Does This Patient Have an Infection of a Chronic Wound?

An 89-year-old nursing home resident with dementia and limited mobility and no other medical illnesses presents with a stage III pressure ulcer in his coccyx region. The wound has been present for 3 months and has recently been increasingly painful. Appropriate support surfaces and transfer mechanisms have been initiated to ensure that pressure in the area has been significantly reduced. On examination, there is no fever, wound erythema, or purulent exudate. Swabs taken from the surface of the wound show moderate growth of mixed bacterial flora. Is this chronic wound infected?

Chronic wounds are those that have not undergone an orderly healing process.¹ They are not always infected. A survey of clinicians in the United States who specialize in wound care found that most practitioners rely solely on the clinical examination to identify infection.² Quiz Ref IDAlthough the classic signs of infection (purulent exudate, heat, edema, and erythema) sometimes aid diagnosis, experts in chronic wounds argue that these features may be absent.³ The current reference standard for diagnosis of infection is a deep tissue biopsy culture.

Knowing the baseline prevalence of infection is crucial to understanding the incremental diagnostic value of various elements of the bedside examination and laboratory testing of wound swabs.⁴ Infected pressure ulcers are one of the leading causes of infections in long-term care facilities and can lead to sepsis.^{5 - 6} Quiz Ref IDPatients with diabetes are prone to infection and have at least a 10-fold greater risk of being hospitalized for soft tissue and bone infections of the foot than individuals without diabetes.⁷ Half of all diabetes-associated foot wounds become clinically infected during the course of hospitalization.⁸ More than 85% of spinal cord–injured patients develop an infected pressure ulcer at least once in their life, and 70% have multiple infected pressure ulcers.⁹

Some clinicians have a low threshold to prescribe empirical antibiotic treatment and determine retrospectively that wounds were infected when they respond to antibiotics. However, this permissive strategy has not been evaluated and it may contribute to unnecessary costs, adverse drug effects, and the development of antibiotic-resistant bacteria.¹⁰ Individuals with diabetic foot ulcers who have previously used antibiotics have an increased risk of ulcer infection with methicillin-resistant Staphylococcus aureus and a resultant increased risk of amputation.¹⁰

Many studies of grafts, flaps, and other surgical procedures for management of chronic wounds exclude infected wounds, but no standard objective criteria exist to define infection in clinical settings. In persons with spinal cord injury and pressure ulcers, ensuring that ulcers are not infected can help identify whether the wound is appropriate for surgical correction. Surgeons evaluating wounds preoperatively by visual inspection before flap closure are unable to reliably determine which wounds are infected,¹¹ which is important because infected chronic wounds have significantly decreased flap survival.¹¹

Thus, it is important to better characterize features of the history and physical examination and noninvasive tests such as swab cultures that might help in the diagnosis of infection of a chronic wound. Unfortunately, a standard technique for obtaining a swab culture does not exist and culture results are not systematically reported; different laboratories might report nonquantitative, semiquantitative, or quantitative culture results. Each type of result requires a specific form of swab and culture, some of which may not be widely available.

This article addresses 2 pertinent clinical questions about chronic wounds: whether the chronic wound is infected and whether there is a preferred swab technique to use for culturing the wound. We will be referring to the diagnosis of infection in existing chronic wounds (for example, pressure ulcers, venous stasis ulcers, and diabetic foot ulcers). This article does not discuss the diagnosis of osteomyelitis because that was the subject of a previous JAMA Rational Clinical Examination article.¹²

We searched MEDLINE, EMBASE, and CINAHL databases through November 18, 2011, with a previously published search strategy used in all articles in the Rational Clinical Examination series.¹² The strategy combines 9 exploded Medical Subject Headings (physical examination, medical history taking, professional competence, sensitivity and specificity, reproducibility of results, observer variation, routine diagnostic tests, decision support techniques, and Bayes theorem). We restricted our search to articles in English and humans. To select studies about infection of chronic wounds, we added as search terms the text words “chronic wound” and “infection.” For each study we identified, we reviewed the bibliography to find additional references.

Articles were included if they were original studies describing historical features, physical examination maneuvers, or laboratory markers or were radiologic studies in the diagnosis of infection of chronic wounds among adult patients; if data could be extracted to construct 2 × 2 tables or the study reported operating characteristics of the diagnostic measure; and if the diagnostic test was compared with the reference standard test (deep tissue biopsy culture; see below) or other commonly used standards (clinical features, swab cultures, and laboratory markers). We included chronic wounds that were surgically debrided as part of the study. Studies of acute wounds (including traumatic wounds, postoperative surgical wounds, and burns) and osteomyelitis were excluded.

