Study Design
This was a cross-sectional survey of patients with chronic conditions aged > 18 years who required healthcare on at least two different levels (e.g., primary care and outpatient specialty clinic). Four different patient profiles were selected: patients with rheumatic disease (rheumatoid arthritis or spondyloarthritis) recruited from outpatient hospital clinics, patients with IBD recruited from outpatient hospital clinics, patients with HIV infection recruited from HIV units or internal medicine clinics and patients with DM who also had a diagnosis of cardiovascular or chronic renal disease, selected from primary care clinics. The survey was handed to patients by their physician or nurse. Patients were instructed to read the survey and complete it voluntarily and anonymously from home, and then return it by pre-paid mail (see the Electronic Supplementary Material (ESM) 1).
Investigators from the specialties involved in the care of these patients were selected from a wide sample of outpatient hospitals and primary care clinics across the different regions of Spain to be representative of the Spanish healthcare system (ESM 2). To minimize selection bias, each physician or nurse (25 from rheumatology teams, 23 from gastroenterology teams and 25 from HIV clinics) handed the survey to the first 25 consecutive patients attending the clinic routinely, regardless of age, sex, disease severity or any other criterion. In addition, 48 primary care physicians each handed the survey to 13 consecutive patients with DM. Surveys were distributed and collected between May and September 2017.
Taking the IEXPAC 2015 (version 11 + 1) scale as the starting point, a survey was developed for this study with the participation of four physicians expert in the treatment of the abovementioned conditions. The survey was reviewed by members of three patients’ associations [the Spanish Association of Patients with Arthritis (CONARTRITIS), the Spanish Association of Patients with Crohn’s Disease and Ulcerative Colitis (ACCU) and the Spanish Federation of Patients with Diabetes (FEDE)] and by members of the Spanish AIDS Multidisciplinary Society (SEISIDA), a Spanish Society composed of patients with HIV and different healthcare professionals. Patients reviewed the content and relevance of the questions, and the language used, and suggested modifications or additional questions. Except for validated questionnaires including IEXPAC, which were not modified, suggestions from patients were included in the final version of the survey.
The study was reviewed and approved by the Clinical Investigation Ethics Committee of the Gregorio Marañón Hospital of Madrid, Spain. The study documentation included printed instructions and information for patients about the anonymous nature of the survey and aggregated data processing, which ensured that patient identification was not possible. As agreed by the Clinical Investigation Ethics Committee, the voluntary return of completed questionnaires was taken as implied consent to participate in the study. No clinical data were collected in this study.
Survey Instrument
The survey included multiple-choice questions that provided information about patient demographics, healthcare-related characteristics, disability (Barthel index) [13], follow-up by different specialists or nurses and treatment characteristics (ESM 3). As part of the survey, we included the whole IEXPAC questionnaire verbatim to assess patients’ experience with healthcare. The first version (version 11 + 1), which was used in this study, was developed in 2015 through a comprehensive process of literature review, expert panel work (including healthcare professional and social organizations, experts in quality of healthcare and patients with chronic conditions) and pilot and field work with 356 patients with chronic conditions to assess questionnaire content and face validity and reliability [2].
The IEXPAC scale was designed to be self-administered by patients. It has 11 items, plus an additional item for patients recently hospitalized. The items refer to the previous 6 months, except the question about hospitalization, which refers to the previous 3 years. For each item, patients responded on a 5-point Likert scale, and responses are transformed into scores: always (score 10), mostly (7.5), sometimes (5), seldom (2.5) or never (0). The scale yields an overall score (sum of individual scores for the 11 items divided by 11) between 0 (worst experience) and 10 (best experience) and allows identification of items in which improvement is needed.
Three factors are derived from the first 11 items of IEXPAC. Factor 1 is named “productive interactions” and refers to the characteristics and content of interactions between patients and professionals designed to improve outcomes. The productive interactions score is the average of the scores for items 1, 2, 5 and 9. Factor 2, the “new relational model,” refers to new forms of patient interaction with the healthcare system, through the internet or with peers. The score for factor 2 is the average of scores for items 3, 7 and 11. Finally, factor 3, named “patient self-management,” refers to the ability of individuals to cope with their diseases, manage their own care and improve their well-being, based on interventions mediated by healthcare professionals. The patient self-management score is the average of scores for items 4, 6, 8 and 10. The 12th item refers to continuity after hospitalization and is described separately. The IEXPAC questionnaire showed acceptable internal consistency (Cronbach’s α 0.76) and independence among the different factors [2].
Statistical Analysis
The survey was exploratory, and no formal hypothesis or pre-specified sample size was calculated. Conservatively, estimates based on a qualitative variable with 50% as the expected prevalence, a confidence interval of 95% and precision of 6% would give a sample size of 267 patients. Assuming that 15% of questionnaires would be incorrectly completed by patients, the total sample size for recruitment would be 314 patients. However, the expected response rate also had to be added to permit calculation of the final number of surveys required. Assuming a response rate of 50%, as seen in other surveys handed to patients [14, 15], our survey would have to be handed to approximately 628 patients. As we included four different patient profiles, we planned to hand the survey to approximately 2500 patients (625 patients with each background disease).
Descriptive information is displayed as mean ± standard deviation (SD) for quantitative variables, and as frequencies or percent for qualitative variables. For results of the IEXPAC questionnaire, we calculated the mean ± SD score for each item, the overall mean ± SD score and the mean ± SD scores for the three abovementioned factors. For this calculation, patients without a response to one IEXPAC item were imputed using the mean value of all respondents for that item. Patients lacking a response to more than one item were excluded from this analysis. The distribution of responses to individual items is also displayed, as is the sum of the percent of “always” plus “mostly” responses to each item.
We used Chi-squared or Fisher’s exact tests to compare proportions and the Student’s t test or analysis of variance (ANOVA) to compare continuous variables. Multiple linear regression models were used to assess the different demographic and healthcare-related variables influencing IEXPAC overall score, and the scores for each factor. Variables that were a priori considered to have the potential to influence patient experience were included in the model. Beta coefficients ± SD with p values are shown. Given the overall descriptive nature of the results, no multiplicity adjustments were made. There was no imputation for missing data.