Correspondence

Contents lists available at ScienceDirect

American Journal of Emergency Medicine

journal homepage:

Serum lactate not associated with Pediatric DKA outcomes

13 November 2019

https://doi.org/10.1016/j.ajem.2019.158407

Dear Editor,

I read with great interest on the study by Cully and colleagues [1], which showed that serum lactate was not associated with clinical out- come in pediatric Diabetic ketoacidosis . While the study is of clin- ical importance, there are several issues that we want to point out for the interpretation of the results. The major problem is related to the design of the study. The authors excluded patients when missing values on serum lactate. While this is reasonable because they want to explore the associ- ation of lactate with clinical outcome, the exclusion can cause bias, mak- ing the study population not representative of all pediatric DKA. There is evidence showing that the healthcare process of laboratory test is related to clinical outcome (e.g. the frequency and timing of lab test is indepen- dently related to the clinical outcome in hospitalized patients, irrespective of the value of the test) [2]. Thus, we deduce that if lactate is not measured for a DKA patient in the emergency department, the lactate can be normal as deemed by the physician and it is not necessary to order the test. In this situation, the patient can be less critically ill than those with lactate mea- sured. Alternatively, the patient can be quickly transferred to the intensive care unit (ICU) and there is not enough stay time in the emergency de- partment (ED) for such test. There is also evidence that length of stay in ED is associated with mortality outcome [3]. No matter which reason is responsible for the missing value, such population can be systematically different from those with lactate measured in the ED. Several approaches can help to address this issue. First, the authors should report how many patients were excluded due to missingness and compare them to those being included. Second, some imputation methods can be employed to handle the missingness [4,5]. Missing lactate can be imputed by building a linear regression model regressing lactate on other variables. Especially, when lactate on ICU admission is available, this can be used as a surrogate for current analysis. Third, sensitivity analysis can be performed to see whether the results are robust to the above-mentioned analyses.

The second issue is related to the statistical approach. While the au-

thors stated that lactate was not related to the clinical outcome, it is ac- tually the problem of low statistical power of the analysis. In the analysis, each clinical outcome was analyzed separately, and the abso- lute number of events was very low in both groups. I suggest the au- thors to combine all these events together to calculate the Composite outcome, which will significantly increase the statistical power and the conclusion can be altered.

There is no conflict of interest.

Tian Ye Department of Emergency Medicine, Affiliated Dongyang Hospital of Wenzhou Medical University, Dongyang, Zhejiang 322100, China

E-mail address: [email protected].

References

1. Cully M, Thompson AD, DePiero AD. Is lactic acidosis predictive of outcomes in pedi- atric diabetic ketoacidosis? Am J Emerg Med 2019. https://doi.org/10.1016/j.ajem. 2019.158449.
2. Agniel D, Kohane IS, Weber GM. Biases in electronic health record data due to pro- cesses within the healthcare system: retrospective observational study. Bmj 2018; 361:k1479. https://doi.org/10.1136/bmj.k1479.
3. Zhang Z, Bokhari F, Guo Y, Goyal H. Prolonged length of stay in the emergency depart- ment and increased risk of hospital mortality in patients with sepsis requiring ICU ad- mission. Emerg Med J 2019;36:82-7. https://doi.org/10.1136/emermed-2018- 208032.
4. Zhang Z. Missing data imputation: focusing on single imputation. Ann Transl Med 2016;4:9. https://doi.org/10.3978/j.issn.2305-5839.2015.12.38.
5. Zhang Z. Multiple imputation with multivariate imputation by chained equation (MICE) package. Ann Transl Med 2016;4:30. https://doi.org/10.3978/j.issn.2305- 5839.2015.12.63.

   Potential selection bias when subjects were excluded because of missing values

   Dear Dr. J. Douglas White,

   Thank you for the opportunity to respond to the letter about our manuscript “Is lactic acidosis predictive of outcomes in pediatric dia- betic ketoacidosis” recently published in “The American Journal of Emergency Medicine”. Our response is listed below.

   We are grateful to the authors for their interest and constructive comments regarding our manuscript “Is lactic acidosis predictive of out- comes in pediatric diabetic ketoacidosis”. This was a retrospective study and as such we acknowledge that one of its limitations is patient en- counters with missing values. At our institution, all patients with known or new onset diabetes mellitus with moderate urine ketones have a critical care profile, basic metabolic panel, and phosphorous col- lected. As we were most interested in whether lactic acidosis was pre- dictive of morbidity and mortality in diabetic ketoacidosis (DKA), we limited our analysis to emergency department patients with severe DKA requiring intensive care admission.

   During the study period of December 1, 2015 to December 1, 2018 we identified 113 patient encounters that met the criteria for severe DKA and who had lactate levels. During this period, there were an addi- tional 9 patient encounters with severe DKA who did not have available lactate levels and were thus not included in our study population. When

   https://doi.org/10.1016/j.ajem.2019.158407

   0735-6757/(C) 2019

   compared to the larger group of severe DKA patients, this group of 9 pa- tient encounters was no different in terms of demographics, vital signs, laboratory findings or length of stay. In addition, these 9 patient encoun- ters also had no documented serious complications or death during their hospitalizations.

   We agree with the authors that a second limitation to our study is the low incidence of morbidity and mortality that limited the power to determine a significant association with the presence of lactic acido- sis. While we analyzed each clinical outcome separately, we also com- bined the events to calculate the total morbidity. We believe these clinical outcomes are accurately reflected in Table 1.

