Prevalence of respiratory symptoms in Norway

J ChronDis 1972.Vol. 25, pp.

519-544.Pergamon Press. Printed in Great Britain

PREVALENCE

OF RESPIRATORY IN NORWAY

SYMPTOMS

WILLIAMHAENSZEL*and ANNA HOUGEN

Cancer Registryof Norway (Received 29 July 1971; in final form 23 March 1972)

INTRODUCTION the prevalence of respiratory symptoms in Norway has been obtained in the course of a study to investigate changes in risks from cardiorespiratory and other diseases among Britons and Norwegians migrating to the United States [l]. The symptom data were collected by mail query using a shortened version of the questionnaire (translated into Norwegian) developed by the Medical Research Council Committee on the Aetiology of Chronic Bronchitis [2], the intent being to compare them with the counterpart data from the United Kingdom and the United States. The study design took note of the fact that mortality, and presumably morbidity, from chronic non-specific lung disease (CNSLD) is much lower in Norway than in the United Kingdom. Apart from this application to migrant studies the Norwegian setting offers other distinctive features relevant for respiratory disease. These include a population distributed between latitudes 58”71” N exposed to severe winters, the absence of large metropolitan conurbations outside the capital city of Oslo, and an unusual pattern of tobacco use. The latter is of interest because of the well-documented excess risk of respiratory symptoms among cigarette smokers [3]. In Norway, pipe smoking accounts for a high proportion of the tobacco consumed. Furthermore, the cigarette sales tax has favored the substitution of hand-rolIed for manufactured cigarettes. The contrasts thus afforded permit more detailed review of effects associated with specific forms of tobacco use. Studies in England and the United States have linked respiratory symptoms and disease with air pollution and other environmental exposures [4-81, but detection of air pollution effects has been obscured by the dominant influence of tobacco use. Populations exposed to low levels of smoking and urbanization can provide an informative contrast to the U.K. and U.S. experience, and this paper attempts a systematic description of respiratory symptom prevalence in Norway with these points in mind. INFORMATION on

*Present Address: Biometry Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20014; while in residence at the &ncer Registry of Norway, William Haenszel was a TraveImg Fellow of the International Agency for Research on Cancer. Work reported here supported in part by Public Health Research Contract PH-64499 from the National Cancer Institute. 519

WILLIAM HAENSZELand ANNA HOUGEN

520 TABLE 1.

Norway, total Age

CHARACTERISTXSOF THEGENERALPOPULATIONAND SIB SAMPLES General population* Males Females No. Percent No. Per cent

Sibs Males Females No. Per cent No. Per cent

6712

100.0

7791

100.0

3887

100.0

4819

100.0

1231 1894 2400 1187

18.3 28.2 35.8 17.7

1552 2009 2692 1538

19.9 25.8 34.6 19.7

570 1308 1271 738

14.7 33.6 32.7 19.0

659 1561 1664 935

13.7 32.4 34.5 19.4

Tobacco use Non-smokerst Ex-smokers Current smokers

1479 1412 3760

22.0 21.0 56.0

5898 324 1540

75.7 4.2 19.8

947 892 2014

24.4 22.9 51.8

3840 201 766

79.7 4.2 15.9

Birthplace Urban Rural non-farm Rural farm

1745 1611 3157

26.0 24.0 47.0

2072 1813 3539

26.6 23.3 45.4

1393 649 1763

35.8 16.7 45.4

1707 819 2161

35.4 17.0 44.8

Current residence Urban Rural

2136 4576

31.8 68.2

2833 4958

36.8 63.6

1432 2455

36.8 63.2

1999 2820

41.5 58.5

Region I (Oslo) II III IV V VI VII

884 2391 791 1227 627 663 129

13.2 35.6 11.8 18.3 9.3 9.9 1.9

1308 2651 914 1378 703 713 124

16.8 34.0 11.7 17.7

344 858 1223 908 241 294 19

8.9 22.1 31.5 23.3 6.2 7.6 0.5

536 992 1525 1132 288 332 14

11.1 20.6 31.6 23.5 6.0 6.9 0.3

35544 45-54 55-64 65-74

yr yr yr yr

Note: Tobacco *Non-augmented

Z:X 1.6

use and birthplace not stated by some respondents. population sample.

tPersons with no history of regular tobacco

use.

.S.S.R.

FIG. 1.

Prevalenceof Respiratory Symptomsin Norway

521

MATERIAL AND METHODS Population samples

The populations studied were drawn from two different sources: (a) a stratified, systematic sample of the general population in the entire country, augmented by additional population samples in four selected counties (Finmark from the north of Norway and Avst-Agder, Vest-Agder and Rogaland from the Southwestern region); (b) a roster of sibs, living in Norway, of Norwegian-born migrants to the United States. The general population and sib samples are described in detail elsewhere [9, lo], and only a brief account is given here. The population sample of persons born 1893-1929 was drawn systematically from the 1960 census lists using a variable sampling fraction, 1.5 per cent for persons born 1893-1912 and 0.75 per cent for those born later. In the four selected counties, the augmented sampling fractions were 3.5 and 1.75 per cent for the respective age groups. The sample for the entire country numbered 18,619 plus an additional 5057 in the four counties. There were 785 deaths reported prior to the start of the survey in 1964, leaving a total of 22,891 eligible respondents. The Norwegian-born migrants surveyed in the United States were asked to provide the names and addresses of brothers and sisters living in Norway. Further information on sibs inadequately identified by this method was obtained by the Cancer Registry of Norway through additional inquiries sent to the migrant, from the Norwegian municipal population registries, and from sibs with whom contact had aIready been established. The latter procedures also identified sibs not reported by the index migrant. Approximately 13,000 sibs were assembled in this manner. The mailing of the questionnaires started in the fall of 1964. The names and addresses of non-respondents were checked in the population registries, and the initial mailing was followed by two reminder letters. The overall response rate was 76 per cent and was higher for the general population sample (80 per cent) than for sibs (66 per cent). The major reason for this difference was probably the better information on identity and addresses available for the general population. The completed questionnaires were carefully checked; those which did not pass the fail-edit for certain respiratory and cardiac symptoms were re-queried. Table 1 shows the general population and sib samples to have similar distributions by age and tobacco use. More sibs were born in urban areas and their residences were concentrated more in southern Norway as expected, since the major waves of migration to the United States originated there. The regional groupings for current residence used in Table 1 are described in Fig. 1. Definition of symptoms

The screening questions followed the format developed by the Medical Research Council Committee on the Aetiology of Chronic Bronchitis [2], which identifies the symptoms of persistent cough and phlegm (PCP) and chronic bronchitis by affirmative replies to a sequence of questions. The proportion remaining positive at each stage and the transition probabilities from one symptom complex to the next state are given below for the combined general population and sib samples. The results for total males and females suffice, since the age variation was minimal.

522

WILLIAMHAENSZELand ANNA HOUGEN

Males

Females

Per cent positive

Probability of remaining positive in next state

Per cent positive

Probability of remaining positive in next state

Cough

25.8

0.48

13.2

0.39

Persistent cough Persistent cough and phlegm Chronic bronchitis

12.5

0.12

5.1

0.70

9.0 1.8

0.20 -

3.5 1.2

0.34 -

The low Norwegian symptom prevalence described in this paper can be traced to the smaller percentage of positive responses to the initial questions on cough. The attrition of positive responses in the later phases was not grossly out of line with experience elsewhere [ 111. Factor adjustment

The indirect method was used, unless the contrary is specifically noted, to adjust for age and type and amount of tobacco use, a condensed version of the age- and smokingclass specific rates in Table 4 for the total sample of Norwegian males and females being employed for this purpose. The same approach was taken to control for age started smoking. The indirect method can, in principle, lead to anomalous results in group comparisons. Inspection of the detailed computations indicated that the differences in the summary contrasts persisted as a rule within subcategories. Under these circumstances, the overall comparisons are not very sensitive to the choice of standard rates or standard populations for weighting purposes and more sophisticated methods were not needed to deal with higher-order interactions. Confidence limits

The analysis has concentrated on the magnitude and pattern of effects rather than on tests of significance. In keeping with this philosophy, the 95 per cent confidence limits of the standardized ratios, which assume that the number of symptom positives can be represented as a Poisson-variate, have been entered in several tables. The assumption is conservative and may lead to overestimating the width of the confidence limits. The numerator of symptom positives for all rates and ratios has been shown routinely for the convenience of the reader interested in making derivative computations. RESULTS

The crude prevalence rates per 100 men and women aged 35-74 yr for PCP and chronic bronchitis are given below for the general population and sib samples. The combined samples include the data from the augmented population samples in four Norwegian counties. The similar age distributions permit direct comparison of the crude rates for general population and sibs and the estimates of symptom prevalence are sensibly equivalent when sampling variability is taken into account.

