Abstract
Abstract
Objectives
To describe the prevalence of sub-optimal monitoring for selected higher-risk medicines in older community-dwelling adults and to evaluate patient characteristics and outcomes associated with sub-optimal monitoring.
Study design
Retrospective observational study (2011–2015) using historical general practice-based cohort data and linked dispensing data from a national pharmacy claims database.
Setting
Irish primary care.
Participants
625 community-dwelling adults aged ≥70 years and prescribed at least one higher-risk medicine during the 5-year study period.
Primary and secondary outcome measures
The primary outcome was the prevalence of sub-optimal laboratory monitoring using a composite measure of published medication monitoring indicators, with a focus on commonly prescribed higher-risk medicines such as diuretics and anticoagulants. Poisson regression was used to assess the patient characteristics associated with sub-optimal monitoring and explanatory variables included the number of medicines, age, sex, deprivation and anxiety/depression symptoms. Logistic regression was used to explore the association between baseline sub-optimal monitoring and the odds of adverse health outcomes (unplanned healthcare utilisation, adverse drug reactions and mortality).
Results
Of 625 participants, the mean age was 77.7 years, 53% were female, the mean number of drugs was 7.3 (SD 3.3) and 499 (79.8%) had ≥1 unmonitored dispensing over 5 years. The number of drugs, deprivation and anxiety/depression symptoms were significantly associated with sub-optimal monitoring, with the strongest association seen for anxiety/depression symptoms (incidence rate ratio: 1.33, 95% CI 1.05 to 1.68). There was a small but significant association between baseline sub-optimal monitoring and emergency department visits at follow-up, but no evidence of an association with unplanned hospital admissions, mortality or adverse drug reactions.
Conclusion
The prevalence of sub-optimal medication monitoring was high, and number of drugs, deprivation and anxiety/depression symptoms were significantly associated with sub-optimal monitoring. However, the public health impact of these findings remains uncertain, as there was no clear evidence of an association between sub-optimal monitoring and adverse health outcomes. Further research is needed to evaluate the effect of improved monitoring strategies and the optimal timing for drug monitoring of higher risk medications.
Keywords: Prescriptions, Polypharmacy, Primary Care, Safety
Strengths and limitations of this study.
This study describes the prevalence and predictors of sub-optimal laboratory monitoring of higher-risk medicines in Irish primary care.
The inclusion of patient reported outcome measures is a key strength and offers critical insights into patient experienced barriers to optimal monitoring.
The absence of universal electronic transfer of results between primary and secondary care could have impacted data quality, potentially leading to an overestimation of sub-optimal monitoring.
This study focused on the prevalence of sub-optimal monitoring by assessing the number of days between dispensing and the relevant test, but we did not assess the outcome of laboratory monitoring.
Introduction
Advances in public health and healthcare has led to a growing population of older people living with multimorbidity and resultant polypharmacy with their attendant risks.1 Prescribing for this population is complex due to the potential for both drug-drug and drug-disease interactions and due to physiological changes associated with ageing.2 The bulk of this prescribing occurs in primary care, including ongoing prescribing of medicines initiated in secondary care, and with that the responsibility for monitoring.3 Older adults are the largest consumers of medicines and due to physiological changes associated with ageing are more susceptible to adverse drug events. It is estimated that up to 10% of emergency admissions in older adults are due to medication related harm, with approximately one-fifth of those related to sub-optimal drug monitoring.4 These estimates are from over 10 years ago, and there is a paucity of recent studies investigating this important patient safety issue.
Explicit indicators of medication appropriateness have been used extensively in pharmaco-epidemiological research to characterise the appropriateness of prescribing and to assess the association between potentially inappropriate prescribing (PIP) and patient outcomes, for example, the US Beers criteria and the recently updated European Screening Tool for Older People’s potentially inappropriate Prescriptions criteria.5 Multiple observational studies have demonstrated an association between PIP, measured using these indicators, and clinical outcomes such as increased emergency admissions, adverse drug reactions (ADRs), functional decline and reduced health related quality of life.6,9 More recently, several research groups have developed explicit indicators for the laboratory monitoring of specified higher-risk medications prescribed in primary care.10 These indicators were developed from consensus validation processes and target inconsistent monitoring for drugs implicated in potentially preventable medication-related harm in primary care.10 11 These indicators assess the completion of laboratory monitoring blood tests for specified higher risk prescriptions. Randomised controlled trials of explicit indicators of medication appropriateness have demonstrated some effectiveness of these tools in improving prescribing with the advantage of being reproducible.12 However, research examining laboratory monitoring of higher risk medicines has been limited to date despite the high frequency of ADRs thought to be related to sub-optimal monitoring. This is due, in part, to routine primary care datasets not collecting this information.13
In the UK, there has been an increase in the rate of laboratory blood monitoring in general practice from an average of 1.5 tests per person in 2000 to 5 tests per person in 2015. The rate of testing increased significantly across all age groups after the introduction of the Quality and Outcomes Framework, a system introduced to incentivise general practitioners (GPs) for providing quality care to patients where financial rewards were linked to reaching certain targets.14 More recent data indicate that the rate of testing is no longer increasing but that there is considerable practice level variation in the rate of testing, perhaps reflecting uncertainty over the most appropriate testing frequencies for different clinical and medication indications.15 In 2012, a retrospective case note review of over 6000 prescribed items in English primary care revealed the prevalence of sub-optimal monitoring to be 6.9% (95% CI 5.2% to 8.9%),3 for medicines identified as needing monitoring. This study examined prescriptions from a random sample of the practice population; however, older adults may be particularly at risk from sub-optimal monitoring due to the physiological changes associated with ageing, particularly decline in renal function.
