Can Dialysis Sufferers Be Precisely Recognized Utilizing Healthcare Claims Information?
Perit Dial Int. 2014 Sep-Oct; 34(6): 643–651.
♦ Background: Whereas medical insurance claims information are sometimes used to estimate the prices of renal alternative remedy in sufferers with end-stage renal illness (ESRD), the accuracy of strategies used to establish sufferers receiving dialysis — particularly peritoneal dialysis (PD) and hemodialysis (HD) — in these information is unknown.
♦ Strategies: The examine inhabitants consisted of all individuals aged 18 – 63 years in a big US built-in well being plan with ESRD and dialysis-related billing codes (i.e., prognosis, procedures) on healthcare encounters between January 1, 2005, and December 31, 2008. Utilizing billing codes for all healthcare encounters inside 30 days of every affected person’s first dialysis-related declare (“index encounter”), we tried to designate every examine topic as both a “PD affected person” or “HD affected person.” Utilizing various home windows of ± 30 days, ± 90 days, and ± 180 days across the index encounter, we reviewed sufferers’ medical data to find out the dialysis modality really obtained. We calculated the optimistic predictive worth (PPV) for every dialysis-related billing code, utilizing info in sufferers’ medical data because the “gold commonplace.”
♦ Outcomes: We recognized a complete of 233 sufferers with proof of ESRD and receipt of dialysis in healthcare claims information. Primarily based on examination of billing codes, 43 and 173 examine topics had been designated PD sufferers and HD sufferers, respectively (14 sufferers had proof of PD and HD, and modality couldn’t be ascertained for 31 sufferers). The PPV of codes used to establish PD sufferers was low primarily based on a ± 30-day medical file overview window (34.9%), and elevated with use of ± 90-day and ± 180-day home windows (each 67.4%). The PPV for codes used to establish HD sufferers was uniformly excessive — 86.7% primarily based on ± 30-day overview, 90.8% primarily based on ± 90-day overview, and 93.1% primarily based on ± 180-day overview.
♦ Conclusions: Whereas HD sufferers might be precisely recognized utilizing billing codes in healthcare claims information, case identification was far more problematic for sufferers receiving PD.
Key phrases: Peritoneal dialysis, hemodialysis, insurance coverage declare overview, claims evaluation, medical data, methodology, epidemiologic strategies, retrospective examine
A lot of what we find out about variations in scientific outcomes and prices of care between sufferers receiving hemodialysis (HD) versus peritoneal dialysis (PD) comes from analyses of Medicare claims information, since individuals with end-stage renal illness (ESRD) are eligible for medical insurance protection below this federal program after the primary three months of dialysis. Complete annual healthcare expenditures have been reported to be considerably decrease amongst Medicare enrollees receiving PD as compared with these receiving HD (1,2).
In most research thus far, identification of sufferers who obtained PD versus HD has been primarily based on varied prognosis and/or process codes used for third-party billing and reimbursement (2-6). To the perfect of our data, nonetheless, the predictive accuracy of those strategies for figuring out PD and HD sufferers in medical insurance claims information is unknown. The likelihood exists, subsequently, that the dialysis modality that some sufferers really obtained was incorrectly designated on claims. To make clear this problem, we examined the predictive accuracy of assorted billing codes as a way of figuring out sufferers who obtained PD versus HD, utilizing info gleaned from digital medical data (EMRs) as our “gold commonplace.”
This retrospective examine was performed at Henry Ford Well being System (HFHS), a complete well being system that gives medical care to roughly 800,000 residents of Detroit, Michigan, and the encompassing areas, with 3.2 million affected person contacts yearly. Well being Alliance Plan (HAP) is a completely owned, not-for-profit well being upkeep group inside HFHS that gives insurance coverage for 475,000 individuals, 125,000 of whom have elected HFHS amenities as their assigned website of care. Research topics had been drawn from the inhabitants of individuals enrolled in HAP with HFHS task, practically 20% of whom are aged 65 years or older.
HFHS makes use of a complete multi-dimensional EMR system that gives clinicians and researchers with real-time entry to computerized medical data (“CarePlus”). CarePlus maintains info on affected person demographics, ambulatory care visits, scientific laboratory and radiology outcomes, and inpatient admissions, in addition to varied different scientific and financial measures. HFHS additionally maintains a big administrative information warehouse containing info on all encounters with HFHS suppliers and amenities, together with ambulatory care visits (outpatient clinic, emergency division), hospital admissions, healthcare companies offered at non-HFHS websites, billing data generated inside inpatient and outpatient settings, and outpatient prescription claims. Billing data embody info on kind of service, date of service, supplier identify and specialty, website of service and kind of encounter, and diagnoses.
