To analyse health care use of cancer patients in general practice we used routinely collected data from the Netherlands Information Network of General Practice (LINH). We compared the health care use of cancer patients during the follow-up phase with age and gender matched control patients from the same practice in a retrospective cohort study.
The Netherlands Information Network of General Practice (LINH) is a representative network consisting of 92 general practices from all over the Netherlands with about 350,000 enlisted patients. In the Netherlands, every inhabitant is listed with a general practitioner. The LINH database holds longitudinal data on GP care, including patient contacts, referrals, prescription medicines, and relevant health problems. Health problems are coded by the GP, using the International Classification of Primary Care (ICPC) [9
]. The database is used for health services research and quality of care research. For more information on the network see http://www.linh.nl
Patients and controls
Based on ICPC-codes [see paragraph below] we selected all patients for whom a diagnosis of cancer was recorded between January 1, 2001 and December 31, 2007. This resulted in 8,703 cancer patients, 44% males and 56% females for whom health care data were available on a minimum of 30
days during the follow-up phase (see further). The age distribution of the cancer patients is shown in Figure .
Number of patients with cancer by gender and age group (absolute numbers; males n=3800, females n=4903).
Cancer patients were matched for age (in five year intervals) and gender with patients without a diagnosis of cancer, enlisted in the same general practice. We could include 8,672 control patients (data not shown).
Type of cancer
For 69.2% of all selected patients with cancer (N=8,703) one of the following cancer types was recorded: breast cancer (ICPC code X76, 27.7% of all patients), colon/rectal cancer (D75, 13.6% of all patients), prostate cancer (Y77, 13.5% of all patients), lung/bronchus cancer (R84, 9.3% of all patients) and cancer of the bladder (U76, 5.1% of all patients).
For 70.0% of the female patients with cancer (N=4,903) one of the following cancer types was recorded: breast cancer (X76, 49.3% of the women), colon/rectal cancer (D75, 11.8% of the women), and malignancies of female reproductive organs (X77, 8.9% of the women).
For 61.2% of the male patients with cancer (N=3,800) one of the following cancer types was recorded: prostate cancer (Y77, 30.9% of the men), colon/rectum cancer (D75, 15.8% of the men) and lung/bronchus cancer (R84, 14.4% of the men).
Other cancer types with low numbers of patients included the following ICPC coded cancer types: A79 (malignancy not otherwise specified (NOS)), B72 (Hodgkin's disease/lymphoma), B73 (leukaemia), B74 (malignant neoplasm blood other), D74 (malignant neoplasm stomach), D76 (malignant neoplasm pancreas), D77 (malignant neoplasm digestive system other/NOS), N74 (malignant neoplasm nervous system), R85 (other malignant neoplasm respiratory), S77 (malignant neoplasm of the skin, excluding squamous and basal cell carcinoma), T71 (malignant neoplasm thyroid), U75 (malignant neoplasm of kidney), U77 (malignant neoplasm urinary other), W72 (malignant neoplasm related to pregnancy), X75 (malignant neoplasm cervix uteri), Y78 (malignant neoplasm male genital other).
Follow-up phase and observation period
We studied use of health care in general practice of cancer patients during the follow-up phase. We defined this phase as starting after the end of the initial treatment and ending before the terminal phase started. Both the initial treatment phase and the terminal phase were excluded, because during these periods patients might have special health care needs that are not representative for the remainder of this period. We set the end of initial treatment to be six months after the first recording of the diagnosis in the electronic medical record (EMR) in the practice and the start of the terminal phase on three months before death or the last registration in the EMR. For reasons of reliability we included only those patients of whom an observation period of at least 30
days was available.
All health care data recorded in the EMR between January 1, 2001 and December 31, 2007 were included in the analyses. Of all patients 78% was still alive at the end of the observation period, 17% had died, and 5% of the patients left the practice because they moved. The length of the observation period differs for every patient, depending on the date of diagnosis. For the most frequently represented types of cancer we found the following median lengths of the observation period: breast cancer (X76) 644
days, colon/rectal cancer (D75) 556
days, prostate cancer (Y77) 598
days, lung/bronchial cancer (R84) 432
days, and cancer of the bladder (U76) 612
Comorbid chronic conditions
For every patient the presence of comorbid chronic conditions was determined using ICPC coded diagnoses recorded during contacts, prescriptions or referrals. We used the classification for comorbid chronic conditions as described by Knottnerus et al. [10
]: sensory conditions (including ear and eye conditions), heart disease, musculoskeletal system, neurological-movement conditions, neurological-pain conditions, psychiatric conditions, respiratory conditions, skin conditions, endocrinological conditions, and urogenital conditions.
Health care use
The EMR encompasses routinely recorded data for every patient enlisted in the practice, including contacts with the practice, morbidity, referrals to other health care providers and drugs prescribed. The total number of contacts with general practice includes face to face consultations with the GP, home visits, telephone contacts for prescriptions, vaccinations, email contacts, and administrative activities. For all data recorded during the observation period numbers were recalculated to a one year’s period.
In the Netherlands the GP functions as the ‘gatekeeper’ of care, meaning that patients need a referral for specialist health care or for other primary health care workers [11
]. The number of referrals can be seen as an indication of the ability of GPs to deal with requests for treatment themselves. We calculated the mean number of new referrals per patient per year.
Cancer patients were compared with age- and gender-matched control patients with regard to the number of contacts with general practice, diagnoses of comorbid conditions, rate of prescribing, and the referral rate in a year.
Furthermore, we studied the influence of having a comorbid chronic condition on health care use. The group of cancer patients without comorbid chronic conditions was compared with the matched control group without comorbid chronic conditions while the group of cancer patients with comorbid chronic conditions was compared with the matched control group with comorbid chronic conditions.
We calculated mean figures for the number of contacts with the practice, prescriptions and referrals, and tested for statistical significance of the differences between the two groups using univariate variance analyses (Student’st
-test). Because of our large sample size, p=0. 01 was set as cut-off value for statistical significance.