Necrotizing Fasciitis Originating in the Hand: A Systematic Review and Meta-Analysis

Abstract

Necrotizing Fasciitis (NF) is a rapidly progressive and potentially lethal infection. This systematic review focused on the treatment and prognosis of the NF of the hand. The literature search was performed in PubMed database, and cohort studies and case reports were deemed eligible. Proportions were pooled to estimate overall mortality and amputation rates, and a logistic regression analysis was performed to evaluate predictors of amputation and mortality. The search algorithm resulted in 51 eligible articles including 48 case reports (54 patients) and 3 larger retrospective cohort studies (107 patients). Most patients had a monomicrobial infection, a surgical debridement on the day of admission, and hospital stay of more than 10 days. The estimated overall amputation rate was 28%, whereas overall mortality was equal to 8%. In the pooled set of case reports, mortality was significantly associated with age older than 54 years and marginally with diabetes mellitus. Necrosis expanding more proximally to the forearm correlated with both amputation and mortality. It is reiterated that early diagnosis of NF is of paramount importance and that early and decisive surgical intervention should have low threshold, especially when potential risk factors are identified.

Introduction

Necrotizing fasciitis (NF) is a rapidly progressive infection mostly affecting the fascia and subcutaneous layers. ¹ Since its first description in the beginning of the previous century, ² the incidence of this unusual infection has been increasing and is unfortunately related to high mortality rates. ³ There is diversity in the clinical manifestation and the course of the disease, as well as in the etiological microorganisms involved. In any case, the prompt surgical intervention is deemed critical for the final outcome. ⁴

Generally, the recorded involvement of the upper extremity varies from 7% to 27% ⁵ with a rather sporadic documentation of the relevant features in the international literature. A specialized approach regarding the NF of the hand region is an even more challenging task, and this study aims to focus on the identification of the specific risk factors, the early diagnostic criteria, the possible reconstructive techniques, and the overall prognosis in this surgical emergency.

Methods

Search Strategy

This systematic review and meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. A preliminary search did not reveal a systematic review on NF of the hand. Two authors (G.C. and M.J.) proceeded independently to the selection of studies; in case of disagreement, team consensus followed. The literature search was performed in “PubMed” database with the end-of-search date set at July 20, 2020.

Search was carried out using the terms (“necrotising fasciitis” OR “necrotizing fasciitis”) AND hand to identify relevant articles. Studies describing the treatment and prognosis of clinically diagnosed NF affecting the hand were deemed eligible; regarding study design, cohort studies (prospective or retrospective), case reposts, and case series were included. Cases in which the disease was further expanded to other anatomical regions were also included in our systematic review; however, non-English language and absence of separate reporting for NF of the hand were considered as exclusion criteria.

Data Extraction

The extraction of data comprised first author’s name, country, year, study design, number of patients, age, sex, and comorbidities. Regarding NF of the hand, the side, other affected anatomical, preceding causative event, and etiological microorganisms were recorded. In addition, data on surgical treatment, hospitalization period, number of surgical debridements, time of first debridement, additional use of hyperbaric oxygen therapy, application of graft or flap reconstruction, amputation, and death were recorded. Two authors (G.C. and M.J.) independently extracted data; in case of disagreement, team consensus followed.

Risk of Bias

This systematic review presents the data reported in cohort studies and case reports. Risk of bias was assessed using the Newcastle-Ottawa scale. ⁶ Regarding mortality, long enough follow-up was considered up to discharge time. A nonexposed population was not defined and no comparator method was identified, concluding in a maximum score of 6 instead of 9. ⁷ Two authors (G.C. and M.J.) independently assessed risk of bias; in case of disagreement, team consensus followed.

Statistical Analysis

To estimate overall mortality and amputation rates, the results of cohort studies were synthesized with a pooled data set of eligible case reports/case series. The overall proportions of mortality and amputation were estimated using the Freeman-Tukey arcsine transformation; random-effects models (DerSimonian-Laird) were implemented. Between-study heterogeneity was evaluated using Q and I ² statistics. Given that the number of eligible studies was smaller than 10, no meta-regression analysis and statistical evaluation of publication bias were performed, in accordance with the Cochrane Handbook for Systematic Reviews of Interventions. ⁸

In the pooled data set of case reports and case series, logistic regression analysis was performed to evaluate predictors of amputation and mortality. Specifically, the associations with the following predictors were studied: age, sex, side, necrosis to other sites, preceding causative event, comorbidities, diabetes, etiological microorganism, Streptococcus, additional hyperbaric oxygen therapy, and number and time of debridement. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. Statistical analysis was performed using STATA/SE version 13 statistical software (Stata Corp., College Station, Texas).