1.3 0.4 1.6

Total Non-smokers$ Smokers8

sample.

1.4 0.4 1.6

10.0 2.8 11.5

8.4 3.7 10.1 2.2 1.3 2.6

Sibs

9.3 2.9 11.1

1.4 0.5 1.6

QCurrent and discontinued

regular smokers.

Males Combined samples*

tNon-augmented population sample. *Persons with no history of regular tobacco use.

*Includes augmented population

9.1 3.0 11.3

Combined samples*

35-54 yr Gen. p0p.t Sibs

Combined samples*

8.7 3.7 10.4 1.2 0.8 1.9

4.2 2.2 8.0

35-54 yr Gen. p0p.t 3.3 2.5 7.8 1.3 1.0 3.2

3.6 2.1 7.2 0.9 0.6 1.8

Sibs

Combined samples*

Females

1.2 0.9 2.8

3.1 2.3 7.9

55-74 yr Gen. p0p.t

1.7 1.4 4.1

3.7 3.1 7.8

Sibs

SEXFOR SMOKERSAND NON-SMOKERS: GENERALPOPULATIONAND SIB

Persistent cough and phlegm 8.2 3.9 3.2 2.1 10.0 7.5 Chronic bronchitis 2.3 2.2 1.1 1.1 0.6 2.7 2.0

55-74 yr Gen. p0p.t

RESPIRATORYSYMPTOMPREVALENCE RATESPER 100 POPULATIONBY AGE AND SAMPLES

Total Non-smokers* Smokers5

TABLE2.

524

WILLIAMHAENSZELand ANNA HOUGEN

Persistent cough and phlegm Males Females

Chronic bronchitis Males

Females 1.2

Combined samples

9.0

3.5

1.8

General population

9.3

3.6

1.9

1.2

Sibs

8.7

3.1

1.8

1.3

The two sources also yielded substantially the same inferences on level of risk and gradients by age and tobacco use (Table 2). Respiratory symptom prevalence for the more detailed smoking classifications and other variables considered in this report was reviewed and the same general tendencies found to persist within each sample. Differences noted were usually within normal sampling variation, so that nothing new is learned from separate analyses of the sib and general population samples. Combination of the two sets of data to secure a more regular and consistent pattern of results was indicated, and beginning with Table 3 the presentations relate to the combined general population and sib experience. TABLE 3.

PER CENT DISTRIBUTION OF SMOKING HABITS BY

AGE

AND SEX: COMBINED

SAMPLES

Males

Total Number Per cent Non-smokers* Ex-smokers Current smokers Pipe, cigars only Cigarettes only Mixed

Females

35-54 yr

55-74 yr

35-54 yr

55-14 yr

5928 100.0 19.1 18.8

6593 100.0 26.4 24.5

6741 100.0 66.8 5.9

7885 100.0 86.6 2.5

15.4 37.3 9.4

18.6 24.0 6.5

27.3 -

10.9 -

*Persons with no history of regular tobacco use.

Age and sex Irregularities noted in the pattern of prevalence rates by lo-yr age groups probably reflect chance variation and the changes with age emerge more clearly in contrasts of two broad age groups. The nature of the age effects is also clarified by control for smoking history (Table 2). For females there was a small rise in PCP prevalence with age for smokers and non-smokers alike and the overall lower risk for older women is obviously due to their larger number of non-smokers. Males exhibited divergent tendencies, a decline with age in the PCP rate for smokers being partially offset by an increase among the numerically less important non-smokers. The age effects for chronic bronchitis appeared stronger than for PCP, since a rise in prevalence of these symptoms could be discerned among smokers and non-smokers of each sex. This description of age effects followed closely those exhibited by a sample of U.S. nativeborn [6], but deviated from the British configuration, in which both sets of symptom prevalence increased with age in all smoking-history categories [4].

52.5

Prevalenceof Respiratory Symptomsin Norway Type and amount of tobacco

The Norwegian practices in tobacco use (see Table 3) influenced the classification adopted for analysis and presentation of results in this paper. It seemed unwise to ignore the substantial numbers of men who smoked a pipe, alone or with cigarettes and to take note only of tobacco consumed in the form of cigarettes as was done in the U.K. and U.S. publications [4, 61. Pipe (and cigar) smokers were identified separately under the headings of ex- and current smokers. Men currently using both cigarettes and pipe were termed ‘mixed’ smokers and classified by consumption of tobacco in all forms. We stress that non-smokers in our terminology means non-regular users of any tobacco, not merely non-smokers of cigarettes. This definition simplifies interpretation of contrasts between male and female non-smokers. In Norway, as elsewhere, smoking history is the most important, single determinant of respiratory symptom prevalence. Table 4 indicates the rates for PCP and chronic bronchitis among smokers to be roughly three times those for non-smokers, a ratio which held approximately within each sex-age group. The prevalence of PCP and chronic bronchitis among ex-smokers was intermediate to that reported by nonsmokers and current smokers, but no firm judgement as to whether the risks among male ex-smokers differed with respect to the manner in which tobacco was consumed, was possible. Detail on maximum rate of use for ex-smokers was omitted, since this did not yield any consistent gradient in symptom prevalence. The experience of ex-smokers in relation to duration of discontinuance is discussed in a succeeding section. The substantial numbers of male pipe and mixed smokers in contrast to cigarette smokers offer an opportunity to investigate whether elevated respiratory symptom prevalence represents a general tobacco effect or one that can be modified by mode of use. Male pipe smokers exhibited a lower overall prevalence for both PCP and chronic bronchitis than either cigarette or mixed smokers, although the disparity would be reduced by control for rate of use. The difference between cigarette and mixed smokers was minimal. Norwegian women smoked cigarettes almost exclusively and no parallel comparisons are possible for them. Table 4 referred to status at time of survey and the current smoker categories contain members who had changed their smoking practices at a past date. However, it was possible to define subgroups of individuals who had consistently followed the same practices throughout life and the contrasts in PCP prevalence between pure pipe and cigarette smokers as determined by current and lifetime criteria are summarized below: Pipe only Lifetime history Current

Cigarettes only Lifetime Current history

35-54 yr

8.7

6.1

14.8

13.0

55-74 yr

7.2

5.0

15.0

13.8

The more restrictive lifetime criteria yielded lower prevalence rates for pipe and cigarette smokers; this probably reflects in part a dosage effect, since persons who had switched from a multiple to a single mode of use may have simultaneously reduced their total tobacco consumption. The major point to be made here, however, is that

526

WILLIAMHAENSZEL and

ANNA HOUGEN

TABLE 4. NU~IBER OF PERSONSWITH RESPIRATORY SYMPTOMSAND PREVALENCERATESPER 100 POPULATION BY AGE,SEX,AND TYPE AND AMOUNTOF TOBACCOUSE: COMBINEDSAMPLES Males 35-54 yr 55-74 yr No. Rate No. Rate Total 574 Non-smokers 34 Smokers* 540 Ex-smokers, total 53 History of pipe and cigar use only 12 History of cigarette use 41 Current smokers, total 487 Pipe, cigars onlyt* 79 < 10 g/day 35 44 IO+g/day Cigarettes onlyt$ 327 i 10 cigarettes/day 65 lo-19 ,, /day 164 /day 96 20+ Mixedt$ ” 81 44 <: 20 g/days 35 2O+g/day§ Total Non-smokers Smokers* Ex-smokers, total History of pipe and cigar use only History of cigarette use Current smokers Pipe, cigars onIyt$ < 10 g/day lO+g/day Cigarettes onlyt* < 10 cigarettes/day 10-19 ,, /day 20+ ,, /day Mixedt/: < 20 g/days SR g/day§

