Although the cornerstone of community-acquired pneumonia (CAP) treatment is now antimicrobial therapy, diagnostic methods, and supportive care, CAP continues to be a leading cause of hospitalisation, morbidity and mortality in the adult population worldwide. The clinical course of CAP is variable, and may include relatively mild disease, or rapid deterioration into respiratory failure, septic shock and death. Therefore, it is important to recognize patients who are at risk for adverse outcomes early in their hospital stay [1]. At the moment, international guidelines suggest to perform the severity assessment at the time of presentation in order to decide the need for hospitalization, intensive care, antimicrobial treatment, and level of monitoring [1].

There are some validated severity assessment tools for CAP, such as the Pneumonia Severity Index (PSI) and CURB-65 score. While these tools offer useful prognostic information, patients that have been initially identified as non-critical patients can later deteriorate while hospitalized. Therefore, physiological abnormalities and acute organ dysfunction may give some other information about impending deterioration than just the severity classification alone [2]. Escalation to mechanical ventilation, vasopressor use, or intensive care unit admission during the first few days of hospitalization is a marker of worse clinical outcomes [2,3] and is of particular concern with early deterioration.

A number of potential factors have been found to signal an unfavorable early course in the past. Previous studies have shown failure to reach early clinical stability associated with abnormal respiratory rate, hypoxemia, hypotension, altered mental status, comorbidities, and higher predicted disease severity at presentation [2,4]. Multi-lobar pulmonary involvement and pleural complications have also been reported to correlate with delayed stability or early treatment failure and are radiological features to be taken into consideration [5] repeatedly. Routine laboratory and inflammatory markers can also help in identification of high risk patients if used in combination with clinical parameters. More recent studies have confirmed that non-stability within the first 72 hours is a predictive factor for increased subsequent mortality and intensive care unit use [6].

These predictors may be different in each healthcare system due to the variation in patient characteristics, comorbidities, delays in presentation, antimicrobial exposure, and access to critical care. In Pakistan, there has been substantial mortality among hospitalized adults with CAP and several factors have been found to be associated with adverse outcome, including the presence of a raised blood urea and altered mental state [7]. Data regarding early clinical deterioration, however, specifically in the first 72 hours after admission to hospital is still sparse at local level. Early identification of easily measurable predictors on admission may help with better surveillance, prompt escalation of care and the appropriate prioritisation of scarce critical-care resources. This study was thus designed to identify clinical, laboratory and radiological predictors of early clinical deterioration in adult patients admitted for community acquired pneumonia (CAP) in a tertiary care hospital in Pakistan.