The aim of this study was to evaluate the laboratory monitoring of specified higher-risk medications prescribed by GPs in a cohort of older (aged ≥70 years) community-dwelling adults over 5 years follow-up (2011–2015). Secondary objectives were to: (i) assess patient characteristics associated with sub-optimal monitoring and (ii) assess whether sub-optimal monitoring is associated with adverse health outcomes (ADRs, emergency admissions and mortality) over time.
Methods
The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were adhered to in the reporting of this study16 (see online supplemental appendix 1).
Patient and public involvement
This research had significant public and patient involvement (PPI) at the design stage. The PI (EW) met with an older person Primary Care PPI group consisting of seven members (five women and two men) in September 2019. The purpose of this meeting was to ask for feedback on key study concepts, including polypharmacy, attitudes towards balancing medication risks and benefits, and experiences of medication-related harm.
Setting
This study was set in general practice in the Ireland, where there is a mix of eligibility and entitlement to services for primary care, with secondary care available for all. Since 2001, all adults aged ≥70 years are entitled to free GP visits, and a majority are entitled to a means tested General Medical Services (GMS) card (that qualifies the patient for free prescription medicines in addition to free GP visits). During the study period, there was no structured programme for chronic disease management in primary care which was delivered on an ad hoc basis between primary and secondary care.17
Study population and study process
The Centre for Primary Care Research (CPCR) elderly cohort was established in 2010 for the purpose of examining the association between potentially inappropriate prescribing (PIP) and patient-reported health outcomes in community-dwelling older (aged ≥70 years) adults.6 This is a retrospective study using data from the CPCR elderly cohort. Patients were recruited from 15 practices in Leinster, Ireland, using proportionate stratified random sampling18 and were eligible to participate if they were aged ≥70 years, taking at least one medication and in receipt of a valid GMS card. Patients were excluded if they were receiving palliative care or unable to complete written patient questionnaires (moderate-severe cognitive impairment, active psychosis, significant visual impairment), as judged by their GP.6 Consent was obtained from participants to link their GP electronic medical record data and patient questionnaire data with their dispensed prescriptions from the National Health Service Executive-Primary Care Reimbursement Service (HSE-PCRS) database. The HSE-PCRS database contains individual patient data, specifically GMS card numbers, date of birth, sex, medication information such as World Health Organization-Anatomical Therapeutic Chemical code, strength, quantity dispensed and date of dispensing, which typically occur on a monthly basis. The HSE-PCRS database only captures dispensed medicines for patients with a valid GMS medical card.
At baseline in 2010, 904 of the 1487 invited eligible participants were recruited, giving a response rate of 61%. Follow-up data collection took place in 2012 and 2016. Dispensed prescriptions data (2010–2016) were obtained retrospectively in 2021, for 793 participants who provided consent, from the HSE-PCRS database and linked to cohort data using patient GMS card numbers. The numbers of participants with PCRS data per annum and the cohort losses to follow-up are described in figure 1. Data for 2016 may have been incomplete due to delays in updating the electronic patient record and scanning hard copies of laboratory results (see online supplemental appendix 2). It was thus decided to examine monitoring prevalence until the end of 2015.
Figure 1. Study flow diagram. GMS, general medical services; PCRS, primary care reimbursement service.
Study outcomes
Primary objective: prevalence of laboratory monitoring
Laboratory monitoring indicators for a selection of higher risk medicines were included from the literature including the UK Data-driven Quality Improvement in Primary Care medication safety criteria,10 the UK pharmacist-led information technology intervention for medication errors (PINCER) prescribing safety indicators11 and the UK National Health Service Specialist Pharmacy Service (NHS SPS) drug monitoring guidelines.19 A full list of these expert consensus-developed indicators was compiled by the research team, which included pharmacists, GPs and a statistician. Drugs with a low prevalence in our dataset (eg, leflunomide), and those where monitoring typically occurs in secondary care (eg, lithium) were excluded. Give that the threshold for optimal monitoring can vary across different clinical guidelines and in clinical practice across different patient contexts, where an indicator was included in more than one set of criteria, the longer recommended time interval for blood monitoring was used (see online supplemental appendix 3).
The primary outcome of interest was the prevalence of sub-optimal laboratory monitoring using a composite measure of all included indicators and defined as the proportion of individuals dispensed a relevant medication with at least one unmonitored dispensing, over the 5-year study period. A dispensing was considered unmonitored if there was no record of the relevant laboratory test for that medicine within the specified period prior to dispensing, for example, if methotrexate was dispensed on a certain date and there was no full blood count recorded in the patient record for the 84 days prior to that date. Monitoring indicators were derived using HSE-PCRS dispensing data and laboratory monitoring data collected from the patient’s GP electronic record for the study period 2010–2015. Each prescription was grouped by claim number, enabling the identification of medicines co-prescribed within the same prescription. The dataset also included a dispensing date for each claim, and this was used as the index date to calculate the time interval in days since the relevant laboratory test, thus enabling the identification of prescriptions for relevant medicines that were dispensed outside the defined monitoring interval.
Secondary objectives
To examine patient characteristics association with sub-optimal laboratory monitoring, the outcome of interest was a count of the number of unmonitored dispensings per person per year. Explanatory variables were chosen based on their potential association with suboptimal monitoring, as identified in the literature and clinical practice. These included age, sex, deprivation, social class, educational attainment, number of repeat medicines, multimorbidity, functional ability and anxiety/depression. To investigate whether sub-optimal monitoring predicted adverse patient outcomes, we examined three binary variables: ADRs (at least one or none), unplanned hospital admissions, unplanned emergency department visits and mortality. The explanatory variable was the number of unmonitored dispensings per person for the baseline period 2011–2012 (allowing for a 1 year run in period) (see online supplemental appendix 4 for a detailed description of all variables including how they were collected and measured).
