Self-reported health and use of health care services in long-term cancer survivors

Abstract

Owing to an increasing number of long-term cancer survivors, the use of health care services and somatic health problems were compared between cancer survivors and a noncancer population. Data from the Nord-Trøndelag Health Survey 2 (HUNT 2, 1995–1997) was merged with the Cancer Registry of Norway. Six cancer subgroups were constructed with diagnosis 5 years prior HUNT 2: testicular cancer (n= 59), colorectal cancer (n= 175), prostate cancer (n= 87), breast cancer (n= 258), gynaecological cancer (n= 153) and lymphoma/leukaemia (n= 83). For each cancer survivor 3 matched noncancer controls were selected from the HUNT 2 survey. The prevalence of common health problems, use of health care services and unfavourably life style parameters were compared between the 2 groups. Cancer survivors used health care services and received social welfare benefits more often than the controls. There was an increased risk of perceiving poor health after a history of cancer. Common health problems and/or unfavourable life style parameters could not explain poor health or the increased use of health care services among cancer survivors. Further studies are needed to investigate the reasons for increased use of health care services and perceived poor health in cancer survivors. © 2004 Wiley-Liss, Inc.

In the recent decade, the incidence of most malignancies has increased, as well as 5-year survival rates in both men and women.1 During 1992–1996, the relative overall 5-year survival rate for all cancer patients in Norway was, respectively, 50% in males and 58%; in females compared to 25% in males and 38%; in females for the period 1957–1961. In 1999, a total of 83,636 cancer survivors with an at-least 5-year survival were alive in Norway.1 These figures have in the last few years led to a paradigm shift: cancer should not only be considered an acute or terminal event but should likewise be viewed as a chronic condition. Quality of life and long-term effects after cancer therapy have therefore become important outcome parameters to be considered by oncologists, cancer surgeons and primary health care officers.

Cancer represents a heterogeneous group of diagnoses where treatment and aetiology are far from uniform. The cancer itself and its treatment may affect the long-term health of cancer survivors at various degrees, causing adverse health conditions, which differ from those observed in the age-matched noncancer population. The recognition of common health problems reducing the overall well being among cancer survivors is of importance in order to identify these individuals' specific needs when presented to the health care service. Furthermore the recognition of long-term effects of treatment may lead to reconsideration and eventual modification of current treatment strategies.

Several reports have focused on long-term medical effects after treatment for childhood cancer as well as adulthood cancer and have described specific health problems related to different treatment modalities.2, 3, 4, 5, 6, 7, 8, 9 Beside causing chronic health conditions, cancer and its treatment increase the risk of secondary malignancies.10, 11, 12, 13

Owing to the increased risk of developing secondary tumours and chronic diseases after treatment for malignancies, cancer survivors represent an important target population for health promotion interventions. Few studies regarding long-term cancer survivors' life style behaviour have been performed.14

Our case-control study aims to study different aspects of somatic health in cancer survivors and their use of health care services compared to controls without cancer in the community. We hypothesised that more cancer survivors (cases) would display common health problems than the noncancer population (controls), which would result in a poor health status as well as an increased use of health care services among the cases. We also expected that cancer survivors would display a favourable lifestyle different from that observed among the controls. With this background we wanted to a) first, determine the general health status in long-term cancer survivors in comparison to age-matched controls; b) second, determine whether cancer survivors' use of health care services differed from that of the age-matched controls; and c) third, explore if common health problems or unfavourable life style parameters could explain the general health status and use of health care services in long-term cancer survivors.

Material and methods

The Nord-Trøndelag Health Survey (the HUNT study)

Nord-Trøndelag is 1 of 19 Norwegian counties, located in the middle of Norway. The county is mostly rural and sparsely populated, the largest city having 20,467 inhabitants. The population of Nord-Trøndelag county is stable (127,000 residents) and homogenous (less than 3% nonCaucasian), making it suitable for epidemiological studies.15

The Norwegian National Health Screening Service conducted the Nord-Trøndelag Health Survey 1995–1997 (HUNT 2), a cross-sectional study consisting of a questionnaire to be completed by the participants, followed by physical examination. The principal objectives in the survey were aimed at the large public adverse health conditions, such as cardiovascular disease, diabetes mellitus, obstructive lung disease, osteoporosis, thyroid diseases and mental health. The study also addressed more general health complaints (pain, nausea etc.) and life style parameters. All residents in Nord-Trøndelag County, aged 20 years or older, were invited. The compliance rate was 71% (approximately 65,000 participants).

In Norway, every individual has a unique 11-digit personal identification number given at birth. As this identification number was linked to the HUNT 2 data, cross-references to other national and regional health registries were possible.

The Cancer Registry of Norway (CRN)

Since 1953 it has been mandatory by law for all hospital departments and histopathology laboratories to report all new cases of cancer to the Cancer Registry of Norway (CRN), Oslo, Norway.16 With the exception of basal cell carcinoma and leukaemia, an evaluation of registration in 1980 showed a completeness of almost 100% for solid tumours.1

The coding system of the CRN allows discrimination of premalignant from invasive malignant tumours, using the 7th revision of the WHO's International Classification of Diseases (ICD-7). The extent of the malignancy was coded according guidelines of the CRN: local, regional, distant and unknown (unknown was defined as either unknown or not classifiable due to insufficient information). Information on treatment modality at the CRN is insufficient.

