Follow-up after curative treatment for colorectal cancer: longitudinal evaluation of patient initiated follow-up in the first 12 months

Two cohorts of patients entering follow-up were recruited prospectively. Cohort A entered OPFU before the service change, between April and December 2011; Cohort B entered follow-up after the implementation of PTFU, between January 2012 and August 2013. Depending on their clinical status, patients in Cohort B could be enrolled in either OPFU (for ongoing symptom management) or PTFU. Patients were identified through hospital records and clinician referral. Eligibility criteria were: completion of all treatment with curative intent for CRC; absence of a second cancer diagnosed during treatment; age 18 years or over; no dementia, severe memory loss, or learning disability preventing questionnaire completion. Written consent was obtained.

Data were collected by self-report using structured questionnaires that were mailed at four time points: on completion of final curative treatment (baseline T0), 4 months later (T1), 8 months later (T2), 12 months later (T3). Reminders were sent after two, four and six weeks to non-respondents. Hospital records were used to check patient survival status prior to mailings. Socio-demographic data and information about long-term conditions (LTCs) were gathered at baseline. Postcode was used to derive the Index of Multiple Deprivation (IMD), a government measure reflecting income, employment, housing and other social indicators for local areas [18]. With consent, clinical records were accessed to gather details of time since diagnosis, all types of treatments received, date of entering FU, timing of recurrence or metastatic disease, and, for PTFU patients, attendance at a SM workshop (yes/no).

Generic health related quality of life (HRQoL) was measured with EQ-5D-5 L. Scores for each of five domains (mobility, self-care, usual activities, pain, anxiety / depression) are combined and converted into a summary utility index; a maximum score of 1 indicating “full health” (no problems on any domain). Participants also completed the EQ-5D visual analogue scale (range 0–100 best possible health) [19, 20]. Cancer-specific QoL was measured using the Functional Assessment of Cancer Therapy: General (FACT-G) yielding four functional subscales (physical, social, emotional, functional wellbeing); higher scores indicate better functioning. [21], and FACT-C Colorectal Cancer Subscale (CCS) higher score indicates better functioning [22].

A tool was designed for this study to assess patient experience of follow-up care, with 8 items generated from a review of follow-up evaluations, covering: reassurance, who to contact with problems, comfort about contacting health professionals (HPs) with problems, feelings of isolation, thoroughness of care, ability to ask questions, sufficient time from HPs, involvement in care decisions. [11, 12, 23]. Responses were scored 0 to 100 (best). Respondents also rated if care was acceptable (yes/no/unsure); met their expectations (exceeded/met/fell short); and their perception of service quality (excellent/very good/good/fair/poor).

The primary outcome was the FACT-G. Other related studies have observed pertinent survivorship issues following completing treatment for CRC, suggest a standard deviation (SD) of 14.6 [24] for FACT-G within the first year after surgery. In order to detect an underlying difference of 10 points in FACT-G whilst making no assumption about statistically significant difference direction, using a two-sided test with size =5% and power = 80%, required at least 35 patients in each follow-up regime (the pre and post change cohorts). With six months planned for recruitment for Cohort A (OPFU), a sample of 35 was considered achievable, even allowing for up to 40% dropout [25‐27]. However, in order to conduct a valid comparison between OPFU and PTFU, a larger cohort was deemed necessary for Cohort B because clinicians would not enrol all patients directly into PTFU. Assuming at least one in three patients is in PTFU by T2, and a dropout rate of not more than 40% by T2, at least 175 patients needed to be recruited to Cohort B to acquire at least 35 PTFU patients at T2.

Baseline characteristics of responders and non-responders were compared. Patients in the post-change B cohort, who started follow-up in PTFU or had been transferred by clinicians to PTFU by T2, were assigned to group B2 for analysis. Patients remaining in OPFU (B) at T2 were assigned to group B1. Summary statistics (numbers, proportions, means, and standard deviations) were calculated for all variables at baseline and T2, and comparisons between follow-up regimes were conducted using appropriate statistical tests (chi-square for categorical variables, Fisher’s exact test for dichotomous variables when an expected cell count of <5 was encountered, unpaired t-test for continuous variables, Mann-Whitney U test for ordinal variables). Forward linear regression modelling was undertaken to explore the effect on patient outcomes (EQ-5D utility index, FACT-G, general QoL score, additional colorectal cancer concerns score (CCS)) after different exposure lengths to PTFU while controlling for demographic factors, co-morbid conditions, and clinical management.

