Evaluating Diagnostic Concordance in Dermatopathology: Insights from a Sub-Saharan African Cohort

Abstract

Background:

Accurate diagnosis of skin disorders relies heavily on clinical examination, supplemented by histopathological evaluation when biopsies are performed. The degree of concordance between clinical and histopathological diagnoses is a key quality indicator and educational tool in dermatologic practice. This study aimed to evaluate clinicopathologic diagnostic concordance in skin biopsies performed at a tertiary dermatology clinic in a sub-Saharan African country.

Methodology:

A retrospective analysis of skin biopsies was conducted to determine the rate of full, partial, and non-concordance between clinical and histopathological diagnoses. Concordance was also assessed across anatomical sites, patient demographics, and diagnostic categories.

Results:

Among the biopsies reviewed, the overall clinicopathologic concordance rate was 64.2%. Full concordance was significantly more common in males than females (55.7% vs 44.3%, p = 0.02). In bivariate analysis, sex was the only variable significantly associated with diagnostic concordance (p = 0.019), with higher concordance observed among male patients. In contrast, multivariable analysis identified eczematous disorders as being associated with a markedly increased likelihood of diagnostic non-concordance compared with tumor categories (AOR = 6.38; 95% CI: 2.06-19.8; p = 0.001). Concordance was highest among tumors, papulosquamous diseases, and infections, with pigmentation disorders also demonstrating relatively high concordance. Partial concordance was most frequently observed in collagen disorders, pigmentation disorders, and inflammatory dermatoses.

Conclusion:

These findings demonstrate a moderate clinicopathologic concordance rate for skin biopsies in this setting, with considerable variability across disease categories and patient characteristics. The results underscore the continued importance of histopathological evaluation, particularly for diagnostically challenging dermatoses, and identify opportunities to improve diagnostic performance. The findings have practical implications for dermatology training by highlighting disease categories with low concordance and may serve as a quality-improvement metric to inform targeted continuing medical education programs aimed at enhancing diagnostic accuracy.

Introduction

The Gold standard for accurate clinical diagnosis of skin disorders, to date, remains thorough clinical examination by trained physicians. This process includes developing a presumptive diagnosis based on patient history and the clinical appearance of skin lesions. The clinical diagnosis, along with a summary of findings, is then submitted to the pathologist together with a skin biopsy to aid in histopathological interpretation, ultimately leading to a definitive diagnosis.

The degree of agreement between the clinical and histopathological diagnoses is an important indicator of diagnostic accuracy and can be categorized in several ways. Full concordance refers to cases where the clinical and histopathological diagnoses are identical. Partial concordance is when the histopathological diagnosis is among the listed differential diagnoses. Discordance occurs when there is no overlap between the histopathological diagnosis and any of the clinical or differential diagnoses.[1]

Reported rates of diagnostic concordance vary, with some studies citing agreement as high as 87%.[2,3] These differences in concordance rates are often attributed to specific disease categories, such as inflammatory dermatoses.[4] In clinical dermatology, only a minority of cases undergo skin biopsy; the majority are managed based on clinical evaluation alone.[5] Nevertheless, skin biopsies are critically important for certain conditions, particularly Papulosquamous inflammatory diseases and neoplastic lesions, where histopathology enhances diagnostic precision and informs management strategies.[2]

There have been very few studies examining diagnostic concordance in skin biopsies in Nigeria. Furthermore, to the best of our knowledge, this is the first study of its kind from northern Nigeria, despite the region's comparatively large population. In addition, our study employed more robust statistical modelling to further interrogate predictors of diagnostic concordance.

Studying diagnostic concordance can also serve as an educational and quality improvement tool. For instance, the concordance rate achieved by dermatology trainees could be used as a metric for evaluating clinical competence. Despite its clinical and academic significance, few studies examining diagnostic concordance in skin biopsies have been conducted in this region.

The aim of this study is to document the clinicopathologic concordance of skin biopsies in a tertiary dermatology clinic in a sub-Saharan African country.