9.7 3.0 11.3 4.7

Persistent cough 553 8.4 64 3.1 489 10.1 103 6.4

Females 35-54 yr 55-14 yr No. Rate No. Rate and phlegm 261 3.9 93 2.1 168 7.5 16 4.0 -

257 174 83 7

3.3 2.5 7.8 3.5

-

-

4.0 8.2 8; 6.4 10.2 24.1 -

7 76 16 32 33 9 -

3.5 8.9 8.9 5.8 14.0 22.5 -

1.1 0.6 2.0 1.5

102 68 34 4

1.3 1.0 3.2 2.0

-

5.8

41

6.3

4.5 13.2 8.7 6.7 11.6 14.8 8.5 15.4 26.2 14.5 11.6 21.7

62 386 88 40 48 237 75 116 43 61 39 19

lY.2 7:2 5.6 10.0 15.0 10.9 17.4 20.6 14.3 13.6 15.2

79 4 75 15

1.3 0.4 1.6 1.3

148 23 125 31

4

1.9

17

2.6

-

-

-

-

11 60 10 4 6 43 7 25 11 7 1 6

1.2 1.6 1.1 0.8 1.6 1.9 0.9 2.4 3.0 1.3 0.3 3.1

14 94 20 13 7 59 15 28 13 15 7 6

1.4 2.9 1.6 1.8 1.5 3.7 2.2 4.2 6.2 3.5 2.4 4.8

6 39 39 18 16 -5

1.5 2.1 -

4 30 30 13 10 6 -

2.0 3.5 -

16 152 152 17 :z -

Chronic bronchitis 2.2 14 1.3 29 2.6 45 1.9 6

-

2; 1.5 3.0 60 L-

3.5 2.3 4.2 15.0 -

*All persons with history of regular tobacco use. tclassified by type and rate of use at time of inquiry. *Includes unknown rate of use. QBased on consumption of all forms of tobacco.

both criteria yielded the same relative relationships. Preparation of Table 4 had systematically covered the lifetime history categories, but this finding led us to discard the latter and to retain only classifications by current status, for which the responses were presumed to be more accurate and reliable. The class intervals for cigarette consumption rate do not conform exactly to those employed in reporting the findings from Britain and the U.S. In Norway the size of the standard package of tobacco used to make hand-rolled cigarettes, dictated a

Prevalence of Respiratory

Symptoms in Norway

527

different grouping to minimize rounding errors in response. The amount of tobacco contained in a hand-rolled cigarette has been investigated previously and an average conversion ratio of 1 g per cigarette established [12]. For both manufactured and hand-rolled cigarettes, the groupings were set at < 10, lo-19 and 20f cigarettes/day. Two classes were retained for pipe and mixed smokers; for pipe smokers IO+ g means in effect 10-19 g and for mixed smokers ~20 g can be read as substantially equivalent to 10-19 g. The detailed distributions by amount indicate that it is difficult to smoke over 20 g daily by pipe alone, so that pipe smokers desiring to increase consumption normally become mixed smokers. Conversely, persons smoking less than 10 g daily appear able to satisfy their needs by one method of use. The breakdown by rate of use in Table 4 reveals that within each category-pipe, cigarette, mixed-the prevalence of PCP and chronic bronchitis symptoms increased with amount consumed. A somewhat steeper gradient by amount was suggested among the younger male cigarette and mixed smokers, particularly for PCP, but little difference between young and older female smokers was noted on this point. The symptom experience of Norwegian cigarette smokers agreed qualitatively with that described for populations in which cigarettes are the major, and almost exclusive, form of tobacco use. In Norway the next question is whether cigarette and pipe smokers present the same ‘dose-response’ gradient. A straightforward approach contrasting pure cigarette and pure pipe smokers is contraindicated by the confounding of pipe smoking and rate of use to which we have alluded. The difficulty can be partially circumvented by contrasting pure cigarette smokers against the pooled experience of pipe and mixed smokers. This course of action provides a full range of dosages for pipe smokers and any deficit in the estimated prevalence for pipe smokers would be conservative because of the contribution of cigarettes to the latter set of rates. An even more conservative contrast takes persons with a lifetime history of cigarette use only as the standard of reference, since their current consumption should more closely approximate maximum rate of use. The results are shown in Table 5 and summarized in Fig. 2. PERCENT

CONSU~PT,ON

SYMPTOM

POSITIVE

O;



IO-19g/DAY

Cigarette

Smokers

Pipe Smokers Only Mwd Smokers

Only And

RESPIRATORY

1.8 2.7 4.5

*Adjusted by direct method to the age distribution

13 22 12

1.3 1.5 3.6

6.4 11.9 16.6

No.

Rate

Pipe and mixed

65 164 96

Persistent cough and phlegm 8.5 30 7.0 42 6.8 15.4 74 13.2 74 11.9 26.2 51 26.2 42 21.1 Chronic bronchitis 7 0.9 4 4 07 g 5 0’8 25 2.4 13 11 3.0 7 3:6 8 4:0

Cigarettes only Lifetime history Current No. Rate No. Rate

for the total sample of Norwegian males.

18 18 14

1.6 3.4 4.7

22 53 24

< 10 g/day 10-19 g/day 20f g/day

9.9 13.8 21.0

89 149 66

73 117 67

9.8 16.4 23.2

140 280 139

Rate

Pipe and mixed

< 10 g/day 10-19 g/day 20+ g/day

All ages* 35-54 yr

15 28 13

75 116 43

2.2 4.2 6.2

10.9 17.4 20.6

9 9 5

43 43 16

2.6 3.0 5.2

12.5 14.4 16.5

Cigarettes only Lifetime Current history No. Rate No. Rate

55-74 yr

14 13 6

47 75 24

No.

3.2

2.1

1.8

6.0 11.9 12.7

Rate

Pipe and mixed

AND AGE-ADJUSTED PREVALENCE RATES PER 100 POPULATION BY AMOUNT OF TOBACCO USE AND AGE: ONLY(CURRENT),CIGARE-ITES ONLY(LIFETIME HISTORY),PIPEAND MIXED,COMBINED SAMPLES

SYMPTOMS

CURRENT SMOKERS-CIGARETTES

OF MEN WITH

No.

MALE

NUMBER

Cigarettes only Lifetime Current history No. Rate No. Rate

TABLE 5.

8

z 8

F 3

E

2 z $ zc;l

Prevalence of Respiratory

Symptoms

529

in Norway

A deficit in prevalence of PCP and chronic bronchitis of the order of 2540 per cent for pipe and mixed smokers measured against pure cigarette smokers (current or lifetime history) suggests an effect for mode of use independent of amount. The fact that 11 of 12 detailed contrasts point in the same direction lends added weight to this inference. The effect of control of such other variables as age started smoking and inhalation practices on the symptom differentials between pure cigarette and other smokers is examined later, but we may note here that these refinements did not change the character of the findings. In general, the tobacco effects delineated in Table 4 require control in the study of other variables and most of the results that follow are expressed as standardized ratios adjusted for age and type and amount of tobacco use. Hand-rolled cigarettes

The present data agreed with the Finland-Norway study [12] on the dominant role of hand-rolled cigarettes in Norway. Of all current cigarette smokers 58.5 per cent used hand-rolled cigarettes exclusively, compared to the preference of 27.1 per cent for manufactured cigarettes. The remainder used both or failed to specify their practice. Table 6 distributes the responses by sex, age and residence. Men relied more on hand-rolled cigarettes, 70 per cent smoking them exclusively, compared to 42 per cent of the women. The smaller amounts consumed by women presumably made it less imperative for them to smoke the cheaper hand-rolled variety. Hand-rolling was most widely practiced by younger males in rural areas. For both sexes the purchase of commercial brands increased with age and urbanization. TABLE 6. PER CENT DISTRIBUTION OF PERSONS USING EXCLUSIVELY HAND-ROLLED OR MANUFACTURED CIGARETTES BY SEX, AGE AND PLACE OF RESIDENCE: CURRENT SMOKERS OF CIGARETTES ONLY,COMBINED SAMPLES