Analysis
Stata V.17 (Stata Corporation, College Station, Texas, USA) was used for all analyses.
The study population was characterised using baseline descriptive statistics. The prevalence of sub-optimal lab monitoring was explored at both the dispensing and patient level. First, the number of dispensings outside the monitoring interval for relevant indicators (numerator) was calculated per patient per year. At the patient level, for each indicator, the number of patients per year with at least one dispensing falling outside the recommended monitoring interval was counted. At the dispensing level, the total number of unmonitored dispensings per patient per year for each indicator was explored. As unobserved monitoring may have occurred immediately before the start of the baseline period, a grace period equal to the monitoring interval for each indicator was applied so analysis for each indicator only included dispensings occurring after this period. For each indicator, the median time since monitoring with inter-quartile range was presented. For the primary outcome, a composite measure of laboratory monitoring was derived whereby a patient/dispensing was considered unmonitored if any of the 14 included monitoring criteria were present.
For the secondary objective, (i) assessing the patient characteristics associated with sub-optimal monitoring, Poisson regression was used with the outcome variable a count of the number of unmonitored dispensings per person per year (and the explanatory variables listed above included). For the secondary objective, (ii) assessing the outcomes of sub-optimal monitoring, logistic regression was used.
The baseline number of medicines, age and sex were included in all models, and standard errors were adjusted for within GP practice clustering and within person clustering (as observations for multiple years within a single patient were used) using a clustered sandwich estimator (using the vce2way (cluster cluster) option in Stata). In accordance with STROBE guidance, adjusted and unadjusted estimates for all models were presented.
Results
Study population
A total of 625 (78.8%) participants had at least one dispensing of a relevant higher-risk drug during the 5-year study period. Participants who were not dispensed any of the included higher-risk drugs during the study period were significantly younger and took less medicines. They also had lower deprivation scores, were more likely to have completed secondary education and were more likely to have private health insurance (see online supplemental appendix 5 for baseline comparisons of these groups).
Baseline descriptive statistics for the 625 participants who were dispensed at least one of the included higher-risk drugs are presented in table 1. The average age was 77.7 years, 53% were female, and the mean number of drugs was 7.3 (SD 3.3). Monitored participants were significantly more likely to have had at least one previous unplanned hospital admission between 2010 and 2012 (table 1).
Table 1. Descriptive characteristics at baseline for those fully monitored according to the included indicators and those with at least one unmonitored dispensing during the 5-year period.
| Characteristic | Total,n=625 | Monitored,n=126 | Sub-optimal monitoring,n=499 | |
| Mean (SD) | Mean (SD) | Mean (SD) | Pvalue | |
| Age, years | 77.7 (5.4) | 77.2 (5.6) | 77.9 (5.3) | 0.24 |
| Deprivation, patient | 1.5 (2.5) | 1.3 (2.3) | 1.6 (2.5) | 0.23 |
| Number of drugs | 7.3 (3.3) | 7.0 (3.3) | 7.4 (3.3) | 0.28 |
| N (%) | N (%) | N (%) | P value | |
| Sex | 0.32 | |||
| Female | 293 (53.1) | 62 (49.2) | 270 (54.1) | |
| Male | 332 (46.9) | 64 (50.8) | 229 (45.9) | |
| Health insurance | 0.31 | |||
| No | 362 (57.9) | 68 (54.0) | 294 (58.9) | |
| Yes | 263 (42.1) | 58 (46.0) | 205 (41.1) | |
| Marital status | 0.78 | |||
| Married | 282 (45.1) | 63 (50.0) | 219 (43.9) | |
| Separated/divorced | 37 (5.9) | 7 (5.6) | 30 (6.0) | |
| Widowed | 194 (31.0) | 36 (28.6) | 158 (31.7) | |
| Never married/single | 111 (17.8) | 20 (15.7) | 91 (18.2) | |
| Missing | 1 (0.2) | 0 (0) | 1 (0.2) | |
| Education | 0.32 | |||
| Basic | 398 (63.7) | 73 (57.9) | 325 (65.1) | |
| Completed secondary | 223 (35.7) | 82 (41.3) | 171 (34.3) | |
| Missing | 4 (0.6) | 1 (0.8) | 3 (0.6) | |
| CCI | 0.50 | |||
| No comorbidities | 294 (47.0) | 54 (42.9) | 240 (48.1) | |
| At least one comorbidity | 330 (52.8) | 72 (57.1) | 258 (51.7) | |
| Missing | 1 (0.2) | 0 (0) | 1 (0.2) | |
| HADS | 0.92 | |||
| Normal score | 569 (91.0) | 115 (91.3) | 454 (91.0) | |
| Screened positive | 56 (9.0) | 11 (8.7) | 45 (9.0) | |
| Previous unplanned hospital admission | 0.03 | |||
| No | 419 (67.0) | 72 (57.1) | 347 (69.5) | |
| Yes | 199 (31.8) | 52 (41.3) | 147 (29.5) | |
| Missing | 7 (1.1) | 2 (1.6) | 5 (1.0) | |
| Occupation | 0.18 | |||
| Unskilled | 236 (37.8) | 41 (32.5) | 194 (39.1) | |
| Skilled | 389 (62.2) | 85 (67.5) | 304 (60.9) | |
CCI, Charlson Comorbidity Index; HADS, Hospital Anxiety and Depression Scale
There was wide variation in the numbers of laboratory tests over the 5 years of manually collected laboratory data from the GP electronic health record. For the entire cohort (n=793), there was a median of four renal blood tests per person (IQR 0–8), rising to a median of five (IQR 0–9) in those who were prescribed at least one of the included higher-risk drugs. Given the large variation in laboratory testing and low monitoring prevalence, a separate analysis was run including the entire cohort, exploring the predictors of having blood monitoring and the practice level variation in laboratory testing (see online supplemental appendix 6).
