Socioeconomic Position, Multimorbidity and Mortality in a Population Cohort: The HUNT Study

Multimorbidity and socioeconomic position are independently associated with mortality. We investigated the association of occupational position and several multimorbidity measures with all-cause mortality. A cohort of people aged 35 to 75 years who participated in the Trøndelag Health Study in 2006–2008 and had occupational data was linked to the Norwegian National Population Registry for all-cause mortality from study entry until 1 February 2019. Logistic regression models for each occupational group were used to analyze associations between the number of conditions and 10-year risk of death. Cox regression models were used to examine associations between combinations of multimorbidity, occupational position, and mortality. Analyses were conducted for men and women. Included were 31,132 adults (16,950 women (54.4%)); occupational groups: high, 7501 (24.1%); low, 15,261 (49.0%)). Increased mortality was associated with lower occupational group, more chronic conditions, and all multimorbidity measures. The joint impact of occupational group and multimorbidity on mortality was greater in men than women. All multimorbidity measures are strongly associated with mortality, with varying occupational gradients. Social differences in multimorbidity are a public health challenge and necessitate consideration in health care. Men in lower occupational groups seem to be a particularly vulnerable group.


"20 Diseases"
Cluster text: "Have you had or do you have any of the following; Myocardial infarction; 4 5 angina pectoris; 5 6 heart failure; 4 other heart disease; stroke 4 5 or brain haemorrhage; kidney disease; 5 asthma; 5 chronic bronchitis, emphysema or chronic obstructive pulmonary disease; diabetes; 4 5 psoriasis; 7 eczema on hands; 8 9 cancer; 5 10 epilepsy; 11 rheumatoid arthritis; 5 12 ankylosing spondylitis; 5 12 sarcoidosis; osteoporosis; 5 13 fibromyalgia 5 and osteoarthritis 5 ?" Separate tick boxes for each diagnosis: Yes, no. For each diagnosis, included were those who affirmed to have or have had the diagnosis. Chronicity is assumed based on medical knowledge and clinical experience.

Headache 5
Seven questions in one block. Question 1: "Have you had headaches in the last year?" Yes/no.

Migraine without aura 14
Of those who affirmed headache last year, migraine without aura was constructed from three of seven questions: 1. "What is the average strength of your headaches?" 1=Mild, 2=Moderate, 3=Strong.
• Pain on one side of the head?" Yes, no.
• Hypersensitivity to light and/or noise?" Yes, no.
Included with migraine: were those who affirmed to headache lasting 0 to 72 hours and at least two of four characteristics (pulsating quality, unilateral location, moderate/severe pain intensity, or aggravation by physical activity) and during headache having at least one of two accompanying symptoms (nausea and/or vomiting or increased sensitivity to light and/or noise). 14 Chronicity is assumed based on medical knowledge and clinical experience.

Chronic headache14
Of those who affirmed headache last year, chronic headache was constructed from two of seven questions: 1. "If yes (headache in the last year): What type of headache? Migraine, other." The HUNT Databank created two variables with range 1: 1) migraine and 2) other headache.
Recoded to dichotomous variable, where 1= More than 14 days.
Included as case with chronic headache were those reporting "other" type of headache and an average frequency of more than 14 days per month.

Chronicity is assumed based on medical knowledge and clinical experience.Pain 5
Index question: "In the last year, have you had pain or stiffness in muscles or joints that has lasted at least 3 consecutive months?" Yes, no.
The follow-up question "If yes: Where have you had this pain or stiffness?" was combined with a figure with arrows and tick boxes at nine locations (neck, upper back, lower back, shoulder, elbow, hand, hip, knee and ankle/foot).

Chronic widespread pain 15
Dichotomous variables were made for each major body area: 1) Trunk (neck, upper and lower back), 2) Upper limb (shoulder, elbow, hand), and 3) Lower limb (hip, knee, foot/ankle), where 1=At least one painful location. A sum (row total) score variable was made for the major body areas and dichotomized, where 1=3, that is one pain in each major body area.
Of those who affirmed to pain or stiffness that has lasted more than three consecutive months, chronic widespread pain was defined as pain at more than three sites in all major body areas (trunk, upper and lower limbs) for more than three months in the last year.

Chronic, local pain
Of those who affirmed to pain or stiffness that has lasted more than three consecutive months, chronic, local pain was defined as pain in the neck or upper back or lower back or shoulder or elbow or hand or hip or knee or ankle/foot, excluding presence of chronic widespread pain, generating nine dichotomous variables.

