A complete list of participant in the QUOVADIS study has been previously published .
Giulio Marchesini MD, Unit of Metabolic Diseases, ‘Alma Mater Studiorum’ University of Bologna, Policlinico S. Orsola, Via Massarenti, 9 I-40138 Bologna, Italy. (fax: +39 051 6364502; e-mail: firstname.lastname@example.org).
Objectives. To investigate weight loss and reasons for attrition in obese patients on long-term continuous care.
Design. Observational study with 36 months of follow-up.
Setting. Fifteen Italian obesity centres applying a continuous care model of medical treatment.
Subjects. One thousand treatment-seeking obese subjects (785 females, median age 45.1 years, median BMI 37.4 kg m−2). Weight loss expectations were systematically recorded at baseline.
Interventions. An initial intensive treatment period (3–6 months) was followed by a less intensive continuous care (a follow-up control every 2–4 months).
Main outcome measures. Attrition, reasons for treatment interruption and BMI change. Data were recorded by telephone interview in dropouts.
Results. Only 157 patients (15.7%) were in continuous treatment at 36 months. The main reasons of attrition were logistics, unsatisfactory results and lack of motivation. The only basal predictor for continuous care was lower Expected One-Year BMI Loss (P = 0.016). The probability of dropout increased systematically for any 5% expected BMI loss (Hazard ratio, 1.05; 96% confidence interval, 1.01–1.09). The mean percentage weight loss was greater in continuers (5.2% vs. 3.0% in dropouts; P = 0.016). However, the dropouts satisfied with the results or confident to lose additional weight without professional help reported a mean weight loss of 9.6% and 6.5% respectively.
Discussion. Continuous care produces long-term weight loss only in a subgroup of obese patients seeking treatment in medical centres. The finding that subgroups of dropouts report long-term weight loss has implication for the treatment of obesity.
Weight loss maintenance is a major problem in the treatment of obesity . This statement is based on robust research data indicating that a large percentage of subjects can achieve a healthy weight loss with both pharmacological and behavioural treatments for obesity but only a minority is able to maintain weight loss after treatment stop . The difficulties associated with long-term weight maintenance led to consider obesity a chronic condition requiring continuous care , a strategy adopted by physicians supporting both a pharmacological [3, 4] and a behavioural approach to obesity [5, 6].
The evidence in favour of long-term (over two years) pharmacological treatment is modest even within randomized, controlled studies [7–10], and attrition remains an unsolved problem [7–11]. The clinical utility of continuous care mainly derives from three findings observed with long-term behaviour treatment for obesity. First, extending the treatment improves weight loss maintenance [1, 12]. Secondly, very good results in weight loss maintenance were obtained after 10–12 years of follow-up in the only study where an intensive 4-year programme was followed by an additional treatment, or, whenever necessary, by an inpatient therapy . Mean weight loss averaged 10.5 kg. Thirdly, good results were obtained in a subgroup of obese subjects who continued treatment for 5 years by the Trevose Behavior Modification Program, a long-term behavioural pressure and support group . However, the effects on weight maintenance were demonstrated only in the small subgroup who accepted and complied with these long-term treatments , and the attendance to a long-term behaviour therapy drops dramatically with time . Unfortunately, patients tend to regain weight once they stop the maintenance programme [2, 17]. Thus, also the clinical utility of a long-term behaviour treatment remains to be fully established.
In Italy, the view of obesity as a chronic condition requiring continuous care has been accepted by several medical centres accredited for the treatment of obesity to the national Health System. However, no study has systematically evaluated the clinical utility of this type of approach. Aim of the present study was to analyse the effect of continuous care on weight loss in a sample of treatment-seeking obese subjects participating in a large observational multicentre Italian study with 36 months of follow-up .
Materials and methods
QUOVADIS study planning and protocol
The QUOVADIS (QUality of life in Obesity: eVAluation and DIsease Surveillance) study planning and protocol were illustrated in detail in a previous paper . QUOVADIS is an observational study started in 1999 on quality of life in obese patients seeking treatments at medical centres accredited by the Italian Health Service for the treatment of obesity. Twenty-five medical centres specialized in the treatment of obesity, enrolled patients between December 1999 and February 2001. The study was purely observational; all centres were expected to treat patients along the lines of their specific programmes, including dieting, cognitive behaviour therapy, drugs, bariatric surgery (<2% of patients).