Results

Selection of Studies

The search algorithm resulted in 191 citations. In total, 51 articles were deemed eligible (Refs 9-11, Supplemental Table 1), resulting in a total sample of 161 patients with NF of the hand. The eligible studies included 48 case reports/case series with overall 54 patients and 3 retrospective cohort studies,^9-11 including a total of 107 patients (Supplemental Tables 1 and 2, respectively).

Synthesis of Results

Table 1 presents a summary for all cohorts taking into account the recorded data for each characteristic. The age ranged from 1 to 86 years with most patients being men and having several underlying comorbidities, commonly including diabetes mellitus. In most cases, there was a triggering event from a recent injury or an infectious systematic disease. The larger part of infections had a monomicrobial background with 3 cases referring to negative culture results. Several different etiological microorganisms were identified (as analytically depicted in Supplemental Tables 1 and 2); group A Streptococcus (GAS) was the most commonly detected bacterium. The vast majority of patients had a surgical debridement on the day of admission. Most patients had to stay in hospital for more than 10 days, and regarding the hand defect reconstruction, a graft was commonly used in 21 cases and a flap applied in 9 patients with 4 pedicled and 5 free flaps. Additional hyperbaric oxygen therapy was rarely used in only 5 cases.

Table 1.

Description of Analyzed Cohorts.

Features	Cohorts
	Nawijn et al 11	Raveendran et al 10	Yeung et al 9	Pooled case reports a (1986-2020)
No. of patients with hand NF	64	23	20	54
Age (median)	N/R	51.5	N/R	54
Male sex	N/R	17/23 (73.9%)	N/R	31/54 (57.4%)
Right side	N/R	N/R	13/20 (65.0%)	27/54 (50.0%)
Expanding necrosis to forearm	52/64 (81.6%)	N/R	9/20 (45.0%)	26/54 (48.2%)
Expanding necrosis more proximally to forearm	10/64 (15.6%)	N/R	3/20 (15.0%)	8/54 (14.8%)
Preceding causative event	N/R	9/23 (39.1%)	16/20 (80.0%)	39/50 (78.0%)
Polymicrobial background	N/R	21/23 (91.3%)	10/20 (50.0%)	7/52 (13.4%)
Group A Streptococcus detection	N/R	0/23 (0.0%)	6/20 (30.0%)	23/52 (44.2%)
Comorbidities	N/R	23/23 (100.0%)	15/20 (75.0%)	33/48 (68.8%)
Diabetes mellitus	N/R	23/23 (100.0%)	3/20 (15.0%)	12/48 (25.0%)
Three or more debridements	N/R	N/R	N/R	20/48 (41.7%)
First debridement on admission	N/R	N/R	N/R	40/47 (85.1%)
Additional hyperbaric oxygen	0/64 (0%)	0/23 (0.0%)	0/20 (0.0%)	5/54 (9.2%)
10 d or more of hospitalization	N/R	23/23 (100.0%)	N/R	34/45 (75.6%)
Graft use	N/R	N/R	N/R	21/47 (44.7%)
Flap reconstruction	N/R	N/R	N/R	9/51 (17.65%)
Amputation	14/64 (21.9%)	6/23 (26.1%)	6/20 (30.0%)	19/52 (36.5%)
Mortality	5/64 (7.8%)	0/23 (0.0%)	4/20 (20.0%)	7/53 (13.2%)

Note. NF = necrotizing fasciitis; N/R = not recorded data for the patients with hand NF.

^aThe total number of cases pertains to those providing relevant data.

Meta-Analysis

The estimated overall amputation rate was 28% (95% CI, 21%-35%; Figure 1). The amputation rate was equal to 37% (95% CI, 25%-50%) in the pooled case reports, 30% (95% CI, 15%-52%) in the retrospective cohort study by Yeung et al, ⁹ 26% (95% CI, 13%-46%) in that of Raveendran et al, ¹⁰ and 22% (95% CI, 14%-33%) in the larger cohort study by Nawijn et al. ¹¹ More than half of the amputations were performed at the finger level.^9-11

Figure 1.

Meta-analysis for the estimated amputation rate.

Note. CI = confidence interval; ES = effect size.

The estimated overall mortality from the NF originated in the hand rate was 8% (95% CI, 2%-18%; Figure 2). The mortality rate was equal to 13% (95% CI, 7%-25%) in the pooled set of case reports, 20% (95% CI, 8%-42%) in the study of Yeung et al, ⁹ 0% (95% CI, 0%-14%) in that of Raveendran et al, ¹⁰ and 8% (95% CI, 3%-17%) in the review of Nawijn et al ¹¹ which provided the largest portion of the study population.

Figure 2.

Meta-analysis for the estimated mortality rate.

Note. CI = confidence interval; ES = effect size.