Methods

Study Design:

We conducted a retrospective review of skin biopsy records of patients who were clinically diagnosed with skin pathologies at the Dermatology clinic of Aminu Kano Teaching Hospital (AKTH) and subsequently underwent histopathological evaluation in the hospital's department of pathology. All biopsies were exclusively obtained using the punch biopsy technique and processed within AKTH.

A total of 404 skin biopsy records were initially identified. Records with both clinical diagnosis by a dermatologist and a histopathological diagnosis were classified as eligible for inclusion in the study. All records with only clinical or histopathological diagnosis, or non-specific diagnosis, were excluded from the study. Repeat biopsies from the same patient were included only once to avoid duplication.

Study Site:

The study was conducted at Aminu Kano Teaching Hospital (AKTH), Kano, Nigeria. AKTH is a 600-bed teaching hospital serving the entire population of Kano State, the most populous state in Nigeria. The hospital also serves the population of neighbouring states, including Katsina, Jigawa, and Bauchi. AKTH operates a twice-weekly dermatology clinic, which about 20,000 outpatients visit yearly. The dermatology clinic runs a robust residency program training dermatology residents.

Study Variables:

Outcome Variables

The primary outcome of interest was concordance between clinical and histopathological diagnosis. Concordance was classified as “Full concordance” for complete alignment between clinical and histopathological diagnosis, “Partial concordance” when the final histopathological diagnosis included at least one of the clinical differential diagnoses, and “No concordance” when there is no agreement between the clinical and histopathological diagnosis. “Overall concordance” was defined as the combination of full and partial concordance.

Explanatory Variables

Explanatory variables were selected based on the existing literature and included: age (0-85 years), age group (0-44, 45-64, >65 years), sex (male or female), and site of Tissue Biopsy (Head/neck, Trunk, Upper limb, Lower limb, and genital region).

Skin diseases were categorized using a standardized classification system adapted from published literature.[1] Individual diagnostic categories were coded numerically (1-24) during data analysis, with each code corresponding to a specific disease category (e.g., tumors, papulosquamous diseases, infections, eczematous disorders). Skin diseases were also broadly grouped into two (Inflammatory dermatoses: Neoplastic diseases 1,2).

Statistical Analysis:

Data analysis was executed using R statistical software, version 4.4.3 (R Statistical Corp., Vienna, Austria). All statistical tests were two-sided, with a type 1 error rate set at 5%. Descriptive statistics were used to summarize patient demographics and biopsy characteristics. Continuous variables were presented as mean with standard deviation (SD) or median with interquartile range (IQR), depending on the distribution, which was assessed using the Shapiro-Wilk test. Categorical variables were summarized using frequencies and percentages.

Bivariate analysis was performed to evaluate the relationship between clinicopathologic concordance and each independent variable using the Chi-square test or Fisher’s exact test for categorical variables and the ANOVA or Kruskal-Wallis rank sum test for non-normally distributed data.

To identify independent predictors of diagnostic concordance between clinical and histopathological diagnoses, a Multinomial logistic regression model was employed. Adjusted odds ratios (aORs) with 95% confidence intervals (CIs) were reported. Statistical significance level was set at p < 0.05.

Ethical Clearance:

Ethical approval for this study was obtained from the Health Research Ethics Committee of AKTH (AKTH/MAC/SUB/12A/P-3/VI/4047).

Results

A total of 404 skin biopsy records were identified, out of which 21 were excluded for failure to meet the inclusion criteria, leaving 383 records for analysis. Table 1 summarizes the demographic and diagnostic profile of included participants.

