HIV/AIDS health services in Manaus, Brazil: patient perception of quality and its influence on adherence to antiretroviral treatment

The final sample included 812 HIV/AIDS patients: 410 patients interviewed at the central hospital and 402 patients interviewed at four decentralized health units. Both samples naturally picked up the 2:1 gender distribution of the epidemic in the state of Amazonas, with 528 men and 281 women participating in the study [13, 14].² Samples were determined using a 95% confidence interval and additional participants were added to account for patient attrition or sample loss due to data quality issues. Table 1 shows the final sample distribution. Additional file 1 provides more details on the sample calculation.

Study participants included men and women living with HIV/AIDS, aged 18 or older, who had at least one prior visit to the health centers. New patients were excluded from the study as the survey to determine satisfaction referred to the last consultation. Patients were selected during their routine consultations, prior to seeing the doctor, until the desired sample was reached. The demographic characteristics of the study participants are shown in Table 2.

We implemented a questionnaire through face-to-face interviews to measure overall patient satisfaction as well as satisfaction with specific aspects of health services. The instrument specifically gauged information about access to the health center, wait time, quality of communication with health professionals, type of exams conducted on the patient (e.g. lab exams, physical exams, etc.), provision of referrals to complementary services (i.e. financial, psychological, and/or alcohol/drug abuse support services), and general perceptions of quality of service. The questionnaire also collected information on patients’ socio-demographic characteristics, such as age, gender, education level, and employment, among others. The questionnaire was developed following the design of other validated tools, including the Health System Responsiveness section of the World Health Organization’s Health Survey, the Satisfaction with HIV/AIDS Treatment Interview Scale (SATIS), the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS), and the Patient Satisfaction Survey for HIV Ambulatory Care (PSS-HIV). Additional file 2 includes the final version of the questionnaire in English.

In addition to the questionnaire, we retrieved patients’ first CD4 count and latest viral load results from the National Disease Notification System (SINAN). Patients’ first CD4 count results were used as a proxy for timely diagnosis or timely initiation of ART, with a CD4 count equal to or greater than 200 copies/mm3 indicating that the patient was either timely diagnosed or had initiated ART in a timely manner [3]. Patients’ most recent available viral load results were used as a measure of adherence to ART, with a viral load equal to or less than 1000 copies/mL indicating that the patient adheres to ART. The data was collected between September and December 2016.

In this paper, we aimed to identify health center characteristics associated with high patient satisfaction. Based on the empirical literature discussed above, we also sought to explore changes in adherence to treatment given high levels of satisfaction with the quality of care. To this end, we looked to answer two main research questions:

We conducted descriptive and regression analyses to identify patient and health center characteristics that influence patient satisfaction, as well as to assess the correlation between patients’ perceptions on the quality of care and adherence to ART. Pearson’s chi-squared tests were used to check for the statistical significance of descriptive results. Logistic regression analysis was used to identify health center characteristics associated with high patient satisfaction and to determine the likelihood of adherence to treatment given high patient satisfaction with the quality of care. Because patient satisfaction with health services and adherence to ART can be affected by health facility factors as well as individual level factors, we explored the use of multi-level logistic models. In the case of adherence to treatment, the likelihood ratio test indicated that a multi-level model was a better fit than the single-level model and it was used accordingly. The analysis was done using information from all patients as well as by comparing results between the two types of health units, centralized versus decentralized. We employed sampling weights to ensure that results reflect the real distribution of patients across the five health centers (Table 1).