Cancer survivors (cases) of the present study

With the permission of the Norwegian Data Inspectorate, the database of the HUNT 2 person-identifiable questionnaire was merged with that of the CRN, identifying every person in HUNT 2 who had a diagnosis of invasive cancer. The linkage was performed in April 2002, with complete incidence data up to December 2001. Every person in HUNT 2 with a malignancy, diagnosed 5 years or more prior HUNT 2 was eligible for our study. If the individual had been diagnosed with more than 1 cancer diagnosis prior to 5 years of HUNT 2, only the first diagnosis and the date of the first diagnosis was considered. As the CRN does not provide sufficient information on relapse and treatment, these parameters were not considered in our study. The individual diagnoses resulted into 6 subgroups of major cancer types: testicular cancer survivors (TCSs), colorectal cancer survivors (ColCaSs), prostatic cancer survivors (ProCaSs), breast cancer survivors (BreastCaSs), gynaecological cancer survivors (cervical and ovarian cancer) (GynCaSs) and lymphoma/leukaemia survivors (LymLeuSs). Smaller groups of 5-year cancer survivors, such as pulmonary cancer, sarcomas etc. as well as larger groups for example skin cancer, were not considered for our report. The extent of disease (local, regional, distant and unknown) was coded as recommended at the CRN.

Controls

For each case, 3 age- and gender-matched individuals were randomly selected from those participating in the HUNT 2 survey, thus for each diagnostic cancer subgroup constructing separate subgroups without cancer (controls). Eligible noncancer participants had no registration at the CRN with either premalignant disease or invasive cancer prior the linkage of HUNT 2 data with the CRN in 2002. The noncancer population comprised 62,006 individuals from the HUNT 2 study.

Data source from the HUNT 2 survey

Our report is entirely based on the responses to a self-reported questionnaire. The HUNT 2 study's demographic data on gender, age, marital status and education were analysed together with the responses to the selected HUNT 2 questions considered relevant for the study (indicated in the Appendix). Poor health status was derived from the single question: “How would you describe your health at present?” The answers were dichotomised into “Poor” and “Well” (Appendix). We identified common health problems and differentiated between patient-experienced health complaints from physician-based adverse health conditions and self-reported unfavourable lifestyle parameters. Back pain, arthralgia, shortness of breath, palpitation, nausea, dyspepsia and diarrhoea, solely describing the patient's experience, were defined as health complaints. Hypertension, heart disease, cerebrovascular episodes, osteoporosis, other musculo-skeletal conditions, diabetes mellitus and thyroid disease, indicating a medical diagnosis, were defined as adverse health conditions. Smoking, obesity and physical inactivity were taken as expression of unfavourable lifestyle parameters. The level of physical activity was assessed by 1 ad hoc-designed question, which described 2 levels of physical activity (Appendix). The first level addressed low physical activities, such as walking, without sweating or breathlessness. The second level identified high physical activities that would lead to sweating and breathlessness. At both levels, respondents were asked to categorise how many hours per week they usually spent with these activities. In the final analyses, the responses were, as previously published by Thorsen et al.,17 recoded. “No physical activity at any level” or “low level physical activity less than 1 hr per week” was defined as “Physical inactivity”. All other responses qualified for “Physical activity”. Height and weight were used to calculate body mass index (BMI; kg/m²) and were measured with the participants wearing light clothes without shoes. Obesity was defined as BMI ≥ 30 kg/m.2

Social welfare benefits excluded participants receiving retirement pensions but included individuals receiving benefits under occupational rehabilitation, sickness benefits, disablement pensions, social security benefits and unemployment benefits (Appendix).

Statistical methods

Descriptive statistics [mean and standard deviation (S.D.)], chi square analyses and logistic regression analyses were established by SPSS 11.5. In bivariate categorical analyses with 5 or fewer expected observations in 1 cell, Fisher's exact test was used to test hypotheses of associations. A 2-tailed p-value less than or equal 0.05 was considered as statistically significant. Owing to missing values for some of the explanatory variables, we explored bivariate logistic regression models that included categories for missing values to retain all of the cases in the analysis and models excluding individuals with missing values. Both approaches gave similar results, and only the results of the first approach are presented here. In the 2 multivariate logistic regression analyses either poor health status or the use of health care service represented the dependent parameter with cancer survivorship as the independent parameter (Model 1). Both analyses were adjusted for health complaints, adverse health conditions as well as life-style parameters (Models 2, 3, 4 and 5). For the multivariate logistic regression analysis, the use of health care services was dichotomised and defined as those who had a consultation with either a general practitioner (GP) and/or a hospital physician within the 12 last months (1) and those with no such contact (0).

Results

Study characteristics

A total of 1,674 cancer survivors were identified with a malignant diagnosis 5 years or more prior the HUNT 2 survey, including all cancer diagnoses. Of these, approximately 50% were eligible for 1 of our diagnostic cancer subgroups, representing 815 cases: 59 TCSs, 175 ColCaSs, 87 ProCaSs, 258 BreastCaSs, 153 GynCaSs and 83 LymLeuSs. A total of 545 (67%) were women and 270 (33%) were men. Only 11 cases (1%) had been diagnosed when they were younger than 20 years. The majority of the childhood cancer survivors belonged to the LymLeuS-group (9 cancer survivors) (data not shown). The LymLeuSs comprised 41% survivors after the diagnosis of leukaemia and 59% survivors after their lymphoma diagnosis (data not shown). The distribution of cancer types among long-term cancer survivors in Nord-Trøndelag differs only slightly from that reported for the whole country of Norway: TCSs: HUNT 2: 4% vs. Norway: 4%, ColCaSs: HUNT 2: 10% vs. Norway: 12%, ProCaSs: HUNT 2: 5% vs. Norway: 5%, BreastCaSs: HUNT 2: 15% vs. Norway: 17%, GynCaSs: HUNT 2: 9% vs. Norway: 9%, LymLeuSs: HUNT 2: 5% vs. Norway: 6%.18

The ProCaSs were the oldest survivors (mean age: 77 years), and the youngest survivors were found in the TCSs-group (mean age: 53 years) (Table I). The cancer survivors had been diagnosed from 1954 to 1992, but the majority had been diagnosed later than 1980 (71%) (Fig. 1). The median interval between the individual's cancer diagnosis and participation in the HUNT 2 survey was 12 years (range: 5–42 years). The interval since diagnosis (observation time) was longer among the GynCaSs (mean 19 years) and TCSs (mean 18 years) than for the other 4 cancer subgroups. No statistically significant differences were found when comparing the categories of marital status or educational level for either of the 6 cancer subgroups with their matched controls (p>0.05). The majority of the cases had been treated for localised cancer (62%) (Table I). The BreastCaSs (27%) as well as the ColCaSs (43%) were most frequently represented in the groups of patients with regional disease, whereas the TCSs (10%) and GynCaSs (10%) represented the largest groups with distant disease.

