Childhood cancer survival has increased over the last 30 years, but long-term effects necessitate continued monitoring of survivors. Since not all of them attend follow-up clinics, this study assesses the efficacy of obtaining information from general practitioners (GPs) through a 5-year rolling postal program.
Survivors were included who had been diagnosed with a malignancy in the West Midlands since 1957 and were not attending central long-term follow-up clinics.
One thousand twenty-seven patients were followed up between 1993 and 2004. Replies were received on 903 (88% response). There were 44 subsequent malignancies and 42 deaths. No medical problems were reported in 341/935 patients (36.5%); in the other 594 endocrine effects were the most common, with visual effects the biggest single problem. Brain tumor survivors had the largest proportion of problems.
UK childhood cancer survival rates have improved dramatically over the last 30 years, such that 75% of these patients now survive 5 years. Five-year survival rates in the West Midlands have increased from 23% for patients diagnosed in the 1960s to 70% for those diagnosed in the 1990s 1. Adverse late sequelae have been well-documented 2–10, and although long-term surveillance is recommended, how and by whom is still a matter of debate 11–14.
Birmingham Children's Hospital NHS Trust (BCH) has been one of the largest centers in the UK treating childhood malignancy for over 30 years. Most patients continue to attend indefinitely, so their health is monitored continually and any long-term problems identified and managed appropriately. Not all West Midlands patients have been treated at BCH, and of those who have, not all continue to attend follow-up clinics for various reasons, including moving away, missed appointments, and clinical discharge. It is therefore necessary to obtain follow-up information from other sources. The hypothesis for this study is that valuable follow-up information can be obtained from general practitioners (GPs) via a postal system. Thus a more complete picture of the outcome of the whole population of childhood cancer survivors can be built up.
The West Midlands Regional Children's Tumor Registry (WMRCTR) was established in 1984 at BCH to record data on all cases of malignancy diagnosed in West Midlands children since 1957. Approval for the WMRCTR (including follow-up) was obtained from all appropriate Ethics Committees.
Since 1993, the WMRCTR has maintained a 5-year rolling program of postal follow-up of the patients who were never treated at BCH, were discharged before the regional service was instituted, have defaulted from follow-up clinics, or have moved out of the region. A questionnaire is sent to the GP every 5 years, seeking medical information, as well as health, education and social issues. If the patient has left the practice, the NHS Tracing Service (NHS STS) is consulted to trace the current GP. The database also records brief treatment details, in order to link up with subsequently reported information on health problems.
In order to achieve consistency and comparison with other reports from our center 1, 10, the same definitions and classifications of “health problems” were used in the construction of the database and the results of this study.
In the West Midlands, 2,331 patients had survived cancer for 5 years or more (Fig. 1). Of these, 1,222 remain on clinical follow-up; this study focuses on the 935 (44%) current survivors who do not attend and are included in the postal system.
Between 1993 and December 2004, 1,027 patients were followed up; replies were received on 903 (response rate 88% at June 1st 2005). Of these, 20 were written refusals to take part for reasons including workload, finance, and concerns over patient consent. The outstanding 124 patients were confirmed through the NHS STS as being alive. The total number of patients recorded as alive is 935, there have been 42 deaths (4%) and 50 (5%) patients are lost to follow-up, owing to emigration 20, GP refusal 20, and no trace 10. These 50 cases are not included in the analysis.
Questionnaires were sent to 805 GPs within the West Midlands region and 222 outside. Of the former, 13% (105) did not reply or refused, compared with 15% (34) of the others, showing very little difference in response rates between local and non-local GPs. The majority of replies (80%) were received within 1 month (median 12 days, range 1 day to 1 year).
The total number followed up included 545 (53%) males and 482 females, with 496 males and 439 females (91% each) still surviving. The current age distribution of those survivors who are not lost to follow-up is shown in Figure 2. The median length of follow-up of survivors was 23 years 1 month (range 2 months–46 years 11 months).
Information was received on 42 deaths, 18 (43%) of which we were previously unaware of. Seventeen patients (41%) died of tumor progression or recurrence, a further 9 (21%) of a subsequent malignancy and the other 16 (38%) of a variety of non-cancer causes, including cardiac conditions 3, neurological conditions 4, suicide 2, and road traffic accidents 2.