Thyroidal disease 5
Cluster text: "Has it ever been verified that you have/have had hypothyroidism or hyperthyroidism?" Separate tick boxes for each condition (yes, no), generating two dichotomous variables, 1=Yes.
For each diagnosis, included were those who affirmed to have or have had the diagnosis. Chronicity is assumed based on medical knowledge and clinical experience.
Irritable bowel syndrome 16 17 Index question: "Have you had stomach pain or discomfort in the last 12 months?" Answers: Yes, much; yes, a little; no. Irritable bowel syndrome was further constructed from four of six followup questions: "If yes: "In the last 3 months, have you had this as often as 1 day a week for at least 3 weeks?" Yes, no. "Is the pain/discomfort relieved by having a bowel movement?" Yes, no. "Is the pain/discomfort related to more frequent or less frequent bowel movements than normal?" Yes,no.
"Is the pain/discomfort related to the stool being softer or harder than usual?" Yes, no.
Included with irritable bowel syndrome were those who affirmed little or much stomach pain or discomfort in the last year, who for as often as 1 day a week for at least 3 weeks in the last 3 months have had at least two of the following: pain/discomfort relieved by having a bowel movement, related to altered frequency of bowel movements, or related to altered stool appearance, resembling a modified version of the Rome criteria. 16 17 Gastro-oesophageal reflux disease 5 18 Cluster text: "To what degree have you had the following problems in the last 12 months?" Options combined type (nausea, heartburn/acid regurgitation, diarrhea, constipation, alternating constipation and diarrhea, and bloating) and frequency (never, a little, or much). Generated one dichotomous variable, heartburn, where 1=Much.
Gastro-oesophageal reflux disease is defined as much heartburn/acid regurgitation in the last 12 months. 18 Anxiety 5 19 Instrument variable: Hospital Anxiety and Depression Scale. 19 Every other statement of 14 statements covers symptoms on anxiety and depression and is scored 0-3. The HUNT Databank constructed a total score for anxiety (HADS-A), if all 7 anxiety items were answered.Anxiety was defined as HADS-A score >=8/21, indicating mild or possible anxiety. [20][21][22] Chronicity is assumed based on medical knowledge and clinical experience.Depression 5 19 Instrument variable: Hospital Anxiety and Depression Scale. 19 Every other statement of 14 statements covers symptoms on anxiety and depression and is scored 0-3. The HUNT Databank constructed total score depression (HADS-D), if all 7 depression items were answered. Depression was defined as HADS-D score >=8/21, indicating mild or possible depression. [20][21][22] Chronicity is assumed based on medical knowledge and clinical experience.Chronic insomnia 5 23 There were nine questions on sleeping pattern in one cluster, including three concerning insomnia. Initial text: "How often in the last 3 months have you "Had difficulty falling asleep at night?" Never/seldom, sometimes, several times a week. "Woken up repeatedly during the night?" Never/seldom, sometimes, several times a week. "Woken too early and couldn't get back to sleep?" Never/seldom, sometimes, several times a week.

Chronic insomnia was defined as in the last 3 months, several times a week, having difficulty falling asleep at night and waking up repeatedly during the night, and waking up too early. A modified version of the diagnostic criteria for insomnia in the International Classification of Sleep Disorders. 23 Alcohol use disorder 24
Instrument variable: Cut down/Annoyed/Guilty/Eye-opener, also known as the CAGE questionnaire. 24 The CAGE questionnaire is a 4-item scale with scores of 0-1. A summary variable was created and dichotomized in which a score of 1 indicates >=2 positive answers.
Alcohol use disorder was defined as CAGE score greater than 2. 25 Chronicity is assumed based on medical knowledge and clinical experience.

Dental health problem
"How would you say your dental health is?" Very, bad, ok, good, very good. Dental health problems were defined as self-reported bad or very bad dental health. 26 Chronicity is assumed based on medical knowledge and clinical experience.

Menopausal hot flashes
Included: Women older than 30 years only. Two questions were used to define menopausal illness: "Do you have/have you had hot flashes due to menopause?" During the day, during the night, day and night, haven't had any.
"If you have had hot flashes, how would you describe them?" Very intense, moderately intense, hardly noticeable.
Included with menopausal hot flashes were those who reported hot flashes occurring daily and/or nightly and of at least moderate severity.
Chronicity is assumed based on medical knowledge 27
One question on nocturia, identical to that of the International Prostate Symptom Scale (IPSS), was asked to men and women older than 30 years.
"How many times do you get up during the night to urinate?" None, 1 time, 2 times, 3 times, 4 times, 5 times or more. Nocturia was defined as two or more voids per night. 28 Chronicity is assumed based on medical knowledge and clinical experience.Urine incontinence 5 29 Excluded: men, 20-29 years. Instrument variable: The Epidemiology of Incontinence in the County of Nord-Trøndelag (EPINCONT) questionnaire. 29 Index question: Do you have involuntary loss of urine? Yes, no. Urine incontinence was constructed from two of six follow up questions. "If yes": "How often do you have involuntary loss of urine?" Less than once a month, once or more per month, once or more per week, every day and/or night "How much urine do you leak each time?" Drops or little, small amount, large amounts.
Self-reported frequency and volume of leakage were multiplied to obtain the validated 4-level Sandvik Severity Index, categorizing incontinence as slight, moderate, severe, and very severe. 29 Urine incontinence were included if severe to very severe.
Chronicity is assumed based on medical knowledge and clinical experience.Prostate symptoms 30 31 Included: men older than 30 years only. Instrument variable: The International Prostate Symptom Scale 30 was slightly modified in HUNT3, 31 becoming a 7-item scale with scores of 0-5 per question.
Included were prostate symptoms of at least moderate severity, i.e. summary score >= 8 points. 30 Chronicity is assumed based on medical knowledge and clinical experience.
Cluster text: "Do you have any of the following eye conditions?" Cataract, glaucoma, and macula degeneration. Separate tick boxes, yes, no.
For each diagnosis, included were those who affirmed to have or have had the diagnosis.
Chronicity is assumed based on medical knowledge and clinical experience.

