Background
Nosocomial pneumonia is among the most frequent complications in trauma patients and is associated with increased mortality and poor prognosis [
1‐
3]. The incidence of nosocomial pneumonia ranges from 4.3 to 38.3% in the literature, and this wide variety may cast doubt on the individual studies’ comparability [
4,
5].
Several types of nosocomial pneumonia have been described in the literature [
6]. Most guidelines on nosocomial pneumonia create a distinction between hospital-acquired pneumonia (HAP) and ventilator-associated pneumonia (VAP) [
7‐
9]. Although VAP essentially is a particular type of HAP, the etiology is not the same. In VAP, endotracheal intubation enables upper respiratory tract colonization by inserting a foreign body; therefore, the two pneumonia types should not be considered equivalent [
10]. Nonetheless, the diagnostic criteria are similar for HAP and VAP in most guidelines, though they differ in the exact duration of mechanical ventilation and the time between mechanical ventilation and pneumonia onset to distinguish VAP from HAP [
7‐
9].
To diagnose hospital-acquired pneumonia, microbiologic diagnostics are superior to clinical symptoms or radiologic examination [
10]. Collecting sputum or tracheal secretions has high sensitivity but low specificity, while bronchoalveolar lavage and comparable methods have both high sensitivity and specificity. However, as fluid is introduced into the lungs, bronchoalveolar lavage is generally unsuitable for non-mechanically ventilated patients and is, therefore, mainly used to diagnose VAP [
11]. Thus, HAP diagnosis is reliant on clinical criteria.
The combination of varying incidence and diagnostic criteria reliance raises the question of what criteria have been previously used to diagnose HAP in trauma patient research [
12,
13]. Potentially, HAP incidence varies because of the use of different diagnostic criteria. Therefore, this systematic review was conducted to create an overview of reported definitions of hospital-acquired pneumonia in trauma research.
Discussion
This systematic review provides a general overview of criteria utilized in trauma patient research to diagnose hospital-acquired pneumonia. In only 46 out of 5749 original studies, well-defined criteria were reported, either pre-defined by published guidelines or clear non-guideline criteria. Forty-eight unique criteria were presented and clustered into five categories: clinical, laboratory, microbiological, radiological, and miscellaneous.
In the 28 studies without pre-defined guideline criteria to diagnose HAP, 37 unique criteria were reported. The heterogeneity in the applied criteria can mainly be attributed to the vast diversity in clinical, laboratory, and microbiological thresholds. For example, when considering leukocyte count as an indicator of HAP, up to five different thresholds were reported, describing both an elevated and decreased leukocyte count as indicative of HAP. One could imagine that a lower cut-off point of leukocytosis (e.g., 10 × 109/L versus 13 × 109/L) may lead to a higher estimate of HAP cases in a research population. Similar threshold differences were observed for body temperature, including “fever” or “febrile” as subjective criteria.
Some studies cited established guidelines as a basis for diagnosis, but the authors added new criteria or deleted pre-defined criteria, thus introducing (potential) aggregate bias. For instance, four studies added “medical record documentation” or “start of antibiotic treatment” as a criterion to diagnose pneumonia in addition to guideline criteria [
12,
28,
30,
61]. Also, several studies added specific criteria (e.g., hypothermia, worsening gas exchange, leukopenia, and bronchoalveolar lavage) to the ATS/IDSA criteria [
12,
22,
34,
61,
64]. Though all are clinically relevant criteria, adjusting pre-defined criteria complicates the comparison of studies that use the same guidelines and increases bias.
Eighteen studies applied existing guideline criteria to diagnose HAP. However, five different guidelines were encountered, leading to a further decrease in uniformity. We encountered a similar variation in body temperature and leukocyte count cut-offs (Table
2) [
7‐
9,
65‐
67]. However, the number of variations was lower for the pre-defined guideline criteria: three versus five cut-offs for body temperature and leukocytosis. The 2015 ATS/IDSA guideline contained no distinct thresholds for hyperthermia and leukocytosis, resulting in differences between studies that used this guideline (Table
2) [
67]. Despite the attempts to generate uniformity in diagnosing HAP by creating and using guideline criteria, the abovementioned differences make it difficult to compare the available literature completely. Only five studies diagnosed HAP based on the exact same criteria.