All obese subjects (body mass index, BMI ≥ 30 kg m−2), consecutively seeking treatment were eligible for the study, provided they were not on active treatment at the time of enrolment, were in the age range between 25 and 65 years, agreed to fill the whole package of self-administered questionnaires, and signed an informed consent to participate. Patients were evaluated at baseline, approximately 1 week before the beginning of the treatment, and after 6 and 12 months. An additional follow-up was scheduled in 18 centres between May 2003 and January 2004, on average 36 months after enrolment.
The present report analyses the 36-month results of the 15 medical centres that treated patients according to a continuous care program. Whilst the theory on which the treatment was based varied between centres, all adopted a low-intensity approach with periodic controls for an indefinite period of time. In general, an initial intensive treatment period (3–6 months) was followed by a less intensive continuous care (in general, a follow-up control every 2–4 months). The very few cases treated by bariatric surgery or bio-enteric intra-gastric balloon positioning during follow-up in these centres were excluded from analysis. Finally, three centres treating morbid obesity were also excluded from analysis: two applied an inpatient care model; the third centre had bariatric surgery as primary treatment.
The ethical committees of the individual centres approved the protocol, after approval by the ethical committee of the coordinating centre (Azienda Ospedaliera di Bologna, Policlinico S. Orsola – Malpighi). All subjects gave written informed consent for participation.
Baseline data collection included weight and height measurement, a detailed Case Report Form and a package of questionnaires for psychological distress and eating behaviour disorders.
Case Report Form. It was filled by physicians at the time of enrolment by directly interviewing patients, and included demographic data as well as a detailed diet history, with specific information on the number and the results of previous attempts to lose weight, on age at first dieting, on Expected One-Year Weight Loss, on Maximum Acceptable and Dream Weight . Expected One-Year Loss was defined as ‘the weight that patients were expecting to lose with treatment in the following 12 months’. To help subjects indicate their expectations, this value was categorized in multiples of 10 kg. Maximum Acceptable Weight was defined as ‘the heaviest body weight that patients could tolerate and accept after treatment’, whilst Dream Weight was defined as ‘the weight that they were dreaming of achieving with treatment, however unrealistic it was’. The form included also a question on the reason(s) for seeking treatment. For this specific purpose, patients were asked to choose the main reason amongst three mutually exclusive answers: (i) improving appearance; (ii) improving future health; (iii) improving present health.
Psychosocial measures. Participants completed a battery of questionnaires specifically meant to detect psychiatric distress, binge eating and body image dissatisfaction, as reported elsewhere . For the purpose of the present study, we used the results of three questionnaires: (i) the Symptom CheckList-90, an easy and validated tool to identify psychopathological distress ; (ii) the Binge Eating Scale, a self-administered questionnaire to measure the severity of the binge eating ; (iii) the Body Uneasiness Test, a self-administered rapid questionnaire, specifically developed to evaluate concern for physical appearance, body image awareness and body parts which most severely contribute to body dissatisfaction .
Weight and height. Weight was measured on a medical-balance and height by a stadiometer by a medical doctor of every centre involved in the study. Patients were dressed with underwear without shoes.
Interview at 36-month follow-up. Patients no longer in continuous treatment (dropouts) were contacted by phone by the same therapists who conducted the baseline evaluations. The telephone interview included several questions on body weight, attrition and its reasons, health status, and mood. For the scope of the present study we employed only the following questions: (i) When did you last weigh yourself? (ii) What was your body weight when you last weighed yourself? (iii) How much do you think your weight is (only if patients did not weigh during the last month)? (iv) Do you regularly weigh yourself? (v) Why did you stop treatment? A structured list of nonmutually excluding answers was provided with this last question. For self-reported weights at follow-up, a +2-kg correction was applied to account for bias .
Attrition. Attrition was evaluated at 36 months analysing the patients’ medical records where the date of the last medical examination and the date of treatment stop were reported.
All data were stored in a large database, provided by CINECA (Casalecchio di Reno, Italy), an Interuniversity Consortium of 15 Italian Universities, through an extranet system. Participating centres accessed the system using a personal user-id and password and inserted patients’ data into the database through electronic forms.
All weight data (in kg) were transformed into BMI units to allow comparison between genders (19). A first descriptive analysis was used to obtain a qualitative evaluation of clinical data, response to questionnaires and patients’ outcomes. anova was used to test the significance of differences between male and female and between subjects on active continuous treatment (continuers) and dropouts. The chi-square test was used to test the significance of difference between continuers and dropouts in relation to the main reason for seeking treatment and in relation to gender. Correlation analysis was also performed to establish links between principal continuous variables and BMI changes from baseline to 36 months. Attrition was also tested using a Cox regression model. Logistic regression analyses were used to identify the determinants of attrition. Data are reported as mean ± standard deviation (SD). The critical value of significance was set at 0.05.
