[Immunoglobulin E, viral antibodies and obstructive lung disease in adults. The relation between antibody level in serum, lung function and non-specific bronchial reactivity]

We have measured immunoglobulin E levels and respiratory virus antibodies and examined their possible role as risk factors for obstructive lung disease in adults. We observed that increased total serum IgE levels were associated with reduced lung function in subjects with obstructive lung disease, but not in asymptomatic subjects. Subjects sensitised to indoor allergens (house dust mites, cats and mould) had reduced lung function and increased, non-specific, bronchial responsiveness compared with individuals who were not sensitised to indoor allergens. Similar relationships were not observed for subjects sensitised to outdoor allergens (birch and timothy). The presence of respiratory virus antibodies was vaguely associated with reduced lung function, but was not related to increased, non-specific, bronchial responsiveness. In adults in this community sensitisation to indoor allergens is a strong predictor of reduced lung function and increased, non-specific, bronchial responsiveness, which are again closely associated with obstructive lung disease.

Effectiveness of postal smoking cessation advice: a randomized controlled trial in young men with reduced FEV1 and asbestos exposure

There have been few community-based randomized, controlled intervention trials for cessation in high-risk smokers. In such a trial we evaluated the effects of postal smoking cessation advice in smokers with asbestos exposure and/or reduced forced expiratory volume in one second (FEV1). All men aged 30-45 yrs (n=22,392) living in 34 municipalities in western Norway were invited to a cross-sectional community survey. Information on smoking habits and occupational asbestos exposure were obtained from self-administered questionnaires and measurements of FEV1 were performed with dry-wedge bellow spirometers. Among 16,393 participants we identified a group of 2,610 smokers with previous occupational asbestos exposure and/or adjusted FEV1 in the lowest quartile. A random half (n=1,300) received a mailed personal letter from a respiratory physician with a person-specific health advice to quit smoking and a pamphlet on smoking cessation. The remaining smokers (n=1,310) acted as controls and did not receive any information. Twelve months after the intervention, information on smoking habits was re-examined using a postal questionnaire. Among the respondents (n=2,282), smoking cessation was reported altogether by 13.7% in the intervention group versus 9.9% in the control group (p<0.01). The 1 yr sustained quit rate (no smoking at all during the last year) was 5.6 versus 35% (p<0.05), respectively. Measurements of carbon monoxide in expired air (with < or = 10 parts per million) confirmed self-reported nonsmoking in samples of the two groups. In a community this simple postal smoking cessation advice from a respiratory physician based on person-specific risk factors improved the 1 yr sustained success rate by 60% in identified high-risk smokers.

[Is it possible to predict nocturnal hypoxemia in snorers?]

We investigated to what extent the initial outpatient clinical findings could predict the results of in-patient nocturnal pulse oximetry in 108 snorers. 30.6% of the patients had oxygen desaturation index (ODI) > or = 5. The lowest SpO2 dip was strongly correlated to ODI (R2 = 0.729). Body mass index (BMI) and FVC were independently correlated to ODI. Relative risk of nocturnal hypoxemia was 3.2 at BMI > 32.0 kg/m2, and 3.0 at FVC < 87% of predicted value, compared with the whole referred group of snorers. Sensitivity of reported apnoea was 0.91, but specificity was only 0.21 with respect to hypoxaemia. Age, sex, hypersomnolence, morning headache, nasal congestion, smoking and consumption of alcohol or sedatives were not predictive of nocturnal hypoxaemia, neither were erythrocyte volume fraction, PaO2, nor PaCO2.

[Pulse oximetry. Reproducibility of two nights' recordings]

Reproducibility of nocturnal ear oximetry was evaluated in 22 in-patients who were screened for sleep apnoea syndrome. Ohmeda Biox II and Rikadenki printer were applied to each patient during two consecutive nights. The average oxygen desaturation index (ODI) was 6.2 for the first night, and 5.7 for the second night. Intra-individual standard deviation (SD) for ODI was 1.8. Average minimum SpO2 was 86.2% for the first night, and 85.6% for the second night, with an intra-individual SD of 2.1%. If ODI is defined as pathological when five or more, one patient (< 5%) with a pathological result during the first night reached a normal value on the second night, and one patient with a normal result on the first night showed a pathological result on the second night. In conclusion, the results from the first night of ear oximetry do not differ significantly from those of the second night in hospitalised patients with suspected sleep apnoea, provided that the method is appropriately standardised.