The reference standard for the diagnosis of infection of chronic wounds is a deep tissue biopsy culture. Microbial loads of greater than 10⁵ of any organism per gram of wound tissue or the presence of any level of β-hemolytic streptococcus is typically considered an indicator of infection of chronic wounds.^{11 ,13 - 14} Deep tissue biopsy can usually be performed in an outpatient setting but is an invasive procedure, often requires local anesthesia, may be contraindicated in patients with leg or foot ulcers from severe peripheral vascular disease, and, in specific situations, requires that the specimen be collected and processed appropriately for aerobic, anaerobic, fungal, and mycobacterial organisms.^{13 ,15}

Other standards used for the diagnosis of infection of chronic wounds include wound swab cultures and laboratory markers. Because these standards are so commonly used in clinical practice and research reports, we evaluated the performance of clinical findings and procedures using these standards of uncertain validity for comparison with their accuracy against the reference standard of deep tissue biopsy culture.

One author (M.R.) identified potential studies by screening the retrieved studies. Three authors independently reviewed studies for quality (M.R., W.W., and S.R.K.) and 1 author (W.W.) extracted the operating characteristics of the diagnostic tests. Each study was rated with a topic-specific quality rating scale that used published principles,¹⁶ as well as a modified quality checklist specific to the Rational Clinical Examination series (Box).¹⁷

Level 1: independent, blind comparison of test (ie, sign, symptom, or investigation) results with a reference standard of anatomy, physiology, diagnosis, or prognosis among a large number of consecutive patients (≥75)

Level 2: independent, blind comparison of test results with a reference standard among a small number of consecutive patients

Level 3: independent, blind comparison of test results with a reference standard among nonconsecutive patients with suspected target condition

Level 4: nonindependent comparison of test results with a reference standard among a “grab” sample of patients who obviously have the target condition (and perhaps healthy individuals)

Level 5: nonindependent comparison of test results with a standard of uncertain validity (which may incorporate the test results in its definition) among “grab” samples of patients (and perhaps healthy individuals)

^aBased on Simel.¹⁷

Likelihood ratios (LRs) predicting the presence of infection, given a positive test result (sensitivity/1 − specificity) and a negative test result (1 − sensitivity/specificity), were calculated for each outcome of interest with published raw data and then rounded off. We calculated 95% CIs by using the method described by Simel et al.¹⁸ When any 2 × 2 table included a value of 0, we added 0.5 to each cell of the 2 × 2 matrix to calculate the 95% CIs for the LRs. A random-effects summary estimate of prevalence was obtained with the derSimonian-Laird approach and heterogeneity was assessed with the I² statistic.

All the included studies reported the number of wounds and the number of patients. Only 2 studies^{19 - 20} included patients who had more than 1 wound. We therefore calculated our results according to the number of wounds in each study, not the number of patients.

The search strategy identified 341 abstracts, from which 15 relevant studies were selected (eFigure).^{19 - 33} Table 1 provides details about the 6 quality level 1 to 3 studies (see eTable 1 for details of the 9 quality level 4-5 studies). The initial agreement between the 3 raters for the level of evidence was 80%, although all differences were resolved with consensus. The 15 studies included 985 patients with a total of 1056 chronic wounds. Eight studies prospectively evaluated the accuracy of findings compared with a reference standard test, although none assessed their interobserver variability.^{20 ,22 ,24 ,26 - 27 ,29 ,31 - 32} Because most of these studies combined wounds arising from multiple etiologies, we could not evaluate the prevalence and accuracy of the various diagnostic features according to patient characteristics or wound type.

The prior probability of infection in a chronic wound can be inferred from the prevalence of infection of chronic wounds in the level 1 to 3 studies (summary prevalence, 45%; 95% CI, 32%-58%; I² = 82%; P < .001 for heterogeneity). The prevalence of infection of chronic wounds in the level 4 to 5 studies was higher (summary prevalence, 59%; 95% CI, 40%-76%; I² = 97%; P < .001 for heterogeneity). The prevalence of infection of chronic wounds in all 15 studies was 53% (95% CI, 40%-67%; I² = 95%; P < .001 for heterogeneity).

