Psoriasis, Psoriatic Arthritis and Type 2 Diabetes Mellitus
Psoriasis, Psoriatic Arthritis and Type 2 Diabetes Mellitus
The initial search provided a total of 274 articles. The search in the Cochrane Library found no results. A preliminary review of the article titles and abstracts restricted the search to 50 articles. Thirteen articles did not fulfil the inclusion criteria. Finally, only 37 papers including 44 studies fulfilled the inclusion criteria. A flowchart of the search process is presented in Figure 1. No evidence for publication bias was found, as indicated by visual inspection of the funnel plot (Fig. 2), proven by nonsignificant Begg's (1·1025, P = 0·2743) and Egger's tests (1·6678, P = 0·1018). Among these studies, most were case–control studies (n = 20), eight were cross-sectional and nine were cohort studies. When a study included different cohorts, each cohort was analysed as a separate study for the meta-analysis (Table 1). Newcastle–Ottawa Scale scores for the studies are given in Table 2.
(Enlarge Image)
Figure 1.
Flow chart of the selection process. The course of the systematic literature review on psoriasis and the risk of type 2 diabetes mellitus, 1966 to December 2012. OR, odds ratio.
(Enlarge Image)
Figure 2.
Publication bias funnel plot for the primary outcome. OR, odds ratio.
Our meta-analysis included 557 697 patients with psoriasis and 5 186 485 controls. The proportion of the total variation in the study estimates, because of heterogeneity, was 98% (heterogeneity test I ) and was significant (P < 0·001). The DerSimonian and Laird heterogeneity test was also significant (Q = 2915·1148, P < 0·0001). Therefore, a random-effects model was also used to perform the meta-analysis. To explore sources of heterogeneity among the studies and to determine how they would influence the estimates, we perform meta-regression and sensitivity analyses. The results of the meta-regression are summarized in Table 3. The sensitivity analysis shows a variation ranging from −2·17% to 1·57%. The sensitivity analysis performs k different meta-analyses (k is the number of included studies) excluding, each time, one of the papers. This process allows determination of the robustness of the obtained results and detection of possible influence points (papers).
The pooled OR from the random-effects analysis for the association between psoriasis and the risk of type 2 diabetes mellitus was determined to be 1·76 (95% CI 1·59–1·96; Fig. 3). The highest risk was for patients with PsA (OR 2·18, 95% CI 1·36–3·50; Fig. 4). We also observed a dose effect in the risk of suffering from type 2 diabetes mellitus, as patients considered as having severe psoriasis had higher risk (OR 2·10, 95% CI 1·73–2·55) than the pooled OR (Fig. 5). Finally, we considered the cohort studies separately and found an RR of 1·50 (95% CI 1·27–1·77; Fig. 6).
(Enlarge Image)
Figure 3.
Psoriasis and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model. Cohen* 2008 is reference 49 and Cohen 2008 is reference 35.
(Enlarge Image)
Figure 4.
Psoriatic arthritis (PsA) and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with PsA compared with subjects without psoriasis. Studies are ordered alphabetically by first author. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model.
(Enlarge Image)
Figure 5.
Patients with severe psoriasis and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with severe psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model.
(Enlarge Image)
Figure 6.
Forest plot for cohort studies. Risk ratios for type 2 diabetes mellitus in subjects with psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled risk ratio (diamond) was calculated by means of a random-effects model.
Results
The initial search provided a total of 274 articles. The search in the Cochrane Library found no results. A preliminary review of the article titles and abstracts restricted the search to 50 articles. Thirteen articles did not fulfil the inclusion criteria. Finally, only 37 papers including 44 studies fulfilled the inclusion criteria. A flowchart of the search process is presented in Figure 1. No evidence for publication bias was found, as indicated by visual inspection of the funnel plot (Fig. 2), proven by nonsignificant Begg's (1·1025, P = 0·2743) and Egger's tests (1·6678, P = 0·1018). Among these studies, most were case–control studies (n = 20), eight were cross-sectional and nine were cohort studies. When a study included different cohorts, each cohort was analysed as a separate study for the meta-analysis (Table 1). Newcastle–Ottawa Scale scores for the studies are given in Table 2.
(Enlarge Image)
Figure 1.
Flow chart of the selection process. The course of the systematic literature review on psoriasis and the risk of type 2 diabetes mellitus, 1966 to December 2012. OR, odds ratio.
(Enlarge Image)
Figure 2.
Publication bias funnel plot for the primary outcome. OR, odds ratio.
Our meta-analysis included 557 697 patients with psoriasis and 5 186 485 controls. The proportion of the total variation in the study estimates, because of heterogeneity, was 98% (heterogeneity test I ) and was significant (P < 0·001). The DerSimonian and Laird heterogeneity test was also significant (Q = 2915·1148, P < 0·0001). Therefore, a random-effects model was also used to perform the meta-analysis. To explore sources of heterogeneity among the studies and to determine how they would influence the estimates, we perform meta-regression and sensitivity analyses. The results of the meta-regression are summarized in Table 3. The sensitivity analysis shows a variation ranging from −2·17% to 1·57%. The sensitivity analysis performs k different meta-analyses (k is the number of included studies) excluding, each time, one of the papers. This process allows determination of the robustness of the obtained results and detection of possible influence points (papers).
The pooled OR from the random-effects analysis for the association between psoriasis and the risk of type 2 diabetes mellitus was determined to be 1·76 (95% CI 1·59–1·96; Fig. 3). The highest risk was for patients with PsA (OR 2·18, 95% CI 1·36–3·50; Fig. 4). We also observed a dose effect in the risk of suffering from type 2 diabetes mellitus, as patients considered as having severe psoriasis had higher risk (OR 2·10, 95% CI 1·73–2·55) than the pooled OR (Fig. 5). Finally, we considered the cohort studies separately and found an RR of 1·50 (95% CI 1·27–1·77; Fig. 6).
(Enlarge Image)
Figure 3.
Psoriasis and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model. Cohen* 2008 is reference 49 and Cohen 2008 is reference 35.
(Enlarge Image)
Figure 4.
Psoriatic arthritis (PsA) and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with PsA compared with subjects without psoriasis. Studies are ordered alphabetically by first author. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model.
(Enlarge Image)
Figure 5.
Patients with severe psoriasis and subsequent risk of type 2 diabetes mellitus. Odds ratios (ORs) for type 2 diabetes mellitus in subjects with severe psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled OR (diamond) was calculated by means of a random-effects model.
(Enlarge Image)
Figure 6.
Forest plot for cohort studies. Risk ratios for type 2 diabetes mellitus in subjects with psoriasis compared with subjects without psoriasis. Studies are ordered by their weight. The point estimate (centre of each blue square) and the statistical size (proportional area of the square) are represented. Horizontal lines indicate 95% confidence intervals (CIs). The pooled risk ratio (diamond) was calculated by means of a random-effects model.
Source...