The semi-proportional hazards model revisited: practical reparametrizations

Reformulations of the semi-proportional hazards model are outlined making estimation and testing of stratum-covariate interaction effects easily accessible within the framework of the stratified Cox proportional hazards model. The method is illustrated by a practical analysis of variables influencing bronchial responsiveness with data from the Hordaland study of obstructive lung disease.

A 3-year clinical study of tunnel restorations

Tunnel restorations (n = 161) performed by 4 dentists were evaluated after a mean of 35 months in situ. Demineralized enamel was to be spared during preparation. The preparations were filled with a cermet glass ionomer (polyalkenoate) cement. Evaluation was carried out using clinical and radiographical criteria. During the observation period, approximately 16% of the restorations were replaced due to caries in dentin and 14% due to marginal ridge fracture. Cavitation in the approximal surface and/or increased radiolucency of the approximal enamel were observed in 34% of the remaining tunnel-restored teeth. A significantly higher frequency of failures were registered when treating patients with a high caries activity, where the initial lesion was large, and where the restoration did not reach the approximal surface.

Serum house dust mite antibodies: predictor of increased bronchial responsiveness in adults of a community

The purpose of this study was to determine whether the presence of serum specific immunoglobulin E (IgE) antibodies was associated with increased bronchial responsiveness in adults. We studied cross-sectionally a random community sample of 18-73 year old adults, of whom 83% (n = 489) performed bronchial responsiveness testing as well as serum measurements of five specific IgE antibodies. In the crude data, 39% of those with house dust mite antibodies (n = 18) had a bronchial responsiveness < or = 32 g.L-1 methacholine compared with 19% in subjects without any of the five specific IgE antibodies (n = 453). The corresponding percentages for subjects with timothy antibodies (n = 16) was 25%, birch antibodies (n = 13) 23%, cat antibodies (n = 10) 40% and mould antibodies (n = 2) 50%. When assessing the multivariate relationship between the presence of one specific IgE antibody and degree of bronchial responsiveness we used a semi-proportional hazards model with the response as a 20% fall in forced expiratory volume in one second (FEV1) from pretest value. Covariates included in the model were: gender, age, pretest FEV1, smoking habits, pack-years, season and other specific IgE antibodies than that examined. The presence of house dust mite antibodies was a significant predictor (p < 0.01) of increased bronchial responsiveness in never- and ex-smokers. Indoor allergic sensitization (house dust mite, cat and mould) was a significant predictor of increased bronchial responsiveness, while outdoor allergic sensitization (timothy and birch) was not. Excluding subjects with obstructive lung disease (n = 39) or including the covariate log total serum IgE as a potential confounder yielded the same result. Thus, in this community, indoor allergic sensitization rather than allergic sensitization per se was related to increased bronchial responsiveness after adjusting for other relevant covariates.

Serum respiratory virus antibodies: predictor of reduced one-second forced expiratory volume (FEV1) in Norwegian adults

BACKGROUND

The purpose of this cross-sectional study was to investigate whether the presence of serum complement antibodies was associated with reduced one-second forced expiratory volume (FEV1) in adults.

METHODS

From a stratified random sample of 18-73 year old adults, we performed measurements of serum complement fixing virus antibodies against influenza type A and B, parainfluenza type 1, 2, and 3, respiratory syncytial virus and adenovirus on 82% (n = 1239).

RESULTS

In the crude data, subjects having five of the seven virus antibodies had significantly lower lung function, given as sex-, age- and height-standardized residuals of FEV1 (SFEV1), compared with those without. After adjusting in addition for smoking habits, lifetime smoking consumption and season, the lung function levels were significantly lower in subjects with influenza type B and respiratory syncytial virus antibodies compared to those without (P < 0.01). Increasing influenza and respiratory syncytial virus antibody titres and increasing numbers of virus antibodies, respectively, were related to progressively lower lung function. Subjects with respiratory symptoms but without obstructive lung disease had lower antibody levels than subjects with obstructive lung disease, but higher levels than asymptomatic subjects. In a final multiple linear regression analysis adjusting in addition for respiratory symptom and disease status as well as for the other respiratory virus antibodies, the presence of respiratory syncytial virus antibodies was a significant predictor for reduced SFEV1 (regression coefficient: -0.226; SE = 0.112; P = 0.04). The magnitude of the effect on lung function remained after excluding subjects reporting symptoms of respiratory infection within 3 weeks prior to the examination (regression coefficient: -0.252; SE = -0.218; P = 0.25).

