Value of blood pressure for detecting serious illness in children attending acute care settings in low- and middle-income countries: a multicentre observational study Authors: Naomi Kemps, Nienke N Hagedoorn, Navin P Boeddha, Abdoulie Faal, Natanael Holband, Amadu E Juliana, Kristina Keitel, Kevin H Van’t Kruys, Elizabeth V Ledger, Chacha D Mangu, Henriette A Moll, Andrew M Prentice, Rainer Tan, Effua Usuf, Stefan A Unger, Clementien L Vermont, Joany M Zachariasse, on behalf of VITaLs Study Group Published: 2026 ABSTRACT AbstractBackgroundIn children, the value of routine blood pressure measurement in first assessment at the emergency department has been questioned. This has, however, not yet been evaluated in low- and middle-income countries (LMICs) where the incidence of serious illness is high. This study assessed the predictive value of low systolic blood pressure to detect serious illness in children attending acute care settings in LMICs.MethodsThis observational study is based on data from children attending three acute care settings in The Gambia and Suriname. Hypotension was defined by age-adjusted reference values according to Advanced Paediatric Life Support (APLS) and Paediatric Advanced Life Support (PALS) cut-offs. The outcome measure was a composite marker of serious illness consisting of admission or referral to a higher level of care. A logistic regression analysis was used to evaluate the association between hypotension and serious illness for each setting separately.Results20 185 visits from Gambia rural, 398 visits from Gambia urban and 1817 visits from Suriname were included. Hypotension was present in 8.3–17.3% of visits and serious illness in 3.9–15.7%. Hypotension was not significantly associated with serious illness in Gambia urban (ORAPLS 3.2 (0.7–10.9) and ORPALS 1.4 (0.1–7.6)) and Suriname (ORAPLS 1.3 (0.9–1.9) and ORPALS 1.7 (0.9–3.2)), and with an inverse association in Gambia rural (ORAPLS 0.6 (0.5–0.7) and ORPALS 0.9 (0.5–1.4)).ConclusionsOur results do not show a significant association between hypotension and serious illness in LMICs. Our results suggest that routine blood pressure measurement does not aid in the evaluation of well-appearing children attending acute care settings in LMICs. WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS In three diverse acute care settings in LMICs, ranging from rural clinic to urban settings, low blood pressure is not significantly associated with serious illness. In three diverse acute care settings in LMICs, ranging from rural clinic to urban settings, low blood pressure is not significantly associated with serious illness. Low blood pressure was relatively common, including in children who were not severely ill. Low blood pressure was relatively common, including in children who were not severely ill. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Blood pressure should not be used as a sole routine screening tool for illness severity in children attending acute care settings in LMICs. Blood pressure should not be used as a sole routine screening tool for illness severity in children attending acute care settings in LMICs. Introduction In acute care settings, early recognition of serious illness is essential to initiate timely and accurate treatment. This is particularly true in low- and middle-income countries (LMICs) where the under-five mortality is substantial and over 30% of in-hospital mortality occurs within 24 hours of admission.14 In these settings, early recognition of children with serious illness poses extra challenges due to lack of necessary resources, late presentation and high prevalence of risk factors for a serious disease course including malnutrition.57 Blood pressure plays an important role in the initial assessment of adults attending acute care, but in children the value of routine blood pressure measurement has been questioned.810 Foremost, research has shown that the sensitivity of hypotension for detecting serious illness in children is low, most likely because hypotension is a late sign of serious illness.8 11 12 This has been shown in several European studies that looked at the value of hypotension by itself or as the Shock Index where blood pressure is adjusted for heart rate.12 13 Second, blood pressure measurement can be stressful for children and time consuming for nurses, and the reliability in young non-cooperative children is often uncertain. These barriers contribute to low measurement rates of blood pressure in emergency departments (EDs) in high-income countries.12 14 15 However, there is little to no research on this topic in LMICs. Due to differences in the numbers of patients and the proportion of seriously ill children, the predictive value of blood pressure in these settings might be affected. In addition, differences in body composition or nutritional status and their potential effect on physiological parameters may suggest that the results from studies in high-income settings cannot be generalised directly to acute care settings in LMICs.7 16 In this study, we aim to investigate the value of hypotension, by itself or in addition