Table 1:

Demographic and Diagnostic Profile of Patients with Skin Biopsies at Aminu Kano Teaching Hospital

Characteristic	N = 383
Age (Years), Median (Q1-Q3)	35.0 (24.0-48.0)
Age Group, n (%)
0-44	270.0 (70.5%)
45-65	88.0 (23.0%)
>65	25.0 (6.5%)
Sex, n (%)
Male	187.0 (49.0%)
Female	196.0 (51.0%)
Site of Skin Tissue Biopsy, n (%)
Head	71.0 (18.5%)
Trunk	102.0 (26.6%)
Upper limb	91.0 (23.8%)
Lower limb	116.0 (30.3%)
Groin/Genital area	3.0 (0.8%)
Category of Skin Disease, n (%)
Eczema	18.0 (4.7%)
Papulosquamous diseases	88.0 (23.0%)
Infections	21.0 (5.5%)
Tumors	141.0 (36.8%)
Autoimmune disorders	15.0 (3.9%)
Drug eruptions	2.0 (0.5%)
Disorders of collagen	11.0 (2.9%)
Cysts	11.0 (2.9%)
Pigmentation Disorders	19.0 (5.0%)
Disorders of Epidermal Maturation	8.0 (2.1%)
Cutaneous deposits	9.0 (2.3%)
Vasculitis	2.0 (0.5%)
Follicular disorders	3.0 (0.8%)
Panniculitis	5.0 (1.3%)
Miscellaneous	25.0 (6.5%)
Keloid	3.0 (0.8%)
Vesicobullous disorders	2.0 (0.5%)

Our result shows that clinicopathologic concordance was moderate overall (Figure 1). Across the 383 biopsies, full concordance occurred in 46.2% (177/383), partial concordance in 18.0% (69/383), and no concordance in 35.8% (137/383), yielding an overall concordance rate of 64.2% (246/383).

Fig 1.

Distribution of Concordance Categories

In bivariate comparisons (Table 2), sex was associated with concordance category (p = 0.019). Males comprised 55.7% of the full concordance group compared with 46.7% of the no concordance group and 36.2% of the partial concordance group. Age and age group were comparable across concordance categories (age p = 0.5; age group p > 0.9). When categories were collapsed into broad classifications, inflammatory dermatoses were 62.3% of the partial concordance group, 53.1% of the full concordance group, and 54.7% of the no concordance group, without evidence of a difference (p = 0.4; Table 2).

Table 2.

Comparison of Patient Demographic and Clinical Characteristics Across Levels of Clinicopathologic Concordance.

Characteristic	Full Concordance (N = 177)	No Concordance (N = 137)	Partial Concordance (N = 69)	p-value
Age, Median (Q1-Q3)	35.0 (23.0-48.0)	35.0 (25.0-48.0)	33.0 (22.0-48.0)	0.51
Age Group, n (%)				>0.92
0-44	127.0 (71.8%)	96.0 (70.1%)	47.0 (68.1%)
45-65	39.0 (22.0%)	31.0 (22.6%)	18.0 (26.1%)
>65	11.0 (6.2%)	10.0 (7.3%)	4.0 (5.8%)
Sex, n (%)				0.0193
Male	98.0 (55.7%)	64.0 (46.7%)	25.0 (36.2%)
Female	79.0 (44.3%)	73.0 (53.3%)	44.0 (63.8%)
Site of Tissue Biopsy, n (%)				0.122
Head/neck	32.0 (18.1%)	28.0 (20.4%)	11.0 (15.9%)
Trunk	40.0 (22.6%)	38.0 (27.7%)	24.0 (34.8%)
Upper limb	39.0 (22.0%)	39.0 (28.5%)	13.0 (18.8%)
Lower limb	65.0 (36.7%)	30.0 (21.9%)	21.0 (30.4%)
Genital region	1.0 (0.6%)	2.0 (1.5%)	0.0 (0.0%)
Category of skin diseases, n (%)				0.32
Eczematous Disorders	4.0 (2.3%)	13.0 (9.5%)	1.0 (1.4%)
Papulosquamous diseases	44.0 (24.9%)	28.0 (20.4%)	16.0 (23.2%)
Infections	9.0 (5.1%)	7.0 (5.1%)	5.0 (7.2%)
Tumors	72.0 (40.7%)	47.0 (34.3%)	22.0 (31.9%)
Autoimmune Disorders	7.0 (4.0%)	4.0 (2.9%)	4.0 (5.8%)
Drug Eruptions	0.0 (0.0%)	0.0 (0.0%)	2.0 (2.9%)
Disorders of Collagen	3.0 (1.7%)	5.0 (3.6%)	3.0 (4.3%)
Cysts	4.0 (2.3%)	4.0 (2.9%)	3.0 (4.3%)
Pigmentation Diseases	8.0 (4.5%)	6.0 (4.4%)	5.0 (7.2%)
Disorders of Epidermal Maturation	5.0 (2.8%)	2.0 (1.5%)	1.0 (1.4%)
Cutaneous Deposits	4.0 (2.3%)	3.0 (2.2%)	2.0 (2.9%)
Vasculitis	1.0 (0.6%)	1.0 (0.7%)	0.0 (0.0%)
Follicular Diseases	2.0 (1.1%)	1.0 (0.7%)	0.0 (0.0%)
Panniculitis	3.0 (1.7%)	1.0 (0.7%)	1.0 (1.4%)
Miscellaneous	7.0 (4.0%)	14.0 (10.2%)	4.0 (5.8%)
Keloids	3.0 (1.7%)	0.0 (0.0%)	0.0 (0.0%)
Vesicobullous diseases	1.0 (0.6%)	1.0 (0.7%)	0.0 (0.0%)
Category of Pathology, n (%)				0.43
Inflammatory dermatoses	94.0 (53.1%)	75.0 (54.7%)	43.0 (62.3%)
Neoplastic diseases	83.0 (46.9%)	62.0 (45.3%)	26.0 (37.7%)