Per cent using exclusively: Manufactured Hand-rolled cigarettes cigarettes

Ratio: Hand-rolled/ manufactured

58.5

27.1

2.2

Males 35-54 yr Urban Rural 55-74 yr Urban Rural

70.0

15.8

4.4

64.6 78.7

19.3 8.3

3.3 9.5

55.5 72.9

28.4 14.3

2.0 5.1

Females 35-54 yr Urban Rural 55-74 yr Urban Rural

42.3

42.9

1.0

36.4 57.9

49.4 27.0

0.7 2.1

23.8 40.7

60.8 44.8

0.4 0.9

Both sexes

If hand-rolled and commercial cigarettes carry different risks, the variation in practices just described would have important implications for respiratory symptom prevalence. To test this possibility, Table 7 compares the standardized prevalence ratios for pure cigarette and mixed smokers using the hand-rolled or manufactured D

~PIRATORY

75 -

Both

696 88 - loo - 114

86 - 130 - 189

100 - 121 - 147 567190 28

109 72

225

49 -

77 61 99 - 177

97 - 122 98 - 148

83 - 100 - 120

Ratio

Males

11

75 22

119

No. with symptoms

Females

53 - 122 - 240

91 - 132 - 185 29 - 50 - 81

78 - 100 - 128

Ratio

Chronic bronchitis

CIGARETTES: CURRENT MALE AND ~mi.4Le

8

34 16

68

No. with symptoms

CXGAREXTE

100

74 -

88 - 105

80 - 112 - 152

II5 - 231 - 413 96 - 137 - I89

*Adjusted for age and tobacco use at time of inquiry.

Note: 95 per cent confidence limits of prevalence ratios shown in italics.

85 - 101

132

105

71 -

84 - 102

69 -

22 yr and over

705 203 232

109 - 125 - 144 89 - 101 - 11.5

100

132

40

11 36

228

Persistent cough and phlegm Males Females No. with No. with Ratio Ratio symptoms symptoms

38 -

47 -

64 - 101

81 - 130

94 - 133 - I82 75 - 105 - I43

100

Ratio

Males

18

17

38 40

124

No. with symptoms

Females

60 -

85 - 116

67 - 126 - 215

44 - 214 - 625 56 - 122 - 232

100

Ratio

Chronic bronchitis

39

13

3 9

69

No. with symptoms

RESPIRATORY SYMPTOM PREVALENCE RATIOS* BY AGE STARTED SMOKING: hfALJ3 (CIGARETTE AND MIXED ONLY) AND FEMALE CURRENT SMOKERS, cOMmNED SAMPLF.3

Under 17 yr 17-19 yr 20-21 yr

Total

TABU 8.

*Adjusted for age and tobacco use at the time of inquiry. *All pure cigarette and mixed smokers reporting current rate of use.

62

487 107

Note: 95 per cent confidence limits shown in italics.

97 - I25

93 - 100 - 108

95 - 104 - II4 78 - 95 - 11.5

Hand-rolled Manufactured

Persistent cough and phlegm Males Females No. with No. with Ratio Ratio symptoms symptoms

SYMPTOM PREVALENCFZRATIOS* FOR USERS OF HAND-ROLLED AND hiANUFACwRED SMOKERS, CGMLUNED SAMPLES

All cigarette smoker@

Type of cigarette

TABLE 7.

531

Prevalence of Respiratory Symptoms in Norway

variety or both. Among males, little difference in risk between hand-rolled and manufactured cigarettes was revealed. The apparent excess female prevalence of PCP and chronic bronchitis for hand-rolled cigarettes was shown by more detailed data to be comined to those smoking < 10 cigarettes/day. The coarse grouping ‘ < 10 cigarettes/day’ employed in controlling for amount exaggerated the risk differential, since women in this category smoked more hand-rolled cigarettes on the average (7-8/day) than commercial brands (2-3/day). More precise groupings by rate of use modified the PCP ratios for women in Table 7 to read: hand-rolled, 111; manufactured, 78 ; both, 125. While not eliminating the differential, this refinement narrowed it to the point where sampling variation may be entertained as an explanation. The negative finding for males, accompanied by a female result not in flat disagreement, leads us to reject tentatively the hypothesis of different respiratory symptom prevalence for hand-rolled and manufactured cigarettes and this factor is ignored in the analysis of other study variables. Age started smoking

Cigarette, but not pipe, smokers were asked at what age they started to smoke, so that analysis of this factor is perforce limited to cigarette and mixed smokers. Tables 8 and 9 were further restricted to current smokers to exclude those who might recently have given up smoking because of the onset of symptoms. A subsequent check of ex-smokers revealed a weak relationship with age started so that inclusion of this group would have diluted only slightly the results reported here. TABLE AMOUNT

9. AGE-ADJUSTED* PCP SMOKED

PREVALENCE RATES PER 100 POPULATION BY SEX, AND AGE STARTED SMOKING: MALE (CIOAREITE AND MIXED) AND FEMALECURRENTSMOKERS,COMBINEDSAMPLES

l-9 g Age started Under 17 yr 17-19 yr 20 yr and over

13.8 10.1 7.4

Amount smoked per day Males Females lo-19 g l-9 g 20+ g 10+ g 20.7 14.7 13.9

24.7 24.5 19.1

>

11.6 5.6

17.6 12.2

*Adjusted by direct method to age distribution for the total sample of Norwegians.

Table 8 showing the prevalence ratios for the two symptom complexes, adjusted for current age and amount smoked, describes an inverse gradient for age started smoking, the highest risks for both sexes being noted among those beginning before age 17. The influence of prolonged delay past age 20 was not very pronounced for PCP. Since the influence of starting age on respiratory symptoms has not been studied extensively, additional details on the relationship of age-adjusted PCP prevalence to starting age and amount smoked are supplied in Table 9. The disparity, percentagewise, between the risk for smokers starting before age 17 and after 20 was greatest among persons using < 10 g/day. Higher rates of tobacco use reduced the relative importance of the age started, although an absolute decline in symptom prevalence among those taking up the habit at older ages persisted. The stronger female expression of the starting age effects in Table 8 can thus be readily explained

WILLIAMHAENSZELand ANNA HOUGEN

532

by the female preponderance of light smokers, for whom the age-started effect was greatest. The questionnaire did not elicit total years of exposure to tobacco, so that duration of exposure in relation to respiratory symptoms could not be investigated directly. The question arises, therefore, whether the age-started gradients represent duration effects in another guise, since in observational, as opposed to experimental, situations the distinction between age started and duration is not clear-cut. For smokers of a given age with an uninterrupted history of tobacco use, age started and duration of exposure represent two facets of the same event with a one-to-one correspondence between the two sets of labels. The observation that the gradients by age started persisted virtually unchanged among persons over 55 yr, as shown in the following adjusted PCP ratios for both sexes combined, argues in favor of a ‘critical age’ effect over and above a duration effect. 35-54 yr

55-14 yr

16 yr and under

123

135

17-19 yr

108

99

88

89

20 yr and over

If starting age reflected only the effect of duration of exposure, its association with prevalence should weaken as the population aged and the relative differences in total exposure among persons with different starting ages diminished. This point is strengthened by noting the short time span represented by the starting ages, which can produce only small differences in total exposure incommensurate with the observed changes in symptom prevalence. The Norwegian findings appear compatible with those reported from Cracow, Poland despite differences in terminology [l 11. Sawicki subtracted breaks in smoking lasting 1 yr or more from total years of exposure, defined ‘duration of smoking period as an independent variable, and found duration of exposure ‘completely overshadowed the effect of (current) age both in males and females’. However, only 12 per cent of Sawicki’s current smokers had given up smoking for 1 yr or longer and his ‘duration’ variable is intimately related to age started smoking. This, coupled with his report of a weak association with current age, leads us to believe that the observations in Norway and Cracow have a similar substantive content. Also, Sawicki’s statement that ‘the effect of long smoking periods considerably reduced the influence of the average number of cigarettes smoked’ shows the interaction between amount smoked and age started, which we noted, but chose to interpret as the impact of a high rate of use in minimizing the age-started effect. Despite any uncertainties in interpretation of the data on age started, the effect of control for this factor in the analysis of associated variables, such as inhalation practices, residence and occupation, required investigation. Prevalence ratios for these variables, adjusted for both age started and amount smoked, were computed and introduction of control for age started found to produce only minor changes. The similar distributions of starting age among users of commercial and hand-rolled cigarettes made its control in this connection unnecessary.