The regression models for patient satisfaction included the following predictors: i) commute time to the health center, as a long commute to the health facility has been associated with lower patient satisfaction [1, 15, 16]; ii) wait time, as patients who experience long wait times to see a health professional are less likely to be satisfied with the quality of health service [5, 6, 15, 17]; iii) convenience of health center’s location,³ which refers to the patient’s perception of the location of the health center and not its actual location (e.g. health facility is not close to patient’s home, but it is on the way to work) and has been associated with higher levels of patient satisfaction [5, 16]; iv) respectful treatment from nurses, since nurses tend to be the health professionals to interact the most with the patients and the quality of communication with them have been found to influence satisfaction levels [6, 15]; and v) respectful treatment from doctors, as patients who report feeling respected by the doctor also report higher levels of satisfaction with services [6, 17]. In addition, the models controlled for the following personal characteristics: i) age, as older patients tend to report higher satisfaction with services [2, 5]; ii) gender, as females tend to report higher patient satisfaction than male patients [5, 6, 16]; iii) education, as higher levels of education have been associated with higher satisfaction [2, 6, 18]; iv) race, since specific racial groups face more discrimination and challenges to access health services, making them more likely to report lower patient satisfaction [19]; v) sexual orientation, as heterosexual patients tend to report lower levels of patient satisfaction than individuals of other sexual identity [20]; vi) income level, as richer patients tend to report lower levels of satisfaction than poorer individuals [2, 5]; vii) place of residence, as patients living in communities with little to no access to specialized HIV/AIDS tend to be more satisfied with any health services they can access than patients in areas with more options for specialized healthcare [9]; and viii) type of health center (centralized versus decentralized), as patients served in decentralized health facilities tend to report higher patient satisfaction [9, 21]. Regression models were conducted in sequence, whereby each predictor was added one at a time to observe changes in patient satisfaction until a final model was achieved with the health center characteristics that best explain variations in patient satisfaction.

In regard to the effect of patience satisfaction on adherence to ART, the regression models included general patient satisfaction as the main explanatory variable. These models controlled for the following patient characteristics: i) age, as older patients are more likely to adhere to treatment than younger individuals [22, 23]; ii) gender, since adherence to ART tends to be lower among female patients than male patients [22, 24, 25]; iii) education, as patients with more levels of education are more likely to adhere to treatment than those with no or lower education levels [24, 25]; iv) race, as racial groups traditionally discriminated against and with difficult access to health services have been found to adhere less to treatment [26‐28]; v) sexual orientation, as treatment continuation tends to be lower among heterosexual patients than among patients of other sexual identity [29‐31]; vi) income level, since richer patients tend to adhere better to ART than poorer patients [22]; vii) place of residence, as patients in communities with little to no access to HIV/AIDS health services tend to show higher adherence to treatment [32]; and viii) type of health center (centralized versus decentralized), as patients in decentralized health facilities tend to adhere better to ART than patients in centralized health units [21, 32].

Almost 82% of all interviewed patients rated the quality of services as ‘excellent’ or ‘good’ (Table 3). However, when disaggregating results between centralized and decentralized health units, we observed higher satisfaction among patients at decentralized facilities (Central Hospital: 81%; Decentralized Health Units: 86%; p-value 0.0003). Differences in satisfaction levels between the two types of health units are more pronounced when looking at the highest satisfaction category, “excellent” (Central Hospital: 19.51%; Decentralized Health Units: 35.59%; p-value: 0.0172).

In regard to access to health services, patients at decentralized facilities were nearly twice as likely to report a commute time of less than 30 min than patients at the centralized unit (Central Hospital: 19.02%; Decentralized Health Units: 34.25%; p-value: 0.0182). Patients at decentralized units were also almost three times more likely to report wait times to see a health professional under 30 min (Central Hospital: 8.05%; Decentralized Health Unit: 23.54%; p-value: 0.0015), whereas patients at the main hospital were more likely to report wait times of one hour or more (Central Hospital: 67.55%; Decentralized Health Units: 40.67%; p-value: 0.001). Almost all patients at decentralized care facilities reported waiting less than 30 min to retrieve their ART medication, compared to only a third of patients at the central hospital who reported the same (Central Hospital: 32.98%; Decentralized Health Units: 92.83%; p-value: 0.0018). The fastest patients at the main hospital could find an available appointment was, on average, 70 days, compared to an average of 31 days for patients at decentralized units (Central Hospital: 69.50 days; Decentralized Health Units: 31.28 days; p-value: < 0.001).

As for the quality of communication with health professionals, patients at decentralized facilities were more likely to report respectful treatment from nurses (Central Hospital: 77.80%; Decentralized Health Units: 92.04%; p-value: 0.0184), that nurses provided information about their health or treatment in a simple and clear way (Central Hospital: 73.90%; Decentralized Health Units: 93.79%; p-value: 0.0092), and that nurses generally answer their questions (Central Hospital: 58.29%; Decentralized Health Units: 83.71%; p-value: 0.0063). Similarly, patients at decentralized health centers were more likely to report that doctors provided information about their health or treatment in a simple and clear way (Central Hospital: 92.68%; Decentralized Health Centers: 97.15%; p-value: 0.0160) and that doctors generally answer their questions (Central Hospital: 82.44%; Decentralized Health Units: 94.76%; p-value: 0.0276).