The baseline characteristics of the whole QUOVADIS sample were described in detail in previous reports [18, 19]. Table 1 presents the principal baseline data of the patients who participated in the 36-month follow-up. The sample consisted of the complete records of 1000 medically treated patients (785 females, 215 males). Females had a higher BMI and a reported higher maximum BMI, earlier age at first dieting, higher weight loss expectations, and a higher frequency of primary motivation for weight loss to improve appearance. Males had a significantly higher BMI at age 20, and a significantly higher Dream and Maximum Acceptable BMI.
Table 1. Baseline data of patients included in the analysis (mean ± SD or percentage)
Females (n = 785)
Males (n = 215)
45.3 ± 11.1
45.0 ± 10.4
BMI (kg m−2)
37.5 ± 6.0
36.6 ± 5.5
BMI at age 20 (kg m−2)
25.4 ± 5.1
26.3 ± 4.5
Maximum BMI (kg m−2)
38.7 ± 6.4
37.6 ± 5.8
Age at first dieting (years)
27.2 ± 11.0
29.6 ± 10.0
Maximum weight loss (%)
16.4 ± 9.1
14.9 ± 10.6
Weight loss expectations (kg m−2)
Maximum acceptable BMI
28.8 ± 4.1
29.8 ± 3.4
25.5 ± 3.2
27.3 ± 2.7
Expected One-Year BMI Loss
10.1 ± 3.6
8.0 ± 3.2
Primary motivation for weight loss (% within sex)
Attrition at follow-up
The attendance of patients to treatment sessions declined sharply with time (Fig. 1). Approximately 20% of patients abandoned treatment soon after the first visit and attrition was as high as 58% at 12 months. After an average of 36 months, only 157 of 1000 patients (15.7%) were in continuous treatment (continuers). Median time to dropout was 227 days (interquartile range, IQR 488 days). The dropout rate was remarkably different between centres, ranging from 61% to 98% (P < 0.0001, chi-square test), but was not different between genders, or between subjects aged under 45 vs. over 45 years, or in subjects trying to lose weight for health concerns versus concern for body appearance.
No data of the diet history could distinguish continuers from subjects who stopped treatment (dropouts). Continuers were significantly older and had a lower Expected One-Year BMI Loss than dropouts (Table 2). Time to dropout was progressively shorter in relation to Expected One-Year BMI Loss [from 315 days (IQR 465) in subjects with expected BMI loss <5 kg m−2, to 231 days (IQR 481), and 202 days (IQR 450) for expected BMI loss in the range 5–10 and above 10 kg m−2 respectively]. No significant differences were found between continuers and dropouts in the score of psychosocial measures (not reported in detail), and in the primary motivation for weight loss (χ2 = 4.337; P = NS). Finally, no difference in attrition rate was found between males and females (χ2 = 0.069; P = NS). By a Cox proportional hazard model, the probability of dropout increased systematically for any 5% Expected One-Year BMI Loss (Hazard ratio, 1.05; 96% confidence interval, 1.01–1.09; P = 0.017).
Table 2. Baseline data in obese subjects, according to dropout at 36 months (mean ± SD)
Continuers (n = 157)
Dropouts (n = 843)
NS, not significant.
47.6 ± 10.7
44.8 ± 11.0
BMI (kg m−2)
37.6 ± 5.8
37.3 ± 5.9
BMI at age 20 (kg m−2)
25.4 ± 5.0
25.7 ± 5.0
Maximum BMI (kg m−2)
38.8 ± 6.3
38.4 ± 6.3
Age at first dieting (years)
28.7 ± 11.3
27.4 ± 10.9
Maximum weight loss (%)
15.8 ± 8.1
16.2 ± 9.6
Weight loss expectations (kg m−2)
Maximum acceptable BMI
28.8 ± 4.1
29.8 ± 3.4
25.5 ± 3.2
27.3 ± 2.7
Expected One-Year BMI Loss
9.0 ± 3.7
9.8 ± 3.6
Logistic regression analysis revealed that the only basal predictor for continuous care after 36 months was Expected One-Year BMI Loss (β = −0.076, χ2 = 5.757, P = 0.016).
Reported reasons for treatment stop
Of the 843 dropouts, approximately one-third was either not traced at the phone interview (n = 273) or did not accept the interview (n = 24). They were considered lost to follow-up.