¹Kruskal-Wallis rank sum test; ²Fisher's Exact Test; ³Pearson's Chi-squared test

Biopsy site was not found to be associated with diagnosis concordance category (p = 0.12), although lower-limb biopsies represented a larger proportion of the full concordance group (36.7%) than the no concordance (21.9%) and partial concordance (30.4%) groups (Table 2). The distribution of detailed skin disease categories across concordance groups was also not found to be statistically different (p = 0.3).

In multinomial regression comparing no concordance with full concordance (Table 3), age, age group, sex, biopsy site, and broad disease classification were not found to predict the likelihood of no concordance. In the detailed disease category model, eczematous disorders were associated with increased odds of no concordance relative to tumors (AOR = 6.38; 95% CI: 2.06–19.8; p = 0.001). Miscellaneous conditions were also associated with higher odds of no concordance compared with tumors (AOR = 3.01; 95% CI: 1.11–8.15; p = 0.031). Several categories (including keloid and drug eruptions) had no observed cases of no concordance, with corresponding zero estimated odds in the model (Table 3).

Table 3.

Multivariable Analysis of Predictors of No Clinicopathologic Concordance.

Characteristic	AOR	95% CI	p-value
"No Concordance"
Age	1.02	0.99, 1.05	0.2
Age group
0-44	—	—
45-65	0.67	0.28, 1.62	0.4
>65	0.46	0.10, 2.21	0.3
Sex
Male	—	—
Female	1.49	0.92, 2.42	0.10
Site of Tissue Biopsy
Head/neck	—	—
Trunk	0.97	0.46, 2.05	>0.9
Upper limb	1.24	0.57, 2.70	0.6
Lower limb	0.47	0.22, 0.99	0.047
Genital region	1.66	0.12, 22.6	0.7
Broad classification of skin diseases
Inflammatory dermatoses	—	—
Neoplastic diseases	1.13	0.66, 1.94	0.7
Category of skin diseases
Tumors	—	—
Eczematous Diseases	6.38	2.06, 19.8	0.001
Papulosquamous Disorders	1.27	0.68, 2.37	0.5
Infections	1.61	0.57, 4.52	0.4
Autoimmune Diseases	0.83	0.24, 2.83	0.8
Drug Eruptions	0.01	0.01, 0.03	<0.001
Disorders of Collagen	3.21	0.78, 13.3	0.11
Cysts	1.45	0.36, 5.88	0.6
Pigmentation Diseases	1.19	0.40, 3.54	0.8
Disorders of Epidermal Maturation	0.97	0.18, 5.14	>0.9
Cutaneous Deposits	1.71	0.39, 7.49	0.5
Vasculitis	2.05	0.12, 35.3	0.6
Follicular Diseases	0.90	0.09, 8.85	>0.9
Panniculitis	0.55	0.06, 4.85	0.6
Miscellaneous	3.01	1.11, 8.15	0.031
Keloid	0.00	0.00, 0.00	<0.001
Vesicobullous diseases	1.22	0.08, 19.3	0.9