Table I. Characteristics for Cancer Survivors (Cases) and Their Matched Controls

	Testicular	Colorectal	Prostate	Breast	Gynaecological	Lymphoma/Leukaemia
1 Mean years and standard deviation (S.D.) 2 No statistical significance was found between cancer survivors and their matched controls. 3 High school or university. 4 Not applicable.
N
Survivors	59	175	87	258	153	83
Controls	177	525	261	774	459	249
Age	53 ± 14.2	72 ± 9.7	77 ± 7.0	69 ± 11.8	64 ± 12.3	60 ± 17.3
Gender (% female)	0	54	0	100	100	48
Married2 (%)
Survivors	68	62	69	55	53	64
Controls	68	61	70	56	60	55
Education2,3
Survivors	27	22	20	26	18	29
Controls	31	27	22	23	23	23
Time since diagnosis1	18 ± 9.4	11 ± 5.9	9 ± 3.8	14 ± 8.3	19 ± 9.4	13 ± 7.5
Extent of disease:
Local n (%)	44 (75)	95 (54)	76 (87)	117 (45)	123 (80)	NA4
Regional n (%)	9 (15)	75 (43)	1 (1)	70 (27)	15 (10)	NA4
Distant n (%)	6 (10)	4 (2)	7 (8)	2 (1)	15 (10)	NA4
Unknown n (%)	0	1 (1)	3 (4)	69 (27)	0	NA4

Figure 1. The cumulative percentage of cancer survivors in relation to the year of diagnosis.

General health status

More cancer survivors perceived a poor health status than their matched controls (49% vs. 42%: p=0.001) (Fig. 2). When evaluating the individual subgroups of cases, in particular ProCaSs (64%) and LymLeuSs (52%) reported poor health more often than their controls (p=0.021 resp. p=0.002).

Figure 2. “How would you describe your health at present?” Percentage of cancer survivors and controls with perceived poor health.

The duration of survival after the cancer diagnosis was not associated with a cancer survivor's perception of poor health (data not shown).

Use of health care services and social welfare benefits

A total of 70% all cases had seen a GP in the last 12 months before the HUNT 2 survey in comparison to 66% of the controls (Table II). Significantly more cases than the controls had used the specialist health care services (43% vs. 28% resp.) (Table II). Even if the analysis was restricted to cases who had lived with their diagnosis for 10 years or longer, the difference in use of specialist health care services remained significant but the difference in the use of GPs disappeared (data not shown). These significant differences, as to the use of specialist physicians and GPs, are particularly obvious for patients with the diagnosis of prostate cancer. LymLeuSs were those who most often visited a specialist physician (55%). There were no statistically significant differences in the use of physical therapy or the use of complementary and alternative medicine between cases and the controls. More cases than controls received social welfare benefits (14% vs. 11%), with particularly large differences for “younger” cancer survivors (TCSs: 17%;; LymLeuSs: 19%;; GynCaSs: 20%;).

Table II. Cancer Survivors' Uses of Health Services within the Last 12 Months Before Hunt 2 as well as Social Welfare Benefits in Comparison to the Matched Controls

	Health care service	Cases (%)	Controls (%)	OR	95% C.I.	p
1 General practitioner. 2 Specialist physician in a hospital. 3 Incl chiropractic therapy. 4 Complementary and alternative medicine. 5 Social welfare benefits.
All:	GP1	70	66	1.19	1.00–1.41	0.05
	Specialist2	43	28	1.96	1.66–2.31	<0.001
	Physical therapy3	14	14	1.03	0.82–1.29	0.82
	CAM4	6	6	0.90	0.64–1.26	0.53
	SWB5	14	11	1.39	1.10–1.76	0.006
Testicular:	GP1	54	51	1.15	0.64–2.07	0.65
	Specialist2	41	27	1.84	1.00–3.41	0.05
	Physical therapy3	17	11	1.70	0.74–3.89	0.21
	CAM4	5	3	1.84	0.42–7.96	0.41
	SWB5	17	10	1.80	0.78–4.16	0.17
Colorectal:	GP1	75	67	1.45	0.98–2.14	0.06
	Specialist2	42	29	1.77	1.24–2.53	0.002
	Physical therapy3	13	12	1.09	0.65–1.84	0.74
	CAM4	3	7	0.39	0.15–1.00	0.05
	SWB5	9	9	1.05	0.58–1.90	0.88
Prostate:	GP1	83	72	1.90	1.02–3.52	0.04
	Specialist2	41	27	1.93	1.16–3.20	0.01
	Physical therapy3	10	11	0.92	0.42–2.03	0.84
	CAM4	5	7	0.69	0.23–2.11	0.52
	SWB5	6	2	2.59	0.77–8.71	0.12
Breast:	GP1	74	68	1.34	0.98–1.84	0.07
	Specialist2	49	27	2.59	1.93–3.46	<0.001
	Physical therapy3	18	16	1.20	0.83–1.74	0.33
	CAM4	8	7	1.19	0.70–2.03	0.52
	SWB5	15	12	1.27	0.85–1.90	0.24
Gynaecological	GP1	58	67	0.70	0.48–1.01	0.06
	Specialist2	29	29	1.01	0.68–1.51	0.96
	Physical therapy3	11	17	0.59	0.33–1.04	0.07
	CAM4	7	8	0.80	0.39–1.65	0.54
	SWB5	20	14	1.51	0.93–2.43	0.09
Lymphoma/Leukaemia:	GP1	66	63	1.17	0.69–1.98	0.55
	Specialist2	55	27	3.39	2.02–5.66	<0.001
	Physical therapy3	16	15	1.06	0.54–2.11	0.86
	CAM4	6	4	1.71	0.56–5.25	0.35
	SWB5	19	13	1.62	0.84–3.13	0.15

Health complaints

Seventy percent of all cases reported at least 1 health complaint of those included in our study compared to 66% of the controls (p=0.019) (Table IIIa). Diarrhoea, as a specific health complaint, was experienced more often by all cases than by controls. Some differences became significant only in some of the cancer subgroups (Table IIIb).