No symptom or sign was evaluated in more than 2 studies, so the LRs from individual studies with their 95% CIs or LR ranges were used to summarize the results.

Symptoms. Two studies (both quality level 2) from the same investigator showed that the likelihood of infection increases when an ulcer causes increasing pain (LR range, 11-20).^{21 ,25} The absence of increasing pain (LR range, 0.64-0.88) was not as useful as its presence.

Signs. Purulent exudate, erythema, heat, and edema (the “classic signs” of wound infection) were not helpful in diagnosing infection of chronic wounds, with LR ranges including values less than 1.0.^{21 ,25} Wounds that had no serous exudates (LR range, 0.57-0.62) or those that were healing rapidly (LR range, 0.29-0.96) were less likely to be infected. Foul odor, which is often considered a sign of infection, did not significantly predict the diagnosis of infection.

Combinations of Findings. Gardner et al²¹ examined the validity of a combination of signs recommended by the Infectious Diseases Society of America for identifying diabetic foot ulcer infections. The society's guidelines recommend using the presence of a purulent exudate or 2 or more indicators of inflammation (pain, erythema, induration, heat, or edema).³⁴ Gardner et al²¹ found that this combination of symptoms and signs has only 52% sensitivity and 46% sensitivity, which yields a positive LR of 0.96 (95% CI, 0.60-1.6) and negative LR of 1.0 (95% CI, 0.61-1.8). Thus, the combination of findings recommended by a panel of experts may not be helpful, pointing out the difficulty in clinically identifying infection of chronic wounds.

Comparisons With Reference Standard. Four studies evaluated the utility of noninvasive techniques (ie, swab cultures or laboratory markers) compared with the reference standard (deep tissue biopsy culture) for the diagnosis of infection of chronic wounds. The 4 studies included a total of 198 patients.^{23 - 24 ,26 - 27}

One of the 2 studies that examined the utility of swab cultures was of quality level 1 and showed that a quantitative swab conducted with Levine technique (wherein a swab is rotated over a 1×1-cm area for 5 seconds with sufficient pressure to extract fluid from within the wound tissue) was helpful in predicting wound infection (positive LR, 6.3; 95% CI, 2.5-15).²¹ This same study demonstrated that a negative swab culture result makes wound infection less likely (LR, 0.47; 95% CI, 0.31-0.73). Cultures of wound exudates or swabs performed with the Z-technique, in which a swab is applied in a zigzag manner covering the entire wound surface while the swab is rotated between the thumb and forefinger, neither predicted nor excluded wound infection, with the CI including 1.0 for positive or negative results (Table 2).²¹ A lower-quality study (level 2) demonstrated that quantitative swab cultures conducted with Levine technique did not predict the likelihood of wound infection, whereas a negative swab culture result predicted the absence of wound infection (LR, 0.35; 95% CI, 0.15-0.85).²⁶ Another study of quality level 2 demonstrated that laboratory inflammatory marker IL-6 in wound fluid did not predict the presence or absence of wound infection.²⁴

Comparisons With Nonreference Standards. Studies that used a standard (eg, semiquantitative swab culture) other than the reference standard (deep tissue biopsy culture) for diagnosing infection of chronic wounds were automatically deemed to be quality level 5. Eight studies used nonreference standards.^{19 - 20 ,28 - 33} There were mixed results regarding the utility of physical examination findings (see eTable 2 for LRs).^{28 ,32} Semiquantitative swab cultures were as predictive as quantitative swab cultures, superficial drainage fluid swab cultures were as predictive as deep wound swab cultures obtained after debridement,^{19 - 20 ,31} and elevated white blood cell count or elevated erythrocyte sedimentation rate was as predictive as clinical features.³⁰ Similarly, the absence of the laboratory marker C-reactive protein was as predictive as clinical features in ruling out infection.^{29 - 30} A handheld infrared thermometer to assess periwound skin temperature may be of use in diagnosing infection of chronic wounds but needs further study.

Only 6 of the 15 studies we identified attained a level 1 to 3 quality rating. The majority of studies compared individual clinical findings with standards of uncertain validity, and none of the findings have been evaluated in multiple studies. Most of the clinical signs to diagnose infection of chronic wounds (eg, foul odor, friable or discolored granulation tissue) are highly subjective and may have poor interrater reliability. With so few studies, we cannot be certain that the symptoms and signs lack utility or whether they might perform better when evaluated in combination with multiple findings. Although wound care experts might value changes in chronic wounds as important indicators of infection (eg, increases in size or amount of exudate), no studies evaluated the diagnostic accuracy of serial assessments.

