At baseline, 57% (n=542) of the invited patients agreed to participate in PRIMAIR. At the 18-month follow-up, 78% (n=425) of the initial participants replied, of whom 69% (n=293) completed the LINQ both at baseline and after 18 months. Non-responders to the LINQ were older than responders (73.4 years vs. 71.1 years, P=0.009) and more often lacked a specific, assigned GP at their PHCC (63.2% vs. 75.6%, P=0.012). The non-response rate was independent of gender, lung function, smoking status, medications, contacts with COPD nurses and/or physiotherapists, level of education, and symptoms/health status (CCQ/CAT/mMRC scores).
Analysis of the study population as a whole
Patient characteristics at baseline and over time
Table 1 shows the main baseline characteristics of the 293 participants who completed the LINQ both at baseline and at 18 months. A subanalysis by gender (not shown in Table 1) revealed that women (57%) had smoked less than men (number of packyears 30.7 vs. 39.2, P=0.001), and more women than men had depression and/or anxiety (27.1% vs. 8.7%, P=0.001). Women also had a lower level of education than men (9 years or less, 62.9% vs. 45.9%, P=0.005). There were no gender differences in lung function measures, medications, current smoking status, comorbidities (other than depression and/or anxiety), or patterns of contacts with health care professionals (GPs, COPD nurses, physiotherapists).
Table 2 shows the changes in patients’ smoking intensity, exacerbations, body mass index, comorbidities, health status, medication, and care provider contacts over the 18-month study period. A total of 6.5% (n=19) of the patients had quit smoking (P<0.001), and 3.4% (P=n.s.) had resumed smoking during the study period. The prevalence of daily smokers was 30.0% at baseline and 27.0% at 18 months (P=n.s.). The proportion of patients who had consulted a physiotherapist increased from 11.9% to 17.4% (P=0.040), and patients who reported having an assigned GP decreased from 75.6% to 65.3% (P=0.001). The proportion of patients who had seen a COPD nurse remained the same over time (39.2% vs. 41.1%, P=n.s.).
Changes in LINQ scores
The mean total LINQ score of the 293 participants did not change over time (10.8 points at baseline vs. 10.9 points at 18 months, that is, 43.5% of the maximum 25 points) (Figure 1).
A logistic regression analysis was performed for the whole study population, with “a lower total LINQ score at 18 months than at baseline” (i.e., reduced information needs over the study period) as the response variable. Three variables from a preliminary univariate analysis were entered into the final multivariate regression model (Table 3). Reduced information needs were associated with having a baseline total LINQ score above the study population mean of 11 points (i.e., great needs for information) and having established contact with a physiotherapist during the study period. Getting a specific, assigned GP during the study period had a close to significant association with reduced information needs. Other variables, such as patients’ age, gender, lung function, medications, education level, and symptoms/health status (CAT/CCQ/mMRC), smoking status, and whether patients had had contacts with a COPD nurse were not associated with changes in information needs.
Both at baseline and 18 months, the patients expressed the greatest needs for information in the LINQ domains “Diet” and “Self-management,” moderate needs for information about “Disease knowledge” and “Exercise,” and the least needs for information about “Medicines” and “Smoking” (Figure 1). Only in domain “Disease knowledge” the change in the mean domain score over time was significant (percent of maximum domain scores: 36% to 51% [P<0.001]), which indicated an increased need for information about disease knowledge in the study population. However, we could not find any significant associations between the changes in the LINQ domain “Disease knowledge” and other clinically significant variables.
Analysis of the subgroups: Four levels of information needs
Although the mean LINQ scores for the study population as a whole remained stable over the study period, 43.3% of the patients reported reduced information needs, and 56.7% reported increased or stable information needs. To present this result more clearly, we divided the patients into four approximately equal-sized groups by baseline LINQ scores (Figure 2). The first group (Figure 2: a) had the lowest information needs (1-7 points, 19.8% of the patients), the second group (Figure 2: b) had medium low information needs (8-10 points, 32.8%), the third group (Figure 2: c) had medium high information needs (11-13 points, 22.2%), and the fourth group (Figure 2: d) had the greatest information needs (14-25 points, 25.3%). We then illustrated the movements of patients from each baseline group to the groups at the 18-month follow-up.
Most patients remained in the same group they had been in at baseline. Patients in the group with the lowest information needs throughout the 18-month study period had tendencies to have lower lung function measures, more exacerbations at baseline, more inhaled medications, more often an assigned GPs, and more contacts with physiotherapists than the patients with highest information needs or who tended to move towards higher information needs over the study period.