Design studies and setting studies. The study was a prospective observational cohort study in the Department of Medicine of a tertiary care teaching hospital. Eligibility to participate was assessed sequentially in the admitted adult patients with CAP during the study period. The aim of the study was to identify clinical, laboratory and radiological factors associated with early clinical deterioration in hospitalized patients with CAP. Study Population Adult (age 18 years and older) patients admitted with community-acquired pneumonia were included. The criteria for CAP were an acute onset of lower respiratory tract infection (LRTI) that occurred in the community or outside of a hospital or health-care facility with symptoms that were compatible with pneumonia (cough, production of sputum, dyspnea, fever, pleuritic chest pain, altered mental status) and a new pulmonary infiltrate seen on chest radiography or computed tomography (CT) scan that was consistent with pneumonia and could not be explained by other diagnoses. HAP (pneumonia that developed 48 hours or more after hospital admission) and VAP (pneumonia associated with mechanical ventilation) patients were excluded. Other exclusion criteria included current pulmonary tuberculosis, post-obstructive pneumonia in the context of a known lung malignancy, severe degree of immunosuppression or an alternative diagnosis that could explain the pulmonary infiltrates. The study excluded patients who were already receiving invasive mechanical ventilation or vasopressor support at the time of the initial hospital assessment, since they achieved the study outcome at baseline. The sampling technique and sample size. A non-probability consecutive sampling technique was used. The patients who met the inclusion criteria within the study period were recruited consecutively up to the desired sample size. The minimum sample size was determined based on the previously reported early clinical deterioration frequency for the specific analysis, a 95% level of confidence, a 5% margin of error and an 80% level of statistical power. For the patients, additional patients were considered to compensate for incompletion, early discharge, transfer, or missing outcome data. Baseline Clinical Assessment The same assessment protocol was used for all participants at admission. A structured data collection proforma was used to collect baseline variables prior to the primary outcome. Demographic factors were age, sex, BMI, and smoking. Clinical data consisted of the duration of illness prior to hospitalization, presence of prior antibiotic treatment and comorbidities. Comorbid diseases were diabetes mellitus, hypertension, chronic obstructive pulmonary disease, bronchial asthma, ischemic heart disease, heart failure, chronic kidney disease, chronic liver disease, previous cerebrovascular disease and malignancy. The following physiological parameters were obtained at presentation: body temperature, heart rate, respiratory rate, systolic and diastolic blood pressure, mean arterial pressure, peripheral oxygen saturation (SpO₂), need for supplemental oxygen, and level of consciousness. Altered mental status was noted at presentation as either new onset disorientation or impaired consciousness. Hypoxemia was considered as SpO₂ <90% on room air or supplemental oxygen needed to keep SpO₂ ≥90%. Tachypnea defined as RR ≥ 30 breaths/min, hypotension documented as SBP<90 mmHg and/or DBP<60 mmHg. Laboratory Evaluation Baseline laboratory assessment was carried out using blood samples collected during routine clinical assessment at admission. Laboratory data recorded were hemoglobin concentration, total leucocyte count, neutrophil count, lymphocyte count, platelet count, blood urea concentration, blood creatinine concentration, blood sodium concentration, blood potassium concentration, random blood glucose, C-reactive protein, liver function tests. The neutrophil to lymphocyte ratio (NLR) was defined as the absolute neutrophil count / absolute lymphocyte count. When clinically warranted, arterial blood gas parameters (pH, PaO2 and PaCO2) were obtained. Blood cultures and sputum cultures were obtained before antibiotic administration whenever clinically feasible; however, microbiological confirmation was not mandatory for inclusion because CAP was primarily diagnosed on clinical and radiological grounds. Radiological Assessment All patients underwent chest radiography at admission. Computed tomography of the chest was performed when clinically indicated. Radiological findings were categorized according to the distribution and extent of pulmonary involvement. The presence of unilateral or bilateral infiltrates, single-lobar or multilobar involvement, consolidation, and pleural effusion was recorded. Multilobar pneumonia was defined as radiographic involvement of two or more pulmonary lobes. Assessment of Pneumonia Severity Disease severity at admission was evaluated using the CURB-65 score. One point was assigned for each of the following: new-onset confusion, blood urea nitrogen >19 mg/dL (>7 mmol/L), respiratory rate ≥30 breaths/min, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg, and