CONCLUSIONS

This cross-sectional community study indicates that respiratory syncytial virus infection or re-infection is an independent predictor for reduced lung function in adults of a wide age range.

Serum house-dust-mite antibodies and reduced FEV1 in adults of a Norwegian community

In this cross-sectional study we investigated whether the presence of specific serum IgE antibodies to house dust mite, timothy, birch, cat, and mold was associated with a reduced FEV1 in adults. We performed complete examinations on 82% of a stratified random sample of 18 to 73-yr-old adults (n = 1,239). Subjects with house-dust-mite antibodies had lower (p = 0.002) sex, age, and height standardized residuals of FEV1 (SFEV1) than those without any specific IgE antibody. This relationship did not differ significantly by sex, age, smoking habit, total serum IgE level, or season, and remained significant after excluding subjects with obstructive lung disease. For house-dust-mite antibodies we also observed a dose-response relationship between antibody levels and impaired lung function. In a final multiple linear regression analysis the presence of house-dust-mite antibodies was the only significant predictor (regression coefficient: -0.425; SE = 0.189; p = 0.02) of reduced SFEV1 after adjusting for smoking habit and lifetime tobacco consumption, season, total serum IgE level, and respiratory-symptom and disease status. Thus, house-dust-mite allergy is an independent predictor of reduced lung function in adults of a wide age range.

Total serum IgE and FEV1 by respiratory symptoms and obstructive lung disease in adults of a Norwegian community

BACKGROUND

The importance of total serum IgE level on lung function impairment has not been established in a general population.

OBJECTIVE

The aim of this cross-sectional community study was to examine the relationship between total serum IgE and level of lung function in adults, and whether this relationship differed by sex, age, smoking habits or by respiratory symptoms and disease status.

METHODS

A stratified random sample of 18-73 year old adults from the general population were invited to spirometry and serum analyses of total and specific IgE. Of 1512 subjects invited, 82% met and performed complete examinations.

RESULTS

Increasing level of total serum IgE was related to reduced lung function (P < 0.01) given as sex, age, and height standardized residuals of one second forced expiratory volume (SFEV1). Subjects with total serum IgE in the highest vs the lowest tertile had a mean SFEV1 of -0.28, corresponding to age and height adjusted FEV1 differences of 120 and 150 mL in women and men, respectively. The relationship between IgE and lung function impairment did not differ significantly by sex, age or smoking habits. In subjects with obstructive lung disease increasing level of total serum IgE was more negatively associated with lung function level than in subjects with respiratory symptoms alone. No relationship was observed in asymptomatic subjects. This was confirmed in a multiple linear regression analysis adjusting for sex, age, smoking habits and lifetime smoking consumption showing that SFEV1 was predicted by an interaction between total serum IgE level and symptom and disease status (P < 0.01). This interaction remained after excluding subjects (n = 105) having specific IgE antibodies.

CONCLUSION

Increasing total serum IgE level was associated with progressively lower lung function in a general adult population after taking other predictors of impaired spirometric lung function into account, though dependent on the subjects' respiratory symptom and disease status. Variation in prevalences of respiratory symptoms and obstructive lung disease in previous examined populations may thus explain conflicting observations of the association between total IgE and airflow impairment.

Cavity design and dimensions of tunnel preparations versus composite resin Class-II preparations

Fourteen pairs of extracted contralateral premolars with small, artificial, standardized, approximal 'caries lesions' were placed in 14 plaster jaws. Fourteen dentists made a mesial tunnel preparation and a distal composite resin preparation on one tooth and vice versa on the contralateral tooth. Less tooth substance was removed in the tunnel preparations than in the class-II preparations, but this difference was not statistically significant when the resin class-II preparations were made without occlusal retention. Twenty-five per cent of the tunnel preparations had residual 'caries', as opposed to 7% in the class-II composite preparations. Tunnel preparations with larger occlusal openings had less residual 'caries'. The morphology of the class-II resin preparations varied considerably, indicating a lack of precise descriptions in the dental literature.

