical challenge.
Specific comments
-
The authors should mention that the USPTFS decided a month ago to revisit the question of screening for kidney disease in high risk groups (page …)
-
Even though ESRD has a legal meaning in the USA, not very relevant to the topic of this paper about early kidney disease, the authors should stick to the nomenclature proposed by a recent KDIGO consensus conference (see Levey et al. Nature Reviews in Nephrology ). In particular, use kidney failure instead of ESRD/ESKD. When the topic is glomerular filtration, use that wording instead of kidney function (page…)
-
The authors allude to the concepts of prediabetes and prehypertension. But this reviewer points to the fact that the levels used to define those entities are currently “generic” , rather than based on previous values in an individual subject. Please discuss.
-
The authors repeatedly mention in the discussion section evidence that even small increases in serum creatinine have prognostic significance. This has indeed been known for decades but is a different topic: AKI . Admittedly, there is growing evidence that AKI and CKD are linked. But that the stability of a biological parameter is prognostically best is all except surprising: the same is true for body weight, mood, blood pressure etc…
Reviewer #2: Thank-you for the opportunity to review this work which highlights the importance of monitoring serum creatinine over time and how this can be a useful tool in detecting possible CKD. This is an important topic as the use of sCr on its own is certainly under-utilised and changes are often missed because they don’t fall into a predefined category.
MAJOR CONCERNS
“Choi- rates of ESRD in Black and White Veterans” doesn’t fit with the rest of the paper including the title; the introduction and conclusion also don’t adequately address this portion of the paper. It feels disjointed from the main point of discussion which is the use of sCr in screening “pre-CKD”. This section and discussion should be removed and possibly considered for another type of publication.
Cases 1 - 3, (lines 93 – 122): where are these cases from? There is no mention of ethics to publish these patient results, which appears to be a clear ethics violation. If so, these cases should be removed and patient consent and ethical approval obtained to publish them.<br />
The authors describe the reasons for not obtaining an ethics waiver for this secondary data analysis. Despite this, the relative ease of obtaining an ethics waiver for secondary data analysis usually means that this is done regardless.
The message of the article and data representation is unclear: do the authors wish to show that sCr is superior to eGFR in this “pre-CKD” stage, should both be used together? Do the authors wish to convey that a “creatinine blind range” does not exist? Or is the aim to demonstrate that continuous variables should not be interpreted in a categorical manner?
MINOR CONCERNS
ABSTRACT<br />
Vague<br />
Doesn’t give a clear picture of the study
INTRODUCTION<br />
51 – 57: needs to state that these stats are from e.g. the US. The authors should consider adding international statistics to complement those from the US.
68: reference KDIGO guidelines, state year
75 – 77: is this reference of the New York Times the most appropriate?
82: within-individual variation not changes (this is repetition of the point made in lines 425 – 427, but should match the language)
82 – 84: reference? If this is a question it should be presented as such
84: “normal GFR above 60” = guidelines (including KDIGO) do not refer to 60 as normal GFR, 60 – 89 is mildly decreased. (see line 126)
93: avoid the use of emotive words such as apparently (also in line 428)
94: “Not meeting KDIGO guidelines”: KDIGO 2.1.3 includes a drop in category (including those with GFR >90). This would appear to include some of the cases listed. Additionally, albuminuria should have been measured for case 2 and 3.
97: “progressive loss of nephrons equivalent to one kidney”: this is based on a single creatinine measurement.
93 – 122: Could any of these shifts be explained by changes in creatinine methodology or standardisation of assays, especially over 15 – 20 years (major differences between assays existed before standardisation and arguably still exist with certain methods).<br />
It would be useful to see a comparison between serial sCr and eGFR measurements on the same figure. There appears to be significant (possibly more pronounced) changes when eGFR is used. As line 87 mentions changes in eGFR may be as useful (and in some situations more useful) than changes in sCr alone.
127 – 142: should there be separate charts for males and females, the differences in creatinine between males and females needs to be discussed somewhere in the paper. Similarly, is this suitable for all ages?
162 – 163: rephrase
METHODS<br />
185 – 193: aim belongs in the introduction, can be adjusted to complement paragraph 178 – 182.
196 – 205: reference sources
224 – 247: not in keeping with the rest of the article or title and conclusion
RESULTS<br />
If eGFR is treated as a continuous variable does inverted sCr still have higher accuracy?
As mentioned, the section on ESRD in black and white veterans doesn’t fit in with the rest of the article.
DISCUSSION<br />
As mentioned, section 4.1 doesn’t fit in with the rest of the article. As the authors note the correlation between illiteracy and CKD is likely not causal.
387: erroneous creatinine blind range. The data presented does not show this is erroneous there is still a relative blind range. A distinction must be made between a population level “blind range” and an individual patient’s serial results. The data and figure 4 in particular demonstrate the lack of predictive ability of sCr above 40ml/min compared to below 40ml/min at a population level. For an individual patient this “blind range” is more relative, and a change in sCr even within the normal range may be predictive. (Note: the terminology “blind range” is problematic).
399 – 400: “rose slowly at first and then more rapidly as mGFR decreased below 60” this refers to a relative blind range. Whether these slow initial changes can be distinguished from analytical and intra-individual variation is the question that needs to be answered before we can say a “blind-range” doesn’t exist for an individual patient.
425 - 432: sCr is indeed very useful when baseline measurements are available. eGFR remains useful when baseline sCr is not available or when large intervals between measurements are found.
425: low analytical variation- if enzymatic methods are used
428: avoid the use of “apparently”
430: reference 56 compares sCr and sCysC with creatinine clearance NOT with mGFR, this does not prove that mGFR has greater physiologic variability. Creatinine clearance is known to be highly variable (partially due to two sources of variability in the measurements of creatinine: serum and urine).
The limitations of sCr for screening should also be discussed: differences in performance and acceptability between enzymatic and Jaffe methods (still widely used in certain parts of the world), the effect of standardizing creatinine assays (an important initiative but one that could also produce shifts in results around the time of standardization- see cases), low InIx means that once-off values are exceedingly difficult to interpret, is a single raised creatinine value predictive (or should there be evidence of chronicity): similarly are there effects from protein rich meals, etc (The influence of a cooked-meat meal on estimated glomerular filtration rate. Annals of Clinical Biochemistry. 2007;44(1):35-42. doi:10.1258/000456307779595995)
CONCLUSION<br />
The discussion recommends using SCr above eGFR while the conclusion recommends the NKF-ASN eGFR for use in pre-CKD and ASC charts. While the use of both together in a complementary fashion is understandable- this needs to be congruent with the discussion, aims and results.