Chronicity is assumed based on medical knowledge and clinical experience.Hypertension 5
Blood pressure in HUNT3 is measured three times at one consultation. The mean of measurement 2 and 3 is calculated by HUNT Databank. If measure 2 or 3 were missing, the other was used as estimate for the mean, https://hunt-db.medisin.ntnu.no/hunt-db/#/variable/144. Hypertension was defined as measured mean systolic BP>= 180 mmHg or diastolic BP >= 110 mmHg or reporting use of antihypertensive medications, excluding self-reported cardiovascular disease, diabetes, or kidney disease, and excluding extreme measures. 26 Chronicity is assumed based on medical knowledge and clinical experience.

Hypercholesterolemia 35
Hypercholesterolemia was defined as total-cholesterol >= 8 mmol/L. 35 Chronicity is assumed based on medical knowledge and clinical experience.  1 An exception to single entity LTC.

Supplementary data 2: Details on operationalization of socioeconomic position.
In the HUNT3 Survey interview, all participants were asked: "What is/was the title of your main occupation?" Free-text answers were manually classified according to the Standard Classifications of Occupations by Statistics Norway, 1 which is based on the European Union's version of the International Standard Classification of Occupations-88. 2 The standard categorize occupations according to skill level and specialization, degree of independence, and manual labor but not social position. 1 Occupations are coded with up to four digits, with increasing detail. One digit indicates major groups; two digits, submajor groups; three digits, minor groups; and four digits, unit groups. The minor occupational group was the highest level of detail available in the HUNT3 Survey.
Occupational socioeconomic position was operationalized using the European Socio-economic Classification scheme. 3 The full version of the scheme requires employment status and size of organization in addition to occupation to assign a class position. We used the simplified class scheme, based on minor occupational group only 3 , as the HUNT3 Survey did not have data corresponding to employment status and size of organization. It is shown that the agreement between three-digit full and simplified version of this scheme is 79.7% for the total workforce. 3 The syntax is available from https://www.iser.essex.ac.uk/archives/esec/matrices-and-syntax. It was performed using SPSS 25.0 (SPSS Inc., Chicago, IL, USA). Table S2 gives details of transformation of data, discrepancies between the Norwegian and European Union standard and the allocated position in the full classification scheme. 2179 individuals had alterations to their occupational data to fit the syntax, 5.7% (2179/38027) of the total sample.
In the HUNT3 Survey data, the minor occupational group was a string variable. To perform the syntax, it had to be altered to a numeric variable. The string "011" changed to numeric value "11," which was manually corrected in the syntax. Within the 3-digit variable, some participants were classified with 1 digit and 2 digits only. These were transformed to the corresponding 3-digit minor group, at the lowest level of detail, by manually adding suffix digits 0 or 00, which is allowed when information is insufficient (section 4.3) 1 and in line with operationalizing of European Socio-economic Classification (see footnote table 1). 3 Norwegian minor groups, which were not found in the European Union standard, were altered to the level of detail in which corresponding groups could be identified. These were Standard Classifications of Occupations by Statistics Norway codes: 112 (corresponding to 2 digits), 25 (corresponding to 1 digit), 251-6 (corresponding to 1 digit), 349 (corresponding to 2 digits), 631 (corresponding to 1 digit), 641 (corresponding to 1 digit), 735 (corresponding to 2 digits), and 745 (corresponding to 2 digits).
In total, 9 classes were created. To increase power and simplify interpretation, the full scheme was collapsed into a 3-class version, with "high" combining class 1 and 2, "middle" combining 3 to 6, and "low" combining 7 to 9. 3 The high occupational class represents large employers, highergrade and lower-grade professionals, administrative and managerial occupations, higher-grade technician occupations, and supervisory occupations. The middle occupational class consist of small employers, self-employed individuals, lower supervisory occupations, and lower technician occupations. The low occupational class contain lower services, sales and clerical occupations, lower technical occupations, and routine occupations.

Supplementary data 3: Estimated 10-year risk of death by all individual and organ-grouped chronic conditions, occupational groups and sex and descriptive data.
Joint effect of occupational group and multimorbidity on 10-year mortality were presented in the article visualized by splines regression of individual chronic conditions for each occupational group and by sex at age 60 years. This appendix presents supplemental visual presentation of the full range 0 to 21 individual chronic conditions (figure S1). In a sensitivity analysis, the full range 0 to 11 organ-grouped chronic conditions was graphed by use of the same method ( figure S2).
The accompanying total frequencies and number of events are presented in Table S3 for individual chronic conditions and Table S4 for organ-grouped chronic conditions, by occupational group and sex. Shading indicates 95% CIs; low, middle, and high indicate occupational groups.