Prevalence of sub-optimal monitoring
During the 5-year study period, there were 45 064 instances of the included higher-risk drugs dispensed, and 11 161 (24.8%) were unmonitored as per the criteria applied. Of the study participants who were prescribed ≥1 higher-risk drug at least once during the 5-year period (n=625), 499 (79.8%) had ≥1 unmonitored dispensing, with a median of 11 per person (IQR 1–27). Of those with sub-optimal monitoring, there was a median of six unmonitored monthly dispensings per year (IQR 2–11). Table 2 lists the prevalence of monitoring for each of the 14 included indicators at the participant level for the 5-year study period, online supplemental appendix 7 gives the annual rates and online supplemental appendix 8 shows the median number of days from dispensing to monitoring at the medication level.
Table 2. Prevalence of sub-optimal monitoring for each of the 14 included indicators over the 5-year study period.
| Criteria | Denominator | N* (%) |
| Patients prescribed a loop diuretic and no record of renal function and electrolytes checked in the last 15 months | 266 | 117 (44.0) |
| Patients prescribed either an ACEI or ARB and no record of renal function and electrolytes checked in the last 15 months | 474 | 295 (62.2) |
| Patient treated with a potassium sparing diuretic and no record of U&Es check in the last 48 weeks | 124 | 74 (59.7) |
| Patient treated with a loop and a thiazide diuretic or metolazone and no record of U&Es check in the last 24 weeks | 13 | 6 (46.2) |
| Patient treated with a potassium sparing diuretic and an ACEI or ARB and no record of U&Es check in the last 48 weeks | 55 | 26 (47.3) |
| Patient treated with an ACE and ARB and no record of U&Es check in the last 24 weeks | 19 | 16 (84.2) |
| Patient treated with a potassium wasting diuretic and no record of U&Es check in the last 48 weeks | 340 | 208 (61.2) |
| Patient treated with a potassium wasting diuretic and digoxin and no record of U&Es check in the last 24 weeks | 29 | 19 (63.3) |
| Patient prescribed amiodarone and had no record of thyroid function test in the last 9 months | 26 | 21 (80.8) |
| Patient treated with azathioprine or methotrexate and no record of FBC check in the last 12 weeks | 13 | 12 (92.3) |
| Patients treated with sulfasalazine and no record of FBC check in the last 24 weeks | 6 | 4 (66.7) |
| Patients prescribed a DOAC and no record of renal function and electrolytes checked in the last 12 months | 65 | 33 (50.8) |
| Patients prescribed levothyroxine and no record of thyroid function test in the last 12 months | 131 | 103 (78.6) |
| Patients prescribed carbimazole and no record of thyroid function test in the last 3 months | 12 | 10 (83.3) |
Number of participants with at least one unmonitored dispensing over the five5-year study period.
ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin receptor blockers; DOAC, direct oral anticoagulant; FBC, full blood count; PINCER, pharmacist-led information technology intervention for medication errors; U&E, urea and electrolytes
Secondary objectives
Patient characteristics associated with sub-optimal laboratory monitoring
Unadjusted associations with sub-optimal monitoring were observed for the number of drugs, deprivation and anxiety/depression symptoms that all persisted in the adjusted model, although the effect size for number of drugs and deprivation was small. The largest effect seen was for anxiety/depression symptoms (incidence rate ratio (IRR) 1.33, 95% CI 1.05 to 1.68) (table 3). Given the large variation in the number of unmonitored dispensings per person, a subgroup analysis was run including those with high levels of sub-optimal monitoring (defined as ≥3 unmonitored dispensings per year). For this group, there was a small borderline significance for the number of drugs (IRR 1.02, 95% CI 1.00 to 1.03) and the largest effect seen for anxiety/depression symptoms (IRR 1.20, 95% CI 1.08 to 1.34) (online supplemental appendix 9).
Table 3. Unadjusted and adjusted Poisson regression examining the effect of patient characteristics on the number of unmonitored dispensings per person per year over 5-year study period for participants prescribed at least one denominator drug that year.
