BJD

British Journal of Dermatology

E P ID EMIOL O GY AN D HE AL TH S ER VI CE S R ESEA RCH

Cancer incidence among patients with a hospital diagnosis of pruritus: a nationwide Danish cohort study S.A. Johannesdottir,1 D.K. Farkas,1 G.R. Vinding,2 L. Pedersen,1 A. Lamberg,1,3 H.T. Sørensen1 and A.B. Olesen3 Departments of 1Clinical Epidemiology and 3Dermatology, Aarhus University Hospital, P.P. Ørumsgade 11, 8000 Aarhus C, Denmark 2 Department of Dermatology, Roskilde Hospital, Health Sciences Faculty, University of Copenhagen, Roskilde, Denmark

Summary Correspondence Anne Braae Olesen. E-mail: [email protected]

Accepted for publication 25 May 2014

Funding sources The study was supported by grants from the Aarhus University Research Foundation, from the Danish Cancer Society R73-A4284-13-S17 and from the Karen Elise Jensen Foundation. The funding sources had no role in the study design, data collection, data analysis, manuscript preparation or publication decisions.

Conflicts of interest Anne Braae Olesen delivered a medical companyfinanced speech in January 2014. This had no relation to the current study. DOI 10.1111/bjd.13157

Background Pruritus is a frequent complaint in patients with cancer. However, no large study has examined pruritus as a marker of undiagnosed cancer. Objectives To examine the association between inpatient, outpatient and emergency hospital diagnoses of pruritus and subsequent cancer diagnoses. Methods In this nationwide Danish cohort study, we used medical databases to identify all patients (n = 12 813) with a diagnosis of pruritus during the period 1978–2011 and followed them until a first-time cancer diagnosis, emigration, death or 31 December 2011. We computed standardized incidence ratios (SIRs) for cancer as the observed to expected number of cancers based on national cancer incidence rates. We calculated the 1-year absolute risk of cancer, treating death as a competing risk. Results The overall SIR of cancer was 1
13 [95% confidence interval (CI) 1
07– 1
20]: 1
22 (95% CI 1
13–1
33) among men and 1
05 (95% CI 0
97–1
14) among women. The SIR was 1
20 (95% CI 1
08–1
33) among patients with a previous diagnosis of dermatological disease and 1
10 (95% CI 1
02–1
18) among patients without such a diagnosis. Both haematological and various solid cancers were observed at increased rates. Overall, the highest SIRs were observed during the first 3 months of follow-up, declining rapidly thereafter. The 1-year absolute risk of a cancer diagnosis was 1
63% and 155 patients with pruritus would have needed to be examined to detect one excess cancer. Conclusions Pruritus may be a marker of occult cancer. Further studies are needed to assess the prognostic benefit of screening for cancer in patients with pruritus.

What’s already known about this topic?

• •

Pruritus is a frequent complaint in patients with cancer. So far, no large study has examined pruritus as a marker of undiagnosed cancer.

What does this study add?

• •

This study demonstrates a twofold increased incidence of cancer among patients with pruritus in the first 3 months after a diagnosis of pruritus, declining rapidly thereafter. The 1-year absolute cancer risk was 1
63% and 155 patients with pruritus would have needed to be examined to detect one excess cancer.

Pruritus (itching) is ‘an unpleasant sensation of the skin leading to the desire to scratch’.1 It may present as a symptom of a wide range of dermatological, neurological, © 2014 British Association of Dermatologists

psychiatric or systemic diseases,1 such as cholestasis, hyperthyroidism, polycythaemia vera and chronic kidney disease.1 British Journal of Dermatology (2014) 171, pp839–846

839

840 Pruritus and cancer, S.A. Johannesdottir et al.

Pruritus is also a frequent complaint of patients with cancer, especially those suffering from haematological malignancies.2 It has been estimated that the prevalence of pruritus is as high as 30% among patients with Hodgkin lymphoma, 10% among patients with non-Hodgkin lymphoma and 5% among patients with leukaemia.2 Despite this recognized link between pruritus and cancer, data are limited regarding pruritus as a marker of undiagnosed cancer. Previous studies have been limited by cross-sectional designs,3–6 short follow-up,7–11 restriction to selected study populations (primarily from dermatology clinics),3–5,9–14 diversity in the definition of pruritus (generalized vs. localized and varying criteria regarding duration and primary skin lesions)3–13 and the comprehensiveness of the work-up for cancer.3–14 Importantly, only one study included a comparison with national cancer incidence.12 However, that study included cancers diagnosed before the start of follow-up and was too small to draw any firm conclusions. Thus, the diversity of prior studies and the lack of comparison groups impede conclusions regarding pruritus as a marker of cancer. To address these shortcomings, we conducted a large comprehensive cohort study in Denmark to assess the association between hospital in- and outpatient diagnoses of pruritus and cancer incidence.