Table IIIa. Health Complaints in all Cases Compared to their Matched Controls

Health problems	Cases (%)	Controls (%)	OR	95% C.I.	p
1 Including back, thorax, and neck as well as shoulder pain. 2 Including any pain reported to hip, knee, ankle, feet, elbow, hands and wrists. 3 ≥1 health complaint defined as one or more of the following: backpain, arthralgia, shortness of breath, palpitation, nausea, dyspepsia and diarrhoea.
Backpain1	47	46	1.07	0.91–1.25	0.41
Arthralgia2	45	44	1.05	0.90–1.23	0.56
Shortness of breath	7	8	0.95	0.70–1.29	0.74
Palpitation	17	19	0.96	0.78–1.20	0.74
Nausea	11	9	1.23	0.94–1.61	0.13
Dyspepsia	29	27	1.12	0.93–1.35	0.23
Diarrhoea	17	10	1.80	1.43–2.27	<0.001
≥1 health complaint3	70	66	1.23	1.03–1.46	0.019

Table IIIb. Health Complaints in Cancer Survivors According to Cancer Subgroups Compared to Matched Control Groups

	Health problems	Cases (%)	Controls (%)	OR	95% C.I.	p
1 Including back, thorax and neck as well as shoulder pains. 2 Including any pain reported to hip, knee, ankle, feet, elbow, hands and wrists.
Testicular	Backpain1	48	41	1.32	0.73–2.38	0.36
	Arthralgia2	36	32	1.16	0.63–2.16	0.63
	Shortness of breath	12	7	1.95	0.72–5.27	0.19
	Palpitation	14	12	1.16	0.48–2.80	0.74
	Nausea	14	7	2.11	0.82–5.43	0.12
	Dyspepsia	36	23	2.02	1.05–3.89	0.036
	Diarrhoea	17	12	1.66	0.72–3.82	0.23
Colorectal	Backpain1	47	44	1.11	0.78–1.56	0.57
	Arthralgia2	42	43	0.96	0.68–1.36	0.82
	Shortness of breath	6	10	0.66	0.33–1.30	0.23
	Palpitation	17	18	0.95	0.59–1.53	0.84
	Nausea	11	8	1.42	0.80–2.55	0.24
	Dyspepsia	31	27	1.25	0.84–1.85	0.28
	Diarrhoea	16	10	1.69	1.01–2.81	0.04
Prostate	Backpain1	48	44	1.19	0.73–1.93	0.49
	Arthralgia2	56	39	1.98	1.21–3.23	0.006
	Shortness of breath	14	10	1.59	0.75–3.38	0.23
	Palpitation	15	13	1.35	0.66–2.77	0.41
	Nausea	10	9	1.34	0.58–3.07	0.49
	Dyspepsia	37	34	1.13	0.62–2.05	0.70
	Diarrhoea	21	7	3.51	1.71–7.25	0.001
Breast	Backpain1	50	47	1.12	0.84–1.48	0.45
	Arthralgia2	48	49	0.95	0.72–1.27	0.75
	Shortness of breath	6	7	0.88	0.48–1.60	0.67
	Palpitation	19	20	1.01	0.70–1.48	0.94
	Nausea	9	11	0.82	0.50–1.36	0.45
	Dyspepsia	26	30	0.82	0.59–1.15	0.25
	Diarrhoea	13	11	1.26	0.81–1.96	0.31
Gynaecological	Backpain1	46	48	0.91	0.63–1.31	0.61
	Arthralgia2	46	48	0.93	0.65–1.34	0.71
	Shortness of breath	3	8	0.44	0.17–1.16	0.10
	Palpitation	16	23	0.73	0.44–1.21	0.22
	Nausea	14	15	0.98	0.52–1.82	0.94
	Dyspepsia	28	28	0.98	0.64–1.50	0.92
	Diarrhoea	23	11	2.50	1.52–4.12	<0.001
Lymphoma/Leukaemia	Backpain1	42	44	0.92	0.56–1.52	0.75
	Arthralgia2	37	37	1.02	0.61–1.70	0.95
	Shortness of breath	11	7	1.55	0.66–3.67	0.32
	Palpitation	22	22	0.98	0.51–1.89	0.96
	Nausea	17	5	3.05	1.36–6.86	0.007
	Dyspepsia	36	20	1.93	1.10–3.38	0.02
	Diarrhoea	21	12	1.61	0.82–3.14	0.16

Significantly more TCSs reported dyspepsia than their control group (36 vs. 23%; p=0.036) (Table IIIb). The cases in the lymphoma/leukaemia survivor group also reported of more dyspepsia than their controls (36% vs. 20%; p=0.02) (Table IIIb). LymLeuSs were 3 times more likely to experience nausea than their controls [OR 3.05 (95% C.I.: 1.36–6.86)]. Three groups of cancer survivors complained significantly more often of diarrhoea: ColCaSs (16% vs.10%; p=0.04), ProCaSs (21% vs. 7%; p=0.001) and GynCaSs (23% vs. 11%;; p<0.001) (Table IIIb).

A total of 56% of the ProCaSs complained of arthralgia compared to 39% of their matched controls (p=0.006).

Cancer survivors with more than 12-year survival did not report more health complaints than those with a shorter survival did (70% in both subgroups, data not shown).

