Introduction
In England, many patients request either new or repeat prescriptions by phoning, writing to or calling at the surgery. Some may use fax or email. Once the prescription is prepared, the patient collects it and takes it to the pharmacy. Some of these prepared prescriptions are never collected. Most research on non-compliance has focussed upon secondary non-compliance (ie, on whether patients take their drugs as intended by the prescribing doctor). A limited amount of research has been conducted on prescriptions issued directly to patients but never dispensed (primary non-compliance). One study found that 20% of prescriptions were not cashed (Rashid, Reference Rashid1982) but others have found lower rates (2.6–7.6%) (Waters et al., Reference Waters, Gould and Lunn1976; Winter, Reference Winter1982; Begg, Reference Begg1984; Stuart, Reference Stuart1985; Beardon et al., Reference Beardon, McGilchrist, McKendrick, McDevitt and MacDonald1993). No study has concentrated on prescriptions requested by the patient but then not collected from the practice.
Although uncollected prescriptions are only a small proportion of the total issued, there is a lack of knowledge about why they are not collected and what, if anything, should be done with them. Their non-collection could be an administrative or a compliance problem. We wished to explore this issue.
Method
We undertook a descriptive cross-sectional analysis to identify what types of prescription items were uncollected and whether the prescription was acute or a repeat. Ethical Committee and Research Governance approval were obtained. Data were collected between November 2003 and March 2004.
All 33 practices in the Gateshead Primary Care Trust (PCT) were invited by letter to take part. We used a definition of a repeat prescription as ‘one printed by a practice computer from its repeat prescribing programme within a given period’ (Harris and Dajda, Reference Harris and Dajda1996). List size data and the number of principals in each practice were obtained from computer records at the Gateshead Primary Care Trust.
There is no agreed definition of an uncollected prescription. For the purpose of this research, it was defined as ‘a prescription that had not been collected for at least four weeks from the issue date and which the practice did not now expect to be collected’. Data on uncollected prescriptions issued from at least four up to eight weeks previously were collected from 16 practices. In three practices that destroyed their prescriptions promptly at the end of each calendar month, a month’s worth of prescriptions that these practices believed would not be collected was recorded instead.
A researcher (MS) visited each participating practice and recorded the items on each uncollected prescription by British National Formulary (BNF) subsection (see Box 1). A frequency list of all items (by BNF subsections) prescribed over the same period was obtained in each practice from the practice computer. The counts of prescription items from each chapter (eg, cardiovascular, respiratory) were compared to the corresponding counts from the rest of the drugs in the study and a χ 2 test performed to see if the differences were significant.
Box 1 Prescription data recorded from uncollected prescriptions
• Practice identification number.
• Prescription number for that practice.
• Patient’s age on the day prescription issued.
• Own home or Aged Person’s Home/Nursing Home.
• Name of drug, dose and instructions.
• Quantity supplied.
• Length of prescription in days where possible to determine.
• BNF category of drug.
• Whether repeat or acute.
• Number of drugs on repeat (determined either from written information on the prescription or from computer data).
Drugs were also ranked according to the Belfast Classification (McGavock and Wilson-Davis, Reference McGavock and Wilson-Davis2000; McGavock, Reference McGavock2004), which is based on a GP’s declared certainty of diagnosis at the moment of prescribing (see Box 2). There are three categories that usually include about 85% of all items prescribed:
• Drugs in the specific diagnosis category include anticoagulants, antidepressants, anticonvulsants, heart preparations, hypoglycaemic, antiparkinsonian, and thyroid/antithyroid agents.
• Drugs in the symptomatic diagnosis category include minor analgesics, antacids, hypnotics and laxatives.
• Drugs in the often presumptive diagnosis category include acid suppressants, antimicrobials, anxiolytics and vaginal preparations.
Box 2 Belfast Classification
We evaluated whether there was a statistical association between the Belfast Classification score and collection of the prescriptions by χ 2 test. Finally, the data were grouped into the number of items collected and uncollected in each BNF chapter.
To attempt to find out why prescriptions were uncollected, 10 practices were revisited. Eight of the 10 practices were visited between one month and five months after the initial visit. The remaining two practices had destroyed most of their prescriptions on the researcher’s first visit. Both practices agreed to the selection of suitable patients for interview, so both tasks (ie, collection of basic data and identification of patients for interview) were done during the same visit in these two practices. Identical data were collected in all the practices, patient records examined and all patients who appeared to have no obvious reason for an uncollected prescription (eg, reissue of prescription, hospital admission) were invited to participate in a telephone interview. Patients for whom such an invitation may have caused embarrassment or distress were excluded, eg, people with mental health problems or those with an uncollected oral contraceptive pill prescription. Semi-structured telephone interviews were carried out with those who responded. An audit was also carried out on patients from the lead authors own practice for whom there was an uncollected prescription: they were telephoned and the purpose of the audit explained and with their permission their comments recorded.