   Sincerely,

   Matthew Cully, DO Amy D. Thompson, MD Andrew D. DePiero, MD

   Matthew Cully, DO^? Amy D. Thompson, MD Andrew D. Depiero, MD

   Nemours Alfred I. duPont Hospital for Children, Attn: Emergency Medicine,

   1600 Rockland Rd., Wilmington, DE 19803, USA.

   ^?Corresponding author. Tel. + 1 302 651 4296; fax + 1 302 651 4227.

   E-mail address: [email protected].

   30 November 2019

   https://doi.org/10.1016/j.ajem.2019.12.001

   Molecular diagnostics, a strategy which does not address the diagnostic deficit in syphilitic meningitis

   The authors of the study where fewer episodes of meningitis were identified by blood culture than by cerebrospinal fluid (CSF) culture did include the caveat that there were some patients who were diag- nosed with meningitis despite negative CSF results “and all these pa- tients were deemed to have viral meningitis” [1]. A comparable scenario was exemplified by a Danish study which documented positive blood cultures and diagnosis of Bacterial meningitis in cases with nega- tive cultures of cerebrospinal fluid [2]. The latter was a retrospective study which covered an 8.5 year period in which 294 patients were identified who had culture negative CSF specimens and positive blood cultures. After a chart review 8 patients were excluded:4 patients had another CSF specimen which produced a positive culture, 2 had a posi- tive CSF Gram stain result, and 2 had Epidural abscesses [2].

   The strategy of Molecular diagnosis, using the polymerase chain re- action(PCR), was proposed by the authors of the recent study to remedy the deficit between CSF and peripheral blood cultures [1]. When applied to the CSF, this strategy, has proven utility in identifying bacterial, viral, and fungal culprit pathogens responsible for single-microbial as well as polymicrobial meningitis [3]. This strategy, however, does not address the problem of suboptimal sensitivity of cerebrospinal fluid PCR for identification of neurosyphilis [4]. In the latter context, a recent meta- analysis showed that the sensitivity of PCR for definite neurosyphilis varied between 40% and 70%, with specificities in the range 60% to 100% [4]. The Venereal Diseases Research Laboratory(VDRL) test on the CSF(with confirmatory test on a blood sample) is an alternative strategy to validate neurosyphilis [5]. In the latter study the sensitivity of cerebrospinal fluid VDRL amounted to only 70.8%, with specificity of 99.3% [5].

   The above observations are concerning for the following reasons:

   There might be a temptation to ascribe the etiology of culture- negative meningitis to an unspecified virus [1], instead of evalu- ating the CSF for a potential treponemal culprit pathogen.

6. The advent of the human immune deficiency virus era has been accompanied by a resurgence of acute syphilitic meningitis [6,7], an entity which has to be included in the differential diag- nosis of single microbial, as well as polymicrobial meningitis, es- pecially in HIV-positive subjects [8-10].
7. Some patients with syphilis can have CSF evidence of neurosyphilis without having neurological symptoms [11]. In the latter study of 466 subjects with serologically validated syph- ilis 88 subjects(68 HIV-positive and 20 HIV-negative) also had neurosyphilis, the latter diagnosis validated by VDRL on the CSF. Neurological symptoms included headache, stiff neck, pho- tophobia, Vision loss ocular inflammation, hearing loss, sensory loss, and gait incoordination. However, among HIV-positive sub- jects, each of those symptoms had a sensitivity for neurosyphilis which varied from 1.5% to 65.1%, the latter statistic being one as- sociated with vision loss. Only hearing loss, sensory loss, and gait incoordination, respectively, were symptoms which were signif- icantly more prevalent in HIV-coinfected than in HIV-negative subjects [11].
8. During the course of its natural history, if left untreated, neurosyphilis is a potential cause of dementia [12,13]. Neverthe- less, in the occasional patient with dementia and neurosyphilis (the latter with no other neurological neurological signs), a course of high-dose intravenous penicillin can result in full re- covery, and an increase in Mini mental state Examination score from a pretreatment level of 22 to a post treatment level of 27, following a course of high dose intravenous penicillin [14]. Equally striking was the case of a 48 year old man with neurosyphilis who presented with a 2 weeks history of head- ache, disorientation, and self-aggressiveness. His Glasgow Coma Scale amounted to 10, and his CSF was characterised by a white cell count of 181 cells/ul (with a predominance of lympho- cytes), and a total protein count of 91.1 mg/dl. After a 14-day course of high-dose intravenous penicillin he experienced marked clinical improvement, with concurrent increase in GCS to a score of 15 [15]. Accordingly, recognition and treatment of neurosyphilis, in all its manifestations, is worthwhile, and clini- cians should have a heightened index of suspicion for this diag- nosis when dealing with HIV-subjects who present either with meningeal signs and symptoms, and/or behavioural disturbance. Furthermore, neurosyphilis should be included in the screen for single microbial as well as for polymocrobial meningitis, even when a culprit pathogen other than Treponema pallidum has al- ready been identified by evaluation of the CSF by PCR.

   Declaration of competing interest

   I have no funding and no conflict of interest.

   Oscar M.P. Jolobe, MRCP

   Manchester Medical Society, Simon Building, Brunswick Street,

   Manchester, M13 9PL, United Kingdom E-mail address: [email protected]

   Received 12 July 2019

   https://doi.org/10.1016/j.ajem.2019.158407