| Characteristic | Unadjusted | Adjusted | ||
| IRR (95% CI) | P value | IRR (95% CI) | P value | |
| Age | 1.00 (0.98 to 1.01) | 0.63 | 0.99 (0.97 to 1.01) | 0.29 |
| Gender | 1.07 (0.87 to 1.33) | 0.56 | 1.04 (0.80 to 1.36) | 0.76 |
| Drugs year | 1.02 (1.01 to 1.04) | 0.001 | 1.02 (1.00 to 1.04) | 0.02 |
| CCI | 0.98 (0.86 to 1.12) | 0.75 | 0.92 (0.79 to 1.08) | 0.31 |
| Deprivation score | 1.04 (1.00 to 1.07) | 0.04 | 1.04 (1.01 to 1.07) | 0.02 |
| Usual activity EQ5D | 1.23 (1.00 to 1.52) | 0.06 | 1.14 (0.89 to 1.46) | 0.31 |
| Social class | 0.93 (0.78 to 1.11) | 0.43 | 0.99 (0.83 to 1.19) | 0.93 |
| Education | 0.93 (0.77 to 1.12) | 0.46 | 1.02 (0.88 to 1.20) | 0.77 |
| HADS | 1.46 (1.16 to 1.84) | 0.001 | 1.33 (1.05 to 1.68) | 0.02 |
CCI, Charlson Comorbidity Index (dichotomised to either no co-morbidity or at least one); EQ5D, European Quality of Life Five Dimension (dichotomised to either no problems or at least some problems); HADS, Hospital Anxiety and Depression Scale (dichotomised to screened positive or not)IRR, incidence rate ratio
Sub-optimal monitoring and adverse health outcomes (adverse drug reactions (ADRs), emergency admissions and mortality) over time
Between 2010 and 2015, 116 of the 625 participants (18.6%) died, 143 (22.9%) had at least one ADR, 122 (19.5%) had at least one unplanned hospital admission and 94 (15.0%) had at least one emergency department visit. There was no evidence of an association identified between sub-optimal monitoring in 2011 and 2012 (measured as the number of unmonitored dispensings per person) and subsequent unplanned hospital admissions, mortality or ADRs. There was evidence of an association between baseline sub-optimal monitoring and emergency department visits at follow-up, although the effect size was small (table 4).
Table 4. Logistic regression examining whether sub-optimal monitoring is associated with adverse health outcomes (ADRs, emergency admissions and mortality) over time.
| Outcome | Unadjusted OR (95% CI) | P value | Adjusted OR (95% CI) | P value |
| ADR over 5 years follow-up | 1.02 (0.99 to 1.03) | 0.35 | 1.02 (0.99 to 1.02) | 0.63 |
| Mortality | 1.02 (1.00 to 1.04) | 0.09 | 1.02 (0.99 to 1.04) | 0.13 |
| Unplanned hospital admission | 0.99 (0.97 to 1.02) | 0.60 | 0.99 (0.97 to 1.01) | 0.66 |
| Emergency department visit | 1.02 (1.00 to 1.04) | 0.03 | 1.02 (1.00 to 1.04) | 0.05 |
ADRsadverse drug reactions
Discussion
Summary of results
The prevalence of sub-optimal monitoring was high in this population of community-dwelling older adults in Irish primary care. Almost 80% of individuals prescribed one of the included higher-risk drugs had ≥1 unmonitored monthly dispensing over the 5-year study period, and almost 24% of all dispensings of included higher-risk drugs were unmonitored. The generalisability of this finding may be limited given the unique context of Irish primary care during the study period, where the management of chronic disease in the community was unfunded and unstructured. Participants prescribed one of the higher-risk drugs of interest were significantly older, took more medicines and experienced higher levels of socioeconomic deprivation. This is unsurprising considering the strong link between socioeconomic deprivation and multimorbidity.1
Patient characteristics associated with sub-optimal monitoring included the number of drugs, socioeconomic deprivation and anxiety/depression symptoms, with the largest effect seen for anxiety/depression symptoms indicating that these patients are at potentially at higher risk for preventable drug related harm.
Our study identified that baseline sub-optimal monitoring was associated with significant increased odds of having an emergency department visit at follow-up. However, the effect size was small, there was no evidence of an association with unplanned hospital admission and the findings are also subject to the influence of potential unmeasured confounders. It is also important to bear in mind that there are inconsistencies across guidelines on optimal monitoring intervals and there is growing recognition that many of these recommendations may be low value or yield and add to an often high level of disease and treatment burden for patients with multimorbidity.20 21
Strengths and limitations
This study adds to the limited literature in the area of laboratory monitoring of higher risk medicines in primary care. A major strength of this study is the inclusion of patient reported outcome measures, and we identified anxiety/depression as a significant predictor of sub-optimal monitoring of higher-risk medicines. Due to the wide variation in the levels of laboratory monitoring, various sensitivity analyses were conducted to further explore the results, and this is another strength of this work. A sub-group analysis was conducted examining the predictors of sub-optimal monitoring limited to participants with high levels of sub-optimal monitoring, and a second analysis was conducted where the number of unmonitored dispensings was truncated due to the long tailed distribution of the variable. Similar results were returned for both models indicating that are results are not being driven by either extreme ends of the spectrum.
The main limitation is the context in which the study was conducted, which has changed over time. There has been markedly increased uptake in the use of electronic communications between primary and secondary care over the past 10 years; however, in 2010, most general practices in Ireland received laboratory results in paper format and scanned to the patient’s electronic record. It is possible that delays and errors in this process affected the quality of data and over-estimated the prevalence of sub-optimal monitoring. Recent improvements in both healthcare technology and the delivery of chronic disease management programmes may mean that the laboratory monitoring practices observed during the study period may not reflect the current landscape. In addition, for this study, we assessed the number of days between dispensing and the relevant laboratory monitoring. We did not assess the outcome of laboratory monitoring. A recent study based in UK primary care indicated that only a quarter of individuals having laboratory blood tests had completely normal results, yet only 48% of tests resulted in any change in management.22 Finally, older people with multiple chronic illness often attend multiple hospital outpatient clinics. It is likely many of these patients had either some or all of their monitoring in secondary care, and although we did include secondary care results where available in the GP electronic health record it is likely that this was not fully captured.