The lack of multiple studies across a broad range of patients is important because the presence or absence of clinical findings might depend on the comorbidities of the patients and the type of chronic wound (eg, pressure ulcer vs venous stasis ulcer) and not just on whether infection is present. Thus, the context in which a wound occurs may have an important influence on the diagnostic accuracy of symptoms and signs that has not been adequately explored in studies of high quality.

The patient has a pressure ulcer and increase in the level of pain, with the absence of erythema or purulent exudate. With a prevalence of 45% infection in the studies of quality level 1 through 3, using the lower end of the LR range (11) for increasing pain increases the posttest probability to 90%. An absence of erythema and purulent exudate does not rule out infection. We are unsure from the information given what technique was used to perform the swab.

Evidence from single studies suggests that the swab culture provides no information (LR CI crosses 1.0), unless Levine technique is used. If the physician assesses the probability of infection as 90% because of the presence of pain, a negative swab result with Levine technique (LR, 0.47) still confers a probability of 81% for infection. Because infection is so likely despite the negative swab result, the need for subsequent diagnostic tests depends on the need for isolating bacteria. This would be important if it would affect the choice of antimicrobial therapy.

Despite heterogeneity, the CI for the prevalence of chronic infection in studies of quality level 1 through 3 of 32% to 58% creates a reasonable range for estimating the probability of infection in individual patients. Quiz Ref IDAn increase in the level of pain might be a useful symptom for identifying infected ulcers, but the absence of pain (in a patient who can perceive pain) should not reassure the clinician that a wound is sterile. This finding needs to be confirmed in future studies by comparing the results to deep tissue biopsy culture because the CIs from the individual findings in our study were broad. Furthermore, the underlying cause of the ulcer and comorbidities of the patient will affect the presence or absence of pain independent of infection.

Quiz Ref IDClassic signs of wound infection, evaluated in isolation from the clinical context and other findings, are not particularly helpful in diagnosing infection in a chronic wound (LR range, 0.8-2.3). The lack of utility of the classic findings for infection is perplexing, as is the finding that the positive LR range of a serous exudate (range, 1.1-1.9) suggests more usefulness than a purulent exudate (range, 0.50-0.74). Overall, the available studies suggest that the character of wound fluid exudates (categorized as serous vs purulent) is most likely not useful as a predictor of infection when the reference standard is a deep tissue biopsy culture. The apparent lack of utility of a combination of findings identified by infectious disease experts (Infectious Diseases Society of America criteria) as useful for diabetic foot infection²¹ is both surprising and disappointing but highlights the difficulty in making the diagnosis.

Quiz Ref IDWhen there is a clinical suspicion of infection (the presence of increasing pain is most useful), a quantitative swab should be performed. A quantitative swab culture performed with Levine technique has the highest-quality evidence of any noninvasive diagnostic test of infection of chronic wounds. An infection of chronic wounds is more likely (LR, 6.3) when a quantitative swab culture result with Levine technique is positive, whereas a negative swab culture result decreases the likelihood (LR, 0.47). Further research with use of an appropriate reference standard (ie, deep tissue biopsy culture) is needed to determine whether wound fluid cultures (obtained with an irrigation-aspiration technique), semiquantitative swab cultures, and laboratory tests such as C-reactive protein level, erythrocyte sedimentation rate, white blood cell count, or procalcitonin level are useful for diagnosing infection in chronic wounds. More specifically, comparative effectiveness studies of treatment guided by clinical diagnosis and Levine technique swab results vs treatment guided by deep tissue biopsy culture would be the most useful data to help physicians in the care of patients with chronic wounds.

Despite our rigorous search, the available literature we found was limited; thus, we have put forth the best evidence possible for identifying infection of a chronic wound. There were no studies identified that addressed the precision of symptoms, signs, or investigations in the diagnosis of chronic wounds. For information regarding the management of infection of chronic wounds, clinical guidelines are available through the Wound Healing Society (http://onlinelibrary.wiley.com /doi/10.1111/wrr.2006.14.issue-6 /issuetoc).^{13 ,35 - 37}

In summary, the presence of increasing pain, along with a quantitative swab culture, might help physicians estimate the probability of infection. Further evidence is required to determine which, if any, type of quantitative swab is most diagnostic.