Materials and methods Setting and design The Danish National Health Service provides tax-supported health care to all Danish residents (5
6 million persons).15 All residents are assigned a unique 10-digit central registration number. This identifier is used to record utilization of health services in various nationwide registries and thus allows for record linkage.16 The study period (1 January 1978–31 December 2011) was chosen based on the availability of registry data. The source population consisted of 8 042 510 persons. Pruritus The Danish National Registry of Patients (DNRP) contains information on all nonpsychiatric hospital inpatient admissions since 1977, and on all outpatient clinic and emergency room visits since 1995.17 Each admission includes information on one primary diagnosis and one or more secondary diagnoses coded according to the International Classification of Diseases (ICD; eighth revision through to 1993 and tenth revision thereafter).17 We used the DNRP to identify all individuals with a primary or secondary inpatient, outpatient or emergency department diagnosis of pruritus recorded in the DNRP during the study period. The primary diagnosis was the main reason for the contact and secondary diagnoses were supplemental diagnoses for other diseases related to the current admission or visit. We considered only the first-time primary or secondary British Journal of Dermatology (2014) 171, pp839–846

pruritus diagnosis recorded in the DNRP. Codes used to identify pruritus included not only unspecified codes for pruritus in general, but also lichenification and prurigo (e.g. prurigo nodularis and prurigo mitis). Patients with a previous or concomitant cancer diagnosis were excluded. Malignancy The Danish Cancer Registry (DCR) contains records of all incident cancers in Denmark since 1943.18 During the study period, tumours were classified according to the topography and histology codes in the third revision of the International Classification of Diseases for Oncology and in the ICD-10. Records of cancers in the DCR are based on notification forms completed by practising physicians and hospital departments, including departments of pathology and forensic medicine, at the time of cancer diagnosis, at autopsy or when changes are made to an initial cancer diagnosis.18 Through linkage to the DCR, we followed all members of the study cohort from the diagnosis of pruritus until the firsttime occurrence of cancer. Follow-up was censored at death, emigration or study end date (31 December 2011). Deaths and emigration are updated on a daily basis in the Danish Civil Registration System.16 Based on ICD-10 codes, we examined cancers at 68 individual sites (Table S1; see Supporting Information). For haematological cancers, the morphology codes overruled the ICD-10 codes to allow categorization according to the DCR’s 2009 recommendation.19 We also aggregated cancers into the following five subgroups: (i) smoking- and alcohol-related cancers; (ii) haematological cancers; (iii) immune-related cancers; (iv) cancers of neurological origin; and (v) a category of ‘all other cancers’.20,21 Comorbidities We hypothesized that pruritus among patients with primary dermatological disease may be a weaker marker of underlying cancer than pruritus in patients without any apparent dermatological cause. Therefore, we used the DNRP to differentiate between patients with pruritis with and without a previous diagnosis of a dermatological disease. We also obtained information on HIV diagnoses to examine if the association between pruritus and anal cancer was more pronouned among HIV-positive patients. Also, because underlying liver disease may confound results for smoking- and alcohol-related cancers, we included information on the presence or absence of liver disease. Statistical analysis We first used descriptive statistics to characterize the study cohort. Next, we multiplied the number of person-years of follow-up by Danish national cancer incidence rates (according to age, sex and year of diagnosis) to determine the number of cancers that would be expected if patients with pruritus had the same cancer risk as that of the general population. We © 2014 British Association of Dermatologists

Pruritus and cancer, S.A. Johannesdottir et al. 841

calculated standardized incidence ratios (SIRs), as the ratio of observed to expected cancers, and associated 95% confidence intervals (CIs), assuming that the observed number of cancers in a specific category followed a Poisson distribution. We used exact 95% CIs when the observed number of cancers was < 10; otherwise, Byar’s approximation was used. We performed analyses for all cancers combined, for the five cancer subgroups, and for specific cancer sites (only cancer sites with a total of at least five observed outcomes are presented). Finally, we stratified the overall results by presence/absence of dermatological disease, by the type of pruritus diagnosis (primary or secondary) and by the setting of the diagnosis (inpatient, outpatient clinic or emergency department). We also stratified the results for smoking- and alcohol-related cancers by the presence or absence of liver disease, and the results for anal cancer by the presence or absence of HIV. We calculated the 1-year absolute risk of cancer, treating death as a competing risk. Also, assuming that cancers occurring during the first year of follow-up were present at the time of pruritus diagnosis, we calculated the number of patients ‘needed to examine at time of diagnosis to detect one excess cancer’ as the reciprocal of the excess risk [(observed number of cancers/follow-up time) – (expected number of cancers/follow-up time)] and calculated 95% CIs as the reciprocal of the CI of the excess risk estimate.22 All analyses were performed using SASâ software (SAS Institute Inc., Cary, NC, U.S.A.). The study was approved by the Danish Data Protection Agency (registration number:

2009-41-3015). Danish legislation does not require that registry-based studies obtain approval by an ethical review board or informed consent from patients.