Prevalence of Respiratory

Symptoms in Norway

533

Inhalation

Some ex-smokers failed to answer the inhalation question, so that analysis was confined to current smokers. Among this latter group inhalation was the common practice; the proportion who denied inhalation ranged from 8.4 per cent among males aged 35-54 to 37.7 per cent among females aged 55-74 (Table 10). Among men little difference between cigarette and mixed smokers was noted and only the older pipe smokers reported substantial numbers of non-inhalers. TABLE

10. PER CENT OF CURRENT SMOKERS WHO DENIED INHALATION AGE,SEX AND TYPE OF TOBACCO USE: COMBINED SAMPLES

Current Pipe, cigar smokers, total only Males 35-54 yr 55-74 yr

8.4 26.9

Females 35-54 yr 55-74 yr

13.8 37.7

16.8 45.0

Cigarette only

Mixed

5.5 15.1

5.9 19.4

13.8 37.7

BY

-

Since symptom prevalence within each category is determined primarily by inhalers, their standardized ratios can exceed 100 only slightly even when low risks prevail for non-inhalers, so that the results for non-inhalers in Table 11 suffice to describe inhalation effects. Despite its subjective nature the inhalation question discriminated for respiratory symptoms. The PCP results and the more scanty data for chronic bronchitis indicated lower risks for non-inhalers within all age-sex groups and each mode of tobacco use. The estimated ratios have wide confidence limits and the result for mixed smokers may not be incompatible with the other findings. The ratios may represent minimum estimates of an inhalation effect; response errors on inhalation status not systematically related to symptom prevalence would tend to inflate the estimated ratios for non-inhalers. TABLE

11.

RESPIRATORY SYMPTOM PREVALENCE RATIOS* FOR NON-INHALERS: AND FEMALE CURRENT SMOKERS,COMBINED SAMPLES

Chronic bronchitis No. with Ratio symptoms

Persistent cough and phlegm No. with Ratio symptoms -

55 56 45 83

MALE

Males Pipe only Cigarette only Mixed

43 38 28 44

- 71 - 81 - 68 - 142

64 29 22 13

23 - 47 - 84 15 - 47 - 110 63191 22 - 107 - 312

11 5 3 3

Females

52 - 75 - 106

33

38 -

11

II - 138

*Adjusted for age and tobacco use at time of inquiry; lOO=rates for all male and female current smokers.

We have noted that simultaneous control for both amount smoked and starting age left the findings almost unaltered. The PCP ratio for non-inhalers (male cigarette and mixed smokers combined) shifted from 54 to 56 with control for starting age,

534

WILLIAM HAENSZEL and ANNAHOUGEN

and the figure of 75 for female smokers remained unchanged. Although the reduced symptom prevalence among non-inhalers appears real and substantial, the small numbers involved rule out inhalation practices as a control variable of consequence. Inhalation effects for respiratory symptoms parallel the observations on lung cancer. While the early epidemiological studies yielded contradictory results, the weight of the evidence accumulated in later investigations supports the inference of higher lung cancer risks among inhalers [13-151. Discontinued smokers

Table 12 investigates the change in respiratory symptom prevalence among exsmokers by length of time discontinued. The act of discontinuance may be triggered by symptoms; to minimize this bias, persons who had discontinued for reasons of respiratory illness or who had stopped smoking within the year prior to the survey were excluded. Amount smoked in the past appeared unrelated to symptom prevalence among ex-smokers, and so was not considered in this analysis. The limited observations available dictated the pooling of data for all ages and a coarse grouping of time intervals. TABLE12. RESPIRATORY SYMPTOM PREVALENCE RATESPER 100 DISCONTINUED SMOKERS BY YEAR OF DISCONTINUANCK~htBINJ3DSAMPLES

Persistent cough and phlegm Chronic bronchitis Males Females Males Females No. with No. with No. with No. with Rate symptoms Rate symptoms Rate symptoms Rate symptoms Yr of discontinuance 1955-1963 1945-1954 Prior to 1945

6.0 6.0 1.7

50 28 46

3.6

9

1.6

13

2.0

5

> 4*o

7

1.7

18

2.3

4

The data do not point to an important and continuing diminution in symptom prevalence as time from cessation of tobacco use increases, indicating that the effects probably become manifest shortly after discontinuance as suggested by casual clinical impressions. Rapid improvement in respiratory symptoms following discontinuance of smoking differs markedly from the picture presented by lung cancer, where reductions in risk are expressed gradually over an extended time interval [16]. Urban-rural residence

Norway experiences by international standards low death rates from chronic and non-specific lung disease (CNSLD), characterized by a modest urban excess in mortality from these causes. The excess mortality from CNSLD in urban Norway has never reached the British level, where the high mortality in large conurbations is a prominent landmark of 2Oth-century vital statistics. The behavior of the symptom data follows the mortality data in some respects. The survey revealed no important or significant urban-rural difference in symptoms of PCP or chronic bronchitis (top half of Table 13). The 95 per cent confidence limits of the ratios included 100 and nothing in the more detailed results leads one to doubt that chance variation could account for the observed findings. Contrasts within the

556

156 400 1024 329 695

103-121-142 84 93-103 93- 99-105 87- 97-108 93-100-108

914

93- 99-105

92-100-109

420

87- 96106

292 695 235 460

102

99-111 97-105 99-113 96-105

98-l 19-145

88908788-

394

671

299

456 157

94104-115

89- 96-94

90-101-113

98-107-117 103-121-142

61% 80- 97 94105-117 76 94-116 96-109-124

96-126-166

93-104-117 78- 91-106

72- 87-104

54 88-100-113 76 96-121 87-101-117

253 75 178

246 128

303 Birthplace 90-102-116 162 80- 95-113 108 329 94 235

103

80- 97-118

121

157 54

403

671 251

518

Place of residence 83- 95-109 215 8388-108-133 94 74 97-114 97-128

100

95-103-111 102-116-132

1127

loo

100

1645

loo

Note: 95 per cent confidence limits shown in italics. *Adjusted for age and tobacco use at time of inquiry.

Urban, total Oslo Other urban Rural, total Non-farm Farm

Urban, total Oslo Other urban Rural

Norway

Both sexes No. with Ratio symptoms

Persistent cough and phlegm Both sexes Males Females No. with No. with No. with Ratio symptoms Ratio symptoms Ratio symptoms

77- 92-109 69- 96-130 73- 90-111

82- 97-115 83-104-131

78-102-133

85-105-130 75-112-161

100

135 43 92

77

139

59

88 29

227

Chronic bronchitis Males No. with symptoms Ratio

91-110-132 66- 96 135 94117-145

63- 84-111

90-109-132

67- 92-124

70- 85-126 89-113 55-

100

118 32 86

51

107

44

69 25

176

Females No*with Ratio symptoms

TABLE13. RESPIRATORY SYMPTOM PREVALENCE RATIOS*BYPLACEOF RESIDENCE AND BYBIRTHPLACE: COMBINED SAMPLES

Y

s

;

B

i 0”

E

B z

%

7 8 & B

536

WILLIAMHAENSZELand ANNA HOUGEN

several regions revealed no striking urban-rural differences and these could also be readily attributed to chance variation. The evidence for elevated symptom prevalence in the one metropolitan area (Oslo) seems inconclusive; the PCP ratio of 116 was of borderline significance, and accompanied by an apparently normal prevalence for chronic bronchitis. Even if a difference in respiratory symptom prevalence between Oslo and other urban areas were established by more extensive observations, it would probably prove to be of modest dimensions. Urban-rural contrasts of age-adjusted rates specific for tobacco use are presented in Table 14 and Fig. 3. The experience of cigarette and mixed smokers was taken to quantitate the effects of tobacco use among men. These results reproduced almost TABLETS. AGE-ADJUSTED*RE.SPIRATORYSYMPTOMPREVALENCEPER 100 PERSONSBYSEX,URBAN-RURAL RESIDENCE AND RATEOFTOBACCO USE:COMBINEDSAMPLES

Total urban

Non-smoker Ex-smokert Current smokert < 10 g/day 10-19 g/day 20+ g/day Non-smoker Ex-smokert Current smokert < 10 g/day lo-19 g/day 20+ g/day

Males Other urban Oslo

3.2 5.6

2.9 7.9

11.4 15.7 21.4

11.4 15.1 23.2

Females Other Oslo urban

Persistent cough and phlegm 3.2 3.4 2.1 2.5 4.7 5.5 4.0 5.5 11.5 15.9 20.7

9.1 15.2 22.0 >

2.4 3.2

5.4

5.5

5.3

6.7

15.5

11.4

14.2 0.9 1.2 2.0

1.8 2.8 4.1

5.1 for the total sample of Norwegians.

and mixed smokers only. PERCENT

SYMPTOM

MALES NON-SMOKERS CVRRENT

2.0 3.0

Rural

13.3

Chronic bronchitis 0.9 0.6 1.3 2.0

0.7 1.4

*Adjusted by direct method to age distribution tcigarette

Rural

Total urban

SMOKERS

clog/DAY IO-19 g/DAY ZZOg/DAY

NON-SMOKERS CURRENT SMOKERS (10 g/DAY IlOg/DAY

FIG.3.