Face-to-face semi-structured interviews with open-ended questions to allow unstructured responses were also held with a member of the primary health care team (PHCT) from each practice who knew about their current system.
Interviews with patients from other practices and the PHCT were transcribed and entered into a qualitative database (QSR NUD*IST) (Scolari, 1995). Codes, categories and themes were identified in PHCT members’ and patients’ narratives. Comments from the lead author’s own patients were entered into an Excel database and analysed.
Results
Collection and analysis of basic data from the practices
Twenty of the 33 practices in the PCT agreed to participate. Data collection was satisfactory in 19. In one practice there was an uneven spread of prescriptions suggesting many of the earlier prescriptions had been destroyed: this practice’s data were therefore excluded from the analysis.
These 19 practices covered 134 056, ie, 63.9% of the total number of patients in the Gateshead Primary Care Trust and 66% of Gateshead GPs. The practices appeared to be a representative sample although there was a slightly lower proportion of single-handed practices than in the PCT as a whole (15% versus 24%).
Proportion of uncollected prescriptions
The total number of items issued by the 19 practices was 165 155, of which 823 were not collected (0.5%). The proportion that was uncollected ranged from 0.032% to 2.4% among practices. Of the uncollected, 214 were acute, 587 repeat and 17 automatic (ie, printed automatically by the computer each month without being requested); the status of five was not recorded.
Belfast Classification
The BNF sections were divided into therapeutic groups based on the certainty of diagnosis according to the Belfast Classification. Of the total number of items issued, 85% could be categorised in this way (see Table 1).
Table 1 Number of prescription items ranked according to the Belfast Classification
There was a statistically significant association between the collection of prescriptions and the classification score (χ 2 = 13.21, df = 2, P = 0.001), with drugs for more specific diagnoses being more likely to be collected.
Analysis by BNF chapters and subsections
Dermatological and respiratory drugs were most likely to be uncollected and cardiovascular and endocrine drugs least likely to be uncollected compared to the rest of the drugs in the study (see Table 2).
Table 2 Uncollected items by BNF chapter
By BNF subsection, antihistamines, compound alginates, corticosteroid inhalers, beta agonist inhalers and emollients were the most likely to remain uncollected, while cardiac drugs (nitrates, ACE inhibitors, calcium channel blockers, diuretics, statins), biguanides and broad-spectrum antibiotics were less likely than average to be uncollected. Some drugs, eg, benzodiazepine hypnotics, were always collected.
Qualitative data
Telephone interviews with patients
From the 10 practices, 254 people had an uncollected prescription (445 items). For 50 people, there was an apparent reason, eg, reissued by the receptionist at least once on the same day (28%), another issued by the receptionist within seven days either prospectively or retrospectively (40%), the patient had had frequent prescriptions and it could be assumed that they had plenty of medication (10%), the prescription had been issued later in a consultation (6%), the prescription belonged to the practice (6%), temporary resident (2%), locum gave patient a prescription they did not want (2%), patient admitted to hospital (2%), father bought OTC medication for the child (2%), doctor told patient to pick up prescription ‘if need be’ (2%).
Twenty-four people were deemed unsuitable to contact and 180 people were therefore invited for telephone interview. The response rate was poor: only 10 agreed to be interviewed. Of these, nine said that they had got a prescription and their medication. Only one acknowledged that he had forgotten and had not obtained the medicine.
Interviews with members of the primary healthcare team
Interviews were held with a member of staff from each of the 20 practices, the 19 practices and the lead author’s practice (12 managers/administrators, 6 receptionists and 3 GPs). In one practice, a doctor was interviewed initially followed by the senior receptionist who knew more about their system. There were therefore 21 interviews. Several administrative reasons for non-collection were identified, eg, if a prescription was missing, misfiled, awaiting signature, at a branch surgery, stuck to another prescription or a current item split from repeat, then it would be reprinted. Practices that relied heavily on pharmacists picking up prescriptions (to be later collected from the pharmacy by the patient) tended to have fewer uncollected prescriptions, whereas practices who asked patients to order well in advance or who printed some items automatically tended to have more. Few practices had a formal system for dealing with uncollected prescriptions and most practices shredded them with no further action. No one could recall any problems other than administrative ones resulting from uncollected prescriptions. Generally, uncollected prescriptions were not perceived as a problem.