The NHS perspective was primarily adopted, comparing costs of OPFU with PTFU in British Pounds, 2013. Two factors were considered: impact on direct costs, reflecting replacement of clinic follow-up visits in OPFU, by self-management in PTFU; unscheduled service use by patients, to explore if removal of outpatient clinic access was associated with increased visits to other health professionals. Resources involved in delivering OPFU and PTFU were obtained from clinic protocols and health professionals, covering the first year of follow-up. Costs were attributed to staff resources using validated national scales [28], and costs of facilities and other physical resources based on actual expenditures. To measure service utilisation, patients were asked at each time point to retrospectively self-report health service use since completing the last questionnaire, using a modified Client Service Receipt Inventory [29]. To minimise confounding, since some patients moved from OPFU to PTFU between T0 and T2, the analysis period was confined to between T1 and T2. Paired group comparison was undertaken (A vs. B; B1 vs. B2; A vs. B2), covering utilisation of the most commonly used services. To assess personal costs, information was requested on travel to clinic appointments, whether accompanied by others, and time lost from work. Travelling distance to the hospital was calculated using postcodes.

The National Research Ethics Service judged this to be a service evaluation. Governance approval was obtained from the Trust Research and Development Department; the study was conducted according to the ethical principles of informed consent and assurance of confidentiality. Storage and processing of data conformed to the Data Protection Act 2000.

Of 124 patients sent baseline questionnaires in Cohort A, 75 (60.5%) were returned in a timely manner; 176 of 239 (73.6%) were returned in Cohort B. Not all patients responded at subsequent time periods; 53 (70.7%) and 79 (44.9%) responses were analysable at T2 respectively (Figs. 1 and 2). Of the 176 patients recruited into the second phase of the study, there were 31 in PTFU at T0; 34 had crossed over by T1, a further 14 had crossed over by T2, and a further 5 at T3.

Comparison of the 251 responders with 112 non-responders at baseline revealed no significant difference in age, sex, or deprivation. Comparison of the characteristics of 132 participants at T2 against the remaining 119 baseline patients from whom a T2 questionnaire was not obtained, showed that those not in the analysis at T2 were more likely to have a recurrence or metastatic disease (data not shown).

Comparison of Cohorts A (n = 75) and B (n = 176) at baseline showed a significantly higher proportion of Cohort B patients had received chemotherapy, and a significantly higher proportion of Cohort A patients reported having arthritis. The groups did not differ on socio-demographic characteristics (mean age 72 years, 40% women) or outcome measures (Table 1).

At baseline, most of the 176 patients in Cohort B were still in OPFU (145 vs. 31 in PTFU). No significant differences were found in socio-demographic characteristics of patients in B1 and B2. Patients retained in OPFU (B1) compared to those selected for PTFU (B2) were more likely to have received chemotherapy, radiotherapy, and multiple treatments. They reported lower self-rated health and physical wellbeing, and worse CCS scores (Table 1)

QoL outcomes were not significantly different between Cohorts A and B at T2. Compared to patients in B1 (OPFU), those in B2 (PTFU) had significantly better HRQoL and CCS scores (Table 2). Regression modelling identified a high number of LTCs as the dominant factor associated with adverse HRQoL outcomes; this was offset by a positive effect for heart disease(HD). Chemotherapy negatively affected CCS scores. Patients in work reported worse CCS scores. Time spent in PTFU was not significantly associated with any outcome. The models explained less than 20% of the variation (Table 3).

There were no significant differences in experience of follow-up care between OPFU (A and B1) and PTFU (B2) with respect to reassurance, access to specialist support, ability to ask questions, time spent with doctors/nurses, and involvement in decision-making. More patients in PTFU (36/37; 97.3%) found their follow-up to be acceptable (vs. not/unsure) than patients in OPFU (24/32; 75%) (p = 0.010). There were no significant differences between groups regarding quality of care or the extent to which it met expectations; most reported high satisfaction (Table 4).

Detailed analysis of the patient pathways under OPFU and PTFU showed that PTFU was up to £142.24 per patient more expensive than OPFU in the first year. Savings in outpatient appointments in PTFU were outweighed by the costs of the self- management workshop and remote surveillance (Online Resource 3 Table, of Resources and Costs). No significant differences were found between utilisation of community and hospital services for any group comparison (data not shown), except GP visits were higher for OPFU patients than PTFU patients (1.84 vs. 1.08 monthly visits, p = 0.024). Most patients (76.9%) reported travelling to clinic by car, the mean (SD) round trip distance between home postcodes and the hospital being 18.73 (12.82) miles. Applying the NHS mileage reimbursement rate (2014) of 50.5p per mile, a move to PTFU would save £28.38 per patient in the first year (assuming OPFU involves four clinic visits vs one in PTFU). The majority (87.9%) were mostly or sometimes accompanied to clinics so additional savings might arise for companions (22.3%) taking time off work.