Logistics was the most frequent reason (51.1%) for treatment stop reported by the 546 interviewed subjects (Table 3). The major obstacles to continue treatment were work and family problems (40.1% and 39.4% respectively). Unsatisfactory results were the third most frequent reason (25.5%); subjects unsatisfied with the results were older than other dropouts (F = 5.647; P = 0.018). Other conditions were reported as reasons for treatment stop. Dropouts who disagreed with treatment programme had a higher BMI (P = 0.003), a higher maximum BMI (P = 0.004) and a higher Expected One-Year BMI Loss (P = 0.016). The dropouts who achieved satisfactory results with treatment had a lower BMI (P = 0.004) and a lower maximum BMI (P < 0.001).
Table 3. Reasons for treatment stop reported by the 546 dropouts during the telephone interview
Reasons for treatment stopa
Total number or % within group
aPatients were allowed to report more than one reason for treatment stop.
Disagreement with treatment plan
Satisfied with treatment results
Confident to lose additional weight without professional help
Living far from the medical centre
Health problems other than obesity
Unsatisfactory results (%)
Unsatisfiedwith weight loss
Unable to keep following treatment programme
Lack/loss of motivation
36-month weight change in continuers and dropouts
At follow-up the mean percentage weight loss was significantly greater in continuers than dropouts (5.2% vs. 3.0%; P = 0.016). The percentage of subjects who achieved a weight loss ≥5% was higher in continuers (40.8%) than in dropouts (22.7%) (χ2 = 22.842; P < 0.001). A similar difference was present for patients achieving a weight loss ≥10% (22.3% in continuers vs. 12.0% in dropouts) (χ2 = 11.977; P = 0.001) (Fig. 2).
Figure 3 describes the mean percentage weight loss at follow-up of dropouts in relation to the reported reasons for treatment stop. Dropouts satisfied with the results reported a mean weight loss of 9.6%, whilst dropouts confident to lose additional weight without professional help reported a mean weight loss of 6.5%; both subgroups of patients lost significantly more weight than other dropouts (P < 0.001 and P = 0.002 respectively). Conversely, subjects stopping treatment because of lack of motivation and unsatisfactory results had a lower final weight loss than other dropouts (P = 0.021 and P = 0.004 respectively).
The study shows that continuous care is effective in producing a long-term weight loss in a subgroup of obese patients seeking treatment. After 36 months, continuers achieved a significant greater weight loss than dropouts (5.2% vs. 3.0%). The 5% weight loss achieved by continuers is considered a satisfactory result in the management of obesity, carrying a positive impact on the medical complications of obesity [24–28]. It also decreases the 4-year cumulative incidence of diabetes by 58% in overweight and obese men and women with impaired glucose tolerance [29, 30].
The study has several strengths. First, its observational structure allows a comprehensive analysis of the effect of weight loss in the ‘real world’ of 15 medical centres scattered throughout Italy, with heterogeneous modalities of treatment but with a common general model of continuous care. Secondly, it included a long follow-up time to evaluate attrition. Thirdly, it recorded a large set of characteristics at baseline, which were used to predict attrition.
However, only a small subgroup of patients achieved a healthy long-term weight loss, thus confirming a large body of research on weight loss behaviour therapy. Three to five years after treatment, a large percentage of patients return to, or even exceed, their pretreatment weight [15, 31–35], and only a minority (approximately 20%) keeps 50% or more weight loss over 4 years . These results indicate that it is urgent to develop new ideas and models of treatment to improve the patients’ long-term weight loss. The continuous care programme used in the Italian centres is a therapeutic option only useful in a minority of patients. Only 15.7% of patients were still on treatment after 36 months and more than half had stopped treatment in the course of the first year. Continuers were older and had a lower Expected One-Year BMI Loss. These data confirm previous observations that initial weight loss expectations distinguish continuers from dropouts in weight loss programmes , whereas several clinical and psychological traits do not predict dropout. Our data do not give any clue as to the mechanism(s) leading continuers to achieve a successful weight loss. An extended treatment might provide patients the best chances to cope with the obstacles to weight maintenance with the help of professional assistance, so as to hold on to behaviour changes . A realistic Expected One-Year BMI Loss may be a psychological factor involved in the persistence in a continuous care model of treatment, and indirectly in weight loss success.