age ≥65 years. The total score ranged from 0 to 5. Where all required variables were available, the Pneumonia Severity Index (PSI) was also calculated to provide additional baseline risk stratification. Severity scores were calculated from information available at initial assessment and were not modified using clinical events occurring after admission. Treatment and Patient Monitoring Patients received standard-of-care management for CAP according to institutional protocols and contemporary international recommendations. Treatment decisions, including antibiotic selection, oxygen supplementation, intravenous fluid therapy, non-invasive ventilation, invasive mechanical ventilation, vasopressor use, and intensive care unit admission, remained at the discretion of the treating physicians and were not influenced by participation in the study. After admission, patients were prospectively monitored for changes in respiratory, cardiovascular, and neurological status. Vital signs, oxygen requirements, escalation of respiratory support, hemodynamic instability, and transfer to a higher level of care were documented during the first 72 hours of hospitalization. Primary Outcome: Early Clinical Deterioration The primary outcome was early clinical deterioration within 72 hours of hospital admission. To ensure objective and reproducible outcome classification, early clinical deterioration was defined as the occurrence of at least one of the following after the baseline assessment: • Transfer to the intensive care unit because of worsening pneumonia; • Initiation of non-invasive ventilatory support or invasive mechanical ventilation because of acute respiratory deterioration; • Development of septic shock requiring vasopressor therapy despite adequate fluid resuscitation; or • All-cause death within 72 hours of admission. Patients who fulfilled at least one of these criteria were categorized as the early clinical deterioration group, whereas patients who survived without any of these events during the first 72 hours were categorized as the non-deterioration group. When more than one deterioration event occurred in the same patient, the earliest event was used to determine the time to deterioration, while all components were recorded separately. Data Quality and Missing Data Data were collected using a predefined structured proforma. Baseline variables were obtained from the initial clinical examination, medical records, laboratory reports, and radiological findings. Data were reviewed for completeness and internal consistency before entry into the statistical database. Patients for whom the primary 72-hour outcome could not be reliably determined were excluded from the primary outcome analysis. The number and proportion of missing observations for individual predictor variables were documented, and no unavailable value was assumed to be normal. Statistical Analysis Data were entered and analyzed using IBM SPSS Statistics for Windows, version 27.0 (IBM Corp., Armonk, NY, USA). Continuous variables were initially assessed for distribution using graphical methods and the Shapiro-Wilk test. Normally distributed continuous variables were expressed as mean ± standard deviation (SD), whereas non-normally distributed variables were reported as median and interquartile range (IQR). Categorical variables were summarized as frequencies and percentages. Patients with early clinical deterioration were compared with those without deterioration. The independent-samples t-test was used to compare normally distributed continuous variables, while the Mann-Whitney U test was used for non-normally distributed continuous variables. Categorical variables were compared using the Pearson chi-square test or Fisher's exact test when expected cell frequencies were small. To identify predictors of early clinical deterioration, univariable binary logistic regression analysis was initially performed. Crude odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Variables considered clinically relevant and variables demonstrating an association at p <0.20 in univariable analysis were considered for inclusion in the multivariable model. A multivariable binary logistic regression analysis was subsequently performed to identify factors independently associated with early clinical deterioration. Adjusted odds ratios (aORs), 95% CIs, and corresponding p-values were reported. The number of variables entered into the multivariable model was restricted according to the number of observed outcome events to minimize model overfitting. Clinically overlapping variables and individual components of composite severity scores were not entered indiscriminately into the same model. Multicollinearity among candidate predictors was assessed using the variance inflation factor (VIF) and tolerance statistics. Model calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, while discriminatory performance was assessed using the area under the receiver operating characteristic curve (AUC/ROC) where appropriate. Statistical tests were two-sided, and a p-value <0.05 was considered statistically significant.