to other vital signs, to detect serious illness in children attending acute care settings in LMICs. We hypothesise that due to the larger volumes of patients and the higher numbers with severe illness and differences in body composition or nutritional status in LMICs in comparison to high-income settings, blood pressure could have value as a routine screening tool in paediatric EDs in LMICs. Insight into the diagnostic accuracy of predictors for serious illness in LMICs may support health care workers in these settings and inform variable selection for the development of prediction rules, early warning scores and guidelines to improve the recognition of serious illness in children. Methods Study design This is a multicentre observational study using routinely collected data of children attending the acute care service or ED of three different settings in The Gambia and Suriname. All data were previously collected for research purposes.1720 Study sites and population The study sites have been described in detail previously.21 In short, data were collected from (1) the MRC Keneba Field Station, a primary healthcare clinic in a rural region Kiang West in The Gambia (hereafter Gambia rural); (2) MRC Fajara and Kanifing General Hospital, two urban tertiary care clinics in The Gambia (Gambia urban); and (3) Academic Hospital Paramaribo, a tertiary care hospital in an urban area in Suriname (Suriname). The inclusion criteria of the studies were defined by the original study for which the data were collected (online supplemental table S1). Most notably, Gambia urban only collected data on children with fever and an indication for a venipuncture, while the other settings included all presenting children. For this study, we excluded children who died at presentation and children with missing data on disposition. We expect the added value of blood pressure to be limited in this group due to these children already presenting as life-threatening emergencies. Data collection Collected data were part of routine emergency care and included general patient characteristics (age, sex, ethnicity, triage urgency, presence of comorbidity) and vital signs measured at triage (heart rate, respiratory rate, temperature, oxygen saturation, non-invasive systolic blood pressure, capillary refill time). Blood pressure measurement was performed at the discretion of the nurse or physician with an oscillometric device, and age-adjusted cuff sizes were mostly available (online supplemental file 1). Clinical indication for measurement differed per setting but included routine triage and on indication in ill-appearing children (online supplemental file 1). In addition, data were obtained on the main reason for acute care visit, clinical symptoms (ill appearance, work of breathing, level of consciousness), treatment at the acute care centre (intravenous medication or fluid bolus, immediate life-saving interventions) and disposition (home, ward admission, intensive care unit (ICU) admission, death at the acute care centre, leave against advice, referral to a higher level of care).22 Data were checked for outliers and completeness. Blood pressure values were excluded if they were deemed to be implausible. Clinical definitions Our main outcome, representing serious illness, was defined as: hospital or ICU admission immediately following the acute care visit, or referral to a higher level of care in settings without an ICU. Hypotension and tachycardia were classified by age-adjusted reference values defined by Advanced Paediatric Life Support (APLS) and Paediatric Advanced Life Support (PALS) (online supplemental file 2).2325 These cut-offs were chosen to represent the range in clinical practice, as they are the two most frequently used clinical reference values. We included systolic blood pressure in two ways: (1) blood pressure as a continuous variable adjusted for age as a z-score, and (2) blood pressure as a dichotomous variable using the APLS and PALS cut-offs to define hypotension versus normal blood pressure. Hypertension was defined according to the American Academy of Pediatrics (AAP) cut-offs as a systolic blood pressure above the 95th percentile according to sex and age or above 130 mm Hg (stage 1), and a systolic blood pressure above the 95th percentile+12 mm Hg or above 140 mm Hg (stage 2).26 Z-scores were calculated separately for each setting based on its respective blood pressure values using the following cut-offs for age groups: 0, 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 14, and 18, and normality was assessed prior to their calculation. Weight was included as weight-for-age z-scores and height as height-for-age z-scores as defined by the WHO.27 Season was categorised as dry versus wet. In The Gambia, dry season was defined as between the months of November to May and wet season between June and October. In Suriname, dry season was defined as between February and April and August and October and wet season from May to July and November to January. In all settings, patients were routinely triaged using conventional triage systems. In Gambia rural, triage was