Study design and study areas

A cross-sectional study was conducted based on a users' survey using the CCAENA© questionnaire (Cuestionario de Continuidad Asistencial Entre Niveles de Atención)²² in the six countries. Two comparable public healthcare networks in urban areas were selected in each country (for further details about the networks and the selection criteria, see Vázquez et al.). ¹⁷

Study population and sample

Study population were residents of the study areas, over 18 years of age, who had at least one chronic condition and had used two care levels (PC doctor and SC doctor or emergency services) in the six months prior to the survey for the same medical condition (either chronic or acute). A sample size of 392 patients per network in each country was estimated to detect a 10% variation in patients' perceptions of continuity of care between healthcare networks, based on 80% power (β = 0.20) and 95% confidence level (α = 0.05) in a bilateral contrast. Patients were selected by simple random sampling, from those waiting in PC centres. As patient registers were unavailable in some of the networks, all patients in waiting rooms, reception and clinical laboratory areas of the primary care centres were approached and those meeting the inclusion criteria were interviewed. Patients who met the criteria and refused to participate ranged from 8.9% in Colombia to 40.6% in Argentina. The final sample size was 4,881 (Table 1).

Questionnaire

A new version of the adapted to the Latin American context and validated version of the CCAENA© questionnaire^22,23 was used. Its content was tailored to contextual and language variants of each country and translated for Brazil, by means of two pre-tests, with cognitive interviews with patients and a pilot test (20 users per country). The tool is divided into two sections: the first reconstructs the care trajectory for a specific episode within the last 6 months, and the second (this paper's object) measures patients' perceptions of continuity.

Data collection and quality

Data were collected by means of face-to-face interviews conducted by specifically trained interviewers in the PC centres from May to December 2015, except in Argentina (April 2016) and Uruguay (February 2016). To ensure the quality and consistency of data, we supervised interviewers in the field, reviewed all questionnaires, re-interviewed 20% of randomly selected participants, and used the double-entry method to control inconsistencies during data entry.

Variables

The outcome variables are:

The explanatory variables are:

Analysis

First a descriptive univariate analysis was performed. Logistic regression models stratified by country were then generated to assess the relationship between the three outcome variables and associated factors. Adjusted odds ratios (OR) and corresponding 95% confidence intervals were obtained. Robust covariance adjustments (employing the network variable) were used to account for correlated observations due to clustering. All the independent variables were included in the analyses. Percentages and adjusted OR were calculated for perceived high levels of continuity of care regarding information transfer and care coherence. Multicollinearity between explanatory variables was tested using the variance inflation factor (VIF), and found to be insignificant (below 1.5). Statistical analyses were performed using Stata, version 12.

Ethical considerations

Ethical approval was obtained from the ethical committees in the participating countries. All interviewees participated on a voluntary basis after signing an informed consent. The right to refuse to participate or withdraw from the survey, anonymity, confidentiality and protection of data were all guaranteed.

Sample characteristics

The sample was predominantly female, with no or primary level education, and middle-aged (40 to 64 years), except in Chile and Colombia. There were notable differences between countries regarding the use of health services outside the public network (from 54% of patients in Mexico to 4% in Uruguay). More than half of the respondents rated their state of health as poor, except in Uruguay (42.3%) and Argentina (44.7%). In all countries except Uruguay (46.8%), more than half the patients suffer from two or more chronic conditions.

The participants perceived high levels of relational continuity with their PC and their SC doctors, with the highest reported in Argentina and Uruguay (over 90%) and the lowest in Chile (66.9% and 70.0%) and Colombia (84.5% and 68.8%) (Table 1).

Perception of care continuity across care levels

Regarding continuity of information, patients perceived that the PC doctor has greater knowledge of the instructions given to them previously by the SC doctor than vice versa (Table 2; see more detailed results in online ^{supplementary table I}). Besides Uruguay, in all countries the proportion of patients perceiving high levels of information continuity is below 50%, with patients in the Brazil networks at the bottom of the scale.

With respect to continuity of clinical management, and care coherence, users perceived that PC doctors were more in agreement with the SC doctor's directions than vice versa (except in Uruguay). The highest levels for both items were found in Argentina and Uruguay (around 95%), and the lowest in Chile and Colombia (under 80%).

The percentage of patients perceiving that doctors collaborated with each other was low in Brazil (25.1%), Chile (47.3%) and Colombia (40.2%), and high in Argentina (92.4%) and Uruguay (94.3%). Information transfer and communication between PC and SC doctors, or its absence was the main reason given in all six countries for believing that doctors do (or do not) collaborate with each other (comprising 30-60% of all answers) (see online ^{supplementary table II}).