Predictors of spirometric test failure: a comparison of the 1983 and 1993 acceptability criteria from the European Community for Coal and Steel

OBJECTIVES

To identify, in this general population study, predictors of spirometric test failure on the 1983 and 1993 acceptability criteria from the European Community for Coal and Steel (ECCS).

METHODS

All men aged 30-46 years living in western Norway (n = 45,380) were invited to join a cross sectional community survey. Respiratory symptoms, smoking habits, and marital status were found from self administered questionnaires, and measurements of forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) were performed with dry wedge bellow spirometers.

RESULTS

Altogether 29,611 subjects (65%) participated in this survey. Spirometric recordings were obtained in 26,803 of these (91%) of whom 1.6% failed the 1983 criterion (< or = 300 ml between the two highest FVC values) and 9.5% failed the 1993 criteria (< or = 5% or 100 ml between the two highest FEV1 and the two highest FVC values). Spirometric failures on both criteria were more prevalent in never smokers, single men, and subjects with respiratory symptoms than in ever smokers, married, and asymptomatic subjects. Failure of the 1993 criteria increased with age and declining height. Morning cough and phlegm, breathlessness uphill, attacks of breathlessness, and wheezing were related to failure of the 1993 criteria after adjustment for demographic variables and smoking, whereas only breathlessness uphill was related to failure of the 1983 criterion.

CONCLUSIONS

In men aged 30-46 years, spirometric test failures on both the 1983 and 1993 ECCS acceptability criteria occurred more often in never smokers than in smokers and ex-smokers after adjustment for other covariables. Spirometric test failure with the 1993 criteria also varied with height and most respiratory symptoms. The higher failure rates found in non-smokers, in shorter, and in single men could be due to late compression of the airways, smaller lung volumes, and poor general health, respectively.

Sequential and average attributable fractions as aids in the selection of preventive strategies

Methods for estimating the attributable fraction based on a multivariate model for a dichotomous response have been extensively developed in the last decade. These methods provide the means for calculating the total attributable fraction for a set of exposure variables possibly adjusted for a set of confounding variables. In this paper, a procedure for stepwise calculation of attributable fractions is outlined. The purpose is to study the effect on risk of disease of preventing several exposures, one at a time, in different orderings. This procedure introduces the concepts of sequential and average attributable fractions as aids for attributing the risk of disease to different exposures. The procedure and concepts are illustrated by analyzing data from a cross-sectional study on the prevalence of some frequent symptoms of lung disease.

Marginal ridge strength of teeth with tunnel preparations

The aim of this study was to investigate how the cavity design of tunnel preparations influences the strength of the marginal ridge and to assess the ability of a restoration to support it. Extracted upper premolars were divided into eight experimental groups and one control group of intact teeth. The size and the position of the tunnel preparation in relation to the marginal ridge differed in four of the groups, the preparations remaining unfilled, while teeth in four corresponding groups were filled. The teeth were subjected to incremental dynamic forces until fracture of the marginal ridge occurred at which point a 'maximum energy-index' was calculated. The relationship between this index and the experimental variables was analysed by the Cox proportional hazards model. The results for both filled and unfilled teeth indicated that the distance from the preparation to the marginal ridge is more influential on weakening the ridge than is the buccopalatinal size of the opening. A conservative tunnel restoration situated 2 mm from the marginal ridge, does not significantly weaken an otherwise intact tooth.

Cancer in association with polymyalgia rheumatica and temporal arteritis

In our prospective controlled study, a total of 185 patients with polymyalgia rheumatica (PMR) and temporal arteritis (TA) diagnosed during 1978-83 and their 925 matched controls were cross checked with the data files at the Cancer Registry of Norway at the end of 1987. Malignancy was registered in 27 patients (14.6%) and 131 controls (14.2%) between 1953 and the end of 1987. Malignancy was registered in 16 (24.6%) of the patients with biopsy demonstrating arteritis temporalis. The hazard rate for developing malignancy after diagnosis for the whole patient population was not significantly different from the controls. The hazard rate for developing malignancy in patients with positive biopsy, however, was 2.35 times higher than in the controls (p = 0.036) and 4.40 times higher than the rest of the patient population (p = 0.007) (Cox proportional hazards model). The general long interval between diagnosis of PMR and/or TA and registration of malignancy (mean 6.5 years) is not consistent with a paraneoplastic mechanism.