performed using the three-level Emergency Triage Assessment and Treatment.28 In Suriname, the five-level Emergency Severity Index system was used, while in Gambia urban, the five-level Integrated Management of Childhood Illness classifications were used.28 29 Data analysis Due to differences in characteristics between the settings, we performed all analyses separately for each site. First, we described the baseline characteristics of visits, comparing visits with and without blood pressure measurement and children with and without hypotension. We explored the distributions of systolic blood pressure visually. Our main analysis consisted of a complete case analysis, where we included only visits with available blood pressure measurements. We analysed the association between serious illness and blood pressure with a univariable logistic regression analysis. We included blood pressure in two ways as described earlier. We conducted several post hoc analyses with the goal of better understanding our results. First, we used multivariable logistic regression to adjust for potential confounders: heart rate/tachycardia, weight, height, ethnicity and dry versus wet season. Second, we evaluated the association between hypertension as defined earlier and the outcome measure with a logistic regression analysis. Finally, we stratified the initial analyses for children under and above 12 years of age. Furthermore, to assess the diagnostic accuracy of hypotension for serious illness, we calculated sensitivity, specificity, positive likelihood ratio and negative likelihood ratio. In a sensitivity analysis, we imputed missing blood pressure measurements using the MICE package in R, which resulted in 20 imputation sets.30 The imputation model included the following variables: patient characteristics (age, sex, ethnicity), all vital signs (heart rate, respiratory rate, oxygen saturation, capillary refill time, temperature, systolic blood pressure, mental status and increased work of breathing), triage urgency and disposition. We repeated the univariable logistic regression analyses using the imputed data. Furthermore, to address selection bias, inverse probability weighting was performed as another sensitivity analysis. Variables included as predictors for inverse probability weighting were child characteristics (age, sex, ethnicity), vital signs (heart rate, respiratory rate, oxygen saturation and temperature) and triage urgency. Lastly, a sensitivity analysis was performed where children who died on the day of presentation were included. All data analyses and visualisations were performed in R V.4.3.2.31 Results Baseline characteristics 40 443 visits from Gambia rural, 501 visits from Gambia urban and 2630 visits from Suriname were included in this study. Of these visits, blood pressure was measured in 20 185 (49.9%), 398 (79.4%) and 1817 (69.1%) children, respectively. Blood pressure was measured more often in older children, but measurement rates were not consistently or markedly higher in the children triaged as (high) urgent (online supplemental file 3). In the visits with blood pressure measured, median age ranged from 3.6 years (IQR 1.8–5.5) to 8.9 years (IQR 3.7–13.0), and the majority in all settings was male (51.1–54.9%). Median weight-for-age z-score ranged from −1.3 to −0.2. Admission to the ward or ICU or referral to a higher level of care ranged from 3.9% to 15.7% across the settings. Rates of hypotension varied between sites and depended on the cut-off used. Based on the APLS cut-offs, rates of hypotension ranged from 8.3% to 17.3%, while using the PALS cut-offs, hypotension ranged from 2.5% to 5.3% of visits (table 1). Hypertension (AAP cut-off at 95th centile) ranged from 15.8% to 34.8%. Characteristics of visits with and without hypotension are shown in online supplemental file 4. We explored the distribution of systolic blood pressure visually (online supplemental file 5). These graphs show that in the group of children who are discharged, there is still a substantial portion of children with hypotension in all three settings, especially in older children. Table 1Baseline characteristics of the children with measured blood pressureGambia ruraln=20 185Gambia urbann=398Surinamen=1817n (%)n (%)n (%)General characteristics Male10 315 (51.1)217 (54.5)997 (54.9) Age in years, median (IQR)8.9 (3.7–13.0)3.6 (1.8–5.5)5.1 (2.1–10.5)Triage label, according to existing triage system* Immediate51 (0.3)0 (0.0)64 (3.5) Very urgent–1 (0.3)771 (42.4) Urgent1381 (6.8)17 (4.3)454 (25) Standard17 801 (88.2)379 (95.2)448 (24.7) Non-urgent–1 (0.3)15 (0.8)Any comorbidityn/a4 (1.0)195 (10.7)Presenting symptoms Ill appearance939 (4.7)73 (18.3)465 (25.6) Increased work of breathing191 (0.9)29 (7.3)188 (10.3) Decreased consciousness231 (1.1)1 (0.2)71 (3.9)Vital signs Temperature ≥38°C1660 (8.2)394 (99.0)117 (6.4) Prolonged capillary refill ≥3 s56 (0.3)0 (0.0)39 (2.2) Hypotension APLS3496 (17.3)33 (8.3)161 (8.9) Hypotension PALS506 (2.5)21 (5.3)54 (2.9) Tachycardia APLS1301 (6.5)211 (53.0)351 (19.3) Hypertension AAP 95th3182 (15.8)109 (27.4)632 (34.8) Hypertension AAP 95th+12413 (2.0)29 (7.3)161 (8.9) Weight-for-age z-score, median (IQR)−1.3 (−1.9 to −0.7)−1.2 (−1.9 to −0.5)−0.2 (−1.1 to 0.8)Treatment Intravenous medication or fluid278 (1.4)n/a111 (6.1) Immediate life-saving intervention22n/an/a35 (1.9)Disposition Admission623 (3.1)14 (3.5)258 (14.2) NICU/PICU admissionn/a0 (0.0)28 (1.5) Referral to a higher level of care159 (0.8)n/an/a Home19 373 (96.0)384 (96.5)1531 (84.3) Death on day of presentation1 (0.0)0 (0.0)2 (0.01)*Suriname: ESI; Gambia urban: IMCI classifications; Gambia rural: ETATAAP, American Academy of Pediatrics; APLS, Advanced Paediatric Life Support; ESI, Emergency Severity Index; ETAT, Emergency Triage Assessment and Treatment; IMCI, Integrated Management of Childhood Illness; NICU, neonatal intensive care unit; PALS, Paediatric Advanced Life Support; PICU, paediatric intensive care unit. Baseline characteristics of the children with measured blood pressure Suriname: ESI; Gambia urban: IMCI classifications; Gambia rural: ETAT AAP, American Academy of Pediatrics; APLS, Advanced Paediatric Life Support; ESI, Emergency Severity Index; ETAT, Emergency Triage Assessment and Treatment; IMCI, Integrated Management of Childhood Illness; NICU, neonatal intensive care unit; PALS, Paediatric Advanced Life Support; PICU, paediatric intensive care unit. Association between blood pressure and serious illness Our results do not show a significant positive association between hypotension and serious illness in children (table 2). In Suriname and Gambia urban, blood pressure was not associated with the outcome measure, either as z-score or as hypotension defined by APLS and PALS. In Gambia rural, blood pressure (z-score) and hypotension (APLS) were associated with the outcome measure, with ORs of 1.11 (95% CI 1.04 to 1.19) and 0.63 (95% CI 0.50 to 0.78), respectively, indicating that a low blood pressure was not associated with serious illness. Given these unexpected results, we conducted several post hoc analyses to exclude potential confounding due to important variables we previously did not include. Based on discussions with the study group, we assessed variables such as heart rate/tachycardia, weight, height, season and ethnicity. However, the inclusion of these variables in the multivariable logistic regression analyses did not change the association between blood pressure or hypotension and the outcome measure for all three settings (online supplemental file 7). Post hoc analyses using the AAP cut-offs for hypertension showed a significant association between the 95th percentile and the outcome measure in Gambia rural (OR 1.31, 95% CI 1.05 to 1.62). No significant associations were found in Gambia urban, in Suriname or for the 95th percentile+12 mm Hg in Gambia rural (online supplemental file 6). Furthermore, stratifying the logistic regression for children under 12 years of age in comparison to children above 12 years of age showed a significant association in Gambia rural for the continuous z-score and hypertension in children under 12 years of age (OR 1.15, 95% CI 1.06 to 1.24 and OR 1.49, 95% CI 1.16 to 1.89, respectively) (online supplemental file 6). These results did not explain the inverse association found in the initial analysis in Gambia rural and did not change the overall conclusion. In two sensitivity analyses where we repeated the logistic regression on the data where we imputed missing blood pressure and performed inverse probability weighting, associations remained similar for all three settings in comparison to the complete case analysis (online supplemental file 8). Lastly, a sensitivity analysis including children who died on the day of presentation showed similar associations (online supplemental file 6). Table 2Logistic regression for the association between blood pressure and admission Gambia ruralGambia urbanSurinameOR (95% CI)OR (95% CI)OR (95% CI)Z-score (continuous) Blood pressure1.11 (1.04 to 1.19)*0.86 (0.49 to 1.47)0.96 (0.85 to 1.10)Z-score (categorical) <−20.82 (0.32 to 1.70)n/a2.93 (1.36 to 6.05)* −2 to −10.88 (0.69 to 1.11)3.92 (0.63 to 30.52)0.78 (0.49 to 1.20) −1 to −00.89 (0.74 to 1.05)2.14 (0.48 to 14.80)0.92 (0.68 to 1.24) 0–1ReferenceReferenceReference 1–21.26 (0.98 to 1.60)2.25 (0.26 to 19.22)0.94 (0.61 to 1.41) >21.40 (0.94 to 2.01)6.39 (0.28 to 73.32)0.83 (0.31 to 1.88)APLS Hypotension0.63 (0.50 to 0.78)*3.22 (0.70 to 10.99)1.31 (0.85 to 1.96)PALS Hypotension0.86 (0.51 to 1.36)1.40 (0.08 to 7.59)1.73 (0.88 to 3.18)*P<0.05.APLS, Advanced Paediatric Life Support; PALS, Paediatric Advanced Life Support. Logistic regression for the association between blood pressure and admission P<0.05. APLS, Advanced Paediatric Life Support; PALS, Paediatric Advanced Life Support. Diagnostic performance The cut-offs for hypotension showed a very high specificity and a very low sensitivity in all settings (table 3). For hypotension according to APLS, sensitivity ranged from 10.8% to 21.4% and specificity ranged from 82.4% to 92.2%. Hypotension according to PALS showed a sensitivity ranging from 2.2% to 7.1% and specificity from 94.8% to 97.5%. Absolute risk for serious illness in children with hypotension