As for accessibility between levels, most patients perceived long waiting times, mainly for consultations with the SC doctor following a referral from PC, but also for follow-up consultations with the PC doctor after the specialist (Table 2). In referrals to SC, under 60% of patients in Argentina and Mexico considered that the waiting time for their consultations was long, whilst in Chile, Brazil and Colombia, around 70% hold this opinion.

Factors associated with continuity between care levels

Relational continuity with the PC doctor and with the SC doctors were the only two factors associated in all countries with both continuity of information and continuity of clinical management between levels (Tables 3, 4 and 5). Similarly, patients perceiving high levels of information transfer between levels were also more likely to report high levels of care coherence in most of the countries (Table 5).

Regarding individual characteristics, the clearest association is between continuity of information and sex: being female tends to be negatively associated with the perception of information transfer from the SC doctor to the PC doctor (Table 3 and 4).

Differences in continuity across care levels in the networks of the six countries

Patients in the networks analysed have a worse perception of transfer of information and accessibility to SC than of care coherence, coinciding with the experience of doctors in these networks.²⁵ Yet, we observe differences between countries: high est levels of continuity of information across levels are perceived in Uruguay, and also higher levels of care coherence in both Uruguay and Argentina, but with less substantial differences. These differences may be attributable to certain characteristics of the networks studied in Uruguay (smaller size, located in smaller urban areas and with co-location of PC and SC doctors in polyclinics) which foster that doctors know each other personally as well as mutual trust and informal communication between them,²⁵ and facilitates the information exchange and agreement on treatments. Regarding Argentina, it may be related to the assignment of specialist as referral teams to PC centres for certain specialities²⁶ which could favour collaboration and agreement among PC and SC doctors, as well as relational continuity with SC, since patients assigned to a particular PC doctor would generally be seen by the same specialists.

The doctors' perspective on equivalent aspects such as the transfer of information or agreement on treatments was generally worse.²⁵ Doctors are probably more critical because they encounter coordination problems with the other care level on a daily basis, whereas patients probably only perceive coordination problems when these have a negative impact on the quality of their care.²⁷ Nevertheless, patients also perceived significant discontinuities between care levels. They perceive them in the transfer of clinical information between levels, which is related to doctors' deficient use of referral and reply letters.²⁰ Also, significant discontinuities were perceived by patients in access to SC following referral by a PC doctor. The long waiting times reported indicate delays or interruptions in the diagnosis and treatment of chronic patients that could result in late diagnosis, upsets and complications, as previous qualitative studies pointed out.²⁵

Relational continuity with PC and SC doctors and information transfer are associated with care continuity between levels

One of the most relevant results of this study is that it is the organisational factors (relational continuity with the PC doctor and the SC doctor, and information transfer between levels), and not the individual characteristics of chronic patients, that are consistently associated in the same way in nearly all countries: relational continuity with perceiving high levels of both types of care continuity between levels, and information transfer with care coherence. This finding is consistent with previous studies.^3,14,28 However, our results also highlight the importance, not only of being seen by the same doctor, but also of the quality of the bond established between patient and doctor in both PC and SC, for continuity of care between levels to exist for chronic patients. Relational continuity appears to favour a doctors' knowledge of the patient's medical history across the care continuum and, at the same time, ease the process of agreeing on a treatment plan with doctors of the other level. Indeed, the frequent staff turnover in Chile and Colombia (more than 60% of doctors on temporary contracts in the first case and nearly 80% in the second, compared to 20% in the other countries),¹⁹ may serve to explain the more negative perception of continuity between care levels in these countries.

Another relevant finding is the association between the perception of information transfer and care coherence, described in qualitative studies,²⁹ that point out that the transfer of information between levels facilitates agreement between PC and SC doctors on patient diagnosis and treatment. This association may explain why patients in Brazil, Chile and Colombia, with a worse perception of information continuity, also have a worse perception of care coherence.

Limitations

Given the lack of records on chronic patients' use of services in some networks, patients were selected in the PC health centres, and they were interviewed on the premises. This may have influenced their responses for instance by inhibiting to express their real opinions or owing to distraction. Also, survey fatigue cannot be excluded. After the pilot test, the original CCAENA recall period was extended from 3 months to 6 given the patients' difficulties in the study countries to access another care level, which may have led to recall bias. Additionally, as this is a cross-sectional study it does not allow us to determine causality but rather association between factors. Furthermore, the dichotomization of indexes to facilitate the interpretation of data could lead to some loss of information.