Results Descriptive data We identified 12 813 patients with pruritus, among whom 10 011 (78
1%) had a primary diagnosis, 2802 (21
9%) had a secondary diagnosis, 3607 (27
5%) had an inpatient diagnosis, 8515 (66
5%) had an outpatient diagnosis and 691 (5
4%) had an emergency department diagnosis (Table 1). The male : female ratio was 3 : 4 and the median age was 53 years [interquartile range (IQR) 36–70]. A dermatological disease had previously been diagnosed in 3917 (30
6%) of patients. Cancer incidence We followed the cohort for a total of 97 968 person-years (median 5
88; IQR 2
27–11
37). During follow-up, we observed 1173 cancers compared with 1039 expected cancers, yielding a SIR of 1
13 (95% CI 1
07–1
20) (Table 1). The point estimates were slightly higher among men (SIR 1
22; 95% CI 1
13–1
33) than women (SIR 1
05; 95% CI 0
97–1
14) and among patients with a dermatological disease (SIR 1
20; 95% CI 1
08–1
33) compared with those

Table 1 Standardized incidence ratios (SIRs) among patients with a diagnosis of pruritus

Type of pruritus diagnosis All patients Primary diagnoses only Inpatient diagnoses Outpatient diagnoses Emergency department diagnoses Sex Female Male Other dermatological disease Yes No Age at diagnosis of pruritus (years)a 0–19 20–39 40–49 50–59 60–69 70–79 ≥ 80 Period of pruritus diagnosis 1978–94 1995–2011

Patients with pruritus, n (%)

Cancers observed (n)

Cancers expected (n)

SIR (95% CI)

12 813 (100
0) 10 011 (78
1) 3607 (27
5) 8515 (66
5) 691 (5
4)

1173 894 474 664 35

1039 785 425 584 30
1

1
13 1
14 1
12 1
14 1
16

7302 (57
0) 5511 (43
0)

598 575

569 470

1
05 (0
97–1
14) 1
22 (1
13–1
33)

3917 (30
6) 8896 (69
4)

381 792

318 720

1
20 (1
08–1
33) 1
10 (1
02–1
18)

962 2927 1867 1964 1836 1750 1507

(7
5) (22
8) (14
6) (15
3) (14
3) (13
7) (11
8)

6 114 135 242 273 269 134

5 85 123 213 259 229 124

1
20 1
34 1
10 1
14 1
05 1
17 1
08

2437 (19
0) 10 376 (81
0)

397 776

357 682

1
11 (1
01–1
23) 1
14 (1
06–1
22)

(1
07–1
20) (1
06–1
22) (1
02–1
22) (1
05–1
23) (0
81–1
61)

(0
44–2
63) (1
10–1
61) (0
92–1
30) (1
00–1
29) (0
93–1
19) (1
04–1
32) (0
91–1
28)

CI, confidence interval. aMedian age 53
33 years (95% CI 36
27–70
27).

© 2014 British Association of Dermatologists

British Journal of Dermatology (2014) 171, pp839–846

842 Pruritus and cancer, S.A. Johannesdottir et al.

without a dermatological disease (SIR 1
10; 95% CI 1
02–1
18). SIRs varied by age at cancer diagnosis: 2
25 (95% CI 1
19–3
84) among those aged 0–29 years; 1
23 (95% CI 1
01–1
49) among those aged 30–49 years; 1
12 (95% CI 1
02–1
23) among those aged 50–69 years; and 1
11 (95% CI 1
02–1
20) among those aged ≥ 70 years. This pattern was particularly evident for haematological cancers (data not shown). Analyses by age group at pruritus diagnosis, calendar period, and type and setting of pruritus diagnoses supported the overall results (Table 1). During follow-up, the SIR decreased from 2
14 (95% CI 1
67–2
70) during the first 3 months to 1
42 (95% CI 1
19– 1
68) during 4–12 months, then approached the incidence in the general population during the remainder of the follow-up (Table 2). The pattern of particularly high SIRs during the first year of follow-up was also observed for each of the cancer subtypes (Table 3). The overall absolute risk of cancer during the first year of follow-up was 1
63%. The number of patients with pruritus needed to examine to detect one excess cancer was 155 (95% CI 106–287). Table 4 presents detailed results for individual cancer subgroups and sites. Overall, we found an increased incidence of smoking- and alcohol-related cancers (SIR 1
13; 95% CI 1
07–1
20). The association between pruritus and these cancers was limited to patients with liver disease (SIR 3
61; 95% CI 2
53–4
99). No substantial increase was found among patients without liver disease (SIR 1
09; 95% CI

0
98–1
20). Pruritus was also associated with haematological cancers (SIR 1
68; 95% CI 1
40–1
99), with a particularly high SIR found for Hodgkin lymphoma (SIR 6
96; 95% CI 4
05–11
15). However, we observed no overall association between pruritus and immune-related cancers (SIR 1
02; 95% CI 0
90–1
14) or pruritus and all other cancers (SIR 1
07; 95% CI 0
96–1
19). Results for individual sites were imprecise, but nevertheless revealed that female patients with pruritus were at increased risk of cancer of the external genitalia (SIR 6
12; 95% CI 3
57–9
81) and anus (SIR 3
13; 95% CI 1
15–6
81). All anal cancers occurred in HIV-negative patients. For both sexes, we observed an increased incidence of skin cancers other than melanoma and basal cell carcinoma (SIR 1
43; 95% CI 1
08–1
86) and a particularly high, albeit imprecise, incidence of Kaposi sarcoma (SIR 35
85; 95% CI 13
16–78
16).