Patients who deteriorated clinically early in the course of the illness were typically older and had more comorbid diseases at baseline. Deterioration on unadjusted comparison was significantly associated with advanced age, diabetes mellitus, COPD, and chronic kidney disease; other factors such as sex, BMI, smoking, hypertension, cardiovascular co-morbidity, chronic liver disease, cerebrovascular disease, malignancy were not significantly different between the groups. The duration of symptoms before admission was also a little longer for patients who deteriorated, while there was no difference between groups in the number of days on which antibiotics were used prior to admission.

Table 1. Baseline demographic and clinical characteristics according to early clinical deterioration

Legend: Values are presented as mean ± SD, median (IQR), or n (%). Independent-samples t-test, Mann–Whitney U test, Pearson's chi-square test, or Fisher's exact test was used as appropriate. COPD = chronic obstructive pulmonary disease; BMI = body mass index.

Patients with early clinical deterioration had significantly more physiological and clinical severity. They had higher rates of heart rate, respiratory rate, hypotension, hypoxemia, altered mental status and supplemental oxygen use, but lower levels of blood pressure and oxygen saturation. The CURB-65 and PSI scores were significantly different between the groups, with higher scores in those who deteriorated, while there was no significant difference in admission temperature.

Table 2. Admission clinical and physiological characteristics according to early clinical deterioration

Legend: Data are presented as mean ± SD, median (IQR), or n (%). BP = blood pressure; SpO₂ = peripheral oxygen saturation; PSI = Pneumonia Severity Index. Independent-samples t-test, Mann–Whitney U test, and Pearson's chi-square test were used as appropriate.

A greater inflammatory and biochemical disturbance on admission was related to early clinical deterioration. Deteriorating patients were found to have elevated leukocytes, neutrophils, NLR, serum urea, creatinine, blood glucose and CRP and low lymphocyte count and serum sodium levels. Radiologically, patients who deteriorated had significantly more bilateral disease, more multilobar involvement and more pleural effusion. Hemoglobin, platelet count, serum potassium, consolidation and culture positivity did not demonstrate statistically significant differences between groups.

Table 3. Laboratory and radiological findings according to early clinical deterioration

Legend: Continuous data are presented as mean ± SD or median (IQR), as appropriate. t = independent-samples t-test statistic; U = Mann–Whitney U statistic; χ² = Pearson's chi-square statistic; NLR = neutrophil-to-lymphocyte ratio; CRP = C-reactive protein. Culture positivity refers to clinically obtained blood and/or sputum cultures.

Of the patients who were hospitalized, around 25% suffered from early clinical deterioration within 72 hours, and most had no clinical deterioration within this time frame. The most common aspect of deterioration was ICU transfer, followed by need for non-invasive ventilatory support, invasive mechanical ventilation, and vasopressor therapy for septic shock. Early mortality was the least common of the composite outcome. Some patients had more than one adverse event as they were not mutually exclusive.

Figure 1. Early clinical deterioration and its components within 72 hours of admission

Legend: Early clinical deterioration was defined as the occurrence of at least one predefined deterioration event within 72 hours of hospital admission. Individual components were not mutually exclusive; therefore, their frequencies should not be summed to obtain the total number of patients with deterioration. ICU = intensive care unit.

Several factors were associated with early clinical deterioration in univariable analysis, which included demographic, comorbid, physiological, inflammatory and radiological factors. In the multivariable logistic regression model, however, tachypnea, hypoxemia, altered mental status, higher NLR and multilobar pulmonary involvement remained independent risk factors. The other variables such as age, diabetes, COPD, chronic kidney disease, hypotension, CRP and pleural effusion lost statistical significance after adjustment, indicating that initial respiratory compromise, neurological status, systemic inflammatory response and pulmonary involvement were the main determinants of independent risk of deterioration.

Table 5. Univariable and multivariable logistic regression analysis of predictors of early clinical deterioration

Legend: Binary logistic regression was performed with early clinical deterioration within 72 hours as the dependent variable. OR = odds ratio; aOR = adjusted odds ratio; CI = confidence interval; COPD = chronic obstructive pulmonary disease; SpO₂ = peripheral oxygen saturation; NLR = neutrophil-to-lymphocyte ratio; CRP = C-reactive protein. Wald χ² statistics are shown for regression coefficients. Variables with p <0.20 in univariable analysis and/or clinical relevance were considered for multivariable analysis. A two-sided p <0.05 was considered statistically significant.