POSITIVE

Prevalence of Respiratory

537

Symptoms in Norway

exactly the experience of pure cigarette smokers, but inclusion of mixed smokers was desirable to augment observations on rural residents in the 20+ g category. Pure pipe smokers were excluded because of the symptom deficit indicated in Table 5, which meant that rural male prevalence in the 1-9 and 10-19 g categories would have been depressed by greater rural representation of pipe smokers. Urban males started smoking at earlier ages so that adjustment for this factor would minimize still further the urban-rural differentials in symptom prevalence. The results do not point to a concentration of the small overall urban-rural differences in symptom prevalence in any particular smoking category; specifically, the question of interaction between urban residence and smoking to produce higher symptom prevalence than would be predicted from the sum of two separate effects does not arise. Urban-rural migration

To check the role of early environmental exposures in symptom prevalence, the second half of Table 13 substituted birthplace for current residence. A query on farm residence permitted the distinction between persons born or not born on a farm. The contrasts of symptom prevalence by birthplace, including those between farm- and other rural-born were essentially negative. The one potentially significant feature was the higher prevalence ratio for persons born in Oslo. The farm-born in the U.S. who moved to metropolitan areas have been described as presenting above-average lung cancer risks [17, 181. The possible presence of a similar phenomenon for respiratory symptoms was explored in Table 15, although the absence of an association with either birthplace or current residence made it doubtful that migration history would discriminate. While the farm-born moving to Oslo or Bergen did display higher prevalence, the finding is only suggestive and sampling variation remains the probable explanation. Repeated confirmation in other settings would be required to establish the presence of this type of migrant effect for respiratory symptoms. TABLE 15. RESPIRATORY SYMPTOM PREVALENCE RATIOS* FOR SELECTED COMBINATIONS OF BIRTHPLACE ANDCURRENTRESIDENCE:TOTALMALESAND FEMALESCOMFJINED

Birthplace

- Current residence

Persistent cough and phlegm No. with Ratio symptoms 100

Chronic bronchitis No. with Ratio symptoms

Norway

- Norway

1645

100

403

Farm

- Rural

91 -

99 - 108

560

87 - 103 - 121

147

Farm

- Urban (excl. Oslo, Bergen)

79 - 100 - 126

74

Farm

- Oslo, Bergen

92 - 119 - 154

61

Rural non-farm

- Rural

89 - 101 - 115

240

Rural non-farm

- Urban

72 -

89 - 110

89 404 152

Urban

- Urban

92 - 102 - 113

Urban

- Rural

80 -

94 - 111

Note: 95 per cent confidence limits shown in italics. *Adjusted for sex, age and tobacco use at time of inquiry.

49 -

85 - 138

16

62 - 110 - 182

15

67 -

90 - 120

49

71 - 108 - 159

26

76 70 -

90 38

94 - 116 99 - 137

538

W~UIAM HAENSZELand ANNAHOUGEN

Region of residence

Table 16 summarizes the variation in symptom prevalence among seven regions of Norway. In the absence of systematic urban-rural differentials within each region, review of the overall regional experience suffices. The major feature was the higher risks in the northernmost regions (VI and VII). The excess risk appeared in each sex-age group and for all smoking categories, and was more marked for women than for men and for chronic bronchitis than for PCP; if anything, the regional excess was less pronounced among heavy smokers. Smokers took up the habit at an earlier age in Finmark (VII) but not Nordland (VI) and control for age started smoking left the findings unchanged. The size of the chronic bronchitis effects and their consistency with the PCP results incline us to accept with fewer reservations the Nordland-Finmark data than those for Oslo, where a normal symptom level of chronic bronchitis prevailed. The other regional findings either did not deviate importantly from the national experience or did not display a consistent age-sex pattern for both sets of symptoms. Table 16 contains a parallel presentation by region of birth, which also identifies the subgroups still residing in the same region. The overlap in birthplace and current residence classtications makes the concordance in the two sets of results less than surprising. Although the contrasts of lifetime residents did not magnify the regional differences, their persistence in the most homogeneous exposure groups (Col. 5 of Table 16) strengthens the inference of true differences in risk for Nordland (VI) and Finmark (VII), and possibly for Oslo city. The ability to detect traces of regional variation in symptom prevalence places in bolder relief the absence of noteworthy urban-rural differences. Social class, occupation

Respondents were asked to report their major lifetime occupation and the replies from males were classified by a standard occupation-industry code of the Norwegian Central Statistical Office. Prevalence ratios were computed for the five groupings listed in Table 17. Professional, administrative, and clerical workers were designated because of social-class differences in respiratory diseases, miners because they constitute a known high-risk group and seamen because they represent a significant fraction of the work force exposed to severe weather conditions. Service employees were segregated from ‘other workers’ because they may have been less exposed to industrial pollutants. The below-average PCP prevalence and the extremely low ratio for chronic bronchitis among professional, administrative, and clerical workers agrees with other evidence on social class effects in chronic respiratory diseases [19]. The low PCP risk among non-smokers in this social stratum tends to rule out the possibility of a statistical artifact through failure to control some specific facet of smoking history, although professional and administrative workers did take up cigarette smoking at a later age than other males (13 per cent prior to age 17 vs 25 per cent). Adjustment for this latter factor raised the PCP prevalence ratio in the professional-administrative class slightly from 80 to 82. The results for miners would be consistent with their assignment to a high-risk category by other investigators [20]. The data on seamen do not set their experience apart unequivocally from other males.

B 5

E 3

i?

8 ,o

iz!i B

z .$ 8 2 % 8 3 d $ 2

LII OL ii1 IT LZZ

IPI Ml LZL IOZ LP

-

LIT 601 P8Z PZl 9z 001

-

-

STIVN

PSI 531 811 56 68 E6 L6 001

o!lm

LfZ f661991 fZI 601 RII 8ZI -

suro$durXs oyex q,!M ‘ON 1Wo.I syyxIo3q yIo1q3

oz ZP ZE 09 101 P6 PS EOf,

swo~dtis qYM ‘ON

muap!sap s!]!ymo.Iq 3yo.Iq3

L6 98 LL IL El L9P L81 s El EL8

:SdflOWl

L5 IEI 6L SEZ ZSE IO& VII 69Zl

EEI lZl 58 EOI 68 16 OZT 86 -

zor zor 89 06 08 I8 66 E6

PSZ

LO1 66 9 9 9E

suroldruds YIP’ ‘ON

98 98

FZI Oil P6Z OII f6

-

-

‘Ll

L8 58 f9 69

a-IS,,=

‘~A!$%I~S!U!UIp~

$11 - +oor

86 Wl a 08

001

PLS 982 11 ZS ZLI LZIT

61 6S 001

o!w’d

W0.L

oy=w -UON 3m~s!s~~d

AlrOLWIdSa~

suro~dtis ql!M ‘ON

IL LPI 68 z9z 6ZP 96E lSZ SW1

091 PM IO1 ETI fOi SO1 Zfl

9Zl ZZI 28 ZOl t76 S6 911 001

o!mI

-

-

66 fO1 99 06 S8 98 ZOI

SJ%f.IOM

sJaql0

S’X.Il~

~fA.IaS

uamaas 'W ‘pq.Iq3

r(EM.ION

‘IW.JO!SSaJOJd

hM.ION

IIA IA A AI III II @ISO) I =0%x

*br!nbu! JO atug le asn omvqol pm ‘a% ‘xas JOJ palsnfpv, ‘s3!p?l! u! uMoys slpuy amapguo3 Jua3 lad ~6 :a)oN

AO.LdKAS

suxojdurh qI!M ‘ON

&'I - 501 921 - EOI OO9-sLz-roT IIT - IS 9IT - P8 -

PUE @&IO3

SOT IOI S9 16 18 98 for

tI3NTIVA3~d

- 9Zl - Lll - 08 - ZOT 06 - S6 - IZI 001

o!wx

asuapysaa u@a[yd pm q8nos lua)s!sJad

F3o.I.