Discussion In this nationwide cohort study, we found a higher than expected (13%) number of cancers among patients with pruritus. The increase was twofold within the first 3 months following diagnosis, then fell rapidly to 1
4-fold during the remainder of the first year and finally approached the incidence in the general population after 10 years of follow-up. Both haematological cancers and various solid cancers occurred at higher rates than expected. The absolute risk of cancer during the first year of follow-up was 1
63%, and the

Table 2 Standardized incidence ratios (SIRs) of cancer by length of follow-up among patients with a diagnosis of pruritus Men

Women

Total

Follow-up time

Number of observed cancers

SIR (95% CI)

Number of observed cancers

SIR (95% CI)

Number of observed cancers

SIR (95% CI)

0–3 months 4–12 months 1–2 years 3–4 years 5–9 years ≥ 10 years

42 68 114 78 150 123

2
58 1
51 1
17 1
05 1
22 1
08

29 64 115 102 143 145

1
72 1
33 1
07 1
17 0
96 0
91

71 132 229 180 293 268

2
14 1
42 1
11 1
12 1
08 0
98

(1
86–3
49) (1
17–1
91) (0
96–1
40) (0
83–1
32) (1
03–1
43) (0
90–1
29)

(1
15–2
47) (1
02–1
70) (0
88–1
28) (0
96–1
43) (0
81–1
13) (0
76–1
07)

(1
67–2
70) (1
19–1
68) (0
97–1
27) (0
96–1
30) (0
96–1
21) (0
86–1
10)

CI, confidence interval.

Table 3 Standardized incidence ratios (SIRs) and 95% confidence intervals (CIs) for subgroups of cancer by length of follow-up among patients with a diagnosis of pruritus

Follow-up time

Smoking- and alcoholrelated cancers

Haematological cancers

Immune-related cancers

Cancers of neurological origin

All other cancers

0–3 months 4–12 months 1–2 years 3–4 years 5–9 years ≥ 10 years

1
35 1
29 1
21 1
45 1
09 0
92

9
99 2
88 1
68 1
06 1
18 1
02

1
19 1
57 0
81 0
97 1
05 0
97

5
81 – – 1
15 0
67 1
03

1
71 1
11 1
15 0
89 1
08 1
03

(0
77–2
19) (0
94–1
74) (0
97–1
49) (1
15–1
80) (0
89–1
32) (0
74–1
15)

British Journal of Dermatology (2014) 171, pp839–846

(6
52–14
64) (1
78–4
40) (1
11–2
45) (0
56–1
81) (0
76–1
76) (0
62–1
57)

(0
57–2
20) (1
10–2
16) (0
59–1
10) (0
69–1
31) (0
83–1
32) (0
76–1
21)

(0
70–20
96)

(0
14–4
14) (0
08–2
43) (0
21–3
01)

(1
00–2
74) (0
75–1
57) (0
90–1
45) (0
64–1
20) (0
87–1
33) (0
83–1
28)

© 2014 British Association of Dermatologists

All cancers Smoking- and alcohol-related cancers Tongue Mouth Tonsil and pharynx Larynx Lung, bronchus, and trachea Oesophagus Stomach Colon including rectosigmoideum Rectum Liver, including intrahepatic bile ducts Pancreas Kidney Urinary bladder Haematological cancers Hodgkin lymphoma Non-Hodgkin lymphoma Multiple myeloma and other plasma cell neoplasms Myeloid leukaemia Lymphocytic leukaemia Metastasis and unspecified cancer in lymph nodes Immune-related cancers Anal canal Cervix External female genitalia Malignant melanoma Basal cell carcinoma Other skin cancer Kaposi sarcoma Cancers of neurological origin Brain All other cancers Gallbladder and biliary tract Breast

Cancer sitea

© 2014 British Association of Dermatologists – – –

1
03 (0
85–1
23)

119 – – – 8 77 30 4 4 4 128 4 2 (0
28–1
27) (0
73–1
16) (1
03–2
18) (7
25–68
27) (0
19–1
81) (0
20–1
86) (0
91–1
30) (0
48–4
56) (0
36–10
60)

2
00 (0
86–3
94) 1
51 (0
69–2
88) 1
65 (0
94–2
68)

8 9 16

0
65 0
93 1
53 26
67 0
71 0
73 1
09 1
78 2
94

1
71 1
68 1
10 2
03 6
38 2
26 1
48

19 15 20 78 8 26 7

(1
03–2
67) (0
94–2
77) (0
67–1
70) (1
61–2
54) (2
75–12
57) (1
47–3
31) (0
59–3
05)

1
02 (0
64–1
54) 3
23 (1
84–5
24)

22 16

(0
62–5
86) (0
54–3
88) (0
33–2
40) (0
88–3
37) (0
98–1
57) (0
75–2
68) (0
60–2
02) (0
59–1
26)

2
29 1
66 1
03 1
83 1
25 1
50 1
15 0
88

1
22 (1
13–1
33) 1
27 (1
12–1
33)

4 5 5 10 73 11 12 30

575 246

163 6 11 17 15 87 25 2 5 4 210 4 135

8 6 6

9 15 6 53 9 18 5

15 6

58 2 7 40

(0
85–1
44) (0
07–2
02) (0
40–2
07) (0
64–1
23)