In the 935 patients still alive and not lost to follow-up, the primary diagnosis was brain/CNS tumor in 300 (32%), leukemia in 166 (18%), and lymphomas and other solid tumors in 469 (50%). Initial treatment comprised: surgery in 66%; radiotherapy in 50%, over half of whom had cranial irradiation; and chemotherapy in 38%. Over half (55%) of the patients receiving chemotherapy alone had no reported problems compared with only 33% of patients undergoing surgery alone. Cranial irradiation was given to 270 patients, of whom 43% were said to be free of long-term effects.
GPs reported no medical problems in 340 (36.4%) patients and in the others, endocrine complications were the most frequent. Eleven (31%) of the 35 patients with recorded cardiac problems had chemotherapy of whom only 4 had received anthracyclines. Table I shows the breakdown of medical problems by original diagnostic group, with examples of some conditions which might be of relevance to GP attendance. Endocrine dysfunction is the biggest reported problem for the leukemia survivors, almost as common as in the brain tumor survivors. As might be expected, the latter had the greatest number of problems per patient.
Table I. Breakdown of Long-Term Effects (LTE) in Childhood Cancer Survivors by Original Diagnostic Group
Total with problems
Many patients had more than one LTE.
Respiratory (inc asthma)
Overall, visual effects were the most common single problem (17% of all survivors); these ranged from complete blindness to nystagmus, particularly affecting retinoblastoma and brain tumor survivors.
Only 229 survivors (24.5%) were confirmed to have children. Thirty-three miscarriages were reported in 23 patients, as well as 1 termination and 1 stillbirth both due to known fetal abnormalities. Health problems were noted in the offspring of 23 patients, including 11 children with malignancies (9 familial retinoblastomas, 1 familial Wilms tumor, and a thyroid carcinoma in the daughter of a woman with MEN IIB).
Not all GPs had the opportunity to respond to the questions about smoking status, or education/employment as the questionnaire was only updated to include these issues in 2001. However, the proportion (15%) of our survivors so far known to smoke is much smaller than the national average of 26% 15, One fifth of GPs who responded did not know whether or not their patients smoked. Questions on education/employment were less fully answered. Only 13% of patients were confirmed by their GP as progressing to further education and 45% of GPs did not know whether or not their patient was employed.
Subsequent tumors were reported in 56 patients, in 17 these were benign. Table II describes the 44 malignancies in the other 39 patients, 20 of which had previously been unknown to us. Patients with lymphomas have the greatest potential for further malignancy; 11 of the 15 tumors in this group developed in Hodgkin lymphoma survivors and 3 of the 5 patients who developed a third malignancy had originally had lymphoma.
Table II. Subsequent Malignancies by Diagnostic Group of Original Tumor
Number of patients
Proportion of whole group (%)
Brain and CNS
Other solid tumors
The most common second malignancy was basal cell carcinoma 11, always in areas of radiotherapy treatment. Brain/CNS tumors were the next most common (six patients, five of whom had had cranial irradiation). There were three soft tissue, four thyroid and four breast tumors, three osteosarcomas and three testicular tumors. Four of the other 10 malignancies developed in the digestive tract.
Long-term follow-up of childhood cancer survivors is essential for three reasons: the monitoring and maintenance of their health and well being; to inform clinical practice; and to enable research into the effects of cancer and its treatment in this special population.
The most comprehensive long-term follow-up information is recorded at hospital follow-up by clinicians with specialist knowledge of the disease and an awareness of potential problems 10. The UK National Institute for Health and Clinical Excellence (NICE) has made recommendations for long-term clinical follow-up procedures 16, stressing the importance of continued communication between the clinical team and primary care. Other discussions of long-term follow-up care models have also focused on low-risk patients who can safely be discharged to telephone or postal follow-up 17, but not all high-risk adolescent or adult patients wish to continue to attend even special adult late effects clinics. Many move away and others may not wish to be reminded of their illness. Thus, clinical discharge of selected patients is not the only reason for seeking information elsewhere and this study sought to evaluate the feasibility of a postal follow-up system.
The results show the value of a specialist tumor registry, which can undertake postal follow-up on a large number of patients, obviating the need for clinical resources to be used 11. Such a model has been proposed 18, in order to complement clinical follow-up. Follow-up notifications are occasionally missed and the fact that we were first informed of several subsequent malignancies and almost half of the late deaths by the GPs confirms the valuable contribution of postal follow-up.