1st 9fI 66 PII 001SO1 ZPI

*SOIJ,Vll

pm

SJ~~OUIS-X9

tialyd

lOtI

swo$dmXs qI!M ‘ON

tz St91

TZI 181 26 662 68E

aBmL33?3S

PZI - El1 - &Of III - 96 - f8 9SZ - 011 - 9f ZZI - 16 - L9 P6 - 6L - L9 001

oyQI ]U~.I.IIl~

-

WNOILVdll330

PLl MPI LOT LTI 66 ZOI 4-H M)l

SJ%fOLUS

swo~dwds yI!M ‘ON

lVIOJ_

96 FOI I8 FL FL 9L PL

suxo$dwKs o!lwI qJ!M ‘ON wq jo uoPaJ u! BurpfsaJ IIPS a=ldqv!8

540

WILLIAMHAENSZELand ANNA HOUGEN

The heterogeneous composition of ‘other workers’ warrants consideration of their modest elevation in symptom prevalence, since the pooled data may conceal high risks in some occupations. The detailed occupations in this category were scanned for above-average PCP prevalence and occupations with a crude prevalence rate of at least 12.0 per cent identified. The resultant list included smelters, iron and other metal workers, textile weavers, stone cutters, brickmakers, concrete and glass workers, leather workers, operators of bulldozers, etc., and bakers, all of which have been reported as at high risk to chronic respiratory diseases [19] ; several of the occupations involve exposures to dusts and metals which may play an etiological role in respiratory illnesses [21]. These ‘high risk’ occupations accounted for the excess prevalence among ‘other workers’ and their exclusion reduced the crude rate for ‘other workers’ to 8.8 per cent, close to the figure for all Norwegian males. The inability to detect interaction between urban residence and smoking history for respiratory symptoms in Norway may stem from the absence of specific urban environmental agents. The ‘high risk’ occupations presumably subject to work-related exposures offer another opportunity to search for interaction. By definition, the list of ‘high-risk’ occupations assembled have above-average symptom prevalence, so that a test must rely on a demonstration of larger increments in prevalence by amount smoked for the ‘high-risk’ occupations as opposed to ‘other workers’, disregarding the information on absolute rates. The table below summarizes the pertinent information on increments in PCP prevalence. The wider spread between non- and currentsmokers, combined with a possibly steeper progression of rates by rate of tobacco use for the ‘high-risk’ occupations suggests interaction, but the effect is not striking and the limited data in hand does not constitute convincing proof on this point. ‘High-risk’ occupations Rate Increment Non- and ex-smokers Current cigarette and mixed smokers 1-9 g/day 10-19 g/day

20+ g/day

5.0 13.9

‘Other workers’ Increment Rate 4.6

8.9 7.6

10.6

6.0 5.4

g’;.> 21.5

DISCUSSION

The respiratory symptom questions have been used by other investigators who have validated the findings by lung function tests and observations on sputum production and abnormalities. The mail query method yields patterns of risk similar to those elicited by structured personal interviews and other screening and detection procedures, although estimates of absolute prevalence differ with the techniques employed. These issues have been discussed elsewhere [6] and we feel that the present Norwegian data can be interpreted and related to those assembled by other investigators. Although the study did not incorporate validation through lung function tests and other detailed clinical observations on subsamples, work of this type was carried out in Bergen, Norway in 1959 [22]. The absence of major urban-rural differentials and the persistence of the same smoking-class gradients in prevalence throughout Norway lead one to believe that the affirmative validation findings of the Bergen study can be

541

Prevalenceof Respiratory Symptomsin Norway

extrapolated to this survey. Further validation of the respiratory symptom data will come from assessment of their prognostic value for subsequent mortality. This is an integral part of the present investigation and the results of 5 yr of follow-up will be reported in a later communication. While the two criteria, PCP and chronic bronchitis, yielded similar risk gradients, too much emphasis should not be placed on this point because the symptoms have been defined by continued positive responses to a series of questions. Parallel symptom patterns are inevitable, so long as the transition rates from PCP+ to chronic bronchitis+ do not vary markedly among subpopulations. The relationship might well be phrased in reverse to state that consistency between the two symptom criteria constitutes evidence for reasonably constant transition rates from PCP+ to chronic bronchitis+. The respiratory symptom prevalence for non-regular smokers provides a benchmark and point of departure for discussion. The rates for non-smokers in Norway, while lower than reported from Britain and Finland cannot be termed exceptionally low; they were of the same order of magnitude as those for U.S. native-born. Table 18 summarizes some comparisons with published data from Britain and the U.S. The difference between British and Norwegian male (but not female) non-smokers has been overstated, since the British data represented the pooled experience of nonsmokers and pipe smokers, in effect non-cigarette smokers. The same formal definition held for the U.S. data, but the small amount of pipe smoking in that country makes this distinction of little practical consequence. The Norwegian male figures corresponding to the British definition of non-smokers are 5.5 and I.3 for PCP and chronic bronchitis respectively, thus bringing the British-Norwegian male contrasts into closer alignment with the female results. Adjustment for the larger urban component in the British experience would scale down, but not eliminate, the British-Norwegian differentials for non-smokers. TABLE18.

AGE-ADJUSTED* RESPIRATORY SYMPTOM PREVALENCE RATESPER100 PERSONS BY SEXAND TOBACCO USE:NORWAY, U.S. NATIVE-BORN AND BRITAIN Males U.S. Norway native-born Britain

Females U.S. Norway native-born Britain

Persistentcough and phlegm

Non-smokers

3.4

3.6

8.2

2.4

1.9

4.4

Ex-smokers Current smokers < 20 cigarettes/day 20+ cigarettes/day

5.5

6.7

13.0

3.8

3.5

6.5

7.9 23

5.8 15

13 22

11 21

25 32

13 26

Chronic bronchitis Non-smokers

0.9

1.0

3.1

0.8

0.5

Ex-smokers Current smokers < 20 cigarettes/day 2O+cigarettesJday

1.3

1.9

6.8

1.8

1.3

2.2 3.6

2.3 4.5

3.1 3.1

8.1 8.2

2.5 11

1.5 1.6

5.5 10

*Unweighted average of age-specific prevalence rates for each smoking class was used for age adjustment; rates for the U.S. native-born taken from (6) and for Britain reconstructed from data in (4).