(0
30–1
24) (1
25–3
70) (0
30–1
78) (1
00–1
74) (3
47–14
39) (0
88–2
34) (0
36–2
57)

(0
86–1
18) (1
15–6
81) (0
47–1
68) (3
57–9
81) (0
45–1
31) (0
65–1
00) (0
86–1
97) (13
93–415
67) (0
30–2
13) (0
21–1
95) (0
92–1
22) (0
28–2
65) (0
89–1
26)

2
09 (0
90–4
11) 1
18 (0
43–2
56) 0
50 (0
18–1
09)

0
65 2
24 0
82 1
33 7
57 1
48 1
10

0
86 (0
48–1
41) 1
93 (0
71–4
21)

1
12 0
56 1
01 0
90

1
01 3
13 0
94 6
12 0
80 0
81 1
33 115
14 0
92 0
76 1
06 1
04 1
06

–

2
41 (0
50–7
05) 0
42 (0
01–2
32) 1
51 (0
31–4
39)

3 1 3 –

1
05 (0
97–1
14) 1
02 (0
87–1
18)

SIR (95% CI)

598 167

Number of observed cancers

Number of observed cancers SIR (95% CI)

Women

Men

Table 4 Standardized incidence ratios (SIRs) for cancer by site among patients with pruritus

282 6 11 17 23 164 55 6 9 8 388 8 137

16 15 22

28 30 26 131 17 44 12

37 22

7 6 8 10 131 13 19 70

1173 413

Number of observed cancers

Total

(0
94–4
82) (0
41–2
42) (0
50–2
30) (0
70–2
68) (0
99–1
41) (0
63–2
03) (0
66–1
71) (0
70–1
13)

(0
75–1
62) (1
30–2
74) (0
66–1
49) (1
40–1
99) (4
05–11
15) (1
35–2
49) (0
67–2
26)

1
02 2
27 0
94 6
12 0
74 0
86 1
43 35
85 0
81 0
74 1
07 1
31 1
07

(0
90–1
14) (0
83–4
94) (0
47–1
68) (3
57–9
81) (0
47–1
11) (0
74–1
01) (1
08–1
86) (13
16–78
16) (0
37–1
54) (0
32–1
46) (0
96–1
19) (0
57–2
58) (0
90–1
27)

2
04 (1
17–3
32) 1
36 (0
76–2
24) 1
01 (0
64–1
54)

1
12 1
92 1
02 1
68 6
96 1
86 1
30

0
95 (0
67–1
31) 2
73 (1
71–4
13)

2
34 1
11 1
17 1
46 1
19 1
19 1
10 0
89

1
13 (1
07–1
20) 1
13 (1
07–1
20)

SIR (95% CI)

82 1 3 4 6 46 19 3 2 2 118 2 35

4 6 7

12 11 10 45 4 18 6

11 8

5 2 4 4 37 2 4 20

381 134

Number of observed cancers

(1
76–12
68) (0
15–4
42) (0
52–4
90) (0
51–4
80) (0
76–1
49) (0
07–2
09) (0
21–1
96) (0
51–1
28)

(0
81–2
75) (1
15–4
14) (0
58–2
24) (1
37–2
52) (1
64–15
46) (1
48–3
95) (0
77–4
57)

0
97 1
26 0
96 4
99 0
66 0
78 1
54 61
21 0
60 0
62 1
24 1
10 0
94

(0
77–1
20) (0
03–7
03) (0
20–2
82) (1
36–12
78) (0
24–1
43) (0
57–1
05) (0
92–2
40) (12
61–178
73) (0
07–2
17) (0
07–2
24) (1
03–1
48) (0
13–3
96) (0
66–1
31)

1
67 (0
45–4
28) 1
78 (0
65–3
89) 1
04 (0
42–2
14)

1
57 2
32 1
22 1
88 6
04 2
50 2
09

0
91 (0
45–1
63) 3
22 (1
39–6
34)

5
44 1
22 1
91 1
87 1
08 0
58 0
76 0
83

1
20 (1
08–1
33) 1
21 (1
02–1
44)

SIR (95% CI)

Dermatological disease positive

200 5 8 13 17 118 36 3 7 6 220 6 102

12 9 15

16 19 16 86 13 26 6

26 14

2 4 4 6 94 11 15 50

792 279

Number of observed cancers

(0
12–3
48) (0
29–2
71) (0
23–2
15) (0
47–2
77) (1
00–1
51) (0
73–2
63) (0
69–2
04) (0
68–1
21)

(0
53–1
50) (1
05–2
73) (0
53–1
50) (1
27–1
96) (3
89–12
49) (1
03–2
31) (0
34–2
04)

1
04 2
69 0
93 6
58 0
77 0
90 1
38 25
35 0
90 0
79 1
00 1
40 1
13

(0
90–1
20) (0
87–6
28) (0
40–1
83) (3
50–11
26) (0
45–1
23) (0
74–1
08) (0
97–1
92) (5
22–74
03) (0
36–1
85) (0
29–1
73) (0
87–1
14) (0
51–3
06) (0
92–1
37)

2
21 (1
14–3
86) 1
17 (0
54–2
23) 1
00 (0
56–1
65)