The continuous care model used in the Italian medical centres is a low-intensity physician–patient contact, with follow-up visits every 2–4 months, because of budget constraints imposed by the Italian National Health System. The scarce resources available are not only dependent on a general politics aimed at a systematic containment of health budgets, but also to the scarce consideration of obesity as a threatening condition requiring adequate treatment per se, independently of complicating diseases. The successful study on continuous care of Bjorvell and Rossner  was based on a very high frequency of contacts (i.e. a minimum of once a week, plus occasional refresh courses) to help patients keep long-term weight loss. Other successful long-term approaches included therapist–patient contact on a weekly or a bi-monthly basis, and the amount of contact with the therapist is associated with weight loss . It is conceivable that a more intensive and expensive model might obtain a greater weight loss and reduce the dropout also in Italian patients. However, the best frequency of a continuous care model has never been determined. Patients may even experience therapy ‘burnout’ with too intensive models , particularly if a weight loss plateau and the monotonous sessions favour the discontinuation of attendance .
The reasons for treatment stop reported by phone interview were multiple and heterogeneous, but not all dropouts should be considered treatment failures. The dropouts satisfied with the results obtained with treatment (7.0% of the whole sample) and those who were confident to lose additional weight without professional help (11.7%) achieved an even larger mean weight loss than continuers. Two cognitive factors may partly account for persistent weight loss behaviours without a continuous contact with a therapist in these two groups: (i) satisfaction with the results, and (ii) confidence in the ability to lose weight without additional professional help. Previous studies had already suggested a role for weight loss satisfaction in long-term weight maintenance. A qualitative retrospective study (in-depth individual interview and group interview) found that obese regainers, but not maintainers, had failed to achieve their initial weight goals and were highly dissatisfied with body weight achieved by treatment . This observation was confirmed by a longitudinal study on the effect of behaviour therapy, where satisfaction with body weight achieved during treatment was associated with better weight maintenance . Our data confirm that patients satisfied with the weight loss do not need a continuous care approach.
The second psychological process ‘confidence in the ability to lose additional weight without professional help’ is a construct close to Bandura's concept of self-efficacy, which refers to a person's belief that he/she is capable of holding on to a specific behaviour . Self-efficacy was demonstrated to be a strong predictor of weight loss success in a few , but not all weight management programmes [31, 44–46]. According to our data, also these patients do not need continuous care to achieve a long-term successful weight loss.
Our results suggest some potential strategies to ameliorate the long-term management of obesity. Not all patients are suitable for a continuous care and not all patients need a continuous care approach to obtain long-term favourable results. Research is needed to identify the candidates to a long-term, expensive treatment, as well as patients who can achieve satisfactory results with a shorter and less expensive approach. The Expected One-Year BMI Loss and age must be carefully considered in treatment plans, together with obstacles to treatment. Treatment plans may be simplified in the elderly, whereas they may be intensified according to the degree of obesity. In young patients, different approaches are need to cope with their desire of rapid and unrealistic weight loss. In order to improve patients’ adherence to treatment, the following items are need to be considered: (i) unrealistic weight loss goals and expectancies; (ii) logistics (e.g. living far from the centre, work, family, medical and economic constraints); (iii) inadequate motivation to weight loss; and (iv) disagreement with treatment plan. Specifically, for subjects in the working age, it is mandatory to adjust treatment plans and visits with the needs of patients.
The study has also some limitations. First, our observations are restricted to obese subjects seeking treatment in a medical setting, and therefore do not provide information on the large number of obese subjects that do not seek treatment or that seek help in nonmedical settings. Secondly, the model of continuous care provided by the 15 centres involved in the study was heterogeneous, with great differences both in the theoretical approach and in the modality of care. Although the population was homogeneous amongst centres, the difference in the care model was probably the reason for the different dropout rate, and this area requires a specific investigation. Thirdly, the reasons for treatment stop were investigated only retrospectively and not immediately at the time of dropout, introducing a memory bias.
Future studies should also evaluate the cost effectiveness of the continuous care model of obesity offered by Italian medical centres, and to evaluate both more intensive and less expensive models. The approach applied in Italian medical centres is based on a low-intensity periodic face-to-face control; more intensive approaches could either increase both the effectiveness and the cost of treatment , or paradoxically reduce attendance because of patients’‘burn-out’ . Cheaper alternatives (e.g. therapist contact by e-mail or by telephone, and the use of telephone reminders) during short-term programmes  have been tested with success and require further validation. Any effort to reduce the cost of the treatment would free resources to tackle larger groups of patients in the present epidemics of obesity.
Conflict of interest statement
The authors declare that they have any conflict of interest in relation to this manuscript.