The present study showed that a combination of baseline patient characteristics, extent of pulmonary involvement, physiological derangement and inflammatory abnormalities were associated with early clinical deterioration in hospitalized patients with community-acquired pneumonia (CAP). Older patients and higher diabetes mellitus, chronic obstructive pulmonary disease, and chronic kidney disease burden were associated with those who were deteriorating. These findings are generally similar to those of the Pakistani study by Iqbal et al., which showed that greater CAP severity and increased age were associated with poor outcomes when they analysed 509 CAP adult in-patient cases in Karachi [8]. In the same way, Ali et al. in their prospective study on CAP mortality in Peshawar, found that age was an independent risk factor for in-hospital mortality [9]. However, in our adjusted model, age and individual comorbidities were no longer significant separately, indicating that these factors contribute to early deterioration in part because of the severity of acute physiological compromise, but also because of other factors. The clinical parameters at admission were particularly well correlated with subsequent deterioration, with respiratory and hemodynamic abnormalities. Patients who worsened were more likely to be tachycardia, tachypnea, have low blood pressure and low saturation levels, altered mental status, and need supplemental oxygen. These results are similar to the results of the international REA-ICU study, which analyzed 6,560 patients who were hospitalized with CAP and found that respiratory rate ≥30 per minute, hypoxemia, abnormal heart rate, elevated blood urea nitrogen, hyponatremia, and extent of the radiographic disease were predictors of admission to the ICU within the first 3 days of hospitalization [10]. This is particularly relevant because the time horizon of that study is very similar to the 72 hour endpoint of deterioration that was used in the present study. In another study using the above criteria for severe CAP, hypoxemia, confusion, and other physiological abnormalities were strongly associated with admission to an intensive care unit [11]. Therefore, parameters available at the bedside may offer valuable early signs of deterioration, before the use of advanced organ support is required. Further supporting the relationship between initial disease severity and short-term adverse outcomes was the significantly higher CURB-65 and PSI severity among deteriorating patients. Iqbal et al. found that a CAP score of CURB-65 ≥ 3 was a standalone predictor of in-hospital mortality in CAP patients in Pakistan [8]. Other severity criteria used to identify severe CAP and the need for higher level care included in the 2019 ATS/IDSA CAP guideline are validated respiratory rate, confusion, hypotension, hypoxemia, uremia and multilobar infiltrates [12]. The present results also highlight that individual physiological parameters were still informative when deterioration was taken into account at an early stage, as opposed to death. This separation is clinically significant, as a patient can be so ill that they need to be transferred to the intensive care unit or need ventilatory support but not die. Laboratory results also showed that the systemic inflammatory and metabolic response was more intense with deterioration. Patients who deteriorated had elevated levels of leukocytes, neutrophils, NLR, urea, creatinine, glucose, and CRP, as well as low levels of serum sodium, with lymphopenia. Previous Pakistani studies corroborate the prognostic role of renal biochemical abnormalities; Irfan et al. observed that elevated blood urea was a significant predictor of mortality in CAP patients admitted to hospital [13] and Adnan et al. in another study in Lahore found that elevated serum creatinine was a useful predictor of in-hospital mortality in CAP patients [14]. The present unadjusted results for urea and creatinine were similar and renal indices were not the most important independent predictors of early deterioration in the final model. Importantly, after adjustment NLR was independently associated with early deterioration. NLR can be a low-cost screening tool that combines neutrophilic inflammatory activation with relative lymphocyte suppression and could thus be particularly useful in low-resource healthcare environments. An NLR cut-off of >3.5 separated 30-day non-survivors significantly from survivors in a meta-analysis of 5220 patients with CAP, and the prognostic value of NLR was significantly greater when reassessed early in the course of CAP [15]. Higher NLR was correlated with an almost doubling of all-cause mortality and with higher odds of admission to the intensive care unit (ICU) by more recent pooled evidence based on 12 observational studies [16]. While the mortality and admission to the intensive care unit did not equal the composite deterioration endpoint in the present study, the results suggested that NLR could be a readily available parameter to assist in clinical determination of severity. Patients who deteriorated were also differentiated by radiological disease extent. Bilateral infiltrates, multilobar involvement and pleural effusion were more common in the deterioration group, and multilobar involvement was independently associated with the