Selection and Characteristics
The results of our literature search are summarized in the flow diagram in Fig. 1. The initial search, run on January 2, 2021, yielded 347 articles, among which 34 duplicates were removed. Subsequently, screening titles and abstracts for key words such as HS, anemia, and hemoglobin narrowed the selection to 23 articles. No clinical trials or cohort studies were found. At the time, 7 studies were found to meet inclusion criteria.30–35 Screening reference sections of eligible studies did not yield additional studies. Among the 7 included studies, there were 2 case-control30, 32 and 5 cross-sectional studies.31,33−35 One study was conducted in Poland,32 2 in Denmark,35 and the remainder in the U.S.30,31,33,34 Table 1 describes characteristics of all included studies.30–35
Risk of Bias
With regards to case-control studies,30,32 the mean (standard deviation [STDEV]) NOS was 5.5 (3.5). For Kimball et al., unclear risk of bias was rated for “adequacy of case definition” due to the use of ICD-9-CM codes as a form of record linkage without independent validation37. “Non-response rate” was also rated as unclear risk of bias although it was mentioned that cases of low severity may have gone untreated, it was not possible to estimate how many of such cases existed within the large data set. Ponikowska et al. received the rating of unclear risk of bias for several categories. Although it was mentioned that a control group of healthy patients was used for comparison, absence of details regarding screening/eligibility criteria made it impossible to assess “comparability of cases and controls.” Furthermore, “ascertainment of exposure was unclear;” while the Hidradenitis Suppurativa Severity Index and Sartorius scale were mentioned, no specific information was present regarding the confirmation of diagnosis through patient interviews or medical records.
With regards to cross-sectional studies31,33−35, the mean (STDEV) NOS was 7.38 (2.58). The cross-sectional study by Miller et al. included two different case groups, distinguished as Miller et al. (1) for the hospital HS group, and Miller et al. (2) for the population HS group. The “sample size” for cross-sectional studies was rated with “unclear risk of bias” when implicitly large but not quantified31 or when quantified but not justified as to why a given number of cases were selected.34 Studies with samples of relatively smaller size were also rated similarly;35 however, unlike other studies with unclear bias, Miller et al. (1), was awarded a star for “representativeness of sample,” because the participation rate for all invited cases was stated (34%), and distribution of age and sex did not differ between participants and non-participants. This study was also the only cross-sectional study to report this data on “non-respondents.” Finally, studies that implied the use of medical records or ICD codes for record linkage to databases but that did not explicitly state so received the rating of unclear risk of bias for “ascertainment of exposure.”31, 34
Hidradenitis Suppurativa and Anemia
Six out of 7 included studies presented either unadjusted ORs or prevalence data that allowed for calculation of unadjusted ORs30,32−35 (Fig. 2). Based on unadjusted ORs, 3 out of 6 potential studies demonstrated statistically significant positive associations between HS and anemia.30,33,34 After pooled estimation, it was found that HS patients were associated with a significantly increased odds of anemia compared to the comparison groups (1.70 [1.61, 1.79]).
Four out of 7 included studies reported adjusted analyses.31,34,35 Two demonstrated a positive association between HS and anemia (ORs, 1.87 [1.02 to 3.43] and 2.20 [1.42 to 3.41]).31,34 The other 2 studies (Miller et al. 1 and 2) presented adjusted ORs demonstrating no association.35 After pooled estimation, we found HS was associated with a significantly increased odds of anemia compared to the comparison groups (OR 1.59 [1.19, 2.11]).
For each study, we also noted associated subtypes of anemia in patients with HS30–35 (Table 1). Five out of 7 studies reported on microcytic anemia.30,32,34,35 Additionally, while Kimball et al. examined the association of “deficiency anemias,” Narla S., Silverberg J.I. and one study by Miller et al. reported on macrocytic anemia. Narla S., Silverberg J.I. specified pernicious anemia as the cause of macrocytic anemia (OR, 1.87 [0.85 to 1.57]).
Reporting Bias
The funnel plot (Fig. 3) displays the evaluation of reporting bias. Three of the seven studies fall at the apex of the triangle due to large sample sizes (weights) and are the only ones that are statistically significant (Ramos et al., Kimball et al. and Solimon et al.). One study fell on the negative side (though it was not statistically significant, Miller et al. (1)). There does appear to be a publication bias with the studies included as the plot is asymmetrical and lacks small and negative studies. The small number of studies serves as a limitation and heterogeneity of the studies was also considerable, I2 = 91%.