0
93 1
75 0
92 1
59 7
30 1
58 0
94

0
96 (0
63–1
41) 2
51 (1
37–4
21)

0
96 1
06 0
84 1
27 1
23 1
47 1
24 0
92

1
10 (1
02–1
18) 1
13 (1
00–1
27)

SIR (95% CI)

Dermatological disease negative

Pruritus and cancer, S.A. Johannesdottir et al. 843

British Journal of Dermatology (2014) 171, pp839–846

(0
63–1
73) (0
56–1
79) (0
57–1
06) (0
56–4
01) (0
20–2
88) (0
09–1
27) 1
08 1
05 0
79 1
72 0
99 0
44 17 13 44 5 3 3 (0
04–1
12) (0
69–3
15) (1
09–2
05) (0
23–6
94) (0
21–6
14) (0
37–3
45) 0
31 1
60 1
52 1
92 1
70 1
35 2 8 42 2 2 4 (0
52–1
34) (0
75–1
84) (0
82–1
27) (0
71–3
65) (0
38–2
76) (0
28–1
46) 0
86 1
21 1
03 1
77 1
19 0
71 0
92 (0
19–2
69) 0
50 (0
10–1
47) 3 3

– – – – – –

– – (0
82–1
27) (0
71–3
65) (0
25–7
49) (0
28–2
63) 1
03 1
77 2
08 1
03 86 7 2 4 Uterus Ovary and fallopian tube Prostate Testicle Thyroid gland Malignant neoplasm at other, ill-defined or unspecified sites

CI, confidence interval. aOnly cancers with a total of at least five observed outcomes are presented.

19 21 86 7 5 7

SIR (95% CI)

19 21

Number of observed cancers Number of observed cancers

SIR (95% CI) Women Men

Cancer sitea

Table 4 (continued)

0
86 (0
52–1
34) 1
21 (0
75–1
84)

SIR (95% CI) SIR (95% CI)

Number of observed cancers Number of observed cancers

Total

SIR (95% CI)

Dermatological disease positive

Number of observed cancers

Dermatological disease negative

844 Pruritus and cancer, S.A. Johannesdottir et al.

British Journal of Dermatology (2014) 171, pp839–846

number of patients with pruritus needed to examine to detect one excess cancer was 155. With regard to size, the current study, which included 12 813 patients with pruritus, far exceeds previous studies,3– 14,23–32 and provides precise and complete long-term followup estimates of increased cancer risk among patients with pruritus with and without a dermatological diagnosis. Several case reports and series supportive of pruritus as a presenting symptom of malignancy have been published.23–32 In four crosssectional studies including between 34 and 109 selected patients with generalized pruritus or pruritus ani,3–6 the prevalence of undiagnosed cancer varied widely, from 1
3% to as high as 26
0%.3–6 In the six studies that followed between 43 and 263 pruritus patients, cancer was found in 2
3–6
4%.7–12 Finally, two studies of 64 patients with generalized pruritus and 447 patients with ‘idiopathic pruritus’ found cancers among 2% and 0
7%, respectively.13,14 The authors did not report how thoroughly patients were examined and followed up for cancer.13,14 Only one previous study on pruritus and cancer risk included a comparison with national cancer rates.12 In this Finnish 6-year follow-up study of 125 patients with pruritus, eight malignancies were detected overall (6
4%), similar to the cancer rate in the general population.12 However, lymphomas occurred more frequently than expected. Limitations of the study included its small size, a selected study population consisting of patients so severely affected by pruritus to warrant inpatient treatment and the inclusion of two cancers that were diagnosed before admission for pruritus.12 Our study confirms the association between pruritus and lymphoma, and extends it to other haematological malignancies and several types of solid cancers. Given the registry-based design of our study with lack of clinical details, we can only speculate on the mechanisms underlying our results. The high rate of cancer early in follow-up suggests that pruritus is caused by an underlying cancer disease rather than vice versa. The pathophysiology of itching is complicated, as it can originate from the skin (pruritoceptive itch, e.g. urticaria and dermatitis), along the afferent pathway of nerves (neuropathic itch) or centrally without neural harm (neurogenic itch through endogenous and exogenous opioid mediators).33,34 Both histaminergic and nonhistaminergic mediators of pruritus are known, including bile salts and acids, bilirubin and other cholephiles.34,35 Several of these mediators are known to accumulate in patients with cholestasis. Thus, cholestasis caused by primary or metastatic cancer of the liver may explain pruritus among some patients in our cohort. Also, it is interesting that we found an increased rate of various cancers linked to viruses that are themselves associated with pruritic diseases, for example condylomas (human papillomavirus) and HIV.34,36 It is important to note that although we included information on underlying diseases that may be associated with both pruritus and cancer (e.g. HIV and liver disease, including viral hepatitis), we cannot entirely rule out that confounding stemming from such diseases explains the association observed. In fact, stratification © 2014 British Association of Dermatologists