Adverse health conditions

Cancer survivors displayed marginally more adverse health conditions than the controls, but the differences did not reach statistical significance except for osteoporosis (Table IVa). The prevalence of osteoporosis was especially high among TCSs, ProCSs and GynCaSs (Table IVb). Hypertension was significantly more frequent among the ProCaSs compared to the controls [OR 1.79 (95% C.I.: 1.04–2.97)] (Table IVb). Any disease to the thyroid gland was reported 4 to 5 times more often among TCSs and LymLeuSs in comparison to their controls (p<0.005). Even though not reaching the level of statistical significance the prevalence of heart disease was higher among TCSs and LymLeuSs than their controls. Significantly more GynCaSs reported of cerebrovascular episodes than their controls (8% vs. 4%; p=0.038). Cerebrovascular episodes were also more frequently reported by the ProCaSs as well as LymLeuSs in comparison to their matched controls.

Table IVa. Adverse Health Conditions in all Cases Compared to their Matched Controls

Health conditions	Cases (%)	Controls (%)	OR	95% C.I.	p
1 ≥1 adverse health condition defined as one or more of the following: hypertension, heart disease, cerebrovascular episode, other musculoskeletal disease, diabetes mellitus and thyroid disease.
Hypertension	28	27	1.02	0.86–1.22	0.81
Heart disease	17	16	1.12	0.91–1.38	0.30
Cerebrovascular	5	4	1.25	0.87–1.79	0.24
Osteoporosis	6	4	1.51	1.07–2.12	0.02
Other musculoskeletal diseases	24	23	1.05	0.87–1.27	0.60
Diabetes mellitus	7	6	1.11	0.81–1.54	0.52
Thyroid disease	10	9	1.19	0.91–1.55	0.20
≥ 1 adverse health condition1	79	78	1.09	0.90–1.32	0.39

Table IVb. Adverse Health Conditions in Cases Compared to their Matched Control Groups

	Health conditions	Cases (%)	Controls (%)	OR	95% C.I.	p
Testicular	Hypertension	17	14	1.24	0.56–2.76	0.60
	Heart disease	15	10	1.69	0.71–4.04	0.23
	Cerebrovascular	2	3	0.59	0.07–5.18	0.64
	Osteoporosis	3	0	—	—	0.032
	Other musculo-skeletal diseases	12	7	1.70	0.64–4.48	0.28
	Diabetes mellitus	2	3	0.70	0.19–14.82	0.63
	Thyroid disease	12	3	4.63	1.41–15.20	0.006
Colorectal	Hypertension	29	30	0.95	0.65–1.39	0.80
	Heart disease	25	20	1.27	0.85–1.91	0.24
	Cerebrovascular	3	5	0.52	0.20–1.37	0.19
	Osteoporosis	7	6	1.15	0.58–2.31	0.69
	Other musculo-skeletal diseases	26	24	1.12	0.76–1.65	0.58
	Diabetes mellitus	6	7	0.82	0.41–1.63	0.56
	Thyroid disease	7	9	0.73	0.38–1.41	0.35
Prostate	Hypertension	36	24	1.76	1.04–2.97	0.034
	Heart disease	31	30	1.08	0.64–1.82	0.79
	Cerebrovascular	11	5	2.21	0.91–5.37	0.08
	Osteoporosis	3	0.4	9.88	1.01–96.58	0.049
	Other musculo-skeletal diseases	22	15	1.59	0.86–2.93	0.14
	Diabetes mellitus	8	8	1.01	0.41–2.46	0.99
	Thyroid disease	2	1	2.02	0.33–12.31	0.44
Breast	Hypertension	29	30	0.93	0.68–1.27	0.66
	Heart disease	14	14	0.94	0.62–1.41	0.76
	Cerebrovascular	4	5	0.85	0.42–1.74	0.66
	Osteoporosis	6	6	0.97	0.53–1.78	0.93
	Other musculo-skeletal diseases	31	30	1.08	0.80–1.47	0.61
	Diabetes mellitus	9	6	1.58	0.94–2.67	0.085
	Thyroid disease	12	12	1.04	0.67–1.60	0.84
Gynaecological	Hypertension	23	28	0.78	0.51–1.19	0.25
	Heart disease	8	10	0.75	0.39–1.45	0.39
	Cerebrovascular	8	4	2.25	1.05–4.81	0.038
	Osteoporosis	10	5	2.25	1.12–4.51	0.022
	Other musculo-skeletal diseases	21	26	0.76	0.49–1.19	0.23
	Diabetes mellitus	5	4	1.44	0.61–3.41	0.41
	Thyroid disease	13	14	0.95	0.55–1.62	0.84
Lymphoma/Leukaemia	Hypertension	28	23	1.34	0.76–2.36	0.32
	Heart disease	18	10	1.89	0.95–3.78	0.071
	Cerebrovascular	7	3	2.71	0.88–8.30	0.082
	Osteoporosis	6	3	2.21	0.68–7.18	0.19
	Other musculo-skeletal diseases	13	16	0.82	0.40–1.69	0.60
	Diabetes mellitus	5	8	0.62	0.20–1.87	0.40
	Thyroid disease	16	4	4.95	2.03–12.07	0.0006

There was no statistically significant difference between those cancer survivors with more than 12-year survival and those with less than 12-year survival regarding number of adverse health conditions (data not shown).

Life style parameters

Marginally more cases reported at least 1 unfavourable life style parameter though the difference did not reach statistical significance (Table V). More cancer survivors were defined as physically inactive, without significant difference for the whole group (Table V). TCSs and GynCaSs were significantly more often defined as physically inactive, whereas LymLeuSs were more physically active than their controls (Table V). Compared to those without a history of cancer, more TCSs and GynCaSs were identified as smokers, whereas smoking was less common among LymLeuSs (Table V).