Comparison with existing literature
A systematic review of the incidence and characteristics of preventable adverse drug events in ambulatory care identified inadequate monitoring (45.4%; range 22.2–69.8%) as the most frequent error of omission that resulted in hospital admission.23 The prevalence of sub-optimal monitoring was higher in our study, compared with similar studies in the USA and UK.324,26 However, these studies included the entire primary care population, and our study included older adults aged ≥70 years. Notwithstanding the different population under investigation, the difference in the prevalence of monitoring was substantial. For example, a recent analysis assessing the implementation of the PINCER indicators in the UK reported that 5% of patients prescribed loop diuretics or angiotensin-converting enzyme inhibitor (ACEI)/angiotensin receptor blockers (ARBs) had sub-optimal monitoring compared 34% in this study.24 Despite the lower prevalence of sub-optimal monitoring, the relative prevalence of monitoring across included indicators is similar to other studies, with lower adherence for amiodarone and methotrexate monitoring.24 Similar factors associated with sub-optimal monitoring such as deprivation and increased number of drugs have been found in other studies.3 26 Anxiety/depression was identified as a significant predictor of sub-optimal monitoring, and no other study including patient-reported outcome measures could be identified for comparison. There is no clear evidence of the impact of sub-optimal monitoring on patient outcomes. In this study, there was no evidence of an association between sub-optimal monitoring and unplanned hospital admissions, but there was a small but significantly increased odds of emergency department visits. The reverse relationship with healthcare usage has been also been reported, whereby higher rates of monitoring were associated with a small but significantly increased rate of unplanned hospital admissions.26 In our population, monitored participants had a significantly higher proportion of unplanned hospital admissions at baseline, compared with participants who were fully monitored. These results reflect the multiple factors at play, whereby patients who have had a recent hospital admission may be more likely to have laboratory blood tests after discharge.
Implications for practice and research
There is evidence that well-developed primary care systems are associated with improved outcomes for patients and a more equitable distribution of healthcare.27,29 As recently as 2018, Ireland was far below European averages with respect to the proportion of the health budget spent on primary care (<5%),30 and this study was set in that context. There have been recent improvements, notably with the introduction of a funded structured Chronic Disease Management programme, where GPs are reimbursed to provide two structured patient visits per year including blood monitoring for specific chronic medical conditions. The results from this study will be an important benchmark for future changes.
Although there were some limitations with respect to our estimates of sub-optimal monitoring in this population, there was significant variation in the prevalence of laboratory testing at both the individual and practice level (online supplemental appendix 2). A comparative analysis of laboratory requesting patterns across 22 general practices in a single district in the UK concluded that the large differences observed in laboratory testing probably result from individual variation in clinical practice.31 In our analysis which included people aged ≥70 years and prescribed ≥1 medicine, we found that increasing age and reduced functional ability were both associated with less frequent renal blood testing. These results may indicate some unmet need in this population.
Our study attempted to quantify the prevalence of necessary testing with respect to monitoring of higher-risk prescriptions, but consideration also needs to be given to unnecessary testing and the impact this may have on resource allocation. Attempting to define unnecessary testing is complex however, with qualitative work suggesting that laboratory testing is sometimes used as a way to be seen to be doing something in the context of patient expectations and as a means for managing diagnostic uncertainty.32 A retrospective cohort study based in the USA that explored suboptimal monitoring of ACEI and ARBs found that higher-risk individuals, such as those with increasing age, multimorbidity and additional risk factors, were more likely to receive appropriate monitoring.33 This finding is particularly interesting given that monitoring intervals are often based on expert consensus rather than individual risk. It raises questions about the potential for over-monitoring in lower-risk individuals, which could represent an inefficient use of healthcare resources and increase the likelihood of detecting spurious results. Over the past 10 years, various campaigns such as the BMJ’s Too Much Medicine and American Board of Internal Medicine’s Choosing Widely Campaign have highlighted the risks of over-testing both at an individual level and due to the opportunity cost of diverting limited healthcare resources.34 35 This is particularly relevant in primary care where there are increasing concerns about managing demand and workload including increased blood testing.21 These findings highlight the complexity of laboratory monitoring, especially when balancing the need for safety with the appropriate allocation of resources. Clinicians and researchers should work to develop systems and protocols for laboratory testing to minimise unnecessary testing but capture those most likely to benefit. With respect to monitoring of higher-risk prescriptions, using computerised clinical decision support inbuilt into practice software is one potential approach.
Uncertainties around responsibility for monitoring for higher risk medicines are likely to contribute to risk and the environmental context, including the organisation of care, and have been identified as important components to managing prescribing.36 Although it is possible that many of these patients had monitoring in secondary care, this was not clear from the patient record, and this finding strengthens the argument for a shared care record to increase visibility of completed monitoring across primary and secondary care.
This analysis identified an association between baseline anxiety/depression symptoms and sub-optimal laboratory monitoring. It has been estimated that mental illness almost doubles the risk of any preventable harm in patients presenting to either primary care or the emergency department.37 It is thus important to screen for comorbid mental illness in older patients, particularly those with multimorbidity to capture this particularly vulnerable population.38
A final consideration is that the monitoring indicators included in this study were developed by expert consensus, and there may be differing opinions as to what constitutes an appropriate monitoring interval and whether a prolonged interval is likely to have significant clinical consequences.39 In this study, sub-optimal monitoring was associated with a small but significant increased odds of an emergency department visit, but there was no evidence of an effect on ADRs, mortality or unplanned admissions, and we may have been underpowered to see an effect. More research is needed to examine the optimal timing of laboratory monitoring for higher-risk medicines and the impact of sub-optimal monitoring of higher risk medicines prescribed in primary care on important clinical endpoints such as unplanned hospital admissions and ADRs. With respect to implementation, future approaches to tackle sub-optimal monitoring may include exploring the effect of shared care electronic health records and technology enabled clinical decision support systems.