Pruritus and cancer, S.A. Johannesdottir et al. 845

by liver disease supported this potential explanation for smoking- and alcohol-related cancers. Our study has strengths. We used data from nationwide medical registries to provide complete prospective follow-up data for the entire Danish population. Compared with previous studies, our results are thus derived from a broader group of patients, with pruritus diagnosed in various clinical settings. Comprehensive assessment has shown that DCR data are more than 95% complete and valid.37 However, the validity of discharge diagnoses recorded in the DNRP is known to vary by specialty,38 and has not been examined for pruritus diagnoses specifically. Completeness is also unknown and some patients with pruritus may be seen only by general practitioners (GPs) outside the hospital setting. It is unclear what effect this potential misclassification would have on the results. It is possible that GPs are prone to refer patients with unexplained pruritus to specialists. Such referral filter bias could result in an overestimation of the true association if referred patients are at higher risk of underlying malignancy. However, in our study, the SIRs did not depend on the type and setting of diagnoses, which suggests that such misclassification would have little impact on our results. Overestimation may also have been induced by detection bias if patients referred to hospitals or outpatient clinics with pruritus are subject to more extensive examination for an underlying cause, thereby increasing the chance of detection of cancer. However, such bias would typically produce high SIRs early in follow-up with a compensatory decrease thereafter,39 which we did not observe. Unfortunately, we were unable to distinguish clearly between localized and generalized pruritus and the presence or absence of primary skin lesions. However, we included data on previous dermatological diseases. Although not substantially different, it revealed that patients with pruritus with a previous dermatological disease might have a slightly higher risk of underlying cancer than patients without dermatological disease. This finding was contrary to our hypothesis, but may reflect that patients with skin diseases are already in contact with the healthcare system or that they react better to changes in symptoms or the development of new symptoms like itch. Alternatively, the association may be partly explained by an imbalanced immune system caused by some skin diseases themselves or by their treatment.40–43 Another concern is that we were unable to differentiate between acute and chronic pruritus. It is likely that patients presenting with pruritus at emergency departments are affected by acute pruritus, while patients referred to other hospital departments or outpatient clinics have chronic pruritus. As only 5% of pruritus diagnoses in our study were from emergency departments, we believe that our results primarily reflect chronic pruritus. Although we included data on sex, age and other diseases, unmeasured confounding factors may be present. For smoking- and alcohol-related cancers, the association was restricted to patients with a diagnosis of liver disease, which may reflect unmeasured confounding. © 2014 British Association of Dermatologists

The clinical implications of the association between pruritus and cancer are unclear. In this study, we found that the number of patients with pruritus needed to be examined at diagnosis to detect one excess cancer was as low as 155. However, we do not know to what extent patients were examined for cancer. Therefore, we could not estimate the number of patients needed to examine in order to prevent one cancer death. Earlier diagnosis of some cancers may require extensive diagnostic work-up, which can be associated with discomfort and physiological stress without improving the prognosis (e.g. oesophageal and liver cancer).44 Thus, based on the current evidence, we cannot recommend extensive diagnostic work-up for cancer in patients with pruritus. In conclusion, we found that men and women affected by itching who seek medical advice at hospital, outpatient clinics or emergency departments are at increased risk of occult cancer, especially during the first 3 months of follow-up.

References 1 St€ander S, Weisshaar E, Mettang T et al. Clinical classification of itch: a position paper of the International Forum for the Study of Itch. Acta Derm Venereol 2007; 87:291–4. 2 Weisshaar E, Dalgard F. Epidemiology of itch: adding to the burden of skin morbidity. Acta Derm Venereol 2009; 89:339–50. 3 Goon AT, Yosipovitch G, Chan YH, Goh CL. Clinical characteristics of generalized idiopathic pruritus in patients from a tertiary referral center in Singapore. Int J Dermatol 2007; 46:1023–6. 4 Rajka G. Investigation of patients suffering from generalized pruritus, with special references to systemic diseases. Acta Derm Venereol 1966; 46:190–4. 5 Lyell A. The itching patient. A review of the causes of pruritus. Scott Med J 1972; 17:334–7. 6 Daniel GL, Longo WE, Vernava AM. Pruritus ani. Causes and concerns. Dis Colon Rectum 1994; 37:670–4. 7 Beare JM. Generalized pruritus. A study of 43 cases. Clin Exp Dermatol 1976; 1:343–52. 8 Kantor GR, Lookingbill DP. Generalized pruritus and systemic disease. J Am Acad Dermatol 1983; 9:375–82. 9 Zirwas MJ, Seraly MP. Pruritus of unknown origin: a retrospective study. J Am Acad Dermatol 2001; 45:892–6. 10 Polat M, Oztas P, Ilhan MN et al. Generalized pruritus: a prospective study concerning etiology. Am J Clin Dermatol 2008; 9:39–44. 11 Sommer F, Hensen P, B€ ockenholt B et al. Underlying diseases and co-factors in patients with severe chronic pruritus: a 3-year retrospective study. Acta Derm Venereol 2007; 87:510–16. 12 Paul R, Paul R, Jansen CT. Itch and malignancy prognosis in generalized pruritus: a 6-year follow-up of 125 patients. J Am Acad Dermatol 1987; 16:1179–82. 13 Forman L. Pruritus and its management. Br Med J 1954; 1:365–7. 14 Rantuccio F. Incidence of malignancy in patients with generalized pruritus. J Am Acad Dermatol 1989; 21:1317. 15 Ministry of Health and Prevention. Healthcare in Denmark. Available at: http://www.sum.dk/Aktuelt/Publikationer/UK_Healthcare_in_ DK.aspx (last accessed February 2014). 16 Pedersen CB. The Danish civil registration system. Scand J Public Health 2011; 39:22–5. 17 Lynge E, Sandegaard JL, Rebolj M. The Danish national patient register. Scand J Public Health 2011; 39:30–3. 18 Gjerstorff ML. The Danish cancer registry. Scand J Public Health 2011; 39:42–5. British Journal of Dermatology (2014) 171, pp839–846