according to the APLS cut-offs ranged from 2.5% to 19.3% in comparison to 3.0% to 15.4% in children with no hypotension. Absolute risk for serious illness in children with hypotension according to the PALS cut-offs ranged from 3.4% to 24.0% in comparison to 3.4% to 15.5% in children with no hypotension (online supplemental file 9). Table 3Diagnostic performance of hypotensionSensitivity (95% CI)Specificity (95% CI)Positive LR (95% CI)Negative LR (95% CI)Gambia rural Hypotension APLS11.9 (9.7 to 14.5)82.4 (81.9 to 83.0)0.68 (0.56 to 0.83)1.07 (1.04 to 1.1) Hypotension PALS2.2 (1.3 to 3.5)97.5 (97.3 to 97.7)0.86 (0.53 to 1.39)1.00 (0.99 to 1.01)Gambia urban Hypotension APLS21.4 (4.7 to 50.8)92.2 (89.0 to 94.7)2.74 (0.95 to 7.92)0.85 (0.65 to 1.12) Hypotension PALS7.1 (0.2 to 33.9)94.8 (92.1 to 96.8)1.37 (0.2 to 9.51)0.98 (0.85 to 1.13)Suriname Hypotension APLS10.8 (7.5 to 15.0)91.5 (90.0 to 92.9)1.28 (0.88 to 1.85)0.97 (0.93 to 1.02) Hypotension PALS4.6 (2.4 to 7.7)97.3 (96.4 to 98.1)1.70 (0.92 to 3.13)0.98 (0.96 to 1.01)APLS, Advanced Paediatric Life Support; LR, likelihood ratio; PALS, Paediatric Advanced Life Support. Diagnostic performance of hypotension APLS, Advanced Paediatric Life Support; LR, likelihood ratio; PALS, Paediatric Advanced Life Support. Discussion In our study across three acute care settings in LMICs, we found no clinically meaningful association between hypotension and serious illness. Adjusting for additional confounders did not change these associations. These results implicate that low blood pressure is not able to identify the most severely ill patients from the general population of children presenting to an acute care setting in LMICs, and the value of routine blood pressure measurement in otherwise well-appearing and haemodynamically stable children is limited. The assessment of low blood pressure in two ways shows that both lower values of blood pressure related to each setting’s distribution (age-adjusted z-scores based on each setting’s blood pressure values) and predefined hypotension cut-off points lack utility as a routine screening tool. Furthermore, when looking at the diagnostic value, hypotension had a high specificity and low sensitivity for the outcome measure, which indicates that it does not perform well in ruling out serious illness. In our study, a relatively high prevalence of hypotension was found, compared with blood pressure data from European paediatric ED visits (8.3–17.3% in the current study vs 3.7% in European data when using the APLS cut-offs and 2.5–5.3% vs 1.2% when using PALS cut-offs).13 However, this does not reflect that children in our study population are more ill, as there is no significant association with our outcome measure as a proxy for serious illness. We hypothesise that normal blood pressure values in children in LMICs are different, possibly due to differences in body composition or a higher prevalence of malnutrition. This has been suggested by earlier studies, which found that systolic blood pressure and diastolic blood pressure are lower in children with malnutrition.3235 We conclude that normal blood pressure values in LMICs and their association with factors such as differences in body composition and nutritional status should be further investigated, as there may be a need for contextualised reference values when defining low blood pressure and possibly also high blood pressure.36 In one setting, Gambia rural, we found a negative association between hypotension and serious illness, indicating that a low blood pressure was most often found in patients discharged home. We undertook several strategies to improve our understanding of these findings. First, we tried to explain this unexpected result by taking into account relevant confounders, such as weight and height, and imputing missing values to account for selection bias. However, these did not provide a solid explanation and did not change this negative association. Second, we explored potential errors in our data and in our statistical analyses. Due to the design of our study, we were unable to check the source data. However, extensive quality checks, including assessment of normality, outliers and implausible values, did not show any major issues. In addition, we do not expect random errors to lead to this inverse association. Third, other explanations are the misclassification of serious illness in this population. However, overnight observation (admission) and transfer to a higher level of care was considered an appropriate outcome measure reflecting severity of illness in this rural primary care station by the local clinical team, and we confirmed that severely ill children are not sent back to the community. In addition, earlier research in Gambia rural has shown significant associations between commonly considered predictive vital signs (heart rate, respiratory rate and oxygen saturation) and the outcome measure, which makes it less likely that misclassification of outcome measure or data quality was the source of this unexpected result and supports our findings.21 Fourth, it is important to note that in our data it