In conclusion, the prevalence of sub-optimal monitoring in this population of community-dwelling older adults was high over a 5-year period. This was in the context of a primary care system without a funded chronic disease management programme. There was an association between number of drugs, deprivation, anxiety/depression and sub-optimal monitoring. Adequate resourcing and organisation of chronic disease management in primary care, as well as limiting low-value or unnecessary testing, is vital to ensure those in most need receive appropriate care. These findings may prove an important benchmark for assessing current and future laboratory monitoring requirements for higher-risk medicines in primary care.
supplementary material
Footnotes
Funding: This work was supported by the Health Research Board (HRB) of Ireland through the HRB Centre for Primary Care Research (grant number: HRC/2007/1) and the HRB Emerging Clinician Scientist Award (grant number: ECSA/2020/002). The funders had no role in study design, data collection and analysis, manuscript preparation, or the decision to submit for publication.
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2024-086446).
Provenance and peer review: Not commissioned; externally peer-reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants and was approved by the RCSI University of Medicine and Health Science’s Human Research Ethics Committee (REC 202303016). Participants gave informed consent to participate in the study before taking part.
Data availability free text: The dataset analysed during the current study is not publicly available in accordance with the consent provided by participants. The participants of this study did not give written consent for their data to be shared publicly. Data may be made available following a reasonable request and ethical approval.
Patient and public involvement: Patients and/or the public were involved in the design, conduct, reporting or dissemination plans of this research. Refer to the Methods section for further details.
Data availability statement
Data are available upon reasonable request.
References
- 1.Fortin M, Stewart M, Poitras M-E, et al. A systematic review of prevalence studies on multimorbidity: toward a more uniform methodology. Ann Fam Med. 2012;10:142–51. doi: 10.1370/afm.1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mangoni AA, Jackson SHD. Age-related changes in pharmacokinetics and pharmacodynamics: basic principles and practical applications. Br J Clin Pharmacol. 2004;57:6–14. doi: 10.1046/j.1365-2125.2003.02007.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Avery AJ, Ghaleb M, Barber N, et al. The prevalence and nature of prescribing and monitoring errors in English general practice: a retrospective case note review. Br J Gen Pract . 2013;63:e543–53. doi: 10.3399/bjgp13X670679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Howard RL, Avery AJ, Slavenburg S, et al. Which drugs cause preventable admissions to hospital? A systematic review. Br J Clin Pharmacol. 2007;63:136–47. doi: 10.1111/j.1365-2125.2006.02698.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.O’Mahony D, Cherubini A, Guiteras AR, et al. STOPP/START criteria for potentially inappropriate prescribing in older people: version 3. Eur Geriatr Med. 2023;14:625–32. doi: 10.1007/s41999-023-00777-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wallace E, McDowell R, Bennett K, et al. Impact of Potentially Inappropriate Prescribing on Adverse Drug Events, Health Related Quality of Life and Emergency Hospital Attendance in Older People Attending General Practice: A Prospective Cohort Study. J Gerontol A Biol Sci Med Sci. 2017;72:271–7. doi: 10.1093/gerona/glw140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cahir C, Moriarty F, Teljeur C, et al. Potentially inappropriate prescribing and vulnerability and hospitalization in older community-dwelling patients. Ann Pharmacother. 2014;48:1546–54. doi: 10.1177/1060028014552821. [DOI] [PubMed] [Google Scholar]
- 8.Cahir C, Bennett K, Teljeur C, et al. Potentially inappropriate prescribing and adverse health outcomes in community dwelling older patients. Br J Clin Pharmacol. 2014;77:201–10. doi: 10.1111/bcp.12161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Moriarty F, Bennett K, Kenny RA, et al. Comparing Potentially Inappropriate Prescribing Tools and Their Association With Patient Outcomes. J Am Geriatr Soc . 2020;68:526–34. doi: 10.1111/jgs.16239. [DOI] [PubMed] [Google Scholar]
- 10.Dreischulte T, Grant AM, McCowan C, et al. Quality and safety of medication use in primary care: consensus validation of a new set of explicit medication assessment criteria and prioritisation of topics for improvement. BMC Clin Pharmacol. 2012;12:5. doi: 10.1186/1472-6904-12-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Avery AJ, Dex GM, Mulvaney C, et al. Development of prescribing-safety indicators for GPs using the RAND Appropriateness Method. Br J Gen Pract. 2011;61:e526–36. doi: 10.3399/bjgp11X588501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dimitrow MS, Airaksinen MSA, Kivelä S-L, et al. Comparison of prescribing criteria to evaluate the appropriateness of drug treatment in individuals aged 65 and older: a systematic review. J Am Geriatr Soc. 2011;59:1521–30. doi: 10.1111/j.1532-5415.2011.03497.x. [DOI] [PubMed] [Google Scholar]
- 13.Steinman MA, Handler SM, Gurwitz JH, et al. Beyond the prescription: medication monitoring and adverse drug events in older adults. J Am Geriatr Soc. 2011;59:1513–20. doi: 10.1111/j.1532-5415.2011.03500.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.O’Sullivan JW, Stevens S, Hobbs FDR, et al. Temporal trends in use of tests in UK primary care, 2000-15: retrospective analysis of 250 million tests. BMJ. 2018;363:k4666. doi: 10.1136/bmj.k4666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jones T, Patel R, Elwenspoek MMC, et al. Variation in laboratory testing for patients with long-term conditions: a longitudinal cohort study in UK primary care. BJGP Open. 2023;7:BJGPO.2022.0139. doi: 10.3399/BJGPO.2022.0139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. PLoS Med. 2007;4:e296. doi: 10.1371/journal.pmed.0040296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.First report of the structured chronic disease management programme in general practice: health service executive. 2022. [25-Jul-2023]. https://www.hse.ie/eng/services/publications/first-report-of-the-structured-chronic-disease-management-programme-in-general-practice.pdf Available. Accessed.