Table V. Unfavourable Lifestyle Parameters in Cases Compared to their Matched Control Groups

	Life style parameters	Cases (%)	Controls (%)	OR	95% C.I.	p
1 BMI ≥30 kg/m^2. 2 Physical inactivity. 3 ≥1 lifestyle parameters defined as one or more of the following: smoking, obesity and physical inactivity.
All	Smoking	19	20	0.96	0.78–1.16	0.60
	Obesity1	22	22	0.96	0.80–1.17	0.70
	Inactivity2	29	27	1.08	0.88–1.32	0.45
	≥1 life style parameter3	53	52	1.02	0.87–1.20	0.81
Testicular	Smoking	34	28	1.34	0.71–2.52	0.36
	Obesity1	14	15	0.87	0.37–2.04	0.75
	Inactivity2	28	15	2.12	1.01–4.42	0.046
Colorectal	Smoking	14	17	0.78	0.49–1.27	0.32
	Obesity1	23	25	0.88	0.59–1.32	0.54
	Inactivity2	25	29	0.82	0.52–1.28	0.37
Prostate	Smoking	22	22	1.02	0.57–1.84	0.94
	Obesity1	15	17	0.89	0.45–1.75	0.74
	Inactivity2	34	28	1.32	0.73–2.38	0.36
Breast	Smoking	16	17	0.94	0.64–1.37	0.74
	Obesity1	26	23	1.14	0.82–1.58	0.42
	Inactivity2	31	32	0.97	0.68–1.40	0.89
Gynaecological	Smoking	27	20	1.44	0.94–2.20	0.091
	Obesity1	18	25	0.68	0.43–1.08	0.099
	Inactivity2	36	24	1.78	1.13–2.80	0.012
Lymphoma/Leukaemia	Smoking	13	29	0.38	0.19–0.75	0.005
	Obesity1	27	21	1.37	0.77–2.43	0.29
	Inactivity2	16	26	0.51	0.25–1.05	0.069

Multivariate logistic regression analyses

Being a cancer survivor increased the risk of experiencing poor health [OR: 1.32; (C.I.: 1.13–1.55)] (Table VI). Health complaints, adverse health conditions and lifestyle parameters did not influence the risk of poor health among the cases. Both health complaints and adverse health conditions had a strong association to poor health, whereas lifestyle parameters had no effect on self-reported poor health. Similar results were found when investigating the effect history of cancer on the use of health care services (OR: 1.37; C.I.: 1.13–1.65) (Table 7). Health complaints, adverse health conditions and life style did not affect the cancer survivors' use of health care services compared to the controls.

Table VI. Effect of History of Cancer on Poor Health Status, Adjusted for Having ≥1 Health Complaints, ≥1 Adverse Health Conditions and ≥1 Lifestyle Parameters

Logistic regression model	OR1	95% C.I.	p
1 OR obtained from multivariate logistic regression analysis. 2 ≥1 health complaint defined as one or more of the following: backpain, arthralgia, shortness of breath, palpitation, nausea, dyspepsia and diarrhoea. 3 ≥1 adverse health condition defined as one or more of the following: hypertension, heart disease, cerebrovascular episode, other musculoskeletal disease, diabetes mellitus and thyroid disease. 4 ≥1 lifestyle parameters defined as one or more of the following: smoking, obesity and physical inactivity
Model 1
History of cancer	1.32	1.13–1.55	0.001
Model 2
History of cancer	1.27	1.07–1.51	0.006
≥1 health complaints2	5.89	4.92–7.04	<0.001
Model 3
History of cancer	1.32	1.12–1.55	0.001
≥1 adverse health conditions3	2.86	2.37–3.45	<0.001
Model 4
History of cancer	1.32	1.13–1.55	0.001
≥1 lifestyle parameters4	1.08	0.94–1.25	0.26
Model 5
History of cancer	1.27	1.06–1.51	0.008
≥1 health complaints2	5.85	4.88–7.02	<0.001
≥1 adverse health conditions3	2.83	2.32–3.44	<0.001
≥1 lifestyle parameters4	1.03	0.88–1.19	0.76

Table VII. Effect of History of Cancer on Use of the Health Care Services, Adjusted for Having ≥1 Health Complaints, ≥1 Adverse Health Condition and ≥1 Lifestyle Parameters

Logistic regression model	OR1	95% C.I.	p
1 OR obtained from multivariate logistic regression analysis. 2 ≥1 health complaint defined as one or more of the following: backpain, arthralgia, shortness of breath, palpitation, nausea, dyspepsia and diarrhoea. 3 ≥1 adverse health condition defined as one or more of the following: hypertension, heart disease, cerebrovascular episode, other musculoskeletal disease, diabetes mellitus and thyroid disease. 4 ≥1 lifestyle parameters defined as one or more of the following: smoking, obesity and physical inactivity.
Model 1
History of cancer	1.36	1.13–1.63	0.001
Model 2
History of cancer	1.33	1.10–1.60	0.001
≥1 health complaints2	1.77	1.5–2.07	<0.001
Model 3
History of cancer	1.35	1.12–1.63	0.001
≥1 adverse health conditions3	1.30	1.09–1.56	0.004
Model 4
History of cancer	1.36	1.13–1.63	0.001
≥1 lifestyle parameters4	0.91	0.78–1.63	0.23
Model 5
History of cancer	1.32	1.10–1.60	0.003
≥1 health complaints2	1.75	1.49–2.05	<0.001
≥1 adverse health conditions3	1.25	1.04–1.50	0.019
≥1 lifestyle parameters4	0.89	0.76–1.04	0.15

Discussion

Our analyses reveal that more long-term cancer survivors report of poor health and have more frequent contact with health care services (general GPs and specialist physicians) than their controls. These differences cannot be explained by the prevalence of common somatic health problems (health complaints and adverse health conditions), even though health problems quantitatively differ from those perceived by the noncancer individuals. Contrary to our assumptions long-term cancer survivors do not display a more health promoting lifestyle than the controls. Not surprisingly, a prior cancer diagnosis is associated with a more frequent use of social welfare benefits.

When planning our investigation, we had expected that the cancer survivors more often than their controls would complain about common health problems. This assumption was confirmed to a lesser degree than expected, with statistically significant differences only for subgroups of cancer survivors. There are at least 3 explanations for our results.