is unknown whether age-appropriate cuff sizes were used. This could have led to generally lower blood pressures being measured if an adult-sized cuff was used. This would, however, not explain the inverse association, but could account for the high rate of hypotension measured. Lastly, we cannot exclude the option that this association is true, that serious illness leads to hypertension and that hypotension is not a warning sign in this setting. However, we consider this unlikely since this association was not found in the other two databases and has not previously been described in literature. Strengths of this study are the use of large databases from LMICs including children presenting with diverse complaints and of all ages. We were able to reuse data from existing research projects, thereby making optimal use of previously collected clinical data and avoiding putting extra strain on settings with already limited resources. Furthermore, the data included a high level of measured blood pressure (49.9–79.4%), which is unique for databases from acute care settings in LMICs. Another strength is that we included blood pressure in multiple ways: continuous as age-adjusted z-scores and categorically with the most frequently clinically used cut-off points for hypotension (APLS and PALS). This resulted in both analysing the z-scores based on each setting’s blood pressure values and the value of the cut-off points in these settings. Ultimately, this is the first study evaluating the value of low blood pressure in children attending acute care settings in LMICs. The first limitation is that this study only included children whose blood pressure was measured. This resulted in a potentially unrepresentative sample regarding age and urgency. However, if blood pressure were to be used as a screening tool in these settings, it would likely not alter the characteristics of the children in whom it is measured, and therefore this selection would reflect current clinical practice. Furthermore, we included a sensitivity analysis where we imputed missing data on blood pressure, which did not change the associations. In addition, we repeated the logistic regression analyses with inverse probability weighting where associations remained similar. With both techniques, we show that the selection bias did not change the predictive value of blood pressure. The second limitation is that we used an outcome measure as a proxy for serious illness, including (ICU) admission or referral to a higher level of care. This outcome may not be perfect, as this proxy might misclassify children, either due to admission reasons for non-serious conditions, such as fractures, or severely ill children not being admitted due to capacity reasons or other contextual factors. However, this outcome could be defined using the available data, includes commonly used markers of illness severity and was determined in consultation with the local research team. Moreover, using this outcome measure makes it possible to compare this study to studies in high-income countries.12 13 Finally, an important limitation is the reliability of blood pressure measurements, which, especially in younger children, can be challenging. Finding the correct blood pressure cuff and assuring children are calm enough to perform a valid measurement is difficult.37 It is unknown in our data whether the age-appropriate cuff size was used, which could have led to lower blood pressure measured due to an adult cuff size being used. This reflects clinical practice and will be a problem in all children presenting to an ED. However, impact will be greater in settings with limited staffing, lower levels of medical training and equipment supply issues. In conclusion, given the lack of a significant association, we do not recommend the use of blood pressure as a sole routine screening tool for well-appearing children presenting to acute care settings in LMICs. Blood pressure should only be measured on indication, for example, in ill-appearing children presenting in shock or for follow-up in admitted patients to evaluate disease course. Further research should focus on the value of hypotension in specific subgroups, such as dehydration or sepsis, to assess in which group the measurement of blood pressure would have value as a screening tool. Furthermore, the value of hypertension in children attending the ED should be evaluated, because some acute diagnoses can present with hypertension rather than hypotension, for example, increased intracranial pressure, acute kidney failure or pain. Finally, an important focus of future research should be on other vital parameters or scoring systems that aid in the assessment of children in these acute care settings, aiming for quick recognition and timely care for the children with the highest need. In low-resource settings, assignment of personnel and resources to these highly urgent children is even more important than in high-resource settings. Finally, there is a need for further studies on contextualised reference values for blood pressure based on setting or nutritional status. Supplementary material