In the second part of the multinomial regression comparing partial concordance with full concordance (Table 4), female sex was associated with higher odds of partial concordance relative to male sex (AOR = 2.30; 95% CI: 1.25–4.23; p = 0.007), while age group, biopsy site, and broad disease classification were not independent predictors. In the detailed disease category model, papulosquamous diseases (AOR = 2.15; 95% CI: 1.06–4.36; p = 0.035), infections (AOR = 3.65; 95% CI: 1.18–11.3; p = 0.025), disorders of collagen (AOR = 5.17; 95% CI: 1.08–24.7; p = 0.039), and pigmentation diseases (AOR = 3.42; 95% CI: 1.09–10.8; p = 0.035) were associated with increased odds of partial concordance relative to tumors (Table 4). Categories including vasculitis and follicular diseases had no observed cases of partial concordance, with corresponding zero estimated odds in the model (Table 4).

Table 4.

Multivariable Analysis of Predictors of Partial Clinicopathologic Concordance.

Characteristic	AOR	95% CI	p-value
"Partial Concordance"
Age	0.98	0.95, 1.01	0.3
Age group
0-44	—	—
45-65	1.94	0.63, 5.95	0.2
>65	2.69	0.37, 19.6	0.3
Sex
Male	—	—
Female	2.30	1.25, 4.23	0.007
Site of Tissue Biopsy
Head/neck	—	—
Trunk	1.47	0.58, 3.75	0.4
Upper limb	0.83	0.29, 2.35	0.7
Lower limb	0.76	0.30, 1.94	0.6
Genital region	0.00	0.00, 0.00	<0.001
Broad classification of skin diseases
Inflammatory dermatoses	—	—
Neoplastic diseases	1.61	0.86, 3.02	0.14
Category of skin diseases
Tumors	—	—
Eczematous Diseases	1.40	0.18, 11.0	0.8
Papulosebaceous Diseases	2.15	1.06, 4.36	0.035
Infections	3.65	1.18, 11.3	0.025
Autoimmune diseases	2.40	0.70, 8.28	0.2
Drug Eruptions	0.00	0.00, 0.00	<0.001
Disorders of Collagen	5.17	1.08, 24.7	0.039
Cysts	3.31	0.72, 15.1	0.12
Pigmentation Diseases	3.42	1.09, 10.8	0.035
Disorders of Epidermal Maturation	1.92	0.23, 15.8	0.5
Cutaneous Deposits	3.58	0.68, 18.8	0.13
Vasculitis	0.00	0.00, 0.00	<0.001
Follicular Diseases	0.00	0.00, 0.00	<0.001
Panniculitis	1.33	0.15, 11.5	0.8
Miscellaneous	1.72	0.45, 6.56	0.4
Keloid	0.00	0.00, 0.00	<0.001
Vesiculobullous diseases	0.00	0.00, 0.00	<0.001

Abbreviations: CI = Confidence Interval, AOR = Adjusted Odds Ratio

Discussion

The overall concordance between clinical and final histopathological diagnoses in our study was 64.2%, which aligns with findings from other regions. Reported concordance rates in the literature vary widely, ranging from 59.8% to 92%. For example, in India, Pradeep et al.[6] found a concordance rate of 59.8% in an analysis of 3,006 skin biopsies. In Saudi Arabia, Al Saif FM[1]reported a concordance of 75.9% across 4,298 biopsies, with full concordance in 28.3% and partial concordance in 47.6% of cases.