First, only the “healthiest” cancer survivors may have responded to the HUNT 2 invitation. This suggestion is supported by the compliance analysis of the HUNT 2 study: those individuals with major health disorders and frequent contact with their GPs tended not to attend.15, 19 Long-term cancer survivors may presumably be over-represented in this group, in particular those with common health problems. There are, however, no available data on the number of cancer survivors in the nonparticipant study, as the Norwegian Data Inspectorate did not allow such relevant analyses to confirm this suggestion. Another aspect is, of course, that cancer survivors with “major” health problems related to their cancer diagnosis or treatment may have died prior their possible participation to the HUNT 2 study and are thus not represented in our cancer population.

Second, only the most common somatic health problems have been addressed in our study, leaving out more cancer specific late effects such as cognitive function, lymphoedema, incontinence etc. Common somatic health problems may indeed be equally prevalent among controls and long-term cancer survivors. It is important to remember that our cases were likely to have had localised or locoregional cancer. They had received less toxic and less intensive treatment, which probably cannot be expected to result in severe long-term health problems.

Third, perceived poor health and the need of frequent contact with the health care services as observed in our cases could be associated with psychosocial disorders more than with somatic health problems. It is well known that anxiety and fatigue remain a health problem in cancer survivors even many years after successful treatment.20, 21, 22, 23 As the HUNT 2 study also addressed these aspects, a future analysis will assess these dimensions for cancer survivors.

Significant differences were only found for diarrhoea and osteoporosis between the whole group of cases and controls, admittedly based on small numbers. These health problems are most probably a consequence of cancer treatment, a suggestion that is supported by the analysis of the different diagnostic cancer subgroups.

Gastrointestinal problems are frequently described as a long-term complication after cancer treatment, especially after abdominal and/or pelvic radiation.24, 25, 26 In our study, the survivors of testicular cancer, colorectal cancer, gynaecological cancer, prostate cancer and lymphoma/leukaemia reported significantly more gastrointestinal problems than the controls. Patients with these cancer types are frequently treated with infra diaphragmatic radiotherapy. However, radiotherapy is not the only cancer treatment modality associated with diarrhoea; subtotal colectomy is also known to cause this problem in survivors after colorectal cancer due to short bowel.27

Osteoporosis is recognised as a possible long-term consequence of cancer treatment28 and the increased prevalence of osteoporosis in our cancer population compares favourably with these authors' review. In cancer patients, several possibilities for the development of osteoporosis after treatment have to be discussed. Normal bone tissue remodelling and maintenance are partly dependent on gonadal hormones.29 Gonadal dysfunction, in particularly early menopause and androgen deprivation, is a frequent finding after treatment for testicular cancer, gynaecological cancer and prostate cancer.8, 28 Furthermore, radiotherapy may damage the structure of bone and cause osteoporosis.25, 30, 31 Finally, the development of osteoporosis is associated with smoking and physical inactivity.32 In our study, the survivors of gynaecological cancer smoke more and are less physical active, contributing to the increased risk of osteoporosis among gynaecological cancer survivors. Patients treated for breast cancer with anti-oestrogen treatment are at lesser risk to develop osteoporosis.33, 34 Most of our breast cancer survivors have probably been treated with tamoxifen as this drug had been introduced in Norway as standard treatment for breast cancer in the 1980s.

During recent years, the oncological community has become aware of long-term cardiovascular risks associated with cancer and its treatment. Even though not reaching the level of statistical significance, we think it is of clinical relevance that survivors after testicular cancer and lymphoma and leukaemia displayed an increased prevalence of cardiovascular disease than the controls. The literature reports emphasise the increased cardiovascular risk in these patients.35, 36, 37, 38, 39 In agreement with the literature, we found a high prevalence of hypertension among the survivors of prostate cancer. It has not been finally determined whether androgen deprivation in prostate cancer patients could lead to hypertension40 or whether hypertension might be related to the aetiology of prostate cancer.41 On the other hand, other authors have shown an inverse association of cancer prostate and the use of antihypertensive medication.42 The increased number of survivors after gynaecological cancer with a cerebrovascular episodes may be related to an early menopause or to lifestyle parameters.43, 44, 45

In agreement with our findings, thyroid disorders are often described among survivors of lymphoma, probably due to mantle field irradiation.46, 47 Surprisingly, a higher prevalence of thyroid disease among TCSs in comparison to the controls was found. In Norway in the 1980s, radiation to the left or both supraclavicular fossae was standard when treating seminomatous testicular cancer with regional lymph node metastases,48 which may explain the results. Furthermore, highly elevated beta human chorionic gonadotrophin (HCG) levels, not infrequently observed in patients with a new testicular cancer, can also induce hyperthyroidism.49, 50, 51 High HCG levels usually normalise with treatment, but hypothetically these transient hormone changes might disturb the hypothalamic-pituitary-thyroid hormone axis at a more permanent level.

Not surprisingly and in agreement with other studies, more cases than controls receive social welfare benefits. In the paper of Verbeck et al., only 67% of the cancer survivors returned to work after 12 months.52. In another study, approximately 17% of the cancer survivors stated inability to work.5

In agreement with our results, Hewitt et al.5 had shown that twice as many cancer survivors compared to those without a history of cancer had visited a doctor in the past year. The increased use of health care services and reports of poor health among cancer survivors could not be explained by health problems and/or unfavourable life style parameters. These observations are neither explained by specific Norwegian guidelines for follow-up of long-term cancer survivors, as such recommendations do not exist. Our results indicate that there most probably exist other reasons for perceived poor health and frequent use of health care services. Hewitt et al.5 found more visits among those cancer survivors, who reported of functional limitations and/or psychological problems. We have not considered any psychological problems or cancer-related health-issues such as cognitive problems, sleeping disorders, sexuality, urinary problems, fear of recurrence or fatigue in our study. All these variables, in particular anxiety of relapse,53 are plausible explanatory factors for poor health and the increased use of health care services.