Norwegian multicentre study of survival and prognostic factors in patients undergoing curative resection for gastric carcinoma. The Norwegian Stomach Cancer Trial

The influence of clinical and tumour-related variables on long-term survival in patients with gastric carcinoma was studied in a national prospective multicentre study. A total of 532 patients underwent curative surgery with a 5-year survival rate of 40 per cent. In a Cox multivariate analysis, age, stage of disease and macroscopic appearance of tumour were the only factors that independently affected survival. The estimated probability of survival to 5 years for a 50-year-old patient with a flattened tumour type and stage I disease was 71 per cent, but 50 per cent for an 80-year-old with the same tumour type and stage. For stage II disease the rates were 56 and 32 per cent, for stage III 39 and 13 per cent and for stage IV 19 and 3 per cent, respectively. A 70-year-old patient with stage II linitis plastica tumour type had an estimated probability of 5-year survival of 20 per cent. Stage of disease was the single most important factor in determining survival. Laurén histopathological type and location of tumour, Karnofsky performance index, weight loss and type of resection had no significant effect on long-term survival.

Is Lauren's histopathological classification of importance in patients with stomach cancer? A national experience. Norwegian Stomach Cancer Trial

This study was performed to investigate the association between Lauren's histopathological classification system and different clinico-pathological characteristics in patients with gastric carcinoma. We revealed that the percentage of intestinal type tumour (ITT) increased with advancing age (52% in patients less than 60 years compared to 73% of octogenarians (P less than 0.001)). The male:female ratio was 1.7 for ITT versus 1.3 for diffuse type tumour (DTT (P = 0.12)). ITT was more common in proximally (fundus) localized tumours than in distal lesions (77% vs 65%; P less than 0.05). The proportion of patients with ITT decreased with advancing stages of the disease (70% in stage I and II vs 52% in stage IV (P less than 0.0001)). More patients with DTT had tumour infiltration in the resection margin (21% vs 9%; P less than 0.001). Intestinal metaplasia was found in 48% of those with ITT compared with 28% of those with DTT (P less than 0.001). No association was found between Lauren's classification and the ABO blood group or between the tumour types and infiltration in lymphatic or blood vessels. We conclude that gastric adenocarcinoma occurs in at least two different biological forms and that differentiation between the two is of relevance for treatment.

A multivariate study of the prognosis of oral squamous cell carcinomas. Are blood group and hemoglobin new prognostic factors?

Because blood groups and hemoglobin concentration have been associated with the risk of the development of some cancers, this study evaluated the significance of ABO and Rhesus blood groups and hemoglobin concentration as prognostic factors in oral squamous cell carcinoma (SCC). The authors examined all registered primary SCC of buccal and maxillary alveolar mucosa in the Norwegian population between 1963 and 1972. The biopsy specimens from these patients were reevaluated and borderline cases excluded. The remaining 111 cases were included in the study, and features recorded on first admission were included in the survival analyses. ABO and Rhesus blood groups were found in 99 of these patients. Multivariate survival analysis showed that tumor size, hemoglobin concentration, stage, and Rhesus blood groups were significant prognostic factors, but sex, age, treatment, duration of symptoms, ABO blood groups, and clinical appearance of the tumors were not. The prognostic value of Rhesus blood groups and hemoglobin concentration has not been previously reported for oral SCC.

Patient and physician treatment delay in patients with stomach cancer in Norway: is it important? The Norwegian Stomach Cancer Trial

The effect and consequences of treatment delay were studied in 1165 patients with stomach cancer included in a Norwegian multicentre study. Median patient delay was 42 days; median doctor delay 37 days; and median total treatment delay 107 days. By Cox proportional hazards model analyses we found that an increase in weight loss was associated with an increase in total delay, whereas a more advanced stage of disease was related to a short total delay. Physician delay was more pronounced in women, increased with increasing Karnofsky performance index, but decreased in patients with stage-IV disease. By logistic regression analyses we found no association between delays and postoperative complication rate. The relationship between physician delay and postoperative mortality was statistical significant, with increasing number of deaths with decreasing delay. In conclusion, there is no evidence that long treatment delay is an important negative factor in relation to outcome of surgery.