542

WILLIAM HAENSZELand ANNA HOUGEN

The Norwegian data for both sexes project the classic pattern of a rise in symptom prevalence with amount smoked. While the liner nuances of local and national practices of tobacco use preclude precise comparisons of published data, the Norwegian results, including intermediate risks for ex-smokers, seem fully consistent with experience elsewhere. The distinctive Norwegian features of pipe smoking and use of hand-rolled cigarettes did not blur the picture of strong smoking effects. The somewhat higher risks for ‘pure’ cigarette smokers indicate that mode of use plays a role in determining symptom response, but we would emphasize how small the differences among cigarette, pipe and mixed smokers were. The higher risks for inhalers and persons who began smoking at an early age round out and reinforce the impression of strong, readily identifiable effects of tobacco use on respiratory symptom prevalence in Norway. These and other points of agreement (social class effects, etc.) with data collected in other countries underscore one aspect of the Norwegian data divergent from British experience (the U.S. data on this point have not been examined)-the inability to demonstrate a clear-cut urban excess in symptom prevalence, which was more pronounced among heavy smokers (interaction). The statement certainly holds for the smaller Norwegian communities lacking a big-city environment and may even apply to Oslo city. Closer inspection indicates, however, that a small, equivocal interaction with smoking history in Oslo might be quite consistent with the British experience. Air pollution in Oslo measured by ‘smoke’ indices [12] would rank as low or very low on a Cpoint British scale (very low, low, moderate, high). In Britain interaction of smoking history with air pollution on respiratory symptom prevalence is not easy to discern in cities with ‘low’ pollution and this feature is best expressed in cities exposed to moderate or heavy air pollution [4]. The combination of a clear Norwegian smoking class gradient without the type of urban-smoking history interaction described in Britain would tend to incriminate an agent(s) in the atmosphere of some English cities that is absent or present only in low concentration in Norway. The insensitive classification of Norwegian urban and rural areas may be partly responsible for the failure to find stronger urban-rural differences in symptom prevalence, but this factor would operate in all countries. Conceivably, community classilications based on topography, climate, specific exposures to industrial wastes, etc., might have pinpointed localities with excess risks. However, general population samples would not be an efficient tool for this purpose, which could be better done by studies in depth of communities exposed to well-defined conditions. The conjunction of high respiratory symptom prevalence and high lung cancer risks in Finmark is an incidental finding worth noting [12]. However, the other epidemiologic features provide no reason to believe that factors responsible for higher symptom prevalence in populations exposed to severe winters in the far North (Nordland, Finmark) are etiologically related to those contributing to elevated urban prevalence. The correspondence in urban-rural patterns presented by symptom prevalence and Norwegian mortality statistics for respiratory diseases (the separate figures for CNSLD were unavailable), including the feature of stronger expression of effects among males, is summarized by the urban-rural ratios of risks given below. To maintain comparability of symptom and mortality data, the former were not adjusted for smoking history:

Prevalence of Respiratory

Symptoms in Norway

543

Males

Females

Mortality from respiratory diseases (ICD 241,470-527, ages 40-69,1959-1962)

1.63

0.93

Persistent cough and phlegm

1.38

1.15

Chronic bronchitis

1.28

1.05

The low Norwegian mortality from CNSLD, which accompanies the modest urban-rural differential in mortality, also follows the symptom data in describing a situation much different from Britain. The British-Norwegian configurations of data prompt the following conjecture: The magnitude of interaction between smoking history and urban residence on symptom prevalence will discriminate better than absolute level of symptom prevalence between countries with high and low mortality from CNSLD. If true, this could support the idea advanced earlier by Reid and Mork l-22,23] that the suspect urban agent acts with tobacco smoke as a co-factor or promoter not only to produce symptoms but also to play a role in determining the progression from symptoms to morbidity to death from CNSLD and related respiratory and cardiovascular causes. SUMMARY

Information on respiratory symptoms using a standardized questionnaire was collected by mail from two groups in Norway-a general population sample of persons born 1893-1929 and sibs of migrants to the United States. The general population and sib samples yielded similar findings and were combined for detailed analysis. An excess symptom prevalence among smokers rising with amount smoked, similar to findings reported from other countries, was noted. Persons who started to smoke at young ages and inhalers also presented higher risks. Cigarette smokers exhibited higher prevalence than pipe smokers after adjustment for rate of use, but no differences between hand-rolled and commercial cigarettes were found. Some regional variation not accompanied by important urban-rural differences in symptom prevalence was observed. Urban residence and a high rate of tobacco use did not interact to produce prevalence rates greater than expected from the separate contributions of these two factors. The consistency of the symptom prevalence information with Norwegian mortality data for respiratory diseases is described and some possible implications of the absence of an urban-rural differential or interaction of urban residence with smoking history in the Norwegian data are discussed.

1.

2. 3. 4. 5.

REFERENCES Reid DD: Studies of disease among migrants and native populations in Great Britain, Norway and the United States. I. Background and design. In Haenszel W (Ed): Epidemiological Study of Cancer and Other Chronic Dil. Nat Cancer Inst Monogr No. 19, US Dept of Health, Education and Welfare, Public Health Service, Wash DC US Govt Print Off, pp. 287-290,1966 Medical Research Council’s Committee on the Aetiology of Chronic Bronchitis. Brit Med J 2: 1665,196o Chronic Bronchitis in Great Britain. A national survey carried out by the respiratory diseases study group of the College of General Practitioners. Brit Med J 2: 973-979, 1961 Lambert PM, Reid DD: Smoking, air pollution and bronchitis in Britain. Lancet 1: 853-857, 1970 Holland WW, Reid DD, Seltser R et al: Respiratory disease in England and the United States. Studies of comparative procedures. Arch Environ Health 10: 338-343, 1965

544 6.

7. 8. 9. 10. Il. 12. 13. 14. 15.

16.

17. 18. 19. 20.

21. 22. 23.

WILLIAMHAENSZELand ANNA HOUGEN Reid DD, Cornfield J, Markush RE et al: Studies of disease among migrants and native populations in Great Britain, Norway and the United States. III. Prevalence of cardiorespiratory symptoms among migrants and native-born in the United States. In Haenszel W (Ed): Epidemiologlcal Study of Cancer and Other Chronic Diseases. Nat Cancer Inst Monogr No. 19, US Dept of Health, Education and Welfare, Public Health Service, Wash DC, US Govt Print Off, pp. 321-346, 1966 Holland WW, Reid DD: The urban factor in chronic bronchitis. Lancet 1: 445-448, 1965 Winkelstein W Jr, Kantor S: Respiratory symptoms and air pollution in an urban population of northeastern United States. Arch Environ Health 18 : 760-769, 1969 The Cancer Registry of Norway: Manuscript in preparation, 1971 Magnus K, Hougen A, Andersen A et al: A study of disease in migrants and their siblings: development of sibling rosters. J Chron Dis 23 : 405410, 1970 Sawicki F: Cough and phlegm in smokers in Cracow in relation to sex, age, smoking period and number of cigarettes smoked. Bull Int Epid Assn 19: 26-36, 1970 Pedersen E, Magnus K, Mork T et al: Lung cancer in Finland and Norway. An epidemiological study. Acta Path Microbial Stand Suppl 199: 7-74, 1969 Schwartz D, Flamant R, Lillouch J et al: Results of a French survey on the role of tobacco, particularly inhalation, in different cancer sites. J Nat Cancer Inst 26: 1085-l 108, 1961 Doll R, Hill AB: Mortality in relation to smoking: Ten years’ observations of British doctors. Brit Med J 1: 1399-1410 and 1460-1467, 1964 Hammond EC: Smoking in relation to the death rates of one million men and women. In Haenszel W (Ed): Epidemiologlcal Study of Cancer and Other Chronic Diseases. Nat Cancer Inst Monogr No. 19, US Dept of Health, Education and Welfare, Public Health Service, Wash DC, US Govt Print Off, pp. 127-204, 1966 Kahn H: The Dom study of smoking and mortality among U.S. veterans: Report on eight and one-half years of observation. In Haenszel W (Ed): Epidemiologlcal Study of Cancer and Other Chronic Diseases. Nat Cancer Inst Monogr No. 19, US Dept of Health, Education and Welfare, Public Health Service, Wash DC, US Govt Print Off, pp. l-125, 1966 Haenszel W, Loveland DB, Sirken MG: Lung cancer mortality as related to residence and smoking histories. I. White males. J Nat Cancer Inst 28: 947-1001, 1962 Haenszel W, Taeuber KE: Lung cancer mortality as related to residence and smoking histories. II. White females. J Nat Cancer Inst 32: 803-838, 1964 Registrar-General: The Registrar-General’s Decennial Supp., England and Wales, 1951, Occupational Mortality. Part I, 1954 Ministry of Pensions and National Insurance: Report on an Inquiry into the Incidence of Incapacity for Work. Part II: Incidence of Incapacity for Work in Different Areas and Occupations, 1965 Hammond EC, Selikoff IJ: The effects of air pollution. Epidemiological evidence. Proc Lnternat Conf of Pneumoconiosis, Johannesburg, 1969, pp. 188-193 Mork T: A comparative study of respiratory disease in England and Wales and Norway. Acta Med Stand Suppl384: l-100, 1962 Reid DD: Environmental factors in respiratory disease. Lancet 1: 1237-1242 and 1289-1294, 1958