Surprisingly, even though the cancer survivors of our study seek more medical help, they do not live a “healthier” life. This could indicate that the main focus of a medical consultation is put on the patient's prior cancer diagnosis. It remains unclear if health prevention measures such as recommended smoking cessation regularly provided to Norwegian cancer survivors or if supplied advises are neglected as suggested by Ko et al.54 On the other hand, we have no data on unfavourably life style parameters prior the cancer diagnosis; it is therefore possible that our cases have made favourable life style changes.14 Our analyses on physical activity indicate that only a few more cases than controls were physical inactive however, with significant differences in some subgroups (TCSs and GynCaSs). Analyses of self-reported physical activity are vulnerable for information bias, and the validation of the measure of self-reported physical activity is a considerable problem. The use of a single dichotomised question (physically inactive and physically active) has been shown to be more valid than multiple response categories.55, 56 Though our final 2-level categorisation of physical activity is not completely comparable to the nowadays-accepted recommendations of at least 3 hr of physical activity a week, we consider our approach as a reasonable utilisation of the available information.

The population of our study of cases and controls are homogenous with regard to their educational level as a measure for socioeconomic status. As socioeconomic level is associated with health, larger socioeconomic differences between cases and controls were expected. Educational level is only 1 possibility to display socioeconomic status.57 We can therefore not rule out socioeconomic health differences between cases and controls if income had been compared. The educational level in the HUNT 2 population is somewhat lower than in Norway as a whole (22% vs. 18%) and the reason for this is unclear,57, 58 but as the differences are small, we still consider our study to be applicable for Norway as a whole.

The distribution of cancer survivors among the subgroups in our study differs slightly from those reported from Norway as a whole, though the differences are so small that we consider it to be representative for Norway and most probably internationally. According to the CRN, the incidence rates of cancer are lower in Nord-Trøndelag than in other areas of Norway, which in part could explain our result. Differences as to cancer survivors in other countries may result from the fact that no large public screening programs were active in Norway before the 1990s. More often than this is the case among our cancer survivors, such programs detect early breast or prostate cancer with excellent prognosis and long-term survival.

The strength of our study is that the cancer survivors and the controls are from the same geographically population and that the HUNT 2 study had a high compliance rate. In comparison to our study, studies carried out in a clinical setting might “over-report” health problems in cancer survivors, as both the investigator as well as the subject are usually informed of the aim of the study.

Our study had 3 significant limitations. First, we had no access on information of treatment modality, but even if such information had been available, our results would still only be limited applicable to generations of long-term cancer survivors after subsequent or today's cancer therapy. Second, we did not have the possibility to validate the answers of the questionnaires by review of medical records of the local hospital or family doctor. On the other hand, there is no reason to believe that the responses from the cases would significantly differ from the responses of the controls. Third, some of the diagnostic cancer subgroups (TCSs and LymLeuSs) are rather small, so small that type II errors cannot be excluded.

In conclusion, the history of cancer increases the likelihood of experiencing poor health, and an increased use of health care service compared to the noncancer population without any plausible somatic explanatory factors. Further studies are therefore needed to explore other reasons for these findings such as psychosocial factors. We also recommend the physicians to be more aware of the necessity to discuss lifestyle behaviour with the cancer survivors.

Acknowledgements

The Nord-Trøndelag Health Study (The HUNT Study) is a collaboration between the HUNT Research Centre, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Verdal, The Norwegian Institute of Public Health, The National Screening Service of Norway and the Nord-Trøndelag County Council.

Appendix

HUNT questions related to health, use of health care service and use of social welfare benefits

• 1
  How would you describe your health at present? Poor = Poor health status/Not very well = Poor health status/Well = Good health status/Extremely well = Good health status.
• 2
  Do you have or have you had cardiovascular disease (acute myocardial infarction and/or angina pectoris)/cerebrovascular episode (haemorrhage and/or embolus)/diabetes mellitus? Yes/No
• 3
  Have you been diagnosed with hyperthyroidism? Have you been diagnosed with hypothyroidism? Have you ever been told that you have or have had goitre? Have you ever had any disease in the thyroid gland? Are you or have you been anti-thyroid drugs (carbimazole)? Have you ever had an operation on your thyroid gland? Have you ever been treated with radioactive iodine? Yes/No
• 4
  Have you in the last 12 months suffered from diarrhoea/nausea/dyspepsia? Never = No/Sometimes = Yes/Often = Yes.
• 5
  Do you use any antihypertensive drugs? Now = Yes/Previous but not now = Yes/Never used = No.
• 6
  Has a doctor ever told you that you have osteoporosis? Yes/No
• 7
  Has a doctor ever told you that you have fibromyalgia/arthritis? Yes/No (other musculoskeletal diseases).
• 8
  Have you experienced any stiffness or pain to: ankles/feet/knees/hips/wrists/hands/elbows? Yes/No
• 9
  Have you experienced any stiffness or pain to: back/shoulders/thorax/neck? Yes/No
• 10
  Anyone who responded “Yes” to any of the following 3 questions was classified as smoker: Do you smoke cigarettes on a daily basis? Do you smoke cigars on a daily basis? Do you smoke a pipe on a daily basis?
• 11
  What has your physical activity been like during the last year in your leisure time? (Image a weekly average within a year, walking to work counts as leisure time.) Physical inactive: no activity or less than 1 hr a week of low-level activity (defined as not sweaty or breathless during the activity). Physical active: more than 1 hr per week of low-level activity or high-level activity (defined as sweaty and breathless during the activity).
• 12
  Have you in the last 12 months been to general practitioner (GP)/specialist hospital physician/physical therapist or chiropractor/homeopath or other practitioner of alternative medicine? Yes/No
• 13
  A social welfare recipient was defined as anyone who replied yes to any of the following question: Do you receive benefits under occupational rehabilitation/sickness benefit/disablement pensions/unemployment benefits/social security benefits? Retirement pension was not included.