Info in CarePlus, whereas saved electronically, isn’t searchable and can’t be harvested in digital format; it subsequently was extracted manually onto hard-copy case-report types that had been developed to be used on this examine. To make sure affected person confidentiality and compliance with the Well being Insurance coverage Portability and Accountability Act (HIPAA) of 1996, no patient-identifying info was extracted. Every affected person within the examine pattern was assigned a novel examine identifier, which was used to hyperlink info from totally different sources. The examine was authorized by the HFHS Institutional Overview Board.
The supply inhabitants for the examine consisted of all individuals, aged 18 – 63 years as of January 1, 2005, who had been enrolled in HAP with HFHS task anytime between January 1, 2005, and December 31, 2008 (the “examine interval”). (Within the US, sufferers aged 65 years or older obtain insurance coverage protection by means of the Medicare program; since we didn’t have entry to Medicare information, sufferers of this age weren’t included in our examine pattern. Medicare can also be this system that gives medical insurance to individuals with ESRD aged < 65 years. Nonetheless, as a result of Medicare doesn't grow to be the first payer for ESRD sufferers of this age till roughly 30 months following illness onset, we restricted our consideration to sufferers between the ages of 18 and 63 years as of the date of initiation of dialysis to get rid of issues of incomplete information seize.) Amongst these individuals, we recognized all these with proof of ESRD and dialysis-related healthcare encounters in the course of the examine interval.
Encounters associated to dialysis had been designated principally utilizing prognosis and/or process codes for dialysis-related companies rendered in both outpatient or inpatient settings (a list of all dialysis-related codes is included within the Appendices). The date of the primary dialysis-related encounter was designated the “index encounter,” and individuals with lower than one yr of HAP enrollment following their index encounter had been excluded from the examine pattern. Proof of ESRD was ascertained primarily based on presence of a number of healthcare encounters with Worldwide Classification of Illnesses (ICD)-9-CM prognosis code 585.6 in the course of the yr previous the index encounter and/or the six-month interval subsequent to it (sufferers had been required to have a prognosis of ESRD within the 12-month interval previous the date of the earliest declare for dialysis, and/or the 6-month interval thereafter).
We categorized all dialysis-related billing codes a priori as indicative of both PD or HD (Appendix and ). We then examined billing codes for all sufferers inside 30 days of their index encounter, and designated (as possible) every affected person as receiving PD or HD; a 30-day interval was used for overview of billing codes, for the reason that code for the index encounter typically was not sufficiently descriptive to allow classification.
APPENDIX A1 –
APPENDIX A2 –
Following this designation utilizing claims information solely, skilled medical abstractors reviewed every affected person’s medical file to find out the dialysis modality really obtained, utilizing various home windows of ± 30 days, ± 90 days, and ± 180 days across the index encounter.
Measures and Analyses
We examined the predictive accuracy of healthcare claims for designating sufferers as receiving PD versus HD, utilizing info within the EMR as our “gold commonplace.” Accordingly, sufferers had been deemed “true-positives” if overview of medical data revealed proof of the designated dialysis modality; they had been deemed “false-positives” if the designated dialysis modality couldn’t be confirmed on this style.
We estimated the predictive accuracy of dialysis-related billing codes for PD and HD, respectively, in healthcare claims utilizing optimistic predictive worth (PPV), outlined because the ratio of the whole variety of sufferers who had been “true-positives” to the whole variety of sufferers who had been both “true-positives” or “false-positives”. Since PPV was anticipated to be dependent upon the timeframe employed for medical file overview, we alternatively employed time home windows of ± 30 days, ± 90 days, and ± 180 days round every affected person’s index encounter (). Ninety-five p.c confidence intervals (95% CI) for PPV had been estimated utilizing a traditional approximation of the binomial distribution.