Occupational dust or gas exposure and prevalences of respiratory symptoms and asthma in a general population

The relationship of occupational airborne, exposure to respiratory symptoms and asthma was examined using a self-administered questionnaire in a cross-sectional survey of a random sample (n = 4,992 subjects) of the general population aged 15-70 yrs of Hordaland county, Norway. The response rate was 90%. Twenty nine percent of the population had a history of occupational dust or gas exposure, 5% reported having been exposed to asbestos at work, and 4% reported quartz exposure. A history of occupational dust or gas exposure was associated with morning cough, chronic cough, phlegm when coughing, breathlessness on exercise, occasional wheezing and a physician's diagnosis of asthma after adjusting for sex, age, smoking habits and urban-rural area of residence. The adjusted relative odds ratios for the respiratory disorders in subjects exposed to dust or gas ranged from 1.6-1.9. The population attributable risk of occupational dust or gas exposure for the respiratory disorders ranged from 11-19%. The study indicates that respiratory disorders are independently associated with occupational airborne exposure in a Norwegian general population sample.

Occupational dust or gas exposure and prevalences of respiratory symptoms and asthma in a general population

The relationship of occupational airborne, exposure to respiratory symptoms and asthma was examined using a self-administered questionnaire in a cross-sectional survey of a random sample (n = 4,992 subjects) of the general population aged 15-70 yrs of Hordaland county, Norway. The response rate was 90%. Twenty nine percent of the population had a history of occupational dust or gas exposure, 5% reported having been exposed to asbestos at work, and 4% reported quartz exposure. A history of occupational dust or gas exposure was associated with morning cough, chronic cough, phlegm when coughing, breathlessness on exercise, occasional wheezing and a physician's diagnosis of asthma after adjusting for sex, age, smoking habits and urban-rural area of residence. The adjusted relative odds ratios for the respiratory disorders in subjects exposed to dust or gas ranged from 1.6-1.9. The population attributable risk of occupational dust or gas exposure for the respiratory disorders ranged from 11-19%. The study indicates that respiratory disorders are independently associated with occupational airborne exposure in a Norwegian general population sample.

Factors related to and consequences of weight loss in patients with stomach cancer. The Norwegian Multicenter experience. Norwegian Stomach Cancer Trial

Of 1165 patients with stomach cancer included in a national, prospective multicenter study with 51 surgical units participating, information about weight loss before diagnosis was available for 855 patients (73%). Median weight loss was 5 kg; 259 patients (31%) experienced no weight loss. By logistic regression analysis the authors found that weight loss increased with age and advancing stages of disease (TNM Stage I-IV), with decreasing Karnofsky index, in Lauren's diffuse versus intestinal tumor type, and with tumors located at the cardia/esophagus. Increasing weight loss reduced the resectability rate significantly, but no association between weight loss and postoperative complication rate was found. The odds ratio for postoperative mortality was 2.5 to 1 for the weight loss group 5 to 10kg versus 0 kg. In conclusion, weight loss reflects a less favorable tumor status. Weight loss did not increase postoperative morbidity but did lead Weight to a higher death rate after surgery.

Postal survey on airborne occupational exposure and respiratory disorders in Norway: causes and consequences of non-response

STUDY OBJECTIVE

The aim was to examine causes for non-response in a community survey, and how non-response influences prevalence estimates of some exposure and disease variables, and associations between the variables.

DESIGN

This was a cross sectional questionnaire study with two reminder letters. The questionnaire asked for information on smoking habits, occupational airborne exposure and respiratory disorders.

SETTING

A random sample of 4992 subjects from the general population aged 15-70 years of Hordaland County, Norway.

MAIN RESULTS

The overall response rate was 90%, with a 63% response to the initial letter. The response rates to the first and second reminder letters were 56% and 36% respectively. In 20% of the non-respondents an uncompleted questionnaire was returned with cause for non-response; in two thirds of these the cause for non-response was that the subject was not resident at the mailing address. A home visit to a random sample of 50 urban non-respondents provided further information on 29 subjects. A wrong address at the Central Population Registry and the subject's feeling of lack of personal benefit from a postal survey were the major reasons for non-response. Smokers were late respondents and subjects with respiratory disorders tended to be early respondents.

CONCLUSION

The main reasons for non-response were a wrong mailing address and a feeling of lack of personal benefit from responding. Using only the initial letter would have changed the estimated prevalence of smokers from 39% to 35%. Otherwise, the estimated prevalence of the exposure and disease variables as well as the associations between them were only slightly changed after including the respondents to the first and second reminder letters.