Similarly, Aslan et al.[7] in Turkey, a clinicopathologic concordance of 76.8% was reported in 3,949 biopsies, with a non-concordance rate of 23.2%. In Sarawak General Hospital, Malaysia, Bin Yap et al.[8]found a concordance rate of 86.8% in 347 biopsies.

Using data from two dermatology clinics in Nigeria, Ogunbor OE et al. reported an overall diagnostic concordance of 60.1%, which is comparable to our findings. [9]

These variations may reflect differences in disease prevalence, diagnostic protocols, training, and institutional resources.

In our study, the highest overall and full concordance rates, 67.1% and 57.4%, respectively, were observed in biopsies from the lower limb and trunk, although this was not statistically significant (p = 0.12). Notably, full concordance was significantly higher in males (55.7%) compared to females (44.3%) (p = 0.02), in contrast to other studies that found no significant gender difference.

The three most common diagnostic categories in our cohort were tumours, Papulosquamous diseases, and infections, accounting for just over 50% of cases. This distribution is broadly similar to findings from other studies, though some differences were noted. For instance, eczematous diseases were relatively less frequent in our study, constituting 4.7% of cases (n = 18), compared to 9.2% (n = 37) reported by Bin Yap et al.[8] In contrast, infectious diseases accounted for 5.5% (n = 21) in our cohort, whereas a study in India reported a much higher prevalence of 66.6% (n = 62). [10]

There was notable variability in concordance across disease categories. Tumors and Papulosquamous diseases showed the highest overall and full concordance, consistent with findings by Al Saif FM [1] in Saudi Arabia and Korfitis C[2] in Greece. Interestingly, pigmentation disorders also exhibited high concordance in our study, similar to observations by Talwar A et al.[11] in India. However, this trend was not observed in Saudi Arabia, potentially due to differing sample sizes and the spectrum of pigmentation disorders included.

We hypothesize that the higher diagnostic concordance observed for tumors and papulosquamous diseases may be attributable to the greater diagnostic familiarity and expertise of most pathologists with these conditions. In contrast, inflammatory dermatoses demonstrated lower diagnostic concordance, likely reflecting the greater diagnostic challenges they pose for most pathologists.

Conversely, eczematous and Vesicobullous disorders demonstrated lower concordance in our cohort. Al Saif FM[1], however, reported high concordance for Vesicobullous diseases, possibly reflecting better access to specialized diagnostic techniques and more frequent clinical exposure to these conditions in that setting.

An analysis of partial concordance in our data revealed that collagen disorders, pigmentation disorders, infectious diseases, and Papulosquamous dermatoses were most frequently associated with this category. Inflammatory dermatoses, in particular, often pose diagnostic challenges due to their overlapping clinical features. This reliance on clinical expertise for accurate diagnosis may contribute to the higher rates of partial concordance.

Furthermore, patients in our setting often seek initial care from non-specialists, leading to treatments that may alter the clinical presentation of skin diseases by the time they reach a dermatologist. We hypothesize that this delay and modification in disease presentation may contribute to the observed partial concordance rates.

To improve the quality of skin disease diagnosis, there is a need to implement quality-improvement strategies in clinical dermatology practice. These include strengthening communication between dermatologists and pathologists through regular clinicopathologic meetings. In addition, enhancing pathologists' knowledge of clinical dermatology, particularly in the area of inflammatory skin diseases, would be beneficial.

Limitations

This is a retrospective study, which inherently limits the ability to control confounding factors and may be prone to information bias, particularly due to incomplete or missing clinical data. Additionally, the study was conducted at a single tertiary hospital, which may affect the generalizability of the findings to other healthcare settings, especially primary or secondary health facilities with differing diagnostic capabilities.

Conclusion

In conclusion, an overall clinicopathologic concordance of 64.2% was documented in our study. The concordance reported varies amongst the different skin categories; with tumors, Papulosquamous dermatoses, and Infections having the most overall and full concordance, and there are some differences in the pattern observed in our study and some other regions.

The correlation study between the different concordance categories has highlighted some unique findings that could be used to further explain the reasons for reported partial concordance in our study.