Interviews with patients from the author’s practice
In 31 out of 103 consecutive uncollected prescriptions in MS’s own practice a reason was apparent (the majority were administrative). Of the 69 people eligible to contact, 57 were successfully telephoned and all were willing to comment. Over three quarters (43/57) claimed to have obtained their medication. Fourteen patients had not obtained their medication: two of these may have benefited from treatment (iron tablets and antihypertensives) but both had shown continued poor compliance in the past. Only seven people reported forgetfulness.
Discussion
Uncollected prescriptions are a small but common problem that varies among practices. Drugs that are linked to a specific diagnosis (and may be essential for the patient’s health) are least likely to remain uncollected. This could be because patients realise the importance of compliance or because people on essential medications who may be elderly, frail or have cognitive impairment have their prescriptions delivered by the pharmacist or have carers who ensure good compliance. Dermatological and respiratory drugs are significantly less likely to be collected than other drugs, perhaps reflecting their common use as necessary.
Many causes for uncollected prescriptions appear to be administrative rather than compliance issues and are rarely due to patient error or forgetfulness. Most of the patients interviewed were certain that they had obtained or had not run out of their medication, though we could not confirm this. No patient or PHCT member could recall any harm resulting to a patient from an uncollected prescription.
Although no work was identified on uncollected prescriptions, two British studies on undispensed prescriptions in 1976 (Waters et al., Reference Waters, Gould and Lunn1976) and in 1993 (Beardon et al., Reference Beardon, McGilchrist, McKendrick, McDevitt and MacDonald1993) found that cardiovascular system drugs were rarely rejected. Research in the USA studied prescriptions (most automatically transmitted to a pharmacy) that had not been claimed at the pharmacy by the patient. This system is very different from that in British General Practice and is probably nearer to the situation whereby a patient is handed a prescription but does not bother to get it dispensed. Medications most frequently unclaimed were anti-infectives (differing from Britain where these are usually prescribed during a consultation). Interestingly, many similar drugs to this study, analgesics, dermatological, anti-inflammatory, respiratory and cough, cold and allergy drugs, were also unclaimed (Craghead and Wartski, Reference Craghead and Wartski1991; Farmer, Reference Farmer and Gumbhir1992; Kirking and Kirking, Reference Kirking and Kirking1993; Hamilton and Hopkins, Reference Hamilton and Hopkins1997; Secnik, Reference Secnik, Pathak and Cohen2000).
One study in the USA studied prescriptions handed in by patients to the pharmacy but later unclaimed: many had obtained their medication elsewhere or had enough left (Hamilton and Hopkins, Reference Hamilton and Hopkins1997). Three further studies in the USA, though in a hospital setting, also found that the much apparent non-compliance could be explained either by administrative reasons or by obtaining correct medication in other ways (Kirking et al., Reference Kirking, Zaleon and Kirking1995; Dong et al., Reference Dong, McCart and Lynch1998; Kinnaird et al., Reference Kinnaird, Cox and Wilson2003).
Strengths and limitations of this study
This study collected almost complete data from the participating practices and is likely to reflect the true extent of the problem. The interview response rate from patients other than the lead researcher’s own was extremely poor and it is likely that those who were interviewed were a very biased sample making it difficult to draw conclusions. Nevertheless, their responses were supported by those of the other groups of interviewees.
The limited data suggest that the causes of uncollected prescriptions are often administrative rather than compliance issues, that patients will ensure that they obtain essential medication and that patients do not come to harm as a result. Current common practice of disposing of uncollected prescriptions therefore seems to be safe.
Further research could focus on securing more interviews and of confirming patients’ statements to have accessed their drugs when there is no record of a prescription. In the future, the introduction of direct electronic transfer of prescriptions to a pharmacy may shift the burden of unclaimed prescriptions to the pharmacy.
Funding body
None.
Ethics Committee
Gateshead and South Tyneside Local Research Ethics committee. Reference number GH2003:43.
Competing interests
None.
Acknowledgements
We would like to thank the GPs, their staff and patients from all of the participating practices, Janette Stephenson, prescribing Advisor at Gateshead Primary Care Trust for allowing MS access to data, Bill Gibson, Statistician, for his advice on data collection and entry and Michael Edwards (student) for assisting with data entry. This work was undertaken as part of the MSc in Prescribing Sciences, University of Liverpool and we would like to thank the staff of the department for their assistance.