When exploring the association between patient satisfaction and all indicators of health services accessibility, health center location, wait time, and interactions with nurses were among the most prominent factors influencing patient satisfaction (Table 4). Patients who believed their health center was conveniently located³ were three times more like to be satisfied with its healthcare services than patients who thought otherwise (adjusted odds ratio (aOR): 2.70; 95% confidence interval (CI): 2.18–3.35; p-value < 0.0001). Patients who generally waited more than three hours to see a doctor were 60% less likely to be satisfied with the services (aOR: 0.40; 95% CI: 0.31–0.51; p-value < 0.0001). Patients who said nurses usually treat them with respect were 50% more likely to be satisfied (aOR: 1.50; 95% CI: 1.03–2.20; p-value: 0.040). Please refer to Additional file 3 for further details.

At the central hospital, patient satisfaction is associated with commute time and health center location (Table 5), whereas patient satisfaction at the decentralized health units is driven by shorter wait times and quality of communication with nurses (Table 6). At the main hospital, patients who reported a commute time over one hour were 52% less likely to be satisfied with health services (aOR: 0.48; 95% CI: 0.23–0.98; p-value: 0.044), while patients who said the health center is conveniently located were nearly three times more likely to be satisfied (aOR: 2.49; 95% CI: 1.07–5.77; p-value: 0.034). At the decentralized health centers, patients who reported wait times to see a doctor above one hour were less likely to be satisfied (30 min-1 h wait: aOR: 0.51, 95% CI: 0.29–0.92, p-value: 0.033; 1–2 h wait: aOR: 0.24, 95% CI: 0.18–0.32, p-value < 0.001; 2–3 h wait: aOR: 0.32, 95% CI: 0.21–0.50, p-value: 0.002; more than 3-h wait: aOR: 0.08, 95% CI: 0.05–0.12, p-value < 0.001). Patients at decentralized units who said nurses generally treat them with respect were nearly four times more likely to be satisfied with health services than patients who reported otherwise (aOR: 3.64, 95% CI: 2.48–5.35, p-value: 0.001). Please refer to Additional files 4 and 5 for further details.

Using patients’ latest available viral load count as a proxy for adherence to ART, we found that high patient satisfaction increases the likelihood of adhering to ART (Table 7). For the purpose of this analysis, patient satisfaction was considered as high when patients rated their general satisfaction with the quality of health services as excellent. We also assumed that patients receiving ART for at least one year and with viral load less than 1000 copies/mL adhered to ART. After controlling for several patient and health center characteristics, patients who rated the overall quality of services as “excellent” were nearly two times more likely to adhere to ART than patients who were less satisfied or not satisfied at all with the services (aOR: 1.99, 95% CI: 1.02–3.89, p-value: 0.043).

Results also show that adherence to treatment is higher at decentralized health units than at the central hospital. Nearly 80% of patients at decentralized facilities had a viral load under 1000 copies/mL, compared to 76% of patients at the main hospital with the same results (Central Hospital: 76.33%; Decentralized Health Units: 79.54%; p-value < 0.001). Please refer to Additional file 6 for further details.

Using patients’ first CD4 count as a measure of timely diagnosis and/or timely initiation of ART, we found that patients at decentralized units tend to be diagnosed and initiate ART more timely than patients at the centralized health unit (Table 8). For the purpose of this analysis, we assumed that patients with a CD4 count equal or greater than 200 copies/mm3 were timely diagnosed or initiated ART timely [3]. A larger proportion of patients at decentralized facilities were timely diagnosed or timely initiated ART when compared to patients at the main hospital (Central Hospital: 60.17%; Decentralized Health Units: 77.73%; p-value: 0.024). In regression analysis, we found that when patients were served at the decentralized health units, they were twice as likely to have been diagnosed or have initiated ART timely than patients served at the central hospital (cOR: 2.31; 95% CI: 1.16–4.62; p-value: 0.027).