Guidelines are continuously updated based on new insights and available literature. The earliest CDC guideline dates from 1988, and the most recent from 2018. During these 30 years, the criteria for pneumonia diagnosis have changed substantially. For example, the 1988 CDC criteria for pneumonia diagnosis were clinical, radiologic, or microbiologic, while body temperature was not included as a criterion [
68]. However, the 2018 guideline provides a more elaborate set of clinical, radiological, microbiological, and laboratory criteria [
7]. As a result, it is more difficult to compare older data sets to more recent studies. Full implementation of or compliance with guideline criteria in clinical practice is hardly feasible for two reasons: patient care and study design. Study subjects are patients; therefore, clinical examination and experience remain decisive in starting pneumonia treatment. The authors understand that an inconclusive or negative X-ray should not delay antibiotic treatment, and awaiting a microbial culture is not mandatory or necessary in seriously ill patients. Using guideline criteria for pneumonia diagnosis in retrospective studies might be impracticable. Also, uniform diagnostic criteria for pneumonia are hard to accomplish in database or registry studies. Nonetheless, these limitations should be mentioned when encountered.
Previously, review studies have been issued on lacking definitions in trauma research, such as fracture-related infections and non-unions of long bones [
69‐
71]. To resolve a lack of definition, these review studies provide a basis for a consensus definition. Subsequently, Delphi method studies can be helpful in reaching consensus. Our study displays the wide variety of clinical criteria for HAP diagnosis in trauma research and exhibits how studies ought to be compared with caution. The comparison of results is essential in trauma patient research and for guidelines. Guideline issuers pursue workable and representable guidelines to help clinicians in decision-making. Continuous improvement is established with the results of clinical studies, for which comparability of results is necessary. Our results emphasize this importance. Though our study addressed a scientific problem rather than a clinical one, it can still impact day-to-day practice.
One established definition of HAP would improve the comparability of trauma research; expert consensus could be a solid foundation to start with. Given the complexity of trauma patients, any diagnostic definition should address potential issues and pitfalls to avoid overdiagnosis. Currently, hyperthermia is incorporated in all guidelines, and sputum and dyspnea (with or without worsening gas exchange) in all but one. Leukocytosis, a common marker for infection, is included in all guidelines. Microbiologic information and evidence of infection aid in diagnosis and treatment; the CDC and ECDC describe several types of respiratory cultures. Radiologic evidence of pneumonia—either radiographic or CT imaging—also supports a diagnosis. Expert consensus should incorporate these criteria. Nonetheless, hypo- or hyperthermia, dyspnea, and leukocytosis are also signs of the systemic inflammatory response syndrome, commonly observed in trauma patients and potentially complicating the diagnostic process [
72]. Also, posttraumatic fever may have a non-infectious origin, such as neurogenic fever, and trauma is associated with an increased immune response, adding to the need for a dedicated leukocytosis threshold [
73,
74]. Lastly, sputum can result from (severe) pulmonary contusion, though unlikely to be purulent [
75]. We propose that a decision-making algorithm includes hyperthermia (≥ 38.5 ℃) and leukocytosis (> 12 × 10
9/L) as major criteria, in addition to microbiologic and radiologic evidence. Dyspnea and (purulent) sputum should be considered minor criteria. Our considerations and recommendations serve as a basis for expert consensus.
Some limitations of this study should be considered. Firstly, PubMed/MEDLINE was the only search engine used in this study, which could result in an incomplete overview of applied clinical criteria for HAP diagnosis. However, the wide variety of clinical criteria and the difficult comparison between studies are evident in the current number of included studies. Secondly, our overview of reported diagnostic criteria did not consider the recommended combinations of these criteria. Not doing so resulted in a more comprehensible overview of used criteria and did not diminish the conclusion of this study.