We recognized a complete of 233 ESRD sufferers with proof of dialysis-related encounters in healthcare claims information in the course of the examine interval; 43 and 173 sufferers had been designated as receiving PD and HD, respectively (14 sufferers had proof of each modalities and had been consequently included in each teams). Dialysis modality couldn’t be decided for 31 sufferers (i.e., their billing codes had been nonspecific). Most sufferers designated as receiving PD had healthcare encounters with present procedural terminology (CPT) code 49421 and/or 90945 (). Nearly all sufferers designated as receiving HD had healthcare encounters with CPT codes 36145 or 90935, ICD-9-CM prognosis code V56.0, and/or ICD-9-CM process codes 38.95 and 39.95.
The PPV of billing codes used to establish PD sufferers was low (34.9%) (95% CI: 20.6%, 49.1%) primarily based on a ± 30-day window (across the index date) for medical file overview; it improved to 67.4% (53.4%, 81.4%) when the window for overview was prolonged to both ± 90 days or ± 180 days (). The PPV of billing codes used to establish HD sufferers was uniformly excessive: 86.7% (81.6%, 91.8%) at ± 30 days, 90.8% (86.4%, 95.1%) at ± 90 days, and 93.1% (89.3%, 96.8%) at ± 180 days (). Among the many mostly encountered codes, CPT-4 code 49421 had a low PPV (40.9%) for PD in a ±30-day window, however excessive (95.5%) with both a ± 90-day or ± 180-day window; the corresponding CPT-4 code for HD (36145) had excessive PPVs for HD at ± 30 days (89.3%), ± 90 days (92.9%), and ± 180 days (96.4%). Process codes for dialysis-related encounters had comparatively low PPVs for PD and comparatively excessive PPVs for HD regardless of time window. For instance, CPT-4 code 90945, which was designated a priori as PD-related, had a PPV of 31.8% at ± 30 days, 45.5% at ± 90 days, and 45.5% at ± 180 days; HD-related CPT-4 code 90935 had corresponding PPVs of 95.3%, 96.9%, and 96.9%, respectively. Different often used HD-related codes additionally had excessive (i.e., >90%) PPVs no matter time window employed.
Our examine is the primary, to the perfect of our data, to look at the predictive accuracy in healthcare claims information of billing (i.e., process, prognosis) codes which can be typically used to establish ESRD sufferers who obtain PD and HD. Our findings point out that whereas sufferers who’re receiving HD could be recognized with cheap accuracy, identification of sufferers receiving PD is far more problematic. Our findings elevate potential issues about comparisons of PD and HD sufferers which can be primarily based on analyses of US healthcare claims information, as their findings are solely as sturdy because the strategies used to establish sufferers receiving these dialysis modalities.
There are two principal explanations for our findings. First, whereas probably the most often used codes used to establish HD sufferers had been particular for this modality, these used to establish PD sufferers — CPT-4 codes 49421 and 90945 — weren’t. Though the PPV for the primary of those codes was low (41%) utilizing a 30-day window, it was fairly excessive (96%) with each the 90-day and 180-day home windows, suggesting that usually there could also be a considerable delay between catheter placement and graduation of PD, indicative of the sometimes deliberate nature of this modality (i.e., whereas it may be utilized in emergency conditions (7), there sometimes is a lag between catheter placement and initiation of PD to permit for restoration from surgical procedure and the event of scar tissue that helps maintain the catheter in place). The delay between surgical procedure and dialysis initiation could also be problematic when it comes to evaluating scientific outcomes and prices, because the date of the process code doesn’t approximate the date that PD is definitely initiated. The PPV of the second code (i.e., CPT-4 90945) remained low regardless of the window used for medical file overview, suggesting that it’s generally used for dialysis modalities apart from PD (e.g., hemofiltration).
Second, the whole variety of sufferers with billing codes for PD was a lot decrease than the quantity with such codes for HD (43 vs 173, respectively), and the quantity with dialysis confirmed by means of medical file overview was even decrease (29 vs 161). Whereas each estimates yield a point-prevalence estimate for PD better than the nationwide common (18% and 15% vs ∼7% (1,8)), robustness of findings is compromised with a small pattern dimension. Moreover, as a result of the variety of PD sufferers was comparatively low, we had been unable to evaluate the PPV of all codes which can be obtainable for reimbursement of PD. For instance, there have been no medical encounters with the accompanying PD-related CPT-4 code 90947.
As well as, billing directives from insurance coverage firms aren’t at all times aligned with companies rendered throughout visits. For instance, the PD-related CPT-4 code 90945 had a PPV of solely 46% over a ± 180-day window. It’s potential that this code is used to invoice for predialysis companies, and that some sufferers are readied for PD, however in the end obtain HD as an alternative. In a current examination of 217 US sufferers who initiated dialysis at a single heart, 124 (57%) opted for training on PD; solely 48% of those sufferers (i.e., 59/124) started dialysis with this modality, nonetheless (9).