- 18.Wallace E, McDowell R, Bennett K, et al. External validation of the Probability of repeated admission (Pra) risk prediction tool in older community-dwelling people attending general practice: a prospective cohort study. BMJ Open. 2016;6:e012336. doi: 10.1136/bmjopen-2016-012336. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.NHS specialist pharmacy service. NHS. 2023. [05-Jul-2023]. https://www.sps.nhs.uk/home/tools/drug-monitoring/ Available. Accessed.
- 20.Buffel du Vaure C, Ravaud P, Baron G, et al. Potential workload in applying clinical practice guidelines for patients with chronic conditions and multimorbidity: a systematic analysis. BMJ Open. 2016;6:e010119. doi: 10.1136/bmjopen-2015-010119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Johansson M, Guyatt G, Montori V. Guidelines should consider clinicians’ time needed to treat. BMJ. 2023;380:e072953. doi: 10.1136/bmj-2022-072953. [DOI] [PubMed] [Google Scholar]
- 22.Watson J, Burrell A, Duncan P, et al. Exploration of reasons for primary care testing (the Why Test study): a UK-wide audit using the Primary care Academic CollaboraTive. Br J Gen Pract. 2024;74:e133–40. doi: 10.3399/BJGP.2023.0191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Thomsen LA, Winterstein AG, Søndergaard B, et al. Systematic review of the incidence and characteristics of preventable adverse drug events in ambulatory care. Ann Pharmacother . 2007;41:1411–26. doi: 10.1345/aph.1H658. [DOI] [PubMed] [Google Scholar]
- 24.Fisher L, Hopcroft LE, Rodgers S, et al. Changes in medication safety indicators in England throughout the covid-19 pandemic using OpenSAFELY: population based, retrospective cohort study of 57 million patients using federated analytics. bmjmed . 2023;2:e000392. doi: 10.1136/bmjmed-2022-000392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wessell AM, Litvin C, Jenkins RG, et al. Medication prescribing and monitoring errors in primary care: a report from the Practice Partner Research Network. Qual Saf Health Care . 2010;19:e21. doi: 10.1136/qshc.2009.034678. [DOI] [PubMed] [Google Scholar]
- 26.McDowell SE, Coleman JJ, Evans SJW, et al. Laboratory monitoring and adverse patient outcomes with antihypertensive therapy in primary care. Pharmacoepidemiol Drug Saf . 2010;19:482–9. doi: 10.1002/pds.1935. [DOI] [PubMed] [Google Scholar]
- 27.Kringos DS, Boerma WGW, Bourgueil Y, et al. The European primary care monitor: structure, process and outcome indicators. BMC Fam Pract. 2010;11:81. doi: 10.1186/1471-2296-11-81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Starfield B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milbank Q. 2005;83:457–502. doi: 10.1111/j.1468-0009.2005.00409.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kringos DS, Boerma W, van der Zee J, et al. Europe’s strong primary care systems are linked to better population health but also to higher health spending. Health Aff (Millwood) 2013;32:686–94. doi: 10.1377/hlthaff.2012.1242. [DOI] [PubMed] [Google Scholar]
- 30.System of Health Accounts 2018. Dublin: Central Statistics Office; 2018. [Google Scholar]
- 31.Smellie WSA, Galloway MJ, Chinn D, et al. Is clinical practice variability the major reason for differences in pathology requesting patterns in general practice? J Clin Pathol. 2002;55:312–4. doi: 10.1136/jcp.55.4.312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Watson J, de Salis I, Banks J, et al. What do tests do for doctors? A qualitative study of blood testing in UK primary care. Fam Pract. 2017;34:735–9. doi: 10.1093/fampra/cmx051. [DOI] [PubMed] [Google Scholar]
- 33.Harrison TN, Reynolds K, Hahn EE, et al. Laboratory monitoring to reduce adverse drug-related events: a mixed methods study. JMCP . 2022;28:16–25. doi: 10.18553/jmcp.2022.28.1.16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Moynihan R, Smith R. Too much medicine? BMJ. 2002;324:859–60. doi: 10.1136/bmj.324.7342.859. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Levinson W, Kallewaard M, Bhatia RS, et al. ‘Choosing Wisely’: a growing international campaign. BMJ Qual Saf . 2015;24:167–74. doi: 10.1136/bmjqs-2014-003821. [DOI] [PubMed] [Google Scholar]
- 36.Riordan DO, Byrne S, Fleming A, et al. GPs’ perspectives on prescribing for older people in primary care: a qualitative study. Br J Clin Pharmacol. 2017;83:1521–31. doi: 10.1111/bcp.13233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Fernholm R, Holzmann MJ, Wachtler C, et al. Patient-related factors associated with an increased risk of being a reported case of preventable harm in first-line health care: a case-control study. BMC Fam Pract. 2020;21:20. doi: 10.1186/s12875-020-1087-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.National Guideline Centre (UK) NICE Guideline. NICE; 2016. Multimorbidity: assessment, prioritisation and management of care for people with commonly occurring multimorbidity. [PubMed] [Google Scholar]
- 39.Spinewine A, Schmader KE, Barber N, et al. Appropriate prescribing in elderly people: how well can it be measured and optimised? Lancet. 2007;370:173–84. doi: 10.1016/S0140-6736(07)61091-5. [DOI] [PubMed] [Google Scholar]