primary outcome. In line with Iqbal et al., who found chest X-ray infiltrates in both lungs to be an independent predictor of mortality in a CAP population within Pakistan [8] this was found to be the case. The international REA-ICU study also reported that multilobar infiltrates or the presence of pleural effusion were independently associated with ICU admission within three days with an adjusted odds ratio of 2.19 [10]. These results are biologically plausible because they are related to the degree of extension of the pulmonary parenchymal process, which could lead to ventilation-perfusion mismatch and impaired gas exchange, and thus to respiratory failure. Around one quarter of the patients in the present study deteriorated, with the most common event being transfer to the intensive care unit (ICU), which was followed by non-invasive ventilation, invasive mechanical ventilation, requirement for vasopressor therapy and early death. The proportion should be interpreted with caution when compared to previous studies as the outcomes were heterogeneous across published studies, ranging from ICU admission, treatment failure, in-hospital mortality, to 30-day mortality. In the REA-ICU cohorts, 4.6% of initially non-critical CAP patients ended up in ICU during days 1–3 [10] compared with a large cohort in the United States, which found about 18% of CAP patients were hospitalized in the ICU [17,19]. However, the events rates of serious populations have been reported in Pakistan significantly higher, for instance Ali et al reported 45% in-hospital mortality in a selected population who had severe CAP with respiratory failure [9]. Thus, the differences in frequency across the studies are mainly due to variations in case mix, severity at enrollment, definition of outcome and period of follow-up rather than to differences in quality of care. The multivariable results led to the main result of the present study: respiratory rate ≥30/min, SpO₂ <90%, altered mental status, elevated NLR, and multilobar involvement were all independent predictors of early clinical deterioration, while the other variables (age, diabetes, COPD, chronic kidney disease, hypotension, CRP, and pleural effusion) were not. The pattern is similar to known severe-CAP patterns, with tachypnea, impaired oxygenation, confusion, and multilobar infiltrates as indicators of acute physiological severity [11,12]. The presence of altered mental status and hypoxemia were also independently associated with pneumonia-related mortality, and other acute physiological changes such as elevated blood urea provided further information [18,20]. These comparisons indicate that the individual clinical parameters of acute respiratory compromise, neurological dysfunction, systemic inflammatory response and radiological disease extent may be more useful than demographic characteristics and comorbidities in predicting early deterioration. These findings would be especially relevant in practice if hospital availability of intensive care (ICU) facilities and advanced biomarker testing is constrained in Pakistan. These parameters are easily accessible at or soon after admission and have minimal added cost, including respiratory rate, oxygen saturation, mental status, complete blood count-derived NLR, and chest radiographic extent. This may result in identification of seemingly stable outpatients with CAP who need more close monitoring and prompt escalation of care. However, these predictors should be used in addition to validated severity assessments and clinical judgment, and should not be used as a substitute for either, especially given that the present outcome was specifically for deterioration within the first 72 hours and is distinct from the mortality endpoints in some comparator studies. Limitations There were some limitations to this study. The single-center observational study design might restrict the applicability of the results to other centers and populations in Pakistan. Relatively short follow-up of 72 hours was performed, but failed to assess subsequent in-hospital or post-discharge outcomes. Certain laboratory/microbiological tests were carried out when clinically indicated only and hence the availability of such tests may not have been consistent. In addition, there was the possibility of residual confounding by unmeasured factors, such as variability in pre-hospital care, pathogen characteristics and response to treatment.

Significant clinical deterioration was present in a large number of patients admitted to hospital with CAP. Among these, tachypnea, hypoxemia, abnormal mental status, and an elevated neutrophil-to-lymphocyte ratio (NLR) were independent predictors of deterioration within 72 hours. The results underscore the need for systematic evaluation of the respiratory condition, neurological status, inflammatory response and extent of radiological disease on admission. These parameters are easily available in clinical routine and could help to identify patients early who need more frequent monitoring or prompt escalation of care, especially in low-resource settings.

Future Perspective

Future multicenter prospective studies with larger samples and longer follow-up should externally validate these predictors and evaluate their integration into a practical risk-prediction model for early deterioration in CAP.