Limitations of our examine ought to be famous. The extent to which coding practices for dialysis at HFHS replicate these at different well being programs throughout the US is unknown. In actual fact, there may be proof suggesting that the precise administrative codes used to invoice for dialysis-related care range considerably by payer (3). Accordingly, the diploma to which codes generally used to establish dialysis sufferers in our examine setting are utilized in different settings is unknown; equally, the PPV of codes generally utilized in different well being plans however not at HFHS is also unknown. Equally, we restricted our consideration to sufferers aged < 63 years to make sure that we might assemble full chronologies of care. Whereas we now have no purpose to consider that diagnostic and process coding on healthcare claims differs considerably throughout third-party payers and subsequently that our outcomes wouldn't be generalizable to individuals aged 65 years or better, this stays a limitation of our examine. Lastly, our pattern dimension was small (n = 233), notably with respect to the variety of sufferers recognized as receiving PD (n=43), which limits the robustness of our findings. Our outcomes subsequently ought to be confirmed in future analysis.
In conclusion, our findings counsel that ESRD sufferers who obtain HD could be precisely recognized utilizing billing codes in healthcare claims information, however that comparable identification of these receiving PD is extra problematic. When endeavor retrospective research utilizing these strategies of case ascertainment, researchers ought to maintain these issues firmly in thoughts.
Help: Funding for this examine was offered by Baxter Healthcare Company, McGaw Park, IL. Baxter Healthcare Company assisted with the interpretation of information, drafting of the manuscript, and the choice to submit the manuscript for publication.
Monetary Disclosure: Ms. Taneja, Mr. Berger, and Dr. Oster are workers of Coverage Evaluation Inc., a contract analysis group that has obtained funding from Baxter Healthcare Company in addition to different biomedical corporations. Dr. Lamerato, Mr. Wolff, and Mr. Sheehan are workers of Henry Ford Well being System. Dr. Sloand is an worker of Baxter Healthcare Company, the sponsor of the examine, and reviews proudly owning inventory in that firm. Mr. Inglese is a former worker of Baxter Healthcare Company, and is presently employed by Hollister Included.
United States, Division of Well being and Human Companies, Nationwide Institutes of Well being, Nationwide Institute of Diabetes and Digestive and Kidney Illnesses, US Renal Information System (USRDS). USRDS 2012 Annual Information Report: Atlas of Persistent Kidney Illness and Finish-Stage Renal Illness in the USA. Bethesda, MD: USRDS; 2012.
Shih YC, Guo A, Simply PM, Mujais S.
Affect of preliminary dialysis modality and modality switches on Medicare expenditures on end-stage renal illness sufferers. Kidney Int
[PubMed] [Google Scholar]
Berger A, Edelsberg J, Inglese GW, Bhattacharyya SK, Oster G.
Price comparability of peritoneal dialysis versus hemodialysis in end-stage renal illness. Am J Manag Care
[PubMed] [Google Scholar]
Berger A, Edelsberg J, Inglese G, Bhattacharyya SK, Oster G.
Identification of sufferers receiving peritoneal dialysis utilizing medical insurance claims information. Clin Ther
[PubMed] [Google Scholar]
Robbins JD, Kim JJ, Zdon G, Chan WW, Jones J.
Useful resource use and affected person care related to persistent kidney illness in a managed care setting. J Manag Care Pharm
[PubMed] [Google Scholar]
Winkelmayer WC, Glynn RJ, Mittleman MA, Levin R, Pliskin JS, Avorn J.
Evaluating mortality of aged sufferers on hemodialysis versus peritoneal dialysis: A propensity rating method. J Am Soc Nephrol
[PubMed] [Google Scholar]
Pressing-start peritoneal dialysis: A top quality enchancment report. Am J Kidney Dis
2012; 59:400–8 [PubMed] [Google Scholar]
Peritoneal dialysis utilization and final result: What are we dealing with?
Perit Dial Int
2007; 27(Suppl 2):S42–7
[PubMed] [Google Scholar]
Liebman SE, Bushinsky DA, Dolan JG, Veazie P.
Variations between dialysis modality choice and initiation. Am J Kidney Dis
[PMC free article] [PubMed] [Google Scholar]