846 Pruritus and cancer, S.A. Johannesdottir et al. 19 Danish Health and Medicines Authority. Cancer register 2007. Available at: http://www.sst.dk/publ/Publ2009/DOKU/cancerreg/ cancerregister2007.pdf (last accessed October 2013). 20 Boffetta P, Hashibe M. Alcohol and cancer. Lancet Oncol 2006; 7:149–56. 21 Boyle P. Cancer, cigarette smoking and premature death in Europe: a review including the Recommendations of European Cancer Experts Consensus Meeting, Helsinki, October 1996. Lung Cancer 1997; 17:1–60. 22 Altman DG. Confidence intervals for the number needed to treat. BMJ 1998; 317:1309–12. 23 Erskine JG, Rowan RM, Alexander JO, Sekoni GA. Pruritus as a presentation of myelomatosis. Br Med J 1977; 1:687–8. 24 Puchner TC Jr, Fink JN. A case of generalized pruritus without rash. Ann Allergy Asthma Immunol 2000; 85:450–6. 25 O’Donnell BF, Alton B, Carney D, O’Loughlin S. Generalized pruritus: when to investigate further. J Am Acad Dermatol 1993; 28:117. 26 Carter JB, Barnes JA, Niell BL, Nardi V. Case records of the Massachusetts General Hospital. Case 24-2013. A 53-year-old woman with erythroderma, pruritus, and lymphadenopathy. N Engl J Med 2013; 369:559–69. 27 Cormia FE. Pruritus, an uncommon but important symptom of systemic carcinoma. Arch Dermatol 1965; 92:36–9. 28 Magilner D. Localized cervical pruritus as the presenting symptom of a spinal cord tumor. Pediatr Emerg Care 2006; 22:746–7. 29 Darken RS, Bogitch R, Leonard J et al. Brainstem glioma presenting as pruritus in children with neurofibromatosis-1. J Pediatr Hematol Oncol 2009; 31:972–6. 30 Peterson AO Jr, Jarratt M. Pruritus and nonspecific nodules preceding myelomonocytic leukemia. J Am Acad Dermatol 1980; 2:496–8. 31 Caravati CM, Richardson DR. Pruritus as a symptom of systemic disease. Va Med Mon (1918) 1969; 96:656–8. 32 Summers CG, MacDonald JT. Paroxysmal facial itch: a presenting sign of childhood brainstem glioma. J Child Neurol 1988; 3:189–92. 33 Twycross R, Greaves MW, Handwerker H et al. Itch: scratching more than the surface. QJM 2003; 96:7–26. 34 Wang H, Yosipovitch G. New insights into the pathophysiology and treatment of chronic itch in patients with end-stage renal

British Journal of Dermatology (2014) 171, pp839–846

35

36 37

38

39 40 41

42

43

44

disease, chronic liver disease, and lymphoma. Int J Dermatol 2010; 49:1–11. Mela M, Mancuso A, Burroughs AK. Review article: pruritus in cholestatic and other liver diseases. Aliment Pharmacol Ther 2003; 17:857–70. Hausen zur H. Papillomaviruses in the causation of human cancers – a brief historical account. Virology 2009; 384:260–5. Storm HH, Michelsen EV, Clemmensen IH, Pihl J. The Danish Cancer Registry—history, content, quality and use. Dan Med Bull 1997; 44:535–9. Andersen TF, Madsen M, Jørgensen J et al. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 1999; 46:263–8. Baron JA, Gridley G, Weiderpass E et al. Venous thromboembolism and cancer. Lancet 1998; 351:1077–80. Frentz G, Olsen JH. Malignant tumours and psoriasis: a follow-up study. Br J Dermatol 1999; 140:237–42. Jensen AØ, Thomsen HF, Engebjerg MC et al. Use of oral glucocorticoids and risk of skin cancer and non-Hodgkin’s lymphoma: a population-based case–control study. Br J Cancer 2009; 100:200–5. Olesen AB, Svaerke C, Farkas DK, Sørensen HT. Systemic sclerosis and the risk of cancer: a nationwide population-based cohort study. Br J Dermatol 2010; 163:800–6. Olesen AB, Engholm G, Storm HH, Thestrup-Pedersen K. The risk of cancer among patients previously hospitalized for atopic dermatitis. J Invest Dermatol 2005; 125:445–9. Walter LC, Covinsky KE. Cancer screening in elderly patients: a framework for individualized decision making. JAMA 2001; 285:2750–6.

Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s website: Table S1. International Classification of Diseases (ICD) codes used in this study.

© 2014 British Association of Dermatologists