Predictors of Mortality and Amputation

Logistic regression analysis examining predictors of mortality from NF originated in the hand demonstrated some statistically significant associations (Table 2). In the pooled data set of case reports, 7 of 27 patients aged 54 years or older (25.9%) died, while no death was noted in the 25 younger patients (P = .010, Fisher exact test). There was also an association between mortality and expanding necrosis; 3 of 25 patients (12%) with necrosis expanding to the forearm and, more importantly, 4 of 8 patients with further expansion (50%) died, whereas no deaths were noted in NF confined to the hand (P = .002, Fisher exact test). Finally, a trend toward increased mortality was identified for patients with diabetes mellitus (OR = 6.38, 95% CI, 0.91-44.71, P = .062).

Table 2.

Results of Univariate Logistic Regression Analysis Examining Risk Factors for Mortality in the Pooled Analysis of Case Reports.

Risk factors	n	Compared categories	Univariate OR (95% CI)	P value
Age	52	≥54 vs <54	Not estimable, due to lack of deaths in younger patients	.010 a
Sex	53	Male vs female	0.53 (0.11-2.63)	.436
Side	53	Right vs left	7.80 (0.87-70.10)	.067
Necrosis to other sites	53	Forearm vs none	Not estimable, due to lack of deaths in patients without expanded necrosis	.002 a
		Distal vs none
Preceding causative event	49	Yes vs no	0.86 (0.08-9.17)	.899
Comorbidities	47	Yes vs no	2.00 (0.20-19.62)	.552
Diabetes	47	Yes vs no	6.38 (0.91-44.71)	.062
Etiological microorganism	50	Polymicrobial vs monomicrobial	1.03 (0.10-10.11)	.981
Participation of group A Streptococcus	50	Yes vs no	1.53 (0.31-7.69)	.603
No. of debridements	48	≥3 vs ≤2	0.51 (0.09-2.95)	.453
Time of the first debridement	47	Late vs on admission	Not estimable, due to lack of deaths in patients with late first debridement	.571
Additional hyperbaric oxygen therapy	53	Yes vs no	Not estimable, due to lack of deaths in patients with patients who received HOT	.575
Amputation	52	Yes vs no	1.36 (0.27-6.85)	.710

Note. n = number of cases; OR = odds ratio; CI = confidence interval; HOT = hyperbaric oxygen therapy.

^aP value derived from the Fisher exact test due to 0 events in an exposure category.

The logistic regression analysis examining predictors of amputation did not reveal any statistically significant correlations (Supplemental Table 3). However, sizeable ORs were related to the expansion and the microbial background of the infection. A sizable effect estimate for amputation was identified in the cases where the infection was expanded more proximally to the forearm (OR = 2.80, 95% CI, 0.50-15.66, P = .241) and when multiple causative pathogens were isolated (OR = 2.40, 95% CI, 0.47-12.18, P = .291). In addition, the same pattern was noted for the late debridement of the NF (OR = 2.22, 95% CI, 0.44-11.32, P = .336), as well as a trend toward an inverse association between amputation and performance of 3 or more debridements (OR 0.33, 95% CI, 0.10-1.17, P = .086).

Risk of Bias

In the Newcastle-Ottawa scale, the quality of eligible studies was mainly compromised by the lack of comparability of factors, but most studies met our criteria on the assessment of outcome and appropriate follow-up which was set at the discharge time.

Discussion

Our systematic review identified 48 case reports and 3 retrospective cohort studies evaluating NF of the hand. Our meta-analysis estimated an amputation rate of 28%, more than half of which were at the level of digit. Cases where the infection was expanded more proximally to the forearm and those that underwent late debridement were more predispose to amputation. The estimated overall mortality rate was 8%; significantly lower than 21.7% reported overall mortality not specific to anatomical location. ¹²

Risk Factors Related to Hand NF

Generally, to reduce risk of mortality and amputation, it is essential to correctly diagnose this condition early and adopt an early and aggressive management strategy. Two of the 3 retrospective cohort studies^9,11 have evaluated risk factors for amputation and mortality. Understanding these gives the opportunity to predict and prevent decline with early diagnosis and intervention. Nawjin et al ¹¹ carried out univariate analysis, identifying that independent risk factors for mortality included history of intravenous drug use, higher age, higher American Society of Anesthesiologists (ASA) Classification Score, increased respiratory rate, capillary blood glucose, serum lactate, or base deficit. Multivariable analysis identified that specifically a base deficit of 3 milliequivalents per liter (mEq/L) or greater and ASA of 3 or 4 were independent risk factors for mortality; especially, ASA of greater than 3 inferred 9 times increase in mortality risk. In addition, Yeung et al ⁹ identified deranged renal and liver function, thrombocytopenia, initial proximal limb involvement, and hypotension on admission as predictive factors for mortality.

The Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) score has been described as an aid to diagnosis of NF, but there are no studies specifically evaluating its predictive accuracy. However, Nawjin et al ¹¹ reported relevant scores in their cohort, and the mean value in their mortality group was 4, which falls below the cutoff for suspicion of NF diagnosis; the mean score in their amputation group was 7. The Laboratory Risk Indicator for Necrotizing Fasciitis score has been validated in 2004, ¹³ but subsequent studies have highlighted lack of diagnostic accuracy when differentiating between NF and severe cellulitis. ¹⁴ In this context, there are risk factors identified in the studies presented in this review that do not feature in the LRINEC score, such as base excess, serum lactate, and platelet count; inclusion of these factors may make for a more robust scoring system.

In addition, LRINEC does not take account of whether the patient is diabetic even though diabetes mellitus was found to be a patient factor in 12 of our 48 identified case reports. Logistic regression analysis showed a trend toward increased mortality for patients with diabetes. Nevertheless, Nawjin et al ¹¹ found no correlation between diabetes diagnosis and amputation risk, but diabetes is an increased risk for severity, particularly with poor glycemic control. ¹⁰

Microbiological Background Pathology

Most cases of NF are caused by bacterial infection with rare reports of infection of fungal origin in immunocompromised patients. ¹⁵ Bacterial infections can be monomicrobial or polymicrobial with synergistic effect. In the upper limb cases identified in our review, a sizable effect estimate for amputation was identified when multiple causative pathogens were isolated. However, most of the cases were monomicrobial, and the most commonly identified organism was GAS. Previous literature not specific to anatomical location reports that monomicrobial forms are less commonly encountered than polymicrobial, accounting for 25% to 45% of necrotizing soft tissue infections (NSTIs), with Staphylococcus aureus, Streptococcus species, Escherichia coli, and Klebsiella being the most commonly identified organisms in polymicrobial infections and Streptococcus pyogenes and S aureus in monomicrobial infection. ¹⁶ Worth highlighting is also the fact that we identified 14 cases where Vibrio vulnificus, a rarely involved microorganism related to seafood handling, was the responsible microorganism.

Post-NF Hand Reconstruction

Reconstruction after upper limb NSTIs is an important consideration. In this review, 21 identified cases of NF of the hand underwent reconstruction with graft only; 9 required flap coverage, 4 of which were pedicled flaps and 5 were free flaps. Choice of pedicled flaps included posterior interosseous artery flap after third ray amputation ¹⁷ and pedicled reverse radial forearm flap ¹⁸ after interphalangeal joint (IPJ) level thumb amputation. In cases where amputation was not required, pedicle flaps used included Quaba flap ¹⁹ and groin flap. ²⁰ Free flap options included free lateral arm flap to amputated thumb (IPJ level), and free flaps to areas of extensive debridement without amputation, including free rectus flap to forearm, ²¹ free anterolateral thigh flap to dorsum of hand, ²² and free groin flap to volar forearm and hand.^23,24

Thumb reconstruction is a particular challenge due to the essential role of the thumb in overall hand function. Both reported thumb reconstruction cases underwent further revision surgery; pedicled reverse radial forearm flap following Metacarpophalageal joint (MCPJ) level thumb amputation was revised with free wrap-around great toe transfer, ¹⁸ and free lateral arm flap following thumb IPJ level terminalization failed, requiring subsequent Split-thickness skin graft (SSG) coverage. ²⁵

Limitations

A limitation of this review pertains to the fact that eligible cases have been treated in different institutions worldwide and by different surgeons during a long interval from 1986 until 2020. In addition, reporting varied across individual studies regarding several features. Concerning cohort studies, no randomized controlled trials were identified; only retrospective noncomparative studies have been published. Regarding the pooled data set of case reports and case series, selection bias should be declared, as those cases may have had special features, prompting treating physicians to proceed to publication.

Conclusion

Necrotizing fasciitis is a rapidly progressive and potential lethal infection. Our review is the first attempt to focus the investigation of this pathology on the hand region. The meta-analysis resulted in an estimated overall amputation rate of 28 and mortality rate of 8%. Furthermore, the logistic regression analysis revealed a statistically significant correlation between mortality and older age and proximally expanding necrosis. In addition, a trend toward increased mortality was identified for patients with diabetes mellitus and toward amputation for infections expanding more proximally to the forearm, for multiple causative pathogens, and for late debridement. It is apparent that early diagnosis is of paramount importance and that decisive surgical intervention should have low threshold, especially when potential risk factors are identified. A series of debridement should aim initially to control the infection and save patient’s life and the subsequent reconstruction to preserve most of hand’s functionality.