GPs are cautious about sharing medical information, as shown by the few who refused to complete the questionnaire without patient consent. Issues of consent and confidentiality have become more significant in the UK since the Data Protection Act 1998 19, although further legislation 20 has allowed information to be passed legally to cancer registries. Almost 90% of GPs provided information, possibly demonstrating their confidence in the cancer registration system. Many studies have investigated GPs' response rates to postal questionnaires 11, 21–23; the consensus is that these rise with their perceived value of the subject matter. Our GPs have confirmed the importance of follow-up of these patients.
It is surprising that a fifth of GPs who responded did not know if their patients smoked, as this group might be particularly vulnerable to the effects of smoking and might have had more advice about it 24–25. However, it is rewarding to see the low percentage of smokers in our group, compared to the national average. It is not surprising that education and employment status were largely unknown, as these are not usually recorded in GP notes. This suggests that, in order to obtain a more complete picture of the patients' lives for quality of life assessment, two questionnaires would be the optimal: one to the GP for medical information, the other to the patient for social data. However, when this was attempted as a pilot exercise by the WMRCTR, only 9/40 questionnaires were returned by the patients (22.5%), possibly indicating their unwillingness to be reminded of their medical history.
The definition of a health problem was carefully considered and while some of those recorded may be unrelated to the disease or its treatment (e.g., asthma), they have been included for completeness, given that the GPs chose to report them. Overall, the proportion of the postal patient reported medical problems differed from those seen in a follow-up group reported from our center 1, although in both groups, the major problem was endocrine dysfunction. In the hospital group, obesity was the most commonly recorded single problem (15.9%), but in the postal group, this was reported in only 1.6% of cases which may reflect GP under-reporting rather than a genuine lower incidence.
The reason for the variations between the two groups is not clear, but it could be related to under-reporting by the patients, who may not perceive the significance of some symptoms. The GPs are probably unaware of the potential risks, whereas in the hospital follow-up clinic, the clinicians specifically test for endocrine deficiencies which may be asymptomatic.
The adverse effects of radiation on the developing brain, particularly growth, endocrine and neuropsychological, are well described, thus the 43% of patients who had no reported problems following cranial irradiation would seem to be an underestimate of the expected figure. Although this group may have received lower doses than patients on hospital follow-up, it raises concerns that some adverse effects are being missed in the primary care setting. The recall of some patients currently on postal follow-up could be justified, particularly if they are at risk of hypopituitarism following cranial irradiation, or of cardiac dysfunction following anthracyclines. There is an ethical precedent for this, in the 2003 UK Department of Health recall of females who received mediastinal irradiation for Hodgkin's lymphoma in childhood, as they were deemed to be at increased risk of breast cancer 26. However, the postal population may simply be healthier and genuinely experience fewer late effects.
This study experienced non-response and missing data inherent to all survey based research. Although under-reporting may be confined to social aspects of health (e.g., smoking), this could also apply to medical information, thus underestimating the scale of the problem. The interval between letters to the GP will be decreased to 3 years, in order to keep more up to date with the health of the survivors.
This study demonstrates the value and feasibility of a postal follow-up program for survivors who would otherwise be lost to follow-up. The high response rate likely indicates that GPs appreciate the value of such work and trust the measures employed by the WMRCTR to protect their patients' information. Without these data, the global picture of the scale of late effects would be restricted and limited to the study of those for whom hospital follow-up is deemed clinically essential and who are willing to attend. These results also demonstrate that long-term effects do exist in some patients who are apparently well enough to be discharged and that more detailed clinical review of certain survivors who are not monitored routinely should be considered.
These survivors appear to differ from those on hospital follow-up and further research is necessary to understand why. These patients may be a selected group, or may truly experience different health outcomes. By assessing the problems experienced by large numbers of survivors, we can determine the late effects they experience and the monitoring they will need. GPs might not be aware of all the potential problems, nor of the existence of hospital long-term follow-up clinics and could be advised about which investigations are appropriate within the primary care setting and when referral for hospital review would be indicated 27. The NICE guidance 16 stressed the importance of the role of a “key worker” to form a bridge between the patient and the clinical team. This role could be filled either by a specialist nurse in the hospital setting, or be transferred to the primary care team for those patients who do not attend. Specific long-term follow-up protocols need to continue to be established for use in the primary care setting 28, 29. Postal follow-up is not simply an academic exercise, but is of clinical benefit to the patients, and makes an important contribution to our understanding of the overall picture of the health of all childhood cancer survivors.
We thank all of the general practitioners who gave their time to respond to our questionnaires, without whose help this research would not be possible.